Medical Grand Rounds

New clinical trials and observational studies are shedding light on ways to improve the health of elderly patients. Here is a brief summary of these trials and how they might influence your clinical practice.

EXERCISE HAS NEWLY DISCOVERED BENEFITS

According to government data,¹ exercise has a dose-dependent effect on rates of all-cause mortality: the more hours one exercises per week, the lower the risk of death. The difference in risk is most pronounced as one goes from no exercise to about 3 hours of exercise per week; above 3 hours per week, the curve flattens out but continues to decline. Hence, we advise patients to engage in about 30 minutes of moderate-intensity exercise every day.

Lately, physical exercise has been found to have other, unexpected benefits.

Exercise helps cognition

ERICKSON KI, PRAKASH RS, VOSS MW, ET AL. AEROBIC FITNESS IS ASSOCIATED WITH HIPPOCAMPAL VOLUME IN ELDERLY HUMANS. HIPPOCAMPUS 2009; 19:1030–1039.

ETGEN T, SANDER D, HUNTGEBURTH U, POPPERT H, FÖRSTL H, BICKEL H. PHYSICAL ACTIVITY AND INCIDENT COGNITIVE IMPAIRMENT IN ELDERLY PERSONS: THE INVADE STUDY. ARCH INTERN MED 2010; 170:186–193.

The hippocampus is a structure deep in the brain that is involved in short-term memory. It atrophies with age, more so with dementia. Erickson² found a correlation between aerobic fitness (as measured by maximum oxygen consumption), hippocampal volume, and spatial memory performance.

Etgen and colleagues³ studied nearly 4,000 older adults in Bavaria for 2 years. Among those reporting no physical activity, 21.4% had cognitive impairment at baseline, compared with 7.3% of those with high activity at baseline. Following those without cognitive impairment over a 2-year period, they found the incidence of new cognitive impairment was 13.9% in those with no physical activity at baseline, 6.7% in those with moderate activity, and 5.1% in those with high activity.

Exercise boosts the effect of influenza vaccine

WOODS JA, KEYLOCK KT, LOWDER T, ET AL. CARDIOVASCULAR EXERCISE TRAINING EXTENDS INFLUENZA VACCINE SEROPROTECTION IN SEDENTARY OLDER ADULTS: THE IMMUNE FUNCTION INTERVENTION TRIAL. J AM GERIATR SOC 2009; 57:2183–2191.

In a study in 144 sedentary but healthy older adults (ages 60 to 83), Woods et al⁴ randomized the participants to undergo either flexibility or cardiovascular training for 10 months, starting 4 months before their annual influenza shot. Exercise extended the duration of antibody protection, with more participants in the cardiovascular group than in the flexibility group showing protection at 24 weeks against all three strains covered by the vaccine: H1N1, H3N2, and influenza B.

PREVENTING FRACTURES

Each year, about 30% of people age 65 or older fall, sustaining serious injuries in 5% to 10% of cases. Unintentional falls are the main cause of hip fractures, which number 300,000 per year. They are also a common cause of death.

Vitamin D prevents fractures, but can there be too much of a good thing?

BISCHOFF-FERRARI HA, WILLETT WC, WONG JB, ET AL. PREVENTION OF NONVERTEBRAL FRACTURES WITH ORAL VITAMIN D AND DOSE DEPENDENCY: A META-ANALYSIS OF RANDOMIZED CONTROLLED TRIALS. ARCH INTERN MED 2009; 169:551–561.

SANDERS KM, STUART AL, WILLIAMSON EJ, ET AL. ANNUAL HIGH-DOSE ORAL VITAMIN D AND FALLS AND FRACTURES IN OLDER WOMEN: A RANDOMIZED CONTROLLED TRIAL. JAMA 2010; 303:1815–1822.

Bischoff-Ferrari⁵ performed a meta-analysis of 12 randomized controlled trials of oral supplemental vitamin D₃ for preventing nonvertebral fractures in people age 65 and older, and eight trials for preventing hip fractures in the same age group. They found that the higher the daily dose of vitamin D, the lower the relative risk of hip fracture. The threshold dose at which supplementation significantly reduced the risk of falling was about 400 units per day. Higher doses of vitamin D reduced both falls and hip fractures by about 20%. The maximal effect was seen with studies using the maximum daily doses, ie, 770 to 800 units per day—not megadoses, but more than most Americans are taking. The threshold serum level of vitamin D of significance was 60 nmol/L (24 ng/mL).

Of interest, the effect on fractures was independent of calcium supplementation. This is important because calcium supplementation over and above ordinary dietary intake may increase the risk of cardiovascular events.^6,7

Despite the benefits of vitamin D, too much may be too much of a good thing. Sanders et al⁸ performed a double-blind, placebo-controlled trial in 2,256 community-dwelling women, age 70 or older, who were considered to be at high risk for fractures. Half received a large oral dose (500,000 units) once a year for 3 to 5 years, and half got placebo. Their initial serum vitamin D level was 49 nmol/L; the level 30 days after a dose in the treatment group was 120 nmol/L.

Contrary to expectations, the incidence of falls was 15% higher in the vitamin D group than in the placebo group (P = .03), and the incidence of fractures was 26% higher (P = .047). The falls and fractures tended to cluster in the first 3 months after the dose in the active treatment group, when serum vitamin D levels were highest.

Comments. Unless future studies suggest a benefit to megadoses of vitamin D or prove calcium supplementation greater than 1,000 mg is safe, the optimal daily intake of vitamin D is likely 1,000 units, with approximately 200 units from diet and 800 units from supplements. A diet rich in low-fat dairy products may not require calcium supplementation. In those consuming a low-calcium diet, supplements of 500 to 1,000 mg/day are likely adequate.

Denosumab, a new drug for preventing fractures

CUMMINGS SR, SAN MARTIN J, MCCLUNG MR, ET AL; FREEDOM TRIAL. DENOSUMAB FOR PREVENTION OF FRACTURES IN POSTMENOPAUSAL WOMEN WITH OSTEOPOROSIS. N ENGL J MED 2009; 361:756–765.

SMITH MR, EGERDIE B, HERNÁNDEZ TORIZ N, ET AL; DENOSUMAB HALT PROSTATE CANCER STUDY GROUP. DENOSUMAB IN MEN RECEIVING ANDROGEN-DEPRIVATION THERAPY FOR PROSTATE CANCER. N ENGL J MED 2009; 361:745–755.

Denosumab (Prolia) is the first of a new class of drugs for the treatment of osteoporosis. It is a monoclonal antibody and member of the tumor necrosis factor superfamily that binds to the receptor activator nuclear factor kappa B (RANK) ligand. It has an antiresorptive effect, preventing osteoclast differentiation and activation. It is given by subcutaneous injection of 60 mg every 6 months; it is cleared by a nonrenal mechanism.

In a randomized controlled trial in 7,868 women between the ages of 60 and 90 who had osteoporosis, Cummings et al⁹ reported that denosumab reduced the 3-year incidence of vertebral fractures by 68% (P < .001), reduced the incidence of hip fractures by 40% (P = .01), and reduced the incidence of nonvertebral fractures by 20% (P = .01). In a trial in men receiving androgen deprivation therapy for prostate cancer, Smith et al¹⁰ reported that denosumab reduced the incidence of vertebral fracture by 62% (P = .006).

Comment. Denosumab was approved by the US Food and Drug Administration (FDA) on June 1, 2010, and is emerging in specialty clinics at the time of this publication. Its potential impact on clinical care is not yet known. It is costly—about $825 (average wholesale price) per injection—but since it is given by injection it may be easier than a yearly infusion of zoledronic acid (Reclast). It has the potential to suppress immune function, although this was not reported in the clinical trials. It may ultimately have a role in treating osteoporosis in men and women, prostate cancer following androgen deprivation, metastatic prostate cancer, metastatic breast cancer, osteoporosis with renal impairment, and other diseases.

DIALYSIS IN THE ELDERLY: A BLEAK STORY

KURELLA TAMURA M, COVINSKY KE, CHERTOW GM, YAFFE K, LANDEFELD CS, MCCOLLOCH CE. FUNCTIONAL STATUS OF ELDERLY ADULTS BEFORE AND AFTER INITIATION OF DIALYSIS. N ENGL J MED 2009; 361:1539–1547.

JASSAL SV, CHIU E, HLADUNEWITH M. LOSS OF INDEPENDENCE IN PATIENTS STARTING DIALYSIS AT 80 YEARS OF AGE OR OLDER (LETTER). N ENGL J MED 2009; 361:1612–1613.

Nursing home residents account for 4% of all patients in end-stage renal disease. However, the benefits of dialysis in older patients are uncertain. The mortality rate during the first year of dialysis is 35% in patients 70 years of age and older and 50% in patients 80 years and older.

Is dialysis helpful in the elderly, ie, does it improve survival and function?

Kurella Tamura et al¹¹ retrospectively identified 3,702 nursing home residents starting dialysis in whom functional assessments had been done. The numbers told a bleak story. Initiation of dialysis was associated with a sharp decline in functional status, as reflected in an increase of 2.8 points on the 28-point Minimum Data Set–Activities of Daily Living (MDS-ADL) scale (the higher the score, the worse the function). MDS-ADL scores stabilized at a plateau for about 6 months and then continued to decline. Moreover, at 12 months, 58% of the patients had died.

The MDS-ADL score is based on seven components: eating, bed mobility, locomotion, transferring, toileting, hygiene, and dressing; function declined in all of these areas when patients started dialysis.

Patients were more likely to decline in activities of daily living after starting dialysis if they were older, were white, had cerebrovascular disease, had a diagnosis of dementia, were hospitalized at the start of dialysis, or had a serum albumin level lower than 3.5 g/dL.

The same thing happens to elders living in the community when they start dialysis. Jassal and colleagues¹² reported that, of 97 community-dwelling patients (mean age 85), 46 (47%) were dead 2 years after starting dialysis. Although 76 (78%) had been living independently at the start of dialysis, only 11 (11%) were still doing so at 2 years.

Comment. These findings indicate that we do not know if hemodialysis improves survival. Hemodialysis may buy about 3 months of stable function, but it clearly does not restore function.

Is this the best we can do? Standard hemodialysis may have flaws, and nocturnal dialysis and peritoneal dialysis are used more in other countries. These dialysis techniques require more study in our older population. The lesson from these two publications on dialysis is that we should attend more carefully to slowing the decline in renal function before patients reach end-stage renal disease.

DABIGATRAN: AN ALTERNATIVE TO WARFARIN FOR ATRIAL FIBRILLATION

CONNOLLY SJ, EZEKOWITZ MD, YUSUF S, ET AL; RE-LY STEERING COMMITTEE AND INVESTIGATORS. DABIGATRAN VERSUS WARFARIN IN PATIENTS WITH ATRIAL FIBRILLATION. N ENGL J MED 2009; 361:1139–1151.

Atrial fibrillation is common, affecting 2.2 million adults. The median age of people who have atrial fibrillation is 75 years, and it is the most common arrhythmia in the elderly. Some 20% of ischemic strokes are attributed to it.^13–15

Warfarin (Coumadin) is still the mainstay of treatment to prevent stroke in patients with atrial fibrillation. In an analysis of pooled data from five clinical trials,¹⁶ the relative risk reduction with warfarin was about 68% in the overall population (number needed to treat 32), 51% in people older than 75 years with no other risk factors (number needed to treat 56), and 85% in people older than 75 years with one or more risk factors (number needed to treat 15).

But warfarin carries a risk of bleeding, and its dose must be periodically adjusted on the basis of the international normalized ratio (INR) of the prothrombin time, so it carries a burden of laboratory monitoring. It is less safe in people who eat erratically, resulting in wide fluctuations in the INR.

Dabigatran (Pradaxa), a direct thrombin inhibitor, is expected to become an alternative to warfarin. It has been approved in Europe but not yet in the United States.

Connolly et al,¹⁷ in a randomized, double-blind trial, assigned 18,113 patients who had atrial fibrillation to receive either dabigatran 110 or 150 mg twice daily or adjusted-dose warfarin in an unblinded fashion. At 2 years, the rates of stroke and systemic embolism were about the same with dabigatran 110 mg as with warfarin but were lower with dabigatran 150 mg (relative risk 0.66, 95% confidence interval [CI] 0.53–0.82, P < .001). The rate of major bleeding was lower with dabigatran 110 mg than with warfarin (2.71% per year vs 3.36% per year, P = .003), but it was similar with dabigatran 150 mg (3.11% per year). Rates of life-threatening bleeding were 1.80% with warfarin, 1.22% with dabigatran 110 mg (P < .05), and 1.45% with dabigatran 150 mg (P < .05).

Comment. I suspect that warfarin’s days are numbered. Dabigatran 110 or 150 mg was as safe and as effective as warfarin in clinical trials, and probably will be more effective than warfarin in clinical practice. It will also probably be safer than warfarin in clinical practice, particularly in challenging settings such as long-term care. On the other hand, it will likely be much more expensive than warfarin.

DEMENTIA

Adverse effects of cholinesterase inhibitors

GILL SS, ANDERSON GM, FISCHER HD, ET AL. SYNCOPE AND ITS CONSEQUENCES IN PATIENTS WITH DEMENTIA RECEIVING CHOLINESTERASE INHIBITORS: A POPULATION-BASED COHORT STUDY. ARCH INTERN MED 2009; 169:867–873.

Cholinesterase inhibitors, eg, donepezil (Aricept), galantamine (Razadyne), and rivastigmine (Exelon), are commonly used to treat Alzheimer disease. However, these drugs carry risks of serious adverse effects.

Gill et al¹⁸ retrospectively reviewed a database from Ontario, Canada, and identified about 20,000 community-dwelling elderly persons admitted to the hospital who had been prescribed cholinesterase inhibitors and about three times as many matched controls.

Several adverse events were more frequent in people receiving cholinesterase inhibitors. Findings (events per 1,000 person-years):

• Hospital visits for syncope: 31.5 vs 18.6, adjusted hazard ratio (HR) 1.76, 95% CI 1.57–1.98
• Hip fractures: 22.4 vs 19.8, HR 1.18, 85% CI 1.04–1.34
• Hospital visits for bradycardia: 6.9 vs 4.4, HR 1.69, 95% CI 1.32–2.15
• Permanent pacemaker insertion: 4.7 vs 3.3, HR 1.49, 95% CI 1.12–2.00.

Comment. This study adds to the concerns that cholinesterase inhibitors, which have only modest cognitive benefits, may increase the risk of falls, injury, and need for pacemaker placement in demented patients. A low threshold to stop medications in this class should be considered when a patient on a cholinesterase inhibitor presents with bradycardia, falls, and syncope.

IVERSON DJ, GRONSETH GS, REGER MA, ET AL; STANDARDS SUBCOMMITTEE OF THE AMERICAN ACADEMY OF NEUROLOGY. PRACTICE PARAMETER UPDATE: EVALUATION AND MANAGEMENT OF DRIVING RISK IN DEMENTIA: REPORT OF THE QUALITY STANDARDS SUBCOMMITTEE OF THE AMERICAN ACADEMY OF NEUROLOGY. NEUROLOGY 2010; 74:1316–1324.

The Clinical Dementia Rating (CDR) is a simple scale that should be applied by clinicians to describe stage of dementia in patients with Alzheimer disease. This scale can be useful in a variety of settings, from prescribing antidementia drugs to determining whether a patient should still drive. Although research protocols utilize a survey or semistructured interview to derive the stage, the clinician can estimate the stage easily in the office, particularly if there is an informant who can comment on performance outside the office.

There are four stages to the CDR¹⁹:

• 0: No dementia
• 0.5: Mild memory deficit but intact function
• 1.0: Moderate memory loss with mild functional impairment
• 2.0: Severe memory loss, moderate functional impairment
• 3.0: Severe memory loss, no significant function outside of the house.

Comment. The first stage (0.5, mild memory deficit but intact function) corresponds to “mild cognitive impairment.” In the clinic, these patients tend to take more notes. They come to the appointment with a little book and they write everything down so they don’t forget. They do arrive at their appointments on time; they are not crashing the car; they are paying their bills.

Patients with CDR stage 1.0 dementia (moderate memory loss with mild functional impairment) may miss appointments, they may confuse their medications, and they may have problems driving. They are still taking care of their basic needs, and they show up for appointments acceptably washed and dressed. However, they are likely having trouble shopping and managing their finances.

Patients with severe memory loss and moderate functional impairment (CDR stage 2.0) may not realize they haven’t bathed for a week or have worn the same clothes repeatedly. They are having trouble with basic activities of daily living, such as bathing and toilet hygiene. However, if you were to encounter them socially and didn’t talk to them for too long, you might think they were normal.

Those with severe memory loss and no significant function outside the house (CDR stage 3.0) are the most severely disabled. Dementia in these individuals is recognizable at a glance, from across the room.

Alzheimer patients progress through the stages, from CDR stage 0.5 at about 1 year to stage 1 by about 2 years, to stage 2 by 5 years, and to stage 3 at 8 or 9 years.²⁰

In prescribing antidementia medications. The CDR can help with prescribing antidementia drugs. No medications are approved by the FDA for stage 0 or 0.5. Cholinesterase inhibitors are approved for stages 1, 2, and 3; memantine (Namenda) is approved for stages 2 and 3.

Advising about driving. The CDR is the only risk predictor with a quality-of-evidence rating of A. More than half of people with stage 0.5 memory impairment are safe drivers; fewer than half of those with stage 1.0 are still safe drivers; and patients with stage 2.0 dementia should not be driving at all.²¹ An adverse rating by a caregiver carries a quality-of-evidence rating of B. Predictors of driving risk with a quality-of-evidence rating of C are decreased mileage due to self-restriction, agitation, or aggression; a crash in the past 1 to 5 years; a citation in the past 2 to 3 years; and a Folstein Mini-Mental State Examination score of 24 or less. Studies also show that a memory-impaired person’s self-rating of safe driving ability or of assurance that he or she avoids unsafe situations is not reliable.²¹

DELIRIUM

Delirium goes by a number of synonyms, eg, “sundowning,” acute confusional state, acute change in mental status, metabolic encephalopathy, toxic encephalopathy (psychosis), acute brain syndrome, and acute toxic psychosis.

Delirium is common in hospitalized elderly patients, occurring in 11% to 42% of elderly hospitalized patients overall, up to 53% of elderly surgical patients on regular hospital floors, 80% of elderly surgical patients in intensive care, and about half of elderly patients after undergoing coronary artery bypass grafting. Unfortunately, it is undiagnosed in 30% to 60% of cases.^22–24

Many pathways can lead to delirium, including hypoxemia, metabolic derangement, drug effects, systemic inflammation, and infection.²⁵

Outcomes can vary from full recovery to death. After 1 year, 50% of those who leave the hospital with some evidence of delirium have not regained their baseline function. Delirium also increases the cost of care and the risk of institutionalization.

Delirium can accelerate dementia

FONG TG, JONES RN, SHI P, ET AL. DELIRIUM ACCELERATES COGNITIVE DECLINE IN ALZHEIMER DISEASE. NEUROLOGY 2009; 72:1570–1575.

Delirium accelerates the course of dementia in patients who had some evidence of dementia before they entered the hospital. Often, the change is noticeable by the family.²⁶

Preventing delirium

INOUYE SK BOGARDUS ST JR, CHARPENTIER PA, ET AL. A MULTICOMPONENT INTERVENTION TO PREVENT DELIRIUM IN HOSPITALIZED OLDER PATIENTS. N ENGL J MED 1999; 340:669–676.

LUNDSTRÖM M, OLOFSSON B, STENVALL M, ET AL. POSTOPERATIVE DELIRIUM IN OLD PATIENTS WITH FEMORAL NECK FRACTURE: A RANDOMIZED INTERVENTION STUDY. AGING CLIN EXP RES 2007; 19:178–186.

Delirium can often be prevented. In a report published in 1999, Inouye et al²⁷ described the outcomes of a program to prevent delirium in hospitalized medically ill elderly patients. Interventions were aimed at optimizing cognitive function, preventing sleep deprivation, avoiding immobility, improving vision and hearing, and treating dehydration. The incidence of delirium was 9.9% in the intervention group vs 15% in the control group, a 40% reduction (P < .05).

Lundström et al²⁸ implemented a similar program for elderly patients with hip fractures. Interventions included staff education and teamwork; active prevention, detection, and treatment of delirium; transfusions if hemoglobin levels were less than 10 g/dL; prompt removal of indwelling urinary catheters, with screening for urinary retention; active prevention and treatment of constipation; and protein-enriched meals. The incidence of delirium was 55% in the intervention group vs 75% in the control group, a 27% reduction.

Comment. Although we have long known that the risk of delirium in medical and surgical patients can be reduced, most hospitals do not have systematic programs to detect delirium and reduce its incidence. Hopefully, reduction in delirium risk will also reduce its adverse consequences, including worsening of dementia and increased mortality.

New clinical trials and observational studies are shedding light on ways to improve the health of elderly patients. Here is a brief summary of these trials and how they might influence your clinical practice.

EXERCISE HAS NEWLY DISCOVERED BENEFITS

According to government data,¹ exercise has a dose-dependent effect on rates of all-cause mortality: the more hours one exercises per week, the lower the risk of death. The difference in risk is most pronounced as one goes from no exercise to about 3 hours of exercise per week; above 3 hours per week, the curve flattens out but continues to decline. Hence, we advise patients to engage in about 30 minutes of moderate-intensity exercise every day.

Lately, physical exercise has been found to have other, unexpected benefits.

Exercise helps cognition

ERICKSON KI, PRAKASH RS, VOSS MW, ET AL. AEROBIC FITNESS IS ASSOCIATED WITH HIPPOCAMPAL VOLUME IN ELDERLY HUMANS. HIPPOCAMPUS 2009; 19:1030–1039.

ETGEN T, SANDER D, HUNTGEBURTH U, POPPERT H, FÖRSTL H, BICKEL H. PHYSICAL ACTIVITY AND INCIDENT COGNITIVE IMPAIRMENT IN ELDERLY PERSONS: THE INVADE STUDY. ARCH INTERN MED 2010; 170:186–193.

The hippocampus is a structure deep in the brain that is involved in short-term memory. It atrophies with age, more so with dementia. Erickson² found a correlation between aerobic fitness (as measured by maximum oxygen consumption), hippocampal volume, and spatial memory performance.

Etgen and colleagues³ studied nearly 4,000 older adults in Bavaria for 2 years. Among those reporting no physical activity, 21.4% had cognitive impairment at baseline, compared with 7.3% of those with high activity at baseline. Following those without cognitive impairment over a 2-year period, they found the incidence of new cognitive impairment was 13.9% in those with no physical activity at baseline, 6.7% in those with moderate activity, and 5.1% in those with high activity.

Exercise boosts the effect of influenza vaccine

WOODS JA, KEYLOCK KT, LOWDER T, ET AL. CARDIOVASCULAR EXERCISE TRAINING EXTENDS INFLUENZA VACCINE SEROPROTECTION IN SEDENTARY OLDER ADULTS: THE IMMUNE FUNCTION INTERVENTION TRIAL. J AM GERIATR SOC 2009; 57:2183–2191.

In a study in 144 sedentary but healthy older adults (ages 60 to 83), Woods et al⁴ randomized the participants to undergo either flexibility or cardiovascular training for 10 months, starting 4 months before their annual influenza shot. Exercise extended the duration of antibody protection, with more participants in the cardiovascular group than in the flexibility group showing protection at 24 weeks against all three strains covered by the vaccine: H1N1, H3N2, and influenza B.

PREVENTING FRACTURES

Each year, about 30% of people age 65 or older fall, sustaining serious injuries in 5% to 10% of cases. Unintentional falls are the main cause of hip fractures, which number 300,000 per year. They are also a common cause of death.

Vitamin D prevents fractures, but can there be too much of a good thing?

BISCHOFF-FERRARI HA, WILLETT WC, WONG JB, ET AL. PREVENTION OF NONVERTEBRAL FRACTURES WITH ORAL VITAMIN D AND DOSE DEPENDENCY: A META-ANALYSIS OF RANDOMIZED CONTROLLED TRIALS. ARCH INTERN MED 2009; 169:551–561.

SANDERS KM, STUART AL, WILLIAMSON EJ, ET AL. ANNUAL HIGH-DOSE ORAL VITAMIN D AND FALLS AND FRACTURES IN OLDER WOMEN: A RANDOMIZED CONTROLLED TRIAL. JAMA 2010; 303:1815–1822.

Bischoff-Ferrari⁵ performed a meta-analysis of 12 randomized controlled trials of oral supplemental vitamin D₃ for preventing nonvertebral fractures in people age 65 and older, and eight trials for preventing hip fractures in the same age group. They found that the higher the daily dose of vitamin D, the lower the relative risk of hip fracture. The threshold dose at which supplementation significantly reduced the risk of falling was about 400 units per day. Higher doses of vitamin D reduced both falls and hip fractures by about 20%. The maximal effect was seen with studies using the maximum daily doses, ie, 770 to 800 units per day—not megadoses, but more than most Americans are taking. The threshold serum level of vitamin D of significance was 60 nmol/L (24 ng/mL).

Of interest, the effect on fractures was independent of calcium supplementation. This is important because calcium supplementation over and above ordinary dietary intake may increase the risk of cardiovascular events.^6,7

Despite the benefits of vitamin D, too much may be too much of a good thing. Sanders et al⁸ performed a double-blind, placebo-controlled trial in 2,256 community-dwelling women, age 70 or older, who were considered to be at high risk for fractures. Half received a large oral dose (500,000 units) once a year for 3 to 5 years, and half got placebo. Their initial serum vitamin D level was 49 nmol/L; the level 30 days after a dose in the treatment group was 120 nmol/L.

Contrary to expectations, the incidence of falls was 15% higher in the vitamin D group than in the placebo group (P = .03), and the incidence of fractures was 26% higher (P = .047). The falls and fractures tended to cluster in the first 3 months after the dose in the active treatment group, when serum vitamin D levels were highest.

Comments. Unless future studies suggest a benefit to megadoses of vitamin D or prove calcium supplementation greater than 1,000 mg is safe, the optimal daily intake of vitamin D is likely 1,000 units, with approximately 200 units from diet and 800 units from supplements. A diet rich in low-fat dairy products may not require calcium supplementation. In those consuming a low-calcium diet, supplements of 500 to 1,000 mg/day are likely adequate.

Denosumab, a new drug for preventing fractures

CUMMINGS SR, SAN MARTIN J, MCCLUNG MR, ET AL; FREEDOM TRIAL. DENOSUMAB FOR PREVENTION OF FRACTURES IN POSTMENOPAUSAL WOMEN WITH OSTEOPOROSIS. N ENGL J MED 2009; 361:756–765.

SMITH MR, EGERDIE B, HERNÁNDEZ TORIZ N, ET AL; DENOSUMAB HALT PROSTATE CANCER STUDY GROUP. DENOSUMAB IN MEN RECEIVING ANDROGEN-DEPRIVATION THERAPY FOR PROSTATE CANCER. N ENGL J MED 2009; 361:745–755.

Denosumab (Prolia) is the first of a new class of drugs for the treatment of osteoporosis. It is a monoclonal antibody and member of the tumor necrosis factor superfamily that binds to the receptor activator nuclear factor kappa B (RANK) ligand. It has an antiresorptive effect, preventing osteoclast differentiation and activation. It is given by subcutaneous injection of 60 mg every 6 months; it is cleared by a nonrenal mechanism.

In a randomized controlled trial in 7,868 women between the ages of 60 and 90 who had osteoporosis, Cummings et al⁹ reported that denosumab reduced the 3-year incidence of vertebral fractures by 68% (P < .001), reduced the incidence of hip fractures by 40% (P = .01), and reduced the incidence of nonvertebral fractures by 20% (P = .01). In a trial in men receiving androgen deprivation therapy for prostate cancer, Smith et al¹⁰ reported that denosumab reduced the incidence of vertebral fracture by 62% (P = .006).

Comment. Denosumab was approved by the US Food and Drug Administration (FDA) on June 1, 2010, and is emerging in specialty clinics at the time of this publication. Its potential impact on clinical care is not yet known. It is costly—about $825 (average wholesale price) per injection—but since it is given by injection it may be easier than a yearly infusion of zoledronic acid (Reclast). It has the potential to suppress immune function, although this was not reported in the clinical trials. It may ultimately have a role in treating osteoporosis in men and women, prostate cancer following androgen deprivation, metastatic prostate cancer, metastatic breast cancer, osteoporosis with renal impairment, and other diseases.

DIALYSIS IN THE ELDERLY: A BLEAK STORY

KURELLA TAMURA M, COVINSKY KE, CHERTOW GM, YAFFE K, LANDEFELD CS, MCCOLLOCH CE. FUNCTIONAL STATUS OF ELDERLY ADULTS BEFORE AND AFTER INITIATION OF DIALYSIS. N ENGL J MED 2009; 361:1539–1547.

JASSAL SV, CHIU E, HLADUNEWITH M. LOSS OF INDEPENDENCE IN PATIENTS STARTING DIALYSIS AT 80 YEARS OF AGE OR OLDER (LETTER). N ENGL J MED 2009; 361:1612–1613.

Nursing home residents account for 4% of all patients in end-stage renal disease. However, the benefits of dialysis in older patients are uncertain. The mortality rate during the first year of dialysis is 35% in patients 70 years of age and older and 50% in patients 80 years and older.

Is dialysis helpful in the elderly, ie, does it improve survival and function?

Kurella Tamura et al¹¹ retrospectively identified 3,702 nursing home residents starting dialysis in whom functional assessments had been done. The numbers told a bleak story. Initiation of dialysis was associated with a sharp decline in functional status, as reflected in an increase of 2.8 points on the 28-point Minimum Data Set–Activities of Daily Living (MDS-ADL) scale (the higher the score, the worse the function). MDS-ADL scores stabilized at a plateau for about 6 months and then continued to decline. Moreover, at 12 months, 58% of the patients had died.

The MDS-ADL score is based on seven components: eating, bed mobility, locomotion, transferring, toileting, hygiene, and dressing; function declined in all of these areas when patients started dialysis.

Patients were more likely to decline in activities of daily living after starting dialysis if they were older, were white, had cerebrovascular disease, had a diagnosis of dementia, were hospitalized at the start of dialysis, or had a serum albumin level lower than 3.5 g/dL.

The same thing happens to elders living in the community when they start dialysis. Jassal and colleagues¹² reported that, of 97 community-dwelling patients (mean age 85), 46 (47%) were dead 2 years after starting dialysis. Although 76 (78%) had been living independently at the start of dialysis, only 11 (11%) were still doing so at 2 years.

Comment. These findings indicate that we do not know if hemodialysis improves survival. Hemodialysis may buy about 3 months of stable function, but it clearly does not restore function.

Is this the best we can do? Standard hemodialysis may have flaws, and nocturnal dialysis and peritoneal dialysis are used more in other countries. These dialysis techniques require more study in our older population. The lesson from these two publications on dialysis is that we should attend more carefully to slowing the decline in renal function before patients reach end-stage renal disease.

DABIGATRAN: AN ALTERNATIVE TO WARFARIN FOR ATRIAL FIBRILLATION

CONNOLLY SJ, EZEKOWITZ MD, YUSUF S, ET AL; RE-LY STEERING COMMITTEE AND INVESTIGATORS. DABIGATRAN VERSUS WARFARIN IN PATIENTS WITH ATRIAL FIBRILLATION. N ENGL J MED 2009; 361:1139–1151.

Atrial fibrillation is common, affecting 2.2 million adults. The median age of people who have atrial fibrillation is 75 years, and it is the most common arrhythmia in the elderly. Some 20% of ischemic strokes are attributed to it.^13–15

Warfarin (Coumadin) is still the mainstay of treatment to prevent stroke in patients with atrial fibrillation. In an analysis of pooled data from five clinical trials,¹⁶ the relative risk reduction with warfarin was about 68% in the overall population (number needed to treat 32), 51% in people older than 75 years with no other risk factors (number needed to treat 56), and 85% in people older than 75 years with one or more risk factors (number needed to treat 15).

But warfarin carries a risk of bleeding, and its dose must be periodically adjusted on the basis of the international normalized ratio (INR) of the prothrombin time, so it carries a burden of laboratory monitoring. It is less safe in people who eat erratically, resulting in wide fluctuations in the INR.

Dabigatran (Pradaxa), a direct thrombin inhibitor, is expected to become an alternative to warfarin. It has been approved in Europe but not yet in the United States.

Connolly et al,¹⁷ in a randomized, double-blind trial, assigned 18,113 patients who had atrial fibrillation to receive either dabigatran 110 or 150 mg twice daily or adjusted-dose warfarin in an unblinded fashion. At 2 years, the rates of stroke and systemic embolism were about the same with dabigatran 110 mg as with warfarin but were lower with dabigatran 150 mg (relative risk 0.66, 95% confidence interval [CI] 0.53–0.82, P < .001). The rate of major bleeding was lower with dabigatran 110 mg than with warfarin (2.71% per year vs 3.36% per year, P = .003), but it was similar with dabigatran 150 mg (3.11% per year). Rates of life-threatening bleeding were 1.80% with warfarin, 1.22% with dabigatran 110 mg (P < .05), and 1.45% with dabigatran 150 mg (P < .05).

Comment. I suspect that warfarin’s days are numbered. Dabigatran 110 or 150 mg was as safe and as effective as warfarin in clinical trials, and probably will be more effective than warfarin in clinical practice. It will also probably be safer than warfarin in clinical practice, particularly in challenging settings such as long-term care. On the other hand, it will likely be much more expensive than warfarin.

DEMENTIA

Adverse effects of cholinesterase inhibitors

GILL SS, ANDERSON GM, FISCHER HD, ET AL. SYNCOPE AND ITS CONSEQUENCES IN PATIENTS WITH DEMENTIA RECEIVING CHOLINESTERASE INHIBITORS: A POPULATION-BASED COHORT STUDY. ARCH INTERN MED 2009; 169:867–873.

Cholinesterase inhibitors, eg, donepezil (Aricept), galantamine (Razadyne), and rivastigmine (Exelon), are commonly used to treat Alzheimer disease. However, these drugs carry risks of serious adverse effects.

Gill et al¹⁸ retrospectively reviewed a database from Ontario, Canada, and identified about 20,000 community-dwelling elderly persons admitted to the hospital who had been prescribed cholinesterase inhibitors and about three times as many matched controls.

Several adverse events were more frequent in people receiving cholinesterase inhibitors. Findings (events per 1,000 person-years):

• Hospital visits for syncope: 31.5 vs 18.6, adjusted hazard ratio (HR) 1.76, 95% CI 1.57–1.98
• Hip fractures: 22.4 vs 19.8, HR 1.18, 85% CI 1.04–1.34
• Hospital visits for bradycardia: 6.9 vs 4.4, HR 1.69, 95% CI 1.32–2.15
• Permanent pacemaker insertion: 4.7 vs 3.3, HR 1.49, 95% CI 1.12–2.00.

Comment. This study adds to the concerns that cholinesterase inhibitors, which have only modest cognitive benefits, may increase the risk of falls, injury, and need for pacemaker placement in demented patients. A low threshold to stop medications in this class should be considered when a patient on a cholinesterase inhibitor presents with bradycardia, falls, and syncope.

IVERSON DJ, GRONSETH GS, REGER MA, ET AL; STANDARDS SUBCOMMITTEE OF THE AMERICAN ACADEMY OF NEUROLOGY. PRACTICE PARAMETER UPDATE: EVALUATION AND MANAGEMENT OF DRIVING RISK IN DEMENTIA: REPORT OF THE QUALITY STANDARDS SUBCOMMITTEE OF THE AMERICAN ACADEMY OF NEUROLOGY. NEUROLOGY 2010; 74:1316–1324.

The Clinical Dementia Rating (CDR) is a simple scale that should be applied by clinicians to describe stage of dementia in patients with Alzheimer disease. This scale can be useful in a variety of settings, from prescribing antidementia drugs to determining whether a patient should still drive. Although research protocols utilize a survey or semistructured interview to derive the stage, the clinician can estimate the stage easily in the office, particularly if there is an informant who can comment on performance outside the office.

There are four stages to the CDR¹⁹:

• 0: No dementia
• 0.5: Mild memory deficit but intact function
• 1.0: Moderate memory loss with mild functional impairment
• 2.0: Severe memory loss, moderate functional impairment
• 3.0: Severe memory loss, no significant function outside of the house.

Comment. The first stage (0.5, mild memory deficit but intact function) corresponds to “mild cognitive impairment.” In the clinic, these patients tend to take more notes. They come to the appointment with a little book and they write everything down so they don’t forget. They do arrive at their appointments on time; they are not crashing the car; they are paying their bills.

Patients with CDR stage 1.0 dementia (moderate memory loss with mild functional impairment) may miss appointments, they may confuse their medications, and they may have problems driving. They are still taking care of their basic needs, and they show up for appointments acceptably washed and dressed. However, they are likely having trouble shopping and managing their finances.

Patients with severe memory loss and moderate functional impairment (CDR stage 2.0) may not realize they haven’t bathed for a week or have worn the same clothes repeatedly. They are having trouble with basic activities of daily living, such as bathing and toilet hygiene. However, if you were to encounter them socially and didn’t talk to them for too long, you might think they were normal.

Those with severe memory loss and no significant function outside the house (CDR stage 3.0) are the most severely disabled. Dementia in these individuals is recognizable at a glance, from across the room.

Alzheimer patients progress through the stages, from CDR stage 0.5 at about 1 year to stage 1 by about 2 years, to stage 2 by 5 years, and to stage 3 at 8 or 9 years.²⁰

In prescribing antidementia medications. The CDR can help with prescribing antidementia drugs. No medications are approved by the FDA for stage 0 or 0.5. Cholinesterase inhibitors are approved for stages 1, 2, and 3; memantine (Namenda) is approved for stages 2 and 3.

Advising about driving. The CDR is the only risk predictor with a quality-of-evidence rating of A. More than half of people with stage 0.5 memory impairment are safe drivers; fewer than half of those with stage 1.0 are still safe drivers; and patients with stage 2.0 dementia should not be driving at all.²¹ An adverse rating by a caregiver carries a quality-of-evidence rating of B. Predictors of driving risk with a quality-of-evidence rating of C are decreased mileage due to self-restriction, agitation, or aggression; a crash in the past 1 to 5 years; a citation in the past 2 to 3 years; and a Folstein Mini-Mental State Examination score of 24 or less. Studies also show that a memory-impaired person’s self-rating of safe driving ability or of assurance that he or she avoids unsafe situations is not reliable.²¹

DELIRIUM

Delirium goes by a number of synonyms, eg, “sundowning,” acute confusional state, acute change in mental status, metabolic encephalopathy, toxic encephalopathy (psychosis), acute brain syndrome, and acute toxic psychosis.

Delirium is common in hospitalized elderly patients, occurring in 11% to 42% of elderly hospitalized patients overall, up to 53% of elderly surgical patients on regular hospital floors, 80% of elderly surgical patients in intensive care, and about half of elderly patients after undergoing coronary artery bypass grafting. Unfortunately, it is undiagnosed in 30% to 60% of cases.^22–24

Many pathways can lead to delirium, including hypoxemia, metabolic derangement, drug effects, systemic inflammation, and infection.²⁵

Outcomes can vary from full recovery to death. After 1 year, 50% of those who leave the hospital with some evidence of delirium have not regained their baseline function. Delirium also increases the cost of care and the risk of institutionalization.

Delirium can accelerate dementia

FONG TG, JONES RN, SHI P, ET AL. DELIRIUM ACCELERATES COGNITIVE DECLINE IN ALZHEIMER DISEASE. NEUROLOGY 2009; 72:1570–1575.

Delirium accelerates the course of dementia in patients who had some evidence of dementia before they entered the hospital. Often, the change is noticeable by the family.²⁶

Preventing delirium

INOUYE SK BOGARDUS ST JR, CHARPENTIER PA, ET AL. A MULTICOMPONENT INTERVENTION TO PREVENT DELIRIUM IN HOSPITALIZED OLDER PATIENTS. N ENGL J MED 1999; 340:669–676.

LUNDSTRÖM M, OLOFSSON B, STENVALL M, ET AL. POSTOPERATIVE DELIRIUM IN OLD PATIENTS WITH FEMORAL NECK FRACTURE: A RANDOMIZED INTERVENTION STUDY. AGING CLIN EXP RES 2007; 19:178–186.

Delirium can often be prevented. In a report published in 1999, Inouye et al²⁷ described the outcomes of a program to prevent delirium in hospitalized medically ill elderly patients. Interventions were aimed at optimizing cognitive function, preventing sleep deprivation, avoiding immobility, improving vision and hearing, and treating dehydration. The incidence of delirium was 9.9% in the intervention group vs 15% in the control group, a 40% reduction (P < .05).

Lundström et al²⁸ implemented a similar program for elderly patients with hip fractures. Interventions included staff education and teamwork; active prevention, detection, and treatment of delirium; transfusions if hemoglobin levels were less than 10 g/dL; prompt removal of indwelling urinary catheters, with screening for urinary retention; active prevention and treatment of constipation; and protein-enriched meals. The incidence of delirium was 55% in the intervention group vs 75% in the control group, a 27% reduction.

Comment. Although we have long known that the risk of delirium in medical and surgical patients can be reduced, most hospitals do not have systematic programs to detect delirium and reduce its incidence. Hopefully, reduction in delirium risk will also reduce its adverse consequences, including worsening of dementia and increased mortality.

New clinical trials and observational studies are shedding light on ways to improve the health of elderly patients. Here is a brief summary of these trials and how they might influence your clinical practice.

EXERCISE HAS NEWLY DISCOVERED BENEFITS

According to government data,¹ exercise has a dose-dependent effect on rates of all-cause mortality: the more hours one exercises per week, the lower the risk of death. The difference in risk is most pronounced as one goes from no exercise to about 3 hours of exercise per week; above 3 hours per week, the curve flattens out but continues to decline. Hence, we advise patients to engage in about 30 minutes of moderate-intensity exercise every day.

Lately, physical exercise has been found to have other, unexpected benefits.

Exercise helps cognition

ERICKSON KI, PRAKASH RS, VOSS MW, ET AL. AEROBIC FITNESS IS ASSOCIATED WITH HIPPOCAMPAL VOLUME IN ELDERLY HUMANS. HIPPOCAMPUS 2009; 19:1030–1039.

ETGEN T, SANDER D, HUNTGEBURTH U, POPPERT H, FÖRSTL H, BICKEL H. PHYSICAL ACTIVITY AND INCIDENT COGNITIVE IMPAIRMENT IN ELDERLY PERSONS: THE INVADE STUDY. ARCH INTERN MED 2010; 170:186–193.

The hippocampus is a structure deep in the brain that is involved in short-term memory. It atrophies with age, more so with dementia. Erickson² found a correlation between aerobic fitness (as measured by maximum oxygen consumption), hippocampal volume, and spatial memory performance.

Etgen and colleagues³ studied nearly 4,000 older adults in Bavaria for 2 years. Among those reporting no physical activity, 21.4% had cognitive impairment at baseline, compared with 7.3% of those with high activity at baseline. Following those without cognitive impairment over a 2-year period, they found the incidence of new cognitive impairment was 13.9% in those with no physical activity at baseline, 6.7% in those with moderate activity, and 5.1% in those with high activity.

Exercise boosts the effect of influenza vaccine

WOODS JA, KEYLOCK KT, LOWDER T, ET AL. CARDIOVASCULAR EXERCISE TRAINING EXTENDS INFLUENZA VACCINE SEROPROTECTION IN SEDENTARY OLDER ADULTS: THE IMMUNE FUNCTION INTERVENTION TRIAL. J AM GERIATR SOC 2009; 57:2183–2191.

In a study in 144 sedentary but healthy older adults (ages 60 to 83), Woods et al⁴ randomized the participants to undergo either flexibility or cardiovascular training for 10 months, starting 4 months before their annual influenza shot. Exercise extended the duration of antibody protection, with more participants in the cardiovascular group than in the flexibility group showing protection at 24 weeks against all three strains covered by the vaccine: H1N1, H3N2, and influenza B.

PREVENTING FRACTURES

Each year, about 30% of people age 65 or older fall, sustaining serious injuries in 5% to 10% of cases. Unintentional falls are the main cause of hip fractures, which number 300,000 per year. They are also a common cause of death.

Vitamin D prevents fractures, but can there be too much of a good thing?

BISCHOFF-FERRARI HA, WILLETT WC, WONG JB, ET AL. PREVENTION OF NONVERTEBRAL FRACTURES WITH ORAL VITAMIN D AND DOSE DEPENDENCY: A META-ANALYSIS OF RANDOMIZED CONTROLLED TRIALS. ARCH INTERN MED 2009; 169:551–561.

SANDERS KM, STUART AL, WILLIAMSON EJ, ET AL. ANNUAL HIGH-DOSE ORAL VITAMIN D AND FALLS AND FRACTURES IN OLDER WOMEN: A RANDOMIZED CONTROLLED TRIAL. JAMA 2010; 303:1815–1822.

Bischoff-Ferrari⁵ performed a meta-analysis of 12 randomized controlled trials of oral supplemental vitamin D₃ for preventing nonvertebral fractures in people age 65 and older, and eight trials for preventing hip fractures in the same age group. They found that the higher the daily dose of vitamin D, the lower the relative risk of hip fracture. The threshold dose at which supplementation significantly reduced the risk of falling was about 400 units per day. Higher doses of vitamin D reduced both falls and hip fractures by about 20%. The maximal effect was seen with studies using the maximum daily doses, ie, 770 to 800 units per day—not megadoses, but more than most Americans are taking. The threshold serum level of vitamin D of significance was 60 nmol/L (24 ng/mL).

Of interest, the effect on fractures was independent of calcium supplementation. This is important because calcium supplementation over and above ordinary dietary intake may increase the risk of cardiovascular events.^6,7

Despite the benefits of vitamin D, too much may be too much of a good thing. Sanders et al⁸ performed a double-blind, placebo-controlled trial in 2,256 community-dwelling women, age 70 or older, who were considered to be at high risk for fractures. Half received a large oral dose (500,000 units) once a year for 3 to 5 years, and half got placebo. Their initial serum vitamin D level was 49 nmol/L; the level 30 days after a dose in the treatment group was 120 nmol/L.

Contrary to expectations, the incidence of falls was 15% higher in the vitamin D group than in the placebo group (P = .03), and the incidence of fractures was 26% higher (P = .047). The falls and fractures tended to cluster in the first 3 months after the dose in the active treatment group, when serum vitamin D levels were highest.

Comments. Unless future studies suggest a benefit to megadoses of vitamin D or prove calcium supplementation greater than 1,000 mg is safe, the optimal daily intake of vitamin D is likely 1,000 units, with approximately 200 units from diet and 800 units from supplements. A diet rich in low-fat dairy products may not require calcium supplementation. In those consuming a low-calcium diet, supplements of 500 to 1,000 mg/day are likely adequate.

Denosumab, a new drug for preventing fractures

CUMMINGS SR, SAN MARTIN J, MCCLUNG MR, ET AL; FREEDOM TRIAL. DENOSUMAB FOR PREVENTION OF FRACTURES IN POSTMENOPAUSAL WOMEN WITH OSTEOPOROSIS. N ENGL J MED 2009; 361:756–765.

SMITH MR, EGERDIE B, HERNÁNDEZ TORIZ N, ET AL; DENOSUMAB HALT PROSTATE CANCER STUDY GROUP. DENOSUMAB IN MEN RECEIVING ANDROGEN-DEPRIVATION THERAPY FOR PROSTATE CANCER. N ENGL J MED 2009; 361:745–755.

Denosumab (Prolia) is the first of a new class of drugs for the treatment of osteoporosis. It is a monoclonal antibody and member of the tumor necrosis factor superfamily that binds to the receptor activator nuclear factor kappa B (RANK) ligand. It has an antiresorptive effect, preventing osteoclast differentiation and activation. It is given by subcutaneous injection of 60 mg every 6 months; it is cleared by a nonrenal mechanism.

In a randomized controlled trial in 7,868 women between the ages of 60 and 90 who had osteoporosis, Cummings et al⁹ reported that denosumab reduced the 3-year incidence of vertebral fractures by 68% (P < .001), reduced the incidence of hip fractures by 40% (P = .01), and reduced the incidence of nonvertebral fractures by 20% (P = .01). In a trial in men receiving androgen deprivation therapy for prostate cancer, Smith et al¹⁰ reported that denosumab reduced the incidence of vertebral fracture by 62% (P = .006).

Comment. Denosumab was approved by the US Food and Drug Administration (FDA) on June 1, 2010, and is emerging in specialty clinics at the time of this publication. Its potential impact on clinical care is not yet known. It is costly—about $825 (average wholesale price) per injection—but since it is given by injection it may be easier than a yearly infusion of zoledronic acid (Reclast). It has the potential to suppress immune function, although this was not reported in the clinical trials. It may ultimately have a role in treating osteoporosis in men and women, prostate cancer following androgen deprivation, metastatic prostate cancer, metastatic breast cancer, osteoporosis with renal impairment, and other diseases.

DIALYSIS IN THE ELDERLY: A BLEAK STORY

KURELLA TAMURA M, COVINSKY KE, CHERTOW GM, YAFFE K, LANDEFELD CS, MCCOLLOCH CE. FUNCTIONAL STATUS OF ELDERLY ADULTS BEFORE AND AFTER INITIATION OF DIALYSIS. N ENGL J MED 2009; 361:1539–1547.

JASSAL SV, CHIU E, HLADUNEWITH M. LOSS OF INDEPENDENCE IN PATIENTS STARTING DIALYSIS AT 80 YEARS OF AGE OR OLDER (LETTER). N ENGL J MED 2009; 361:1612–1613.

Nursing home residents account for 4% of all patients in end-stage renal disease. However, the benefits of dialysis in older patients are uncertain. The mortality rate during the first year of dialysis is 35% in patients 70 years of age and older and 50% in patients 80 years and older.

Is dialysis helpful in the elderly, ie, does it improve survival and function?

Kurella Tamura et al¹¹ retrospectively identified 3,702 nursing home residents starting dialysis in whom functional assessments had been done. The numbers told a bleak story. Initiation of dialysis was associated with a sharp decline in functional status, as reflected in an increase of 2.8 points on the 28-point Minimum Data Set–Activities of Daily Living (MDS-ADL) scale (the higher the score, the worse the function). MDS-ADL scores stabilized at a plateau for about 6 months and then continued to decline. Moreover, at 12 months, 58% of the patients had died.

The MDS-ADL score is based on seven components: eating, bed mobility, locomotion, transferring, toileting, hygiene, and dressing; function declined in all of these areas when patients started dialysis.

Patients were more likely to decline in activities of daily living after starting dialysis if they were older, were white, had cerebrovascular disease, had a diagnosis of dementia, were hospitalized at the start of dialysis, or had a serum albumin level lower than 3.5 g/dL.

The same thing happens to elders living in the community when they start dialysis. Jassal and colleagues¹² reported that, of 97 community-dwelling patients (mean age 85), 46 (47%) were dead 2 years after starting dialysis. Although 76 (78%) had been living independently at the start of dialysis, only 11 (11%) were still doing so at 2 years.

Comment. These findings indicate that we do not know if hemodialysis improves survival. Hemodialysis may buy about 3 months of stable function, but it clearly does not restore function.

Is this the best we can do? Standard hemodialysis may have flaws, and nocturnal dialysis and peritoneal dialysis are used more in other countries. These dialysis techniques require more study in our older population. The lesson from these two publications on dialysis is that we should attend more carefully to slowing the decline in renal function before patients reach end-stage renal disease.

DABIGATRAN: AN ALTERNATIVE TO WARFARIN FOR ATRIAL FIBRILLATION

CONNOLLY SJ, EZEKOWITZ MD, YUSUF S, ET AL; RE-LY STEERING COMMITTEE AND INVESTIGATORS. DABIGATRAN VERSUS WARFARIN IN PATIENTS WITH ATRIAL FIBRILLATION. N ENGL J MED 2009; 361:1139–1151.

Atrial fibrillation is common, affecting 2.2 million adults. The median age of people who have atrial fibrillation is 75 years, and it is the most common arrhythmia in the elderly. Some 20% of ischemic strokes are attributed to it.^13–15

Warfarin (Coumadin) is still the mainstay of treatment to prevent stroke in patients with atrial fibrillation. In an analysis of pooled data from five clinical trials,¹⁶ the relative risk reduction with warfarin was about 68% in the overall population (number needed to treat 32), 51% in people older than 75 years with no other risk factors (number needed to treat 56), and 85% in people older than 75 years with one or more risk factors (number needed to treat 15).

But warfarin carries a risk of bleeding, and its dose must be periodically adjusted on the basis of the international normalized ratio (INR) of the prothrombin time, so it carries a burden of laboratory monitoring. It is less safe in people who eat erratically, resulting in wide fluctuations in the INR.

Dabigatran (Pradaxa), a direct thrombin inhibitor, is expected to become an alternative to warfarin. It has been approved in Europe but not yet in the United States.

Connolly et al,¹⁷ in a randomized, double-blind trial, assigned 18,113 patients who had atrial fibrillation to receive either dabigatran 110 or 150 mg twice daily or adjusted-dose warfarin in an unblinded fashion. At 2 years, the rates of stroke and systemic embolism were about the same with dabigatran 110 mg as with warfarin but were lower with dabigatran 150 mg (relative risk 0.66, 95% confidence interval [CI] 0.53–0.82, P < .001). The rate of major bleeding was lower with dabigatran 110 mg than with warfarin (2.71% per year vs 3.36% per year, P = .003), but it was similar with dabigatran 150 mg (3.11% per year). Rates of life-threatening bleeding were 1.80% with warfarin, 1.22% with dabigatran 110 mg (P < .05), and 1.45% with dabigatran 150 mg (P < .05).

Comment. I suspect that warfarin’s days are numbered. Dabigatran 110 or 150 mg was as safe and as effective as warfarin in clinical trials, and probably will be more effective than warfarin in clinical practice. It will also probably be safer than warfarin in clinical practice, particularly in challenging settings such as long-term care. On the other hand, it will likely be much more expensive than warfarin.

DEMENTIA

Adverse effects of cholinesterase inhibitors

GILL SS, ANDERSON GM, FISCHER HD, ET AL. SYNCOPE AND ITS CONSEQUENCES IN PATIENTS WITH DEMENTIA RECEIVING CHOLINESTERASE INHIBITORS: A POPULATION-BASED COHORT STUDY. ARCH INTERN MED 2009; 169:867–873.

Cholinesterase inhibitors, eg, donepezil (Aricept), galantamine (Razadyne), and rivastigmine (Exelon), are commonly used to treat Alzheimer disease. However, these drugs carry risks of serious adverse effects.

Gill et al¹⁸ retrospectively reviewed a database from Ontario, Canada, and identified about 20,000 community-dwelling elderly persons admitted to the hospital who had been prescribed cholinesterase inhibitors and about three times as many matched controls.

Several adverse events were more frequent in people receiving cholinesterase inhibitors. Findings (events per 1,000 person-years):

• Hospital visits for syncope: 31.5 vs 18.6, adjusted hazard ratio (HR) 1.76, 95% CI 1.57–1.98
• Hip fractures: 22.4 vs 19.8, HR 1.18, 85% CI 1.04–1.34
• Hospital visits for bradycardia: 6.9 vs 4.4, HR 1.69, 95% CI 1.32–2.15
• Permanent pacemaker insertion: 4.7 vs 3.3, HR 1.49, 95% CI 1.12–2.00.

Comment. This study adds to the concerns that cholinesterase inhibitors, which have only modest cognitive benefits, may increase the risk of falls, injury, and need for pacemaker placement in demented patients. A low threshold to stop medications in this class should be considered when a patient on a cholinesterase inhibitor presents with bradycardia, falls, and syncope.

IVERSON DJ, GRONSETH GS, REGER MA, ET AL; STANDARDS SUBCOMMITTEE OF THE AMERICAN ACADEMY OF NEUROLOGY. PRACTICE PARAMETER UPDATE: EVALUATION AND MANAGEMENT OF DRIVING RISK IN DEMENTIA: REPORT OF THE QUALITY STANDARDS SUBCOMMITTEE OF THE AMERICAN ACADEMY OF NEUROLOGY. NEUROLOGY 2010; 74:1316–1324.

The Clinical Dementia Rating (CDR) is a simple scale that should be applied by clinicians to describe stage of dementia in patients with Alzheimer disease. This scale can be useful in a variety of settings, from prescribing antidementia drugs to determining whether a patient should still drive. Although research protocols utilize a survey or semistructured interview to derive the stage, the clinician can estimate the stage easily in the office, particularly if there is an informant who can comment on performance outside the office.

There are four stages to the CDR¹⁹:

• 0: No dementia
• 0.5: Mild memory deficit but intact function
• 1.0: Moderate memory loss with mild functional impairment
• 2.0: Severe memory loss, moderate functional impairment
• 3.0: Severe memory loss, no significant function outside of the house.

Comment. The first stage (0.5, mild memory deficit but intact function) corresponds to “mild cognitive impairment.” In the clinic, these patients tend to take more notes. They come to the appointment with a little book and they write everything down so they don’t forget. They do arrive at their appointments on time; they are not crashing the car; they are paying their bills.

Patients with CDR stage 1.0 dementia (moderate memory loss with mild functional impairment) may miss appointments, they may confuse their medications, and they may have problems driving. They are still taking care of their basic needs, and they show up for appointments acceptably washed and dressed. However, they are likely having trouble shopping and managing their finances.

Patients with severe memory loss and moderate functional impairment (CDR stage 2.0) may not realize they haven’t bathed for a week or have worn the same clothes repeatedly. They are having trouble with basic activities of daily living, such as bathing and toilet hygiene. However, if you were to encounter them socially and didn’t talk to them for too long, you might think they were normal.

Those with severe memory loss and no significant function outside the house (CDR stage 3.0) are the most severely disabled. Dementia in these individuals is recognizable at a glance, from across the room.

Alzheimer patients progress through the stages, from CDR stage 0.5 at about 1 year to stage 1 by about 2 years, to stage 2 by 5 years, and to stage 3 at 8 or 9 years.²⁰

In prescribing antidementia medications. The CDR can help with prescribing antidementia drugs. No medications are approved by the FDA for stage 0 or 0.5. Cholinesterase inhibitors are approved for stages 1, 2, and 3; memantine (Namenda) is approved for stages 2 and 3.

Advising about driving. The CDR is the only risk predictor with a quality-of-evidence rating of A. More than half of people with stage 0.5 memory impairment are safe drivers; fewer than half of those with stage 1.0 are still safe drivers; and patients with stage 2.0 dementia should not be driving at all.²¹ An adverse rating by a caregiver carries a quality-of-evidence rating of B. Predictors of driving risk with a quality-of-evidence rating of C are decreased mileage due to self-restriction, agitation, or aggression; a crash in the past 1 to 5 years; a citation in the past 2 to 3 years; and a Folstein Mini-Mental State Examination score of 24 or less. Studies also show that a memory-impaired person’s self-rating of safe driving ability or of assurance that he or she avoids unsafe situations is not reliable.²¹

DELIRIUM

Delirium goes by a number of synonyms, eg, “sundowning,” acute confusional state, acute change in mental status, metabolic encephalopathy, toxic encephalopathy (psychosis), acute brain syndrome, and acute toxic psychosis.

Delirium is common in hospitalized elderly patients, occurring in 11% to 42% of elderly hospitalized patients overall, up to 53% of elderly surgical patients on regular hospital floors, 80% of elderly surgical patients in intensive care, and about half of elderly patients after undergoing coronary artery bypass grafting. Unfortunately, it is undiagnosed in 30% to 60% of cases.^22–24

Many pathways can lead to delirium, including hypoxemia, metabolic derangement, drug effects, systemic inflammation, and infection.²⁵

Outcomes can vary from full recovery to death. After 1 year, 50% of those who leave the hospital with some evidence of delirium have not regained their baseline function. Delirium also increases the cost of care and the risk of institutionalization.

Delirium can accelerate dementia

FONG TG, JONES RN, SHI P, ET AL. DELIRIUM ACCELERATES COGNITIVE DECLINE IN ALZHEIMER DISEASE. NEUROLOGY 2009; 72:1570–1575.

Delirium accelerates the course of dementia in patients who had some evidence of dementia before they entered the hospital. Often, the change is noticeable by the family.²⁶

Preventing delirium

INOUYE SK BOGARDUS ST JR, CHARPENTIER PA, ET AL. A MULTICOMPONENT INTERVENTION TO PREVENT DELIRIUM IN HOSPITALIZED OLDER PATIENTS. N ENGL J MED 1999; 340:669–676.

LUNDSTRÖM M, OLOFSSON B, STENVALL M, ET AL. POSTOPERATIVE DELIRIUM IN OLD PATIENTS WITH FEMORAL NECK FRACTURE: A RANDOMIZED INTERVENTION STUDY. AGING CLIN EXP RES 2007; 19:178–186.

Delirium can often be prevented. In a report published in 1999, Inouye et al²⁷ described the outcomes of a program to prevent delirium in hospitalized medically ill elderly patients. Interventions were aimed at optimizing cognitive function, preventing sleep deprivation, avoiding immobility, improving vision and hearing, and treating dehydration. The incidence of delirium was 9.9% in the intervention group vs 15% in the control group, a 40% reduction (P < .05).

Lundström et al²⁸ implemented a similar program for elderly patients with hip fractures. Interventions included staff education and teamwork; active prevention, detection, and treatment of delirium; transfusions if hemoglobin levels were less than 10 g/dL; prompt removal of indwelling urinary catheters, with screening for urinary retention; active prevention and treatment of constipation; and protein-enriched meals. The incidence of delirium was 55% in the intervention group vs 75% in the control group, a 27% reduction.

Comment. Although we have long known that the risk of delirium in medical and surgical patients can be reduced, most hospitals do not have systematic programs to detect delirium and reduce its incidence. Hopefully, reduction in delirium risk will also reduce its adverse consequences, including worsening of dementia and increased mortality.

Efforts to modify the relentless course of Alzheimer disease have until now been based on altering the production or clearance of beta-amyloid, the protein found in plaques in the brains of patients with the disease. Results have been disappointing, possibly because our models of the disease—mostly based on the rare, inherited form—may not be applicable to the much more common sporadic form.

Ely Lilly’s recent announcement that it is halting research into semagacestat, a drug designed to reduce amyloid production, only cast further doubt on viability of the amyloid hypothesis as a framework for effective treatments for Alzheimer disease.

Because of the close association of sporadic Alzheimer disease with vascular disease and type 2 diabetes mellitus, increased efforts to treat and prevent these conditions may be the best approach to reducing the incidence of Alzheimer disease.

This article will discuss current thinking of the pathophysiology of Alzheimer disease, with special attention to potential prevention and treatment strategies.

THE CANONICAL VIEW: AMYLOID IS THE CAUSE

The canonical view is that the toxic effects of beta-amyloid are the cause of neuronal dysfunction and loss in Alzheimer disease.

Beta-amyloid is a small peptide, 38 to 42 amino acids long, that accumulates in the extracellular plaque that characterizes Alzheimer pathology. Small amounts of extracellular beta-amyloid can be detected in the brains of elderly people who die of other causes, but the brains of people who die with severe Alzheimer disease show extensive accumulation of plaques.

The amyloid precursor protein is cleaved by normal constitutive enzymes, leaving beta-amyloid as a fragment. The beta-amyloid forms into fibrillar aggregations, which further clump into the extracellular plaque. Plaques can occur in the normal aging process in relatively low amounts. However, in Alzheimer disease, through some unknown trigger, the immune system appears to become activated in reference to the plaque. Microglial cells—the brain’s macrophages—invade the plaque and trigger a cycle of inflammation. The inflammation and its by-products cause local neuronal damage, which seems to propagate the inflammatory cycle to an even greater extent through a feed-forward loop. The damage leads to metabolic stress in the neuron and collapse of the cytoskeleton into a neurofibrillary tangle. Once the neurofibrillary tangle is forming, the neuron is probably on the path to certain death.

This pathway might be interrupted at several points, and in fact, much of the drug development world is working on possible ways to do so.

GENETIC VS SPORADIC DISEASE: WHAT ARE THE KEY DIFFERENCES?

Although the autosomal dominant form of the disease accounts for probably only 1% or 2% of all cases of Alzheimer disease, most animal models and hence much of the basic research and drug testing in Alzheimer disease are based on those dominant mutations. The pathology—the plaques and tangles—in Alzheimer disease in older adults is identical to that in younger adults, but the origins of the disease may not be the same. Therefore, the experimental model for one may not be relevant to the other.

In the last several years, some have questioned whether the amyloid hypothesis applies to all Alzheimer disease.^1,2 Arguments go back to at least 2002, when Bishop and Robinson in an article entitled “The amyloid hypothesis: Let sleeping dogmas lie?”³ criticized the hypothesis and suggested that the beta-amyloid peptide appeared to be neuroprotective, not neurotoxic, in most situations. They suggested we await the outcome of antiamyloid therapeutic trials to determine whether the amyloid hypothesis truly explains the disorder.

The antiamyloid trials have now been under way for some time, and we have no definitive answer. Data from the phase II study of the monoclonal antibody agent bapineuzumab suggests there might be some small clinical impact of removing amyloid from the brain through immunotherapy mechanisms, but the benefits thus far are not robust.

COULD AMYLOID BE NEUROPROTECTIVE?

A pivotal question might be, “What if sick neurons made amyloid, instead of amyloid making neurons sick?” A corollary question is, “What if the effect were bidirectional?”

It is possible that in certain concentrations amyloid is neurotoxic, but in other concentrations, it actually facilitates neuronal repair, healing, and connection.

REDUCING METABOLIC STRESS: THE KEY TO PREVENTION?

If our current models of drug therapy are not effective against sporadic Alzheimer disease, perhaps focusing on prevention would be more fruitful.

Consider diabetes mellitus as an analogy. Its manifestations include polydipsia, polyuria, fatigue, and elevated glucose and hemoglobin A^1c. Its complications are cardiovascular disease, nephropathy, and retinopathy. Yet diabetes mellitus encompasses two different diseases—type 1 and type 2—with different underlying pathophysiology. We do not treat them the same way. We may be moving toward a similar view of Alzheimer disease.

Links have been hypothesized between vascular risks and dementia. Diabetes, hypertension, dyslipidemia, and obesity might lead to dementia in a process abetted by oxidative stress, endothelial dysfunction, insulin resistance, inflammation, adiposity, and subcortical vascular disease. All of these could be targets of intervention to prevent and treat dementia.⁴

Instead of a beta-amyloid trigger, let us hypothesize that metabolic stress is the initiating element of the Alzheimer cascade, which then triggers beta-amyloid overproduction or underclearance, and the immune activation damages neurons. By lessening metabolic stress or by preventing immune activation, it may, in theory, be possible to prevent neurons from entering into the terminal pathway of tangle formation and cell death.

LINKS BETWEEN ALZHEIMER DISEASE AND DIABETES

Rates of dementia of all causes are higher in people with diabetes. The strongest effect has been noted in vascular dementia, but Alzheimer disease was also found to be associated with diabetes.⁵ The Framingham Heart Study⁶ found the association between dementia and diabetes was significant only when other risk factors for Alzheimer disease were minimal: in an otherwise healthy population, diabetes alone appears to trigger the risk for dementia. But in a population with a lot of vascular comorbidities, the association between diabetes and dementia is not as clear. Perhaps the magnitude of the risk is overwhelmed by greater cerebrovascular and cardiovascular morbidity.

A systematic review⁷ supported the notion that the risk of dementia is higher in people with diabetes, and even raised the issue of whether we should consider Alzheimer disease “type 3 diabetes.”

Testing of the reverse hypothesis—diabetes is more common in people with Alzheimer disease—also is supportive: diabetes mellitus and even impaired fasting glucose are approximately twice as common in people with Alzheimer disease than in those without.⁸ Fasting blood glucose levels increase steadily with age, but after age 65, they are higher in people with Alzheimer disease than in those without.

Glucose has some direct effects on brain metabolism that might explain the higher risk. Chronic hyperglycemia is associated with excessive production of free radicals, which leads to reactive oxygen species. These are toxic to neuronal membranes as well as to mitochondria, where many of the reactive oxygen species are generated. Free radicals also facilitate the inflammatory response.

We also see greater neuronal and mitochondrial calcium influx in the presence of hyperglycemia. The excess calcium interferes with mitochondrial metabolism and may trigger the cascade of apoptosis when it reaches critical levels in neurons.

Chronic hyperglycemia is also associated with increased advanced glycation end-products. These are toxic molecules produced by the persistent exposure of proteins to high sugar levels and may be facilitated by the presence of reactive oxygen species that catalyze the reactions between the sugars and the peptides. Glycation end-products are commonly recognized as the same as those occurring during browning of meat (the Maillard reaction).

Hyperglycemia also potentiates neuronal damage from ischemia. Animal experiments show that brain infarction in the presence of hyperglycemia results in worse damage than the same degree of ischemia in the absence of hyperglycemia. Hyperglycemia may exaggerate other blows to neuronal function such as those from small strokes or microvascular ischemia.

AN ALTERNATIVE TO THE AMYLOID HYPOTHESIS: THE ‘MITOCHONDRIAL CASCADE HYPOTHESIS’

Swerdlow and Khan⁹ have proposed an alternative to the amyloid hypothesis as the cause of Alzheimer disease, known as the “mitochondrial cascade hypothesis.” According to this model, as we age we accumulate more wear-and-tear from oxidative mitochondrial damage, especially the accumulation of toxins leading to reduced cell metabolic activity. This triggers the “3-R response”:

Reset. When toxins alter cell metabolism, neurons try to repair themselves by manufacturing beta-amyloid, which is a “repair-and-reset” synaptic signaling molecule that reduces energy production. Under the lower energy state, beta-pleated sheets develop from beta-amyloid, which aggregate and form amyloid plaque.

Remove. Many cells undergo programmed death when faced with oxidative stress. The first step in neuronal loss is reduced synaptic connections and, hence, losses in neuronal communication. This results in impaired cognition.

Replace. Some cells that are faced with metabolic stress re-enter the cell cycle by undergoing cell division. Neurons, however, are terminally postmitotic and die if they try to divide: by synthesizing cell division proteins, duplicating chromosomes, and reorganizing the complex internal structure, the cell cannot work properly and cell division fails. In the mitochondrial cascade hypothesis, neurofibrillary tangles result from this attempted remodeling of the cytoskeletal filaments, furthering neuronal dysfunction.

ALZHEIMER DISEASE AND STROKE: MORE ALIKE THAN WE THOUGHT?

Although historically clinicians and researchers have tried to distinguish between Alzheimer disease and vascular dementia, growing evidence indicates that the two disorders overlap significantly and that the pathologies may be synergistic.

Alzheimer disease has been hypothesized as being a vascular disorder.¹⁰ It shares many of the risk factors of vascular disease, and preclinical detection of Alzheimer disease is possible from measurements of regional cerebral perfusion. Cerebrovascular and neurodegenerative pathology are parallel in Alzheimer disease and vascular disease.

Pure Alzheimer disease and vascular disease are two ends of a pathologic continuum.¹¹ At one end is “pure” Alzheimer disease, in which patients die only with histologic findings of plaques and neurofibrillary tangles. This form may occur only in patients with the autosomal dominant early-onset form. At the other end of the spectrum are people who have serious vascular disease, multiple strokes, and microvascular ischemia and who die demented but with no evidence of the plaques and tangles of Alzheimer disease.

Between these poles is a spectrum of overlapping pathology that is either Alzheimer disease-dominant or vascular disease-dominant, with varying degrees of amyloid plaque and evidence of microvascular infarcts. Cerebral amyloid angiopathy (the accumulation of beta-amyloid in the wall of arteries in the brain) bridges the syndromes.¹² In some drug studies that attempted removing amyloid from the brain, vascular permeability was altered, resulting in brain edema.

Along the same lines as Kalaria’s model,¹¹ Snowden et al¹³ found at autopsy of aged Catholic nuns that for some the accumulation of Alzheimer pathology alone was insufficient to cause dementia, but dementia was nearly universal in nuns with the same burden of Alzheimer pathology commingled with vascular pathology.

DOES INFLAMMATION PLAY A ROLE?

The inflammatory state is a recognized risk factor for Alzheimer disease, but the clinical data are mixed. Epidemiologic evidence is strong: patients who regularly take nonsteroidal anti-inflammatory drugs (NSAIDs) or steroids for chronic, systemic inflammatory diseases (eg, arthritis) have a 45% to 60% reduced risk for Alzheimer disease.^14,15

However, multiple clinical trials in patients with Alzheimer disease have failed to show a benefit of taking anti-inflammatory drugs. One preliminary report suggested that indomethacin (Indocin) might offer benefit, but because of gastrointestinal side effects its usefulness in an elderly population is limited.

Diabetes and inflammation are also closely linked: hyperinsulinemia is proinflammatory, promoting the formation of reactive oxygen species, inhibiting the degradation of oxidized proteins, and increasing the risk for lipid per-oxidation. Insulin acts synergistically with endotoxins to raise inflammatory markers, eg, proinflammatory cytokines and C-reactive protein.¹⁶

It is possible that anti-inflammatory drugs may not work in Alzheimer disease because inflammation in the brain is mediated more by microglial cells than by prostaglandin pathways. In Alzheimer disease, inflammation is mediated by activated microglial cells, which invade plaques with their processes; these are not evident in the diffuse beta-amyloid-rich plaques seen in typical aging. The trigger for their activation is unclear, but the activated microglial cells and the invasion of plaques are seen in transgenic mouse models of Alzheimer disease, and activation is seen when beta-amyloid is injected into the brain of a healthy mouse.¹⁷

Activated microglial cells enlarge and their metabolic rate increases, with a surge in the production of proteins and other protein-mediated inflammatory markers such as alpha-antichymotrypsin, alpha-antitrypsin, serum amyloid P, C-reactive protein, nitric oxide, and proinflammatory cytokines. It is unlikely that it is healthy for cells to be exposed to these inflammatory products. Some of the cytokines are now targets of drug development for Alzheimer disease, and agents targeting these pathways have already been developed for connective tissue diseases.

In a controversial pilot study, Tobinick et al¹⁸ studied the use of etanercept (Enbrel), an inhibitor of tumor necrosis factor-alpha (an inflammatory cytokine). They injected etanercept weekly into the spinal canal in 15 patients with mild to severe Alzheimer disease, for 6 months. Patients improved in the Mini-Mental State Examination by more than two points during the study. Patent issues surrounding use of this drug in Alzheimer disease may delay further trials.

Thiazolidinediones block microglial cell activation

The reactive microglial phenotype can be prevented in cell culture by peroxisome proliferator-activated receptor (PPAR) gamma agonists. These include the antidiabetic thiazolidinediones such as pioglitazone (Actos), troglitazone (Rezulin), and rosiglitazone (Avandia), and indomethacin and other NSAIDs.

Using a Veterans Administration database of more than 142,000 patients, Miller et al¹⁹ retrospectively found that patients who took a thiazolidinedione for diabetes had a 20% lower risk of developing Alzheimer disease compared with users of insulin or metformin (Glucophage).

However, rosiglitazone showed no benefit against Alzheimer disease in a large clinical trial,²⁰ but this may be because it is rapidly cleared from the brain. Pioglitazone is not actively exported from the brain, so it may be a better candidate, but pharmaceutical industry interest in this agent is low because its patent will soon expire.

Fish oil is another PPAR-gamma agonist, and some studies indicate that eating fish may protect against developing Alzheimer disease; it may also be therapeutic if the disease is present. Double-blind controlled studies have not been carried out and likely will not because of patent issues: the costs of such studies are high, and the potential payback is low.

ESTROGEN: PROTECTIVE OR NOT?

Whether taking estrogen is a risk factor or is protective has not yet been determined. Estrogen directly affects neurons. It increases the number of dendritic spines, which are associated with improved memory. Meta-analyses suggest that hormone replacement therapy reduces the risk of dementia by about one-third. ^21,22 Both positive and negative prospective studies exist, but all are complicated by serious methodologic flaws.^23,24

Combined analysis of about 7,500 women from two double-blind, randomized, placebo-controlled trials of the Women’s Health Initiative Memory Study found that the risks of dementia and mild cognitive impairment were increased by hormone replacement therapy. The hazard ratio for dementia was found to be 1.76 (P < .005), amounting to 23 new cases of dementia per 10,000 prescriptions annually.²⁵

Patient selection may account for the conflicting results in different studies. Epidemiologic studies consisted mostly of newly postmenopausal women and those who were being treated for symptoms of vasomotor instability. In contrast, the Women’s Health Initiative enrolled only women older than 65 and excluded women with vasomotor instability. Other studies indicate that the greatest cognitive improvements with hormone therapies are seen in women with vasomotor symptoms.

WHICH RISK FACTORS CAN WE CONTROL?

In summary, some of the risk factors for Alzheimer disease can be modified if we do the following.

Aggressively manage diabetes and cardiovascular disease. Vascular risk factors significantly increase dementia risk, providing good targets for prevention: clinicians should aggressively help their patients control diabetes, hypertension, and hyperlipidemia.²⁶ However, aggressive control of hypertension in a patient with already-existing dementia may exacerbate the condition, so caution is warranted.

Optimize diet. Dietary measures include high intake of antioxidants (which are especially high in brightly colored and tart-flavored fruits and vegetables) and polyunsaturated fats.²⁶ Eating a Mediterranean-type diet that includes a high intake of cold-water ocean fish is recommended. Fish should not be fried: the high temperatures may destroy the omega-3 fatty acids, and the high fat content may inhibit their absorption.

Weigh the risks and benefits of estrogen. Although estrogen replacement therapy for postmenopausal women has had mixed results for controlling dementia, it appears to be clinically indicated to control vasomotor symptoms and likely does not increase the risk of dementia for newly menopausal women. Risks and benefits should be carefully weighed for each patient.

Optimize exercise. People who are physically active in midlife have a lower risk of Alzheimer disease.²⁷ Those who adopt new physical activity late in life may also gain some protective or restorative benefit.²⁸

Many measures, such as taking anti-inflammatory or antihypertensive drugs, probably have a very small incremental benefit over time, so it is difficult to measure significant effects during the course of a typical clinical trial.

Clinicians are already recommending actions to reduce the risk of dementia by focusing on lowering cardiovascular risk. Hopefully, as these actions become more commonly practiced as lifelong habits in those reaching the age of risk for Alzheimer disease, we will see a reduced incidence of that devastating and much-feared illness.

Efforts to modify the relentless course of Alzheimer disease have until now been based on altering the production or clearance of beta-amyloid, the protein found in plaques in the brains of patients with the disease. Results have been disappointing, possibly because our models of the disease—mostly based on the rare, inherited form—may not be applicable to the much more common sporadic form.

Ely Lilly’s recent announcement that it is halting research into semagacestat, a drug designed to reduce amyloid production, only cast further doubt on viability of the amyloid hypothesis as a framework for effective treatments for Alzheimer disease.

Because of the close association of sporadic Alzheimer disease with vascular disease and type 2 diabetes mellitus, increased efforts to treat and prevent these conditions may be the best approach to reducing the incidence of Alzheimer disease.

This article will discuss current thinking of the pathophysiology of Alzheimer disease, with special attention to potential prevention and treatment strategies.

THE CANONICAL VIEW: AMYLOID IS THE CAUSE

The canonical view is that the toxic effects of beta-amyloid are the cause of neuronal dysfunction and loss in Alzheimer disease.

Beta-amyloid is a small peptide, 38 to 42 amino acids long, that accumulates in the extracellular plaque that characterizes Alzheimer pathology. Small amounts of extracellular beta-amyloid can be detected in the brains of elderly people who die of other causes, but the brains of people who die with severe Alzheimer disease show extensive accumulation of plaques.

The amyloid precursor protein is cleaved by normal constitutive enzymes, leaving beta-amyloid as a fragment. The beta-amyloid forms into fibrillar aggregations, which further clump into the extracellular plaque. Plaques can occur in the normal aging process in relatively low amounts. However, in Alzheimer disease, through some unknown trigger, the immune system appears to become activated in reference to the plaque. Microglial cells—the brain’s macrophages—invade the plaque and trigger a cycle of inflammation. The inflammation and its by-products cause local neuronal damage, which seems to propagate the inflammatory cycle to an even greater extent through a feed-forward loop. The damage leads to metabolic stress in the neuron and collapse of the cytoskeleton into a neurofibrillary tangle. Once the neurofibrillary tangle is forming, the neuron is probably on the path to certain death.

This pathway might be interrupted at several points, and in fact, much of the drug development world is working on possible ways to do so.

GENETIC VS SPORADIC DISEASE: WHAT ARE THE KEY DIFFERENCES?

Although the autosomal dominant form of the disease accounts for probably only 1% or 2% of all cases of Alzheimer disease, most animal models and hence much of the basic research and drug testing in Alzheimer disease are based on those dominant mutations. The pathology—the plaques and tangles—in Alzheimer disease in older adults is identical to that in younger adults, but the origins of the disease may not be the same. Therefore, the experimental model for one may not be relevant to the other.

In the last several years, some have questioned whether the amyloid hypothesis applies to all Alzheimer disease.^1,2 Arguments go back to at least 2002, when Bishop and Robinson in an article entitled “The amyloid hypothesis: Let sleeping dogmas lie?”³ criticized the hypothesis and suggested that the beta-amyloid peptide appeared to be neuroprotective, not neurotoxic, in most situations. They suggested we await the outcome of antiamyloid therapeutic trials to determine whether the amyloid hypothesis truly explains the disorder.

The antiamyloid trials have now been under way for some time, and we have no definitive answer. Data from the phase II study of the monoclonal antibody agent bapineuzumab suggests there might be some small clinical impact of removing amyloid from the brain through immunotherapy mechanisms, but the benefits thus far are not robust.

COULD AMYLOID BE NEUROPROTECTIVE?

A pivotal question might be, “What if sick neurons made amyloid, instead of amyloid making neurons sick?” A corollary question is, “What if the effect were bidirectional?”

It is possible that in certain concentrations amyloid is neurotoxic, but in other concentrations, it actually facilitates neuronal repair, healing, and connection.

REDUCING METABOLIC STRESS: THE KEY TO PREVENTION?

If our current models of drug therapy are not effective against sporadic Alzheimer disease, perhaps focusing on prevention would be more fruitful.

Consider diabetes mellitus as an analogy. Its manifestations include polydipsia, polyuria, fatigue, and elevated glucose and hemoglobin A^1c. Its complications are cardiovascular disease, nephropathy, and retinopathy. Yet diabetes mellitus encompasses two different diseases—type 1 and type 2—with different underlying pathophysiology. We do not treat them the same way. We may be moving toward a similar view of Alzheimer disease.

Links have been hypothesized between vascular risks and dementia. Diabetes, hypertension, dyslipidemia, and obesity might lead to dementia in a process abetted by oxidative stress, endothelial dysfunction, insulin resistance, inflammation, adiposity, and subcortical vascular disease. All of these could be targets of intervention to prevent and treat dementia.⁴

Instead of a beta-amyloid trigger, let us hypothesize that metabolic stress is the initiating element of the Alzheimer cascade, which then triggers beta-amyloid overproduction or underclearance, and the immune activation damages neurons. By lessening metabolic stress or by preventing immune activation, it may, in theory, be possible to prevent neurons from entering into the terminal pathway of tangle formation and cell death.

LINKS BETWEEN ALZHEIMER DISEASE AND DIABETES

Rates of dementia of all causes are higher in people with diabetes. The strongest effect has been noted in vascular dementia, but Alzheimer disease was also found to be associated with diabetes.⁵ The Framingham Heart Study⁶ found the association between dementia and diabetes was significant only when other risk factors for Alzheimer disease were minimal: in an otherwise healthy population, diabetes alone appears to trigger the risk for dementia. But in a population with a lot of vascular comorbidities, the association between diabetes and dementia is not as clear. Perhaps the magnitude of the risk is overwhelmed by greater cerebrovascular and cardiovascular morbidity.

A systematic review⁷ supported the notion that the risk of dementia is higher in people with diabetes, and even raised the issue of whether we should consider Alzheimer disease “type 3 diabetes.”

Testing of the reverse hypothesis—diabetes is more common in people with Alzheimer disease—also is supportive: diabetes mellitus and even impaired fasting glucose are approximately twice as common in people with Alzheimer disease than in those without.⁸ Fasting blood glucose levels increase steadily with age, but after age 65, they are higher in people with Alzheimer disease than in those without.

Glucose has some direct effects on brain metabolism that might explain the higher risk. Chronic hyperglycemia is associated with excessive production of free radicals, which leads to reactive oxygen species. These are toxic to neuronal membranes as well as to mitochondria, where many of the reactive oxygen species are generated. Free radicals also facilitate the inflammatory response.

We also see greater neuronal and mitochondrial calcium influx in the presence of hyperglycemia. The excess calcium interferes with mitochondrial metabolism and may trigger the cascade of apoptosis when it reaches critical levels in neurons.

Chronic hyperglycemia is also associated with increased advanced glycation end-products. These are toxic molecules produced by the persistent exposure of proteins to high sugar levels and may be facilitated by the presence of reactive oxygen species that catalyze the reactions between the sugars and the peptides. Glycation end-products are commonly recognized as the same as those occurring during browning of meat (the Maillard reaction).

Hyperglycemia also potentiates neuronal damage from ischemia. Animal experiments show that brain infarction in the presence of hyperglycemia results in worse damage than the same degree of ischemia in the absence of hyperglycemia. Hyperglycemia may exaggerate other blows to neuronal function such as those from small strokes or microvascular ischemia.

AN ALTERNATIVE TO THE AMYLOID HYPOTHESIS: THE ‘MITOCHONDRIAL CASCADE HYPOTHESIS’

Swerdlow and Khan⁹ have proposed an alternative to the amyloid hypothesis as the cause of Alzheimer disease, known as the “mitochondrial cascade hypothesis.” According to this model, as we age we accumulate more wear-and-tear from oxidative mitochondrial damage, especially the accumulation of toxins leading to reduced cell metabolic activity. This triggers the “3-R response”:

Reset. When toxins alter cell metabolism, neurons try to repair themselves by manufacturing beta-amyloid, which is a “repair-and-reset” synaptic signaling molecule that reduces energy production. Under the lower energy state, beta-pleated sheets develop from beta-amyloid, which aggregate and form amyloid plaque.

Remove. Many cells undergo programmed death when faced with oxidative stress. The first step in neuronal loss is reduced synaptic connections and, hence, losses in neuronal communication. This results in impaired cognition.

Replace. Some cells that are faced with metabolic stress re-enter the cell cycle by undergoing cell division. Neurons, however, are terminally postmitotic and die if they try to divide: by synthesizing cell division proteins, duplicating chromosomes, and reorganizing the complex internal structure, the cell cannot work properly and cell division fails. In the mitochondrial cascade hypothesis, neurofibrillary tangles result from this attempted remodeling of the cytoskeletal filaments, furthering neuronal dysfunction.

ALZHEIMER DISEASE AND STROKE: MORE ALIKE THAN WE THOUGHT?

Although historically clinicians and researchers have tried to distinguish between Alzheimer disease and vascular dementia, growing evidence indicates that the two disorders overlap significantly and that the pathologies may be synergistic.

Alzheimer disease has been hypothesized as being a vascular disorder.¹⁰ It shares many of the risk factors of vascular disease, and preclinical detection of Alzheimer disease is possible from measurements of regional cerebral perfusion. Cerebrovascular and neurodegenerative pathology are parallel in Alzheimer disease and vascular disease.

Pure Alzheimer disease and vascular disease are two ends of a pathologic continuum.¹¹ At one end is “pure” Alzheimer disease, in which patients die only with histologic findings of plaques and neurofibrillary tangles. This form may occur only in patients with the autosomal dominant early-onset form. At the other end of the spectrum are people who have serious vascular disease, multiple strokes, and microvascular ischemia and who die demented but with no evidence of the plaques and tangles of Alzheimer disease.

Between these poles is a spectrum of overlapping pathology that is either Alzheimer disease-dominant or vascular disease-dominant, with varying degrees of amyloid plaque and evidence of microvascular infarcts. Cerebral amyloid angiopathy (the accumulation of beta-amyloid in the wall of arteries in the brain) bridges the syndromes.¹² In some drug studies that attempted removing amyloid from the brain, vascular permeability was altered, resulting in brain edema.

Along the same lines as Kalaria’s model,¹¹ Snowden et al¹³ found at autopsy of aged Catholic nuns that for some the accumulation of Alzheimer pathology alone was insufficient to cause dementia, but dementia was nearly universal in nuns with the same burden of Alzheimer pathology commingled with vascular pathology.

DOES INFLAMMATION PLAY A ROLE?

The inflammatory state is a recognized risk factor for Alzheimer disease, but the clinical data are mixed. Epidemiologic evidence is strong: patients who regularly take nonsteroidal anti-inflammatory drugs (NSAIDs) or steroids for chronic, systemic inflammatory diseases (eg, arthritis) have a 45% to 60% reduced risk for Alzheimer disease.^14,15

However, multiple clinical trials in patients with Alzheimer disease have failed to show a benefit of taking anti-inflammatory drugs. One preliminary report suggested that indomethacin (Indocin) might offer benefit, but because of gastrointestinal side effects its usefulness in an elderly population is limited.

Diabetes and inflammation are also closely linked: hyperinsulinemia is proinflammatory, promoting the formation of reactive oxygen species, inhibiting the degradation of oxidized proteins, and increasing the risk for lipid per-oxidation. Insulin acts synergistically with endotoxins to raise inflammatory markers, eg, proinflammatory cytokines and C-reactive protein.¹⁶

It is possible that anti-inflammatory drugs may not work in Alzheimer disease because inflammation in the brain is mediated more by microglial cells than by prostaglandin pathways. In Alzheimer disease, inflammation is mediated by activated microglial cells, which invade plaques with their processes; these are not evident in the diffuse beta-amyloid-rich plaques seen in typical aging. The trigger for their activation is unclear, but the activated microglial cells and the invasion of plaques are seen in transgenic mouse models of Alzheimer disease, and activation is seen when beta-amyloid is injected into the brain of a healthy mouse.¹⁷

Activated microglial cells enlarge and their metabolic rate increases, with a surge in the production of proteins and other protein-mediated inflammatory markers such as alpha-antichymotrypsin, alpha-antitrypsin, serum amyloid P, C-reactive protein, nitric oxide, and proinflammatory cytokines. It is unlikely that it is healthy for cells to be exposed to these inflammatory products. Some of the cytokines are now targets of drug development for Alzheimer disease, and agents targeting these pathways have already been developed for connective tissue diseases.

In a controversial pilot study, Tobinick et al¹⁸ studied the use of etanercept (Enbrel), an inhibitor of tumor necrosis factor-alpha (an inflammatory cytokine). They injected etanercept weekly into the spinal canal in 15 patients with mild to severe Alzheimer disease, for 6 months. Patients improved in the Mini-Mental State Examination by more than two points during the study. Patent issues surrounding use of this drug in Alzheimer disease may delay further trials.

Thiazolidinediones block microglial cell activation

The reactive microglial phenotype can be prevented in cell culture by peroxisome proliferator-activated receptor (PPAR) gamma agonists. These include the antidiabetic thiazolidinediones such as pioglitazone (Actos), troglitazone (Rezulin), and rosiglitazone (Avandia), and indomethacin and other NSAIDs.

Using a Veterans Administration database of more than 142,000 patients, Miller et al¹⁹ retrospectively found that patients who took a thiazolidinedione for diabetes had a 20% lower risk of developing Alzheimer disease compared with users of insulin or metformin (Glucophage).

However, rosiglitazone showed no benefit against Alzheimer disease in a large clinical trial,²⁰ but this may be because it is rapidly cleared from the brain. Pioglitazone is not actively exported from the brain, so it may be a better candidate, but pharmaceutical industry interest in this agent is low because its patent will soon expire.

Fish oil is another PPAR-gamma agonist, and some studies indicate that eating fish may protect against developing Alzheimer disease; it may also be therapeutic if the disease is present. Double-blind controlled studies have not been carried out and likely will not because of patent issues: the costs of such studies are high, and the potential payback is low.

ESTROGEN: PROTECTIVE OR NOT?

Whether taking estrogen is a risk factor or is protective has not yet been determined. Estrogen directly affects neurons. It increases the number of dendritic spines, which are associated with improved memory. Meta-analyses suggest that hormone replacement therapy reduces the risk of dementia by about one-third. ^21,22 Both positive and negative prospective studies exist, but all are complicated by serious methodologic flaws.^23,24

Combined analysis of about 7,500 women from two double-blind, randomized, placebo-controlled trials of the Women’s Health Initiative Memory Study found that the risks of dementia and mild cognitive impairment were increased by hormone replacement therapy. The hazard ratio for dementia was found to be 1.76 (P < .005), amounting to 23 new cases of dementia per 10,000 prescriptions annually.²⁵

Patient selection may account for the conflicting results in different studies. Epidemiologic studies consisted mostly of newly postmenopausal women and those who were being treated for symptoms of vasomotor instability. In contrast, the Women’s Health Initiative enrolled only women older than 65 and excluded women with vasomotor instability. Other studies indicate that the greatest cognitive improvements with hormone therapies are seen in women with vasomotor symptoms.

WHICH RISK FACTORS CAN WE CONTROL?

In summary, some of the risk factors for Alzheimer disease can be modified if we do the following.

Aggressively manage diabetes and cardiovascular disease. Vascular risk factors significantly increase dementia risk, providing good targets for prevention: clinicians should aggressively help their patients control diabetes, hypertension, and hyperlipidemia.²⁶ However, aggressive control of hypertension in a patient with already-existing dementia may exacerbate the condition, so caution is warranted.

Optimize diet. Dietary measures include high intake of antioxidants (which are especially high in brightly colored and tart-flavored fruits and vegetables) and polyunsaturated fats.²⁶ Eating a Mediterranean-type diet that includes a high intake of cold-water ocean fish is recommended. Fish should not be fried: the high temperatures may destroy the omega-3 fatty acids, and the high fat content may inhibit their absorption.

Weigh the risks and benefits of estrogen. Although estrogen replacement therapy for postmenopausal women has had mixed results for controlling dementia, it appears to be clinically indicated to control vasomotor symptoms and likely does not increase the risk of dementia for newly menopausal women. Risks and benefits should be carefully weighed for each patient.

Optimize exercise. People who are physically active in midlife have a lower risk of Alzheimer disease.²⁷ Those who adopt new physical activity late in life may also gain some protective or restorative benefit.²⁸

Many measures, such as taking anti-inflammatory or antihypertensive drugs, probably have a very small incremental benefit over time, so it is difficult to measure significant effects during the course of a typical clinical trial.

Clinicians are already recommending actions to reduce the risk of dementia by focusing on lowering cardiovascular risk. Hopefully, as these actions become more commonly practiced as lifelong habits in those reaching the age of risk for Alzheimer disease, we will see a reduced incidence of that devastating and much-feared illness.

Efforts to modify the relentless course of Alzheimer disease have until now been based on altering the production or clearance of beta-amyloid, the protein found in plaques in the brains of patients with the disease. Results have been disappointing, possibly because our models of the disease—mostly based on the rare, inherited form—may not be applicable to the much more common sporadic form.

Ely Lilly’s recent announcement that it is halting research into semagacestat, a drug designed to reduce amyloid production, only cast further doubt on viability of the amyloid hypothesis as a framework for effective treatments for Alzheimer disease.

Because of the close association of sporadic Alzheimer disease with vascular disease and type 2 diabetes mellitus, increased efforts to treat and prevent these conditions may be the best approach to reducing the incidence of Alzheimer disease.

This article will discuss current thinking of the pathophysiology of Alzheimer disease, with special attention to potential prevention and treatment strategies.

THE CANONICAL VIEW: AMYLOID IS THE CAUSE

The canonical view is that the toxic effects of beta-amyloid are the cause of neuronal dysfunction and loss in Alzheimer disease.

Beta-amyloid is a small peptide, 38 to 42 amino acids long, that accumulates in the extracellular plaque that characterizes Alzheimer pathology. Small amounts of extracellular beta-amyloid can be detected in the brains of elderly people who die of other causes, but the brains of people who die with severe Alzheimer disease show extensive accumulation of plaques.

The amyloid precursor protein is cleaved by normal constitutive enzymes, leaving beta-amyloid as a fragment. The beta-amyloid forms into fibrillar aggregations, which further clump into the extracellular plaque. Plaques can occur in the normal aging process in relatively low amounts. However, in Alzheimer disease, through some unknown trigger, the immune system appears to become activated in reference to the plaque. Microglial cells—the brain’s macrophages—invade the plaque and trigger a cycle of inflammation. The inflammation and its by-products cause local neuronal damage, which seems to propagate the inflammatory cycle to an even greater extent through a feed-forward loop. The damage leads to metabolic stress in the neuron and collapse of the cytoskeleton into a neurofibrillary tangle. Once the neurofibrillary tangle is forming, the neuron is probably on the path to certain death.

This pathway might be interrupted at several points, and in fact, much of the drug development world is working on possible ways to do so.

GENETIC VS SPORADIC DISEASE: WHAT ARE THE KEY DIFFERENCES?

Although the autosomal dominant form of the disease accounts for probably only 1% or 2% of all cases of Alzheimer disease, most animal models and hence much of the basic research and drug testing in Alzheimer disease are based on those dominant mutations. The pathology—the plaques and tangles—in Alzheimer disease in older adults is identical to that in younger adults, but the origins of the disease may not be the same. Therefore, the experimental model for one may not be relevant to the other.

In the last several years, some have questioned whether the amyloid hypothesis applies to all Alzheimer disease.^1,2 Arguments go back to at least 2002, when Bishop and Robinson in an article entitled “The amyloid hypothesis: Let sleeping dogmas lie?”³ criticized the hypothesis and suggested that the beta-amyloid peptide appeared to be neuroprotective, not neurotoxic, in most situations. They suggested we await the outcome of antiamyloid therapeutic trials to determine whether the amyloid hypothesis truly explains the disorder.

The antiamyloid trials have now been under way for some time, and we have no definitive answer. Data from the phase II study of the monoclonal antibody agent bapineuzumab suggests there might be some small clinical impact of removing amyloid from the brain through immunotherapy mechanisms, but the benefits thus far are not robust.

COULD AMYLOID BE NEUROPROTECTIVE?

A pivotal question might be, “What if sick neurons made amyloid, instead of amyloid making neurons sick?” A corollary question is, “What if the effect were bidirectional?”

It is possible that in certain concentrations amyloid is neurotoxic, but in other concentrations, it actually facilitates neuronal repair, healing, and connection.

REDUCING METABOLIC STRESS: THE KEY TO PREVENTION?

If our current models of drug therapy are not effective against sporadic Alzheimer disease, perhaps focusing on prevention would be more fruitful.

Consider diabetes mellitus as an analogy. Its manifestations include polydipsia, polyuria, fatigue, and elevated glucose and hemoglobin A^1c. Its complications are cardiovascular disease, nephropathy, and retinopathy. Yet diabetes mellitus encompasses two different diseases—type 1 and type 2—with different underlying pathophysiology. We do not treat them the same way. We may be moving toward a similar view of Alzheimer disease.

Links have been hypothesized between vascular risks and dementia. Diabetes, hypertension, dyslipidemia, and obesity might lead to dementia in a process abetted by oxidative stress, endothelial dysfunction, insulin resistance, inflammation, adiposity, and subcortical vascular disease. All of these could be targets of intervention to prevent and treat dementia.⁴

Instead of a beta-amyloid trigger, let us hypothesize that metabolic stress is the initiating element of the Alzheimer cascade, which then triggers beta-amyloid overproduction or underclearance, and the immune activation damages neurons. By lessening metabolic stress or by preventing immune activation, it may, in theory, be possible to prevent neurons from entering into the terminal pathway of tangle formation and cell death.

LINKS BETWEEN ALZHEIMER DISEASE AND DIABETES

Rates of dementia of all causes are higher in people with diabetes. The strongest effect has been noted in vascular dementia, but Alzheimer disease was also found to be associated with diabetes.⁵ The Framingham Heart Study⁶ found the association between dementia and diabetes was significant only when other risk factors for Alzheimer disease were minimal: in an otherwise healthy population, diabetes alone appears to trigger the risk for dementia. But in a population with a lot of vascular comorbidities, the association between diabetes and dementia is not as clear. Perhaps the magnitude of the risk is overwhelmed by greater cerebrovascular and cardiovascular morbidity.

A systematic review⁷ supported the notion that the risk of dementia is higher in people with diabetes, and even raised the issue of whether we should consider Alzheimer disease “type 3 diabetes.”

Testing of the reverse hypothesis—diabetes is more common in people with Alzheimer disease—also is supportive: diabetes mellitus and even impaired fasting glucose are approximately twice as common in people with Alzheimer disease than in those without.⁸ Fasting blood glucose levels increase steadily with age, but after age 65, they are higher in people with Alzheimer disease than in those without.

Glucose has some direct effects on brain metabolism that might explain the higher risk. Chronic hyperglycemia is associated with excessive production of free radicals, which leads to reactive oxygen species. These are toxic to neuronal membranes as well as to mitochondria, where many of the reactive oxygen species are generated. Free radicals also facilitate the inflammatory response.

We also see greater neuronal and mitochondrial calcium influx in the presence of hyperglycemia. The excess calcium interferes with mitochondrial metabolism and may trigger the cascade of apoptosis when it reaches critical levels in neurons.

Chronic hyperglycemia is also associated with increased advanced glycation end-products. These are toxic molecules produced by the persistent exposure of proteins to high sugar levels and may be facilitated by the presence of reactive oxygen species that catalyze the reactions between the sugars and the peptides. Glycation end-products are commonly recognized as the same as those occurring during browning of meat (the Maillard reaction).

Hyperglycemia also potentiates neuronal damage from ischemia. Animal experiments show that brain infarction in the presence of hyperglycemia results in worse damage than the same degree of ischemia in the absence of hyperglycemia. Hyperglycemia may exaggerate other blows to neuronal function such as those from small strokes or microvascular ischemia.

AN ALTERNATIVE TO THE AMYLOID HYPOTHESIS: THE ‘MITOCHONDRIAL CASCADE HYPOTHESIS’

Swerdlow and Khan⁹ have proposed an alternative to the amyloid hypothesis as the cause of Alzheimer disease, known as the “mitochondrial cascade hypothesis.” According to this model, as we age we accumulate more wear-and-tear from oxidative mitochondrial damage, especially the accumulation of toxins leading to reduced cell metabolic activity. This triggers the “3-R response”:

Reset. When toxins alter cell metabolism, neurons try to repair themselves by manufacturing beta-amyloid, which is a “repair-and-reset” synaptic signaling molecule that reduces energy production. Under the lower energy state, beta-pleated sheets develop from beta-amyloid, which aggregate and form amyloid plaque.

Remove. Many cells undergo programmed death when faced with oxidative stress. The first step in neuronal loss is reduced synaptic connections and, hence, losses in neuronal communication. This results in impaired cognition.

Replace. Some cells that are faced with metabolic stress re-enter the cell cycle by undergoing cell division. Neurons, however, are terminally postmitotic and die if they try to divide: by synthesizing cell division proteins, duplicating chromosomes, and reorganizing the complex internal structure, the cell cannot work properly and cell division fails. In the mitochondrial cascade hypothesis, neurofibrillary tangles result from this attempted remodeling of the cytoskeletal filaments, furthering neuronal dysfunction.

ALZHEIMER DISEASE AND STROKE: MORE ALIKE THAN WE THOUGHT?

Although historically clinicians and researchers have tried to distinguish between Alzheimer disease and vascular dementia, growing evidence indicates that the two disorders overlap significantly and that the pathologies may be synergistic.

Alzheimer disease has been hypothesized as being a vascular disorder.¹⁰ It shares many of the risk factors of vascular disease, and preclinical detection of Alzheimer disease is possible from measurements of regional cerebral perfusion. Cerebrovascular and neurodegenerative pathology are parallel in Alzheimer disease and vascular disease.

Pure Alzheimer disease and vascular disease are two ends of a pathologic continuum.¹¹ At one end is “pure” Alzheimer disease, in which patients die only with histologic findings of plaques and neurofibrillary tangles. This form may occur only in patients with the autosomal dominant early-onset form. At the other end of the spectrum are people who have serious vascular disease, multiple strokes, and microvascular ischemia and who die demented but with no evidence of the plaques and tangles of Alzheimer disease.

Between these poles is a spectrum of overlapping pathology that is either Alzheimer disease-dominant or vascular disease-dominant, with varying degrees of amyloid plaque and evidence of microvascular infarcts. Cerebral amyloid angiopathy (the accumulation of beta-amyloid in the wall of arteries in the brain) bridges the syndromes.¹² In some drug studies that attempted removing amyloid from the brain, vascular permeability was altered, resulting in brain edema.

Along the same lines as Kalaria’s model,¹¹ Snowden et al¹³ found at autopsy of aged Catholic nuns that for some the accumulation of Alzheimer pathology alone was insufficient to cause dementia, but dementia was nearly universal in nuns with the same burden of Alzheimer pathology commingled with vascular pathology.

DOES INFLAMMATION PLAY A ROLE?

The inflammatory state is a recognized risk factor for Alzheimer disease, but the clinical data are mixed. Epidemiologic evidence is strong: patients who regularly take nonsteroidal anti-inflammatory drugs (NSAIDs) or steroids for chronic, systemic inflammatory diseases (eg, arthritis) have a 45% to 60% reduced risk for Alzheimer disease.^14,15

However, multiple clinical trials in patients with Alzheimer disease have failed to show a benefit of taking anti-inflammatory drugs. One preliminary report suggested that indomethacin (Indocin) might offer benefit, but because of gastrointestinal side effects its usefulness in an elderly population is limited.

Diabetes and inflammation are also closely linked: hyperinsulinemia is proinflammatory, promoting the formation of reactive oxygen species, inhibiting the degradation of oxidized proteins, and increasing the risk for lipid per-oxidation. Insulin acts synergistically with endotoxins to raise inflammatory markers, eg, proinflammatory cytokines and C-reactive protein.¹⁶

It is possible that anti-inflammatory drugs may not work in Alzheimer disease because inflammation in the brain is mediated more by microglial cells than by prostaglandin pathways. In Alzheimer disease, inflammation is mediated by activated microglial cells, which invade plaques with their processes; these are not evident in the diffuse beta-amyloid-rich plaques seen in typical aging. The trigger for their activation is unclear, but the activated microglial cells and the invasion of plaques are seen in transgenic mouse models of Alzheimer disease, and activation is seen when beta-amyloid is injected into the brain of a healthy mouse.¹⁷

Activated microglial cells enlarge and their metabolic rate increases, with a surge in the production of proteins and other protein-mediated inflammatory markers such as alpha-antichymotrypsin, alpha-antitrypsin, serum amyloid P, C-reactive protein, nitric oxide, and proinflammatory cytokines. It is unlikely that it is healthy for cells to be exposed to these inflammatory products. Some of the cytokines are now targets of drug development for Alzheimer disease, and agents targeting these pathways have already been developed for connective tissue diseases.

In a controversial pilot study, Tobinick et al¹⁸ studied the use of etanercept (Enbrel), an inhibitor of tumor necrosis factor-alpha (an inflammatory cytokine). They injected etanercept weekly into the spinal canal in 15 patients with mild to severe Alzheimer disease, for 6 months. Patients improved in the Mini-Mental State Examination by more than two points during the study. Patent issues surrounding use of this drug in Alzheimer disease may delay further trials.

Thiazolidinediones block microglial cell activation

The reactive microglial phenotype can be prevented in cell culture by peroxisome proliferator-activated receptor (PPAR) gamma agonists. These include the antidiabetic thiazolidinediones such as pioglitazone (Actos), troglitazone (Rezulin), and rosiglitazone (Avandia), and indomethacin and other NSAIDs.

Using a Veterans Administration database of more than 142,000 patients, Miller et al¹⁹ retrospectively found that patients who took a thiazolidinedione for diabetes had a 20% lower risk of developing Alzheimer disease compared with users of insulin or metformin (Glucophage).

However, rosiglitazone showed no benefit against Alzheimer disease in a large clinical trial,²⁰ but this may be because it is rapidly cleared from the brain. Pioglitazone is not actively exported from the brain, so it may be a better candidate, but pharmaceutical industry interest in this agent is low because its patent will soon expire.

Fish oil is another PPAR-gamma agonist, and some studies indicate that eating fish may protect against developing Alzheimer disease; it may also be therapeutic if the disease is present. Double-blind controlled studies have not been carried out and likely will not because of patent issues: the costs of such studies are high, and the potential payback is low.

ESTROGEN: PROTECTIVE OR NOT?

Whether taking estrogen is a risk factor or is protective has not yet been determined. Estrogen directly affects neurons. It increases the number of dendritic spines, which are associated with improved memory. Meta-analyses suggest that hormone replacement therapy reduces the risk of dementia by about one-third. ^21,22 Both positive and negative prospective studies exist, but all are complicated by serious methodologic flaws.^23,24

Combined analysis of about 7,500 women from two double-blind, randomized, placebo-controlled trials of the Women’s Health Initiative Memory Study found that the risks of dementia and mild cognitive impairment were increased by hormone replacement therapy. The hazard ratio for dementia was found to be 1.76 (P < .005), amounting to 23 new cases of dementia per 10,000 prescriptions annually.²⁵

Patient selection may account for the conflicting results in different studies. Epidemiologic studies consisted mostly of newly postmenopausal women and those who were being treated for symptoms of vasomotor instability. In contrast, the Women’s Health Initiative enrolled only women older than 65 and excluded women with vasomotor instability. Other studies indicate that the greatest cognitive improvements with hormone therapies are seen in women with vasomotor symptoms.

WHICH RISK FACTORS CAN WE CONTROL?

In summary, some of the risk factors for Alzheimer disease can be modified if we do the following.

Aggressively manage diabetes and cardiovascular disease. Vascular risk factors significantly increase dementia risk, providing good targets for prevention: clinicians should aggressively help their patients control diabetes, hypertension, and hyperlipidemia.²⁶ However, aggressive control of hypertension in a patient with already-existing dementia may exacerbate the condition, so caution is warranted.

Optimize diet. Dietary measures include high intake of antioxidants (which are especially high in brightly colored and tart-flavored fruits and vegetables) and polyunsaturated fats.²⁶ Eating a Mediterranean-type diet that includes a high intake of cold-water ocean fish is recommended. Fish should not be fried: the high temperatures may destroy the omega-3 fatty acids, and the high fat content may inhibit their absorption.

Weigh the risks and benefits of estrogen. Although estrogen replacement therapy for postmenopausal women has had mixed results for controlling dementia, it appears to be clinically indicated to control vasomotor symptoms and likely does not increase the risk of dementia for newly menopausal women. Risks and benefits should be carefully weighed for each patient.

Optimize exercise. People who are physically active in midlife have a lower risk of Alzheimer disease.²⁷ Those who adopt new physical activity late in life may also gain some protective or restorative benefit.²⁸

Many measures, such as taking anti-inflammatory or antihypertensive drugs, probably have a very small incremental benefit over time, so it is difficult to measure significant effects during the course of a typical clinical trial.

Clinicians are already recommending actions to reduce the risk of dementia by focusing on lowering cardiovascular risk. Hopefully, as these actions become more commonly practiced as lifelong habits in those reaching the age of risk for Alzheimer disease, we will see a reduced incidence of that devastating and much-feared illness.

Despite all the attention paid to ST-segment-elevation myocardial infarction (MI), in terms of sheer numbers, non-ST-elevation MI and unstable angina are where the action is. Acute coronary syndromes account for 2.43 million hospital discharges per year. Of these, 0.46 million are for ST-elevation MI and 1.97 million are for non-ST-elevation MI and unstable angina.^1,2

A number of recent studies have begun to answer some of the pressing questions about treating these types of acute coronary syndromes. In this article, I update the reader on these studies, along with recent findings regarding stenting and antiplatelet agents. As you will see, they are all interconnected.

TO CATHETERIZE IS BETTER THAN NOT TO CATHETERIZE

In the 1990s, a topic of debate was whether patients presenting with unstable angina or non-ST-elevation MI should routinely undergo catheterization or whether they would do just as well with a conservative approach, ie, undergoing catheterization only if they developed recurrent, spontaneous, or stress-induced ischemia. Now, the data are reasonably clear and favor an aggressive strategy.³

Mehta et al⁴ performed a meta-analysis of seven randomized controlled trials (N = 9,212 patients) of aggressive vs conservative angiography and revascularization for non-ST-elevation MI or unstable angina. The results favored the aggressive strategy. At 17 months of follow-up, death or MI had occurred in 7.4% of patients who received the aggressive therapy compared with 11.0% of those who received the conservative therapy, for an odds ratio of 0.82 (P = .001).

The CRUSADE (Can Rapid Risk Stratification of Unstable Angina Patients Suppress Adverse Outcomes With Early Implemention of the ACC/AHA Guidelines?) Quality Improvement Initiative⁵ analyzed data from a registry of 17,926 patients with non-ST-elevation acute coronary syndrome who were at high risk because of positive cardiac markers or ischemic electrocardiographic changes. Overall, 2.0% of patients who received early invasive care (catheterization within the first 48 hours) died in the hospital compared with 6.2% of those who got no early invasive care, for an adjusted odds ratio of 0.63 (95% confidence interval [CI] 0.52–0.77).

The investigators also stratified the patients into those at low, medium, and high risk, using the criteria of the PURSUIT (Platelet Glycoprotein IIb/IIIa in Unstable Angina: Receptor Suppression Using Integrilin [eptifibatide] Therapy) risk score. There were fewer deaths with early invasive therapy in each risk group, and the risk reduction was greatest in the high-risk group.⁵

Bavry et al⁶ performed an updated meta-analysis of randomized trials. At a mean follow-up of 24 months, the relative risk of death from any cause was 0.75 in patients who received early invasive therapy.

In another meta-analysis, O’Donoghue et al⁷ found that the odds ratio of death, MI, or rehospitalization with acute coronary syndromes was 0.73 (95% CI 0.55–0.98) in men who received invasive vs conservative therapy; in women it was 0.81 (95% CI 0.65–1.01). In women, the benefit was statistically significant in those who had elevations of creatine kinase MB or troponin but not in those who did not, though the benefit in men appeared to be less dependent on the presence of biomarker abnormalities.

MUST ANGIOGRAPHY BE DONE IN THE FIRST 24 HOURS?

Although a number of trials showed that a routine invasive strategy leads to better outcomes than a conservative strategy, until recently we had no information as to whether the catheterization needed to be done early (eg, within the first 24 hours) or if it could be delayed a day or two while the patient received medical therapy.

Mehta et al⁸ conducted a trial to find out: the Timing of Intervention in Acute Coronary Syndrome (TIMACS) trial. Patients were included if they had unstable angina or non-ST-elevation MI, presented to a hospital within 24 hours of the onset of symptoms, and had two of three high-risk features: age 60 years or older, elevated cardiac biomarkers, or electrocardiographic findings compatible with ischemia. All received standard medical therapy, and 3,031 were randomly assigned to undergo angiography either within 24 hours after randomization or 36 or more hours after randomization.

At 6 months, the primary outcome of death, new MI, or stroke had occurred in 9.6% of the patients in the early-intervention group and in 11.3% of those in the delayed-intervention group, but the difference was not statistically significant. However, the difference in the rate of a secondary end point, death, MI, or refractory ischemia, was statistically significant: 9.5% vs 12.9%, P = .003, owing mainly to less refractory ischemia with early intervention.

The patients were also stratified into two groups by baseline risk. The rate of the primary outcome was significantly lower with early intervention in high-risk patients, but not in those at intermediate or low risk. Thus, early intervention may be beneficial in patients at high risk, such as those with ongoing chest pain, but not necessarily in those at low risk.

LEAVE NO LESION BEHIND?

Coronary artery disease often affects more than one segment. Until recently, it was not known whether we should stent all stenotic segments in patients presenting with non-ST-elevation MI or unstable angina, or only the “culprit lesion.”

Shishehbor et al⁹ examined data from a Cleveland Clinic registry of 1,240 patients with acute coronary syndrome and multivessel coronary artery disease who underwent bare-metal stenting. The median follow-up was 2.3 years. Using a propensity model to match patients in the two groups with similar baseline characteristics, they found that the rate of repeat revascularization was less with multivessel intervention than with culprit-only stenting, as was the rate of the combined end point of death, MI, or revascularization, but not that of all-cause mortality or the composite of death or MI.

BARE-METAL VS DRUG-ELUTING STENTS: BALANCING THE RISKS AND BENEFITS

After a patient receives a stent, two bad things can happen: the artery can close up again either gradually, in a process called restenosis, or suddenly, via thrombosis.

Drug-eluting stents were invented to solve the problem of restenosis, and they work very well. Stone et al¹⁰ pooled the data from four double-blind trials of sirolimus (Rapamune) stents and five double-blind trials of paclitaxel (Taxol) stents and found that, at 4 years, the rates of target-lesion revascularization (for restenosis) were 7.8% with sirolimus stents vs 23.6% with bare-metal stents (P < .001), and 10.1% with paclitaxel stents vs 20.0% with bare-metal stents (P < .001).

Thrombosis was much less common in these studies, occurring in 1.2% of the sirolimus stent groups vs 0.6% of the bare-metal stent groups (P = .20), and in 1.3% of the paclitaxel stent groups vs 0.9% of the bare-metal stent groups (P = .30).¹⁰

However, drug-eluting stents appear to increase the risk of thrombosis later on, ie, after 1 year. Bavry et al,¹¹ in a meta-analysis, calculated that when stent thrombosis occurred, the median time after implantation was 15.5 months with sirolimus stents vs 4 months with bare-metal stents (P = .0052), and 18 months with paclitaxel stents vs 3.5 months with bare-metal stents (P = .04). The absolute risk of very late stent thrombosis after 1 year was very low, with five events per 1,000 patients with drug-eluting stents vs no events with bare-metal stents (P = .02). Nevertheless, this finding has practical implications. How long must patients continue dual antiplatelet therapy? And what if a patient needs surgery a year later?

Restenosis is not always so gradual

Although stent thrombosis is serious and often fatal, bare-metal stent restenosis is not always benign either, despite the classic view that stent restenosis is a gradual process that results in exertional angina. Reviewing 1,186 cases of bare-metal stent restenosis in 984 patients at Cleveland Clinic, Chen et al¹² reported that 9.5% of cases presented as acute MI (2.2% as ST-elevation MI and 7.3% as non-ST-elevation MI), and 26.4% as unstable angina requiring hospitalization.

A Mayo Clinic study¹³ corroborated these findings. The 10-year incidence of clinical bare-metal stent restenosis was 18.1%, and the incidence of MI was 2.1%. The 10-year rate of bare-metal stent thrombosis was 2%. Off-label use, primarily in saphenous vein grafts, increased the incidence; other correlates were prior MI, peripheral arterial disease, and ulcerated lesions.

Furthermore, bare-metal stent thrombosis can also occur later. We saw a case that occurred 13 years after the procedure, 3 days after the patient stopped taking aspirin because he was experiencing flu-like symptoms, ran out of aspirin, and felt too sick to go out and buy more. The presentation was with ST-elevation MI. The patient recovered after treatment with intracoronary abciximab (ReoPro), percutaneous thrombectomy, balloon angioplasty, and, eventually, bypass surgery.¹⁴

No difference in risk of death with drug-eluting vs bare-metal stents

Even though drug-eluting stents pose a slightly higher risk of thrombosis than bare-metal stents, the risk of death is no higher.¹⁵

I believe the reason is that there are competing risks, and that the higher risk of thrombosis with first-generation drug-eluting stents and the higher risk of restenosis with bare-metal stents essentially cancel each other out. For most patients, there is an absolute benefit with drug-eluting stents, which reduce the need for revascularization with no effect in terms of either increasing or decreasing the risk of MI or death. Second-generation drug-eluting stents may have advantages in reducing rates of death or MI compared with first-generation drug-eluting stents, though this remains to be proven conclusively.

The right revascularization for the right patient

Bavry and I¹⁶ developed an algorithm for deciding on revascularization, posing a series of questions:

• Does the patient need any form of revascularization?
• Is he or she at higher risk of both stent thrombosis and restenosis, as in patients with diabetes, diffuse multivessel disease with bifurcation lesions, or chronic total occlusions? If so, coronary artery bypass grafting remains an excellent option.
• Does he or she have a low risk of restenosis, as in patients without diabetes with focal lesions in large vessels? If so, one could consider a bare-metal stent, which would probably be more cost-effective than a drug-eluting stent in this situation.
• Does the patient have relative contraindications to drug-eluting stents? Examples are a history of noncompliance with medical therapy, financial issues such as lack of insurance that would make buying clopidogrel (Plavix) a problem, long-term anticoagulation, or anticipated need for surgery in the next few years.

If a drug-eluting stent is used, certain measures can help ensure that it is used optimally. It should often be placed under high pressure with a noncompliant balloon so that it achieves contact with the artery wall all around. One should consider intravascular ultrasonographic guidance to make sure the stent is well opposed if it is in a very calcified lesion. Dual antiplatelet therapy with clopidogrel and aspirin should be given for at least 1 year, and if there is no bleeding, perhaps longer, pending further data.¹⁶

LEAVE NO PLATELET ACTIVATED?

Platelets have several types of receptors that, when bound by their respective ligands, lead to platelet activation and aggregation and, ultimately, thrombus formation. Antagonists to some of these receptors are available or are being developed.¹⁷

For long-term therapy, blocking the process “upstream,” ie, preventing platelet activation, is better than blocking it “downstream,” ie, preventing aggregation. For example, clopidogrel, ticlopipine (Ticlid), and prasugrel (Effient) have active metabolites that bind to a subtype of the adenosine diphosphate receptor and prevent platelet activation, whereas the glycoprotein IIb/IIIa inhibitors such as abciximab work downstream, binding to a different receptor and preventing aggregation.¹⁸

The evidence for using dual antiplatelet therapy (ie, aspirin plus clopidogrel) in patients with acute coronary syndromes without ST-elevation is very well established.

The Clopidogrel in Unstable Angina to Prevent Recurrent Events (CURE) trial,¹⁹ published in 2001, found a 20% relative risk reduction and a 2% absolute risk reduction in the incidence of MI, stroke, or cardiovascular death in patients randomly assigned to receive clopidogrel plus aspirin for 1 year vs aspirin alone for 1 year (P < .001). In the subgroup of patients who underwent percutaneous coronary intervention, the relative risk reduction in the incidence of MI or cardiovascular death at 1 year of follow-up was 31% (P = .002).²⁰

As a result of these findings, the cardiology society guidelines²¹ recommend a year of dual antiplatelet therapy after acute coronary syndromes, regardless of whether the patient is treated medically, percutaneously, or surgically.

But what happens after clopidogrel is withdrawn? Ho et al²² retrospectively analyzed data from Veterans Affairs hospitals and found a spike in the incidence of death or MI in the first 90 days after stopping clopidogrel treatment. This was true in medically treated patients as well as in those treated with percutaneous coronary interventions, in those with or without diabetes mellitus, in those who received a drug-eluting stent or a bare-metal stent, and in those treated longer than 9 months.

The investigators concluded that there might be a “clopidogrel rebound effect.” However, I believe that a true rebound effect, such as after withdrawal of heparin or warfarin, is biologically unlikely with clopidogrel, since clopidogrel irreversibly binds to its receptor for the 7- to 10-day life span of the platelet. Rather, I believe the phenomenon must be due to withdrawal of protection in patients at risk.

In stable patients, dual therapy is not as beneficial

Would dual antiplatelet therapy with clopidogrel and aspirin also benefit patients at risk of atherothrombotic events but without acute coronary syndromes?

The Clopidogrel for High Atherothrombotic Risk and Ischemic Stabilization, Management, and Avoidance (CHARISMA) trial²³ included 15,603 patients with either clinically evident but stable cardiovascular disease or multiple risk factors for athero-thrombosis. They were randomly assigned to receive either clopidogrel 75 mg/day plus aspirin 75 to 162 mg/day or placebo plus aspirin. At a median of 28 months, the groups did not differ significantly in the rate of MI, stroke, or death from cardiovascular causes.

However, the subgroup of patients who had documented prior MI, ischemic stroke, or symptomatic peripheral arterial disease did appear to derive significant benefit from dual therapy.²⁴ In this subgroup, the rate of MI, stroke, or cardiovascular death at a median follow-up of 27.6 months was 8.8% with placebo plus aspirin compared with 7.3% with clopidogrel plus aspirin, for a hazard ratio of 0.83 (95% CI 0.72–0.96, P = .01). Unstented patients with stable coronary artery disease but without prior MI derived no benefit.

Bleeding and thrombosis: The Scylla and Charybdis of antiplatelet therapy

However, with dual antiplatelet therapy, we steer between the Scylla of bleeding and the Charybdis of thrombosis.²⁵

In the CHARISMA subgroup who had prior MI, ischemic stroke, or symptomatic peripheral arterial disease, the incidence of moderate or severe bleeding was higher with dual therapy than with aspirin alone, but the rates converged after about 1 year of treatment.²⁴ Further, there was no difference in fatal bleeding or intracranial bleeding, although the rate of moderate bleeding (defined as the need for transfusion) was higher with dual therapy (2.0% vs 1.3%, P = .004).

I believe the data indicate that if a patient can tolerate dual antiplatelet therapy for 9 to 12 months without any bleeding issues, he or she is unlikely to have a major bleeding episode if dual therapy is continued beyond this time.

About half of bleeding events in patients on chronic antiplatelet therapy are gastrointestinal. To address this risk, in 2008 an expert committee from the American College of Cardiology, American College of Gastroenterology, and American Heart Association issued a consensus document²⁶ in which they recommended assessing gastrointestinal risk factors in patients on antiplatelet therapy, such as history of ulcers (and testing for and treating Helicobacter pylori infection if present), history of gastrointestinal bleeding, concomitant anticoagulant therapy, and dual antiplatelet therapy. If any of these were present, the committee recommended considering a proton pump inhibitor. The committee also recommended a proton pump inhibitor for patients on antiplatelet therapy who have more than one of the following: age 60 years or more, corticosteroid use, or dyspepsia or gastroesophageal reflux symptoms.

Some ex vivo platelet studies and observational analyses have suggested that there might be an adverse interaction between clopidogrel and proton pump inhibitors due to a blunting of clopidogrel’s antiplatelet effect. A large randomized clinical trial was designed and launched to determine if a single-pill combination of the proton pump inhibitor omeprazole (Prilosec) and clopidogrel would be safer than clopidogrel alone when added to aspirin. Called COGENT-1 (Clopidogrel and the Optimization of GI Events Trial), it was halted early in 2009 when it lost its funding. However, preliminary data did not show an adverse interaction between clopidogrel and omeprazole.

What is the right dose of aspirin?

Steinhubl et al²⁷ performed a post hoc observational analysis of data from the CHARISMA trial. Their findings suggested that higher doses of aspirin are not more effective than lower doses for chronic therapy. Furthermore, in the group receiving clopidogrel plus aspirin, the incidence of severe or life-threatening bleeding was significantly greater with aspirin doses higher than 100 mg than with doses lower than 100 mg, 2.6% vs 1.7%, P = .040.

A randomized, controlled trial called Clopidogrel Optimal Loading Dose Usage to Reduce Recurrent Events/Optimal Antiplatelet Strategy for Interventions (CURRENT/OASIS 7)²⁸ recently reported that higher-dose aspirin (ie, 325 mg) may be better than lower dose aspirin (ie, 81 mg) in patients with acute coronary syndromes undergoing percutaneous coronary intervention and receiving clopidogrel. During this 30-day study, there was no increase in overall bleeding with the higher dose of aspirin, though gastrointestinal bleeding was slightly increased.²⁹ In a factorial design, the second part of this trial found that a higher-dose clopidogrel regimen reduced stent thrombosis.²⁹

In vitro, response to clopidogrel shows a normal bell-shaped distribution.³⁰ In theory, therefore, patients who are hyperresponders may be at higher risk of bleeding, and those who are hyporesponders may be at risk of ischemic events.

A clinical trial is under way to examine whether hyporesponders should get higher doses. Called GRAVITAS (Gauging Responsiveness With a VerifyNow Assay Impact on Thrombosis and Safety), it will use a point-of-care platelet assay and then allocate patients to receive either standard therapy or double the dose of clopidogrel. The primary end point will be the rate of cardiovascular death, nonfatal MI, or stent thrombosis at 6 months.

Is prasugrel better than clopidogrel?

Prasugrel (Effient) is a new drug of the same class as clopidogrel, ie, a thienopyridine, with its active metabolite binding to the same platelet receptor as clopidogrel and inhibiting platelet aggregation more rapidly, more consistently, and to a greater extent than clopidogrel. Prasugrel was recently approved by the Food and Drug Administration. But is it better?³¹

The Trial to Assess Improvement in Therapeutic Outcomes by Optimizing Platelet Inhibition With Prasugrel–Thrombolysis in Myocardial Infarction (TRITON-TIMI 38) compared prasugrel and clopidogrel in 13,608 patients with moderate- to high-risk acute coronary syndromes who were scheduled to undergo percutaneous coronary intervention.³²

Overall, prasugrel was better. At 15 months, the incidence of the primary end point (death from cardiovascular causes, nonfatal MI, or nonfatal stroke) was significantly lower with prasugrel therapy than with clopidogrel in the entire cohort (9.9% vs 12.1%, hazard ratio 0.81, 95% CI 0.73–0.90, P < .001), in the subgroup with ST-segment elevation MI, and in the subgroup with unstable angina or non-ST-elevation MI.

However, there was a price to pay. The rate of major bleeding was higher with prasugrel (2.4% vs 1.8%, hazard ratio 1.32, 95% CI 1.03–1.68, P = .03). Assessing the balance between the risk and the benefit, the investigators identified three subgroups who did not derive a net clinical benefit from prasugrel: patients who had had a previous stroke or transient ischemic attack (this group actually had a net harm from prasugrel), patients 75 years of age or older, and patients weighing less than 60 kg (132 pounds).

More work is needed to determine which patients are best served by standard-dose clopidogrel, higher doses of clopidogrel, platelet-assay-guided dosing of clopidogrel, or prasugrel.²⁴

Short-acting, potent intravenous platelet blockade with an agent such as cangrelor is theoretically appealing, but further research is necessary.^33,34 Ticagrelor, a reversible adenosine diphosphate receptor antagonist, provides yet another potential option in antiplatelet therapy for acute coronary syndromes. In the recent PLATO trial (Study of Platelet Inhibition and Patient Outcomes), compared with clopidogrel, ticagrelor reduced the risk of ischemic events, including death.^35,36 Here, too, there was more major bleeding (unrelated to coronary artery bypass grafting) with ticagrelor.

Thus, clinical assessment of an individual patient’s ischemic and bleeding risks will continue to be critical as therapeutic strategies evolve.

Despite all the attention paid to ST-segment-elevation myocardial infarction (MI), in terms of sheer numbers, non-ST-elevation MI and unstable angina are where the action is. Acute coronary syndromes account for 2.43 million hospital discharges per year. Of these, 0.46 million are for ST-elevation MI and 1.97 million are for non-ST-elevation MI and unstable angina.^1,2

A number of recent studies have begun to answer some of the pressing questions about treating these types of acute coronary syndromes. In this article, I update the reader on these studies, along with recent findings regarding stenting and antiplatelet agents. As you will see, they are all interconnected.

TO CATHETERIZE IS BETTER THAN NOT TO CATHETERIZE

In the 1990s, a topic of debate was whether patients presenting with unstable angina or non-ST-elevation MI should routinely undergo catheterization or whether they would do just as well with a conservative approach, ie, undergoing catheterization only if they developed recurrent, spontaneous, or stress-induced ischemia. Now, the data are reasonably clear and favor an aggressive strategy.³

Mehta et al⁴ performed a meta-analysis of seven randomized controlled trials (N = 9,212 patients) of aggressive vs conservative angiography and revascularization for non-ST-elevation MI or unstable angina. The results favored the aggressive strategy. At 17 months of follow-up, death or MI had occurred in 7.4% of patients who received the aggressive therapy compared with 11.0% of those who received the conservative therapy, for an odds ratio of 0.82 (P = .001).

The CRUSADE (Can Rapid Risk Stratification of Unstable Angina Patients Suppress Adverse Outcomes With Early Implemention of the ACC/AHA Guidelines?) Quality Improvement Initiative⁵ analyzed data from a registry of 17,926 patients with non-ST-elevation acute coronary syndrome who were at high risk because of positive cardiac markers or ischemic electrocardiographic changes. Overall, 2.0% of patients who received early invasive care (catheterization within the first 48 hours) died in the hospital compared with 6.2% of those who got no early invasive care, for an adjusted odds ratio of 0.63 (95% confidence interval [CI] 0.52–0.77).

The investigators also stratified the patients into those at low, medium, and high risk, using the criteria of the PURSUIT (Platelet Glycoprotein IIb/IIIa in Unstable Angina: Receptor Suppression Using Integrilin [eptifibatide] Therapy) risk score. There were fewer deaths with early invasive therapy in each risk group, and the risk reduction was greatest in the high-risk group.⁵

Bavry et al⁶ performed an updated meta-analysis of randomized trials. At a mean follow-up of 24 months, the relative risk of death from any cause was 0.75 in patients who received early invasive therapy.

In another meta-analysis, O’Donoghue et al⁷ found that the odds ratio of death, MI, or rehospitalization with acute coronary syndromes was 0.73 (95% CI 0.55–0.98) in men who received invasive vs conservative therapy; in women it was 0.81 (95% CI 0.65–1.01). In women, the benefit was statistically significant in those who had elevations of creatine kinase MB or troponin but not in those who did not, though the benefit in men appeared to be less dependent on the presence of biomarker abnormalities.

MUST ANGIOGRAPHY BE DONE IN THE FIRST 24 HOURS?

Although a number of trials showed that a routine invasive strategy leads to better outcomes than a conservative strategy, until recently we had no information as to whether the catheterization needed to be done early (eg, within the first 24 hours) or if it could be delayed a day or two while the patient received medical therapy.

Mehta et al⁸ conducted a trial to find out: the Timing of Intervention in Acute Coronary Syndrome (TIMACS) trial. Patients were included if they had unstable angina or non-ST-elevation MI, presented to a hospital within 24 hours of the onset of symptoms, and had two of three high-risk features: age 60 years or older, elevated cardiac biomarkers, or electrocardiographic findings compatible with ischemia. All received standard medical therapy, and 3,031 were randomly assigned to undergo angiography either within 24 hours after randomization or 36 or more hours after randomization.

At 6 months, the primary outcome of death, new MI, or stroke had occurred in 9.6% of the patients in the early-intervention group and in 11.3% of those in the delayed-intervention group, but the difference was not statistically significant. However, the difference in the rate of a secondary end point, death, MI, or refractory ischemia, was statistically significant: 9.5% vs 12.9%, P = .003, owing mainly to less refractory ischemia with early intervention.

The patients were also stratified into two groups by baseline risk. The rate of the primary outcome was significantly lower with early intervention in high-risk patients, but not in those at intermediate or low risk. Thus, early intervention may be beneficial in patients at high risk, such as those with ongoing chest pain, but not necessarily in those at low risk.

LEAVE NO LESION BEHIND?

Coronary artery disease often affects more than one segment. Until recently, it was not known whether we should stent all stenotic segments in patients presenting with non-ST-elevation MI or unstable angina, or only the “culprit lesion.”

Shishehbor et al⁹ examined data from a Cleveland Clinic registry of 1,240 patients with acute coronary syndrome and multivessel coronary artery disease who underwent bare-metal stenting. The median follow-up was 2.3 years. Using a propensity model to match patients in the two groups with similar baseline characteristics, they found that the rate of repeat revascularization was less with multivessel intervention than with culprit-only stenting, as was the rate of the combined end point of death, MI, or revascularization, but not that of all-cause mortality or the composite of death or MI.

BARE-METAL VS DRUG-ELUTING STENTS: BALANCING THE RISKS AND BENEFITS

After a patient receives a stent, two bad things can happen: the artery can close up again either gradually, in a process called restenosis, or suddenly, via thrombosis.

Drug-eluting stents were invented to solve the problem of restenosis, and they work very well. Stone et al¹⁰ pooled the data from four double-blind trials of sirolimus (Rapamune) stents and five double-blind trials of paclitaxel (Taxol) stents and found that, at 4 years, the rates of target-lesion revascularization (for restenosis) were 7.8% with sirolimus stents vs 23.6% with bare-metal stents (P < .001), and 10.1% with paclitaxel stents vs 20.0% with bare-metal stents (P < .001).

Thrombosis was much less common in these studies, occurring in 1.2% of the sirolimus stent groups vs 0.6% of the bare-metal stent groups (P = .20), and in 1.3% of the paclitaxel stent groups vs 0.9% of the bare-metal stent groups (P = .30).¹⁰

However, drug-eluting stents appear to increase the risk of thrombosis later on, ie, after 1 year. Bavry et al,¹¹ in a meta-analysis, calculated that when stent thrombosis occurred, the median time after implantation was 15.5 months with sirolimus stents vs 4 months with bare-metal stents (P = .0052), and 18 months with paclitaxel stents vs 3.5 months with bare-metal stents (P = .04). The absolute risk of very late stent thrombosis after 1 year was very low, with five events per 1,000 patients with drug-eluting stents vs no events with bare-metal stents (P = .02). Nevertheless, this finding has practical implications. How long must patients continue dual antiplatelet therapy? And what if a patient needs surgery a year later?

Restenosis is not always so gradual

Although stent thrombosis is serious and often fatal, bare-metal stent restenosis is not always benign either, despite the classic view that stent restenosis is a gradual process that results in exertional angina. Reviewing 1,186 cases of bare-metal stent restenosis in 984 patients at Cleveland Clinic, Chen et al¹² reported that 9.5% of cases presented as acute MI (2.2% as ST-elevation MI and 7.3% as non-ST-elevation MI), and 26.4% as unstable angina requiring hospitalization.

A Mayo Clinic study¹³ corroborated these findings. The 10-year incidence of clinical bare-metal stent restenosis was 18.1%, and the incidence of MI was 2.1%. The 10-year rate of bare-metal stent thrombosis was 2%. Off-label use, primarily in saphenous vein grafts, increased the incidence; other correlates were prior MI, peripheral arterial disease, and ulcerated lesions.

Furthermore, bare-metal stent thrombosis can also occur later. We saw a case that occurred 13 years after the procedure, 3 days after the patient stopped taking aspirin because he was experiencing flu-like symptoms, ran out of aspirin, and felt too sick to go out and buy more. The presentation was with ST-elevation MI. The patient recovered after treatment with intracoronary abciximab (ReoPro), percutaneous thrombectomy, balloon angioplasty, and, eventually, bypass surgery.¹⁴

No difference in risk of death with drug-eluting vs bare-metal stents

Even though drug-eluting stents pose a slightly higher risk of thrombosis than bare-metal stents, the risk of death is no higher.¹⁵

I believe the reason is that there are competing risks, and that the higher risk of thrombosis with first-generation drug-eluting stents and the higher risk of restenosis with bare-metal stents essentially cancel each other out. For most patients, there is an absolute benefit with drug-eluting stents, which reduce the need for revascularization with no effect in terms of either increasing or decreasing the risk of MI or death. Second-generation drug-eluting stents may have advantages in reducing rates of death or MI compared with first-generation drug-eluting stents, though this remains to be proven conclusively.

The right revascularization for the right patient

Bavry and I¹⁶ developed an algorithm for deciding on revascularization, posing a series of questions:

• Does the patient need any form of revascularization?
• Is he or she at higher risk of both stent thrombosis and restenosis, as in patients with diabetes, diffuse multivessel disease with bifurcation lesions, or chronic total occlusions? If so, coronary artery bypass grafting remains an excellent option.
• Does he or she have a low risk of restenosis, as in patients without diabetes with focal lesions in large vessels? If so, one could consider a bare-metal stent, which would probably be more cost-effective than a drug-eluting stent in this situation.
• Does the patient have relative contraindications to drug-eluting stents? Examples are a history of noncompliance with medical therapy, financial issues such as lack of insurance that would make buying clopidogrel (Plavix) a problem, long-term anticoagulation, or anticipated need for surgery in the next few years.

If a drug-eluting stent is used, certain measures can help ensure that it is used optimally. It should often be placed under high pressure with a noncompliant balloon so that it achieves contact with the artery wall all around. One should consider intravascular ultrasonographic guidance to make sure the stent is well opposed if it is in a very calcified lesion. Dual antiplatelet therapy with clopidogrel and aspirin should be given for at least 1 year, and if there is no bleeding, perhaps longer, pending further data.¹⁶

LEAVE NO PLATELET ACTIVATED?

Platelets have several types of receptors that, when bound by their respective ligands, lead to platelet activation and aggregation and, ultimately, thrombus formation. Antagonists to some of these receptors are available or are being developed.¹⁷

For long-term therapy, blocking the process “upstream,” ie, preventing platelet activation, is better than blocking it “downstream,” ie, preventing aggregation. For example, clopidogrel, ticlopipine (Ticlid), and prasugrel (Effient) have active metabolites that bind to a subtype of the adenosine diphosphate receptor and prevent platelet activation, whereas the glycoprotein IIb/IIIa inhibitors such as abciximab work downstream, binding to a different receptor and preventing aggregation.¹⁸

The evidence for using dual antiplatelet therapy (ie, aspirin plus clopidogrel) in patients with acute coronary syndromes without ST-elevation is very well established.

The Clopidogrel in Unstable Angina to Prevent Recurrent Events (CURE) trial,¹⁹ published in 2001, found a 20% relative risk reduction and a 2% absolute risk reduction in the incidence of MI, stroke, or cardiovascular death in patients randomly assigned to receive clopidogrel plus aspirin for 1 year vs aspirin alone for 1 year (P < .001). In the subgroup of patients who underwent percutaneous coronary intervention, the relative risk reduction in the incidence of MI or cardiovascular death at 1 year of follow-up was 31% (P = .002).²⁰

As a result of these findings, the cardiology society guidelines²¹ recommend a year of dual antiplatelet therapy after acute coronary syndromes, regardless of whether the patient is treated medically, percutaneously, or surgically.

But what happens after clopidogrel is withdrawn? Ho et al²² retrospectively analyzed data from Veterans Affairs hospitals and found a spike in the incidence of death or MI in the first 90 days after stopping clopidogrel treatment. This was true in medically treated patients as well as in those treated with percutaneous coronary interventions, in those with or without diabetes mellitus, in those who received a drug-eluting stent or a bare-metal stent, and in those treated longer than 9 months.

The investigators concluded that there might be a “clopidogrel rebound effect.” However, I believe that a true rebound effect, such as after withdrawal of heparin or warfarin, is biologically unlikely with clopidogrel, since clopidogrel irreversibly binds to its receptor for the 7- to 10-day life span of the platelet. Rather, I believe the phenomenon must be due to withdrawal of protection in patients at risk.

In stable patients, dual therapy is not as beneficial

Would dual antiplatelet therapy with clopidogrel and aspirin also benefit patients at risk of atherothrombotic events but without acute coronary syndromes?

The Clopidogrel for High Atherothrombotic Risk and Ischemic Stabilization, Management, and Avoidance (CHARISMA) trial²³ included 15,603 patients with either clinically evident but stable cardiovascular disease or multiple risk factors for athero-thrombosis. They were randomly assigned to receive either clopidogrel 75 mg/day plus aspirin 75 to 162 mg/day or placebo plus aspirin. At a median of 28 months, the groups did not differ significantly in the rate of MI, stroke, or death from cardiovascular causes.

However, the subgroup of patients who had documented prior MI, ischemic stroke, or symptomatic peripheral arterial disease did appear to derive significant benefit from dual therapy.²⁴ In this subgroup, the rate of MI, stroke, or cardiovascular death at a median follow-up of 27.6 months was 8.8% with placebo plus aspirin compared with 7.3% with clopidogrel plus aspirin, for a hazard ratio of 0.83 (95% CI 0.72–0.96, P = .01). Unstented patients with stable coronary artery disease but without prior MI derived no benefit.

Bleeding and thrombosis: The Scylla and Charybdis of antiplatelet therapy

However, with dual antiplatelet therapy, we steer between the Scylla of bleeding and the Charybdis of thrombosis.²⁵

In the CHARISMA subgroup who had prior MI, ischemic stroke, or symptomatic peripheral arterial disease, the incidence of moderate or severe bleeding was higher with dual therapy than with aspirin alone, but the rates converged after about 1 year of treatment.²⁴ Further, there was no difference in fatal bleeding or intracranial bleeding, although the rate of moderate bleeding (defined as the need for transfusion) was higher with dual therapy (2.0% vs 1.3%, P = .004).

I believe the data indicate that if a patient can tolerate dual antiplatelet therapy for 9 to 12 months without any bleeding issues, he or she is unlikely to have a major bleeding episode if dual therapy is continued beyond this time.

About half of bleeding events in patients on chronic antiplatelet therapy are gastrointestinal. To address this risk, in 2008 an expert committee from the American College of Cardiology, American College of Gastroenterology, and American Heart Association issued a consensus document²⁶ in which they recommended assessing gastrointestinal risk factors in patients on antiplatelet therapy, such as history of ulcers (and testing for and treating Helicobacter pylori infection if present), history of gastrointestinal bleeding, concomitant anticoagulant therapy, and dual antiplatelet therapy. If any of these were present, the committee recommended considering a proton pump inhibitor. The committee also recommended a proton pump inhibitor for patients on antiplatelet therapy who have more than one of the following: age 60 years or more, corticosteroid use, or dyspepsia or gastroesophageal reflux symptoms.

Some ex vivo platelet studies and observational analyses have suggested that there might be an adverse interaction between clopidogrel and proton pump inhibitors due to a blunting of clopidogrel’s antiplatelet effect. A large randomized clinical trial was designed and launched to determine if a single-pill combination of the proton pump inhibitor omeprazole (Prilosec) and clopidogrel would be safer than clopidogrel alone when added to aspirin. Called COGENT-1 (Clopidogrel and the Optimization of GI Events Trial), it was halted early in 2009 when it lost its funding. However, preliminary data did not show an adverse interaction between clopidogrel and omeprazole.

What is the right dose of aspirin?

Steinhubl et al²⁷ performed a post hoc observational analysis of data from the CHARISMA trial. Their findings suggested that higher doses of aspirin are not more effective than lower doses for chronic therapy. Furthermore, in the group receiving clopidogrel plus aspirin, the incidence of severe or life-threatening bleeding was significantly greater with aspirin doses higher than 100 mg than with doses lower than 100 mg, 2.6% vs 1.7%, P = .040.

A randomized, controlled trial called Clopidogrel Optimal Loading Dose Usage to Reduce Recurrent Events/Optimal Antiplatelet Strategy for Interventions (CURRENT/OASIS 7)²⁸ recently reported that higher-dose aspirin (ie, 325 mg) may be better than lower dose aspirin (ie, 81 mg) in patients with acute coronary syndromes undergoing percutaneous coronary intervention and receiving clopidogrel. During this 30-day study, there was no increase in overall bleeding with the higher dose of aspirin, though gastrointestinal bleeding was slightly increased.²⁹ In a factorial design, the second part of this trial found that a higher-dose clopidogrel regimen reduced stent thrombosis.²⁹

In vitro, response to clopidogrel shows a normal bell-shaped distribution.³⁰ In theory, therefore, patients who are hyperresponders may be at higher risk of bleeding, and those who are hyporesponders may be at risk of ischemic events.

A clinical trial is under way to examine whether hyporesponders should get higher doses. Called GRAVITAS (Gauging Responsiveness With a VerifyNow Assay Impact on Thrombosis and Safety), it will use a point-of-care platelet assay and then allocate patients to receive either standard therapy or double the dose of clopidogrel. The primary end point will be the rate of cardiovascular death, nonfatal MI, or stent thrombosis at 6 months.

Is prasugrel better than clopidogrel?

Prasugrel (Effient) is a new drug of the same class as clopidogrel, ie, a thienopyridine, with its active metabolite binding to the same platelet receptor as clopidogrel and inhibiting platelet aggregation more rapidly, more consistently, and to a greater extent than clopidogrel. Prasugrel was recently approved by the Food and Drug Administration. But is it better?³¹

The Trial to Assess Improvement in Therapeutic Outcomes by Optimizing Platelet Inhibition With Prasugrel–Thrombolysis in Myocardial Infarction (TRITON-TIMI 38) compared prasugrel and clopidogrel in 13,608 patients with moderate- to high-risk acute coronary syndromes who were scheduled to undergo percutaneous coronary intervention.³²

Overall, prasugrel was better. At 15 months, the incidence of the primary end point (death from cardiovascular causes, nonfatal MI, or nonfatal stroke) was significantly lower with prasugrel therapy than with clopidogrel in the entire cohort (9.9% vs 12.1%, hazard ratio 0.81, 95% CI 0.73–0.90, P < .001), in the subgroup with ST-segment elevation MI, and in the subgroup with unstable angina or non-ST-elevation MI.

However, there was a price to pay. The rate of major bleeding was higher with prasugrel (2.4% vs 1.8%, hazard ratio 1.32, 95% CI 1.03–1.68, P = .03). Assessing the balance between the risk and the benefit, the investigators identified three subgroups who did not derive a net clinical benefit from prasugrel: patients who had had a previous stroke or transient ischemic attack (this group actually had a net harm from prasugrel), patients 75 years of age or older, and patients weighing less than 60 kg (132 pounds).

More work is needed to determine which patients are best served by standard-dose clopidogrel, higher doses of clopidogrel, platelet-assay-guided dosing of clopidogrel, or prasugrel.²⁴

Short-acting, potent intravenous platelet blockade with an agent such as cangrelor is theoretically appealing, but further research is necessary.^33,34 Ticagrelor, a reversible adenosine diphosphate receptor antagonist, provides yet another potential option in antiplatelet therapy for acute coronary syndromes. In the recent PLATO trial (Study of Platelet Inhibition and Patient Outcomes), compared with clopidogrel, ticagrelor reduced the risk of ischemic events, including death.^35,36 Here, too, there was more major bleeding (unrelated to coronary artery bypass grafting) with ticagrelor.

Thus, clinical assessment of an individual patient’s ischemic and bleeding risks will continue to be critical as therapeutic strategies evolve.

Despite all the attention paid to ST-segment-elevation myocardial infarction (MI), in terms of sheer numbers, non-ST-elevation MI and unstable angina are where the action is. Acute coronary syndromes account for 2.43 million hospital discharges per year. Of these, 0.46 million are for ST-elevation MI and 1.97 million are for non-ST-elevation MI and unstable angina.^1,2

A number of recent studies have begun to answer some of the pressing questions about treating these types of acute coronary syndromes. In this article, I update the reader on these studies, along with recent findings regarding stenting and antiplatelet agents. As you will see, they are all interconnected.

TO CATHETERIZE IS BETTER THAN NOT TO CATHETERIZE

In the 1990s, a topic of debate was whether patients presenting with unstable angina or non-ST-elevation MI should routinely undergo catheterization or whether they would do just as well with a conservative approach, ie, undergoing catheterization only if they developed recurrent, spontaneous, or stress-induced ischemia. Now, the data are reasonably clear and favor an aggressive strategy.³

Mehta et al⁴ performed a meta-analysis of seven randomized controlled trials (N = 9,212 patients) of aggressive vs conservative angiography and revascularization for non-ST-elevation MI or unstable angina. The results favored the aggressive strategy. At 17 months of follow-up, death or MI had occurred in 7.4% of patients who received the aggressive therapy compared with 11.0% of those who received the conservative therapy, for an odds ratio of 0.82 (P = .001).

The CRUSADE (Can Rapid Risk Stratification of Unstable Angina Patients Suppress Adverse Outcomes With Early Implemention of the ACC/AHA Guidelines?) Quality Improvement Initiative⁵ analyzed data from a registry of 17,926 patients with non-ST-elevation acute coronary syndrome who were at high risk because of positive cardiac markers or ischemic electrocardiographic changes. Overall, 2.0% of patients who received early invasive care (catheterization within the first 48 hours) died in the hospital compared with 6.2% of those who got no early invasive care, for an adjusted odds ratio of 0.63 (95% confidence interval [CI] 0.52–0.77).

The investigators also stratified the patients into those at low, medium, and high risk, using the criteria of the PURSUIT (Platelet Glycoprotein IIb/IIIa in Unstable Angina: Receptor Suppression Using Integrilin [eptifibatide] Therapy) risk score. There were fewer deaths with early invasive therapy in each risk group, and the risk reduction was greatest in the high-risk group.⁵

Bavry et al⁶ performed an updated meta-analysis of randomized trials. At a mean follow-up of 24 months, the relative risk of death from any cause was 0.75 in patients who received early invasive therapy.

In another meta-analysis, O’Donoghue et al⁷ found that the odds ratio of death, MI, or rehospitalization with acute coronary syndromes was 0.73 (95% CI 0.55–0.98) in men who received invasive vs conservative therapy; in women it was 0.81 (95% CI 0.65–1.01). In women, the benefit was statistically significant in those who had elevations of creatine kinase MB or troponin but not in those who did not, though the benefit in men appeared to be less dependent on the presence of biomarker abnormalities.

MUST ANGIOGRAPHY BE DONE IN THE FIRST 24 HOURS?

Although a number of trials showed that a routine invasive strategy leads to better outcomes than a conservative strategy, until recently we had no information as to whether the catheterization needed to be done early (eg, within the first 24 hours) or if it could be delayed a day or two while the patient received medical therapy.

Mehta et al⁸ conducted a trial to find out: the Timing of Intervention in Acute Coronary Syndrome (TIMACS) trial. Patients were included if they had unstable angina or non-ST-elevation MI, presented to a hospital within 24 hours of the onset of symptoms, and had two of three high-risk features: age 60 years or older, elevated cardiac biomarkers, or electrocardiographic findings compatible with ischemia. All received standard medical therapy, and 3,031 were randomly assigned to undergo angiography either within 24 hours after randomization or 36 or more hours after randomization.

At 6 months, the primary outcome of death, new MI, or stroke had occurred in 9.6% of the patients in the early-intervention group and in 11.3% of those in the delayed-intervention group, but the difference was not statistically significant. However, the difference in the rate of a secondary end point, death, MI, or refractory ischemia, was statistically significant: 9.5% vs 12.9%, P = .003, owing mainly to less refractory ischemia with early intervention.

The patients were also stratified into two groups by baseline risk. The rate of the primary outcome was significantly lower with early intervention in high-risk patients, but not in those at intermediate or low risk. Thus, early intervention may be beneficial in patients at high risk, such as those with ongoing chest pain, but not necessarily in those at low risk.

LEAVE NO LESION BEHIND?

Coronary artery disease often affects more than one segment. Until recently, it was not known whether we should stent all stenotic segments in patients presenting with non-ST-elevation MI or unstable angina, or only the “culprit lesion.”

Shishehbor et al⁹ examined data from a Cleveland Clinic registry of 1,240 patients with acute coronary syndrome and multivessel coronary artery disease who underwent bare-metal stenting. The median follow-up was 2.3 years. Using a propensity model to match patients in the two groups with similar baseline characteristics, they found that the rate of repeat revascularization was less with multivessel intervention than with culprit-only stenting, as was the rate of the combined end point of death, MI, or revascularization, but not that of all-cause mortality or the composite of death or MI.

BARE-METAL VS DRUG-ELUTING STENTS: BALANCING THE RISKS AND BENEFITS

After a patient receives a stent, two bad things can happen: the artery can close up again either gradually, in a process called restenosis, or suddenly, via thrombosis.

Drug-eluting stents were invented to solve the problem of restenosis, and they work very well. Stone et al¹⁰ pooled the data from four double-blind trials of sirolimus (Rapamune) stents and five double-blind trials of paclitaxel (Taxol) stents and found that, at 4 years, the rates of target-lesion revascularization (for restenosis) were 7.8% with sirolimus stents vs 23.6% with bare-metal stents (P < .001), and 10.1% with paclitaxel stents vs 20.0% with bare-metal stents (P < .001).

Thrombosis was much less common in these studies, occurring in 1.2% of the sirolimus stent groups vs 0.6% of the bare-metal stent groups (P = .20), and in 1.3% of the paclitaxel stent groups vs 0.9% of the bare-metal stent groups (P = .30).¹⁰

However, drug-eluting stents appear to increase the risk of thrombosis later on, ie, after 1 year. Bavry et al,¹¹ in a meta-analysis, calculated that when stent thrombosis occurred, the median time after implantation was 15.5 months with sirolimus stents vs 4 months with bare-metal stents (P = .0052), and 18 months with paclitaxel stents vs 3.5 months with bare-metal stents (P = .04). The absolute risk of very late stent thrombosis after 1 year was very low, with five events per 1,000 patients with drug-eluting stents vs no events with bare-metal stents (P = .02). Nevertheless, this finding has practical implications. How long must patients continue dual antiplatelet therapy? And what if a patient needs surgery a year later?

Restenosis is not always so gradual

Although stent thrombosis is serious and often fatal, bare-metal stent restenosis is not always benign either, despite the classic view that stent restenosis is a gradual process that results in exertional angina. Reviewing 1,186 cases of bare-metal stent restenosis in 984 patients at Cleveland Clinic, Chen et al¹² reported that 9.5% of cases presented as acute MI (2.2% as ST-elevation MI and 7.3% as non-ST-elevation MI), and 26.4% as unstable angina requiring hospitalization.

A Mayo Clinic study¹³ corroborated these findings. The 10-year incidence of clinical bare-metal stent restenosis was 18.1%, and the incidence of MI was 2.1%. The 10-year rate of bare-metal stent thrombosis was 2%. Off-label use, primarily in saphenous vein grafts, increased the incidence; other correlates were prior MI, peripheral arterial disease, and ulcerated lesions.

Furthermore, bare-metal stent thrombosis can also occur later. We saw a case that occurred 13 years after the procedure, 3 days after the patient stopped taking aspirin because he was experiencing flu-like symptoms, ran out of aspirin, and felt too sick to go out and buy more. The presentation was with ST-elevation MI. The patient recovered after treatment with intracoronary abciximab (ReoPro), percutaneous thrombectomy, balloon angioplasty, and, eventually, bypass surgery.¹⁴

No difference in risk of death with drug-eluting vs bare-metal stents

Even though drug-eluting stents pose a slightly higher risk of thrombosis than bare-metal stents, the risk of death is no higher.¹⁵

I believe the reason is that there are competing risks, and that the higher risk of thrombosis with first-generation drug-eluting stents and the higher risk of restenosis with bare-metal stents essentially cancel each other out. For most patients, there is an absolute benefit with drug-eluting stents, which reduce the need for revascularization with no effect in terms of either increasing or decreasing the risk of MI or death. Second-generation drug-eluting stents may have advantages in reducing rates of death or MI compared with first-generation drug-eluting stents, though this remains to be proven conclusively.

The right revascularization for the right patient

Bavry and I¹⁶ developed an algorithm for deciding on revascularization, posing a series of questions:

• Does the patient need any form of revascularization?
• Is he or she at higher risk of both stent thrombosis and restenosis, as in patients with diabetes, diffuse multivessel disease with bifurcation lesions, or chronic total occlusions? If so, coronary artery bypass grafting remains an excellent option.
• Does he or she have a low risk of restenosis, as in patients without diabetes with focal lesions in large vessels? If so, one could consider a bare-metal stent, which would probably be more cost-effective than a drug-eluting stent in this situation.
• Does the patient have relative contraindications to drug-eluting stents? Examples are a history of noncompliance with medical therapy, financial issues such as lack of insurance that would make buying clopidogrel (Plavix) a problem, long-term anticoagulation, or anticipated need for surgery in the next few years.

If a drug-eluting stent is used, certain measures can help ensure that it is used optimally. It should often be placed under high pressure with a noncompliant balloon so that it achieves contact with the artery wall all around. One should consider intravascular ultrasonographic guidance to make sure the stent is well opposed if it is in a very calcified lesion. Dual antiplatelet therapy with clopidogrel and aspirin should be given for at least 1 year, and if there is no bleeding, perhaps longer, pending further data.¹⁶

LEAVE NO PLATELET ACTIVATED?

Platelets have several types of receptors that, when bound by their respective ligands, lead to platelet activation and aggregation and, ultimately, thrombus formation. Antagonists to some of these receptors are available or are being developed.¹⁷

For long-term therapy, blocking the process “upstream,” ie, preventing platelet activation, is better than blocking it “downstream,” ie, preventing aggregation. For example, clopidogrel, ticlopipine (Ticlid), and prasugrel (Effient) have active metabolites that bind to a subtype of the adenosine diphosphate receptor and prevent platelet activation, whereas the glycoprotein IIb/IIIa inhibitors such as abciximab work downstream, binding to a different receptor and preventing aggregation.¹⁸

The evidence for using dual antiplatelet therapy (ie, aspirin plus clopidogrel) in patients with acute coronary syndromes without ST-elevation is very well established.

The Clopidogrel in Unstable Angina to Prevent Recurrent Events (CURE) trial,¹⁹ published in 2001, found a 20% relative risk reduction and a 2% absolute risk reduction in the incidence of MI, stroke, or cardiovascular death in patients randomly assigned to receive clopidogrel plus aspirin for 1 year vs aspirin alone for 1 year (P < .001). In the subgroup of patients who underwent percutaneous coronary intervention, the relative risk reduction in the incidence of MI or cardiovascular death at 1 year of follow-up was 31% (P = .002).²⁰

As a result of these findings, the cardiology society guidelines²¹ recommend a year of dual antiplatelet therapy after acute coronary syndromes, regardless of whether the patient is treated medically, percutaneously, or surgically.

But what happens after clopidogrel is withdrawn? Ho et al²² retrospectively analyzed data from Veterans Affairs hospitals and found a spike in the incidence of death or MI in the first 90 days after stopping clopidogrel treatment. This was true in medically treated patients as well as in those treated with percutaneous coronary interventions, in those with or without diabetes mellitus, in those who received a drug-eluting stent or a bare-metal stent, and in those treated longer than 9 months.

The investigators concluded that there might be a “clopidogrel rebound effect.” However, I believe that a true rebound effect, such as after withdrawal of heparin or warfarin, is biologically unlikely with clopidogrel, since clopidogrel irreversibly binds to its receptor for the 7- to 10-day life span of the platelet. Rather, I believe the phenomenon must be due to withdrawal of protection in patients at risk.

In stable patients, dual therapy is not as beneficial

Would dual antiplatelet therapy with clopidogrel and aspirin also benefit patients at risk of atherothrombotic events but without acute coronary syndromes?

The Clopidogrel for High Atherothrombotic Risk and Ischemic Stabilization, Management, and Avoidance (CHARISMA) trial²³ included 15,603 patients with either clinically evident but stable cardiovascular disease or multiple risk factors for athero-thrombosis. They were randomly assigned to receive either clopidogrel 75 mg/day plus aspirin 75 to 162 mg/day or placebo plus aspirin. At a median of 28 months, the groups did not differ significantly in the rate of MI, stroke, or death from cardiovascular causes.

However, the subgroup of patients who had documented prior MI, ischemic stroke, or symptomatic peripheral arterial disease did appear to derive significant benefit from dual therapy.²⁴ In this subgroup, the rate of MI, stroke, or cardiovascular death at a median follow-up of 27.6 months was 8.8% with placebo plus aspirin compared with 7.3% with clopidogrel plus aspirin, for a hazard ratio of 0.83 (95% CI 0.72–0.96, P = .01). Unstented patients with stable coronary artery disease but without prior MI derived no benefit.

Bleeding and thrombosis: The Scylla and Charybdis of antiplatelet therapy

However, with dual antiplatelet therapy, we steer between the Scylla of bleeding and the Charybdis of thrombosis.²⁵

In the CHARISMA subgroup who had prior MI, ischemic stroke, or symptomatic peripheral arterial disease, the incidence of moderate or severe bleeding was higher with dual therapy than with aspirin alone, but the rates converged after about 1 year of treatment.²⁴ Further, there was no difference in fatal bleeding or intracranial bleeding, although the rate of moderate bleeding (defined as the need for transfusion) was higher with dual therapy (2.0% vs 1.3%, P = .004).

I believe the data indicate that if a patient can tolerate dual antiplatelet therapy for 9 to 12 months without any bleeding issues, he or she is unlikely to have a major bleeding episode if dual therapy is continued beyond this time.

About half of bleeding events in patients on chronic antiplatelet therapy are gastrointestinal. To address this risk, in 2008 an expert committee from the American College of Cardiology, American College of Gastroenterology, and American Heart Association issued a consensus document²⁶ in which they recommended assessing gastrointestinal risk factors in patients on antiplatelet therapy, such as history of ulcers (and testing for and treating Helicobacter pylori infection if present), history of gastrointestinal bleeding, concomitant anticoagulant therapy, and dual antiplatelet therapy. If any of these were present, the committee recommended considering a proton pump inhibitor. The committee also recommended a proton pump inhibitor for patients on antiplatelet therapy who have more than one of the following: age 60 years or more, corticosteroid use, or dyspepsia or gastroesophageal reflux symptoms.

Some ex vivo platelet studies and observational analyses have suggested that there might be an adverse interaction between clopidogrel and proton pump inhibitors due to a blunting of clopidogrel’s antiplatelet effect. A large randomized clinical trial was designed and launched to determine if a single-pill combination of the proton pump inhibitor omeprazole (Prilosec) and clopidogrel would be safer than clopidogrel alone when added to aspirin. Called COGENT-1 (Clopidogrel and the Optimization of GI Events Trial), it was halted early in 2009 when it lost its funding. However, preliminary data did not show an adverse interaction between clopidogrel and omeprazole.

What is the right dose of aspirin?

Steinhubl et al²⁷ performed a post hoc observational analysis of data from the CHARISMA trial. Their findings suggested that higher doses of aspirin are not more effective than lower doses for chronic therapy. Furthermore, in the group receiving clopidogrel plus aspirin, the incidence of severe or life-threatening bleeding was significantly greater with aspirin doses higher than 100 mg than with doses lower than 100 mg, 2.6% vs 1.7%, P = .040.

A randomized, controlled trial called Clopidogrel Optimal Loading Dose Usage to Reduce Recurrent Events/Optimal Antiplatelet Strategy for Interventions (CURRENT/OASIS 7)²⁸ recently reported that higher-dose aspirin (ie, 325 mg) may be better than lower dose aspirin (ie, 81 mg) in patients with acute coronary syndromes undergoing percutaneous coronary intervention and receiving clopidogrel. During this 30-day study, there was no increase in overall bleeding with the higher dose of aspirin, though gastrointestinal bleeding was slightly increased.²⁹ In a factorial design, the second part of this trial found that a higher-dose clopidogrel regimen reduced stent thrombosis.²⁹

In vitro, response to clopidogrel shows a normal bell-shaped distribution.³⁰ In theory, therefore, patients who are hyperresponders may be at higher risk of bleeding, and those who are hyporesponders may be at risk of ischemic events.

A clinical trial is under way to examine whether hyporesponders should get higher doses. Called GRAVITAS (Gauging Responsiveness With a VerifyNow Assay Impact on Thrombosis and Safety), it will use a point-of-care platelet assay and then allocate patients to receive either standard therapy or double the dose of clopidogrel. The primary end point will be the rate of cardiovascular death, nonfatal MI, or stent thrombosis at 6 months.

Is prasugrel better than clopidogrel?

Prasugrel (Effient) is a new drug of the same class as clopidogrel, ie, a thienopyridine, with its active metabolite binding to the same platelet receptor as clopidogrel and inhibiting platelet aggregation more rapidly, more consistently, and to a greater extent than clopidogrel. Prasugrel was recently approved by the Food and Drug Administration. But is it better?³¹

The Trial to Assess Improvement in Therapeutic Outcomes by Optimizing Platelet Inhibition With Prasugrel–Thrombolysis in Myocardial Infarction (TRITON-TIMI 38) compared prasugrel and clopidogrel in 13,608 patients with moderate- to high-risk acute coronary syndromes who were scheduled to undergo percutaneous coronary intervention.³²

Overall, prasugrel was better. At 15 months, the incidence of the primary end point (death from cardiovascular causes, nonfatal MI, or nonfatal stroke) was significantly lower with prasugrel therapy than with clopidogrel in the entire cohort (9.9% vs 12.1%, hazard ratio 0.81, 95% CI 0.73–0.90, P < .001), in the subgroup with ST-segment elevation MI, and in the subgroup with unstable angina or non-ST-elevation MI.

However, there was a price to pay. The rate of major bleeding was higher with prasugrel (2.4% vs 1.8%, hazard ratio 1.32, 95% CI 1.03–1.68, P = .03). Assessing the balance between the risk and the benefit, the investigators identified three subgroups who did not derive a net clinical benefit from prasugrel: patients who had had a previous stroke or transient ischemic attack (this group actually had a net harm from prasugrel), patients 75 years of age or older, and patients weighing less than 60 kg (132 pounds).

More work is needed to determine which patients are best served by standard-dose clopidogrel, higher doses of clopidogrel, platelet-assay-guided dosing of clopidogrel, or prasugrel.²⁴

Short-acting, potent intravenous platelet blockade with an agent such as cangrelor is theoretically appealing, but further research is necessary.^33,34 Ticagrelor, a reversible adenosine diphosphate receptor antagonist, provides yet another potential option in antiplatelet therapy for acute coronary syndromes. In the recent PLATO trial (Study of Platelet Inhibition and Patient Outcomes), compared with clopidogrel, ticagrelor reduced the risk of ischemic events, including death.^35,36 Here, too, there was more major bleeding (unrelated to coronary artery bypass grafting) with ticagrelor.

Thus, clinical assessment of an individual patient’s ischemic and bleeding risks will continue to be critical as therapeutic strategies evolve.

Hypertension is difficult to diagnose, and its treatment is difficult to monitor optimally on the basis of traditional office blood pressure measurements. To better protect our patients from the effects of undiagnosed or poorly controlled hypertension, we need to consider other options, such as ambulatory 24-hour blood pressure monitoring, automated measurement in the office, measurement in the patient’s home, and devices that analyze the peripheral pulse wave to estimate the central blood pressure and other indices of arterial stiffness.

MANUAL OFFICE MEASUREMENT HAS INHERENT LIMITATIONS

Office blood pressure measurements do provide enormous information about cardiovascular risk and the risk of death, as shown in epidemiologic studies. A meta-analysis¹ of 61 prospective observational studies that included more than 1 million patients showed that office blood pressure levels clearly correlate with increased risk of death from cardiovascular disease and stroke.

But blood pressure is a dynamic measure with inherent minute-to-minute variability, and measurement will not be accurate if the correct technique is not followed. Traditional office sphygmomanometry is a snapshot and does not accurately reflect a patient’s blood pressure in the real world and in real time.

Recently, unique patterns of blood pressure have been identified that may not be detected in the physician’s office. It is clear from several clinical trials that some patients’ blood pressure is transiently elevated in the first few minutes during office measurements (the “white coat effect”). In addition, when office measurements are compared with out-of-office measurements, several patterns of hypertension emerge that have prognostic value. These patterns are white coat hypertension, masked hypertension, nocturnal hypertension, and failure of the blood pressure to dip during sleep.

WHITE COAT EFFECT

The white coat effect is described as a transient elevation in office blood pressure caused by an alerting reaction when the pressure is measured by a physician or a nurse. It may last for several minutes. The magnitude of blood pressure elevation has been noted to be higher when measured by a physician than when measured by a nurse. Multiple blood pressure measurements taken over 5 to 10 minutes help eliminate the white coat effect. In a recent study,² 36% of patients with hypertension demonstrated the white coat effect.

In a study by Mancia et al,³ 46 patients underwent intra-arterial blood pressure monitoring for 2 days, during which time a physician or a nurse would check their blood pressure repeatedly over 10 minutes. This study found that most patients demonstrated the white coat effect: the blood pressure was higher in the first few measurements, but came down after 5 minutes. The white coat effect was as much as 22.6 ± 1.8 mm Hg when blood pressure was measured by a physician and was lower when measured by a nurse.

WHITE COAT HYPERTENSION

In contrast to the white coat effect, which is transient, white coat hypertension is defined as persistent elevation of office blood pressure measurements with normal blood pressure levels when measured outside the physician’s office. Depending on the population sampled, the prevalence of white coat hypertension ranges from 12% to 20%, but this is understandably difficult or almost impossible to detect with traditional office blood pressure measurements alone.^4–7

MASKED HYPERTENSION

Patients with normal blood pressure in the physician’s office but high blood pressure during daily life were found to have a higher risk of cardiovascular events. This condition is called masked hypertension.⁸ For clinicians, the danger lies in underestimating the patient’s risk of cardiovascular events and, thus, undertreating the hypertension. Preliminary data on masked hypertension show that the rates of end-organ damage and cardiovascular events are slightly higher in patients with masked hypertension than in patients with sustained hypertension.

NOCTURNAL HYPERTENSION

Elevated nighttime blood pressure (>125/75 mm Hg) is considered nocturnal hypertension and is generally considered a subgroup of masked hypertension.⁹

In the African American Study of Kidney Disease and Hypertension (AASK),^10,11 although most patients achieved their blood pressure goal during the trial, they were noted to have relentless progression of renal disease. On ambulatory 24-hour blood pressure monitoring during the cohort phase of the study,¹⁰ a high prevalence of elevated nighttime blood pressure (66%) was found. Further analysis showed that the elevated nighttime blood pressure was associated with worse hypertension-related end-organ damage. It is still unclear if lowering nighttime blood pressure improves clinical outcomes in this high-risk population.

DIPPING VS NONDIPPING

The mean blood pressure during sleep should normally decrease by 10% to 20% compared with daytime readings. “Nondipping,” ie, the lack of this nocturnal dip in blood pressure, carries a higher risk of death from cardiovascular causes, even if the person is otherwise normotensive.^12,13 Nondipping is commonly noted in African Americans, patients with diabetes, and those with chronic kidney disease.

A study by Lurbe et al¹⁴ of patients with type 1 diabetes mellitus who underwent ambulatory 24-hour blood pressure monitoring found that the onset of the nondipping phenomenon preceded microalbuminuria (a risk factor for kidney disease). Data from our institution¹⁵ showed that nondipping was associated with a greater decline in glomerular filtration rate when compared with dipping.

The lack of reproducibility of a person’s dipping status has been a barrier in relying on this as a prognostic measure. White and Larocca¹⁶ found that only about half of the patients who appeared to be nondippers on one 24-hour recording still were nondippers on a second recording 4 to 8 weeks later. Compared with nondipping, nocturnal hypertension is a more stable blood pressure pattern that is being increasingly recognized in patients undergoing 24-hour blood pressure monitoring.

AUTOMATIC BLOOD PRESSURE DEVICES

An automated in-office blood pressure measurement device is one way to minimize the white coat effect and obtain a more accurate blood pressure assessment. Devices such as BpTRU (BpTRU Medical Devices Ltd, Coquitlam, BC, Canada) are programmed to take a series of automatic, oscillometric readings at regular intervals while the patient is left alone in a quiet room. BpTRU has been validated in several clinical trials and has been shown to overcome the white coat effect to some extent. Myers et al¹⁷ compared 24-hour blood pressure readings with those obtained by a family physician, by a research technician, and by the BpTRU device and found that the BpTRU readings were much closer to the average of awake blood pressure readings on 24-hour blood pressure monitoring.

AMBULATORY 24-HOUR BLOOD PRESSURE MONITORING

• Presence or absence of the nocturnal dip (the normal 10% to 20% drop in blood pressure at night during sleep)
• Morning surge (which in some studies was associated with higher incidence of stroke)
• Supine hypertension and sudden fluctuations in blood pressure seen in patients with autonomic failure.

Studies have shown that basing antihypertensive therapy on ambulatory 24-hour blood pressure monitoring results in better control of hypertension and lowers the rate of cardiovascular events.^18,19

Perloff et al¹⁸ found that in patients whose hypertension was considered well controlled on the basis of office blood pressure measurements, those with higher blood pressures on ambulatory 24-hour monitoring had higher cardiovascular morbidity and mortality rates.

More recently, Clement et al¹⁹ showed that patients being treated for hypertension who have higher average ambulatory 24-hour blood pressures had a higher risk of cardiovascular events and cardiovascular death.

After following 790 patients for 3.7 years, Verdecchia et al²⁰ concluded that controlling hypertension on the basis of ambulatory 24-hour blood pressure readings rather than traditional office measurements lowered the risk of cardiovascular disease.

‘Normal’ blood pressure on ambulatory 24-hour monitoring

It should be noted that the normal average blood pressure on ambulatory 24-hour monitoring tends to be lower than that on traditional office readings. According to the 2007 European guidelines,²¹ an average 24-hour blood pressure above the range of 125/80 to 130/80 mm Hg is considered diagnostic of hypertension.

The bottom line on ambulatory 24-hour monitoring: Not perfect, but helpful

Ambulatory 24-hour blood pressure monitoring is not perfect. It interferes with the patient’s activities and with sleep, and this can affect the readings. It is also expensive, and Medicare and Medicaid cover it only if the patient is diagnosed with white coat hypertension, based on stringent criteria that include three elevated clinic blood pressure measurements and two normal out-of-clinic blood pressure measurements and no evidence of end-organ damage. Despite these issues, almost all national guidelines for the management of hypertension recommend ambulatory 24-hour blood pressure monitoring to improve cardiovascular risk prediction and to measure the variability in blood pressure levels.

USING THE INTERNET IN MANAGING HYPERTENSION

Green et al²² studied a new model of care using home blood pressure monitoring via the Internet, and provided feedback and intervention to the patient via a pharmacist to achieve blood pressure goals. Patients measured their blood pressure at home on at least 2 days a week (two measurements each time), using an automatic oscillometric monitor (Omron Hem-705-CP, Kyoto, Japan), and entered the results in an electronic medical record on the Internet. In the intervention group, a pharmacist communicated with each patient by either phone or e-mail every 2 weeks, making changes to their antihypertensive regimens as needed.

Patients in the intervention group had an average reduction in blood pressure of 14 mm Hg from baseline, and their blood pressure was much better controlled compared with the control groups, who were being passively monitored or were receiving usual care based on office blood pressure readings.

MEASURING ARTERIAL STIFFNESS TO ASSES RISK OF END-ORGAN DAMAGE

Mean arterial blood pressure, derived from the extremes of systolic and diastolic pressure as measured with a traditional sphygmomanometer, is a product of cardiac output and total peripheral vascular resistance. In contrast, central aortic blood pressure, the central augmentation index, and pulse wave velocity are measures derived from brachial blood pressure as well as arterial pulse wave tracings. They provide additional information on arterial stiffness and help stratify patients at increased cardiovascular risk.

The art of evaluating the arterial pulse wave with the fingertips while examining a patient and diagnosing various ailments was well known and practiced by ancient Greek and Chinese physicians. Although this was less recognized in Western medicine, it was the pulse wave recording on a sphygmograph that was used to measure human blood pressure in the 19th century.²³ In the early 20th century, this art was lost with the invention of the mercury sphygmomanometer.

Arterial stiffness indices—ie, central aortic blood pressure, the central augmentation index, and pulse wave velocity—can now be measured noninvasively and have been shown to correlate very well with measurements obtained via a central arterial catheter. In the past, the only way to measure central blood pressure was directly via a central arterial catheter. New devices now measure arterial stiffness indices indirectly by applanation tonometry and pulse wave analysis (reviewed by O’Rourke et al²⁵).

Several trials have shown that these arterial indices have a better prognostic value than the mean arterial pressure or the brachial pulse pressure. For example, the Baltimore Longitudinal Study of Aging²⁶ followed 100 normotensive individuals for 5 years and found that those with a higher pulse wave velocity had a greater chance of developing incident hypertension. Other studies showed that pulse wave velocity and other indices of arterial stiffness are associated with dysfunction of the microvasculature in the brain, with higher cardiovascular risk, and a higher risk of death.

A major limitation in measuring these arterial stiffness indices is that they are derived values and require measurement of brachial blood pressure in addition to the pulse wave tracing.

Recent hypertension guidelines^21,27,28 released during the past 2 years in Europe, Latin America, and Japan have recommended measurement of arterial stiffness as part of a comprehensive evaluation of patients with hypertension.

EXCITING TIMES IN HYPERTENSION

These are exciting times in the field of hypertension. With advances in technology, we have new devices and techniques that provide a closer view of the hemodynamic changes and blood pressures experienced by vital organs. In addition, we can now go beyond the physician’s office and evaluate blood pressure changes that occur during the course of a usual day in a patient’s life. This enables us to make better decisions in the management of their hypertension, embodying Dr. Harvey Cushing’s teaching that the physician’s obligation is to “view the man in his world.”²⁹

Hypertension is difficult to diagnose, and its treatment is difficult to monitor optimally on the basis of traditional office blood pressure measurements. To better protect our patients from the effects of undiagnosed or poorly controlled hypertension, we need to consider other options, such as ambulatory 24-hour blood pressure monitoring, automated measurement in the office, measurement in the patient’s home, and devices that analyze the peripheral pulse wave to estimate the central blood pressure and other indices of arterial stiffness.

MANUAL OFFICE MEASUREMENT HAS INHERENT LIMITATIONS

Office blood pressure measurements do provide enormous information about cardiovascular risk and the risk of death, as shown in epidemiologic studies. A meta-analysis¹ of 61 prospective observational studies that included more than 1 million patients showed that office blood pressure levels clearly correlate with increased risk of death from cardiovascular disease and stroke.

But blood pressure is a dynamic measure with inherent minute-to-minute variability, and measurement will not be accurate if the correct technique is not followed. Traditional office sphygmomanometry is a snapshot and does not accurately reflect a patient’s blood pressure in the real world and in real time.

Recently, unique patterns of blood pressure have been identified that may not be detected in the physician’s office. It is clear from several clinical trials that some patients’ blood pressure is transiently elevated in the first few minutes during office measurements (the “white coat effect”). In addition, when office measurements are compared with out-of-office measurements, several patterns of hypertension emerge that have prognostic value. These patterns are white coat hypertension, masked hypertension, nocturnal hypertension, and failure of the blood pressure to dip during sleep.

WHITE COAT EFFECT

The white coat effect is described as a transient elevation in office blood pressure caused by an alerting reaction when the pressure is measured by a physician or a nurse. It may last for several minutes. The magnitude of blood pressure elevation has been noted to be higher when measured by a physician than when measured by a nurse. Multiple blood pressure measurements taken over 5 to 10 minutes help eliminate the white coat effect. In a recent study,² 36% of patients with hypertension demonstrated the white coat effect.

In a study by Mancia et al,³ 46 patients underwent intra-arterial blood pressure monitoring for 2 days, during which time a physician or a nurse would check their blood pressure repeatedly over 10 minutes. This study found that most patients demonstrated the white coat effect: the blood pressure was higher in the first few measurements, but came down after 5 minutes. The white coat effect was as much as 22.6 ± 1.8 mm Hg when blood pressure was measured by a physician and was lower when measured by a nurse.

WHITE COAT HYPERTENSION

In contrast to the white coat effect, which is transient, white coat hypertension is defined as persistent elevation of office blood pressure measurements with normal blood pressure levels when measured outside the physician’s office. Depending on the population sampled, the prevalence of white coat hypertension ranges from 12% to 20%, but this is understandably difficult or almost impossible to detect with traditional office blood pressure measurements alone.^4–7

MASKED HYPERTENSION

Patients with normal blood pressure in the physician’s office but high blood pressure during daily life were found to have a higher risk of cardiovascular events. This condition is called masked hypertension.⁸ For clinicians, the danger lies in underestimating the patient’s risk of cardiovascular events and, thus, undertreating the hypertension. Preliminary data on masked hypertension show that the rates of end-organ damage and cardiovascular events are slightly higher in patients with masked hypertension than in patients with sustained hypertension.

NOCTURNAL HYPERTENSION

Elevated nighttime blood pressure (>125/75 mm Hg) is considered nocturnal hypertension and is generally considered a subgroup of masked hypertension.⁹

In the African American Study of Kidney Disease and Hypertension (AASK),^10,11 although most patients achieved their blood pressure goal during the trial, they were noted to have relentless progression of renal disease. On ambulatory 24-hour blood pressure monitoring during the cohort phase of the study,¹⁰ a high prevalence of elevated nighttime blood pressure (66%) was found. Further analysis showed that the elevated nighttime blood pressure was associated with worse hypertension-related end-organ damage. It is still unclear if lowering nighttime blood pressure improves clinical outcomes in this high-risk population.

DIPPING VS NONDIPPING

The mean blood pressure during sleep should normally decrease by 10% to 20% compared with daytime readings. “Nondipping,” ie, the lack of this nocturnal dip in blood pressure, carries a higher risk of death from cardiovascular causes, even if the person is otherwise normotensive.^12,13 Nondipping is commonly noted in African Americans, patients with diabetes, and those with chronic kidney disease.

A study by Lurbe et al¹⁴ of patients with type 1 diabetes mellitus who underwent ambulatory 24-hour blood pressure monitoring found that the onset of the nondipping phenomenon preceded microalbuminuria (a risk factor for kidney disease). Data from our institution¹⁵ showed that nondipping was associated with a greater decline in glomerular filtration rate when compared with dipping.

The lack of reproducibility of a person’s dipping status has been a barrier in relying on this as a prognostic measure. White and Larocca¹⁶ found that only about half of the patients who appeared to be nondippers on one 24-hour recording still were nondippers on a second recording 4 to 8 weeks later. Compared with nondipping, nocturnal hypertension is a more stable blood pressure pattern that is being increasingly recognized in patients undergoing 24-hour blood pressure monitoring.

AUTOMATIC BLOOD PRESSURE DEVICES

An automated in-office blood pressure measurement device is one way to minimize the white coat effect and obtain a more accurate blood pressure assessment. Devices such as BpTRU (BpTRU Medical Devices Ltd, Coquitlam, BC, Canada) are programmed to take a series of automatic, oscillometric readings at regular intervals while the patient is left alone in a quiet room. BpTRU has been validated in several clinical trials and has been shown to overcome the white coat effect to some extent. Myers et al¹⁷ compared 24-hour blood pressure readings with those obtained by a family physician, by a research technician, and by the BpTRU device and found that the BpTRU readings were much closer to the average of awake blood pressure readings on 24-hour blood pressure monitoring.

AMBULATORY 24-HOUR BLOOD PRESSURE MONITORING

• Presence or absence of the nocturnal dip (the normal 10% to 20% drop in blood pressure at night during sleep)
• Morning surge (which in some studies was associated with higher incidence of stroke)
• Supine hypertension and sudden fluctuations in blood pressure seen in patients with autonomic failure.

Studies have shown that basing antihypertensive therapy on ambulatory 24-hour blood pressure monitoring results in better control of hypertension and lowers the rate of cardiovascular events.^18,19

Perloff et al¹⁸ found that in patients whose hypertension was considered well controlled on the basis of office blood pressure measurements, those with higher blood pressures on ambulatory 24-hour monitoring had higher cardiovascular morbidity and mortality rates.

More recently, Clement et al¹⁹ showed that patients being treated for hypertension who have higher average ambulatory 24-hour blood pressures had a higher risk of cardiovascular events and cardiovascular death.

After following 790 patients for 3.7 years, Verdecchia et al²⁰ concluded that controlling hypertension on the basis of ambulatory 24-hour blood pressure readings rather than traditional office measurements lowered the risk of cardiovascular disease.

‘Normal’ blood pressure on ambulatory 24-hour monitoring

It should be noted that the normal average blood pressure on ambulatory 24-hour monitoring tends to be lower than that on traditional office readings. According to the 2007 European guidelines,²¹ an average 24-hour blood pressure above the range of 125/80 to 130/80 mm Hg is considered diagnostic of hypertension.

The bottom line on ambulatory 24-hour monitoring: Not perfect, but helpful

Ambulatory 24-hour blood pressure monitoring is not perfect. It interferes with the patient’s activities and with sleep, and this can affect the readings. It is also expensive, and Medicare and Medicaid cover it only if the patient is diagnosed with white coat hypertension, based on stringent criteria that include three elevated clinic blood pressure measurements and two normal out-of-clinic blood pressure measurements and no evidence of end-organ damage. Despite these issues, almost all national guidelines for the management of hypertension recommend ambulatory 24-hour blood pressure monitoring to improve cardiovascular risk prediction and to measure the variability in blood pressure levels.

USING THE INTERNET IN MANAGING HYPERTENSION

Green et al²² studied a new model of care using home blood pressure monitoring via the Internet, and provided feedback and intervention to the patient via a pharmacist to achieve blood pressure goals. Patients measured their blood pressure at home on at least 2 days a week (two measurements each time), using an automatic oscillometric monitor (Omron Hem-705-CP, Kyoto, Japan), and entered the results in an electronic medical record on the Internet. In the intervention group, a pharmacist communicated with each patient by either phone or e-mail every 2 weeks, making changes to their antihypertensive regimens as needed.

Patients in the intervention group had an average reduction in blood pressure of 14 mm Hg from baseline, and their blood pressure was much better controlled compared with the control groups, who were being passively monitored or were receiving usual care based on office blood pressure readings.

MEASURING ARTERIAL STIFFNESS TO ASSES RISK OF END-ORGAN DAMAGE

Mean arterial blood pressure, derived from the extremes of systolic and diastolic pressure as measured with a traditional sphygmomanometer, is a product of cardiac output and total peripheral vascular resistance. In contrast, central aortic blood pressure, the central augmentation index, and pulse wave velocity are measures derived from brachial blood pressure as well as arterial pulse wave tracings. They provide additional information on arterial stiffness and help stratify patients at increased cardiovascular risk.

The art of evaluating the arterial pulse wave with the fingertips while examining a patient and diagnosing various ailments was well known and practiced by ancient Greek and Chinese physicians. Although this was less recognized in Western medicine, it was the pulse wave recording on a sphygmograph that was used to measure human blood pressure in the 19th century.²³ In the early 20th century, this art was lost with the invention of the mercury sphygmomanometer.

Arterial stiffness indices—ie, central aortic blood pressure, the central augmentation index, and pulse wave velocity—can now be measured noninvasively and have been shown to correlate very well with measurements obtained via a central arterial catheter. In the past, the only way to measure central blood pressure was directly via a central arterial catheter. New devices now measure arterial stiffness indices indirectly by applanation tonometry and pulse wave analysis (reviewed by O’Rourke et al²⁵).

Several trials have shown that these arterial indices have a better prognostic value than the mean arterial pressure or the brachial pulse pressure. For example, the Baltimore Longitudinal Study of Aging²⁶ followed 100 normotensive individuals for 5 years and found that those with a higher pulse wave velocity had a greater chance of developing incident hypertension. Other studies showed that pulse wave velocity and other indices of arterial stiffness are associated with dysfunction of the microvasculature in the brain, with higher cardiovascular risk, and a higher risk of death.

A major limitation in measuring these arterial stiffness indices is that they are derived values and require measurement of brachial blood pressure in addition to the pulse wave tracing.

Recent hypertension guidelines^21,27,28 released during the past 2 years in Europe, Latin America, and Japan have recommended measurement of arterial stiffness as part of a comprehensive evaluation of patients with hypertension.

EXCITING TIMES IN HYPERTENSION

These are exciting times in the field of hypertension. With advances in technology, we have new devices and techniques that provide a closer view of the hemodynamic changes and blood pressures experienced by vital organs. In addition, we can now go beyond the physician’s office and evaluate blood pressure changes that occur during the course of a usual day in a patient’s life. This enables us to make better decisions in the management of their hypertension, embodying Dr. Harvey Cushing’s teaching that the physician’s obligation is to “view the man in his world.”²⁹

Hypertension is difficult to diagnose, and its treatment is difficult to monitor optimally on the basis of traditional office blood pressure measurements. To better protect our patients from the effects of undiagnosed or poorly controlled hypertension, we need to consider other options, such as ambulatory 24-hour blood pressure monitoring, automated measurement in the office, measurement in the patient’s home, and devices that analyze the peripheral pulse wave to estimate the central blood pressure and other indices of arterial stiffness.

MANUAL OFFICE MEASUREMENT HAS INHERENT LIMITATIONS

Office blood pressure measurements do provide enormous information about cardiovascular risk and the risk of death, as shown in epidemiologic studies. A meta-analysis¹ of 61 prospective observational studies that included more than 1 million patients showed that office blood pressure levels clearly correlate with increased risk of death from cardiovascular disease and stroke.

But blood pressure is a dynamic measure with inherent minute-to-minute variability, and measurement will not be accurate if the correct technique is not followed. Traditional office sphygmomanometry is a snapshot and does not accurately reflect a patient’s blood pressure in the real world and in real time.

Recently, unique patterns of blood pressure have been identified that may not be detected in the physician’s office. It is clear from several clinical trials that some patients’ blood pressure is transiently elevated in the first few minutes during office measurements (the “white coat effect”). In addition, when office measurements are compared with out-of-office measurements, several patterns of hypertension emerge that have prognostic value. These patterns are white coat hypertension, masked hypertension, nocturnal hypertension, and failure of the blood pressure to dip during sleep.

WHITE COAT EFFECT

The white coat effect is described as a transient elevation in office blood pressure caused by an alerting reaction when the pressure is measured by a physician or a nurse. It may last for several minutes. The magnitude of blood pressure elevation has been noted to be higher when measured by a physician than when measured by a nurse. Multiple blood pressure measurements taken over 5 to 10 minutes help eliminate the white coat effect. In a recent study,² 36% of patients with hypertension demonstrated the white coat effect.

In a study by Mancia et al,³ 46 patients underwent intra-arterial blood pressure monitoring for 2 days, during which time a physician or a nurse would check their blood pressure repeatedly over 10 minutes. This study found that most patients demonstrated the white coat effect: the blood pressure was higher in the first few measurements, but came down after 5 minutes. The white coat effect was as much as 22.6 ± 1.8 mm Hg when blood pressure was measured by a physician and was lower when measured by a nurse.

WHITE COAT HYPERTENSION

In contrast to the white coat effect, which is transient, white coat hypertension is defined as persistent elevation of office blood pressure measurements with normal blood pressure levels when measured outside the physician’s office. Depending on the population sampled, the prevalence of white coat hypertension ranges from 12% to 20%, but this is understandably difficult or almost impossible to detect with traditional office blood pressure measurements alone.^4–7

MASKED HYPERTENSION

Patients with normal blood pressure in the physician’s office but high blood pressure during daily life were found to have a higher risk of cardiovascular events. This condition is called masked hypertension.⁸ For clinicians, the danger lies in underestimating the patient’s risk of cardiovascular events and, thus, undertreating the hypertension. Preliminary data on masked hypertension show that the rates of end-organ damage and cardiovascular events are slightly higher in patients with masked hypertension than in patients with sustained hypertension.

NOCTURNAL HYPERTENSION

Elevated nighttime blood pressure (>125/75 mm Hg) is considered nocturnal hypertension and is generally considered a subgroup of masked hypertension.⁹

In the African American Study of Kidney Disease and Hypertension (AASK),^10,11 although most patients achieved their blood pressure goal during the trial, they were noted to have relentless progression of renal disease. On ambulatory 24-hour blood pressure monitoring during the cohort phase of the study,¹⁰ a high prevalence of elevated nighttime blood pressure (66%) was found. Further analysis showed that the elevated nighttime blood pressure was associated with worse hypertension-related end-organ damage. It is still unclear if lowering nighttime blood pressure improves clinical outcomes in this high-risk population.

DIPPING VS NONDIPPING

The mean blood pressure during sleep should normally decrease by 10% to 20% compared with daytime readings. “Nondipping,” ie, the lack of this nocturnal dip in blood pressure, carries a higher risk of death from cardiovascular causes, even if the person is otherwise normotensive.^12,13 Nondipping is commonly noted in African Americans, patients with diabetes, and those with chronic kidney disease.

A study by Lurbe et al¹⁴ of patients with type 1 diabetes mellitus who underwent ambulatory 24-hour blood pressure monitoring found that the onset of the nondipping phenomenon preceded microalbuminuria (a risk factor for kidney disease). Data from our institution¹⁵ showed that nondipping was associated with a greater decline in glomerular filtration rate when compared with dipping.

The lack of reproducibility of a person’s dipping status has been a barrier in relying on this as a prognostic measure. White and Larocca¹⁶ found that only about half of the patients who appeared to be nondippers on one 24-hour recording still were nondippers on a second recording 4 to 8 weeks later. Compared with nondipping, nocturnal hypertension is a more stable blood pressure pattern that is being increasingly recognized in patients undergoing 24-hour blood pressure monitoring.

AUTOMATIC BLOOD PRESSURE DEVICES

An automated in-office blood pressure measurement device is one way to minimize the white coat effect and obtain a more accurate blood pressure assessment. Devices such as BpTRU (BpTRU Medical Devices Ltd, Coquitlam, BC, Canada) are programmed to take a series of automatic, oscillometric readings at regular intervals while the patient is left alone in a quiet room. BpTRU has been validated in several clinical trials and has been shown to overcome the white coat effect to some extent. Myers et al¹⁷ compared 24-hour blood pressure readings with those obtained by a family physician, by a research technician, and by the BpTRU device and found that the BpTRU readings were much closer to the average of awake blood pressure readings on 24-hour blood pressure monitoring.

AMBULATORY 24-HOUR BLOOD PRESSURE MONITORING

• Presence or absence of the nocturnal dip (the normal 10% to 20% drop in blood pressure at night during sleep)
• Morning surge (which in some studies was associated with higher incidence of stroke)
• Supine hypertension and sudden fluctuations in blood pressure seen in patients with autonomic failure.

Studies have shown that basing antihypertensive therapy on ambulatory 24-hour blood pressure monitoring results in better control of hypertension and lowers the rate of cardiovascular events.^18,19

Perloff et al¹⁸ found that in patients whose hypertension was considered well controlled on the basis of office blood pressure measurements, those with higher blood pressures on ambulatory 24-hour monitoring had higher cardiovascular morbidity and mortality rates.

More recently, Clement et al¹⁹ showed that patients being treated for hypertension who have higher average ambulatory 24-hour blood pressures had a higher risk of cardiovascular events and cardiovascular death.

After following 790 patients for 3.7 years, Verdecchia et al²⁰ concluded that controlling hypertension on the basis of ambulatory 24-hour blood pressure readings rather than traditional office measurements lowered the risk of cardiovascular disease.

‘Normal’ blood pressure on ambulatory 24-hour monitoring

It should be noted that the normal average blood pressure on ambulatory 24-hour monitoring tends to be lower than that on traditional office readings. According to the 2007 European guidelines,²¹ an average 24-hour blood pressure above the range of 125/80 to 130/80 mm Hg is considered diagnostic of hypertension.

The bottom line on ambulatory 24-hour monitoring: Not perfect, but helpful

Ambulatory 24-hour blood pressure monitoring is not perfect. It interferes with the patient’s activities and with sleep, and this can affect the readings. It is also expensive, and Medicare and Medicaid cover it only if the patient is diagnosed with white coat hypertension, based on stringent criteria that include three elevated clinic blood pressure measurements and two normal out-of-clinic blood pressure measurements and no evidence of end-organ damage. Despite these issues, almost all national guidelines for the management of hypertension recommend ambulatory 24-hour blood pressure monitoring to improve cardiovascular risk prediction and to measure the variability in blood pressure levels.

USING THE INTERNET IN MANAGING HYPERTENSION

Green et al²² studied a new model of care using home blood pressure monitoring via the Internet, and provided feedback and intervention to the patient via a pharmacist to achieve blood pressure goals. Patients measured their blood pressure at home on at least 2 days a week (two measurements each time), using an automatic oscillometric monitor (Omron Hem-705-CP, Kyoto, Japan), and entered the results in an electronic medical record on the Internet. In the intervention group, a pharmacist communicated with each patient by either phone or e-mail every 2 weeks, making changes to their antihypertensive regimens as needed.

Patients in the intervention group had an average reduction in blood pressure of 14 mm Hg from baseline, and their blood pressure was much better controlled compared with the control groups, who were being passively monitored or were receiving usual care based on office blood pressure readings.

MEASURING ARTERIAL STIFFNESS TO ASSES RISK OF END-ORGAN DAMAGE

Mean arterial blood pressure, derived from the extremes of systolic and diastolic pressure as measured with a traditional sphygmomanometer, is a product of cardiac output and total peripheral vascular resistance. In contrast, central aortic blood pressure, the central augmentation index, and pulse wave velocity are measures derived from brachial blood pressure as well as arterial pulse wave tracings. They provide additional information on arterial stiffness and help stratify patients at increased cardiovascular risk.

The art of evaluating the arterial pulse wave with the fingertips while examining a patient and diagnosing various ailments was well known and practiced by ancient Greek and Chinese physicians. Although this was less recognized in Western medicine, it was the pulse wave recording on a sphygmograph that was used to measure human blood pressure in the 19th century.²³ In the early 20th century, this art was lost with the invention of the mercury sphygmomanometer.

Arterial stiffness indices—ie, central aortic blood pressure, the central augmentation index, and pulse wave velocity—can now be measured noninvasively and have been shown to correlate very well with measurements obtained via a central arterial catheter. In the past, the only way to measure central blood pressure was directly via a central arterial catheter. New devices now measure arterial stiffness indices indirectly by applanation tonometry and pulse wave analysis (reviewed by O’Rourke et al²⁵).

Several trials have shown that these arterial indices have a better prognostic value than the mean arterial pressure or the brachial pulse pressure. For example, the Baltimore Longitudinal Study of Aging²⁶ followed 100 normotensive individuals for 5 years and found that those with a higher pulse wave velocity had a greater chance of developing incident hypertension. Other studies showed that pulse wave velocity and other indices of arterial stiffness are associated with dysfunction of the microvasculature in the brain, with higher cardiovascular risk, and a higher risk of death.

A major limitation in measuring these arterial stiffness indices is that they are derived values and require measurement of brachial blood pressure in addition to the pulse wave tracing.

Recent hypertension guidelines^21,27,28 released during the past 2 years in Europe, Latin America, and Japan have recommended measurement of arterial stiffness as part of a comprehensive evaluation of patients with hypertension.

EXCITING TIMES IN HYPERTENSION

These are exciting times in the field of hypertension. With advances in technology, we have new devices and techniques that provide a closer view of the hemodynamic changes and blood pressures experienced by vital organs. In addition, we can now go beyond the physician’s office and evaluate blood pressure changes that occur during the course of a usual day in a patient’s life. This enables us to make better decisions in the management of their hypertension, embodying Dr. Harvey Cushing’s teaching that the physician’s obligation is to “view the man in his world.”²⁹

A 69-year-old ohio man with leukemia was treated in another state in late June. During the car trip back to Ohio, he developed a sore throat, fever, cough, and nasal congestion. He was admitted to Cleveland Clinic with a presumed diagnosis of neutropenic fever; his absolute neutrophil count was 0.4 × 10⁹/L (reference range 1.8–7.7). His chest radiograph was normal. He was treated with empiric broad-spectrum antimicrobials. On his second day in the hospital, he was tested for influenza by a polymerase chain reaction (PCR) test, which was positive for influenza A. He was moved to a private room and started on oseltamivir (Tamiflu) and rimantadine (Flumadine). The patient’s previous roommate subsequently tested positive for influenza A, as did two health care workers working on the ward. All patients on the floor received prophylactic oseltamivir.

The patient’s condition worsened, and he subsequently went into respiratory distress with diffuse pulmonary infiltrates. He was transferred to the intensive care unit, where he was intubated. Influenza A was isolated from a bronchoscopic specimen. He subsequently recovered after a prolonged course and was discharged on hospital day 50. Testing by the Ohio Department of Health confirmed that this was the 2009 pandemic influenza A (H1N1) virus.

THE CHALLENGES WE FACE

We are now in the midst of an influenza pandemic of the 2009 influenza A (H1N1) virus, with pandemic defined as “worldwide sustained community transmission.” The circulation of seasonal and 2009 pandemic influenza A (H1N1) strains will make this flu season both interesting and challenging.

The approaches to vaccination, prophylaxis, and treatment will be more complex. As of this writing (mid-September 2009), it is clear that we will be giving two influenza vaccines this season: a trivalent vaccine for seasonal influenza, and a monovalent vaccine for pandemic H1N1. It appears the monovalent vaccine may require only one dose to provide protective immunity.¹ Fortunately, the vast majority of cases of pandemic H1N1 are relatively mild and uncomplicated. Still, some people are at higher risk of complications, including young patients, pregnant women, and people with immune deficiency or concomitant health conditions that put them at higher risk of flu-associated complications. Thus, clinicians will need to be educated about whom to test, who needs prophylaxis, and who should not be treated.

As our case demonstrates, unsuspected cases of influenza in hospitalized patients or health care workers working with influenza pose the greatest threat for transmission of influenza within the hospital. Adults hospitalized with influenza tend to present late (more than 48 hours after the onset of symptoms) and tend to have prolonged illness.² Ambulatory adults shed virus for 3 to 6 days; virus shedding is more prolonged for hospitalized patients. Antiviral agents started within 4 days of illness enhance viral clearance and are associated with a shorter stay.³ Therefore, we should have a low threshold for testing for influenza and for isolating all suspected cases.

This is also creating a paradigm shift for health care workers, who are notorious for working through an illness. If you are sick, stay home! This applies whether you have pandemic H1N1 or something else.

EPIDEMIOLOGY OF PANDEMIC 2009 INFLUENZA A (H1N1) VIRUS

The location of cases can now be found on Google Maps; the US Centers for Disease Control and Prevention (CDC) provides weekly influenza reports at www.cdc.gov/flu/weekly/fluactivity.htm.

Pandemic H1N1 appeared in the spring of 2009, and cases continued to mount all summer in the United States (when influenza is normally absent) and around the world. In Mexico in March and April 2009, 2,155 cases of pneumonia, 821 hospitalizations, and 100 deaths were reported.⁴

In contrast with seasonal influenza, children and younger adults were hit the hardest in Mexico. The age group 5 through 59 years accounted for 87% of the deaths (usually, they account for about 17%) and 71% of the cases of severe pneumonia (usually, they account for 32%). These observations may be explained in part by the possibility that people who were alive during the 1957 pandemic (which was an H1N1 strain) have some immunity to the new virus. However, the case-fatality rate was highest in people age 65 and older.⁴

As of July 2009, there were more than 43,000 confirmed cases of pandemic H1N1 in the United States, and actual cases probably exceed 1 million, with more than 400 deaths. An underlying risk factor was identified in more than half of the fatal cases.⁵ Ten percent of the women who died were pregnant.

Pandemic H1N1 has several distinctive epidemiologic features:

• The distribution of cases is similar across multiple geographic areas.
• The distribution of cases by age group is markedly different than that of seasonal influenza, with more cases in school children and fewer cases in older adults.
• Fewer cases have been reported in older adults, but this group has the highest case-fatality rate.

2009 PANDEMIC H1N1 IS A MONGREL

There are three types of influenza viruses, designated A, B, and C. Type A undergoes antigenic shift (rapid changes) and antigenic drift (gradual changes) from year to year, and so it is the type associated with pandemics. In contrast, type B undergoes antigenic drift only, and type C is relatively stable.

Influenza virus is subtyped on the basis of surface glycoproteins: 16 hemagglutinins and nine neuraminidases. The circulating subtypes change every year; the current circulating human subtypes are a seasonal subtype of H1N1 that is different than the pandemic H1N1 subtype, and H3N2.

The 2009 pandemic H1N1 is a new virus never seen before in North America.⁶ Genetically, it is a mongrel, coming from three recognized sources (pigs, birds, and humans) which were combined in pigs.⁷ It is similar to subtypes that circulated in the 1920s through the 1940s.

Most influenza in the Western world comes from Asia every fall, and its arrival is probably facilitated by air travel. The spread is usually unidirectional and is unlikely to contribute to long-term viral evolution.⁸ It appears that 2009 H1N1 virus is the predominant strain circulating in the current influenza season in the Southern Hemisphere. Virologic studies indicate that the H1N1 virus strain has remained antigenically stable since it appeared in April 2009. Thus, it appears likely that the strain selected by the United States for vaccine manufacturing will match the currently circulating seasonal and pandemic H1N1 strains.

VACCINATION IS THE FIRST LINE OF DEFENSE

In addition to the trivalent vaccine against seasonal influenza, a monovalent vaccine for pandemic H1N1 virus is being produced. The CDC has indicated that 45 million doses of pandemic influenza vaccine are expected in October 2009, with an average of 20 million doses each week thereafter. It is anticipated that half of these will be in multidose vials, that 20% will be in prefilled syringes for children over 5 years old and for pregnant women, and that 20% will be in the form of live-attenuated influenza vaccine (nasal spray). The inhaled vaccine should not be given to children under 2 years old, to children under 5 years old who have recurrent wheezing, or to anyone with severe asthma. Neither vaccine should be given to people allergic to hen eggs, from which the vaccine is produced.

An ample supply of the seasonal trivalent vaccine should be available. Once the CDC has more information about specific product availability of the pandemic H1N1 vaccine, that vaccine will be distributed. It can be given concurrently with seasonal influenza vaccine.

Several definitions should be kept in mind when discussing vaccination strategies. Supply is the number of vaccine doses available for distribution. Availability is the ability of a person recommended to be vaccinated to do so in a local venue. Prioritization is the recommendation to vaccination venues to selectively use vaccine for certain population groups first. Targeting is the recommendation that immunization programs encourage and promote vaccination for certain population groups.

The Advisory Committee on Immunization Practices and the CDC recommend both seasonal and H1N1 vaccinations for anyone 6 months of age or older who is at risk of becoming ill or of transmitting the viruses to others. Based on a review of epidemiologic data, the recommendation is for targeting the following five groups for H1N1 vaccination: children and young adults aged 6 months through 24 years; pregnant women; health care workers and emergency medical service workers; people ages 25 through 64 years who have certain health conditions (eg, diabetes, heart disease, lung disease); and people who live with or care for children younger than 6 months of age. This represents approximately 159 million people in the United States.

If the estimates for the vaccine supply are met, and if pandemic H1N1 vaccine requires only a single injection, there should be no need for prioritization of vaccine. If the supply of pandemic H1N1 vaccine is inadequate, then those groups who are targeted would also receive the first doses of the pandemic H1N1 vaccine. It should be used only with caution after consideration of potential benefits and risks in people who have had Guillain-Barré syndrome during the previous 6 weeks, in people with altered immunocompetence, or in people with medical conditions predisposing to influenza complications.

A mass vaccination campaign involving two separate flu vaccines can pose challenges in execution and messaging for public health officials and politicians. In 1976, an aggressive vaccination program turned into a disaster, as there was no pandemic and the vaccine was associated with adverse effects such as Guillain-Barré syndrome. The government and the medical profession need to prepare for a vaccine controversy and to communicate and continue to explain the plan to the public. As pointed out in a recent op-ed piece,⁹ we would hope that all expectant women in the fall flu season will get the flu vaccines. We also know that, normally, one in seven pregnancies would be expected to miscarry. The challenge for public health officials and physicians will be to explain to these patients that there may be an association rather than a causal relationship.

In health care workers, the average vaccination rate is only 37%. We should be doing much better. Cleveland Clinic previously increased the rate of vaccination among its employees via a program in which all workers must either be vaccinated or formally declare (on an internal Web site) that they decline to be vaccinated.¹⁰ This season, even more resources are being directed at decreasing the barriers to flu vaccinations for our health care workers with the support from hospital leadership.

INFECTION CONTROL IN THE HOSPITAL AND IN THE COMMUNITY

Influenza is very contagious and is spread in droplets via sneezing and coughing (within a 3-foot radius), or via unwashed hands—thus the infection-control campaigns urging you to cover your cough and wash your hands.

As noted, for patients being admitted or transferred to the hospital, we need to have a low threshold for testing for influenza and for isolating patients suspected of having influenza. For patients with suspected or proven seasonal influenza, transmission precautions are those recommended by the CDC for droplet precautions (www.cdc.gov/ncidod/dhqp/gl_isolation_droplet.html). A face mask is deemed adequate to protect transmission when coming within 3 feet of an infected person. CDC guidelines for pandemic H1N1 recommends airborne-transmission-based precautions for health care workers who are in close contact with patients with proven or possible H1N1 (www.cdc.gov/ncidod/dhqp/gl_isolation_airborne.html). This recommendation implies the use of fit-tested N95 respirators and negative air pressure rooms (if available).

The recent Institute of Medicine report, Respiratory Protection for Healthcare Workers in the Workplace Against Novel H1N1 Influenza A (www.iom.edu/CMS/3740/71769/72967/72970.aspx) endorses the current CDC guidelines and recommends following these guidelines until we have evidence that other forms of protection or guidelines are equally or more effective.

Personally, I am against this requirement because it creates a terrible administrative burden with no proven benefit. Requiring a respirator means requiring fit-testing, and this will negatively affect our ability to deliver patient care. Recent studies have shown that surgical masks may not be as effective¹¹ but are probably sufficient. Lim et al¹² reported that 79 (37%) of 212 workers who responded to a survey experienced headaches while wearing N95 masks. This remains a controversial issue.

Besides getting the flu shot, what can one do to avoid getting influenza or transmitting to others?

• Cover your cough (cough etiquette) and sneeze.
• Practice good hand hygiene.
• Avoid close contact with people who are sick.
• Do not go to school or work if sick.

A recent study of influenza in households suggested that having the person with flu and household contacts wear face masks and practice hand hygiene within the first 36 hours decreased transmission of flu within the household.¹³

The United States does have a national influenza pandemic plan that outlines specific roles in the event of a pandemic, and I urge you to peruse it at www.hhs.gov/pandemicflu/plan/.

RECOGNIZING AND DIAGNOSING INFLUENZA

The familiar signs and symptoms of influenza—fever, cough, muscle aches, and headache—are nonspecific. Call et al¹⁴ analyzed the diagnostic accuracy of symptoms and signs of influenza and found that fever and cough during an epidemic suggest but do not confirm influenza, and that sneezing in those over age 60 argues against influenza. They concluded that signs and symptoms can tell us whether a patient has an influenza-like illness, but do not confirm or exclude the diagnosis of influenza: “Clinicians need to consider whether influenza is circulating in their communities, and then either treat patients with influenza-like illness empirically or obtain a rapid influenza test.”¹⁴

The signs and symptoms of pandemic 2009 H1N1 are the same as for seasonal flu, except that about 25% of patients with pandemic flu develop gastrointestinal symptoms. It has not been more virulent than seasonal influenza to date.

Should you order a test for influenza?

Most people with influenza are neither tested nor treated. Before ordering a test for influenza, ask, “Does this patient actually have influenza?” Patients diagnosed with “influenza” may have a range of infectious and noninfectious causes, such as vasculitis, endocarditis, or any other condition that can cause a fever and cough.

If I truly suspect influenza, I would still only order a test if the results would change how I manage the patient—for example, a patient being admitted to the hospital where isolation would be required.

Pandemic H1N1 will be detected only as influenza A in our current PCR screen for human influenza. The test does not differentiate between seasonal strains of influenza A (which is resistant to oseltamivir) and pandemic H1N1 (which is susceptible to oseltamivir). This means if you intend to treat, you will have to address further complexity.

Testing for influenza

The clinician should be familiar with the types of tests available. Each test has advantages and disadvantages¹⁵:

Rapid antigen assay is a point-of-care test that can give results in 15 minutes but unfortunately is only 20% to 30% sensitive, so a negative result does not exclude the diagnosis. The positive predictive value is high, meaning a positive test means the patient does have the flu.

Direct fluorescent antibody testing takes about 2.5 hours to complete and requires special training for technicians. It has a sensitivity of 47%, a positive predictive value of 95%, and a negative predictive value of 92%.

PCR testing takes about 6 hours and has a sensitivity of 98%, a positive predictive value of 100%, and a negative predictive value of 98%. This is probably the best test, in view of its all-around performance, but it is not a point-of-care test.

Culture takes 2 to 3 days, has a sensitivity of 89%, a positive predictive value of 100%, and a negative predictive value of 88%.

These tests can determine that the patient has influenza A, but a confirmatory test is always required to confirm pandemic H1N1. This confirmatory testing can be done by the CDC, by state public health laboratories, and by commercial reference laboratories.

ANTIVIRAL TREATMENT

Since influenza test results do not specify whether the patient has seasonal or pandemic influenza, treatment decisions are a sticky wicket. Most patients with pandemic H1N1 do not need to be tested or treated.

Several drugs are approved for treating influenza and shorten the duration of symptoms by about 1 day. The earlier the treatment is started, the better: the time of antiviral initiation affects influenza viral load and the duration of viral shedding.³

The neuraminidase inhibitors oseltamivir and zanamivir (Relenza) block release of virus from the cell. Resistance to oseltamivir is emerging in seasonal influenza A, while most pandemic H1N1 strains are susceptible.

Oseltamivir resistance in pandemic H1N1

A total of 11 cases of oseltamivir-resistant pandemic H1N1 have been confirmed worldwide, including 3 in the United States (2 in immunosuppressed patients in Seattle, WA). Ten of the 11 cases occurred with oseltamivir exposure. All involved a histidine-to-tyrosine substitution at position 275 (H275Y) of the neuraminidase gene. Most were susceptible to zanamivir.

Supplies of oseltamivir and zanamivir are limited, so they should be used only in those who will benefit the most, ie, those at higher risk of influenza complications. These include children under 5 years old, adults age 65 and older, children and adolescents on long-term aspirin therapy, pregnant women, patients who have chronic conditions or who are immunosuppressed, and residents of long-term care facilities.

A 69-year-old ohio man with leukemia was treated in another state in late June. During the car trip back to Ohio, he developed a sore throat, fever, cough, and nasal congestion. He was admitted to Cleveland Clinic with a presumed diagnosis of neutropenic fever; his absolute neutrophil count was 0.4 × 10⁹/L (reference range 1.8–7.7). His chest radiograph was normal. He was treated with empiric broad-spectrum antimicrobials. On his second day in the hospital, he was tested for influenza by a polymerase chain reaction (PCR) test, which was positive for influenza A. He was moved to a private room and started on oseltamivir (Tamiflu) and rimantadine (Flumadine). The patient’s previous roommate subsequently tested positive for influenza A, as did two health care workers working on the ward. All patients on the floor received prophylactic oseltamivir.

The patient’s condition worsened, and he subsequently went into respiratory distress with diffuse pulmonary infiltrates. He was transferred to the intensive care unit, where he was intubated. Influenza A was isolated from a bronchoscopic specimen. He subsequently recovered after a prolonged course and was discharged on hospital day 50. Testing by the Ohio Department of Health confirmed that this was the 2009 pandemic influenza A (H1N1) virus.

THE CHALLENGES WE FACE

We are now in the midst of an influenza pandemic of the 2009 influenza A (H1N1) virus, with pandemic defined as “worldwide sustained community transmission.” The circulation of seasonal and 2009 pandemic influenza A (H1N1) strains will make this flu season both interesting and challenging.

The approaches to vaccination, prophylaxis, and treatment will be more complex. As of this writing (mid-September 2009), it is clear that we will be giving two influenza vaccines this season: a trivalent vaccine for seasonal influenza, and a monovalent vaccine for pandemic H1N1. It appears the monovalent vaccine may require only one dose to provide protective immunity.¹ Fortunately, the vast majority of cases of pandemic H1N1 are relatively mild and uncomplicated. Still, some people are at higher risk of complications, including young patients, pregnant women, and people with immune deficiency or concomitant health conditions that put them at higher risk of flu-associated complications. Thus, clinicians will need to be educated about whom to test, who needs prophylaxis, and who should not be treated.

As our case demonstrates, unsuspected cases of influenza in hospitalized patients or health care workers working with influenza pose the greatest threat for transmission of influenza within the hospital. Adults hospitalized with influenza tend to present late (more than 48 hours after the onset of symptoms) and tend to have prolonged illness.² Ambulatory adults shed virus for 3 to 6 days; virus shedding is more prolonged for hospitalized patients. Antiviral agents started within 4 days of illness enhance viral clearance and are associated with a shorter stay.³ Therefore, we should have a low threshold for testing for influenza and for isolating all suspected cases.

This is also creating a paradigm shift for health care workers, who are notorious for working through an illness. If you are sick, stay home! This applies whether you have pandemic H1N1 or something else.

EPIDEMIOLOGY OF PANDEMIC 2009 INFLUENZA A (H1N1) VIRUS

The location of cases can now be found on Google Maps; the US Centers for Disease Control and Prevention (CDC) provides weekly influenza reports at www.cdc.gov/flu/weekly/fluactivity.htm.

Pandemic H1N1 appeared in the spring of 2009, and cases continued to mount all summer in the United States (when influenza is normally absent) and around the world. In Mexico in March and April 2009, 2,155 cases of pneumonia, 821 hospitalizations, and 100 deaths were reported.⁴

In contrast with seasonal influenza, children and younger adults were hit the hardest in Mexico. The age group 5 through 59 years accounted for 87% of the deaths (usually, they account for about 17%) and 71% of the cases of severe pneumonia (usually, they account for 32%). These observations may be explained in part by the possibility that people who were alive during the 1957 pandemic (which was an H1N1 strain) have some immunity to the new virus. However, the case-fatality rate was highest in people age 65 and older.⁴

As of July 2009, there were more than 43,000 confirmed cases of pandemic H1N1 in the United States, and actual cases probably exceed 1 million, with more than 400 deaths. An underlying risk factor was identified in more than half of the fatal cases.⁵ Ten percent of the women who died were pregnant.

Pandemic H1N1 has several distinctive epidemiologic features:

• The distribution of cases is similar across multiple geographic areas.
• The distribution of cases by age group is markedly different than that of seasonal influenza, with more cases in school children and fewer cases in older adults.
• Fewer cases have been reported in older adults, but this group has the highest case-fatality rate.

2009 PANDEMIC H1N1 IS A MONGREL

There are three types of influenza viruses, designated A, B, and C. Type A undergoes antigenic shift (rapid changes) and antigenic drift (gradual changes) from year to year, and so it is the type associated with pandemics. In contrast, type B undergoes antigenic drift only, and type C is relatively stable.

Influenza virus is subtyped on the basis of surface glycoproteins: 16 hemagglutinins and nine neuraminidases. The circulating subtypes change every year; the current circulating human subtypes are a seasonal subtype of H1N1 that is different than the pandemic H1N1 subtype, and H3N2.

The 2009 pandemic H1N1 is a new virus never seen before in North America.⁶ Genetically, it is a mongrel, coming from three recognized sources (pigs, birds, and humans) which were combined in pigs.⁷ It is similar to subtypes that circulated in the 1920s through the 1940s.

Most influenza in the Western world comes from Asia every fall, and its arrival is probably facilitated by air travel. The spread is usually unidirectional and is unlikely to contribute to long-term viral evolution.⁸ It appears that 2009 H1N1 virus is the predominant strain circulating in the current influenza season in the Southern Hemisphere. Virologic studies indicate that the H1N1 virus strain has remained antigenically stable since it appeared in April 2009. Thus, it appears likely that the strain selected by the United States for vaccine manufacturing will match the currently circulating seasonal and pandemic H1N1 strains.

VACCINATION IS THE FIRST LINE OF DEFENSE

In addition to the trivalent vaccine against seasonal influenza, a monovalent vaccine for pandemic H1N1 virus is being produced. The CDC has indicated that 45 million doses of pandemic influenza vaccine are expected in October 2009, with an average of 20 million doses each week thereafter. It is anticipated that half of these will be in multidose vials, that 20% will be in prefilled syringes for children over 5 years old and for pregnant women, and that 20% will be in the form of live-attenuated influenza vaccine (nasal spray). The inhaled vaccine should not be given to children under 2 years old, to children under 5 years old who have recurrent wheezing, or to anyone with severe asthma. Neither vaccine should be given to people allergic to hen eggs, from which the vaccine is produced.

An ample supply of the seasonal trivalent vaccine should be available. Once the CDC has more information about specific product availability of the pandemic H1N1 vaccine, that vaccine will be distributed. It can be given concurrently with seasonal influenza vaccine.

Several definitions should be kept in mind when discussing vaccination strategies. Supply is the number of vaccine doses available for distribution. Availability is the ability of a person recommended to be vaccinated to do so in a local venue. Prioritization is the recommendation to vaccination venues to selectively use vaccine for certain population groups first. Targeting is the recommendation that immunization programs encourage and promote vaccination for certain population groups.

The Advisory Committee on Immunization Practices and the CDC recommend both seasonal and H1N1 vaccinations for anyone 6 months of age or older who is at risk of becoming ill or of transmitting the viruses to others. Based on a review of epidemiologic data, the recommendation is for targeting the following five groups for H1N1 vaccination: children and young adults aged 6 months through 24 years; pregnant women; health care workers and emergency medical service workers; people ages 25 through 64 years who have certain health conditions (eg, diabetes, heart disease, lung disease); and people who live with or care for children younger than 6 months of age. This represents approximately 159 million people in the United States.

If the estimates for the vaccine supply are met, and if pandemic H1N1 vaccine requires only a single injection, there should be no need for prioritization of vaccine. If the supply of pandemic H1N1 vaccine is inadequate, then those groups who are targeted would also receive the first doses of the pandemic H1N1 vaccine. It should be used only with caution after consideration of potential benefits and risks in people who have had Guillain-Barré syndrome during the previous 6 weeks, in people with altered immunocompetence, or in people with medical conditions predisposing to influenza complications.

A mass vaccination campaign involving two separate flu vaccines can pose challenges in execution and messaging for public health officials and politicians. In 1976, an aggressive vaccination program turned into a disaster, as there was no pandemic and the vaccine was associated with adverse effects such as Guillain-Barré syndrome. The government and the medical profession need to prepare for a vaccine controversy and to communicate and continue to explain the plan to the public. As pointed out in a recent op-ed piece,⁹ we would hope that all expectant women in the fall flu season will get the flu vaccines. We also know that, normally, one in seven pregnancies would be expected to miscarry. The challenge for public health officials and physicians will be to explain to these patients that there may be an association rather than a causal relationship.

In health care workers, the average vaccination rate is only 37%. We should be doing much better. Cleveland Clinic previously increased the rate of vaccination among its employees via a program in which all workers must either be vaccinated or formally declare (on an internal Web site) that they decline to be vaccinated.¹⁰ This season, even more resources are being directed at decreasing the barriers to flu vaccinations for our health care workers with the support from hospital leadership.

INFECTION CONTROL IN THE HOSPITAL AND IN THE COMMUNITY

Influenza is very contagious and is spread in droplets via sneezing and coughing (within a 3-foot radius), or via unwashed hands—thus the infection-control campaigns urging you to cover your cough and wash your hands.

As noted, for patients being admitted or transferred to the hospital, we need to have a low threshold for testing for influenza and for isolating patients suspected of having influenza. For patients with suspected or proven seasonal influenza, transmission precautions are those recommended by the CDC for droplet precautions (www.cdc.gov/ncidod/dhqp/gl_isolation_droplet.html). A face mask is deemed adequate to protect transmission when coming within 3 feet of an infected person. CDC guidelines for pandemic H1N1 recommends airborne-transmission-based precautions for health care workers who are in close contact with patients with proven or possible H1N1 (www.cdc.gov/ncidod/dhqp/gl_isolation_airborne.html). This recommendation implies the use of fit-tested N95 respirators and negative air pressure rooms (if available).

The recent Institute of Medicine report, Respiratory Protection for Healthcare Workers in the Workplace Against Novel H1N1 Influenza A (www.iom.edu/CMS/3740/71769/72967/72970.aspx) endorses the current CDC guidelines and recommends following these guidelines until we have evidence that other forms of protection or guidelines are equally or more effective.

Personally, I am against this requirement because it creates a terrible administrative burden with no proven benefit. Requiring a respirator means requiring fit-testing, and this will negatively affect our ability to deliver patient care. Recent studies have shown that surgical masks may not be as effective¹¹ but are probably sufficient. Lim et al¹² reported that 79 (37%) of 212 workers who responded to a survey experienced headaches while wearing N95 masks. This remains a controversial issue.

Besides getting the flu shot, what can one do to avoid getting influenza or transmitting to others?

• Cover your cough (cough etiquette) and sneeze.
• Practice good hand hygiene.
• Avoid close contact with people who are sick.
• Do not go to school or work if sick.

A recent study of influenza in households suggested that having the person with flu and household contacts wear face masks and practice hand hygiene within the first 36 hours decreased transmission of flu within the household.¹³

The United States does have a national influenza pandemic plan that outlines specific roles in the event of a pandemic, and I urge you to peruse it at www.hhs.gov/pandemicflu/plan/.

RECOGNIZING AND DIAGNOSING INFLUENZA

The familiar signs and symptoms of influenza—fever, cough, muscle aches, and headache—are nonspecific. Call et al¹⁴ analyzed the diagnostic accuracy of symptoms and signs of influenza and found that fever and cough during an epidemic suggest but do not confirm influenza, and that sneezing in those over age 60 argues against influenza. They concluded that signs and symptoms can tell us whether a patient has an influenza-like illness, but do not confirm or exclude the diagnosis of influenza: “Clinicians need to consider whether influenza is circulating in their communities, and then either treat patients with influenza-like illness empirically or obtain a rapid influenza test.”¹⁴

The signs and symptoms of pandemic 2009 H1N1 are the same as for seasonal flu, except that about 25% of patients with pandemic flu develop gastrointestinal symptoms. It has not been more virulent than seasonal influenza to date.

Should you order a test for influenza?

Most people with influenza are neither tested nor treated. Before ordering a test for influenza, ask, “Does this patient actually have influenza?” Patients diagnosed with “influenza” may have a range of infectious and noninfectious causes, such as vasculitis, endocarditis, or any other condition that can cause a fever and cough.

If I truly suspect influenza, I would still only order a test if the results would change how I manage the patient—for example, a patient being admitted to the hospital where isolation would be required.

Pandemic H1N1 will be detected only as influenza A in our current PCR screen for human influenza. The test does not differentiate between seasonal strains of influenza A (which is resistant to oseltamivir) and pandemic H1N1 (which is susceptible to oseltamivir). This means if you intend to treat, you will have to address further complexity.

Testing for influenza

The clinician should be familiar with the types of tests available. Each test has advantages and disadvantages¹⁵:

Rapid antigen assay is a point-of-care test that can give results in 15 minutes but unfortunately is only 20% to 30% sensitive, so a negative result does not exclude the diagnosis. The positive predictive value is high, meaning a positive test means the patient does have the flu.

Direct fluorescent antibody testing takes about 2.5 hours to complete and requires special training for technicians. It has a sensitivity of 47%, a positive predictive value of 95%, and a negative predictive value of 92%.

PCR testing takes about 6 hours and has a sensitivity of 98%, a positive predictive value of 100%, and a negative predictive value of 98%. This is probably the best test, in view of its all-around performance, but it is not a point-of-care test.

Culture takes 2 to 3 days, has a sensitivity of 89%, a positive predictive value of 100%, and a negative predictive value of 88%.

These tests can determine that the patient has influenza A, but a confirmatory test is always required to confirm pandemic H1N1. This confirmatory testing can be done by the CDC, by state public health laboratories, and by commercial reference laboratories.

ANTIVIRAL TREATMENT

Since influenza test results do not specify whether the patient has seasonal or pandemic influenza, treatment decisions are a sticky wicket. Most patients with pandemic H1N1 do not need to be tested or treated.

Several drugs are approved for treating influenza and shorten the duration of symptoms by about 1 day. The earlier the treatment is started, the better: the time of antiviral initiation affects influenza viral load and the duration of viral shedding.³

The neuraminidase inhibitors oseltamivir and zanamivir (Relenza) block release of virus from the cell. Resistance to oseltamivir is emerging in seasonal influenza A, while most pandemic H1N1 strains are susceptible.

Oseltamivir resistance in pandemic H1N1

A total of 11 cases of oseltamivir-resistant pandemic H1N1 have been confirmed worldwide, including 3 in the United States (2 in immunosuppressed patients in Seattle, WA). Ten of the 11 cases occurred with oseltamivir exposure. All involved a histidine-to-tyrosine substitution at position 275 (H275Y) of the neuraminidase gene. Most were susceptible to zanamivir.

Supplies of oseltamivir and zanamivir are limited, so they should be used only in those who will benefit the most, ie, those at higher risk of influenza complications. These include children under 5 years old, adults age 65 and older, children and adolescents on long-term aspirin therapy, pregnant women, patients who have chronic conditions or who are immunosuppressed, and residents of long-term care facilities.

A 69-year-old ohio man with leukemia was treated in another state in late June. During the car trip back to Ohio, he developed a sore throat, fever, cough, and nasal congestion. He was admitted to Cleveland Clinic with a presumed diagnosis of neutropenic fever; his absolute neutrophil count was 0.4 × 10⁹/L (reference range 1.8–7.7). His chest radiograph was normal. He was treated with empiric broad-spectrum antimicrobials. On his second day in the hospital, he was tested for influenza by a polymerase chain reaction (PCR) test, which was positive for influenza A. He was moved to a private room and started on oseltamivir (Tamiflu) and rimantadine (Flumadine). The patient’s previous roommate subsequently tested positive for influenza A, as did two health care workers working on the ward. All patients on the floor received prophylactic oseltamivir.

The patient’s condition worsened, and he subsequently went into respiratory distress with diffuse pulmonary infiltrates. He was transferred to the intensive care unit, where he was intubated. Influenza A was isolated from a bronchoscopic specimen. He subsequently recovered after a prolonged course and was discharged on hospital day 50. Testing by the Ohio Department of Health confirmed that this was the 2009 pandemic influenza A (H1N1) virus.

THE CHALLENGES WE FACE

We are now in the midst of an influenza pandemic of the 2009 influenza A (H1N1) virus, with pandemic defined as “worldwide sustained community transmission.” The circulation of seasonal and 2009 pandemic influenza A (H1N1) strains will make this flu season both interesting and challenging.

The approaches to vaccination, prophylaxis, and treatment will be more complex. As of this writing (mid-September 2009), it is clear that we will be giving two influenza vaccines this season: a trivalent vaccine for seasonal influenza, and a monovalent vaccine for pandemic H1N1. It appears the monovalent vaccine may require only one dose to provide protective immunity.¹ Fortunately, the vast majority of cases of pandemic H1N1 are relatively mild and uncomplicated. Still, some people are at higher risk of complications, including young patients, pregnant women, and people with immune deficiency or concomitant health conditions that put them at higher risk of flu-associated complications. Thus, clinicians will need to be educated about whom to test, who needs prophylaxis, and who should not be treated.

As our case demonstrates, unsuspected cases of influenza in hospitalized patients or health care workers working with influenza pose the greatest threat for transmission of influenza within the hospital. Adults hospitalized with influenza tend to present late (more than 48 hours after the onset of symptoms) and tend to have prolonged illness.² Ambulatory adults shed virus for 3 to 6 days; virus shedding is more prolonged for hospitalized patients. Antiviral agents started within 4 days of illness enhance viral clearance and are associated with a shorter stay.³ Therefore, we should have a low threshold for testing for influenza and for isolating all suspected cases.

This is also creating a paradigm shift for health care workers, who are notorious for working through an illness. If you are sick, stay home! This applies whether you have pandemic H1N1 or something else.

EPIDEMIOLOGY OF PANDEMIC 2009 INFLUENZA A (H1N1) VIRUS

The location of cases can now be found on Google Maps; the US Centers for Disease Control and Prevention (CDC) provides weekly influenza reports at www.cdc.gov/flu/weekly/fluactivity.htm.

Pandemic H1N1 appeared in the spring of 2009, and cases continued to mount all summer in the United States (when influenza is normally absent) and around the world. In Mexico in March and April 2009, 2,155 cases of pneumonia, 821 hospitalizations, and 100 deaths were reported.⁴

In contrast with seasonal influenza, children and younger adults were hit the hardest in Mexico. The age group 5 through 59 years accounted for 87% of the deaths (usually, they account for about 17%) and 71% of the cases of severe pneumonia (usually, they account for 32%). These observations may be explained in part by the possibility that people who were alive during the 1957 pandemic (which was an H1N1 strain) have some immunity to the new virus. However, the case-fatality rate was highest in people age 65 and older.⁴

As of July 2009, there were more than 43,000 confirmed cases of pandemic H1N1 in the United States, and actual cases probably exceed 1 million, with more than 400 deaths. An underlying risk factor was identified in more than half of the fatal cases.⁵ Ten percent of the women who died were pregnant.

Pandemic H1N1 has several distinctive epidemiologic features:

• The distribution of cases is similar across multiple geographic areas.
• The distribution of cases by age group is markedly different than that of seasonal influenza, with more cases in school children and fewer cases in older adults.
• Fewer cases have been reported in older adults, but this group has the highest case-fatality rate.

2009 PANDEMIC H1N1 IS A MONGREL

There are three types of influenza viruses, designated A, B, and C. Type A undergoes antigenic shift (rapid changes) and antigenic drift (gradual changes) from year to year, and so it is the type associated with pandemics. In contrast, type B undergoes antigenic drift only, and type C is relatively stable.

Influenza virus is subtyped on the basis of surface glycoproteins: 16 hemagglutinins and nine neuraminidases. The circulating subtypes change every year; the current circulating human subtypes are a seasonal subtype of H1N1 that is different than the pandemic H1N1 subtype, and H3N2.

The 2009 pandemic H1N1 is a new virus never seen before in North America.⁶ Genetically, it is a mongrel, coming from three recognized sources (pigs, birds, and humans) which were combined in pigs.⁷ It is similar to subtypes that circulated in the 1920s through the 1940s.

Most influenza in the Western world comes from Asia every fall, and its arrival is probably facilitated by air travel. The spread is usually unidirectional and is unlikely to contribute to long-term viral evolution.⁸ It appears that 2009 H1N1 virus is the predominant strain circulating in the current influenza season in the Southern Hemisphere. Virologic studies indicate that the H1N1 virus strain has remained antigenically stable since it appeared in April 2009. Thus, it appears likely that the strain selected by the United States for vaccine manufacturing will match the currently circulating seasonal and pandemic H1N1 strains.

VACCINATION IS THE FIRST LINE OF DEFENSE

In addition to the trivalent vaccine against seasonal influenza, a monovalent vaccine for pandemic H1N1 virus is being produced. The CDC has indicated that 45 million doses of pandemic influenza vaccine are expected in October 2009, with an average of 20 million doses each week thereafter. It is anticipated that half of these will be in multidose vials, that 20% will be in prefilled syringes for children over 5 years old and for pregnant women, and that 20% will be in the form of live-attenuated influenza vaccine (nasal spray). The inhaled vaccine should not be given to children under 2 years old, to children under 5 years old who have recurrent wheezing, or to anyone with severe asthma. Neither vaccine should be given to people allergic to hen eggs, from which the vaccine is produced.

An ample supply of the seasonal trivalent vaccine should be available. Once the CDC has more information about specific product availability of the pandemic H1N1 vaccine, that vaccine will be distributed. It can be given concurrently with seasonal influenza vaccine.

Several definitions should be kept in mind when discussing vaccination strategies. Supply is the number of vaccine doses available for distribution. Availability is the ability of a person recommended to be vaccinated to do so in a local venue. Prioritization is the recommendation to vaccination venues to selectively use vaccine for certain population groups first. Targeting is the recommendation that immunization programs encourage and promote vaccination for certain population groups.

The Advisory Committee on Immunization Practices and the CDC recommend both seasonal and H1N1 vaccinations for anyone 6 months of age or older who is at risk of becoming ill or of transmitting the viruses to others. Based on a review of epidemiologic data, the recommendation is for targeting the following five groups for H1N1 vaccination: children and young adults aged 6 months through 24 years; pregnant women; health care workers and emergency medical service workers; people ages 25 through 64 years who have certain health conditions (eg, diabetes, heart disease, lung disease); and people who live with or care for children younger than 6 months of age. This represents approximately 159 million people in the United States.

If the estimates for the vaccine supply are met, and if pandemic H1N1 vaccine requires only a single injection, there should be no need for prioritization of vaccine. If the supply of pandemic H1N1 vaccine is inadequate, then those groups who are targeted would also receive the first doses of the pandemic H1N1 vaccine. It should be used only with caution after consideration of potential benefits and risks in people who have had Guillain-Barré syndrome during the previous 6 weeks, in people with altered immunocompetence, or in people with medical conditions predisposing to influenza complications.

A mass vaccination campaign involving two separate flu vaccines can pose challenges in execution and messaging for public health officials and politicians. In 1976, an aggressive vaccination program turned into a disaster, as there was no pandemic and the vaccine was associated with adverse effects such as Guillain-Barré syndrome. The government and the medical profession need to prepare for a vaccine controversy and to communicate and continue to explain the plan to the public. As pointed out in a recent op-ed piece,⁹ we would hope that all expectant women in the fall flu season will get the flu vaccines. We also know that, normally, one in seven pregnancies would be expected to miscarry. The challenge for public health officials and physicians will be to explain to these patients that there may be an association rather than a causal relationship.

In health care workers, the average vaccination rate is only 37%. We should be doing much better. Cleveland Clinic previously increased the rate of vaccination among its employees via a program in which all workers must either be vaccinated or formally declare (on an internal Web site) that they decline to be vaccinated.¹⁰ This season, even more resources are being directed at decreasing the barriers to flu vaccinations for our health care workers with the support from hospital leadership.

INFECTION CONTROL IN THE HOSPITAL AND IN THE COMMUNITY

Influenza is very contagious and is spread in droplets via sneezing and coughing (within a 3-foot radius), or via unwashed hands—thus the infection-control campaigns urging you to cover your cough and wash your hands.

As noted, for patients being admitted or transferred to the hospital, we need to have a low threshold for testing for influenza and for isolating patients suspected of having influenza. For patients with suspected or proven seasonal influenza, transmission precautions are those recommended by the CDC for droplet precautions (www.cdc.gov/ncidod/dhqp/gl_isolation_droplet.html). A face mask is deemed adequate to protect transmission when coming within 3 feet of an infected person. CDC guidelines for pandemic H1N1 recommends airborne-transmission-based precautions for health care workers who are in close contact with patients with proven or possible H1N1 (www.cdc.gov/ncidod/dhqp/gl_isolation_airborne.html). This recommendation implies the use of fit-tested N95 respirators and negative air pressure rooms (if available).

The recent Institute of Medicine report, Respiratory Protection for Healthcare Workers in the Workplace Against Novel H1N1 Influenza A (www.iom.edu/CMS/3740/71769/72967/72970.aspx) endorses the current CDC guidelines and recommends following these guidelines until we have evidence that other forms of protection or guidelines are equally or more effective.

Personally, I am against this requirement because it creates a terrible administrative burden with no proven benefit. Requiring a respirator means requiring fit-testing, and this will negatively affect our ability to deliver patient care. Recent studies have shown that surgical masks may not be as effective¹¹ but are probably sufficient. Lim et al¹² reported that 79 (37%) of 212 workers who responded to a survey experienced headaches while wearing N95 masks. This remains a controversial issue.

Besides getting the flu shot, what can one do to avoid getting influenza or transmitting to others?

• Cover your cough (cough etiquette) and sneeze.
• Practice good hand hygiene.
• Avoid close contact with people who are sick.
• Do not go to school or work if sick.

A recent study of influenza in households suggested that having the person with flu and household contacts wear face masks and practice hand hygiene within the first 36 hours decreased transmission of flu within the household.¹³

The United States does have a national influenza pandemic plan that outlines specific roles in the event of a pandemic, and I urge you to peruse it at www.hhs.gov/pandemicflu/plan/.

RECOGNIZING AND DIAGNOSING INFLUENZA

The familiar signs and symptoms of influenza—fever, cough, muscle aches, and headache—are nonspecific. Call et al¹⁴ analyzed the diagnostic accuracy of symptoms and signs of influenza and found that fever and cough during an epidemic suggest but do not confirm influenza, and that sneezing in those over age 60 argues against influenza. They concluded that signs and symptoms can tell us whether a patient has an influenza-like illness, but do not confirm or exclude the diagnosis of influenza: “Clinicians need to consider whether influenza is circulating in their communities, and then either treat patients with influenza-like illness empirically or obtain a rapid influenza test.”¹⁴

The signs and symptoms of pandemic 2009 H1N1 are the same as for seasonal flu, except that about 25% of patients with pandemic flu develop gastrointestinal symptoms. It has not been more virulent than seasonal influenza to date.

Should you order a test for influenza?

Most people with influenza are neither tested nor treated. Before ordering a test for influenza, ask, “Does this patient actually have influenza?” Patients diagnosed with “influenza” may have a range of infectious and noninfectious causes, such as vasculitis, endocarditis, or any other condition that can cause a fever and cough.

If I truly suspect influenza, I would still only order a test if the results would change how I manage the patient—for example, a patient being admitted to the hospital where isolation would be required.

Pandemic H1N1 will be detected only as influenza A in our current PCR screen for human influenza. The test does not differentiate between seasonal strains of influenza A (which is resistant to oseltamivir) and pandemic H1N1 (which is susceptible to oseltamivir). This means if you intend to treat, you will have to address further complexity.

Testing for influenza

The clinician should be familiar with the types of tests available. Each test has advantages and disadvantages¹⁵:

Rapid antigen assay is a point-of-care test that can give results in 15 minutes but unfortunately is only 20% to 30% sensitive, so a negative result does not exclude the diagnosis. The positive predictive value is high, meaning a positive test means the patient does have the flu.

Direct fluorescent antibody testing takes about 2.5 hours to complete and requires special training for technicians. It has a sensitivity of 47%, a positive predictive value of 95%, and a negative predictive value of 92%.

PCR testing takes about 6 hours and has a sensitivity of 98%, a positive predictive value of 100%, and a negative predictive value of 98%. This is probably the best test, in view of its all-around performance, but it is not a point-of-care test.

Culture takes 2 to 3 days, has a sensitivity of 89%, a positive predictive value of 100%, and a negative predictive value of 88%.

These tests can determine that the patient has influenza A, but a confirmatory test is always required to confirm pandemic H1N1. This confirmatory testing can be done by the CDC, by state public health laboratories, and by commercial reference laboratories.

ANTIVIRAL TREATMENT

Since influenza test results do not specify whether the patient has seasonal or pandemic influenza, treatment decisions are a sticky wicket. Most patients with pandemic H1N1 do not need to be tested or treated.

Several drugs are approved for treating influenza and shorten the duration of symptoms by about 1 day. The earlier the treatment is started, the better: the time of antiviral initiation affects influenza viral load and the duration of viral shedding.³

The neuraminidase inhibitors oseltamivir and zanamivir (Relenza) block release of virus from the cell. Resistance to oseltamivir is emerging in seasonal influenza A, while most pandemic H1N1 strains are susceptible.

Oseltamivir resistance in pandemic H1N1

A total of 11 cases of oseltamivir-resistant pandemic H1N1 have been confirmed worldwide, including 3 in the United States (2 in immunosuppressed patients in Seattle, WA). Ten of the 11 cases occurred with oseltamivir exposure. All involved a histidine-to-tyrosine substitution at position 275 (H275Y) of the neuraminidase gene. Most were susceptible to zanamivir.

Supplies of oseltamivir and zanamivir are limited, so they should be used only in those who will benefit the most, ie, those at higher risk of influenza complications. These include children under 5 years old, adults age 65 and older, children and adolescents on long-term aspirin therapy, pregnant women, patients who have chronic conditions or who are immunosuppressed, and residents of long-term care facilities.

In spite of some recent studies, or perhaps because of them, we still are unsure about how best to screen for and prevent prostate cancer. Two large trials of screening with prostate-specific antigen (PSA) measurements came to seemingly opposite conclusions.^1,2 Furthermore, a large trial of selenium and vitamin E found that these agents have no value as preventive agents.³

See related editorial

Nevertheless, negative studies also advance science, and steady progress is being made in prostate cancer research. In this paper I briefly summarize and comment on some of the recent findings.

TO SCREEN OR NOT TO SCREEN?

All cases of prostate cancer are clinically relevant in that they can cause anxiety or can lead to treatment-related morbidity. The challenge is to detect the minority of cases of cancer that are biologically significant, ie, those that will cause serious illness or death.

Many men have prostate cancer

In the United States, the lifetime probability of developing prostate cancer is 1 in 6, and the probability increases with age. Prostate cancer is primarily a disease of the Western world, but it is becoming more common in other areas as well.

Risk factors for prostate cancer are age, race, and family history. Clinically apparent disease is very rare in men younger than 40 years; until recently, most guidelines suggested that screening for it should begin at age 50. African American men have the highest risk of developing and dying of prostate cancer, for reasons that are not clear. In the past, this finding was attributed to disparities in access and less aggressive therapy in black men, but recent studies suggest the differences persist even in the absence of these factors, suggesting there is a biological difference in cancers between blacks and whites. Having a father or brother who had prostate cancer increases one’s risk twofold (threefold if the father or brother was affected before the age of 60); having a father and a brother with prostate cancer increases one’s risk fourfold, and true hereditary cancer raises the risk fivefold.⁴

But relatively few men die of it

The Scandinavian Prostate Cancer Group⁵ randomized 695 men with early prostate cancer (mostly discovered by digital rectal examination or by symptoms) to undergo either radical prostatectomy or a program of watchful waiting. In 8.2 years of follow-up, 8.6% of the men in the surgery group and 14.4% of those in the watchful waiting group died of prostate cancer. Thus, we can conclude that surgery is beneficial in this situation.

Adapated from Bill-Axelson A, et al. Radical prostatectomy versus watchful waiting in early prostate cancer. N Engl J Med 2005; 352:1977–1984. Copyright 2005 Massachusetts Medical Society. All rights reserved.

Despite these calculations, in contemporary practice in the United States, about 90% of men with newly diagnosed low-grade prostate cancer choose to be treated.⁶ This high level of intervention reflects our current inability to predict which cancers will remain indolent vs which will progress and the lack of validated markers that tell us when to intervene in patients who are managed expectantly and not lose the chance for cure. Most often, patients and their physicians, who are paid to intervene, deal with this uncertainty by choosing the high likelihood of cure with early intervention despite treatment-related morbidity.

What PSA has wrought

When PSA screening was introduced in the late 1980s and early 1990s, it brought about several changes in the epidemiology and clinical profile of this disease that led us to believe that it was making a meaningful difference.

A spike in the apparent incidence of prostate cancer occurred in the late 1980s and early 1990s with the introduction of PSA screening. The spike was temporary, representing detection of preexisting cases. Now, the incidence may have leveled off.⁷

A shift in the stages of cancers detected. In 1982, half of men with newly diagnosed prostate cancer had incurable disease.⁸ Five years after the introduction of PSA testing, 95% had curable disease.⁹

An increase in the rate of cure after radical prostatectomy was seen.

A decrease in the death rate from prostate cancer since the early 1990s has been noted, which is likely due not only to earlier detection but also to earlier and better treatment.

PSA screening has low specificity. PSA is more sensitive than digital rectal examination, but most men with “elevated” PSA do not have prostate cancer. Nevertheless, although it is not a perfect screening test, it is still the best cancer marker that we have.

In the Prostate Cancer Prevention Trial (PCPT),¹⁰ finasteride (Proscar) decreased the incidence of prostate cancer by about 25% over 7 years. But there were also lessons to be learned from the placebo group, which underwent PSA testing every year and prostate biopsy at the end of the study.

We used to think the cutoff PSA level that had high sensitivity and specificity for finding cancer was 4 ng/mL. However, in the PCPT, 6.6% of men with PSA levels below 0.5 ng/mL were found to have cancer, and 12.5% of those cancers were high-grade. Of those with PSA levels of 3.1 to 4.0 ng/mL, 26.9% had cancer, and 25.0% of the cancers were high-grade. These data demonstrate that there is no PSA level below which risk of cancer is zero, and that there is no PSA cutoff with sufficient sensitivity and specificity to be clinically useful.

The PCPT risk calculator (http://deb.uthscsa.edu/URORiskCalc/Pages/uroriskcalc.jsp) is a wonderful tool that came out of that study. It uses seven variables—race, age, PSA level, family history of prostate cancer, findings on digital rectal examination, whether the patient has ever undergone a prostate biopsy, and whether the patient is taking finasteride—and calculates the patient’s risk of harboring prostate cancer and, more important, the risk of having high-grade prostate cancer. This tool allows estimation of individual risk and helps identify who is at risk of cancer that may require therapy.

Other factors can affect PSA levels. Men with a higher body mass index have lower PSA levels. The reason is not clear; it may be a hormonal effect, or heavier men may simply have higher blood volume, which may dilute the PSA. Furthermore, there are genetic differences that make some men secrete more PSA, but this effect is probably not clinically important. And a study by Hamilton et al¹¹ suggested that statin drugs lower PSA levels. As these findings are confirmed, in the future it may be necessary to adjust PSA levels to account for their effects before deciding on the need for biopsy.

Two new, conflicting studies

Two large trials of PSA screening, published simultaneously in March 2009, came to opposite conclusions.

The European Randomized Study of Screening for Prostate Cancer² randomized 162,243 men between the ages of 55 and 69 to undergo PSA screening at an average of once every 4 years or to a control group. Most of the participating centers used a PSA level of 3.0 ng/mL as an indication for biopsy. At an average follow-up time of 8.8 years, 214 men had died of prostate cancer in the screening group, compared with 326 in the control group, for an adjusted rate ratio of 0.80 (95% confidence interval [CI] 0.65–0.98, P = .04). In other words, screening decreased the risk of death from prostate cancer by 20%.

The Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial,¹ conducted in the United States, came to the opposite conclusion, ie, that there is no benefit from PSA screening. This study was smaller, with 76,693 men between ages 55 and 74 randomly assigned to receive PSA testing every year for 6 years and digital rectal examination for 4 years, or usual care. A PSA level of more than 4.0 ng/mL was considered to be positive for prostate cancer. At 7 years, of those who reported undergoing no more than one PSA test at baseline, 48 men had died of prostate cancer in the screening group, compared with 41 in the control group (rate ratio 1.16, 95% CI 0.76–1.76).

Why were the findings different? The PLCO investigators offered several possible explanations for their negative results. The PSA threshold of 4 ng/mL that was used in that study may not be effective. More than half the men in the control group actually had a PSA test in the first 6 years of the study, potentially diluting any effect of testing. (This was the most worrisome flaw in the study, in my opinion.) About 44% of the men in the study had already had one or more PSA tests at baseline, which would have eliminated cancers detectable on screening from the study, and not all men who were advised to undergo biopsy actually did so. The follow-up time may not yet be long enough for the benefit to be apparent. Most important, in their opinion, treatment for prostate cancer improved during the time of the trial, so that fewer men than expected died of prostate cancer in both groups.

Improvements to PSA screening

Derivatives of PSA have been used in an attempt to improve its performance characteristics for detecting cancer.

PSA density, defined as serum PSA divided by prostate volume, has some predictive power but requires performance of transrectal ultrasonography. It is therefore not a good screening test in the primary care setting.

PSA velocity or doubling time, based on the rate of change over time, is predictive of prostate cancer, but is highly dependent on the absolute value of PSA and does not add independent information to the variables defined in the PCPT risk calculator or other standard predictive variables.¹²

A PSA level between the ages of 44 and 50 may predict the lifetime risk of prostate cancer, according to a study by Lilja et al¹³ in Sweden. This finding suggests that we should measure PSA early in life and screen men who have higher values more frequently or with better strategies. This recommendation has been adopted by the American Urological Association, which released updated screening guidelines in April 2009 (available at www.auanet.org/content/guidelines-and-quality-care/clinical-guidelines/main-reports/psa09.pdf).

New markers under study

A number of new biological markers probably will improve our ability to detect prostate cancer, although they are not yet ready for widespread use.

Urinary PCA3. Prostate cancer gene 3 (PCA3) codes for a messenger RNA that is highly overexpressed in the urine of men with prostate cancer. Urine is collected after prostate massage. Marks et al¹⁴ reported that PCA3 scores predicted biopsy outcomes in men with serum PSA levels of 2.5 ng/mL or higher.

Serum EPCA-2 (early prostate cancer antigen 2) is another candidate marker undergoing study.

Gene fusions, specifically of TRMPSS2 and the ETS gene family, are detectable in high levels in the urine of some men with prostate cancer, and appear to be very promising markers for detection.

Metabolomics is a technique that uses mass spectroscopy to detect the metabolic signature of cancer. Sreekumar et al¹⁵ identified sarcosine as a potential marker of prostate cancer using this technique.

Some data suggest that we can use genetic tests to screen for prostate cancer, but the tests are not yet as good as we would like.

Zheng et al¹⁶ reported that 16 singlenucleotide polymorphisms (SNPs) in five chromosomal regions plus a family history of prostate cancer have a cumulative association with prostate cancer: men who had any five or more of these SNPs had a risk of prostate cancer nearly 10 times as high as men without any of them. However, the number of men who actually fall into this category is so low that routine use in the general population is not cost-effective; it may, however, be useful in men with a family history of prostate cancer.

Other SNPs have been linked to prostate cancer (reviewed by Witte¹⁷). Having any one of these loci increases one’s risk only modestly, however. Only about 2% of the population has five or more of these SNPS, and the sensitivity is about only about 16%.

A commercially available DNA test (Decode Genetics, Reykjavik, Iceland) can detect eight variants that, according to the company, account for about half of all cases of prostate cancer.

Prostate cancer screening: My interpretation

I believe the two new studies of PSA screening suggest there is a modest benefit from screening in terms of preventing deaths from prostate cancer. But I also believe we should be more judicious in recommending treatment for men whom we know have biologically indolent tumors, although we cannot yet identify them perfectly.

In the past, we used an arbitrary PSA cutoff to detect prostate cancer of any grade, and men with high levels were advised to have a biopsy. Currently, we use continuous-risk models to look for any cancer and biologically significant cancers. These involve nomograms, a risk calculator, and new markers.

We use the PCPT risk calculator routinely in our practice. I recommend—completely arbitrarily—that a man undergo biopsy if he has a 10% or higher risk of high-grade cancer, but not if the risk is less. I believe this is more accurate than a simple PSA cutoff value.

CAN WE PREVENT PROSTATE CANCER?

Prostate cancer is a significant public health risk, with 186,000 new cases and 26,000 deaths yearly. Its risk factors (age, race, and genes) are not modifiable. The benefit of screening in terms of preventing deaths is not as good as we would like, and therapy is associated with morbidity. That leaves prevention as a potential way to reduce the morbidity and perhaps mortality of prostate cancer and its therapy.

Epidemiologic studies suggest that certain lifestyle factors may increase the risk, ie, consumption of fat, red meat, fried foods, and dairy; high calcium intake; smoking; total calories; and body size. Other factors may decrease the risk: plant-based foods and vegetables, especially lycopene-containing foods such as tomatoes, cruciferous vegetables, soy, and legumes, specific nutrients such as carotenoids, lycopene, total antioxidants, fish oil (omega-3 fatty acids), and moderate to vigorous exercise. However, there have been few randomized trials to determine if any of these agents are beneficial.

Selenium and vitamin E do not prevent prostate cancer, lung cancer, colorectal cancer, other primary cancers, or deaths. The Selenium and Vitamin E Cancer Prevention Trial (SELECT)³ involved 35,533 men 55 years of age or older (or 50 and older if they were African American). They were randomized to receive one of four treatments: selenium 200 μg/day plus vitamin E placebo, vitamin E 400 IU/day plus selenium placebo, selenium plus vitamin E, or double placebo. At a median follow-up of 5.46 years, compared with the placebo group, the hazard ratio for prostate cancer was 1.04 in the selenium-only group, 1.13 in the vitamin E-only group, and 1.05 in the selenium-plus-vitamin E group. None of the differences was statistically significant.

The Physician’s Health Study¹⁸ also found that vitamin E at the same dose given every other day does not prevent prostate cancer.

Finasteride prevents prostate cancer. The PCPT¹⁹ included 18,882 men, 55 years of age or older, who had PSA levels of 3.0 ng/mL or less and normal findings on digital rectal examination. Treatment was with finasteride 5 mg/day or placebo. At 7 years, prostate cancer had been discovered in 18.4% of the finasteride group vs 24.4% of the placebo group, a 24.8% reduction (95% CI 18.6–30.6, P < .001). Sexual side effects were more common in the men who received finasteride, while urinary symptoms were more common in the placebo group.

At the time of the original PCPT report in 2003,¹⁹ tumors of Gleason grade 7 or higher were more common in the finasteride group, accounting for 37.0% of the tumors discovered, than in the placebo group (22.2%), creating concern that finasteride might somehow cause the tumors that occurred to be more aggressive. However, a subsequent analysis²⁰ found the opposite to be true, ie, that finasteride decreases the risk of high-grade cancers. A companion quality-of-life study showed that chronic use of finasteride had clinically insignificant effects on sexual function, and the PCPT and other studies have shown benefits of finasteride in reducing lower urinary tract symptoms due to benign prostatic hyperplasia (BPH), reducing the risk of acute urinary retention and the need for surgical intervention for BPH, and reducing the risk of prostatitis.

Dutasteride also prevents prostate cancer. A large-scale trial of another 5-alpha reductase inhibitor, dutasteride (Avodart), was reported by Andriole at the annual meeting of the American Urological Association in April 2009.²¹ The Reduction by Dutasteride of Prostate Events (REDUCE) trial included men who were 50 to 75 years old, inclusively, and who had PSA levels between 2.5 and 10 ng/mL, prostate volume less than 80 cc, and one prior negative prostate biopsy within 6 months of enrollment, representing a group at high risk for cancer on a subsequent biopsy. The trial accrued 8,231 men. At 4 years, prostate cancer had occurred in 659 men in the dutasteride group vs 857 in the placebo group, a 23% reduction (P < .0001). Interestingly, no significant increase in Gleason grade 8 to grade 10 tumors was observed in the study.

Preliminary analyses also suggest that dutasteride enhanced the utility of PSA as a diagnostic test for prostate cancer, had beneficial effects on BPH, and was generally well tolerated. The fact that the results of REDUCE were congruent with those of the PCPT with respect to the magnitude of risk reduction, beneficial effects on benign prostatic hypertrophy, minimal toxicity, and no issues related to tumor grade suggests a class effect for 5-alpha reductase inhibitors, and suggests that these agents should be used more liberally for the prevention of prostate cancer.

There is current debate about whether 5-alpha reductase inhibitors should be used by all men at risk of prostate cancer or only by those at high risk. However, the American Urological Association and the American Society of Clinical Oncology have issued guidelines stating that men at risk should consider this intervention.²²

In spite of some recent studies, or perhaps because of them, we still are unsure about how best to screen for and prevent prostate cancer. Two large trials of screening with prostate-specific antigen (PSA) measurements came to seemingly opposite conclusions.^1,2 Furthermore, a large trial of selenium and vitamin E found that these agents have no value as preventive agents.³

See related editorial

Nevertheless, negative studies also advance science, and steady progress is being made in prostate cancer research. In this paper I briefly summarize and comment on some of the recent findings.

TO SCREEN OR NOT TO SCREEN?

All cases of prostate cancer are clinically relevant in that they can cause anxiety or can lead to treatment-related morbidity. The challenge is to detect the minority of cases of cancer that are biologically significant, ie, those that will cause serious illness or death.

Many men have prostate cancer

In the United States, the lifetime probability of developing prostate cancer is 1 in 6, and the probability increases with age. Prostate cancer is primarily a disease of the Western world, but it is becoming more common in other areas as well.

Risk factors for prostate cancer are age, race, and family history. Clinically apparent disease is very rare in men younger than 40 years; until recently, most guidelines suggested that screening for it should begin at age 50. African American men have the highest risk of developing and dying of prostate cancer, for reasons that are not clear. In the past, this finding was attributed to disparities in access and less aggressive therapy in black men, but recent studies suggest the differences persist even in the absence of these factors, suggesting there is a biological difference in cancers between blacks and whites. Having a father or brother who had prostate cancer increases one’s risk twofold (threefold if the father or brother was affected before the age of 60); having a father and a brother with prostate cancer increases one’s risk fourfold, and true hereditary cancer raises the risk fivefold.⁴

But relatively few men die of it

The Scandinavian Prostate Cancer Group⁵ randomized 695 men with early prostate cancer (mostly discovered by digital rectal examination or by symptoms) to undergo either radical prostatectomy or a program of watchful waiting. In 8.2 years of follow-up, 8.6% of the men in the surgery group and 14.4% of those in the watchful waiting group died of prostate cancer. Thus, we can conclude that surgery is beneficial in this situation.