CASE 1 › Joe S is a 41-year-old African-American man who comes to your clinic after his employee health screening revealed elevated triglycerides. The patient has a 3-year history of type 2 diabetes mellitus (T2DM); he also has a history of hypertension, gastroesophageal reflux disease, and obstructive sleep apnea. Mr. S tells you he takes metformin 1000 mg twice daily, but stopped taking his glipizide because he didn’t think it was helping his blood sugar. His last hemoglobin (Hb) A1c result was 8.8%, and he is very resistant to starting insulin therapy.

The patient’s other medications include enalapril 10 mg/d, atorvastatin 10 mg/d, and omeprazole 20 mg/d. Mr. S weighs 255.6 lbs (body mass index=34.7), his BP is 140/88 mm Hg, and his heart rate is 82 beats per minute. Laboratory values include: serum creatinine, 1.01 mg/dL; estimated glomerular filtration rate (eGFR) >100 mL/min/1.73 m²; potassium (K), 4.3 mmol/L; serum phosphorous (Phos), 2.8 mg/dL; magnesium (Mg), 1.9 mg/dL; total cholesterol, 167 mg/dL; low-density lipoprotein (LDL), 78 mg/dL; high-density lipoprotein (HDL), 38 mg/dL; and triglycerides, 256 mg/dL.

CASE 2  › Susan R, a 68-year-old Caucasian woman, returns to your clinic for a follow-up visit 3 months after you prescribed dapagliflozin 10 mg/d for her T2DM. Her glucose levels have improved, but she complains of vaginal pruritus and is worried that she has a yeast infection.

You diagnose vulvovaginal candidiasis in this patient and prescribe a single dose of fluconazole 150 mg. After reviewing her laboratory test results, you notice that since starting the dapagliflozin, her HbA1c level has improved slightly from 9.8% to 9.3%, but is still not where it needs to be. Her eGFR is 49 mL/min/1.73 m².

What would you recommend to improve control of these patients’ blood glucose levels?

Sodium-glucose cotransporter 2 (SGLT2) inhibitors are the newest class of agents to enter the T2DM management arena. They act in the proximal renal tubules to decrease the reabsorption of glucose by targeting the SGLT2 transmembrane protein, which reabsorbs about 90% of the body’s glucose.^1,2 The class is currently made up of 3 agents—canagliflozin, dapagliflozin, and empagliflozin—all of which are approved by the US Food and Drug Administration (FDA) for the treatment of T2DM (TABLE 1).²

The American Diabetes Association and the European Association for the Study of Diabetes published updated guidelines for T2DM management in 2015.¹ In addition to lifestyle modifications, the guidelines recommend the use of metformin as first-line therapy unless it is contraindicated or patients are unable to tolerate it (eg, because of gastrointestinal adverse effects). They recommend other pharmacologic therapies as second-line options based on specific patient characteristics. Thus, SGLT2 inhibitors may be used as add-on therapy after metformin, or as a first-line option if metformin is contraindicated or not tolerated. Because the mechanism of action of SGLT2 inhibitors is independent of insulin secretion, these agents may be used at any stage of the diabetes continuum.

SGLT2 agents as monotherapy, or as add-on therapy

All SGLT2 agents have been studied as monotherapy accompanied by diet and exercise and shown to produce HbA1c reductions of 0.34% to 1.11%.^3-6 In trials, the effect was similar regardless of study duration (18-104 weeks); generally, higher doses corresponded with larger HbA1c reductions.^3-6

Used as monotherapy, SGLT2 inhibitors produce HbA1c reductions of as much as 1.11%.

SGLT2 inhibitors have also been studied as add-on therapy to several oral agents including metformin, sulfonylureas, thiazolidinediones (TZDs), and the combination of metformin plus sulfonylureas or TZDs or dipeptidyl peptidase IV (DPP-IV) inhibitors.¹ When used in any of these combinations, each SGLT2 agent demonstrated a consistent HbA1c lowering effect of 0.62% to 1.19%.^7-14

Additionally, SGLT2 inhibitors have been studied in combination with insulin therapy (median or mean daily doses >60 units), which yielded further reductions in HbA1c of 0.58% to 1.02% without significant insulin adjustments or an increase in major hypoglycemia events.^15-17 Patients receiving insulin and an SGLT2 inhibitor had lower insulin doses and more weight loss compared to placebo groups.

SGLT2 inhibitors offer additional benefits

Secondary analyses of most studies of SGLT2 inhibitors include changes in BP and weight from baseline as well as minor changes (some positive, some not) in several lipid parameters.^{3-5,7-9,13-15,17-24} In general, these effects do not appear to be dose-dependent (with the exception of canagliflozin and its associated lipid effects²⁵) and are similar among the 3 medications.^{3-5,7-9,13-15,17-24} (For more on who would benefit from these agents, see “When to consider an SGLT2 inhibitor” above.)

BP reduction. Although the mean baseline BP was controlled in most studies, SGLT2 inhibitors have been shown to significantly reduce BP. Reductions in BP with all 3 SGLT2 medications range from approximately 2 to 5 mm Hg systolic and 0.5 to 2.5 mm Hg diastolic, which may be due to weight loss and diuresis.^{4-8,10-16,20-23} While the reductions were modest at best, one study involving empagliflozin reported that more than one-quarter of patients with uncontrolled BP at baseline achieved a BP <130/80 mm Hg 24 weeks later.⁵ While these agents should not be used solely for their BP lowering effects, they may help a small number of patients with mildly elevated BP achieve their goal without an additional antihypertensive agent.

Weight reduction. Modest weight loss, likely due to the loss of calories through urine, was seen with SGLT2 inhibitors in most studies, with reductions persisting beyond one year of use. In most studies, including those involving obese patients on insulin therapy,^15,17,21 patients’ body weights were reduced by approximately 2 to 4 kg from baseline.^{3-16,18,21-23,26}

Lipid effects. Although the mechanism is unclear, use of SGLT2 inhibitors can have varying effects on lipid panels. In most studies, total and LDL cholesterol levels were increased with elevations ranging from 0.7 to 10 mg/dL.^{3,7,8,18,19,22,23} Conversely, at least one study demonstrated mild reductions in total and LDL cholesterol levels with higher doses of empagliflozin.¹³ Additionally, modest reductions in triglycerides and increases in HDL across all doses of canagliflozin, dapagliflozin, and empagliflozin have been seen.^8,9,13,15,19 While the clinical relevance of these lipid changes is unknown, monitoring is recommended.²

These agents are well tolerated

SGLT2 inhibitors were generally well tolerated in studies. The most common adverse effects include mycotic infections (2.4%-21.6%) and urinary tract infections (UTIs) (4.0%-19.6%) (both with higher incidences in females); volume-related effects such as dizziness and hypotension (0.3%-8.3%); and nasopharyngitis (5.4%-18.3%).^{4-14,16-23,26-28} Hypoglycemia was observed more often when an SGLT2 inhibitor was used in combination with a sulfonylurea or insulin therapy.^{4-14,16-23,26-28} The number of times adverse events led to discontinuation was low and similar to that in control groups.^{4-14,16-23,26-28}

Mycotic and urinary infections should be diagnosed and treated according to current standards of care and do not require discontinuation of the SGLT2 inhibitor. Canagli-flozin therapy was associated with electrolyte abnormalities including hyperkalemia, hypermagnesemia, and hyperphosphatemia.²⁵ Thus, levels should be monitored periodically, especially in patients predisposed to elevations due to other conditions or medications.²⁵

Two additional warnings are worth noting

Diabetic ketoacidosis (DKA) has been reported with all 3 agents, and bone fractures have been reported with canagliflozin.

The FDA issued a warning in May 2015 regarding the increased risk of DKA with the use of SGLT2 inhibitor single and combination products.²⁹ This warning was prompted by several case reports of DKA with uncharacteristically mild to moderate glucose elevations in patients with type 1 diabetes mellitus (T1DM) and T2DM who were taking an SGLT2 inhibitor. The absence of significant hyperglycemia delayed diagnosis in many cases. Therefore, patients should be counseled on the signs and symptoms of DKA, as well as when to seek medical attention.

SGLT2 inhibitors can reduce BP by about 2 to 5 mm Hg, systolic, and 0.5 to 2.5 mm Hg, diastolic.

Patients with diabetes and symptoms of ketoacidosis (eg, difficulty breathing, nausea, vomiting, abdominal pain, confusion, and fatigue) should be evaluated regardless of current blood glucose levels, and SGLT2 inhibitors should be discontinued if acidosis is confirmed. Identified potential triggers include illness, reduced food and fluid intake, reduced insulin dose, and history of alcohol intake. Use of SGLT2 inhibitors should be avoided in patients with T1DM until safety and efficacy are established in large randomized controlled trials. The European Medicines Agency announced that a thorough review of all currently approved SGLT2 agents is underway to evaluate the risk for DKA.³⁰

In addition, the FDA called for a revision of the label of canagliflozin to reflect a strengthened warning about an increased risk of bone fractures and decreased bone mineral density (BMD).³¹ Fractures can occur as early as 12 weeks after initiating treatment and with only minor trauma.³¹

Over a 2-year period, canagliflozin also significantly decreased BMD in the hip and lower spine compared to placebo.³¹ Patients should be evaluated for additional risk factors for fracture before taking canagliflozin.³¹ The FDA is continuing to evaluate whether the other approved SGLT2 inhibitors are associated with an increased risk for fractures.

Drug interactions: Proceed carefully with diuretics

The number of drugs that interact with SGLT2 inhibitors is minimal. Because these agents can cause volume-related effects such as hypotension, dizziness, and osmotic diuresis, patients—particularly the elderly and those with renal impairment—taking concomitant diuretics, especially loop diuretics, may be at increased risk for these effects and should be monitored accordingly.^2,25

Canagliflozin is primarily metabolized via glucuronidation by the uridine 5'-diphospho-glucuronosyltransferase (UGT) enzymes. Therefore, UGT enzyme inducers (eg, rifampin, phenytoin, phenobarbital, ritonavir) decrease canagliflozin’s serum concentration. If a patient has an eGFR >60 mL/min/1.73 m² and is tolerating a dose of 100 mg/d, consider increasing the dose to 300 mg/d during concomitant treatment.

In addition, researchers have found that canagliflozin increases serum levels of digoxin by between 20% and 36%.²⁵ Experts suspect this occurs because canagliflozin inhibits P-glycoprotein efflux of digoxin. Although monitoring of digoxin levels is recommended, this interaction is considered to be minor.²⁵

Cost consideration: SGLT2 inhibitors are more expensive

The SGLT2 inhibitors are available only as brand name products and are more expensive than agents that have generic options (eg, metformin, sulfonylureas, TZDs). The average wholesale cost is approximately $400 for a 30-day supply of all SGLT2 agents.³² When considering an SGLT2 inhibitor, the patient should ideally have medication prescription coverage. Depending on the specific insurance plan, these agents are classified as tier 2 to 4, which is comparable to other oral brand name options.

Research looks at CV outcomes and cancer risk

Cardiovascular (CV) risk reduction. To date, only one study evaluating the effect of SGLT2 inhibitors on CV outcomes is complete.³³ Two large randomized controlled trials involving canagliflozin and dapagliflozin designed to evaluate treatment effects on major CV endpoints are ongoing.^34,35

In the EMPA-REG OUTCOME trial,³³ researchers found that empagliflozin had beneficial effects on CV outcomes, making it one of the only antidiabetic agents on the market to have such benefits. The study, which involved more than 7000 patients with a history of T2DM and existing cardiovascular disease (CVD), found that 10.5% of patients in the empagliflozin group vs 12.1% in the placebo group died from a CV cause or experienced a nonfatal myocardial infarction or stroke over a median of 3.1 years. Results were similar with both doses (10 mg vs 25 mg) of empagliflozin. The mechanisms behind the CV benefits are likely multifactorial and may be related to reductions in weight and BP,³³ but additional research is needed to fully elucidate the role of empagliflozin in this population.

Canagliflozin is being evaluated in the Canagliflozin Cardiovascular Assessment Study (CANVAS) for its effect on major CV events—CV death, nonfatal myocardial infarction, and nonfatal stroke—in patients with either a history of CVD or who are at increased risk of CVD and have uncontrolled diabetes.³⁴ The trial is expected to wrap up in June 2017.

And dapagliflozin is being studied in the DECLARE-TIMI 58 trial (the Effect of Dapagliflozin on the Incidence of Cardiovascular Events) in patients with T2DM and either known CVD or at least 2 risk factors for CVD.³⁵ The study is designed to assess dapagliflozin’s effect on the incidence of CV death, myocardial infarction, and ischemic stroke and has an estimated completion date of April 2019, which will provide a median follow-up of 4.5 years.

Cancer. All 3 agents have been examined for any possible carcinogenic links. In 2011, the FDA issued a request for further investigation surrounding the risk of cancer associated with dapagliflozin.³⁶ As of November 2013, 10 of 6045 patients treated with dapagliflozin developed bladder cancer compared to 1 of 3512 controls.³⁶ Furthermore, 9 of 2223 patients treated with dapagliflozin developed breast cancer compared to 1 of 1053 controls.³⁶

Although the trials were not designed to detect an increase in risk, the number of observed cases warranted further investigation. No official warning for breast cancer exists since the characteristics of the malignancies led the FDA to believe dapagliflozin was unlikely the cause.³⁶

When considering an SGLT2 inhibitor, the patient should ideally have medication prescription coverage.

Given what we know to date, it appears to be prudent to avoid prescribing SGLT2 inhibitors in patients with active bladder cancer, and to use them with caution in those with a history of the disease.²

Other studies. Initially, animal studies suggested an increased risk of various malignancies associated with canagliflozin use in rats,³⁷ but consistent results were not seen in human studies. Similarly, at least one study found that empagliflozin was associated with lung cancer and melanoma, but closer examination found that most patients who developed these cancers had risk factors.³⁸ Large, long-term studies of these agents in various populations are needed to thoroughly investigate possible carcinogenicity.

Additional considerations: Kidney function, age, and pregnancy

Consider avoiding SGLT2 inhibitors in patients with moderate kidney dysfunction (eGFR 30-59 mL/min/1.73 m²). Studies have shown that SGLT2 inhibitors are not as effective at lowering blood glucose in those with reduced eGFR, although adverse events were similar to those in placebo groups.^24,39,40 Dapagliflozin is not recommended in patients with an eGFR <60 mL/min/1.73 m² due to lack of efficacy.^2,24 Empagliflozin does not require dose adjustments if eGFR is ≥45 mL/min/1.73 m². A lower dose of canagliflozin (ie, 100 mg/d) is recommended in those with an eGFR of 45 to 59 mL/min/1.73 m².² All agents are contraindicated in patients with severe renal impairment (eGFR <30 mL/min/1.73 m²).

Older patients are at higher risk for dehydration, hypotension, and falls; therefore, SGLT2 inhibitors should be used with caution in this population. Similarly, they should not be used in patients with T1DM and should be avoided in those with active, or a history of, DKA.

There are no data on the use of SGLT2 inhibitors in pregnancy; thus, these agents should be avoided unless the potential benefits outweigh the potential risks to the unborn fetus.²

CASE 1 › An SGLT2 inhibitor is an acceptable option for Mr. S. Because he is resistant to starting insulin therapy and his HbA1c is <9%, an additional oral medication is reasonable. Adding an SGLT2 inhibitor may reduce his HbA1c up to ~1%, and education on lifestyle modifications may help bring him to goal. An SGLT2 inhibitor may also benefit his BP and weight, both of which could be improved.

The FDA issued a warning in light of several case reports of DKA with unusually mild glucose elevations in patients taking an SGLT2 inhibitor.

Given the drugs he’s taking, drug interactions should not be an issue, and his renal function and pertinent labs (K, Phos, Mg) are within normal limits. Nevertheless, monitor these labs periodically and monitor Mr. S for adverse effects, such as UTIs, although these are more common in women. Canagliflozin is the preferred SGLT2 inhibitor on his insurance formulary, so you could initiate therapy at 100 mg/d, administered prior to the first meal, and increase to 300 mg/d if needed. As an alternative, consider prescribing the metformin/canagliflozin combination agent.

CASE 2 › Ms. R is likely experiencing a yeast infection as an adverse effect of the dapagliflozin. Although one yeast infection is insufficient grounds for discontinuation of the drug, recurrent infections should prompt a risk-to-benefit analysis to determine whether it’s worth continuing the medication. Her recent eGFR (<60 mL/min/1.73 m²) is, however, a contraindication to dapagliflozin, and therapy should be discontinued. Canagliflozin and empagliflozin may be considered since her eGFR is >45 mL/min/1.73 m², but given her current HbA1c and recent adverse drug event, alternative therapies, such as basal insulin, are more appropriate treatment choices.

CORRESPONDENCE
Katelin M. Lisenby, PharmD, BCPS, University of Alabama College of Community Health Sciences, University Medical Center, Box 870374, Tuscaloosa, AL 35487; [email protected].

CASE 1 › Joe S is a 41-year-old African-American man who comes to your clinic after his employee health screening revealed elevated triglycerides. The patient has a 3-year history of type 2 diabetes mellitus (T2DM); he also has a history of hypertension, gastroesophageal reflux disease, and obstructive sleep apnea. Mr. S tells you he takes metformin 1000 mg twice daily, but stopped taking his glipizide because he didn’t think it was helping his blood sugar. His last hemoglobin (Hb) A1c result was 8.8%, and he is very resistant to starting insulin therapy.

The patient’s other medications include enalapril 10 mg/d, atorvastatin 10 mg/d, and omeprazole 20 mg/d. Mr. S weighs 255.6 lbs (body mass index=34.7), his BP is 140/88 mm Hg, and his heart rate is 82 beats per minute. Laboratory values include: serum creatinine, 1.01 mg/dL; estimated glomerular filtration rate (eGFR) >100 mL/min/1.73 m²; potassium (K), 4.3 mmol/L; serum phosphorous (Phos), 2.8 mg/dL; magnesium (Mg), 1.9 mg/dL; total cholesterol, 167 mg/dL; low-density lipoprotein (LDL), 78 mg/dL; high-density lipoprotein (HDL), 38 mg/dL; and triglycerides, 256 mg/dL.

CASE 2  › Susan R, a 68-year-old Caucasian woman, returns to your clinic for a follow-up visit 3 months after you prescribed dapagliflozin 10 mg/d for her T2DM. Her glucose levels have improved, but she complains of vaginal pruritus and is worried that she has a yeast infection.

You diagnose vulvovaginal candidiasis in this patient and prescribe a single dose of fluconazole 150 mg. After reviewing her laboratory test results, you notice that since starting the dapagliflozin, her HbA1c level has improved slightly from 9.8% to 9.3%, but is still not where it needs to be. Her eGFR is 49 mL/min/1.73 m².

What would you recommend to improve control of these patients’ blood glucose levels?

Sodium-glucose cotransporter 2 (SGLT2) inhibitors are the newest class of agents to enter the T2DM management arena. They act in the proximal renal tubules to decrease the reabsorption of glucose by targeting the SGLT2 transmembrane protein, which reabsorbs about 90% of the body’s glucose.^1,2 The class is currently made up of 3 agents—canagliflozin, dapagliflozin, and empagliflozin—all of which are approved by the US Food and Drug Administration (FDA) for the treatment of T2DM (TABLE 1).²

The American Diabetes Association and the European Association for the Study of Diabetes published updated guidelines for T2DM management in 2015.¹ In addition to lifestyle modifications, the guidelines recommend the use of metformin as first-line therapy unless it is contraindicated or patients are unable to tolerate it (eg, because of gastrointestinal adverse effects). They recommend other pharmacologic therapies as second-line options based on specific patient characteristics. Thus, SGLT2 inhibitors may be used as add-on therapy after metformin, or as a first-line option if metformin is contraindicated or not tolerated. Because the mechanism of action of SGLT2 inhibitors is independent of insulin secretion, these agents may be used at any stage of the diabetes continuum.

SGLT2 agents as monotherapy, or as add-on therapy

All SGLT2 agents have been studied as monotherapy accompanied by diet and exercise and shown to produce HbA1c reductions of 0.34% to 1.11%.^3-6 In trials, the effect was similar regardless of study duration (18-104 weeks); generally, higher doses corresponded with larger HbA1c reductions.^3-6

Used as monotherapy, SGLT2 inhibitors produce HbA1c reductions of as much as 1.11%.

SGLT2 inhibitors have also been studied as add-on therapy to several oral agents including metformin, sulfonylureas, thiazolidinediones (TZDs), and the combination of metformin plus sulfonylureas or TZDs or dipeptidyl peptidase IV (DPP-IV) inhibitors.¹ When used in any of these combinations, each SGLT2 agent demonstrated a consistent HbA1c lowering effect of 0.62% to 1.19%.^7-14

Additionally, SGLT2 inhibitors have been studied in combination with insulin therapy (median or mean daily doses >60 units), which yielded further reductions in HbA1c of 0.58% to 1.02% without significant insulin adjustments or an increase in major hypoglycemia events.^15-17 Patients receiving insulin and an SGLT2 inhibitor had lower insulin doses and more weight loss compared to placebo groups.

SGLT2 inhibitors offer additional benefits

Secondary analyses of most studies of SGLT2 inhibitors include changes in BP and weight from baseline as well as minor changes (some positive, some not) in several lipid parameters.^{3-5,7-9,13-15,17-24} In general, these effects do not appear to be dose-dependent (with the exception of canagliflozin and its associated lipid effects²⁵) and are similar among the 3 medications.^{3-5,7-9,13-15,17-24} (For more on who would benefit from these agents, see “When to consider an SGLT2 inhibitor” above.)

BP reduction. Although the mean baseline BP was controlled in most studies, SGLT2 inhibitors have been shown to significantly reduce BP. Reductions in BP with all 3 SGLT2 medications range from approximately 2 to 5 mm Hg systolic and 0.5 to 2.5 mm Hg diastolic, which may be due to weight loss and diuresis.^{4-8,10-16,20-23} While the reductions were modest at best, one study involving empagliflozin reported that more than one-quarter of patients with uncontrolled BP at baseline achieved a BP <130/80 mm Hg 24 weeks later.⁵ While these agents should not be used solely for their BP lowering effects, they may help a small number of patients with mildly elevated BP achieve their goal without an additional antihypertensive agent.

Weight reduction. Modest weight loss, likely due to the loss of calories through urine, was seen with SGLT2 inhibitors in most studies, with reductions persisting beyond one year of use. In most studies, including those involving obese patients on insulin therapy,^15,17,21 patients’ body weights were reduced by approximately 2 to 4 kg from baseline.^{3-16,18,21-23,26}

Lipid effects. Although the mechanism is unclear, use of SGLT2 inhibitors can have varying effects on lipid panels. In most studies, total and LDL cholesterol levels were increased with elevations ranging from 0.7 to 10 mg/dL.^{3,7,8,18,19,22,23} Conversely, at least one study demonstrated mild reductions in total and LDL cholesterol levels with higher doses of empagliflozin.¹³ Additionally, modest reductions in triglycerides and increases in HDL across all doses of canagliflozin, dapagliflozin, and empagliflozin have been seen.^8,9,13,15,19 While the clinical relevance of these lipid changes is unknown, monitoring is recommended.²

These agents are well tolerated

SGLT2 inhibitors were generally well tolerated in studies. The most common adverse effects include mycotic infections (2.4%-21.6%) and urinary tract infections (UTIs) (4.0%-19.6%) (both with higher incidences in females); volume-related effects such as dizziness and hypotension (0.3%-8.3%); and nasopharyngitis (5.4%-18.3%).^{4-14,16-23,26-28} Hypoglycemia was observed more often when an SGLT2 inhibitor was used in combination with a sulfonylurea or insulin therapy.^{4-14,16-23,26-28} The number of times adverse events led to discontinuation was low and similar to that in control groups.^{4-14,16-23,26-28}

Mycotic and urinary infections should be diagnosed and treated according to current standards of care and do not require discontinuation of the SGLT2 inhibitor. Canagli-flozin therapy was associated with electrolyte abnormalities including hyperkalemia, hypermagnesemia, and hyperphosphatemia.²⁵ Thus, levels should be monitored periodically, especially in patients predisposed to elevations due to other conditions or medications.²⁵

Two additional warnings are worth noting

Diabetic ketoacidosis (DKA) has been reported with all 3 agents, and bone fractures have been reported with canagliflozin.

The FDA issued a warning in May 2015 regarding the increased risk of DKA with the use of SGLT2 inhibitor single and combination products.²⁹ This warning was prompted by several case reports of DKA with uncharacteristically mild to moderate glucose elevations in patients with type 1 diabetes mellitus (T1DM) and T2DM who were taking an SGLT2 inhibitor. The absence of significant hyperglycemia delayed diagnosis in many cases. Therefore, patients should be counseled on the signs and symptoms of DKA, as well as when to seek medical attention.

SGLT2 inhibitors can reduce BP by about 2 to 5 mm Hg, systolic, and 0.5 to 2.5 mm Hg, diastolic.

Patients with diabetes and symptoms of ketoacidosis (eg, difficulty breathing, nausea, vomiting, abdominal pain, confusion, and fatigue) should be evaluated regardless of current blood glucose levels, and SGLT2 inhibitors should be discontinued if acidosis is confirmed. Identified potential triggers include illness, reduced food and fluid intake, reduced insulin dose, and history of alcohol intake. Use of SGLT2 inhibitors should be avoided in patients with T1DM until safety and efficacy are established in large randomized controlled trials. The European Medicines Agency announced that a thorough review of all currently approved SGLT2 agents is underway to evaluate the risk for DKA.³⁰

In addition, the FDA called for a revision of the label of canagliflozin to reflect a strengthened warning about an increased risk of bone fractures and decreased bone mineral density (BMD).³¹ Fractures can occur as early as 12 weeks after initiating treatment and with only minor trauma.³¹

Over a 2-year period, canagliflozin also significantly decreased BMD in the hip and lower spine compared to placebo.³¹ Patients should be evaluated for additional risk factors for fracture before taking canagliflozin.³¹ The FDA is continuing to evaluate whether the other approved SGLT2 inhibitors are associated with an increased risk for fractures.

Drug interactions: Proceed carefully with diuretics

The number of drugs that interact with SGLT2 inhibitors is minimal. Because these agents can cause volume-related effects such as hypotension, dizziness, and osmotic diuresis, patients—particularly the elderly and those with renal impairment—taking concomitant diuretics, especially loop diuretics, may be at increased risk for these effects and should be monitored accordingly.^2,25

Canagliflozin is primarily metabolized via glucuronidation by the uridine 5'-diphospho-glucuronosyltransferase (UGT) enzymes. Therefore, UGT enzyme inducers (eg, rifampin, phenytoin, phenobarbital, ritonavir) decrease canagliflozin’s serum concentration. If a patient has an eGFR >60 mL/min/1.73 m² and is tolerating a dose of 100 mg/d, consider increasing the dose to 300 mg/d during concomitant treatment.

In addition, researchers have found that canagliflozin increases serum levels of digoxin by between 20% and 36%.²⁵ Experts suspect this occurs because canagliflozin inhibits P-glycoprotein efflux of digoxin. Although monitoring of digoxin levels is recommended, this interaction is considered to be minor.²⁵

Cost consideration: SGLT2 inhibitors are more expensive

The SGLT2 inhibitors are available only as brand name products and are more expensive than agents that have generic options (eg, metformin, sulfonylureas, TZDs). The average wholesale cost is approximately $400 for a 30-day supply of all SGLT2 agents.³² When considering an SGLT2 inhibitor, the patient should ideally have medication prescription coverage. Depending on the specific insurance plan, these agents are classified as tier 2 to 4, which is comparable to other oral brand name options.

Research looks at CV outcomes and cancer risk

Cardiovascular (CV) risk reduction. To date, only one study evaluating the effect of SGLT2 inhibitors on CV outcomes is complete.³³ Two large randomized controlled trials involving canagliflozin and dapagliflozin designed to evaluate treatment effects on major CV endpoints are ongoing.^34,35

In the EMPA-REG OUTCOME trial,³³ researchers found that empagliflozin had beneficial effects on CV outcomes, making it one of the only antidiabetic agents on the market to have such benefits. The study, which involved more than 7000 patients with a history of T2DM and existing cardiovascular disease (CVD), found that 10.5% of patients in the empagliflozin group vs 12.1% in the placebo group died from a CV cause or experienced a nonfatal myocardial infarction or stroke over a median of 3.1 years. Results were similar with both doses (10 mg vs 25 mg) of empagliflozin. The mechanisms behind the CV benefits are likely multifactorial and may be related to reductions in weight and BP,³³ but additional research is needed to fully elucidate the role of empagliflozin in this population.

Canagliflozin is being evaluated in the Canagliflozin Cardiovascular Assessment Study (CANVAS) for its effect on major CV events—CV death, nonfatal myocardial infarction, and nonfatal stroke—in patients with either a history of CVD or who are at increased risk of CVD and have uncontrolled diabetes.³⁴ The trial is expected to wrap up in June 2017.

And dapagliflozin is being studied in the DECLARE-TIMI 58 trial (the Effect of Dapagliflozin on the Incidence of Cardiovascular Events) in patients with T2DM and either known CVD or at least 2 risk factors for CVD.³⁵ The study is designed to assess dapagliflozin’s effect on the incidence of CV death, myocardial infarction, and ischemic stroke and has an estimated completion date of April 2019, which will provide a median follow-up of 4.5 years.

Cancer. All 3 agents have been examined for any possible carcinogenic links. In 2011, the FDA issued a request for further investigation surrounding the risk of cancer associated with dapagliflozin.³⁶ As of November 2013, 10 of 6045 patients treated with dapagliflozin developed bladder cancer compared to 1 of 3512 controls.³⁶ Furthermore, 9 of 2223 patients treated with dapagliflozin developed breast cancer compared to 1 of 1053 controls.³⁶

Although the trials were not designed to detect an increase in risk, the number of observed cases warranted further investigation. No official warning for breast cancer exists since the characteristics of the malignancies led the FDA to believe dapagliflozin was unlikely the cause.³⁶

When considering an SGLT2 inhibitor, the patient should ideally have medication prescription coverage.

Given what we know to date, it appears to be prudent to avoid prescribing SGLT2 inhibitors in patients with active bladder cancer, and to use them with caution in those with a history of the disease.²

Other studies. Initially, animal studies suggested an increased risk of various malignancies associated with canagliflozin use in rats,³⁷ but consistent results were not seen in human studies. Similarly, at least one study found that empagliflozin was associated with lung cancer and melanoma, but closer examination found that most patients who developed these cancers had risk factors.³⁸ Large, long-term studies of these agents in various populations are needed to thoroughly investigate possible carcinogenicity.

Additional considerations: Kidney function, age, and pregnancy

Consider avoiding SGLT2 inhibitors in patients with moderate kidney dysfunction (eGFR 30-59 mL/min/1.73 m²). Studies have shown that SGLT2 inhibitors are not as effective at lowering blood glucose in those with reduced eGFR, although adverse events were similar to those in placebo groups.^24,39,40 Dapagliflozin is not recommended in patients with an eGFR <60 mL/min/1.73 m² due to lack of efficacy.^2,24 Empagliflozin does not require dose adjustments if eGFR is ≥45 mL/min/1.73 m². A lower dose of canagliflozin (ie, 100 mg/d) is recommended in those with an eGFR of 45 to 59 mL/min/1.73 m².² All agents are contraindicated in patients with severe renal impairment (eGFR <30 mL/min/1.73 m²).

Older patients are at higher risk for dehydration, hypotension, and falls; therefore, SGLT2 inhibitors should be used with caution in this population. Similarly, they should not be used in patients with T1DM and should be avoided in those with active, or a history of, DKA.

There are no data on the use of SGLT2 inhibitors in pregnancy; thus, these agents should be avoided unless the potential benefits outweigh the potential risks to the unborn fetus.²

CASE 1 › An SGLT2 inhibitor is an acceptable option for Mr. S. Because he is resistant to starting insulin therapy and his HbA1c is <9%, an additional oral medication is reasonable. Adding an SGLT2 inhibitor may reduce his HbA1c up to ~1%, and education on lifestyle modifications may help bring him to goal. An SGLT2 inhibitor may also benefit his BP and weight, both of which could be improved.

The FDA issued a warning in light of several case reports of DKA with unusually mild glucose elevations in patients taking an SGLT2 inhibitor.

Given the drugs he’s taking, drug interactions should not be an issue, and his renal function and pertinent labs (K, Phos, Mg) are within normal limits. Nevertheless, monitor these labs periodically and monitor Mr. S for adverse effects, such as UTIs, although these are more common in women. Canagliflozin is the preferred SGLT2 inhibitor on his insurance formulary, so you could initiate therapy at 100 mg/d, administered prior to the first meal, and increase to 300 mg/d if needed. As an alternative, consider prescribing the metformin/canagliflozin combination agent.

CASE 2 › Ms. R is likely experiencing a yeast infection as an adverse effect of the dapagliflozin. Although one yeast infection is insufficient grounds for discontinuation of the drug, recurrent infections should prompt a risk-to-benefit analysis to determine whether it’s worth continuing the medication. Her recent eGFR (<60 mL/min/1.73 m²) is, however, a contraindication to dapagliflozin, and therapy should be discontinued. Canagliflozin and empagliflozin may be considered since her eGFR is >45 mL/min/1.73 m², but given her current HbA1c and recent adverse drug event, alternative therapies, such as basal insulin, are more appropriate treatment choices.

CORRESPONDENCE
Katelin M. Lisenby, PharmD, BCPS, University of Alabama College of Community Health Sciences, University Medical Center, Box 870374, Tuscaloosa, AL 35487; [email protected].

CASE 1 › Joe S is a 41-year-old African-American man who comes to your clinic after his employee health screening revealed elevated triglycerides. The patient has a 3-year history of type 2 diabetes mellitus (T2DM); he also has a history of hypertension, gastroesophageal reflux disease, and obstructive sleep apnea. Mr. S tells you he takes metformin 1000 mg twice daily, but stopped taking his glipizide because he didn’t think it was helping his blood sugar. His last hemoglobin (Hb) A1c result was 8.8%, and he is very resistant to starting insulin therapy.

The patient’s other medications include enalapril 10 mg/d, atorvastatin 10 mg/d, and omeprazole 20 mg/d. Mr. S weighs 255.6 lbs (body mass index=34.7), his BP is 140/88 mm Hg, and his heart rate is 82 beats per minute. Laboratory values include: serum creatinine, 1.01 mg/dL; estimated glomerular filtration rate (eGFR) >100 mL/min/1.73 m²; potassium (K), 4.3 mmol/L; serum phosphorous (Phos), 2.8 mg/dL; magnesium (Mg), 1.9 mg/dL; total cholesterol, 167 mg/dL; low-density lipoprotein (LDL), 78 mg/dL; high-density lipoprotein (HDL), 38 mg/dL; and triglycerides, 256 mg/dL.

CASE 2  › Susan R, a 68-year-old Caucasian woman, returns to your clinic for a follow-up visit 3 months after you prescribed dapagliflozin 10 mg/d for her T2DM. Her glucose levels have improved, but she complains of vaginal pruritus and is worried that she has a yeast infection.

You diagnose vulvovaginal candidiasis in this patient and prescribe a single dose of fluconazole 150 mg. After reviewing her laboratory test results, you notice that since starting the dapagliflozin, her HbA1c level has improved slightly from 9.8% to 9.3%, but is still not where it needs to be. Her eGFR is 49 mL/min/1.73 m².

What would you recommend to improve control of these patients’ blood glucose levels?

Sodium-glucose cotransporter 2 (SGLT2) inhibitors are the newest class of agents to enter the T2DM management arena. They act in the proximal renal tubules to decrease the reabsorption of glucose by targeting the SGLT2 transmembrane protein, which reabsorbs about 90% of the body’s glucose.^1,2 The class is currently made up of 3 agents—canagliflozin, dapagliflozin, and empagliflozin—all of which are approved by the US Food and Drug Administration (FDA) for the treatment of T2DM (TABLE 1).²

The American Diabetes Association and the European Association for the Study of Diabetes published updated guidelines for T2DM management in 2015.¹ In addition to lifestyle modifications, the guidelines recommend the use of metformin as first-line therapy unless it is contraindicated or patients are unable to tolerate it (eg, because of gastrointestinal adverse effects). They recommend other pharmacologic therapies as second-line options based on specific patient characteristics. Thus, SGLT2 inhibitors may be used as add-on therapy after metformin, or as a first-line option if metformin is contraindicated or not tolerated. Because the mechanism of action of SGLT2 inhibitors is independent of insulin secretion, these agents may be used at any stage of the diabetes continuum.

SGLT2 agents as monotherapy, or as add-on therapy

All SGLT2 agents have been studied as monotherapy accompanied by diet and exercise and shown to produce HbA1c reductions of 0.34% to 1.11%.^3-6 In trials, the effect was similar regardless of study duration (18-104 weeks); generally, higher doses corresponded with larger HbA1c reductions.^3-6

Used as monotherapy, SGLT2 inhibitors produce HbA1c reductions of as much as 1.11%.

SGLT2 inhibitors have also been studied as add-on therapy to several oral agents including metformin, sulfonylureas, thiazolidinediones (TZDs), and the combination of metformin plus sulfonylureas or TZDs or dipeptidyl peptidase IV (DPP-IV) inhibitors.¹ When used in any of these combinations, each SGLT2 agent demonstrated a consistent HbA1c lowering effect of 0.62% to 1.19%.^7-14

Additionally, SGLT2 inhibitors have been studied in combination with insulin therapy (median or mean daily doses >60 units), which yielded further reductions in HbA1c of 0.58% to 1.02% without significant insulin adjustments or an increase in major hypoglycemia events.^15-17 Patients receiving insulin and an SGLT2 inhibitor had lower insulin doses and more weight loss compared to placebo groups.

SGLT2 inhibitors offer additional benefits

Secondary analyses of most studies of SGLT2 inhibitors include changes in BP and weight from baseline as well as minor changes (some positive, some not) in several lipid parameters.^{3-5,7-9,13-15,17-24} In general, these effects do not appear to be dose-dependent (with the exception of canagliflozin and its associated lipid effects²⁵) and are similar among the 3 medications.^{3-5,7-9,13-15,17-24} (For more on who would benefit from these agents, see “When to consider an SGLT2 inhibitor” above.)

BP reduction. Although the mean baseline BP was controlled in most studies, SGLT2 inhibitors have been shown to significantly reduce BP. Reductions in BP with all 3 SGLT2 medications range from approximately 2 to 5 mm Hg systolic and 0.5 to 2.5 mm Hg diastolic, which may be due to weight loss and diuresis.^{4-8,10-16,20-23} While the reductions were modest at best, one study involving empagliflozin reported that more than one-quarter of patients with uncontrolled BP at baseline achieved a BP <130/80 mm Hg 24 weeks later.⁵ While these agents should not be used solely for their BP lowering effects, they may help a small number of patients with mildly elevated BP achieve their goal without an additional antihypertensive agent.

Weight reduction. Modest weight loss, likely due to the loss of calories through urine, was seen with SGLT2 inhibitors in most studies, with reductions persisting beyond one year of use. In most studies, including those involving obese patients on insulin therapy,^15,17,21 patients’ body weights were reduced by approximately 2 to 4 kg from baseline.^{3-16,18,21-23,26}

Lipid effects. Although the mechanism is unclear, use of SGLT2 inhibitors can have varying effects on lipid panels. In most studies, total and LDL cholesterol levels were increased with elevations ranging from 0.7 to 10 mg/dL.^{3,7,8,18,19,22,23} Conversely, at least one study demonstrated mild reductions in total and LDL cholesterol levels with higher doses of empagliflozin.¹³ Additionally, modest reductions in triglycerides and increases in HDL across all doses of canagliflozin, dapagliflozin, and empagliflozin have been seen.^8,9,13,15,19 While the clinical relevance of these lipid changes is unknown, monitoring is recommended.²

These agents are well tolerated

SGLT2 inhibitors were generally well tolerated in studies. The most common adverse effects include mycotic infections (2.4%-21.6%) and urinary tract infections (UTIs) (4.0%-19.6%) (both with higher incidences in females); volume-related effects such as dizziness and hypotension (0.3%-8.3%); and nasopharyngitis (5.4%-18.3%).^{4-14,16-23,26-28} Hypoglycemia was observed more often when an SGLT2 inhibitor was used in combination with a sulfonylurea or insulin therapy.^{4-14,16-23,26-28} The number of times adverse events led to discontinuation was low and similar to that in control groups.^{4-14,16-23,26-28}

Mycotic and urinary infections should be diagnosed and treated according to current standards of care and do not require discontinuation of the SGLT2 inhibitor. Canagli-flozin therapy was associated with electrolyte abnormalities including hyperkalemia, hypermagnesemia, and hyperphosphatemia.²⁵ Thus, levels should be monitored periodically, especially in patients predisposed to elevations due to other conditions or medications.²⁵

Two additional warnings are worth noting

Diabetic ketoacidosis (DKA) has been reported with all 3 agents, and bone fractures have been reported with canagliflozin.

The FDA issued a warning in May 2015 regarding the increased risk of DKA with the use of SGLT2 inhibitor single and combination products.²⁹ This warning was prompted by several case reports of DKA with uncharacteristically mild to moderate glucose elevations in patients with type 1 diabetes mellitus (T1DM) and T2DM who were taking an SGLT2 inhibitor. The absence of significant hyperglycemia delayed diagnosis in many cases. Therefore, patients should be counseled on the signs and symptoms of DKA, as well as when to seek medical attention.

SGLT2 inhibitors can reduce BP by about 2 to 5 mm Hg, systolic, and 0.5 to 2.5 mm Hg, diastolic.

Patients with diabetes and symptoms of ketoacidosis (eg, difficulty breathing, nausea, vomiting, abdominal pain, confusion, and fatigue) should be evaluated regardless of current blood glucose levels, and SGLT2 inhibitors should be discontinued if acidosis is confirmed. Identified potential triggers include illness, reduced food and fluid intake, reduced insulin dose, and history of alcohol intake. Use of SGLT2 inhibitors should be avoided in patients with T1DM until safety and efficacy are established in large randomized controlled trials. The European Medicines Agency announced that a thorough review of all currently approved SGLT2 agents is underway to evaluate the risk for DKA.³⁰

In addition, the FDA called for a revision of the label of canagliflozin to reflect a strengthened warning about an increased risk of bone fractures and decreased bone mineral density (BMD).³¹ Fractures can occur as early as 12 weeks after initiating treatment and with only minor trauma.³¹

Over a 2-year period, canagliflozin also significantly decreased BMD in the hip and lower spine compared to placebo.³¹ Patients should be evaluated for additional risk factors for fracture before taking canagliflozin.³¹ The FDA is continuing to evaluate whether the other approved SGLT2 inhibitors are associated with an increased risk for fractures.

Drug interactions: Proceed carefully with diuretics

The number of drugs that interact with SGLT2 inhibitors is minimal. Because these agents can cause volume-related effects such as hypotension, dizziness, and osmotic diuresis, patients—particularly the elderly and those with renal impairment—taking concomitant diuretics, especially loop diuretics, may be at increased risk for these effects and should be monitored accordingly.^2,25

Canagliflozin is primarily metabolized via glucuronidation by the uridine 5'-diphospho-glucuronosyltransferase (UGT) enzymes. Therefore, UGT enzyme inducers (eg, rifampin, phenytoin, phenobarbital, ritonavir) decrease canagliflozin’s serum concentration. If a patient has an eGFR >60 mL/min/1.73 m² and is tolerating a dose of 100 mg/d, consider increasing the dose to 300 mg/d during concomitant treatment.

In addition, researchers have found that canagliflozin increases serum levels of digoxin by between 20% and 36%.²⁵ Experts suspect this occurs because canagliflozin inhibits P-glycoprotein efflux of digoxin. Although monitoring of digoxin levels is recommended, this interaction is considered to be minor.²⁵

Cost consideration: SGLT2 inhibitors are more expensive

The SGLT2 inhibitors are available only as brand name products and are more expensive than agents that have generic options (eg, metformin, sulfonylureas, TZDs). The average wholesale cost is approximately $400 for a 30-day supply of all SGLT2 agents.³² When considering an SGLT2 inhibitor, the patient should ideally have medication prescription coverage. Depending on the specific insurance plan, these agents are classified as tier 2 to 4, which is comparable to other oral brand name options.

Research looks at CV outcomes and cancer risk

Cardiovascular (CV) risk reduction. To date, only one study evaluating the effect of SGLT2 inhibitors on CV outcomes is complete.³³ Two large randomized controlled trials involving canagliflozin and dapagliflozin designed to evaluate treatment effects on major CV endpoints are ongoing.^34,35

In the EMPA-REG OUTCOME trial,³³ researchers found that empagliflozin had beneficial effects on CV outcomes, making it one of the only antidiabetic agents on the market to have such benefits. The study, which involved more than 7000 patients with a history of T2DM and existing cardiovascular disease (CVD), found that 10.5% of patients in the empagliflozin group vs 12.1% in the placebo group died from a CV cause or experienced a nonfatal myocardial infarction or stroke over a median of 3.1 years. Results were similar with both doses (10 mg vs 25 mg) of empagliflozin. The mechanisms behind the CV benefits are likely multifactorial and may be related to reductions in weight and BP,³³ but additional research is needed to fully elucidate the role of empagliflozin in this population.

Canagliflozin is being evaluated in the Canagliflozin Cardiovascular Assessment Study (CANVAS) for its effect on major CV events—CV death, nonfatal myocardial infarction, and nonfatal stroke—in patients with either a history of CVD or who are at increased risk of CVD and have uncontrolled diabetes.³⁴ The trial is expected to wrap up in June 2017.

And dapagliflozin is being studied in the DECLARE-TIMI 58 trial (the Effect of Dapagliflozin on the Incidence of Cardiovascular Events) in patients with T2DM and either known CVD or at least 2 risk factors for CVD.³⁵ The study is designed to assess dapagliflozin’s effect on the incidence of CV death, myocardial infarction, and ischemic stroke and has an estimated completion date of April 2019, which will provide a median follow-up of 4.5 years.

Cancer. All 3 agents have been examined for any possible carcinogenic links. In 2011, the FDA issued a request for further investigation surrounding the risk of cancer associated with dapagliflozin.³⁶ As of November 2013, 10 of 6045 patients treated with dapagliflozin developed bladder cancer compared to 1 of 3512 controls.³⁶ Furthermore, 9 of 2223 patients treated with dapagliflozin developed breast cancer compared to 1 of 1053 controls.³⁶

Although the trials were not designed to detect an increase in risk, the number of observed cases warranted further investigation. No official warning for breast cancer exists since the characteristics of the malignancies led the FDA to believe dapagliflozin was unlikely the cause.³⁶

When considering an SGLT2 inhibitor, the patient should ideally have medication prescription coverage.

Given what we know to date, it appears to be prudent to avoid prescribing SGLT2 inhibitors in patients with active bladder cancer, and to use them with caution in those with a history of the disease.²

Other studies. Initially, animal studies suggested an increased risk of various malignancies associated with canagliflozin use in rats,³⁷ but consistent results were not seen in human studies. Similarly, at least one study found that empagliflozin was associated with lung cancer and melanoma, but closer examination found that most patients who developed these cancers had risk factors.³⁸ Large, long-term studies of these agents in various populations are needed to thoroughly investigate possible carcinogenicity.

Additional considerations: Kidney function, age, and pregnancy

Consider avoiding SGLT2 inhibitors in patients with moderate kidney dysfunction (eGFR 30-59 mL/min/1.73 m²). Studies have shown that SGLT2 inhibitors are not as effective at lowering blood glucose in those with reduced eGFR, although adverse events were similar to those in placebo groups.^24,39,40 Dapagliflozin is not recommended in patients with an eGFR <60 mL/min/1.73 m² due to lack of efficacy.^2,24 Empagliflozin does not require dose adjustments if eGFR is ≥45 mL/min/1.73 m². A lower dose of canagliflozin (ie, 100 mg/d) is recommended in those with an eGFR of 45 to 59 mL/min/1.73 m².² All agents are contraindicated in patients with severe renal impairment (eGFR <30 mL/min/1.73 m²).

Older patients are at higher risk for dehydration, hypotension, and falls; therefore, SGLT2 inhibitors should be used with caution in this population. Similarly, they should not be used in patients with T1DM and should be avoided in those with active, or a history of, DKA.

There are no data on the use of SGLT2 inhibitors in pregnancy; thus, these agents should be avoided unless the potential benefits outweigh the potential risks to the unborn fetus.²

CASE 1 › An SGLT2 inhibitor is an acceptable option for Mr. S. Because he is resistant to starting insulin therapy and his HbA1c is <9%, an additional oral medication is reasonable. Adding an SGLT2 inhibitor may reduce his HbA1c up to ~1%, and education on lifestyle modifications may help bring him to goal. An SGLT2 inhibitor may also benefit his BP and weight, both of which could be improved.

The FDA issued a warning in light of several case reports of DKA with unusually mild glucose elevations in patients taking an SGLT2 inhibitor.

Given the drugs he’s taking, drug interactions should not be an issue, and his renal function and pertinent labs (K, Phos, Mg) are within normal limits. Nevertheless, monitor these labs periodically and monitor Mr. S for adverse effects, such as UTIs, although these are more common in women. Canagliflozin is the preferred SGLT2 inhibitor on his insurance formulary, so you could initiate therapy at 100 mg/d, administered prior to the first meal, and increase to 300 mg/d if needed. As an alternative, consider prescribing the metformin/canagliflozin combination agent.

CASE 2 › Ms. R is likely experiencing a yeast infection as an adverse effect of the dapagliflozin. Although one yeast infection is insufficient grounds for discontinuation of the drug, recurrent infections should prompt a risk-to-benefit analysis to determine whether it’s worth continuing the medication. Her recent eGFR (<60 mL/min/1.73 m²) is, however, a contraindication to dapagliflozin, and therapy should be discontinued. Canagliflozin and empagliflozin may be considered since her eGFR is >45 mL/min/1.73 m², but given her current HbA1c and recent adverse drug event, alternative therapies, such as basal insulin, are more appropriate treatment choices.

CORRESPONDENCE
Katelin M. Lisenby, PharmD, BCPS, University of Alabama College of Community Health Sciences, University Medical Center, Box 870374, Tuscaloosa, AL 35487; [email protected].

The care of people with pain has been wrought with ineffective and unnecessary treatment, including the misuse of opioids, largely because we do not have an accurate conceptualization of pain. The absence of animal and human models of central nervous system (CNS) pain processing ensures that our understanding of pain will remain incomplete for the foreseeable future, but enough evidence exists to help family physicians develop an understanding of pain that goes beyond what we learned in medical school and that can help us more effectively treat patients with pain.

In this review, we will briefly discuss the established concepts of nociceptive and neuropathic pain. And then, with those concepts in mind, we will explore a third type of pain that for lack of a better term, we will call “pain for psychological reasons.” We hypothesize that this pain may be the consequence of changes in nervous system function that arise from developmental trauma, other traumatic experiences in a patient’s life, or mental health disorders. It is this third type of pain that may offer us insights into conditions such as fibromyalgia.

While we do not yet have validated diagnostic criteria for this third type of pain, we believe that there is enough information to present initial criteria so that one may distinguish it from nociceptive and neuropathic pain.

Nociceptive and neuropathic pain: The current paradigm

Nociceptive pain. The sensory pain experience, or nociceptive pain, is produced by noxious stimuli that either damage, or are capable of damaging, tissues (eg, burns, cuts, fractures, inflammation, and increased pressure in a hollow viscus). Noxious stimuli are detected at the molecular level by specific pain sensory receptors embedded in our tissues called nociceptors.

The process by which noxious stimuli lead to the experience of sensory pain consists of 4 steps—transduction, transmission, modulation, and perception—which are described in “From periphery to brain: The process of nociceptive pain.”^1-4

Neuropathic pain. While nociceptive pain can be easily traced from a peripheral nociceptive fiber to the brain and typically resolves when the nociceptive stimulus stops, neuropathic pain (NPP) results from changes to the function of the nervous system and is typically caused by injury to the nerves. Such changes, referred to as neuronal sensitization, may not quickly resolve, as is the case with postherpetic neuralgia. In fact, the changes can become permanent. NPP fundamentally differs from nociceptive pain because it results from changes in the central processing of pain that can lead a person to perceive pain sensations even in the absence of tissue pathology.

This third type of pain may be the consequence of changes in nervous system function that arise from developmental trauma, other traumatic experiences in a patient's life, or mental health disorders.

Common causes of NPP that persists even after tissue damage has healed include trauma (eg, amputation of a limb), ischemia (eg, pressure palsy), disease (eg, the metabolic injury of diabetes or the injury caused by a shingles infection), and drug treatment (eg, chemotherapy). The underlying mechanisms of NPP and the neuronal plasticity (the ability of the nervous system to rewire itself) that initiate and then maintain NPP are important areas of active research that may eventually lead to the development of more effective treatments.

Timing is critical. Neuroplastic changes in the nervous system following nerve injury are time-dependent. Synaptic plasticity can occur within seconds to minutes, while cellular plasticity occurs within hours to days. Synaptic and cellular plasticity happen relatively fast and may be reversible.

In contrast, systems plasticity (when new CNS neuronal connections are formed in response to nerve injury) takes place over the months and years following nerve injury and is often irreversible. When we recognize NPP and intervene before system neuroplastic changes occur, it may be possible to prevent pain from becoming chronic (TABLE 1⁵). In cases of nerve injury, researchers have long suspected that early and aggressive pain treatment within the first few months that may include sympathetic and peripheral neural blockade reduces the likelihood that the patient will have chronic pain.^6,7

It’s time to update our understanding of pain

The International Association for the Study of Pain (IASP)—a group of health care providers, scientists, and policymakers seeking to improve pain relief worldwide—notes in its definition of pain that the complaint, “I hurt” does not necessarily imply that there is a painful stimulus in the form of tissue injury.⁸ Yet most of us have been taught to think of pain solely as the result of tissue pathology, and we assume that emotional factors merely modify how the physical damage is perceived. This traditional concept of pain is incomplete. It leads clinicians to misdiagnose the cause of pain, initiate expensive and unnecessary treatment, engage in well-meaning but misguided prescribing behavior, and miss opportunities to help patients.

Pain in the absence of any pathophysiologic cause or injury

The clinician’s search for a pain diagnosis is typically predicated on the notion that there must be an underlying tissue injury of severity equal to the severity of the patient’s pain complaints. This approach to a pain evaluation rests on 2 assumptions that are not true for all patients:

1. Pain is simply a sensory experience that is always caused by tissue damage of some type.
2. The severity of the pain experienced by a patient should be tightly bound to the severity of the pain stimulus (ie, tissue damage).

These assumptions are true of acute nociceptive pain, they may or may not be true for NPP, but they do not apply to the third type of pain—pain for psychological reasons. While tissue pathology in humans and animals with nociceptive pain is usually visible, measurable, and correlates with observed pain behaviors, the damage to nerve tissue and the ensuing changes in nervous system function with NPP are not always visible or able to be imaged. These changes produce pain that can appear more severe than expected based on a brief exam. Some of the time, however, characteristic symptoms and physical signs of NPP will be present, and perhaps electrodiagnostic or other tests will be abnormal, thus providing some objective sense of changes in nervous system function.

In contrast, pain behavior due to the third type of pain usually appears very much out of proportion, and unbound to, tissue pathology. Furthermore, the patient’s pain behaviors often reflect heightened emotional pain processing (TABLE 2⁹). The resulting emotionally charged presentation can be alarming and suggestive of extreme tissue injury, but there may be absolutely no evidence of tissue injury or pathology.

Functional change in the CNS

There is evidence from experimental studies that psychologic factors change nervous system function. In one review, the authors concluded, “Pain…can vary widely between people and even within an individual depending on…the psychological state of the person.”¹⁰ In a second review, the authors concluded that our emotional state has an enormous influence on pain; a negative emotional state increases pain, whereas a positive state lowers pain.¹¹

But can psychological factors induce long-term changes in nervous system function analogous to the systems neuroplasticity responsible for irreversible changes in NPP? And can psychologically induced changes in nervous system sensory processing lead to pain without any tissue or nerve damage?

We theorize that a functional change in the CNS can occur in response to certain emotional states or traumatic experiences (eg, child abuse, assault, accidents). (More on this in a bit.) When such changes occur, mildly painful stimuli are amplified and processed through overly sensitized, dysregulated, ramped-up emotional and somatosensory pain circuits in the brain. This is analogous to the functional changes in the nervous system that occur with NPP; however, when the nervous system changes are due to psychological factors, there may be no tissue or nerve injury.

Childhood trauma influences adult pain. One of the more compelling narratives emerging in health care has to do with the influence that childhood developmental trauma can have on health, including pain. In his chapter on the impact of early life trauma on health and disease, Lanius states:¹²

“Women were 50% more likely than men to have experienced 5 or more categories of adverse childhood experiences. We believe that here is a key to what in mainstream epidemiology appears as women’s natural proneness to ill-defined health problems like fibromyalgia, chronic fatigue syndrome, obesity, irritable bowel syndrome, and chronic non-malignant pain syndromes. In light of our findings, we now see these as medical constructs, artifacts resulting from medical blindness to social realities and ignorance of the impact of gender.”

Brain activity in response to emotional insult mimics physical pain, and it is difficult to tell from images of brain activity whether a person is experiencing one or the other.

Lanius¹² suggests that adverse childhood experiences¹³ (trauma such as abuse and sexual assault) can lead to long-term changes within the nervous system, including areas of pain processing. My coauthor and I describe these changes here in terms of nervous system sensitization or dysregulation, and we believe that these changes lead to a bias toward hyperactivation of emotional pain circuits, which leads to the emotionally laden pain behaviors that often seem out of proportion to tissue pathology.

The care of people with pain has been wrought with ineffective and unnecessary treatment, including the misuse of opioids, largely because we do not have an accurate conceptualization of pain. The absence of animal and human models of central nervous system (CNS) pain processing ensures that our understanding of pain will remain incomplete for the foreseeable future, but enough evidence exists to help family physicians develop an understanding of pain that goes beyond what we learned in medical school and that can help us more effectively treat patients with pain.

In this review, we will briefly discuss the established concepts of nociceptive and neuropathic pain. And then, with those concepts in mind, we will explore a third type of pain that for lack of a better term, we will call “pain for psychological reasons.” We hypothesize that this pain may be the consequence of changes in nervous system function that arise from developmental trauma, other traumatic experiences in a patient’s life, or mental health disorders. It is this third type of pain that may offer us insights into conditions such as fibromyalgia.

While we do not yet have validated diagnostic criteria for this third type of pain, we believe that there is enough information to present initial criteria so that one may distinguish it from nociceptive and neuropathic pain.

Nociceptive and neuropathic pain: The current paradigm

Nociceptive pain. The sensory pain experience, or nociceptive pain, is produced by noxious stimuli that either damage, or are capable of damaging, tissues (eg, burns, cuts, fractures, inflammation, and increased pressure in a hollow viscus). Noxious stimuli are detected at the molecular level by specific pain sensory receptors embedded in our tissues called nociceptors.

The process by which noxious stimuli lead to the experience of sensory pain consists of 4 steps—transduction, transmission, modulation, and perception—which are described in “From periphery to brain: The process of nociceptive pain.”^1-4

Neuropathic pain. While nociceptive pain can be easily traced from a peripheral nociceptive fiber to the brain and typically resolves when the nociceptive stimulus stops, neuropathic pain (NPP) results from changes to the function of the nervous system and is typically caused by injury to the nerves. Such changes, referred to as neuronal sensitization, may not quickly resolve, as is the case with postherpetic neuralgia. In fact, the changes can become permanent. NPP fundamentally differs from nociceptive pain because it results from changes in the central processing of pain that can lead a person to perceive pain sensations even in the absence of tissue pathology.

This third type of pain may be the consequence of changes in nervous system function that arise from developmental trauma, other traumatic experiences in a patient's life, or mental health disorders.

Common causes of NPP that persists even after tissue damage has healed include trauma (eg, amputation of a limb), ischemia (eg, pressure palsy), disease (eg, the metabolic injury of diabetes or the injury caused by a shingles infection), and drug treatment (eg, chemotherapy). The underlying mechanisms of NPP and the neuronal plasticity (the ability of the nervous system to rewire itself) that initiate and then maintain NPP are important areas of active research that may eventually lead to the development of more effective treatments.

Timing is critical. Neuroplastic changes in the nervous system following nerve injury are time-dependent. Synaptic plasticity can occur within seconds to minutes, while cellular plasticity occurs within hours to days. Synaptic and cellular plasticity happen relatively fast and may be reversible.

In contrast, systems plasticity (when new CNS neuronal connections are formed in response to nerve injury) takes place over the months and years following nerve injury and is often irreversible. When we recognize NPP and intervene before system neuroplastic changes occur, it may be possible to prevent pain from becoming chronic (TABLE 1⁵). In cases of nerve injury, researchers have long suspected that early and aggressive pain treatment within the first few months that may include sympathetic and peripheral neural blockade reduces the likelihood that the patient will have chronic pain.^6,7

It’s time to update our understanding of pain

The International Association for the Study of Pain (IASP)—a group of health care providers, scientists, and policymakers seeking to improve pain relief worldwide—notes in its definition of pain that the complaint, “I hurt” does not necessarily imply that there is a painful stimulus in the form of tissue injury.⁸ Yet most of us have been taught to think of pain solely as the result of tissue pathology, and we assume that emotional factors merely modify how the physical damage is perceived. This traditional concept of pain is incomplete. It leads clinicians to misdiagnose the cause of pain, initiate expensive and unnecessary treatment, engage in well-meaning but misguided prescribing behavior, and miss opportunities to help patients.

Pain in the absence of any pathophysiologic cause or injury

The clinician’s search for a pain diagnosis is typically predicated on the notion that there must be an underlying tissue injury of severity equal to the severity of the patient’s pain complaints. This approach to a pain evaluation rests on 2 assumptions that are not true for all patients:

1. Pain is simply a sensory experience that is always caused by tissue damage of some type.
2. The severity of the pain experienced by a patient should be tightly bound to the severity of the pain stimulus (ie, tissue damage).

These assumptions are true of acute nociceptive pain, they may or may not be true for NPP, but they do not apply to the third type of pain—pain for psychological reasons. While tissue pathology in humans and animals with nociceptive pain is usually visible, measurable, and correlates with observed pain behaviors, the damage to nerve tissue and the ensuing changes in nervous system function with NPP are not always visible or able to be imaged. These changes produce pain that can appear more severe than expected based on a brief exam. Some of the time, however, characteristic symptoms and physical signs of NPP will be present, and perhaps electrodiagnostic or other tests will be abnormal, thus providing some objective sense of changes in nervous system function.

In contrast, pain behavior due to the third type of pain usually appears very much out of proportion, and unbound to, tissue pathology. Furthermore, the patient’s pain behaviors often reflect heightened emotional pain processing (TABLE 2⁹). The resulting emotionally charged presentation can be alarming and suggestive of extreme tissue injury, but there may be absolutely no evidence of tissue injury or pathology.

Functional change in the CNS

There is evidence from experimental studies that psychologic factors change nervous system function. In one review, the authors concluded, “Pain…can vary widely between people and even within an individual depending on…the psychological state of the person.”¹⁰ In a second review, the authors concluded that our emotional state has an enormous influence on pain; a negative emotional state increases pain, whereas a positive state lowers pain.¹¹

But can psychological factors induce long-term changes in nervous system function analogous to the systems neuroplasticity responsible for irreversible changes in NPP? And can psychologically induced changes in nervous system sensory processing lead to pain without any tissue or nerve damage?

We theorize that a functional change in the CNS can occur in response to certain emotional states or traumatic experiences (eg, child abuse, assault, accidents). (More on this in a bit.) When such changes occur, mildly painful stimuli are amplified and processed through overly sensitized, dysregulated, ramped-up emotional and somatosensory pain circuits in the brain. This is analogous to the functional changes in the nervous system that occur with NPP; however, when the nervous system changes are due to psychological factors, there may be no tissue or nerve injury.

Childhood trauma influences adult pain. One of the more compelling narratives emerging in health care has to do with the influence that childhood developmental trauma can have on health, including pain. In his chapter on the impact of early life trauma on health and disease, Lanius states:¹²

“Women were 50% more likely than men to have experienced 5 or more categories of adverse childhood experiences. We believe that here is a key to what in mainstream epidemiology appears as women’s natural proneness to ill-defined health problems like fibromyalgia, chronic fatigue syndrome, obesity, irritable bowel syndrome, and chronic non-malignant pain syndromes. In light of our findings, we now see these as medical constructs, artifacts resulting from medical blindness to social realities and ignorance of the impact of gender.”

Brain activity in response to emotional insult mimics physical pain, and it is difficult to tell from images of brain activity whether a person is experiencing one or the other.

Lanius¹² suggests that adverse childhood experiences¹³ (trauma such as abuse and sexual assault) can lead to long-term changes within the nervous system, including areas of pain processing. My coauthor and I describe these changes here in terms of nervous system sensitization or dysregulation, and we believe that these changes lead to a bias toward hyperactivation of emotional pain circuits, which leads to the emotionally laden pain behaviors that often seem out of proportion to tissue pathology.

The care of people with pain has been wrought with ineffective and unnecessary treatment, including the misuse of opioids, largely because we do not have an accurate conceptualization of pain. The absence of animal and human models of central nervous system (CNS) pain processing ensures that our understanding of pain will remain incomplete for the foreseeable future, but enough evidence exists to help family physicians develop an understanding of pain that goes beyond what we learned in medical school and that can help us more effectively treat patients with pain.

In this review, we will briefly discuss the established concepts of nociceptive and neuropathic pain. And then, with those concepts in mind, we will explore a third type of pain that for lack of a better term, we will call “pain for psychological reasons.” We hypothesize that this pain may be the consequence of changes in nervous system function that arise from developmental trauma, other traumatic experiences in a patient’s life, or mental health disorders. It is this third type of pain that may offer us insights into conditions such as fibromyalgia.

While we do not yet have validated diagnostic criteria for this third type of pain, we believe that there is enough information to present initial criteria so that one may distinguish it from nociceptive and neuropathic pain.

Nociceptive and neuropathic pain: The current paradigm

Nociceptive pain. The sensory pain experience, or nociceptive pain, is produced by noxious stimuli that either damage, or are capable of damaging, tissues (eg, burns, cuts, fractures, inflammation, and increased pressure in a hollow viscus). Noxious stimuli are detected at the molecular level by specific pain sensory receptors embedded in our tissues called nociceptors.

The process by which noxious stimuli lead to the experience of sensory pain consists of 4 steps—transduction, transmission, modulation, and perception—which are described in “From periphery to brain: The process of nociceptive pain.”^1-4

Neuropathic pain. While nociceptive pain can be easily traced from a peripheral nociceptive fiber to the brain and typically resolves when the nociceptive stimulus stops, neuropathic pain (NPP) results from changes to the function of the nervous system and is typically caused by injury to the nerves. Such changes, referred to as neuronal sensitization, may not quickly resolve, as is the case with postherpetic neuralgia. In fact, the changes can become permanent. NPP fundamentally differs from nociceptive pain because it results from changes in the central processing of pain that can lead a person to perceive pain sensations even in the absence of tissue pathology.

This third type of pain may be the consequence of changes in nervous system function that arise from developmental trauma, other traumatic experiences in a patient's life, or mental health disorders.

Common causes of NPP that persists even after tissue damage has healed include trauma (eg, amputation of a limb), ischemia (eg, pressure palsy), disease (eg, the metabolic injury of diabetes or the injury caused by a shingles infection), and drug treatment (eg, chemotherapy). The underlying mechanisms of NPP and the neuronal plasticity (the ability of the nervous system to rewire itself) that initiate and then maintain NPP are important areas of active research that may eventually lead to the development of more effective treatments.

Timing is critical. Neuroplastic changes in the nervous system following nerve injury are time-dependent. Synaptic plasticity can occur within seconds to minutes, while cellular plasticity occurs within hours to days. Synaptic and cellular plasticity happen relatively fast and may be reversible.

In contrast, systems plasticity (when new CNS neuronal connections are formed in response to nerve injury) takes place over the months and years following nerve injury and is often irreversible. When we recognize NPP and intervene before system neuroplastic changes occur, it may be possible to prevent pain from becoming chronic (TABLE 1⁵). In cases of nerve injury, researchers have long suspected that early and aggressive pain treatment within the first few months that may include sympathetic and peripheral neural blockade reduces the likelihood that the patient will have chronic pain.^6,7

It’s time to update our understanding of pain

The International Association for the Study of Pain (IASP)—a group of health care providers, scientists, and policymakers seeking to improve pain relief worldwide—notes in its definition of pain that the complaint, “I hurt” does not necessarily imply that there is a painful stimulus in the form of tissue injury.⁸ Yet most of us have been taught to think of pain solely as the result of tissue pathology, and we assume that emotional factors merely modify how the physical damage is perceived. This traditional concept of pain is incomplete. It leads clinicians to misdiagnose the cause of pain, initiate expensive and unnecessary treatment, engage in well-meaning but misguided prescribing behavior, and miss opportunities to help patients.

Pain in the absence of any pathophysiologic cause or injury

The clinician’s search for a pain diagnosis is typically predicated on the notion that there must be an underlying tissue injury of severity equal to the severity of the patient’s pain complaints. This approach to a pain evaluation rests on 2 assumptions that are not true for all patients:

1. Pain is simply a sensory experience that is always caused by tissue damage of some type.
2. The severity of the pain experienced by a patient should be tightly bound to the severity of the pain stimulus (ie, tissue damage).

These assumptions are true of acute nociceptive pain, they may or may not be true for NPP, but they do not apply to the third type of pain—pain for psychological reasons. While tissue pathology in humans and animals with nociceptive pain is usually visible, measurable, and correlates with observed pain behaviors, the damage to nerve tissue and the ensuing changes in nervous system function with NPP are not always visible or able to be imaged. These changes produce pain that can appear more severe than expected based on a brief exam. Some of the time, however, characteristic symptoms and physical signs of NPP will be present, and perhaps electrodiagnostic or other tests will be abnormal, thus providing some objective sense of changes in nervous system function.

In contrast, pain behavior due to the third type of pain usually appears very much out of proportion, and unbound to, tissue pathology. Furthermore, the patient’s pain behaviors often reflect heightened emotional pain processing (TABLE 2⁹). The resulting emotionally charged presentation can be alarming and suggestive of extreme tissue injury, but there may be absolutely no evidence of tissue injury or pathology.

Functional change in the CNS

There is evidence from experimental studies that psychologic factors change nervous system function. In one review, the authors concluded, “Pain…can vary widely between people and even within an individual depending on…the psychological state of the person.”¹⁰ In a second review, the authors concluded that our emotional state has an enormous influence on pain; a negative emotional state increases pain, whereas a positive state lowers pain.¹¹

But can psychological factors induce long-term changes in nervous system function analogous to the systems neuroplasticity responsible for irreversible changes in NPP? And can psychologically induced changes in nervous system sensory processing lead to pain without any tissue or nerve damage?

We theorize that a functional change in the CNS can occur in response to certain emotional states or traumatic experiences (eg, child abuse, assault, accidents). (More on this in a bit.) When such changes occur, mildly painful stimuli are amplified and processed through overly sensitized, dysregulated, ramped-up emotional and somatosensory pain circuits in the brain. This is analogous to the functional changes in the nervous system that occur with NPP; however, when the nervous system changes are due to psychological factors, there may be no tissue or nerve injury.

Childhood trauma influences adult pain. One of the more compelling narratives emerging in health care has to do with the influence that childhood developmental trauma can have on health, including pain. In his chapter on the impact of early life trauma on health and disease, Lanius states:¹²

“Women were 50% more likely than men to have experienced 5 or more categories of adverse childhood experiences. We believe that here is a key to what in mainstream epidemiology appears as women’s natural proneness to ill-defined health problems like fibromyalgia, chronic fatigue syndrome, obesity, irritable bowel syndrome, and chronic non-malignant pain syndromes. In light of our findings, we now see these as medical constructs, artifacts resulting from medical blindness to social realities and ignorance of the impact of gender.”

Brain activity in response to emotional insult mimics physical pain, and it is difficult to tell from images of brain activity whether a person is experiencing one or the other.

Lanius¹² suggests that adverse childhood experiences¹³ (trauma such as abuse and sexual assault) can lead to long-term changes within the nervous system, including areas of pain processing. My coauthor and I describe these changes here in terms of nervous system sensitization or dysregulation, and we believe that these changes lead to a bias toward hyperactivation of emotional pain circuits, which leads to the emotionally laden pain behaviors that often seem out of proportion to tissue pathology.

A 40-year-old woman with hypertrophic   obstructive cardiomyopathy presents to the hematology clinic for a second opinion regarding a history of headaches and fatigue for the past 10 years. She has been diagnosed with idiopathic erythrocytosis, presumed to be due to polycythemia vera. She periodically undergoes phlebotomy to keep her hematocrit below 41%, and this markedly improves her headaches. She denies shortness of breath, cough, fever, weight loss, joint pain, and visual or other neurologic symptoms. She has never reported pruritus related to bathing or exposure to water.

She does not smoke, drink alcohol, or use illicit drugs. She works as a pharmacy technician. She says her father died of cancer (no further details available) and describes a family history of gastrointestinal malignancy in her grandfather and paternal aunt. She takes aspirin, metoprolol, and spironolactone for her cardiomyopathy.

Physical examination reveals generalized plethora, more marked on her cheeks and face, and mild bilateral pitting pedal edema. No lymphadenopathy or hepatosplenomegaly can be palpated. Other systems, including the cardiac, respiratory, and nervous systems, are normal.

ERYTHROCYTOSIS AND POLYCYTHEMIA VERA

1. In patients with erythrocytosis, which of the following is not characteristic of polycythemia vera?

• Erythromelalgia and postbathing pruritus
• Splenomegaly
• History of thrombosis
• Gout
• Hematuria

Erythrocytosis—an abnormally high concentration of red blood cells in the peripheral blood—is a laboratory finding. It often reflects an increase in the total quantity or mass of red blood cells in the body (polycythemia) but can sometimes be due to decreased plasma volume (spurious polycythemia).¹ Erythrocytosis can be caused by a number of diseases, hereditary and acquired, and can be classified as primary or secondary (Table 1).

Symptoms arise from an increase in the total blood volume and red blood cell mass, often leading to dilated capillaries and other blood vessels. Symptoms can occur regardless of the cause and classically include headache (often described as diffuse heaviness), dizziness, and a tendency for bleeding or thrombosis.² Symptoms are relieved when the hematocrit is lowered.

Several features in the history and physical examination of a patient being evaluated for erythrocytosis can suggest an underlying cause. Smoking, chronic respiratory insufficiency, and congenital cyanotic heart disease point to secondary erythrocytosis and can usually be identified at the outset. A history of occupational exposure to carbon monoxide (such as engine exhaust) should be elicited carefully. A family history of erythrocytosis should raise suspicion of a heritable condition such as a hemoglobinopathy associated with increased oxygen affinity or rare forms of primary erythrocytosis associated with endogenous overproduction of erythropoietin or activating mutations of the erythropoietin receptor.³ Iatrogenic causes such as androgen supplementation, erythropoietin abuse, and postrenal-transplant erythrocytosis should also be considered.

Secretion of erythropoietin or erythropoietinlike proteins by a malignant neoplasm is a rare but important cause of erythrocytosis. For example, renal cell carcinoma may present with erythrocytosis secondary to excessive erythropoietin production, and hematuria can be an early symptom.

Polycythemia vera

Polycythemia vera, a myeloproliferative neoplasm, is characterized by increased red blood cell production independent of the mechanisms that normally regulate erythropoiesis. The bone marrow shows a panmyelosis that is often accompanied by leukocytosis or thrombocytosis, or both, in the peripheral blood.

Symptoms such as severe itching after exposure to hot water (aquagenic pruritus) and periodic attacks of redness, swelling, and pain in the hands or feet, or both (erythromelalgia), have been described in patients with polycythemia vera. Splenomegaly is relatively common, seen in approximately two-thirds of patients.⁴ Hyperuricemia (from increased cell turnover) and gout are also associated with polycythemia vera, as is a history of arterial and venous thrombosis.⁵

Hematuria is not commonly seen in polycythemia vera, although bleeding from the bladder, vagina, or uterus has been described.

CASE RESUMED: INITIAL LABORATORY TESTS

Results of our patient’s initial laboratory tests are:

• Hemoglobin 16.9 g/dL (reference range 11.5–15.5)
• Hematocrit 48.8% (36.0–46.0)
• Mean corpuscular volume 85.2 fL (80–100)
• Platelet count 328 × 10⁹/L (150–400)
• White blood cell count 9.14 × 10⁹/L (3.7–11.0)
• Absolute neutrophil count 5.95 × 10⁹/L (1.45–7.5)
• Blood urea nitrogen 12 mg/dL (8–25)
• Creatinine 0.5 mg/dL (0.7–1.4)
• Lactate dehydrogenase 180 U/L (100–220)
• Uric acid 3.0 mg/dL (2.0–7.0)
• Thyroid-stimulating hormone 2.2 µU/mL (0.4–5.5).

The patient undergoes additional tests, including a serum erythropoietin level and hemoglobinopathy screen. Bone marrow aspiration and biopsy are performed, with cytogenetic analysis, chromosomal microarray analysis, and molecular testing for mutation of the Janus kinase 2 (JAK2) gene.

CONFIRMING SUSPECTED POLYCYTHEMIA VERA

2. In patients with suspected polycythemia vera, which of the following laboratory tests is most useful in making the diagnosis?

• Hemoglobin, hematocrit, and red blood cell mass
• Serum erythropoietin level
• Arterial blood gases with co-oximetry
• Testing for the JAK2 mutation
• Bone marrow aspiration and biopsy

The aim of the initial workup of erythrocytosis is to differentiate polycythemia vera from secondary causes of erythrocytosis.

Hemoglobin, hematocrit, red cell mass

Erythrocytosis is defined by an abnormal elevation in the hematocrit (> 48% in women or > 49% in men), hemoglobin concentration (> 16.0 g/dL in women or > 16.5 g/dL in men), or red blood cell mass. The red blood cell count should not be used as a surrogate for red blood cell mass, since some anemias (especially thalassemia minor) can be associated with an increase in the number of red blood cells but a low hemoglobin concentration.

Isotope dilution techniques to determine the red cell mass and plasma volume can differentiate true erythrocytosis from a spurious elevation due to a decrease in plasma volume.^6,7 However, this is an expensive, time-consuming test that is not widely available and so is rarely performed.⁸

JAK2 mutation testing

The initial evaluation of a patient with erythrocytosis has changed significantly in the past 10 years with the discovery of the JAK2 gene and its role in the pathogenesis of polycythemia vera and other myeloproliferative neoplasms.

JAK2, located at 9p24, codes for a tyrosine kinase important for signal transduction in hematopoietic cells. Mutations in this gene have been shown to promote hypersensitivity to cytokines, including erythropoietin.⁹ The most common somatic mutation occurs within exon 14 at base pair 1849 and results in a phenylalanine-for-valine amino acid substitution in the JAK2 protein, designated V617F. Less commonly, mutations occur elsewhere in exons 12 to 15, with more than 50 different mutations described; nonpolymorphic mutations are assumed to have biologic effects similar to those of V617F.

Taken together, the JAK2 V617F and non-V617F mutations have a diagnostic sensitivity of 98% to 100% for polycythemia vera. For practical purposes, this means that the presence of a JAK2 mutation can be used as a clonal marker to distinguish polycythemia vera from reactive secondary causes of erythrocytosis. A JAK2 mutation is one of three major diagnostic criteria for polycythemia vera in the 2016 revision to the 2008 World Health Organization criteria (Table 2).¹⁰ Of note, this mutation is not specific for polycythemia vera and can also be found in other myeloproliferative neoplasms, including primary myelofibrosis and essential thrombocythemia.

Absence of a JAK2 mutation makes polycythemia vera unlikely, so this test is most useful in making the diagnosis.

Serum erythropoietin

Serum erythropoietin testing can be very useful to distinguish polycythemia vera from secondary erythrocytosis. Low levels suggest polycythemia vera, while high levels are seen in secondary processes.¹¹

This test is best used along with JAK2 V617F mutation analysis as an initial step in evaluating patients with erythrocytosis. When JAK2 V617F mutation analysis is negative, a low serum erythropoietin level should prompt further testing for non-V617F JAK2 mutations, whereas a normal or elevated erythropoietin level should be evaluated further with tests to distinguish hereditary from acquired secondary causes of erythrocytosis.

Arterial blood gas analysis and co-oximetry

Arterial blood gas analysis can reveal hypoxemia, pointing to a cardiorespiratory process driving the erythrocytosis, whereas co-oximetry can be used to identify the presence and amount of carboxyhemoglobin in the blood.

Bone marrow biopsy

An increase in pleomorphic megakaryocytes in the bone marrow without stainable iron is often described as characteristic in polycythemia vera patients, but it is not diagnostic. Panmyelosis with increased cellularity is the norm but can be seen in other myeloproliferative neoplasms. The morphologic features of bone marrow are now included as one of the major diagnostic criteria for polycythemia vera (Table 2).

OUR PATIENT’S FURTHER WORKUP

Our patient’s erythropoietin level is 34.2 mIU/mL (reference range 4.7–28.6). Her oxygen saturation is 96%, and her carboxyhemoglobin level is 1.1% (0–5).

She undergoes bone marrow biopsy. Analysis finds that the marrow is normocellular (60%) with trilineage hematopoiesis and decreased stainable iron.

Cytogenetic analysis shows a 46,XX[20] karyotype. Chromosomal microarray analysis shows no pathogenic copy-number changes. There is no detectable JAK2 V617F or exon 12-to-15 mutation.

The patient’s erythrocytosis and abnormal hemoglobin electrophoresis study raise suspicion for a variant type of hemoglobin that has a higher affinity for oxygen than normal.

3. What is the next best step to evaluate this patient?

• Red-cell oxygen equilibrium curve to calculate the P50 (the partial pressure of oxygen that is required to saturate 50% of the hemoglobin.)
• High-performance liquid chromatography
• Globin gene DNA sequencing
• Testing 2,3-bisphosphoglycerate mutase (BPGM) activity

Nearly 200 mutational variants in alpha and beta globin chains that lead to an increased affinity of hemoglobin for oxygen have been reported.¹² While not all mutations are clinically significant, increased oxygen affinity variants can lead to impaired oxygen delivery to tissues, especially the kidneys, resulting in a physiologic increase in erythropoietin and erythrocytosis.

In patients being evaluated for a high-oxygen-affinity hemoglobinopathy, a two-step approach has been outlined.¹³ The first involves measuring the oxygen-binding properties of a freshly collected sample of blood by directly measuring the oxygen saturation of the hemoglobin and pO₂ using a co-oximeter. This information is used to create a red cell oxygen equilibrium curve and to calculate the P50. A low P50 correlates with an abnormally high affinity of hemoglobin for oxygen.

The second step is to identify the abnormal hemoglobin. High-performance liquid chromatography is now widely available as a screening test but does not detect all variants. For many years, sequencing of globin chain DNA has been a gold standard for identifying specific mutations. Subsequent to analyzing a catalog of known hemoglobin variants, mass spectrometry can serve as a screening and identification technique. Mass spectroscopy can also detect known rare variants with posttranslational modifications¹⁴ that are not recognized by DNA analysis. Mass spectroscopy and DNA sequencing are complementary techniques available only in specialized reference laboratories.

Erythrocytosis due to BPGM deficiency is very rare. Clinical and laboratory features mimic those of high-oxygen-affinity hemoglobin, but patients do not have a demonstrable mutation in alpha or beta globin genes. The level of BPGM is low, and the diagnosis is established by measuring BPGM levels and sequencing the BPGM gene.¹⁵

RESULTS OF THE ADDITIONAL WORKUP

In our patient, hemoglobin electrophoresis reveals an abnormal hemoglobin variant. High-performance liquid chromatography reveals an abnormal peak that comprises approximately 23.7% of the total hemoglobin, consistent with an alpha globin variant. Further characterization (using a sample of venous blood) shows an oxygen dissociation P50 of 22 mm Hg (normal 24–30 mm Hg) (Figure 1).

Mass spectrometry identifies the variant as hemoglobin Tarrant. This variant is characterized by a substitution of asparagine for aspartic acid at position 126 of the alpha globin chain, a known site of contact between the alpha 1 and beta 1 chains.¹⁶ It has been seen in patients of Hispanic heritage and clinically correlates with mild erythrocytosis. Indeed, this woman’s mother was from Mexico.

EDUCATING PATIENTS

4. What should patients know about their high-oxygen-affinity hemoglobinopathy?

• High altitudes and air travel can be risky
• Pregnancy may have adverse outcomes
• Systemic anticoagulation may lower the risk of venous thromboembolism
• Periodic phlebotomy may help control symptoms

Most patients with high-oxygen-affinity hemoglobin do not require specific clinical management but only counseling and education about their condition. Establishing an accurate diagnosis is important in order to avoid further inappropriate, invasive, and expensive testing.

Although exposure to high altitudes may be associated with decreased ambient oxygen levels, hypoxia is usually not a problem because of hemoglobin’s high affinity for oxygen.

Impaired delivery of oxygen across the placenta may be anticipated in a mother with high-oxygen-affinity hemoglobin, but this has not been observed clinically.¹⁷

Compared with patients with polycythemia vera, patients with high-oxygen-affinity hemoglobin have fewer complications from hyperviscosity and thrombosis, even with comparable degrees of erythrocytosis.

Although patients usually do not require treatment, phlebotomy may be helpful for symptoms that can be attributed to the higher hemoglobin concentration.

Our patient continues to be seen in clinic for periodic blood counts and phlebotomy for her headaches, as required.

HEMOGLOBIN: RELAXED OR TENSE

Normal adult hemoglobin is a tetramer composed of two pairs of globin polypeptide chains: alpha and beta (Figure 2). The intrinsic properties of the constituent globin chains and their allosteric conformation—as well as extrinsic factors including temperature, pH, and the binding of hydrogen ion and 2,3-BPG—play important roles in modifying the affinity of hemoglobin for oxygen. The major modulator of hemoglobin-oxygen affinity in human erythrocytes is 2,3-BPG.

The hemoglobin tetramer, consisting of two identical halves, alpha 1-beta 1 and alpha 2-beta 2, oscillates between two quaternary conformations, “relaxed” (fully oxygenated) and “tense” (fully deoxygenated).¹⁸ High-oxygen-affinity hemoglobins can result from factors that enhance the relaxed state, either by stabilizing the relaxed state or by destabilizing the tense state. Structural modifications in hemoglobin typically affect the main contacts involved in the transition from the deoxygenated to the oxygenated state, the 2,3-BPG binding sites, the heme pocket, or elongation of globin chains by various mutations. In hemoglobin Tarrant, the mutation prevents formation of noncovalent salt bridges in the alpha 1-beta 1 contact that normally stabilize the deoxygenated conformation of hemoglobin. As a result, the deoxygenated (tense) state is destabilized, shifting the allosteric equilibrium in favor of the oxygenated (relaxed) state with consequent high oxygen affinity.¹⁶

MORE ABOUT HIGH-OXYGEN-AFFINITY HEMOGLOBINS

The first case of erythrocytosis due to an abnormal hemoglobin was identified in 1966. This was an alpha chain variant with an arginine-to-leucine substitution at position 92, named hemoglobin Chesapeake.¹⁹

High-oxygen-affinity hemoglobin variants are usually transmitted as autosomal dominant traits. Patients are most often identified because of unexplained erythrocytosis detected on a routine blood cell count, as in our patient.

Not all high-oxygen-affinity hemoglobinopathies are associated with erythrocytosis. The degree of increased oxygen affinity may only be mild or the abnormal hemoglobin may be slightly unstable, thereby masking the usual clinical signs and symptoms.

Therapeutic phlebotomy should be used cautiously since it can decrease delivery of oxygen to tissues. A subset of patients whose symptoms are related to an elevated red cell mass may experience some relief, as did our patient.

A 40-year-old woman with hypertrophic   obstructive cardiomyopathy presents to the hematology clinic for a second opinion regarding a history of headaches and fatigue for the past 10 years. She has been diagnosed with idiopathic erythrocytosis, presumed to be due to polycythemia vera. She periodically undergoes phlebotomy to keep her hematocrit below 41%, and this markedly improves her headaches. She denies shortness of breath, cough, fever, weight loss, joint pain, and visual or other neurologic symptoms. She has never reported pruritus related to bathing or exposure to water.

She does not smoke, drink alcohol, or use illicit drugs. She works as a pharmacy technician. She says her father died of cancer (no further details available) and describes a family history of gastrointestinal malignancy in her grandfather and paternal aunt. She takes aspirin, metoprolol, and spironolactone for her cardiomyopathy.

Physical examination reveals generalized plethora, more marked on her cheeks and face, and mild bilateral pitting pedal edema. No lymphadenopathy or hepatosplenomegaly can be palpated. Other systems, including the cardiac, respiratory, and nervous systems, are normal.

ERYTHROCYTOSIS AND POLYCYTHEMIA VERA

1. In patients with erythrocytosis, which of the following is not characteristic of polycythemia vera?

• Erythromelalgia and postbathing pruritus
• Splenomegaly
• History of thrombosis
• Gout
• Hematuria

Erythrocytosis—an abnormally high concentration of red blood cells in the peripheral blood—is a laboratory finding. It often reflects an increase in the total quantity or mass of red blood cells in the body (polycythemia) but can sometimes be due to decreased plasma volume (spurious polycythemia).¹ Erythrocytosis can be caused by a number of diseases, hereditary and acquired, and can be classified as primary or secondary (Table 1).

Symptoms arise from an increase in the total blood volume and red blood cell mass, often leading to dilated capillaries and other blood vessels. Symptoms can occur regardless of the cause and classically include headache (often described as diffuse heaviness), dizziness, and a tendency for bleeding or thrombosis.² Symptoms are relieved when the hematocrit is lowered.

Several features in the history and physical examination of a patient being evaluated for erythrocytosis can suggest an underlying cause. Smoking, chronic respiratory insufficiency, and congenital cyanotic heart disease point to secondary erythrocytosis and can usually be identified at the outset. A history of occupational exposure to carbon monoxide (such as engine exhaust) should be elicited carefully. A family history of erythrocytosis should raise suspicion of a heritable condition such as a hemoglobinopathy associated with increased oxygen affinity or rare forms of primary erythrocytosis associated with endogenous overproduction of erythropoietin or activating mutations of the erythropoietin receptor.³ Iatrogenic causes such as androgen supplementation, erythropoietin abuse, and postrenal-transplant erythrocytosis should also be considered.

Secretion of erythropoietin or erythropoietinlike proteins by a malignant neoplasm is a rare but important cause of erythrocytosis. For example, renal cell carcinoma may present with erythrocytosis secondary to excessive erythropoietin production, and hematuria can be an early symptom.

Polycythemia vera

Polycythemia vera, a myeloproliferative neoplasm, is characterized by increased red blood cell production independent of the mechanisms that normally regulate erythropoiesis. The bone marrow shows a panmyelosis that is often accompanied by leukocytosis or thrombocytosis, or both, in the peripheral blood.

Symptoms such as severe itching after exposure to hot water (aquagenic pruritus) and periodic attacks of redness, swelling, and pain in the hands or feet, or both (erythromelalgia), have been described in patients with polycythemia vera. Splenomegaly is relatively common, seen in approximately two-thirds of patients.⁴ Hyperuricemia (from increased cell turnover) and gout are also associated with polycythemia vera, as is a history of arterial and venous thrombosis.⁵

Hematuria is not commonly seen in polycythemia vera, although bleeding from the bladder, vagina, or uterus has been described.

CASE RESUMED: INITIAL LABORATORY TESTS

Results of our patient’s initial laboratory tests are:

• Hemoglobin 16.9 g/dL (reference range 11.5–15.5)
• Hematocrit 48.8% (36.0–46.0)
• Mean corpuscular volume 85.2 fL (80–100)
• Platelet count 328 × 10⁹/L (150–400)
• White blood cell count 9.14 × 10⁹/L (3.7–11.0)
• Absolute neutrophil count 5.95 × 10⁹/L (1.45–7.5)
• Blood urea nitrogen 12 mg/dL (8–25)
• Creatinine 0.5 mg/dL (0.7–1.4)
• Lactate dehydrogenase 180 U/L (100–220)
• Uric acid 3.0 mg/dL (2.0–7.0)
• Thyroid-stimulating hormone 2.2 µU/mL (0.4–5.5).

The patient undergoes additional tests, including a serum erythropoietin level and hemoglobinopathy screen. Bone marrow aspiration and biopsy are performed, with cytogenetic analysis, chromosomal microarray analysis, and molecular testing for mutation of the Janus kinase 2 (JAK2) gene.

CONFIRMING SUSPECTED POLYCYTHEMIA VERA

2. In patients with suspected polycythemia vera, which of the following laboratory tests is most useful in making the diagnosis?

• Hemoglobin, hematocrit, and red blood cell mass
• Serum erythropoietin level
• Arterial blood gases with co-oximetry
• Testing for the JAK2 mutation
• Bone marrow aspiration and biopsy

The aim of the initial workup of erythrocytosis is to differentiate polycythemia vera from secondary causes of erythrocytosis.

Hemoglobin, hematocrit, red cell mass

Erythrocytosis is defined by an abnormal elevation in the hematocrit (> 48% in women or > 49% in men), hemoglobin concentration (> 16.0 g/dL in women or > 16.5 g/dL in men), or red blood cell mass. The red blood cell count should not be used as a surrogate for red blood cell mass, since some anemias (especially thalassemia minor) can be associated with an increase in the number of red blood cells but a low hemoglobin concentration.

Isotope dilution techniques to determine the red cell mass and plasma volume can differentiate true erythrocytosis from a spurious elevation due to a decrease in plasma volume.^6,7 However, this is an expensive, time-consuming test that is not widely available and so is rarely performed.⁸

JAK2 mutation testing

The initial evaluation of a patient with erythrocytosis has changed significantly in the past 10 years with the discovery of the JAK2 gene and its role in the pathogenesis of polycythemia vera and other myeloproliferative neoplasms.

JAK2, located at 9p24, codes for a tyrosine kinase important for signal transduction in hematopoietic cells. Mutations in this gene have been shown to promote hypersensitivity to cytokines, including erythropoietin.⁹ The most common somatic mutation occurs within exon 14 at base pair 1849 and results in a phenylalanine-for-valine amino acid substitution in the JAK2 protein, designated V617F. Less commonly, mutations occur elsewhere in exons 12 to 15, with more than 50 different mutations described; nonpolymorphic mutations are assumed to have biologic effects similar to those of V617F.

Taken together, the JAK2 V617F and non-V617F mutations have a diagnostic sensitivity of 98% to 100% for polycythemia vera. For practical purposes, this means that the presence of a JAK2 mutation can be used as a clonal marker to distinguish polycythemia vera from reactive secondary causes of erythrocytosis. A JAK2 mutation is one of three major diagnostic criteria for polycythemia vera in the 2016 revision to the 2008 World Health Organization criteria (Table 2).¹⁰ Of note, this mutation is not specific for polycythemia vera and can also be found in other myeloproliferative neoplasms, including primary myelofibrosis and essential thrombocythemia.

Absence of a JAK2 mutation makes polycythemia vera unlikely, so this test is most useful in making the diagnosis.

Serum erythropoietin

Serum erythropoietin testing can be very useful to distinguish polycythemia vera from secondary erythrocytosis. Low levels suggest polycythemia vera, while high levels are seen in secondary processes.¹¹

This test is best used along with JAK2 V617F mutation analysis as an initial step in evaluating patients with erythrocytosis. When JAK2 V617F mutation analysis is negative, a low serum erythropoietin level should prompt further testing for non-V617F JAK2 mutations, whereas a normal or elevated erythropoietin level should be evaluated further with tests to distinguish hereditary from acquired secondary causes of erythrocytosis.

Arterial blood gas analysis and co-oximetry

Arterial blood gas analysis can reveal hypoxemia, pointing to a cardiorespiratory process driving the erythrocytosis, whereas co-oximetry can be used to identify the presence and amount of carboxyhemoglobin in the blood.

Bone marrow biopsy

An increase in pleomorphic megakaryocytes in the bone marrow without stainable iron is often described as characteristic in polycythemia vera patients, but it is not diagnostic. Panmyelosis with increased cellularity is the norm but can be seen in other myeloproliferative neoplasms. The morphologic features of bone marrow are now included as one of the major diagnostic criteria for polycythemia vera (Table 2).

OUR PATIENT’S FURTHER WORKUP

Our patient’s erythropoietin level is 34.2 mIU/mL (reference range 4.7–28.6). Her oxygen saturation is 96%, and her carboxyhemoglobin level is 1.1% (0–5).

She undergoes bone marrow biopsy. Analysis finds that the marrow is normocellular (60%) with trilineage hematopoiesis and decreased stainable iron.

Cytogenetic analysis shows a 46,XX[20] karyotype. Chromosomal microarray analysis shows no pathogenic copy-number changes. There is no detectable JAK2 V617F or exon 12-to-15 mutation.

The patient’s erythrocytosis and abnormal hemoglobin electrophoresis study raise suspicion for a variant type of hemoglobin that has a higher affinity for oxygen than normal.

3. What is the next best step to evaluate this patient?

• Red-cell oxygen equilibrium curve to calculate the P50 (the partial pressure of oxygen that is required to saturate 50% of the hemoglobin.)
• High-performance liquid chromatography
• Globin gene DNA sequencing
• Testing 2,3-bisphosphoglycerate mutase (BPGM) activity

Nearly 200 mutational variants in alpha and beta globin chains that lead to an increased affinity of hemoglobin for oxygen have been reported.¹² While not all mutations are clinically significant, increased oxygen affinity variants can lead to impaired oxygen delivery to tissues, especially the kidneys, resulting in a physiologic increase in erythropoietin and erythrocytosis.

In patients being evaluated for a high-oxygen-affinity hemoglobinopathy, a two-step approach has been outlined.¹³ The first involves measuring the oxygen-binding properties of a freshly collected sample of blood by directly measuring the oxygen saturation of the hemoglobin and pO₂ using a co-oximeter. This information is used to create a red cell oxygen equilibrium curve and to calculate the P50. A low P50 correlates with an abnormally high affinity of hemoglobin for oxygen.

The second step is to identify the abnormal hemoglobin. High-performance liquid chromatography is now widely available as a screening test but does not detect all variants. For many years, sequencing of globin chain DNA has been a gold standard for identifying specific mutations. Subsequent to analyzing a catalog of known hemoglobin variants, mass spectrometry can serve as a screening and identification technique. Mass spectroscopy can also detect known rare variants with posttranslational modifications¹⁴ that are not recognized by DNA analysis. Mass spectroscopy and DNA sequencing are complementary techniques available only in specialized reference laboratories.

Erythrocytosis due to BPGM deficiency is very rare. Clinical and laboratory features mimic those of high-oxygen-affinity hemoglobin, but patients do not have a demonstrable mutation in alpha or beta globin genes. The level of BPGM is low, and the diagnosis is established by measuring BPGM levels and sequencing the BPGM gene.¹⁵

RESULTS OF THE ADDITIONAL WORKUP

In our patient, hemoglobin electrophoresis reveals an abnormal hemoglobin variant. High-performance liquid chromatography reveals an abnormal peak that comprises approximately 23.7% of the total hemoglobin, consistent with an alpha globin variant. Further characterization (using a sample of venous blood) shows an oxygen dissociation P50 of 22 mm Hg (normal 24–30 mm Hg) (Figure 1).

Mass spectrometry identifies the variant as hemoglobin Tarrant. This variant is characterized by a substitution of asparagine for aspartic acid at position 126 of the alpha globin chain, a known site of contact between the alpha 1 and beta 1 chains.¹⁶ It has been seen in patients of Hispanic heritage and clinically correlates with mild erythrocytosis. Indeed, this woman’s mother was from Mexico.

EDUCATING PATIENTS

4. What should patients know about their high-oxygen-affinity hemoglobinopathy?

• High altitudes and air travel can be risky
• Pregnancy may have adverse outcomes
• Systemic anticoagulation may lower the risk of venous thromboembolism
• Periodic phlebotomy may help control symptoms

Most patients with high-oxygen-affinity hemoglobin do not require specific clinical management but only counseling and education about their condition. Establishing an accurate diagnosis is important in order to avoid further inappropriate, invasive, and expensive testing.

Although exposure to high altitudes may be associated with decreased ambient oxygen levels, hypoxia is usually not a problem because of hemoglobin’s high affinity for oxygen.

Impaired delivery of oxygen across the placenta may be anticipated in a mother with high-oxygen-affinity hemoglobin, but this has not been observed clinically.¹⁷

Compared with patients with polycythemia vera, patients with high-oxygen-affinity hemoglobin have fewer complications from hyperviscosity and thrombosis, even with comparable degrees of erythrocytosis.

Although patients usually do not require treatment, phlebotomy may be helpful for symptoms that can be attributed to the higher hemoglobin concentration.

Our patient continues to be seen in clinic for periodic blood counts and phlebotomy for her headaches, as required.

HEMOGLOBIN: RELAXED OR TENSE

Normal adult hemoglobin is a tetramer composed of two pairs of globin polypeptide chains: alpha and beta (Figure 2). The intrinsic properties of the constituent globin chains and their allosteric conformation—as well as extrinsic factors including temperature, pH, and the binding of hydrogen ion and 2,3-BPG—play important roles in modifying the affinity of hemoglobin for oxygen. The major modulator of hemoglobin-oxygen affinity in human erythrocytes is 2,3-BPG.

The hemoglobin tetramer, consisting of two identical halves, alpha 1-beta 1 and alpha 2-beta 2, oscillates between two quaternary conformations, “relaxed” (fully oxygenated) and “tense” (fully deoxygenated).¹⁸ High-oxygen-affinity hemoglobins can result from factors that enhance the relaxed state, either by stabilizing the relaxed state or by destabilizing the tense state. Structural modifications in hemoglobin typically affect the main contacts involved in the transition from the deoxygenated to the oxygenated state, the 2,3-BPG binding sites, the heme pocket, or elongation of globin chains by various mutations. In hemoglobin Tarrant, the mutation prevents formation of noncovalent salt bridges in the alpha 1-beta 1 contact that normally stabilize the deoxygenated conformation of hemoglobin. As a result, the deoxygenated (tense) state is destabilized, shifting the allosteric equilibrium in favor of the oxygenated (relaxed) state with consequent high oxygen affinity.¹⁶

MORE ABOUT HIGH-OXYGEN-AFFINITY HEMOGLOBINS

The first case of erythrocytosis due to an abnormal hemoglobin was identified in 1966. This was an alpha chain variant with an arginine-to-leucine substitution at position 92, named hemoglobin Chesapeake.¹⁹

High-oxygen-affinity hemoglobin variants are usually transmitted as autosomal dominant traits. Patients are most often identified because of unexplained erythrocytosis detected on a routine blood cell count, as in our patient.

Not all high-oxygen-affinity hemoglobinopathies are associated with erythrocytosis. The degree of increased oxygen affinity may only be mild or the abnormal hemoglobin may be slightly unstable, thereby masking the usual clinical signs and symptoms.

Therapeutic phlebotomy should be used cautiously since it can decrease delivery of oxygen to tissues. A subset of patients whose symptoms are related to an elevated red cell mass may experience some relief, as did our patient.

A 40-year-old woman with hypertrophic   obstructive cardiomyopathy presents to the hematology clinic for a second opinion regarding a history of headaches and fatigue for the past 10 years. She has been diagnosed with idiopathic erythrocytosis, presumed to be due to polycythemia vera. She periodically undergoes phlebotomy to keep her hematocrit below 41%, and this markedly improves her headaches. She denies shortness of breath, cough, fever, weight loss, joint pain, and visual or other neurologic symptoms. She has never reported pruritus related to bathing or exposure to water.

She does not smoke, drink alcohol, or use illicit drugs. She works as a pharmacy technician. She says her father died of cancer (no further details available) and describes a family history of gastrointestinal malignancy in her grandfather and paternal aunt. She takes aspirin, metoprolol, and spironolactone for her cardiomyopathy.

Physical examination reveals generalized plethora, more marked on her cheeks and face, and mild bilateral pitting pedal edema. No lymphadenopathy or hepatosplenomegaly can be palpated. Other systems, including the cardiac, respiratory, and nervous systems, are normal.

ERYTHROCYTOSIS AND POLYCYTHEMIA VERA

1. In patients with erythrocytosis, which of the following is not characteristic of polycythemia vera?

• Erythromelalgia and postbathing pruritus
• Splenomegaly
• History of thrombosis
• Gout
• Hematuria

Erythrocytosis—an abnormally high concentration of red blood cells in the peripheral blood—is a laboratory finding. It often reflects an increase in the total quantity or mass of red blood cells in the body (polycythemia) but can sometimes be due to decreased plasma volume (spurious polycythemia).¹ Erythrocytosis can be caused by a number of diseases, hereditary and acquired, and can be classified as primary or secondary (Table 1).

Symptoms arise from an increase in the total blood volume and red blood cell mass, often leading to dilated capillaries and other blood vessels. Symptoms can occur regardless of the cause and classically include headache (often described as diffuse heaviness), dizziness, and a tendency for bleeding or thrombosis.² Symptoms are relieved when the hematocrit is lowered.

Several features in the history and physical examination of a patient being evaluated for erythrocytosis can suggest an underlying cause. Smoking, chronic respiratory insufficiency, and congenital cyanotic heart disease point to secondary erythrocytosis and can usually be identified at the outset. A history of occupational exposure to carbon monoxide (such as engine exhaust) should be elicited carefully. A family history of erythrocytosis should raise suspicion of a heritable condition such as a hemoglobinopathy associated with increased oxygen affinity or rare forms of primary erythrocytosis associated with endogenous overproduction of erythropoietin or activating mutations of the erythropoietin receptor.³ Iatrogenic causes such as androgen supplementation, erythropoietin abuse, and postrenal-transplant erythrocytosis should also be considered.

Secretion of erythropoietin or erythropoietinlike proteins by a malignant neoplasm is a rare but important cause of erythrocytosis. For example, renal cell carcinoma may present with erythrocytosis secondary to excessive erythropoietin production, and hematuria can be an early symptom.

Polycythemia vera

Polycythemia vera, a myeloproliferative neoplasm, is characterized by increased red blood cell production independent of the mechanisms that normally regulate erythropoiesis. The bone marrow shows a panmyelosis that is often accompanied by leukocytosis or thrombocytosis, or both, in the peripheral blood.

Symptoms such as severe itching after exposure to hot water (aquagenic pruritus) and periodic attacks of redness, swelling, and pain in the hands or feet, or both (erythromelalgia), have been described in patients with polycythemia vera. Splenomegaly is relatively common, seen in approximately two-thirds of patients.⁴ Hyperuricemia (from increased cell turnover) and gout are also associated with polycythemia vera, as is a history of arterial and venous thrombosis.⁵

Hematuria is not commonly seen in polycythemia vera, although bleeding from the bladder, vagina, or uterus has been described.

CASE RESUMED: INITIAL LABORATORY TESTS

Results of our patient’s initial laboratory tests are:

• Hemoglobin 16.9 g/dL (reference range 11.5–15.5)
• Hematocrit 48.8% (36.0–46.0)
• Mean corpuscular volume 85.2 fL (80–100)
• Platelet count 328 × 10⁹/L (150–400)
• White blood cell count 9.14 × 10⁹/L (3.7–11.0)
• Absolute neutrophil count 5.95 × 10⁹/L (1.45–7.5)
• Blood urea nitrogen 12 mg/dL (8–25)
• Creatinine 0.5 mg/dL (0.7–1.4)
• Lactate dehydrogenase 180 U/L (100–220)
• Uric acid 3.0 mg/dL (2.0–7.0)
• Thyroid-stimulating hormone 2.2 µU/mL (0.4–5.5).

The patient undergoes additional tests, including a serum erythropoietin level and hemoglobinopathy screen. Bone marrow aspiration and biopsy are performed, with cytogenetic analysis, chromosomal microarray analysis, and molecular testing for mutation of the Janus kinase 2 (JAK2) gene.

CONFIRMING SUSPECTED POLYCYTHEMIA VERA

2. In patients with suspected polycythemia vera, which of the following laboratory tests is most useful in making the diagnosis?

• Hemoglobin, hematocrit, and red blood cell mass
• Serum erythropoietin level
• Arterial blood gases with co-oximetry
• Testing for the JAK2 mutation
• Bone marrow aspiration and biopsy

The aim of the initial workup of erythrocytosis is to differentiate polycythemia vera from secondary causes of erythrocytosis.

Hemoglobin, hematocrit, red cell mass

Erythrocytosis is defined by an abnormal elevation in the hematocrit (> 48% in women or > 49% in men), hemoglobin concentration (> 16.0 g/dL in women or > 16.5 g/dL in men), or red blood cell mass. The red blood cell count should not be used as a surrogate for red blood cell mass, since some anemias (especially thalassemia minor) can be associated with an increase in the number of red blood cells but a low hemoglobin concentration.

Isotope dilution techniques to determine the red cell mass and plasma volume can differentiate true erythrocytosis from a spurious elevation due to a decrease in plasma volume.^6,7 However, this is an expensive, time-consuming test that is not widely available and so is rarely performed.⁸

JAK2 mutation testing

The initial evaluation of a patient with erythrocytosis has changed significantly in the past 10 years with the discovery of the JAK2 gene and its role in the pathogenesis of polycythemia vera and other myeloproliferative neoplasms.

JAK2, located at 9p24, codes for a tyrosine kinase important for signal transduction in hematopoietic cells. Mutations in this gene have been shown to promote hypersensitivity to cytokines, including erythropoietin.⁹ The most common somatic mutation occurs within exon 14 at base pair 1849 and results in a phenylalanine-for-valine amino acid substitution in the JAK2 protein, designated V617F. Less commonly, mutations occur elsewhere in exons 12 to 15, with more than 50 different mutations described; nonpolymorphic mutations are assumed to have biologic effects similar to those of V617F.

Taken together, the JAK2 V617F and non-V617F mutations have a diagnostic sensitivity of 98% to 100% for polycythemia vera. For practical purposes, this means that the presence of a JAK2 mutation can be used as a clonal marker to distinguish polycythemia vera from reactive secondary causes of erythrocytosis. A JAK2 mutation is one of three major diagnostic criteria for polycythemia vera in the 2016 revision to the 2008 World Health Organization criteria (Table 2).¹⁰ Of note, this mutation is not specific for polycythemia vera and can also be found in other myeloproliferative neoplasms, including primary myelofibrosis and essential thrombocythemia.

Absence of a JAK2 mutation makes polycythemia vera unlikely, so this test is most useful in making the diagnosis.

Serum erythropoietin

Serum erythropoietin testing can be very useful to distinguish polycythemia vera from secondary erythrocytosis. Low levels suggest polycythemia vera, while high levels are seen in secondary processes.¹¹

This test is best used along with JAK2 V617F mutation analysis as an initial step in evaluating patients with erythrocytosis. When JAK2 V617F mutation analysis is negative, a low serum erythropoietin level should prompt further testing for non-V617F JAK2 mutations, whereas a normal or elevated erythropoietin level should be evaluated further with tests to distinguish hereditary from acquired secondary causes of erythrocytosis.

Arterial blood gas analysis and co-oximetry

Arterial blood gas analysis can reveal hypoxemia, pointing to a cardiorespiratory process driving the erythrocytosis, whereas co-oximetry can be used to identify the presence and amount of carboxyhemoglobin in the blood.

Bone marrow biopsy

An increase in pleomorphic megakaryocytes in the bone marrow without stainable iron is often described as characteristic in polycythemia vera patients, but it is not diagnostic. Panmyelosis with increased cellularity is the norm but can be seen in other myeloproliferative neoplasms. The morphologic features of bone marrow are now included as one of the major diagnostic criteria for polycythemia vera (Table 2).

OUR PATIENT’S FURTHER WORKUP

Our patient’s erythropoietin level is 34.2 mIU/mL (reference range 4.7–28.6). Her oxygen saturation is 96%, and her carboxyhemoglobin level is 1.1% (0–5).

She undergoes bone marrow biopsy. Analysis finds that the marrow is normocellular (60%) with trilineage hematopoiesis and decreased stainable iron.

Cytogenetic analysis shows a 46,XX[20] karyotype. Chromosomal microarray analysis shows no pathogenic copy-number changes. There is no detectable JAK2 V617F or exon 12-to-15 mutation.

The patient’s erythrocytosis and abnormal hemoglobin electrophoresis study raise suspicion for a variant type of hemoglobin that has a higher affinity for oxygen than normal.

3. What is the next best step to evaluate this patient?

• Red-cell oxygen equilibrium curve to calculate the P50 (the partial pressure of oxygen that is required to saturate 50% of the hemoglobin.)
• High-performance liquid chromatography
• Globin gene DNA sequencing
• Testing 2,3-bisphosphoglycerate mutase (BPGM) activity

Nearly 200 mutational variants in alpha and beta globin chains that lead to an increased affinity of hemoglobin for oxygen have been reported.¹² While not all mutations are clinically significant, increased oxygen affinity variants can lead to impaired oxygen delivery to tissues, especially the kidneys, resulting in a physiologic increase in erythropoietin and erythrocytosis.

In patients being evaluated for a high-oxygen-affinity hemoglobinopathy, a two-step approach has been outlined.¹³ The first involves measuring the oxygen-binding properties of a freshly collected sample of blood by directly measuring the oxygen saturation of the hemoglobin and pO₂ using a co-oximeter. This information is used to create a red cell oxygen equilibrium curve and to calculate the P50. A low P50 correlates with an abnormally high affinity of hemoglobin for oxygen.

The second step is to identify the abnormal hemoglobin. High-performance liquid chromatography is now widely available as a screening test but does not detect all variants. For many years, sequencing of globin chain DNA has been a gold standard for identifying specific mutations. Subsequent to analyzing a catalog of known hemoglobin variants, mass spectrometry can serve as a screening and identification technique. Mass spectroscopy can also detect known rare variants with posttranslational modifications¹⁴ that are not recognized by DNA analysis. Mass spectroscopy and DNA sequencing are complementary techniques available only in specialized reference laboratories.

Erythrocytosis due to BPGM deficiency is very rare. Clinical and laboratory features mimic those of high-oxygen-affinity hemoglobin, but patients do not have a demonstrable mutation in alpha or beta globin genes. The level of BPGM is low, and the diagnosis is established by measuring BPGM levels and sequencing the BPGM gene.¹⁵

RESULTS OF THE ADDITIONAL WORKUP

In our patient, hemoglobin electrophoresis reveals an abnormal hemoglobin variant. High-performance liquid chromatography reveals an abnormal peak that comprises approximately 23.7% of the total hemoglobin, consistent with an alpha globin variant. Further characterization (using a sample of venous blood) shows an oxygen dissociation P50 of 22 mm Hg (normal 24–30 mm Hg) (Figure 1).

Mass spectrometry identifies the variant as hemoglobin Tarrant. This variant is characterized by a substitution of asparagine for aspartic acid at position 126 of the alpha globin chain, a known site of contact between the alpha 1 and beta 1 chains.¹⁶ It has been seen in patients of Hispanic heritage and clinically correlates with mild erythrocytosis. Indeed, this woman’s mother was from Mexico.

EDUCATING PATIENTS

4. What should patients know about their high-oxygen-affinity hemoglobinopathy?

• High altitudes and air travel can be risky
• Pregnancy may have adverse outcomes
• Systemic anticoagulation may lower the risk of venous thromboembolism
• Periodic phlebotomy may help control symptoms

Most patients with high-oxygen-affinity hemoglobin do not require specific clinical management but only counseling and education about their condition. Establishing an accurate diagnosis is important in order to avoid further inappropriate, invasive, and expensive testing.

Although exposure to high altitudes may be associated with decreased ambient oxygen levels, hypoxia is usually not a problem because of hemoglobin’s high affinity for oxygen.

Impaired delivery of oxygen across the placenta may be anticipated in a mother with high-oxygen-affinity hemoglobin, but this has not been observed clinically.¹⁷

Compared with patients with polycythemia vera, patients with high-oxygen-affinity hemoglobin have fewer complications from hyperviscosity and thrombosis, even with comparable degrees of erythrocytosis.

Although patients usually do not require treatment, phlebotomy may be helpful for symptoms that can be attributed to the higher hemoglobin concentration.

Our patient continues to be seen in clinic for periodic blood counts and phlebotomy for her headaches, as required.

HEMOGLOBIN: RELAXED OR TENSE

Normal adult hemoglobin is a tetramer composed of two pairs of globin polypeptide chains: alpha and beta (Figure 2). The intrinsic properties of the constituent globin chains and their allosteric conformation—as well as extrinsic factors including temperature, pH, and the binding of hydrogen ion and 2,3-BPG—play important roles in modifying the affinity of hemoglobin for oxygen. The major modulator of hemoglobin-oxygen affinity in human erythrocytes is 2,3-BPG.

The hemoglobin tetramer, consisting of two identical halves, alpha 1-beta 1 and alpha 2-beta 2, oscillates between two quaternary conformations, “relaxed” (fully oxygenated) and “tense” (fully deoxygenated).¹⁸ High-oxygen-affinity hemoglobins can result from factors that enhance the relaxed state, either by stabilizing the relaxed state or by destabilizing the tense state. Structural modifications in hemoglobin typically affect the main contacts involved in the transition from the deoxygenated to the oxygenated state, the 2,3-BPG binding sites, the heme pocket, or elongation of globin chains by various mutations. In hemoglobin Tarrant, the mutation prevents formation of noncovalent salt bridges in the alpha 1-beta 1 contact that normally stabilize the deoxygenated conformation of hemoglobin. As a result, the deoxygenated (tense) state is destabilized, shifting the allosteric equilibrium in favor of the oxygenated (relaxed) state with consequent high oxygen affinity.¹⁶

MORE ABOUT HIGH-OXYGEN-AFFINITY HEMOGLOBINS

The first case of erythrocytosis due to an abnormal hemoglobin was identified in 1966. This was an alpha chain variant with an arginine-to-leucine substitution at position 92, named hemoglobin Chesapeake.¹⁹

High-oxygen-affinity hemoglobin variants are usually transmitted as autosomal dominant traits. Patients are most often identified because of unexplained erythrocytosis detected on a routine blood cell count, as in our patient.

Not all high-oxygen-affinity hemoglobinopathies are associated with erythrocytosis. The degree of increased oxygen affinity may only be mild or the abnormal hemoglobin may be slightly unstable, thereby masking the usual clinical signs and symptoms.

Therapeutic phlebotomy should be used cautiously since it can decrease delivery of oxygen to tissues. A subset of patients whose symptoms are related to an elevated red cell mass may experience some relief, as did our patient.

Conventional treatment for a large symptomatic Zenker diverticulum is to surgically either remove it (diverticulectomy) or obliterate it by repositioning and securing it after cutting into the cricopharyngeus muscle (diverticulopexy with cricopharyngeal myotomy). But high rates of complications with these procedures in elderly patients have led to the development of endoscopic procedures that are safer and have similar or higher success rates.

ESOPHAGEAL POUCH IN SPACE BETWEEN MUSCLES

Zenker diverticulum is an outpouching of the esophageal mucosa into the potential space of the Killian triangle, the area between the inferior pharyngeal constrictor and the cricopharyngeus muscle. Increased cricopharyngeal tone coupled with insufficient relaxation results in a pressure gradient that eventually causes esophageal outpouchings. The problem is induced by age-related fibrosis and atrophy, esophageal spasms related to gastroesophageal reflux disease, and idiopathic cricopharyngeal spasms.¹

USUALLY PRESENTS WITH DYSPHAGIA

Zenker diverticulum has an estimated incidence of about 2 per 100,000 per year, but this is likely low because many cases are asymptomatic. It occurs mostly in men and people of Northern European descent in their 60s and 70s and rarely before age 40.²

About 80% to 90% of patients present with dysphagia. Other signs and symptoms include regurgitation of undigested foods and medications, halitosis, hoarseness, chronic cough, and aspiration.

Rare complications of Zenker diverticulum include recurrent pulmonary infection, tracheal fistula, ulceration, hemorrhage, squamous cell carcinoma (incidence 0.4% to 1.5%), vocal cord paralysis, and fistula to the prevertebral ligament with cervical osteomyelitis.¹

The diagnosis is suspected based on symptoms but should be confirmed with a barium esophagram that shows an outpouching of contrast from the main contrast column.

A watch-and-wait approach should be used for patients with mild symptoms (ie, mild or intermittent dysphagia) and minor functional limitations. Patients should be counseled to eat small amounts of food at a time, to chew thoroughly, and to sip liquids between bites.

TREATMENT IS SURGERY OR ENDOSCOPY

Patients with more than mild symptoms who are candidates for intervention should be offered treatment. The goal is to open the septum between the diverticulum and the main esophageal lumen so that food can be propelled from the hypopharynx to the main esophageal lumen without obstruction. Because increased cricopharyngeal tone plays a large role in the development and propagation of a diverticulum, most experts recommend cricopharyngomyotomy in addition to any treatment strategy.

Open surgery

In centers that do not offer flexible endoscopy, and for patients with a large diverticulum, open surgery is the sole option. It is done under general anesthesia with a left cervical approach. The length of the cricopharyngeal myotomy can vary from 2 to 6 cm.^3,4 This is followed by one of three options:

• Diverticulectomy
• Diverticulopexy, in which the diverticulum is repositioned and sutured against the prevertebral fascia
• Diverticular inversion, in which the diverticulum is inverted into the esophageal lumen and then oversewn.⁴

Success rates for open surgery vary from study to study but are usually around 90%.⁴ Complications are reported in 10% to 30% of cases and include mediastinitis, severe recurrent laryngeal nerve injury, and a 1% to 2% chance of death. These rates seem high, but the older age of most of these patients puts them at high risk.⁵

Endoscopic procedures

All endoscopic procedures involve incision of the wall separating the diverticulum from the esophageal lumen to relieve the obstruction.

Rigid endoscopy is used in centers that do not offer flexible endoscopy for patients who are not candidates for surgery. It is done under general anesthesia. With the patient’s neck completely hyperextended, a rigid endoscope is passed into the oral cavity, and diverticulotomy and cricopharyngeal myotomy are performed.

This technique has been extensively evaluated and, although effective, carries up to an 8% risk of complications, including perforation. It is not recommended for patients with a small diverticulum, a high body mass index, or difficulty with neck extension.⁶

Flexible endoscopy is the first-line therapy for most patients in many centers. With the flexible endoscope, the diverticulotomy and the cricopharyngeal myotomy can be performed in an outpatient endoscopy suite, and general anesthesia is not required.³

Initial studies of outcomes using these approaches have been promising,^3,7–9 with substantial reduction in dysphagia, regurgitation, and chronic cough. Unfortunately, this technique is associated with a recurrence rate as high as 25%. However, repeat endoscopic therapy in patients with recurrence results in success rates similar to those for first-time treatment.

Complication rates are low. In a series of 150 patients, four adverse events occurred—three cases of fever and one of pneumonia. In addition, one patient had subcutaneous emphysema that resolved spontaneously.⁷

Improved device on horizon

A potentially major advance in the endoscopic treatment of Zenker diverticulum is the development of a diverticuloscope, a soft tube with a V-shaped end. This tube is inserted through the mouth with one leg of the “V” protecting the anterior esophageal wall and the other leg protecting the posterior diverticular lumen. The endoscope is passed through this tube to perform the diverticulotomy. In one report,¹⁰ diverticuloscope-guided diverticulotomy led to fewer complications, shorter procedural time, and higher success rates compared with the standard technique.¹⁰ This device is not yet available in the United States.

No head-to-head comparison of outcomes with various flexible endoscopic techniques for treating Zenker diverticulum has been published, nor are there data yet on the success of surgery if endoscopic therapy fails.

We recommend flexible endoscopy for the initial treatment of Zenker diverticulum in most patients, but its availability is limited in the United States, as many practitioners do not have adequate experience with the technique.

TAKE-HOME POINTS

• Zenker diverticulum is an outpouching of the esophageal mucosa. It is often asymptomatic. Patients with minimal symptoms and with no functional impairment can be followed conservatively.
• A head-to-head comparison of surgical and endoscopic therapy has not yet been done, and optimal patient selection criteria are lacking. Therefore, treatment recommendations are based primarily on expert opinion.
• We recommend flexible endoscopic therapy as initial treatment in patients with a small diverticulum, but this may be limited by the unavailability of a surgeon experienced in this technique.
• For a large diverticulum, we recommend an open cervical procedure involving excision of the diverticulum or a diverticulopexy for patients who are good surgical candidates.

Conventional treatment for a large symptomatic Zenker diverticulum is to surgically either remove it (diverticulectomy) or obliterate it by repositioning and securing it after cutting into the cricopharyngeus muscle (diverticulopexy with cricopharyngeal myotomy). But high rates of complications with these procedures in elderly patients have led to the development of endoscopic procedures that are safer and have similar or higher success rates.

ESOPHAGEAL POUCH IN SPACE BETWEEN MUSCLES

Zenker diverticulum is an outpouching of the esophageal mucosa into the potential space of the Killian triangle, the area between the inferior pharyngeal constrictor and the cricopharyngeus muscle. Increased cricopharyngeal tone coupled with insufficient relaxation results in a pressure gradient that eventually causes esophageal outpouchings. The problem is induced by age-related fibrosis and atrophy, esophageal spasms related to gastroesophageal reflux disease, and idiopathic cricopharyngeal spasms.¹

USUALLY PRESENTS WITH DYSPHAGIA

Zenker diverticulum has an estimated incidence of about 2 per 100,000 per year, but this is likely low because many cases are asymptomatic. It occurs mostly in men and people of Northern European descent in their 60s and 70s and rarely before age 40.²

About 80% to 90% of patients present with dysphagia. Other signs and symptoms include regurgitation of undigested foods and medications, halitosis, hoarseness, chronic cough, and aspiration.

Rare complications of Zenker diverticulum include recurrent pulmonary infection, tracheal fistula, ulceration, hemorrhage, squamous cell carcinoma (incidence 0.4% to 1.5%), vocal cord paralysis, and fistula to the prevertebral ligament with cervical osteomyelitis.¹

The diagnosis is suspected based on symptoms but should be confirmed with a barium esophagram that shows an outpouching of contrast from the main contrast column.

A watch-and-wait approach should be used for patients with mild symptoms (ie, mild or intermittent dysphagia) and minor functional limitations. Patients should be counseled to eat small amounts of food at a time, to chew thoroughly, and to sip liquids between bites.

TREATMENT IS SURGERY OR ENDOSCOPY

Patients with more than mild symptoms who are candidates for intervention should be offered treatment. The goal is to open the septum between the diverticulum and the main esophageal lumen so that food can be propelled from the hypopharynx to the main esophageal lumen without obstruction. Because increased cricopharyngeal tone plays a large role in the development and propagation of a diverticulum, most experts recommend cricopharyngomyotomy in addition to any treatment strategy.

Open surgery

In centers that do not offer flexible endoscopy, and for patients with a large diverticulum, open surgery is the sole option. It is done under general anesthesia with a left cervical approach. The length of the cricopharyngeal myotomy can vary from 2 to 6 cm.^3,4 This is followed by one of three options:

• Diverticulectomy
• Diverticulopexy, in which the diverticulum is repositioned and sutured against the prevertebral fascia
• Diverticular inversion, in which the diverticulum is inverted into the esophageal lumen and then oversewn.⁴

Success rates for open surgery vary from study to study but are usually around 90%.⁴ Complications are reported in 10% to 30% of cases and include mediastinitis, severe recurrent laryngeal nerve injury, and a 1% to 2% chance of death. These rates seem high, but the older age of most of these patients puts them at high risk.⁵

Endoscopic procedures

All endoscopic procedures involve incision of the wall separating the diverticulum from the esophageal lumen to relieve the obstruction.

Rigid endoscopy is used in centers that do not offer flexible endoscopy for patients who are not candidates for surgery. It is done under general anesthesia. With the patient’s neck completely hyperextended, a rigid endoscope is passed into the oral cavity, and diverticulotomy and cricopharyngeal myotomy are performed.

This technique has been extensively evaluated and, although effective, carries up to an 8% risk of complications, including perforation. It is not recommended for patients with a small diverticulum, a high body mass index, or difficulty with neck extension.⁶

Flexible endoscopy is the first-line therapy for most patients in many centers. With the flexible endoscope, the diverticulotomy and the cricopharyngeal myotomy can be performed in an outpatient endoscopy suite, and general anesthesia is not required.³

Initial studies of outcomes using these approaches have been promising,^3,7–9 with substantial reduction in dysphagia, regurgitation, and chronic cough. Unfortunately, this technique is associated with a recurrence rate as high as 25%. However, repeat endoscopic therapy in patients with recurrence results in success rates similar to those for first-time treatment.

Complication rates are low. In a series of 150 patients, four adverse events occurred—three cases of fever and one of pneumonia. In addition, one patient had subcutaneous emphysema that resolved spontaneously.⁷

Improved device on horizon

A potentially major advance in the endoscopic treatment of Zenker diverticulum is the development of a diverticuloscope, a soft tube with a V-shaped end. This tube is inserted through the mouth with one leg of the “V” protecting the anterior esophageal wall and the other leg protecting the posterior diverticular lumen. The endoscope is passed through this tube to perform the diverticulotomy. In one report,¹⁰ diverticuloscope-guided diverticulotomy led to fewer complications, shorter procedural time, and higher success rates compared with the standard technique.¹⁰ This device is not yet available in the United States.

No head-to-head comparison of outcomes with various flexible endoscopic techniques for treating Zenker diverticulum has been published, nor are there data yet on the success of surgery if endoscopic therapy fails.

We recommend flexible endoscopy for the initial treatment of Zenker diverticulum in most patients, but its availability is limited in the United States, as many practitioners do not have adequate experience with the technique.

TAKE-HOME POINTS

• Zenker diverticulum is an outpouching of the esophageal mucosa. It is often asymptomatic. Patients with minimal symptoms and with no functional impairment can be followed conservatively.
• A head-to-head comparison of surgical and endoscopic therapy has not yet been done, and optimal patient selection criteria are lacking. Therefore, treatment recommendations are based primarily on expert opinion.
• We recommend flexible endoscopic therapy as initial treatment in patients with a small diverticulum, but this may be limited by the unavailability of a surgeon experienced in this technique.
• For a large diverticulum, we recommend an open cervical procedure involving excision of the diverticulum or a diverticulopexy for patients who are good surgical candidates.

Conventional treatment for a large symptomatic Zenker diverticulum is to surgically either remove it (diverticulectomy) or obliterate it by repositioning and securing it after cutting into the cricopharyngeus muscle (diverticulopexy with cricopharyngeal myotomy). But high rates of complications with these procedures in elderly patients have led to the development of endoscopic procedures that are safer and have similar or higher success rates.

ESOPHAGEAL POUCH IN SPACE BETWEEN MUSCLES

Zenker diverticulum is an outpouching of the esophageal mucosa into the potential space of the Killian triangle, the area between the inferior pharyngeal constrictor and the cricopharyngeus muscle. Increased cricopharyngeal tone coupled with insufficient relaxation results in a pressure gradient that eventually causes esophageal outpouchings. The problem is induced by age-related fibrosis and atrophy, esophageal spasms related to gastroesophageal reflux disease, and idiopathic cricopharyngeal spasms.¹

USUALLY PRESENTS WITH DYSPHAGIA

Zenker diverticulum has an estimated incidence of about 2 per 100,000 per year, but this is likely low because many cases are asymptomatic. It occurs mostly in men and people of Northern European descent in their 60s and 70s and rarely before age 40.²

About 80% to 90% of patients present with dysphagia. Other signs and symptoms include regurgitation of undigested foods and medications, halitosis, hoarseness, chronic cough, and aspiration.

Rare complications of Zenker diverticulum include recurrent pulmonary infection, tracheal fistula, ulceration, hemorrhage, squamous cell carcinoma (incidence 0.4% to 1.5%), vocal cord paralysis, and fistula to the prevertebral ligament with cervical osteomyelitis.¹

The diagnosis is suspected based on symptoms but should be confirmed with a barium esophagram that shows an outpouching of contrast from the main contrast column.

A watch-and-wait approach should be used for patients with mild symptoms (ie, mild or intermittent dysphagia) and minor functional limitations. Patients should be counseled to eat small amounts of food at a time, to chew thoroughly, and to sip liquids between bites.

TREATMENT IS SURGERY OR ENDOSCOPY

Patients with more than mild symptoms who are candidates for intervention should be offered treatment. The goal is to open the septum between the diverticulum and the main esophageal lumen so that food can be propelled from the hypopharynx to the main esophageal lumen without obstruction. Because increased cricopharyngeal tone plays a large role in the development and propagation of a diverticulum, most experts recommend cricopharyngomyotomy in addition to any treatment strategy.

Open surgery

In centers that do not offer flexible endoscopy, and for patients with a large diverticulum, open surgery is the sole option. It is done under general anesthesia with a left cervical approach. The length of the cricopharyngeal myotomy can vary from 2 to 6 cm.^3,4 This is followed by one of three options:

• Diverticulectomy
• Diverticulopexy, in which the diverticulum is repositioned and sutured against the prevertebral fascia
• Diverticular inversion, in which the diverticulum is inverted into the esophageal lumen and then oversewn.⁴

Success rates for open surgery vary from study to study but are usually around 90%.⁴ Complications are reported in 10% to 30% of cases and include mediastinitis, severe recurrent laryngeal nerve injury, and a 1% to 2% chance of death. These rates seem high, but the older age of most of these patients puts them at high risk.⁵

Endoscopic procedures

All endoscopic procedures involve incision of the wall separating the diverticulum from the esophageal lumen to relieve the obstruction.

Rigid endoscopy is used in centers that do not offer flexible endoscopy for patients who are not candidates for surgery. It is done under general anesthesia. With the patient’s neck completely hyperextended, a rigid endoscope is passed into the oral cavity, and diverticulotomy and cricopharyngeal myotomy are performed.

This technique has been extensively evaluated and, although effective, carries up to an 8% risk of complications, including perforation. It is not recommended for patients with a small diverticulum, a high body mass index, or difficulty with neck extension.⁶

Flexible endoscopy is the first-line therapy for most patients in many centers. With the flexible endoscope, the diverticulotomy and the cricopharyngeal myotomy can be performed in an outpatient endoscopy suite, and general anesthesia is not required.³

Initial studies of outcomes using these approaches have been promising,^3,7–9 with substantial reduction in dysphagia, regurgitation, and chronic cough. Unfortunately, this technique is associated with a recurrence rate as high as 25%. However, repeat endoscopic therapy in patients with recurrence results in success rates similar to those for first-time treatment.

Complication rates are low. In a series of 150 patients, four adverse events occurred—three cases of fever and one of pneumonia. In addition, one patient had subcutaneous emphysema that resolved spontaneously.⁷

Improved device on horizon

A potentially major advance in the endoscopic treatment of Zenker diverticulum is the development of a diverticuloscope, a soft tube with a V-shaped end. This tube is inserted through the mouth with one leg of the “V” protecting the anterior esophageal wall and the other leg protecting the posterior diverticular lumen. The endoscope is passed through this tube to perform the diverticulotomy. In one report,¹⁰ diverticuloscope-guided diverticulotomy led to fewer complications, shorter procedural time, and higher success rates compared with the standard technique.¹⁰ This device is not yet available in the United States.

No head-to-head comparison of outcomes with various flexible endoscopic techniques for treating Zenker diverticulum has been published, nor are there data yet on the success of surgery if endoscopic therapy fails.

We recommend flexible endoscopy for the initial treatment of Zenker diverticulum in most patients, but its availability is limited in the United States, as many practitioners do not have adequate experience with the technique.

TAKE-HOME POINTS

• Zenker diverticulum is an outpouching of the esophageal mucosa. It is often asymptomatic. Patients with minimal symptoms and with no functional impairment can be followed conservatively.
• A head-to-head comparison of surgical and endoscopic therapy has not yet been done, and optimal patient selection criteria are lacking. Therefore, treatment recommendations are based primarily on expert opinion.
• We recommend flexible endoscopic therapy as initial treatment in patients with a small diverticulum, but this may be limited by the unavailability of a surgeon experienced in this technique.
• For a large diverticulum, we recommend an open cervical procedure involving excision of the diverticulum or a diverticulopexy for patients who are good surgical candidates.

The article “Total pancreatectomy and islet cell autotransplantation: Definitive treatment for chronic pancreatitis” (Arce KM, Lin YK, Stevens T, Walsh RM, Hatipoglu BA. Cleve Clin J Med 2016; 83:435–442) incorrectly stated that Paul Lacy and David Scharp performed research at the University of Washington at Seattle. They did their work at Washington University in St. Louis, Missouri.

The article “Total pancreatectomy and islet cell autotransplantation: Definitive treatment for chronic pancreatitis” (Arce KM, Lin YK, Stevens T, Walsh RM, Hatipoglu BA. Cleve Clin J Med 2016; 83:435–442) incorrectly stated that Paul Lacy and David Scharp performed research at the University of Washington at Seattle. They did their work at Washington University in St. Louis, Missouri.

The article “Total pancreatectomy and islet cell autotransplantation: Definitive treatment for chronic pancreatitis” (Arce KM, Lin YK, Stevens T, Walsh RM, Hatipoglu BA. Cleve Clin J Med 2016; 83:435–442) incorrectly stated that Paul Lacy and David Scharp performed research at the University of Washington at Seattle. They did their work at Washington University in St. Louis, Missouri.

A 72-year-old woman with type 2 diabetes mellitus, hypertension, and atrial fibrillation on anticoagulation was brought to the emergency department by her husband after 1 day of altered mental status with acute onset. Her husband reported that she had been minimally arousable, and the physical examination revealed that she was stuporous and withdrew extremities only from noxious stimuli.

Results of initial laboratory tests revealed a creatinine level of 2.4 mg/dL (reference range 0.7–1.4), hemoglobin 12.1 g/dL (12–16), platelet count 16 × 10⁹/L (150–400), white blood cell count of 7.7 × 10⁹/L (3.7–11), and international normalized ratio of 2.1. A peripheral blood smear is shown in Figure 1.

Computed tomography showed evidence of chronic small vascular ischemia. Magnetic resonance imaging of the brain showed numerous foci of restricted diffusion within the supratentorial and infratentorial areas, suggesting microembolic phenomena.

The peripheral blood smear was compatible with microangiopathic hemolytic anemia, which can occur in thrombotic thrombocytopenic purpura (TTP), hemolytic uremic syndrome, malignant hypertension, scleroderma, antiphospholipid antibody syndrome, systemic lupus erythematosus, eclampsia, renal allograft rejection, hematopoietic stem cell transplant, and severe sepsis.^1,2

In addition to hemolytic anemia, the patient also had neurologic abnormalities, renal involvement, and thrombocytopenia. The hemolytic anemia and thrombocytopenia were sufficient to raise our suspicion of TTP and to consider initiation of plasma exchange. Only 5% of patients with TTP demonstrate the classic pentad of clinical features,¹ ie, thrombocytopenia, microangiopathic hemolytic anemia, fluctuating neurologic signs, renal impairment, and fever.

In 1991, when plasma exchange was introduced for TTP, the survival rate of patients increased from 10% to 78%.^1,3 Thus, the diagnosis of TTP is an urgent indication for plasma exchange. We normally do plasma exchange daily until the platelet levels improve.

Our patient received methylprednisone 125 mg intravenously every 12 hours and plasma exchange daily. After three cycles of plasma exchange, she regained normal consciousness, and her platelet count had increased to 20.5 × 10⁹/L on the day of discharge from our hospital.

TTP is a life-threatening hematologic disorder. Evidence of microangiopathic hemolytic anemia on a peripheral blood smear is vital to the suspicion of TTP. The diagnosis should be confirmed by ADAMTS13 testing, which should show decreased activity (< 10%) or increased inhibition, or both. Rapid management with plasma exchange and steroids can lead to a satisfactory outcome.

Acknowledgment: We are particularly grateful to Dr. Vivian Arguello (Director of Flow Cytometry, Department of Pathology, Einstein Medical Center, Philadelphia) for her kind support with the blood smear image.

A 72-year-old woman with type 2 diabetes mellitus, hypertension, and atrial fibrillation on anticoagulation was brought to the emergency department by her husband after 1 day of altered mental status with acute onset. Her husband reported that she had been minimally arousable, and the physical examination revealed that she was stuporous and withdrew extremities only from noxious stimuli.

Results of initial laboratory tests revealed a creatinine level of 2.4 mg/dL (reference range 0.7–1.4), hemoglobin 12.1 g/dL (12–16), platelet count 16 × 10⁹/L (150–400), white blood cell count of 7.7 × 10⁹/L (3.7–11), and international normalized ratio of 2.1. A peripheral blood smear is shown in Figure 1.

Computed tomography showed evidence of chronic small vascular ischemia. Magnetic resonance imaging of the brain showed numerous foci of restricted diffusion within the supratentorial and infratentorial areas, suggesting microembolic phenomena.

The peripheral blood smear was compatible with microangiopathic hemolytic anemia, which can occur in thrombotic thrombocytopenic purpura (TTP), hemolytic uremic syndrome, malignant hypertension, scleroderma, antiphospholipid antibody syndrome, systemic lupus erythematosus, eclampsia, renal allograft rejection, hematopoietic stem cell transplant, and severe sepsis.^1,2

In addition to hemolytic anemia, the patient also had neurologic abnormalities, renal involvement, and thrombocytopenia. The hemolytic anemia and thrombocytopenia were sufficient to raise our suspicion of TTP and to consider initiation of plasma exchange. Only 5% of patients with TTP demonstrate the classic pentad of clinical features,¹ ie, thrombocytopenia, microangiopathic hemolytic anemia, fluctuating neurologic signs, renal impairment, and fever.

In 1991, when plasma exchange was introduced for TTP, the survival rate of patients increased from 10% to 78%.^1,3 Thus, the diagnosis of TTP is an urgent indication for plasma exchange. We normally do plasma exchange daily until the platelet levels improve.

Our patient received methylprednisone 125 mg intravenously every 12 hours and plasma exchange daily. After three cycles of plasma exchange, she regained normal consciousness, and her platelet count had increased to 20.5 × 10⁹/L on the day of discharge from our hospital.

TTP is a life-threatening hematologic disorder. Evidence of microangiopathic hemolytic anemia on a peripheral blood smear is vital to the suspicion of TTP. The diagnosis should be confirmed by ADAMTS13 testing, which should show decreased activity (< 10%) or increased inhibition, or both. Rapid management with plasma exchange and steroids can lead to a satisfactory outcome.

Acknowledgment: We are particularly grateful to Dr. Vivian Arguello (Director of Flow Cytometry, Department of Pathology, Einstein Medical Center, Philadelphia) for her kind support with the blood smear image.

A 72-year-old woman with type 2 diabetes mellitus, hypertension, and atrial fibrillation on anticoagulation was brought to the emergency department by her husband after 1 day of altered mental status with acute onset. Her husband reported that she had been minimally arousable, and the physical examination revealed that she was stuporous and withdrew extremities only from noxious stimuli.

Results of initial laboratory tests revealed a creatinine level of 2.4 mg/dL (reference range 0.7–1.4), hemoglobin 12.1 g/dL (12–16), platelet count 16 × 10⁹/L (150–400), white blood cell count of 7.7 × 10⁹/L (3.7–11), and international normalized ratio of 2.1. A peripheral blood smear is shown in Figure 1.

Computed tomography showed evidence of chronic small vascular ischemia. Magnetic resonance imaging of the brain showed numerous foci of restricted diffusion within the supratentorial and infratentorial areas, suggesting microembolic phenomena.

The peripheral blood smear was compatible with microangiopathic hemolytic anemia, which can occur in thrombotic thrombocytopenic purpura (TTP), hemolytic uremic syndrome, malignant hypertension, scleroderma, antiphospholipid antibody syndrome, systemic lupus erythematosus, eclampsia, renal allograft rejection, hematopoietic stem cell transplant, and severe sepsis.^1,2

In addition to hemolytic anemia, the patient also had neurologic abnormalities, renal involvement, and thrombocytopenia. The hemolytic anemia and thrombocytopenia were sufficient to raise our suspicion of TTP and to consider initiation of plasma exchange. Only 5% of patients with TTP demonstrate the classic pentad of clinical features,¹ ie, thrombocytopenia, microangiopathic hemolytic anemia, fluctuating neurologic signs, renal impairment, and fever.

In 1991, when plasma exchange was introduced for TTP, the survival rate of patients increased from 10% to 78%.^1,3 Thus, the diagnosis of TTP is an urgent indication for plasma exchange. We normally do plasma exchange daily until the platelet levels improve.

Our patient received methylprednisone 125 mg intravenously every 12 hours and plasma exchange daily. After three cycles of plasma exchange, she regained normal consciousness, and her platelet count had increased to 20.5 × 10⁹/L on the day of discharge from our hospital.

TTP is a life-threatening hematologic disorder. Evidence of microangiopathic hemolytic anemia on a peripheral blood smear is vital to the suspicion of TTP. The diagnosis should be confirmed by ADAMTS13 testing, which should show decreased activity (< 10%) or increased inhibition, or both. Rapid management with plasma exchange and steroids can lead to a satisfactory outcome.

Acknowledgment: We are particularly grateful to Dr. Vivian Arguello (Director of Flow Cytometry, Department of Pathology, Einstein Medical Center, Philadelphia) for her kind support with the blood smear image.

A 39-year-old man with human immunodeficiency virus (HIV) infection and a CD4 cell count of 528 × 10⁶/L without treatment was referred for evaluation of periarticular, indurated, erythematous papules and nodules on the knees and elbows and purpuric lesions on the ankles (Figure 1). He has had recurrent fever, arthralgia, and mild constitutional symptoms during the past month. He also reported a diagnosis of polyclonal immunoglobulin A gammopathy.

Punch biopsy of the purpuric lesions was performed, and histologic study showed leukocytoclastic vasculitis with eosinophilic necrosis of the epithelium.

Treatment with a systemic corticosteroid was started. The purpuric lesions disappeared after 3 weeks of therapy, but the nodules over the extensor surface of both knees showed no improvement (Figure 2). Subsequent biopsy of late-stage lesions (3 months after the start of therapy) demonstrated perivascular fibrosis with small, persistent foci of vasculitis, and confirmed the diagnosis of HIV-associated nodular erythema elevatum diutinum (EED). Antiretroviral therapy was started in addition to intralesional corticosteroids and topical dapsone 5% gel, with resolution of the lesions 1 month later.

ERYTHEMA ELEVATUM DIUTINUM

EED is an uncommon chronic dermatosis, classified as a fibrosing form of leukocytoclastic vasculitis and characterized clinically by violaceous papules, plaques, and nodules, usually distributed acrally and symmetrically over extensor surfaces. The histopathologic picture depends on the stage of the lesion. Features of leukocytoclastic vasculitis are found in early-stage lesions, while a fibrotic replacement of the dermis with small persistent foci of vasculitis is typical of late-stage lesions.

EED has clinical and histopathologic similarities to Sweet syndrome, but EED is distinguished from neutrophilic dermatosis by vasculitis.

In HIV-infected patients, it is important to include pruritic papular eruption in the differential diagnosis. It is characterized by chronic bilaterally symmetric pruritic papules on the trunk and extremities and is the most common cutaneous noninfectious manifestation of HIV.

The clinical presentation of EED may also be easily confused with Kaposi sarcoma and bacillary angiomatosis.²

AN EMERGING HIV-RELATED DERMATOSIS

EED is emerging as a specific HIV-associated dermatosis, with 20 cases reported in the medical literature as of this writing.³

The cause of EED is not known, but it is often associated with streptococcal infection, monoclonal IgA gammopathy, hematologic malignancy, cryoglobulinemia, rheumatoid arthritis, and autoimmune disease.³ The stimulus could be immune-complex deposition in blood vessels triggered by HIV infection, or by another infection acting as an antigenic stimulus.⁴ The nodular variant of EED is even rarer, but it evolves most often in HIV-positive individuals.^5,6

Oral dapsone is the treatment of choice but is less effective in late-stage fibrotic lesions.⁷ Treatment courses tend to be long and recurrence is common.⁸ Intralesional, topical, and oral corticosteroids, topical dapsone 5% gel,⁹ tetracycline and nicotinamide, sulfonamides, colchicine, chloroquine, and surgical excision are other options.

Our patient’s presentation reminds us to consider EED in HIV-infected patients and illustrates the importance of histologic diagnosis to differentiate EED from assumed Kaposi sarcoma in patients with HIV. EED can also be the first clinical sign of HIV infection. It is important to rule out underlying disorders such as HIV infection, because directed therapy is often the best management.

A 39-year-old man with human immunodeficiency virus (HIV) infection and a CD4 cell count of 528 × 10⁶/L without treatment was referred for evaluation of periarticular, indurated, erythematous papules and nodules on the knees and elbows and purpuric lesions on the ankles (Figure 1). He has had recurrent fever, arthralgia, and mild constitutional symptoms during the past month. He also reported a diagnosis of polyclonal immunoglobulin A gammopathy.

Punch biopsy of the purpuric lesions was performed, and histologic study showed leukocytoclastic vasculitis with eosinophilic necrosis of the epithelium.

Treatment with a systemic corticosteroid was started. The purpuric lesions disappeared after 3 weeks of therapy, but the nodules over the extensor surface of both knees showed no improvement (Figure 2). Subsequent biopsy of late-stage lesions (3 months after the start of therapy) demonstrated perivascular fibrosis with small, persistent foci of vasculitis, and confirmed the diagnosis of HIV-associated nodular erythema elevatum diutinum (EED). Antiretroviral therapy was started in addition to intralesional corticosteroids and topical dapsone 5% gel, with resolution of the lesions 1 month later.

ERYTHEMA ELEVATUM DIUTINUM

EED is an uncommon chronic dermatosis, classified as a fibrosing form of leukocytoclastic vasculitis and characterized clinically by violaceous papules, plaques, and nodules, usually distributed acrally and symmetrically over extensor surfaces. The histopathologic picture depends on the stage of the lesion. Features of leukocytoclastic vasculitis are found in early-stage lesions, while a fibrotic replacement of the dermis with small persistent foci of vasculitis is typical of late-stage lesions.

EED has clinical and histopathologic similarities to Sweet syndrome, but EED is distinguished from neutrophilic dermatosis by vasculitis.

In HIV-infected patients, it is important to include pruritic papular eruption in the differential diagnosis. It is characterized by chronic bilaterally symmetric pruritic papules on the trunk and extremities and is the most common cutaneous noninfectious manifestation of HIV.

The clinical presentation of EED may also be easily confused with Kaposi sarcoma and bacillary angiomatosis.²

AN EMERGING HIV-RELATED DERMATOSIS

EED is emerging as a specific HIV-associated dermatosis, with 20 cases reported in the medical literature as of this writing.³

The cause of EED is not known, but it is often associated with streptococcal infection, monoclonal IgA gammopathy, hematologic malignancy, cryoglobulinemia, rheumatoid arthritis, and autoimmune disease.³ The stimulus could be immune-complex deposition in blood vessels triggered by HIV infection, or by another infection acting as an antigenic stimulus.⁴ The nodular variant of EED is even rarer, but it evolves most often in HIV-positive individuals.^5,6

Oral dapsone is the treatment of choice but is less effective in late-stage fibrotic lesions.⁷ Treatment courses tend to be long and recurrence is common.⁸ Intralesional, topical, and oral corticosteroids, topical dapsone 5% gel,⁹ tetracycline and nicotinamide, sulfonamides, colchicine, chloroquine, and surgical excision are other options.

Our patient’s presentation reminds us to consider EED in HIV-infected patients and illustrates the importance of histologic diagnosis to differentiate EED from assumed Kaposi sarcoma in patients with HIV. EED can also be the first clinical sign of HIV infection. It is important to rule out underlying disorders such as HIV infection, because directed therapy is often the best management.

A 39-year-old man with human immunodeficiency virus (HIV) infection and a CD4 cell count of 528 × 10⁶/L without treatment was referred for evaluation of periarticular, indurated, erythematous papules and nodules on the knees and elbows and purpuric lesions on the ankles (Figure 1). He has had recurrent fever, arthralgia, and mild constitutional symptoms during the past month. He also reported a diagnosis of polyclonal immunoglobulin A gammopathy.

Punch biopsy of the purpuric lesions was performed, and histologic study showed leukocytoclastic vasculitis with eosinophilic necrosis of the epithelium.

Treatment with a systemic corticosteroid was started. The purpuric lesions disappeared after 3 weeks of therapy, but the nodules over the extensor surface of both knees showed no improvement (Figure 2). Subsequent biopsy of late-stage lesions (3 months after the start of therapy) demonstrated perivascular fibrosis with small, persistent foci of vasculitis, and confirmed the diagnosis of HIV-associated nodular erythema elevatum diutinum (EED). Antiretroviral therapy was started in addition to intralesional corticosteroids and topical dapsone 5% gel, with resolution of the lesions 1 month later.

ERYTHEMA ELEVATUM DIUTINUM

EED is an uncommon chronic dermatosis, classified as a fibrosing form of leukocytoclastic vasculitis and characterized clinically by violaceous papules, plaques, and nodules, usually distributed acrally and symmetrically over extensor surfaces. The histopathologic picture depends on the stage of the lesion. Features of leukocytoclastic vasculitis are found in early-stage lesions, while a fibrotic replacement of the dermis with small persistent foci of vasculitis is typical of late-stage lesions.

EED has clinical and histopathologic similarities to Sweet syndrome, but EED is distinguished from neutrophilic dermatosis by vasculitis.

In HIV-infected patients, it is important to include pruritic papular eruption in the differential diagnosis. It is characterized by chronic bilaterally symmetric pruritic papules on the trunk and extremities and is the most common cutaneous noninfectious manifestation of HIV.

The clinical presentation of EED may also be easily confused with Kaposi sarcoma and bacillary angiomatosis.²

AN EMERGING HIV-RELATED DERMATOSIS

EED is emerging as a specific HIV-associated dermatosis, with 20 cases reported in the medical literature as of this writing.³

The cause of EED is not known, but it is often associated with streptococcal infection, monoclonal IgA gammopathy, hematologic malignancy, cryoglobulinemia, rheumatoid arthritis, and autoimmune disease.³ The stimulus could be immune-complex deposition in blood vessels triggered by HIV infection, or by another infection acting as an antigenic stimulus.⁴ The nodular variant of EED is even rarer, but it evolves most often in HIV-positive individuals.^5,6

Oral dapsone is the treatment of choice but is less effective in late-stage fibrotic lesions.⁷ Treatment courses tend to be long and recurrence is common.⁸ Intralesional, topical, and oral corticosteroids, topical dapsone 5% gel,⁹ tetracycline and nicotinamide, sulfonamides, colchicine, chloroquine, and surgical excision are other options.

Our patient’s presentation reminds us to consider EED in HIV-infected patients and illustrates the importance of histologic diagnosis to differentiate EED from assumed Kaposi sarcoma in patients with HIV. EED can also be the first clinical sign of HIV infection. It is important to rule out underlying disorders such as HIV infection, because directed therapy is often the best management.

A 39-year-old Filipino man presented with nausea, vomiting, and abdominal pain of 2 weeks’ duration. He did not report trauma, and he had no history of medical illness or surgery.

On arrival, his blood pressure was 123/83 mm Hg, pulse 122 beats per minute, respiratory rate 18 breaths per minute, and temperature 100.7°F (38.1°C). On physical examination, he exhibited marked tenderness of the right upper quadrant on palpation. The abdomen was otherwise soft with no guarding or rebound tenderness.

Results of initial laboratory testing were as follows:

• Leukocyte count 17.0 × 10⁹/L (reference range 4.5–11.0)
• Serum glucose 558 mg/dL without ketoacidosis
• Aspartate aminotransferase 109 U/L (2–40)
• Alanine aminotranferase 28 U/L (2–50)
• Total serum bilirubin 4.0 mg/dL (0.0–1.5).

Plain chest radiography showed dramatic elevation of the right hemidiaphragm with a large subphrenic air-fluid level (Figure 1). Abdominal computed tomography (CT) demonstrated a multiloculated hepatic abscess 18 × 13.5 cm subjacent to the diaphragm (Figure 2). Cultures of blood and the abscess yielded Klebsiella pneumoniae. The patient recovered after percutaneous drainage and a course of ceftriaxone.

PRIMARY KLEBSIELLA LIVER ABSCESS

K pneumoniae, a gram-negative aerobic encapsulated bacillus of the normal human intestinal flora, is closely related to Escherichia coli, historically the most frequent bacterial cause of pyogenic liver abscess.¹ Over the last 30 years, K pneumoniae has eclipsed E coli as the most common causative agent, with the epicenter of this trend being located in Taiwan and South Korea, perhaps because rates of fecal Klebsiella carriage in that region are particularly high.^1,2

Concurrently, there has been increasing recognition—initially across Asia, but lately in Europe and the Western Hemisphere—of the so-called invasive Klebsiella liver abscess (KLA) syndrome, virtually unique to the hypervirulent K1 and K2 capsular serotypes of K pneumoniae prevalent in Asia.^3–6 This community-acquired syndrome is characterized by hematogenous deposition of the organism at distant sites, such as the lung, soft tissues, central nervous system, and eyes. Impairment of phagocytic function, as occurs in diabetes mellitus, and the resistance to phagocytosis conferred by the K1 and K2 serotypes have been identified as predisposing factors for dissemination.^7,8 The mucoid phenotype of K pneumoniae, very common in Asian isolates of the K1 and K2 serotypes, is also associated with hypervirulence and extrahepatic spread, presumably through evasion of phagocytosis and complement-mediated opsonization.^2,9

Our patient’s risk factors for KLA were his Asian origin and uncontrolled diabetes. No evidence of remote infection was detected during his hospitalization.

HEMIDIAPHRAGM ELEVATION

Acquired hemidiaphragm elevation is most commonly unilateral and typically represents an incidental radiologic finding attributable to paralysis of the corresponding diaphragm after phrenic nerve injury caused by trauma, surgery, or infection. Unilateral diaphragmatic paralysis is classically confirmed by performing a fluoroscopic sniff test, which is positive if the affected hemidiaphragm is observed in real time to paradoxically move upward during forced inhalation.¹⁰ This condition is usually asymptomatic at rest but could cause exertional dyspnea and contribute to ventilatory failure when pulmonary disease coexists.¹¹

Occasionally, as in our patient, hemidiaphragm elevation is part of the presentation of active abdominal pathology that displaces the corresponding hemidiaphragm cephalad by mass effect. Examples of such space-occupying abdominal lesions include infections, malignancy, hepatosplenomegaly, and pneumoperitoneum from a ruptured viscus. Pneumoperitoneum is suggested by the presence of an air crescent immediately subjacent to the affected hemidiaphragm on an upright radiograph accompanied by peritoneal signs.

Although there was subphrenic air on this patient’s initial chest radiograph, it was actually part of an air-fluid level without associated peritoneal signs. An air-fluid level is characterized by a sharp horizontal demarcation between the lighter gas component floating at the top and the heavier fluid component settling on the bottom (Figure 1). The subsequent CT excluded free intra-abdominal air while identifying a large hepatic abscess as the cause of hemidiaphragm elevation. In trauma victims, CT is also helpful in ruling out diaphragmatic rupture, which can have a similar radiographic appearance.¹²

Our patient’s presentation was a reminder that an elevated hemidiaphragm may reflect abdominal pathology and that subphrenic air in this context need not be either “free” or a surgical emergency. Drainage of the abscess restored the normal position of our patient’s right hemidiaphragm (Figure 3).

A 39-year-old Filipino man presented with nausea, vomiting, and abdominal pain of 2 weeks’ duration. He did not report trauma, and he had no history of medical illness or surgery.

On arrival, his blood pressure was 123/83 mm Hg, pulse 122 beats per minute, respiratory rate 18 breaths per minute, and temperature 100.7°F (38.1°C). On physical examination, he exhibited marked tenderness of the right upper quadrant on palpation. The abdomen was otherwise soft with no guarding or rebound tenderness.

Results of initial laboratory testing were as follows:

• Leukocyte count 17.0 × 10⁹/L (reference range 4.5–11.0)
• Serum glucose 558 mg/dL without ketoacidosis
• Aspartate aminotransferase 109 U/L (2–40)
• Alanine aminotranferase 28 U/L (2–50)
• Total serum bilirubin 4.0 mg/dL (0.0–1.5).

Plain chest radiography showed dramatic elevation of the right hemidiaphragm with a large subphrenic air-fluid level (Figure 1). Abdominal computed tomography (CT) demonstrated a multiloculated hepatic abscess 18 × 13.5 cm subjacent to the diaphragm (Figure 2). Cultures of blood and the abscess yielded Klebsiella pneumoniae. The patient recovered after percutaneous drainage and a course of ceftriaxone.

PRIMARY KLEBSIELLA LIVER ABSCESS

K pneumoniae, a gram-negative aerobic encapsulated bacillus of the normal human intestinal flora, is closely related to Escherichia coli, historically the most frequent bacterial cause of pyogenic liver abscess.¹ Over the last 30 years, K pneumoniae has eclipsed E coli as the most common causative agent, with the epicenter of this trend being located in Taiwan and South Korea, perhaps because rates of fecal Klebsiella carriage in that region are particularly high.^1,2

Concurrently, there has been increasing recognition—initially across Asia, but lately in Europe and the Western Hemisphere—of the so-called invasive Klebsiella liver abscess (KLA) syndrome, virtually unique to the hypervirulent K1 and K2 capsular serotypes of K pneumoniae prevalent in Asia.^3–6 This community-acquired syndrome is characterized by hematogenous deposition of the organism at distant sites, such as the lung, soft tissues, central nervous system, and eyes. Impairment of phagocytic function, as occurs in diabetes mellitus, and the resistance to phagocytosis conferred by the K1 and K2 serotypes have been identified as predisposing factors for dissemination.^7,8 The mucoid phenotype of K pneumoniae, very common in Asian isolates of the K1 and K2 serotypes, is also associated with hypervirulence and extrahepatic spread, presumably through evasion of phagocytosis and complement-mediated opsonization.^2,9

Our patient’s risk factors for KLA were his Asian origin and uncontrolled diabetes. No evidence of remote infection was detected during his hospitalization.

HEMIDIAPHRAGM ELEVATION

Acquired hemidiaphragm elevation is most commonly unilateral and typically represents an incidental radiologic finding attributable to paralysis of the corresponding diaphragm after phrenic nerve injury caused by trauma, surgery, or infection. Unilateral diaphragmatic paralysis is classically confirmed by performing a fluoroscopic sniff test, which is positive if the affected hemidiaphragm is observed in real time to paradoxically move upward during forced inhalation.¹⁰ This condition is usually asymptomatic at rest but could cause exertional dyspnea and contribute to ventilatory failure when pulmonary disease coexists.¹¹

Occasionally, as in our patient, hemidiaphragm elevation is part of the presentation of active abdominal pathology that displaces the corresponding hemidiaphragm cephalad by mass effect. Examples of such space-occupying abdominal lesions include infections, malignancy, hepatosplenomegaly, and pneumoperitoneum from a ruptured viscus. Pneumoperitoneum is suggested by the presence of an air crescent immediately subjacent to the affected hemidiaphragm on an upright radiograph accompanied by peritoneal signs.

Although there was subphrenic air on this patient’s initial chest radiograph, it was actually part of an air-fluid level without associated peritoneal signs. An air-fluid level is characterized by a sharp horizontal demarcation between the lighter gas component floating at the top and the heavier fluid component settling on the bottom (Figure 1). The subsequent CT excluded free intra-abdominal air while identifying a large hepatic abscess as the cause of hemidiaphragm elevation. In trauma victims, CT is also helpful in ruling out diaphragmatic rupture, which can have a similar radiographic appearance.¹²

Our patient’s presentation was a reminder that an elevated hemidiaphragm may reflect abdominal pathology and that subphrenic air in this context need not be either “free” or a surgical emergency. Drainage of the abscess restored the normal position of our patient’s right hemidiaphragm (Figure 3).

A 39-year-old Filipino man presented with nausea, vomiting, and abdominal pain of 2 weeks’ duration. He did not report trauma, and he had no history of medical illness or surgery.

On arrival, his blood pressure was 123/83 mm Hg, pulse 122 beats per minute, respiratory rate 18 breaths per minute, and temperature 100.7°F (38.1°C). On physical examination, he exhibited marked tenderness of the right upper quadrant on palpation. The abdomen was otherwise soft with no guarding or rebound tenderness.

Results of initial laboratory testing were as follows:

• Leukocyte count 17.0 × 10⁹/L (reference range 4.5–11.0)
• Serum glucose 558 mg/dL without ketoacidosis
• Aspartate aminotransferase 109 U/L (2–40)
• Alanine aminotranferase 28 U/L (2–50)
• Total serum bilirubin 4.0 mg/dL (0.0–1.5).

Plain chest radiography showed dramatic elevation of the right hemidiaphragm with a large subphrenic air-fluid level (Figure 1). Abdominal computed tomography (CT) demonstrated a multiloculated hepatic abscess 18 × 13.5 cm subjacent to the diaphragm (Figure 2). Cultures of blood and the abscess yielded Klebsiella pneumoniae. The patient recovered after percutaneous drainage and a course of ceftriaxone.

PRIMARY KLEBSIELLA LIVER ABSCESS

K pneumoniae, a gram-negative aerobic encapsulated bacillus of the normal human intestinal flora, is closely related to Escherichia coli, historically the most frequent bacterial cause of pyogenic liver abscess.¹ Over the last 30 years, K pneumoniae has eclipsed E coli as the most common causative agent, with the epicenter of this trend being located in Taiwan and South Korea, perhaps because rates of fecal Klebsiella carriage in that region are particularly high.^1,2

Concurrently, there has been increasing recognition—initially across Asia, but lately in Europe and the Western Hemisphere—of the so-called invasive Klebsiella liver abscess (KLA) syndrome, virtually unique to the hypervirulent K1 and K2 capsular serotypes of K pneumoniae prevalent in Asia.^3–6 This community-acquired syndrome is characterized by hematogenous deposition of the organism at distant sites, such as the lung, soft tissues, central nervous system, and eyes. Impairment of phagocytic function, as occurs in diabetes mellitus, and the resistance to phagocytosis conferred by the K1 and K2 serotypes have been identified as predisposing factors for dissemination.^7,8 The mucoid phenotype of K pneumoniae, very common in Asian isolates of the K1 and K2 serotypes, is also associated with hypervirulence and extrahepatic spread, presumably through evasion of phagocytosis and complement-mediated opsonization.^2,9

Our patient’s risk factors for KLA were his Asian origin and uncontrolled diabetes. No evidence of remote infection was detected during his hospitalization.

HEMIDIAPHRAGM ELEVATION

Acquired hemidiaphragm elevation is most commonly unilateral and typically represents an incidental radiologic finding attributable to paralysis of the corresponding diaphragm after phrenic nerve injury caused by trauma, surgery, or infection. Unilateral diaphragmatic paralysis is classically confirmed by performing a fluoroscopic sniff test, which is positive if the affected hemidiaphragm is observed in real time to paradoxically move upward during forced inhalation.¹⁰ This condition is usually asymptomatic at rest but could cause exertional dyspnea and contribute to ventilatory failure when pulmonary disease coexists.¹¹

Occasionally, as in our patient, hemidiaphragm elevation is part of the presentation of active abdominal pathology that displaces the corresponding hemidiaphragm cephalad by mass effect. Examples of such space-occupying abdominal lesions include infections, malignancy, hepatosplenomegaly, and pneumoperitoneum from a ruptured viscus. Pneumoperitoneum is suggested by the presence of an air crescent immediately subjacent to the affected hemidiaphragm on an upright radiograph accompanied by peritoneal signs.

Although there was subphrenic air on this patient’s initial chest radiograph, it was actually part of an air-fluid level without associated peritoneal signs. An air-fluid level is characterized by a sharp horizontal demarcation between the lighter gas component floating at the top and the heavier fluid component settling on the bottom (Figure 1). The subsequent CT excluded free intra-abdominal air while identifying a large hepatic abscess as the cause of hemidiaphragm elevation. In trauma victims, CT is also helpful in ruling out diaphragmatic rupture, which can have a similar radiographic appearance.¹²

Our patient’s presentation was a reminder that an elevated hemidiaphragm may reflect abdominal pathology and that subphrenic air in this context need not be either “free” or a surgical emergency. Drainage of the abscess restored the normal position of our patient’s right hemidiaphragm (Figure 3).

Our goal at the Cleveland Clinic Journal of Medicine is to provide timely, readily digestible, and useful clinical information to our readers. To do so, we need authors who buy into our educational mission, but we also need conscientious peer reviewers and an editorial staff capable of turning “doctorese” into readily understood English.

Our physician deputy editors Pelin Batur and James Pile solicit articles, guide authors, draft and revise CME questions, and assist greatly in the peer review process. Our nonphysician editors edit manuscripts to achieve a consistent editorial style in all of our published papers, but they serve many other key functions. They  manage the business of publishing a monthly journal at a time of drastically shrinking advertising revenue, they ensure that Journal content complies with rules for CME material published in print and online, and they keep up with continuous changes in online publishing. The Journal receives significant funding from Cleveland Clinic’s Education Institute, funding we need to pursue our role as an independent, peer-reviewed conveyer of clinical information.

I write this to emphasize that real people manage all of these tasks, and to gratefully acknowledge two people who are leaving the Journal: Mr. Joseph Dennehy and Dr. James Pile.

Long-time sales and marketing director Joe Dennehy played a key role in the Journal’s rise from relative obscurity about 20 years ago. This period was marked by many hospital-based journals closing up shop. Published since 1931, the Journal was relaunched in 1995 as a resource for postgraduate medical education with a national and international reach. Joe was brought in to guide and implement the marketing of the Journal as an independent, high-quality, clinical educational journal to increase its attractiveness as an advertising medium for the pharmaceutical industry so we could at least partly defray the significant expenses of publishing.

Joe is well known to medical publishers and media buyers, and over the past 20 years he became the face of the Journal in sales and marketing circles. In 2010, the Association of Medical Media recognized Joe’s achievements at its 18th Annual Nexus Representatives of the Year Awards, which acknowledge outstanding sales and marketing directors for their superior service, professionalism, and communication of ideas. Joe exemplifies these qualities and has been an inspiration to those of us who know and work with him. He has fully understood the sincerity of our mission and has never asked us to stray from it. Although Joe and his wife Holly live in New York, they are part of the Cleveland-based Journal family, and we will miss them. We wish them happiness and good health.

Dr. James Pile is an internal medicine hospitalist, an infectious disease specialist, and a superior medical educator. His work with us for the past several years has enhanced the Journal’s educational value for practicing hospitalists. His working familiarity with clinical leaders in the Society of Hospital Medicine has provided us with willing and skilled peer reviewers. Jim is now transitioning to a role as director of the internal medicine residency program at MetroHealth Medical Center, also here in Cleveland. He will continue to be a clinical resource for us as author and reviewer, but his ever-calm demeanor and clinical common sense will be hard to replace.

The Journal continues to evolve with the publishing times. We are assuming a greater presence in the digital world and adjusting to an environment of ever-diminishing advertising revenue. But with the work of our editorial team (I invite you to periodically glance at our masthead to note our team of writers, managers, and production staff), we intend to stay true to our educational mission. Our personal thanks to Joe and Jim for their contributions of the past, and to our current team in Scientific Publications for their ongoing and very personal commitment to providing the highest quality medical education that we can.

Our goal at the Cleveland Clinic Journal of Medicine is to provide timely, readily digestible, and useful clinical information to our readers. To do so, we need authors who buy into our educational mission, but we also need conscientious peer reviewers and an editorial staff capable of turning “doctorese” into readily understood English.

Our physician deputy editors Pelin Batur and James Pile solicit articles, guide authors, draft and revise CME questions, and assist greatly in the peer review process. Our nonphysician editors edit manuscripts to achieve a consistent editorial style in all of our published papers, but they serve many other key functions. They  manage the business of publishing a monthly journal at a time of drastically shrinking advertising revenue, they ensure that Journal content complies with rules for CME material published in print and online, and they keep up with continuous changes in online publishing. The Journal receives significant funding from Cleveland Clinic’s Education Institute, funding we need to pursue our role as an independent, peer-reviewed conveyer of clinical information.

I write this to emphasize that real people manage all of these tasks, and to gratefully acknowledge two people who are leaving the Journal: Mr. Joseph Dennehy and Dr. James Pile.

Long-time sales and marketing director Joe Dennehy played a key role in the Journal’s rise from relative obscurity about 20 years ago. This period was marked by many hospital-based journals closing up shop. Published since 1931, the Journal was relaunched in 1995 as a resource for postgraduate medical education with a national and international reach. Joe was brought in to guide and implement the marketing of the Journal as an independent, high-quality, clinical educational journal to increase its attractiveness as an advertising medium for the pharmaceutical industry so we could at least partly defray the significant expenses of publishing.

Joe is well known to medical publishers and media buyers, and over the past 20 years he became the face of the Journal in sales and marketing circles. In 2010, the Association of Medical Media recognized Joe’s achievements at its 18th Annual Nexus Representatives of the Year Awards, which acknowledge outstanding sales and marketing directors for their superior service, professionalism, and communication of ideas. Joe exemplifies these qualities and has been an inspiration to those of us who know and work with him. He has fully understood the sincerity of our mission and has never asked us to stray from it. Although Joe and his wife Holly live in New York, they are part of the Cleveland-based Journal family, and we will miss them. We wish them happiness and good health.

Dr. James Pile is an internal medicine hospitalist, an infectious disease specialist, and a superior medical educator. His work with us for the past several years has enhanced the Journal’s educational value for practicing hospitalists. His working familiarity with clinical leaders in the Society of Hospital Medicine has provided us with willing and skilled peer reviewers. Jim is now transitioning to a role as director of the internal medicine residency program at MetroHealth Medical Center, also here in Cleveland. He will continue to be a clinical resource for us as author and reviewer, but his ever-calm demeanor and clinical common sense will be hard to replace.

The Journal continues to evolve with the publishing times. We are assuming a greater presence in the digital world and adjusting to an environment of ever-diminishing advertising revenue. But with the work of our editorial team (I invite you to periodically glance at our masthead to note our team of writers, managers, and production staff), we intend to stay true to our educational mission. Our personal thanks to Joe and Jim for their contributions of the past, and to our current team in Scientific Publications for their ongoing and very personal commitment to providing the highest quality medical education that we can.

Our goal at the Cleveland Clinic Journal of Medicine is to provide timely, readily digestible, and useful clinical information to our readers. To do so, we need authors who buy into our educational mission, but we also need conscientious peer reviewers and an editorial staff capable of turning “doctorese” into readily understood English.

Our physician deputy editors Pelin Batur and James Pile solicit articles, guide authors, draft and revise CME questions, and assist greatly in the peer review process. Our nonphysician editors edit manuscripts to achieve a consistent editorial style in all of our published papers, but they serve many other key functions. They  manage the business of publishing a monthly journal at a time of drastically shrinking advertising revenue, they ensure that Journal content complies with rules for CME material published in print and online, and they keep up with continuous changes in online publishing. The Journal receives significant funding from Cleveland Clinic’s Education Institute, funding we need to pursue our role as an independent, peer-reviewed conveyer of clinical information.

I write this to emphasize that real people manage all of these tasks, and to gratefully acknowledge two people who are leaving the Journal: Mr. Joseph Dennehy and Dr. James Pile.

Long-time sales and marketing director Joe Dennehy played a key role in the Journal’s rise from relative obscurity about 20 years ago. This period was marked by many hospital-based journals closing up shop. Published since 1931, the Journal was relaunched in 1995 as a resource for postgraduate medical education with a national and international reach. Joe was brought in to guide and implement the marketing of the Journal as an independent, high-quality, clinical educational journal to increase its attractiveness as an advertising medium for the pharmaceutical industry so we could at least partly defray the significant expenses of publishing.

Joe is well known to medical publishers and media buyers, and over the past 20 years he became the face of the Journal in sales and marketing circles. In 2010, the Association of Medical Media recognized Joe’s achievements at its 18th Annual Nexus Representatives of the Year Awards, which acknowledge outstanding sales and marketing directors for their superior service, professionalism, and communication of ideas. Joe exemplifies these qualities and has been an inspiration to those of us who know and work with him. He has fully understood the sincerity of our mission and has never asked us to stray from it. Although Joe and his wife Holly live in New York, they are part of the Cleveland-based Journal family, and we will miss them. We wish them happiness and good health.

Dr. James Pile is an internal medicine hospitalist, an infectious disease specialist, and a superior medical educator. His work with us for the past several years has enhanced the Journal’s educational value for practicing hospitalists. His working familiarity with clinical leaders in the Society of Hospital Medicine has provided us with willing and skilled peer reviewers. Jim is now transitioning to a role as director of the internal medicine residency program at MetroHealth Medical Center, also here in Cleveland. He will continue to be a clinical resource for us as author and reviewer, but his ever-calm demeanor and clinical common sense will be hard to replace.

The Journal continues to evolve with the publishing times. We are assuming a greater presence in the digital world and adjusting to an environment of ever-diminishing advertising revenue. But with the work of our editorial team (I invite you to periodically glance at our masthead to note our team of writers, managers, and production staff), we intend to stay true to our educational mission. Our personal thanks to Joe and Jim for their contributions of the past, and to our current team in Scientific Publications for their ongoing and very personal commitment to providing the highest quality medical education that we can.

Percutaneous coronary intervention for acute coronary syndromes has evolved, and so, hand in hand, has antiplatelet therapy. With the advent of clopidogrel and newer agents, several studies demonstrated the benefits of dual antiplatelet therapy in preventing major vascular ischemic complications. The findings culminated in a guideline recommendation for at least 12 months of dual antiplatelet therapy after placement of a drug-eluting stent, when feasible—a class I recommendation (treatment should be given), level of evidence B (limited populations evaluated).^1,2 But extending dual antiplatelet therapy beyond 12 months had no strong favorable evidence until the recent Dual Antiplatelet Therapy (DAPT) study³ shed light on this topic.

Here, we review the evidence thus far on the optimal duration of dual antiplatelet therapy in the secondary prevention of coronary artery disease.

PLATELETS IN ACUTE CORONARY SYNDROMES AND STENT THROMBOSIS

Acute coronary syndromes begin with fissuring or ulceration of a vulnerable atherosclerotic plaque, followed by thrombosis and occlusion, mediated by platelet adhesion, activation, and aggregation (Figure 1). Transient occlusion results in unstable angina or non-ST-elevation myocardial infarction, while total occlusion usually results in ST-elevation myocardial infarction.

Platelet aggregation is prominent among the mechanisms leading to stent thrombosis and vaso-occlusive ischemic complications after percutaneous coronary intervention. Thus, antiplatelet agents play a vital role in both primary and secondary prevention of cardiovascular events.^4–6

Adhesion, activation, and aggregation

Adhesion. Disruption of the vascular endothelium as a result of vulnerable plaque fissuring or ulceration exposes subendothelial thrombogenic collagen and von Willebrand factor to blood. Collagen engages platelets through their glycoprotein (GP) Ia, IIa, and VI receptors, and von Willebrand factor binds platelets through the GP Ib-IX-V receptor.

Activation. Once platelets adhere to the subendothelium, they undergo a conformational change and become activated. Simultaneous release of various autocrine and paracrine mediators including adenosine diphosphate, serotonin, epinephrine, thromboxane, and various ligand-receptor interactions all contribute to the activation cascade.  Adenosine diphosphate binds to the platelet receptor P2Y₁, leading to an increase in intracellular calcium, and it binds to P2Y₁₂, leading to a decrease in cyclic adenosine monophosphate, both of which cause GP IIb/IIIa receptor activation. Thromboxane A₂ released by platelets by cyclo-oxygenase 1 binds to alpha or beta variant receptors and contributes to GP IIb/IIIa activation through elevation of intracellular calcium levels.

Aggregation and thrombosis. Exposure of tissue factor to plasma following plaque rupture activates the coagulation cascade via the extrinsic pathway, which generates thrombin, a powerful platelet activator that causes thrombus formation via fibrin. Thrombin binds to protease-activated receptors PAR-1 and PAR-4 on platelets, causing an increase in intracellular calcium and a decrease in cyclic adenosine monophosphate with subsequent GP IIb/IIIa activation. GP IIb/IIIa facilitates platelet aggregation by binding to fibrinogen and forming a stable platelet thrombus.

In the early stages of thrombus formation, platelets predominate (“white” thrombi); further organization with fibrin results in older “red” thrombi. The stages of thrombi vary in non-ST-elevation and ST-elevation myocardial infarction and are prognostic markers of death.^4–8

PERCUTANEOUS INTERVENTION, RESTENOSIS, AND STENT THROMBOSIS

Percutaneous coronary intervention, the preferred means of revascularization for many patients, is performed emergently in patients with ST-elevation myocardial infarction, urgently in those with acute coronary syndromes without ST elevation, and electively in those with stable ischemic symptoms.

Percutaneous revascularization techniques have evolved from balloon angioplasty to bare-metal stents to drug-eluting stents, but each of these procedures has been associated with a periprocedural and postprocedural risk of thrombosis.

Balloon angioplasty was associated with vascular intimal injury, inciting elastic vascular recoil and smooth muscle cell proliferation leading to restenosis.

Bare-metal stents reduced the restenosis rate by eliminating vascular recoil, although restenosis still occurred within the stent because of neointimal proliferation of vascular smooth muscle cells. This was an important limitation, as both acute and subacute stent thrombosis were refractory to aggressive anticoagulation regimens that were associated with major bleeding complications and longer hospital length of stay. Stenting became mainstream practice only after the ISAR⁹ and STARS¹⁰ trials showed that dual antiplatelet therapy controlled stent thrombosis.

Drug-eluting stents coated with anti-proliferative and anti-inflammatory polymers markedly reduced in-stent restenosis rates by suppressing the initial vascular smooth-muscle proliferative response. However, they were still associated with late and very late stent thrombosis with incomplete endothelialization, even up to 40 months after implantation. Proposed mechanisms include incomplete stent apposition and inflammatory hypersensitivity reactions to the polymer coating. Incomplete stent apposition associated with low-velocity blood flow at the junction of the stent strut and vessel wall, together with delayed endothelialization, promotes platelet adhesion and aggregation, followed by thrombus formation.¹¹

Second-generation drug-eluting stents have thinner struts and more biocompatible polymers and are thought to favor more complete re-endothelialization, reducing the rates of stent thrombosis.^8,12,13

Predictors of early stent thrombosis

The Dutch Stent Thrombosis Registry and other studies looked at risk factors for stent thrombosis.^14,15

Procedure-related factors included:

• Stent undersizing
• Residual uncovered dissections after angioplasty
• Longer stents
• Low flow after angioplasty (< 3 on the 0–3 Thrombolysis in Myocardial Infarction [TIMI] scale).

Lesion-related factors included:

• Intermediate coronary artery disease both proximal and distal to the culprit lesions
• Bifurcation lesions.

Patient-related factors included:

• Low left ventricular ejection fraction
• Diabetes mellitus
• Peripheral arterial disease
Premature discontinuation of clopidogrel.

ANTIPLATELET AGENTS: MECHANISM OF ACTION

Various pathways play synergistic roles in platelet activation and aggregation and thrombus formation, and different antiplatelet agents inhibit these specific pathways, thus complementing each other and having additive effects (Figure 2, Table 1).^5,16–21

Aspirin inhibits cyclo-oxygenase 1

Cyclo-oxygenase 1, found in platelets, endothelial cells, and other cells, catalyzes the conversion of arachidonic acid to thromboxane A₂. Aspirin irreversibly inhibits cyclo-oxygenase 1 by acetylating its serine residue, preventing formation of thromboxane A₂ and preventing platelet activation and aggregation.

P2Y₁₂ ADP receptor antagonists

Clopidogrel and prasugrel are thienopyridine agents that irreversibly inhibit the P2Y₁₂ receptor, thereby preventing binding of adenosine diphosphate and the subsequent platelet activation-aggregation cascade. They are both prodrugs and require conversion by cytochrome P450 enzymes to active metabolites. Prasugrel is 10 times more potent than clopidogrel due to more efficient formation of its active metabolite, and it achieves a comparable effect on platelet inhibition 30 minutes faster than the peak effect of clopidogrel at 6 hours. The overall peak inhibitory effect of prasugrel is twice that of clopidogrel.²²

Ticagrelor, a cyclopentyl-triazolo-pyrimidine, directly and reversibly inhibits the P2Y₁₂ ADP receptor. Unlike clopidogrel and prasugrel, it does not need to be converted to an active metabolite, and it noncompetitively inhibits P2Y₁₂ at a site different from the adenosine diphosphate binding site.²³ Like prasugrel, ticagrelor inhibits platelet function more rapidly and more completely than clopidogrel.

Cangrelor, an intravenously administered analogue of adenosine triphosphate, reversibly inhibits the P2Y₁₂ receptor. It has undergone phase 3 trials but is not yet approved for clinical use.²⁴

WHY DUAL ANTIPLATELET THERAPY?

Aspirin is good, clopidogrel is better

Aspirin has a well-validated role in both primary and secondary prevention of coronary and noncoronary atherosclerotic vascular disease.

The CAPRIE trial found clopidogrel monotherapy to be superior to aspirin monotherapy in patients with established atherosclerotic vascular disease.²⁵

After stenting, short-term dual therapy is better than short-term warfarin

Thrombotic complications in the early postprocedural period were a major limitation of stenting, and existing anticoagulation regimens were ineffective in preventing them.^26,27

The ISAR trial studied the benefit of combined antiplatelet vs anticoagulant therapy after stent placement. Patients randomized to receive combined aspirin plus ticlopidine (an early P2Y₁₂ inhibitor) had significantly lower rates of primary cardiac, hemorrhagic, and vascular events at 30 days.⁹ Two other trials confirmed this finding.^28,29

STARS¹⁰ also confirmed the benefit of aspirin and ticlopidine after stenting. Patients were randomly assigned to aspirin alone, aspirin plus warfarin, or aspirin plus ticlopidine after stent placement. The rate of stent thrombosis at 30 days was significantly lower in the dual antiplatelet group than in the other two groups. The dual antiplatelet group had a higher rate of bleeding than the aspirin-alone group, but the rate was similar to that of the aspirin-plus-warfarin group.

Long-term dual antiplatelet therapy is beneficial in several situations

ISAR and STARS were landmark trials that showed stent thrombosis could be reduced by dual antiplatelet therapy for a 30-day period. However, the long-term role of dual antiplatelet therapy was still unknown.

The CURE trial^30–32 randomized patients presenting with acute coronary syndromes without ST elevation to receive clopidogrel plus aspirin or placebo plus aspirin for 3 to 12 months. The rate of the primary end point (cardiac death, nonfatal myocardial infarction, or stroke) was significantly lower in the clopidogrel-plus-aspirin group. A similar benefit of dual antiplatelet therapy was seen in the subgroup of patients who underwent percutaneous coronary intervention. Both pretreatment with clopidogrel plus aspirin for a median of 10 days prior to percutaneous intervention and continuing it for a mean of 9 months reduced major adverse cardiovascular events.

The CREDO trial²⁰ found that the combination of clopidogrel and aspirin significantly reduced the incidence of death, myocardial infarction, or stroke at 1 year after percutaneous coronary intervention. A subgroup of patients in this trial who had a longer pretreatment interval with a loading clopidogrel dose showed a benefit at 28 days, which was not as evident with a shorter loading dose interval.

The CLARITY-TIMI 28 trial^33,34 showed the advantage of adding clopidogrel to aspirin in patients receiving fibrinolytic therapy for ST-elevation myocardial infarction. Adding clopidogrel both improved the patency of the infarct-related artery and reduced ischemic complications. In patients who subsequently underwent percutaneous coronary intervention and stenting, clopidogrel pretreatment was associated with a significant decrease in ischemic complications before and after the procedure. There was no significant increase in bleeding complications in either group.

COMMIT/CCS 2³⁵ also showed the benefit of dual antiplatelet therapy in patients with ST-elevation myocardial infarction. Clopidogrel added to aspirin during the short-term in-hospital or postdischarge treatment period significantly reduced a composite end point of reinfarction, death, or stroke as well as death from any cause.

The CHARISMA trial^36–38 aimed to determine if patients who were more stable (ie, no recent acute coronary syndrome event or percutaneous coronary intervention) would benefit. Overall, CHARISMA showed no benefit of adding clopidogrel to aspirin compared with aspirin alone in a broad population of patients with established vascular disease (secondary prevention) or risk factors for vascular disease (primary prevention).

But importantly, though no benefit was seen in the primary prevention group, the large subgroup of patients with established atherosclerotic vascular disease (12,153 of the 15,603 patients in the trial) did benefit from dual antiplatelet therapy.^36,37 This subgroup showed an overall reduction in absolute risk of 1.5% (relative risk 0.88, P = .046) over a median follow-up of 27.6 months. This benefit was even more apparent in the 9,478 patients with prior myocardial infarction, stroke, or peripheral artery disease, for whom the relative risk reduction was 17.1% (P = .01) and the reduction in absolute risk 1.5%.³⁸

These results are comparable to the 2% absolute risk reduction in the CURE trial for similar end points over 9 months. In both studies, there was no significant increase in the risk of major bleeding or intracranial bleeding in the clopidogrel-plus-aspirin groups, although minor bleeding was increased by dual antiplatelet therapy.

The rate of severe bleeding, which was the primary safety end point in CHARISMA, was not significantly different in the clopidogrel-plus-aspirin group compared with the placebo-plus-aspirin group (relative risk 1.25, 95% CI 0.97–1.61, P = .09).

Thus, although the CHARISMA findings were negative overall, the positive finding observed in the predominant subgroup of patients with established vascular disease can therefore be considered supportive of the results of the subsequent trials discussed below.

The PEGASUS-TIMI 54 trial³⁹ studied the benefit of adding ticagrelor (60 or 90 mg) to low-dose aspirin in patients with stable coronary artery disease who had had a myocardial infarction 1 to 3 years earlier.

Confirming the results of the CHARISMA subgroup analysis, the incidence of the ischemic primary efficacy end point (a composite of cardiovascular death, myocardial infarction, and stroke) was significantly lower in both groups receiving ticagrelor plus aspirin compared with those receiving placebo plus aspirin. The Kaplan-Meier rate at 3 years for the ticagrelor 90 mg-plus-aspirin group was 7.85% vs 9.04% for the placebo-plus-aspirin group (hazard ratio 0.85, 95% confidence interval [CI] 0.75–0.96, P = .008). The rate for the ticagrelor 60 mg-plus-aspirin group was 7.77% vs 9.04% for the placebo-plus-aspirin group (hazard ratio 0.84, 95% CI 0.74–0.95, P = .004).

The rates of all TIMI major and minor bleeding, as well as bleeding requiring transfusion or discontinuation of the study drug, were significantly higher in both ticagrelor dosing groups than in the placebo group (P < .01 for both groups vs placebo). The rates of fatal bleeding and nonfatal intracranial hemorrhage were not significantly higher. Although there was an overall reduction in ischemic end points with the addition of ticagrelor, there was also a significantly higher incidence of bleeding in this group.

Comment. Thus, with or without percutaneous coronary intervention in acute coronary syndrome as well as in stable coronary artery disease, dual antiplatelet therapy was shown to improve outcomes and decrease ischemic complications compared with aspirin alone. It provided benefit in the setting of acute coronary syndrome (in the CURE trial) and percutaneous coronary intervention (in the CREDO trial) for up to 1 year.

Major questions remained to be addressed:

• Do the results of CREDO, which was performed before the current interventional era and the use of drug-eluting stents, reflect outcomes after current interventional practice?
• Could shorter periods of dual antiplatelet therapy be sufficient, especially with newer stents with less risk of late thrombosis?
• Does the benefit of dual antiplatelet therapy extend beyond the 1-year time period tested in those trials to date?

RECOMMENDATIONS FOR DOSING

The American College of Cardiology Foundation/American Heart Association guidelines for dosing of antiplatelet agents for non-ST-elevation myocardial infarction are summarized in Table 2, and those for ST-elevation myocardial infarction are summarized in Table 3.^1,2

WOULD SHORTER THERAPY AFTER STENTING WORK AS WELL?

The American College of Cardiology Foundation/American Heart Association currently recommend dual antiplatelet therapy for at least 12 months after drug-eluting stent placement, with shorter courses appropriate for patients who develop excessive bleeding complications or who are at high risk of bleeding.

Four trials (Table 4) evaluated whether shorter durations of dual antiplatelet therapy would suffice: SECURITY,⁴⁰ EXCELLENT,⁴¹ OPTIMIZE,⁴² and RESET.⁴³ All of them showed that short-duration therapy was not inferior to standard-duration therapy.⁴⁴ These studies were comparable in that:

• Patients were randomized at the time of percutaneous coronary intervention or within 24 hours of it.
• Most patients received a second-generation drug-eluting stent, with the following exceptions: in EXCELLENT,⁴¹ one-fourth of patients received a Cypher first-generation drug-eluting stent, and in RESET,⁴³ approximately one-fourth of the patients received a sirolimus-eluting stent in the standard-duration group for short lesions. Those patients with longer lesions in the RESET standard-duration group received an evero­limus drug-eluting stent.
• The second antiplatelet added to aspirin in all studies was clopidogrel, with the exception of the SECURITY trial, in which fewer than 2% of patients received ticagrelor or prasugrel.⁴⁰
• All the trials except RESET excluded patients who had had a myocardial infarction within 72 hours, and thus most patients studied had a lower risk profile.
• All of the trials sought to study noninferiority of short- vs standard-duration dual antiplatelet therapy, defined as the occurrence of a primary end point at 1 year (a composite of cardiovascular death, myocardial infarction, stroke, stent thrombosis, target vessel failure or revascularization, or bleeding).

Their low-risk patient populations and infrequent end points rendered these studies underpowered to make definitive conclusions about the relative efficacy of 6-months vs 12-months of dual antiplatelet therapy.

WOULD LONGER THERAPY BE BETTER?

The PRODIGY trial⁴⁵ assessed durations of dual antiplatelet therapy both shorter and longer than the conventional 1 year, randomizing patients undergoing placement of a bare-metal stent, first-generation drug-eluting stent, or second-generation drug-eluting stent to receive aspirin and clopidogrel for either 6 months or 24 months. The study showed no significant difference in primary outcomes in the short- or long-duration groups.

Other trials that compared the standard 12 months of dual antiplatelet therapy with extended duration beyond 12 months were DAPT,³ ARCTIC-Interruption,⁴⁶ and DES-LATE.⁴⁷ The trials were comparable in that:

• All patients were randomized after completing 12 months of dual antiplatelet therapy following drug-eluting stent placement.
• All patients who were included had been free of major cardiac ischemic events or bleeding during the 12 months following stent placement.
• The primary aim of all three studies was to compare primary end points in groups receiving aspirin alone vs extended dual antiplatelet therapy. The primary end point was a composite of death due to a cardiovascular cause, nonfatal myocardial infarction, stroke, or stent thrombosis.
• The principal safety end point was bleeding.

Although the two earlier studies (ARCTIC-Interruption and DES-LATE) did not show any benefit of extended dual antiplatelet therapy compared with the standard 12-month duration, the recent DAPT study did.

The DAPT study

The DAPT study³ was an international, multicenter, placebo-controlled, double-blind randomized trial designed to examine the benefit of dual antiplatelet therapy beyond 1 year in a patient population large enough to provide definitive assessment of benefit and risk.

A total of 9,961 patients who received drug-eluting stents were randomized after 12 months of dual antiplatelet therapy to receive either a thienopyridine (clopidogrel or prasugrel) plus aspirin or placebo plus aspirin. They were followed for an additional 18 months. The coprimary efficacy end points were stent thrombosis and a composite of death, myocardial infarction, or stroke, while the primary safety end point was moderate or severe bleeding. The patients were also observed from months 30 to 33 on aspirin alone after stopping the thienopyridine.

Results. Longer therapy substantially reduced the risks of stent thrombosis (hazard ratio [HR] 0.29, 95% confidence interval [CI] 0.17–0.48) and the composite ischemic end point (HR 0.71, 95% CI 0.59–0.85). Follow-up during the 3-month thienopyridine discontinuation phase starting at 30 months revealed convergence of the ischemic event-rate curves in the two groups, which suggested that continuing dual antiplatelet therapy beyond 30 months might have been beneficial. Myocardial infarction unrelated to stent thrombosis accounted for 55% of the treatment benefit of dual antiplatelet therapy.

The risk of bleeding was higher in the thienopyridine group during the treatment period (2.5% vs 1.6%, P = .001). There was also a higher rate of noncardiovascular mortality in the thienopyridine group, although this difference may have been due to chance.^3,48

Why were the results different?

All three trials included first- and second-generation drug-eluting stents, with different proportions in different trials. In ARCTIC-Interruption,⁴⁶ 43% of the patients in the continuation group had a first-generation stent, as did 64% of the patients in the dual antiplatelet group of DES-LATE.⁴⁷ In the DAPT trial,³ 38% of the patients in the longer-duration arm had a first-generation stent, and in 26% of cases it was a paclitaxel-eluting stent.

Only clopidogrel was used as the second antiplatelet agent in DES-LATE, whereas prasugrel was used in 10% of patients in ARCTIC-Interruption and 35% in DAPT.

Yet none of these differences seem to explain the differences in outcome among the studies. ARCTIC-Interruption and DES-LATE did not show any benefit of continued dual antiplatelet therapy beyond 12 months. DAPT showed benefit of extended therapy with prasugrel or with clopidogrel, and with first-generation or second-generation drug-eluting stents. The most likely explanation for the different results was that DAPT was the only trial sufficiently powered to definitively assess the end points, including stent thrombosis.

A balance between ischemic efficacy and bleeding risk is the major consideration with any antithrombotic and antiplatelet therapy. In the three largest trials we discussed (the vascular disease subgroups of CHARISMA,³⁸ PEGASUS,³⁹ and DAPT³), comparison of the prespecified efficacy and safety end points of each trial suggests that dual antiplatelet therapy has a net benefit, particularly given the irreversible nature of ischemic end points.

In CHARISMA,³⁸ 60 cardiovascular deaths, myocardial infarctions, or strokes were prevented per year per 10,000 patients treated, at the cost of 28 excess moderate bleeding events.

In PEGASUS,³⁹ 42 cardiovascular deaths, myocardial infarctions, or strokes were prevented, at the cost of 79 excess bleeding events requiring transfusion.

In DAPT (a selected population who had tolerated dual antiplatelet therapy for 1 year), 106 deaths, myocardial infarctions, or stroke events were prevented, at the cost of 47 excess moderate bleeding events^.3

Indirect comparisons between trials are problematic, given different end point definitions, populations, and background therapies. But their results suggest that less-intensive inhibition with clopidogrel as the second antiplatelet long-term (as in CHARISMA) may provide the best balance of benefit vs risk.

BALANCING RISK AND BENEFIT

The evidence is unequivocal that dual antiplatelet therapy suppresses coronary ischemic complications resulting from thrombosis at sites of spontaneous plaque rupture following acute coronary syndromes or mechanical plaque disruption and foreign body implantation associated with percutaneous coronary intervention.

Three large-scale trials (DAPT,³ PEGASUS,³⁹ and the secondary prevention subgroup of CHARISMA³⁸) showed that the protective effect of dual antiplatelet therapy continues with prolonged therapy in patients who have experienced an acute coronary syndrome event or have received a drug-eluting stent. That benefit seems to be due to the action of these therapies on the culprit vessel (the one that caused the acute coronary syndrome or the site of stenting), as well as nonculprit arteries, emphasizing that dual antiplatelet therapy protects against atherosclerosis progression and future plaque rupture events.

For the durations studied in the longest trials thus far, 30 months (DAPT³) and 36 months (PEGASUS³⁹), event curves continue to diverge, indicating that the advantage of dual antiplatelet therapy may persist for an indefinite period of time. Thus, indefinite therapy with dual antiplatelet agents can be supported, particularly in patients with advanced coronary artery disease or those who have had multiple coronary events.

We believe that the balance of evidence suggests that smaller studies that failed to show a benefit of longer-term therapy were underpowered to do so.

The ischemic protection is associated with the adverse effect of increased bleeding risk. Unfortunately, there has been little success in guiding dual antiplatelet therapy based on ischemic vs bleeding risk, in part because the same factors that predict risk of ischemic complications seem to predict increased susceptibility to bleeding. Nevertheless, indirect comparisons between studies suggest that for longer-term therapy clopidogrel may be superior to ticagrelor or prasugrel: the absolute excess bleeding risk with dual antiplatelet therapy vs aspirin in the CHARISMA secondary prevention subgroup was less than that in PEGASUS, with similar absolute reductions in ischemic events. So while the TRITON-TIMI 38²² and PLATO²³ trials support the superiority of prasugrel or ticagrelor over clopidogrel for the first year after acute coronary syndrome, subsequent years of therapy may best be provided with clopidogrel.

Some patients may have identifiable factors that place them at very high risk of bleeding—need for surgical procedures, need for anticoagulation, or occurrence of bleeding complications or excessive “nuisance bleeding.” In those patients, the data suggest that dual antiplatelet therapy could be discontinued after 6 months, or perhaps even 3 months in the highest bleeding risk circumstances after second-generation drug-eluting stent placement.

WOEST⁴⁹ was an open-label randomized controlled trial that studied the safety of antiplatelet regimens in patients on anticoagulation requiring percutaneous coronary interventions. Patients were randomized to double therapy with anticoagulant and clopidogrel vs triple therapy with additional aspirin following percutaneous coronary intervention. The primary end point was bleeding events within 1 year. Clopidogrel without aspirin was associated with significantly fewer bleeding events compared with triple therapy, with no increase in adverse ischemic events. The strategy tested in the WOEST trial seems reasonable in the specific group of patients who require ongoing anticoagulant therapy after drug-eluting stent placement, recognizing that the trial was somewhat underpowered to make definitive conclusions, particularly in patients at high risk for stent thrombosis.

Based on the results of PEGASUS and the CHARISMA subgroup with established ischemic burden, in which dual antiplatelet therapy was started after an interruption following the index coronary event, it is also reasonable to restart long-term dual antiplatelet therapy in patients who require interruption for short-term indications such as a surgical procedure.            

Percutaneous coronary intervention for acute coronary syndromes has evolved, and so, hand in hand, has antiplatelet therapy. With the advent of clopidogrel and newer agents, several studies demonstrated the benefits of dual antiplatelet therapy in preventing major vascular ischemic complications. The findings culminated in a guideline recommendation for at least 12 months of dual antiplatelet therapy after placement of a drug-eluting stent, when feasible—a class I recommendation (treatment should be given), level of evidence B (limited populations evaluated).^1,2 But extending dual antiplatelet therapy beyond 12 months had no strong favorable evidence until the recent Dual Antiplatelet Therapy (DAPT) study³ shed light on this topic.

Here, we review the evidence thus far on the optimal duration of dual antiplatelet therapy in the secondary prevention of coronary artery disease.

PLATELETS IN ACUTE CORONARY SYNDROMES AND STENT THROMBOSIS

Acute coronary syndromes begin with fissuring or ulceration of a vulnerable atherosclerotic plaque, followed by thrombosis and occlusion, mediated by platelet adhesion, activation, and aggregation (Figure 1). Transient occlusion results in unstable angina or non-ST-elevation myocardial infarction, while total occlusion usually results in ST-elevation myocardial infarction.

Platelet aggregation is prominent among the mechanisms leading to stent thrombosis and vaso-occlusive ischemic complications after percutaneous coronary intervention. Thus, antiplatelet agents play a vital role in both primary and secondary prevention of cardiovascular events.^4–6

Adhesion, activation, and aggregation

Adhesion. Disruption of the vascular endothelium as a result of vulnerable plaque fissuring or ulceration exposes subendothelial thrombogenic collagen and von Willebrand factor to blood. Collagen engages platelets through their glycoprotein (GP) Ia, IIa, and VI receptors, and von Willebrand factor binds platelets through the GP Ib-IX-V receptor.

Activation. Once platelets adhere to the subendothelium, they undergo a conformational change and become activated. Simultaneous release of various autocrine and paracrine mediators including adenosine diphosphate, serotonin, epinephrine, thromboxane, and various ligand-receptor interactions all contribute to the activation cascade.  Adenosine diphosphate binds to the platelet receptor P2Y₁, leading to an increase in intracellular calcium, and it binds to P2Y₁₂, leading to a decrease in cyclic adenosine monophosphate, both of which cause GP IIb/IIIa receptor activation. Thromboxane A₂ released by platelets by cyclo-oxygenase 1 binds to alpha or beta variant receptors and contributes to GP IIb/IIIa activation through elevation of intracellular calcium levels.

Aggregation and thrombosis. Exposure of tissue factor to plasma following plaque rupture activates the coagulation cascade via the extrinsic pathway, which generates thrombin, a powerful platelet activator that causes thrombus formation via fibrin. Thrombin binds to protease-activated receptors PAR-1 and PAR-4 on platelets, causing an increase in intracellular calcium and a decrease in cyclic adenosine monophosphate with subsequent GP IIb/IIIa activation. GP IIb/IIIa facilitates platelet aggregation by binding to fibrinogen and forming a stable platelet thrombus.

In the early stages of thrombus formation, platelets predominate (“white” thrombi); further organization with fibrin results in older “red” thrombi. The stages of thrombi vary in non-ST-elevation and ST-elevation myocardial infarction and are prognostic markers of death.^4–8

PERCUTANEOUS INTERVENTION, RESTENOSIS, AND STENT THROMBOSIS

Percutaneous coronary intervention, the preferred means of revascularization for many patients, is performed emergently in patients with ST-elevation myocardial infarction, urgently in those with acute coronary syndromes without ST elevation, and electively in those with stable ischemic symptoms.

Percutaneous revascularization techniques have evolved from balloon angioplasty to bare-metal stents to drug-eluting stents, but each of these procedures has been associated with a periprocedural and postprocedural risk of thrombosis.

Balloon angioplasty was associated with vascular intimal injury, inciting elastic vascular recoil and smooth muscle cell proliferation leading to restenosis.

Bare-metal stents reduced the restenosis rate by eliminating vascular recoil, although restenosis still occurred within the stent because of neointimal proliferation of vascular smooth muscle cells. This was an important limitation, as both acute and subacute stent thrombosis were refractory to aggressive anticoagulation regimens that were associated with major bleeding complications and longer hospital length of stay. Stenting became mainstream practice only after the ISAR⁹ and STARS¹⁰ trials showed that dual antiplatelet therapy controlled stent thrombosis.

Drug-eluting stents coated with anti-proliferative and anti-inflammatory polymers markedly reduced in-stent restenosis rates by suppressing the initial vascular smooth-muscle proliferative response. However, they were still associated with late and very late stent thrombosis with incomplete endothelialization, even up to 40 months after implantation. Proposed mechanisms include incomplete stent apposition and inflammatory hypersensitivity reactions to the polymer coating. Incomplete stent apposition associated with low-velocity blood flow at the junction of the stent strut and vessel wall, together with delayed endothelialization, promotes platelet adhesion and aggregation, followed by thrombus formation.¹¹

Second-generation drug-eluting stents have thinner struts and more biocompatible polymers and are thought to favor more complete re-endothelialization, reducing the rates of stent thrombosis.^8,12,13

Predictors of early stent thrombosis

The Dutch Stent Thrombosis Registry and other studies looked at risk factors for stent thrombosis.^14,15

Procedure-related factors included:

• Stent undersizing
• Residual uncovered dissections after angioplasty
• Longer stents
• Low flow after angioplasty (< 3 on the 0–3 Thrombolysis in Myocardial Infarction [TIMI] scale).

Lesion-related factors included:

• Intermediate coronary artery disease both proximal and distal to the culprit lesions
• Bifurcation lesions.

Patient-related factors included:

• Low left ventricular ejection fraction
• Diabetes mellitus
• Peripheral arterial disease
Premature discontinuation of clopidogrel.

ANTIPLATELET AGENTS: MECHANISM OF ACTION

Various pathways play synergistic roles in platelet activation and aggregation and thrombus formation, and different antiplatelet agents inhibit these specific pathways, thus complementing each other and having additive effects (Figure 2, Table 1).^5,16–21

Aspirin inhibits cyclo-oxygenase 1

Cyclo-oxygenase 1, found in platelets, endothelial cells, and other cells, catalyzes the conversion of arachidonic acid to thromboxane A₂. Aspirin irreversibly inhibits cyclo-oxygenase 1 by acetylating its serine residue, preventing formation of thromboxane A₂ and preventing platelet activation and aggregation.

P2Y₁₂ ADP receptor antagonists

Clopidogrel and prasugrel are thienopyridine agents that irreversibly inhibit the P2Y₁₂ receptor, thereby preventing binding of adenosine diphosphate and the subsequent platelet activation-aggregation cascade. They are both prodrugs and require conversion by cytochrome P450 enzymes to active metabolites. Prasugrel is 10 times more potent than clopidogrel due to more efficient formation of its active metabolite, and it achieves a comparable effect on platelet inhibition 30 minutes faster than the peak effect of clopidogrel at 6 hours. The overall peak inhibitory effect of prasugrel is twice that of clopidogrel.²²

Ticagrelor, a cyclopentyl-triazolo-pyrimidine, directly and reversibly inhibits the P2Y₁₂ ADP receptor. Unlike clopidogrel and prasugrel, it does not need to be converted to an active metabolite, and it noncompetitively inhibits P2Y₁₂ at a site different from the adenosine diphosphate binding site.²³ Like prasugrel, ticagrelor inhibits platelet function more rapidly and more completely than clopidogrel.

Cangrelor, an intravenously administered analogue of adenosine triphosphate, reversibly inhibits the P2Y₁₂ receptor. It has undergone phase 3 trials but is not yet approved for clinical use.²⁴

WHY DUAL ANTIPLATELET THERAPY?

Aspirin is good, clopidogrel is better

Aspirin has a well-validated role in both primary and secondary prevention of coronary and noncoronary atherosclerotic vascular disease.

The CAPRIE trial found clopidogrel monotherapy to be superior to aspirin monotherapy in patients with established atherosclerotic vascular disease.²⁵

After stenting, short-term dual therapy is better than short-term warfarin

Thrombotic complications in the early postprocedural period were a major limitation of stenting, and existing anticoagulation regimens were ineffective in preventing them.^26,27

The ISAR trial studied the benefit of combined antiplatelet vs anticoagulant therapy after stent placement. Patients randomized to receive combined aspirin plus ticlopidine (an early P2Y₁₂ inhibitor) had significantly lower rates of primary cardiac, hemorrhagic, and vascular events at 30 days.⁹ Two other trials confirmed this finding.^28,29

STARS¹⁰ also confirmed the benefit of aspirin and ticlopidine after stenting. Patients were randomly assigned to aspirin alone, aspirin plus warfarin, or aspirin plus ticlopidine after stent placement. The rate of stent thrombosis at 30 days was significantly lower in the dual antiplatelet group than in the other two groups. The dual antiplatelet group had a higher rate of bleeding than the aspirin-alone group, but the rate was similar to that of the aspirin-plus-warfarin group.

Long-term dual antiplatelet therapy is beneficial in several situations

ISAR and STARS were landmark trials that showed stent thrombosis could be reduced by dual antiplatelet therapy for a 30-day period. However, the long-term role of dual antiplatelet therapy was still unknown.

The CURE trial^30–32 randomized patients presenting with acute coronary syndromes without ST elevation to receive clopidogrel plus aspirin or placebo plus aspirin for 3 to 12 months. The rate of the primary end point (cardiac death, nonfatal myocardial infarction, or stroke) was significantly lower in the clopidogrel-plus-aspirin group. A similar benefit of dual antiplatelet therapy was seen in the subgroup of patients who underwent percutaneous coronary intervention. Both pretreatment with clopidogrel plus aspirin for a median of 10 days prior to percutaneous intervention and continuing it for a mean of 9 months reduced major adverse cardiovascular events.

The CREDO trial²⁰ found that the combination of clopidogrel and aspirin significantly reduced the incidence of death, myocardial infarction, or stroke at 1 year after percutaneous coronary intervention. A subgroup of patients in this trial who had a longer pretreatment interval with a loading clopidogrel dose showed a benefit at 28 days, which was not as evident with a shorter loading dose interval.

The CLARITY-TIMI 28 trial^33,34 showed the advantage of adding clopidogrel to aspirin in patients receiving fibrinolytic therapy for ST-elevation myocardial infarction. Adding clopidogrel both improved the patency of the infarct-related artery and reduced ischemic complications. In patients who subsequently underwent percutaneous coronary intervention and stenting, clopidogrel pretreatment was associated with a significant decrease in ischemic complications before and after the procedure. There was no significant increase in bleeding complications in either group.

COMMIT/CCS 2³⁵ also showed the benefit of dual antiplatelet therapy in patients with ST-elevation myocardial infarction. Clopidogrel added to aspirin during the short-term in-hospital or postdischarge treatment period significantly reduced a composite end point of reinfarction, death, or stroke as well as death from any cause.

The CHARISMA trial^36–38 aimed to determine if patients who were more stable (ie, no recent acute coronary syndrome event or percutaneous coronary intervention) would benefit. Overall, CHARISMA showed no benefit of adding clopidogrel to aspirin compared with aspirin alone in a broad population of patients with established vascular disease (secondary prevention) or risk factors for vascular disease (primary prevention).

But importantly, though no benefit was seen in the primary prevention group, the large subgroup of patients with established atherosclerotic vascular disease (12,153 of the 15,603 patients in the trial) did benefit from dual antiplatelet therapy.^36,37 This subgroup showed an overall reduction in absolute risk of 1.5% (relative risk 0.88, P = .046) over a median follow-up of 27.6 months. This benefit was even more apparent in the 9,478 patients with prior myocardial infarction, stroke, or peripheral artery disease, for whom the relative risk reduction was 17.1% (P = .01) and the reduction in absolute risk 1.5%.³⁸

These results are comparable to the 2% absolute risk reduction in the CURE trial for similar end points over 9 months. In both studies, there was no significant increase in the risk of major bleeding or intracranial bleeding in the clopidogrel-plus-aspirin groups, although minor bleeding was increased by dual antiplatelet therapy.

The rate of severe bleeding, which was the primary safety end point in CHARISMA, was not significantly different in the clopidogrel-plus-aspirin group compared with the placebo-plus-aspirin group (relative risk 1.25, 95% CI 0.97–1.61, P = .09).

Thus, although the CHARISMA findings were negative overall, the positive finding observed in the predominant subgroup of patients with established vascular disease can therefore be considered supportive of the results of the subsequent trials discussed below.

The PEGASUS-TIMI 54 trial³⁹ studied the benefit of adding ticagrelor (60 or 90 mg) to low-dose aspirin in patients with stable coronary artery disease who had had a myocardial infarction 1 to 3 years earlier.

Confirming the results of the CHARISMA subgroup analysis, the incidence of the ischemic primary efficacy end point (a composite of cardiovascular death, myocardial infarction, and stroke) was significantly lower in both groups receiving ticagrelor plus aspirin compared with those receiving placebo plus aspirin. The Kaplan-Meier rate at 3 years for the ticagrelor 90 mg-plus-aspirin group was 7.85% vs 9.04% for the placebo-plus-aspirin group (hazard ratio 0.85, 95% confidence interval [CI] 0.75–0.96, P = .008). The rate for the ticagrelor 60 mg-plus-aspirin group was 7.77% vs 9.04% for the placebo-plus-aspirin group (hazard ratio 0.84, 95% CI 0.74–0.95, P = .004).

The rates of all TIMI major and minor bleeding, as well as bleeding requiring transfusion or discontinuation of the study drug, were significantly higher in both ticagrelor dosing groups than in the placebo group (P < .01 for both groups vs placebo). The rates of fatal bleeding and nonfatal intracranial hemorrhage were not significantly higher. Although there was an overall reduction in ischemic end points with the addition of ticagrelor, there was also a significantly higher incidence of bleeding in this group.

Comment. Thus, with or without percutaneous coronary intervention in acute coronary syndrome as well as in stable coronary artery disease, dual antiplatelet therapy was shown to improve outcomes and decrease ischemic complications compared with aspirin alone. It provided benefit in the setting of acute coronary syndrome (in the CURE trial) and percutaneous coronary intervention (in the CREDO trial) for up to 1 year.

Major questions remained to be addressed:

• Do the results of CREDO, which was performed before the current interventional era and the use of drug-eluting stents, reflect outcomes after current interventional practice?
• Could shorter periods of dual antiplatelet therapy be sufficient, especially with newer stents with less risk of late thrombosis?
• Does the benefit of dual antiplatelet therapy extend beyond the 1-year time period tested in those trials to date?

RECOMMENDATIONS FOR DOSING

The American College of Cardiology Foundation/American Heart Association guidelines for dosing of antiplatelet agents for non-ST-elevation myocardial infarction are summarized in Table 2, and those for ST-elevation myocardial infarction are summarized in Table 3.^1,2

WOULD SHORTER THERAPY AFTER STENTING WORK AS WELL?

The American College of Cardiology Foundation/American Heart Association currently recommend dual antiplatelet therapy for at least 12 months after drug-eluting stent placement, with shorter courses appropriate for patients who develop excessive bleeding complications or who are at high risk of bleeding.

Four trials (Table 4) evaluated whether shorter durations of dual antiplatelet therapy would suffice: SECURITY,⁴⁰ EXCELLENT,⁴¹ OPTIMIZE,⁴² and RESET.⁴³ All of them showed that short-duration therapy was not inferior to standard-duration therapy.⁴⁴ These studies were comparable in that:

• Patients were randomized at the time of percutaneous coronary intervention or within 24 hours of it.
• Most patients received a second-generation drug-eluting stent, with the following exceptions: in EXCELLENT,⁴¹ one-fourth of patients received a Cypher first-generation drug-eluting stent, and in RESET,⁴³ approximately one-fourth of the patients received a sirolimus-eluting stent in the standard-duration group for short lesions. Those patients with longer lesions in the RESET standard-duration group received an evero­limus drug-eluting stent.
• The second antiplatelet added to aspirin in all studies was clopidogrel, with the exception of the SECURITY trial, in which fewer than 2% of patients received ticagrelor or prasugrel.⁴⁰
• All the trials except RESET excluded patients who had had a myocardial infarction within 72 hours, and thus most patients studied had a lower risk profile.
• All of the trials sought to study noninferiority of short- vs standard-duration dual antiplatelet therapy, defined as the occurrence of a primary end point at 1 year (a composite of cardiovascular death, myocardial infarction, stroke, stent thrombosis, target vessel failure or revascularization, or bleeding).

Their low-risk patient populations and infrequent end points rendered these studies underpowered to make definitive conclusions about the relative efficacy of 6-months vs 12-months of dual antiplatelet therapy.

WOULD LONGER THERAPY BE BETTER?

The PRODIGY trial⁴⁵ assessed durations of dual antiplatelet therapy both shorter and longer than the conventional 1 year, randomizing patients undergoing placement of a bare-metal stent, first-generation drug-eluting stent, or second-generation drug-eluting stent to receive aspirin and clopidogrel for either 6 months or 24 months. The study showed no significant difference in primary outcomes in the short- or long-duration groups.

Other trials that compared the standard 12 months of dual antiplatelet therapy with extended duration beyond 12 months were DAPT,³ ARCTIC-Interruption,⁴⁶ and DES-LATE.⁴⁷ The trials were comparable in that:

• All patients were randomized after completing 12 months of dual antiplatelet therapy following drug-eluting stent placement.
• All patients who were included had been free of major cardiac ischemic events or bleeding during the 12 months following stent placement.
• The primary aim of all three studies was to compare primary end points in groups receiving aspirin alone vs extended dual antiplatelet therapy. The primary end point was a composite of death due to a cardiovascular cause, nonfatal myocardial infarction, stroke, or stent thrombosis.
• The principal safety end point was bleeding.

Although the two earlier studies (ARCTIC-Interruption and DES-LATE) did not show any benefit of extended dual antiplatelet therapy compared with the standard 12-month duration, the recent DAPT study did.

The DAPT study

The DAPT study³ was an international, multicenter, placebo-controlled, double-blind randomized trial designed to examine the benefit of dual antiplatelet therapy beyond 1 year in a patient population large enough to provide definitive assessment of benefit and risk.

A total of 9,961 patients who received drug-eluting stents were randomized after 12 months of dual antiplatelet therapy to receive either a thienopyridine (clopidogrel or prasugrel) plus aspirin or placebo plus aspirin. They were followed for an additional 18 months. The coprimary efficacy end points were stent thrombosis and a composite of death, myocardial infarction, or stroke, while the primary safety end point was moderate or severe bleeding. The patients were also observed from months 30 to 33 on aspirin alone after stopping the thienopyridine.

Results. Longer therapy substantially reduced the risks of stent thrombosis (hazard ratio [HR] 0.29, 95% confidence interval [CI] 0.17–0.48) and the composite ischemic end point (HR 0.71, 95% CI 0.59–0.85). Follow-up during the 3-month thienopyridine discontinuation phase starting at 30 months revealed convergence of the ischemic event-rate curves in the two groups, which suggested that continuing dual antiplatelet therapy beyond 30 months might have been beneficial. Myocardial infarction unrelated to stent thrombosis accounted for 55% of the treatment benefit of dual antiplatelet therapy.

The risk of bleeding was higher in the thienopyridine group during the treatment period (2.5% vs 1.6%, P = .001). There was also a higher rate of noncardiovascular mortality in the thienopyridine group, although this difference may have been due to chance.^3,48

Why were the results different?

All three trials included first- and second-generation drug-eluting stents, with different proportions in different trials. In ARCTIC-Interruption,⁴⁶ 43% of the patients in the continuation group had a first-generation stent, as did 64% of the patients in the dual antiplatelet group of DES-LATE.⁴⁷ In the DAPT trial,³ 38% of the patients in the longer-duration arm had a first-generation stent, and in 26% of cases it was a paclitaxel-eluting stent.

Only clopidogrel was used as the second antiplatelet agent in DES-LATE, whereas prasugrel was used in 10% of patients in ARCTIC-Interruption and 35% in DAPT.

Yet none of these differences seem to explain the differences in outcome among the studies. ARCTIC-Interruption and DES-LATE did not show any benefit of continued dual antiplatelet therapy beyond 12 months. DAPT showed benefit of extended therapy with prasugrel or with clopidogrel, and with first-generation or second-generation drug-eluting stents. The most likely explanation for the different results was that DAPT was the only trial sufficiently powered to definitively assess the end points, including stent thrombosis.

A balance between ischemic efficacy and bleeding risk is the major consideration with any antithrombotic and antiplatelet therapy. In the three largest trials we discussed (the vascular disease subgroups of CHARISMA,³⁸ PEGASUS,³⁹ and DAPT³), comparison of the prespecified efficacy and safety end points of each trial suggests that dual antiplatelet therapy has a net benefit, particularly given the irreversible nature of ischemic end points.

In CHARISMA,³⁸ 60 cardiovascular deaths, myocardial infarctions, or strokes were prevented per year per 10,000 patients treated, at the cost of 28 excess moderate bleeding events.

In PEGASUS,³⁹ 42 cardiovascular deaths, myocardial infarctions, or strokes were prevented, at the cost of 79 excess bleeding events requiring transfusion.

In DAPT (a selected population who had tolerated dual antiplatelet therapy for 1 year), 106 deaths, myocardial infarctions, or stroke events were prevented, at the cost of 47 excess moderate bleeding events^.3

Indirect comparisons between trials are problematic, given different end point definitions, populations, and background therapies. But their results suggest that less-intensive inhibition with clopidogrel as the second antiplatelet long-term (as in CHARISMA) may provide the best balance of benefit vs risk.

BALANCING RISK AND BENEFIT

The evidence is unequivocal that dual antiplatelet therapy suppresses coronary ischemic complications resulting from thrombosis at sites of spontaneous plaque rupture following acute coronary syndromes or mechanical plaque disruption and foreign body implantation associated with percutaneous coronary intervention.

Three large-scale trials (DAPT,³ PEGASUS,³⁹ and the secondary prevention subgroup of CHARISMA³⁸) showed that the protective effect of dual antiplatelet therapy continues with prolonged therapy in patients who have experienced an acute coronary syndrome event or have received a drug-eluting stent. That benefit seems to be due to the action of these therapies on the culprit vessel (the one that caused the acute coronary syndrome or the site of stenting), as well as nonculprit arteries, emphasizing that dual antiplatelet therapy protects against atherosclerosis progression and future plaque rupture events.

For the durations studied in the longest trials thus far, 30 months (DAPT³) and 36 months (PEGASUS³⁹), event curves continue to diverge, indicating that the advantage of dual antiplatelet therapy may persist for an indefinite period of time. Thus, indefinite therapy with dual antiplatelet agents can be supported, particularly in patients with advanced coronary artery disease or those who have had multiple coronary events.

We believe that the balance of evidence suggests that smaller studies that failed to show a benefit of longer-term therapy were underpowered to do so.

The ischemic protection is associated with the adverse effect of increased bleeding risk. Unfortunately, there has been little success in guiding dual antiplatelet therapy based on ischemic vs bleeding risk, in part because the same factors that predict risk of ischemic complications seem to predict increased susceptibility to bleeding. Nevertheless, indirect comparisons between studies suggest that for longer-term therapy clopidogrel may be superior to ticagrelor or prasugrel: the absolute excess bleeding risk with dual antiplatelet therapy vs aspirin in the CHARISMA secondary prevention subgroup was less than that in PEGASUS, with similar absolute reductions in ischemic events. So while the TRITON-TIMI 38²² and PLATO²³ trials support the superiority of prasugrel or ticagrelor over clopidogrel for the first year after acute coronary syndrome, subsequent years of therapy may best be provided with clopidogrel.

Some patients may have identifiable factors that place them at very high risk of bleeding—need for surgical procedures, need for anticoagulation, or occurrence of bleeding complications or excessive “nuisance bleeding.” In those patients, the data suggest that dual antiplatelet therapy could be discontinued after 6 months, or perhaps even 3 months in the highest bleeding risk circumstances after second-generation drug-eluting stent placement.

WOEST⁴⁹ was an open-label randomized controlled trial that studied the safety of antiplatelet regimens in patients on anticoagulation requiring percutaneous coronary interventions. Patients were randomized to double therapy with anticoagulant and clopidogrel vs triple therapy with additional aspirin following percutaneous coronary intervention. The primary end point was bleeding events within 1 year. Clopidogrel without aspirin was associated with significantly fewer bleeding events compared with triple therapy, with no increase in adverse ischemic events. The strategy tested in the WOEST trial seems reasonable in the specific group of patients who require ongoing anticoagulant therapy after drug-eluting stent placement, recognizing that the trial was somewhat underpowered to make definitive conclusions, particularly in patients at high risk for stent thrombosis.

Based on the results of PEGASUS and the CHARISMA subgroup with established ischemic burden, in which dual antiplatelet therapy was started after an interruption following the index coronary event, it is also reasonable to restart long-term dual antiplatelet therapy in patients who require interruption for short-term indications such as a surgical procedure.            

Percutaneous coronary intervention for acute coronary syndromes has evolved, and so, hand in hand, has antiplatelet therapy. With the advent of clopidogrel and newer agents, several studies demonstrated the benefits of dual antiplatelet therapy in preventing major vascular ischemic complications. The findings culminated in a guideline recommendation for at least 12 months of dual antiplatelet therapy after placement of a drug-eluting stent, when feasible—a class I recommendation (treatment should be given), level of evidence B (limited populations evaluated).^1,2 But extending dual antiplatelet therapy beyond 12 months had no strong favorable evidence until the recent Dual Antiplatelet Therapy (DAPT) study³ shed light on this topic.

Here, we review the evidence thus far on the optimal duration of dual antiplatelet therapy in the secondary prevention of coronary artery disease.

PLATELETS IN ACUTE CORONARY SYNDROMES AND STENT THROMBOSIS

Acute coronary syndromes begin with fissuring or ulceration of a vulnerable atherosclerotic plaque, followed by thrombosis and occlusion, mediated by platelet adhesion, activation, and aggregation (Figure 1). Transient occlusion results in unstable angina or non-ST-elevation myocardial infarction, while total occlusion usually results in ST-elevation myocardial infarction.

Platelet aggregation is prominent among the mechanisms leading to stent thrombosis and vaso-occlusive ischemic complications after percutaneous coronary intervention. Thus, antiplatelet agents play a vital role in both primary and secondary prevention of cardiovascular events.^4–6

Adhesion, activation, and aggregation

Adhesion. Disruption of the vascular endothelium as a result of vulnerable plaque fissuring or ulceration exposes subendothelial thrombogenic collagen and von Willebrand factor to blood. Collagen engages platelets through their glycoprotein (GP) Ia, IIa, and VI receptors, and von Willebrand factor binds platelets through the GP Ib-IX-V receptor.

Activation. Once platelets adhere to the subendothelium, they undergo a conformational change and become activated. Simultaneous release of various autocrine and paracrine mediators including adenosine diphosphate, serotonin, epinephrine, thromboxane, and various ligand-receptor interactions all contribute to the activation cascade.  Adenosine diphosphate binds to the platelet receptor P2Y₁, leading to an increase in intracellular calcium, and it binds to P2Y₁₂, leading to a decrease in cyclic adenosine monophosphate, both of which cause GP IIb/IIIa receptor activation. Thromboxane A₂ released by platelets by cyclo-oxygenase 1 binds to alpha or beta variant receptors and contributes to GP IIb/IIIa activation through elevation of intracellular calcium levels.

Aggregation and thrombosis. Exposure of tissue factor to plasma following plaque rupture activates the coagulation cascade via the extrinsic pathway, which generates thrombin, a powerful platelet activator that causes thrombus formation via fibrin. Thrombin binds to protease-activated receptors PAR-1 and PAR-4 on platelets, causing an increase in intracellular calcium and a decrease in cyclic adenosine monophosphate with subsequent GP IIb/IIIa activation. GP IIb/IIIa facilitates platelet aggregation by binding to fibrinogen and forming a stable platelet thrombus.

In the early stages of thrombus formation, platelets predominate (“white” thrombi); further organization with fibrin results in older “red” thrombi. The stages of thrombi vary in non-ST-elevation and ST-elevation myocardial infarction and are prognostic markers of death.^4–8

PERCUTANEOUS INTERVENTION, RESTENOSIS, AND STENT THROMBOSIS

Percutaneous coronary intervention, the preferred means of revascularization for many patients, is performed emergently in patients with ST-elevation myocardial infarction, urgently in those with acute coronary syndromes without ST elevation, and electively in those with stable ischemic symptoms.

Percutaneous revascularization techniques have evolved from balloon angioplasty to bare-metal stents to drug-eluting stents, but each of these procedures has been associated with a periprocedural and postprocedural risk of thrombosis.

Balloon angioplasty was associated with vascular intimal injury, inciting elastic vascular recoil and smooth muscle cell proliferation leading to restenosis.

Bare-metal stents reduced the restenosis rate by eliminating vascular recoil, although restenosis still occurred within the stent because of neointimal proliferation of vascular smooth muscle cells. This was an important limitation, as both acute and subacute stent thrombosis were refractory to aggressive anticoagulation regimens that were associated with major bleeding complications and longer hospital length of stay. Stenting became mainstream practice only after the ISAR⁹ and STARS¹⁰ trials showed that dual antiplatelet therapy controlled stent thrombosis.

Drug-eluting stents coated with anti-proliferative and anti-inflammatory polymers markedly reduced in-stent restenosis rates by suppressing the initial vascular smooth-muscle proliferative response. However, they were still associated with late and very late stent thrombosis with incomplete endothelialization, even up to 40 months after implantation. Proposed mechanisms include incomplete stent apposition and inflammatory hypersensitivity reactions to the polymer coating. Incomplete stent apposition associated with low-velocity blood flow at the junction of the stent strut and vessel wall, together with delayed endothelialization, promotes platelet adhesion and aggregation, followed by thrombus formation.¹¹

Second-generation drug-eluting stents have thinner struts and more biocompatible polymers and are thought to favor more complete re-endothelialization, reducing the rates of stent thrombosis.^8,12,13

Predictors of early stent thrombosis

The Dutch Stent Thrombosis Registry and other studies looked at risk factors for stent thrombosis.^14,15

Procedure-related factors included:

• Stent undersizing
• Residual uncovered dissections after angioplasty
• Longer stents
• Low flow after angioplasty (< 3 on the 0–3 Thrombolysis in Myocardial Infarction [TIMI] scale).

Lesion-related factors included:

• Intermediate coronary artery disease both proximal and distal to the culprit lesions
• Bifurcation lesions.

Patient-related factors included:

• Low left ventricular ejection fraction
• Diabetes mellitus
• Peripheral arterial disease
Premature discontinuation of clopidogrel.

ANTIPLATELET AGENTS: MECHANISM OF ACTION

Various pathways play synergistic roles in platelet activation and aggregation and thrombus formation, and different antiplatelet agents inhibit these specific pathways, thus complementing each other and having additive effects (Figure 2, Table 1).^5,16–21

Aspirin inhibits cyclo-oxygenase 1

Cyclo-oxygenase 1, found in platelets, endothelial cells, and other cells, catalyzes the conversion of arachidonic acid to thromboxane A₂. Aspirin irreversibly inhibits cyclo-oxygenase 1 by acetylating its serine residue, preventing formation of thromboxane A₂ and preventing platelet activation and aggregation.

P2Y₁₂ ADP receptor antagonists

Clopidogrel and prasugrel are thienopyridine agents that irreversibly inhibit the P2Y₁₂ receptor, thereby preventing binding of adenosine diphosphate and the subsequent platelet activation-aggregation cascade. They are both prodrugs and require conversion by cytochrome P450 enzymes to active metabolites. Prasugrel is 10 times more potent than clopidogrel due to more efficient formation of its active metabolite, and it achieves a comparable effect on platelet inhibition 30 minutes faster than the peak effect of clopidogrel at 6 hours. The overall peak inhibitory effect of prasugrel is twice that of clopidogrel.²²

Ticagrelor, a cyclopentyl-triazolo-pyrimidine, directly and reversibly inhibits the P2Y₁₂ ADP receptor. Unlike clopidogrel and prasugrel, it does not need to be converted to an active metabolite, and it noncompetitively inhibits P2Y₁₂ at a site different from the adenosine diphosphate binding site.²³ Like prasugrel, ticagrelor inhibits platelet function more rapidly and more completely than clopidogrel.

Cangrelor, an intravenously administered analogue of adenosine triphosphate, reversibly inhibits the P2Y₁₂ receptor. It has undergone phase 3 trials but is not yet approved for clinical use.²⁴

WHY DUAL ANTIPLATELET THERAPY?

Aspirin is good, clopidogrel is better

Aspirin has a well-validated role in both primary and secondary prevention of coronary and noncoronary atherosclerotic vascular disease.

The CAPRIE trial found clopidogrel monotherapy to be superior to aspirin monotherapy in patients with established atherosclerotic vascular disease.²⁵

After stenting, short-term dual therapy is better than short-term warfarin

Thrombotic complications in the early postprocedural period were a major limitation of stenting, and existing anticoagulation regimens were ineffective in preventing them.^26,27

The ISAR trial studied the benefit of combined antiplatelet vs anticoagulant therapy after stent placement. Patients randomized to receive combined aspirin plus ticlopidine (an early P2Y₁₂ inhibitor) had significantly lower rates of primary cardiac, hemorrhagic, and vascular events at 30 days.⁹ Two other trials confirmed this finding.^28,29

STARS¹⁰ also confirmed the benefit of aspirin and ticlopidine after stenting. Patients were randomly assigned to aspirin alone, aspirin plus warfarin, or aspirin plus ticlopidine after stent placement. The rate of stent thrombosis at 30 days was significantly lower in the dual antiplatelet group than in the other two groups. The dual antiplatelet group had a higher rate of bleeding than the aspirin-alone group, but the rate was similar to that of the aspirin-plus-warfarin group.

Long-term dual antiplatelet therapy is beneficial in several situations

ISAR and STARS were landmark trials that showed stent thrombosis could be reduced by dual antiplatelet therapy for a 30-day period. However, the long-term role of dual antiplatelet therapy was still unknown.

The CURE trial^30–32 randomized patients presenting with acute coronary syndromes without ST elevation to receive clopidogrel plus aspirin or placebo plus aspirin for 3 to 12 months. The rate of the primary end point (cardiac death, nonfatal myocardial infarction, or stroke) was significantly lower in the clopidogrel-plus-aspirin group. A similar benefit of dual antiplatelet therapy was seen in the subgroup of patients who underwent percutaneous coronary intervention. Both pretreatment with clopidogrel plus aspirin for a median of 10 days prior to percutaneous intervention and continuing it for a mean of 9 months reduced major adverse cardiovascular events.

The CREDO trial²⁰ found that the combination of clopidogrel and aspirin significantly reduced the incidence of death, myocardial infarction, or stroke at 1 year after percutaneous coronary intervention. A subgroup of patients in this trial who had a longer pretreatment interval with a loading clopidogrel dose showed a benefit at 28 days, which was not as evident with a shorter loading dose interval.

The CLARITY-TIMI 28 trial^33,34 showed the advantage of adding clopidogrel to aspirin in patients receiving fibrinolytic therapy for ST-elevation myocardial infarction. Adding clopidogrel both improved the patency of the infarct-related artery and reduced ischemic complications. In patients who subsequently underwent percutaneous coronary intervention and stenting, clopidogrel pretreatment was associated with a significant decrease in ischemic complications before and after the procedure. There was no significant increase in bleeding complications in either group.

COMMIT/CCS 2³⁵ also showed the benefit of dual antiplatelet therapy in patients with ST-elevation myocardial infarction. Clopidogrel added to aspirin during the short-term in-hospital or postdischarge treatment period significantly reduced a composite end point of reinfarction, death, or stroke as well as death from any cause.

The CHARISMA trial^36–38 aimed to determine if patients who were more stable (ie, no recent acute coronary syndrome event or percutaneous coronary intervention) would benefit. Overall, CHARISMA showed no benefit of adding clopidogrel to aspirin compared with aspirin alone in a broad population of patients with established vascular disease (secondary prevention) or risk factors for vascular disease (primary prevention).

But importantly, though no benefit was seen in the primary prevention group, the large subgroup of patients with established atherosclerotic vascular disease (12,153 of the 15,603 patients in the trial) did benefit from dual antiplatelet therapy.^36,37 This subgroup showed an overall reduction in absolute risk of 1.5% (relative risk 0.88, P = .046) over a median follow-up of 27.6 months. This benefit was even more apparent in the 9,478 patients with prior myocardial infarction, stroke, or peripheral artery disease, for whom the relative risk reduction was 17.1% (P = .01) and the reduction in absolute risk 1.5%.³⁸

These results are comparable to the 2% absolute risk reduction in the CURE trial for similar end points over 9 months. In both studies, there was no significant increase in the risk of major bleeding or intracranial bleeding in the clopidogrel-plus-aspirin groups, although minor bleeding was increased by dual antiplatelet therapy.

The rate of severe bleeding, which was the primary safety end point in CHARISMA, was not significantly different in the clopidogrel-plus-aspirin group compared with the placebo-plus-aspirin group (relative risk 1.25, 95% CI 0.97–1.61, P = .09).

Thus, although the CHARISMA findings were negative overall, the positive finding observed in the predominant subgroup of patients with established vascular disease can therefore be considered supportive of the results of the subsequent trials discussed below.

The PEGASUS-TIMI 54 trial³⁹ studied the benefit of adding ticagrelor (60 or 90 mg) to low-dose aspirin in patients with stable coronary artery disease who had had a myocardial infarction 1 to 3 years earlier.

Confirming the results of the CHARISMA subgroup analysis, the incidence of the ischemic primary efficacy end point (a composite of cardiovascular death, myocardial infarction, and stroke) was significantly lower in both groups receiving ticagrelor plus aspirin compared with those receiving placebo plus aspirin. The Kaplan-Meier rate at 3 years for the ticagrelor 90 mg-plus-aspirin group was 7.85% vs 9.04% for the placebo-plus-aspirin group (hazard ratio 0.85, 95% confidence interval [CI] 0.75–0.96, P = .008). The rate for the ticagrelor 60 mg-plus-aspirin group was 7.77% vs 9.04% for the placebo-plus-aspirin group (hazard ratio 0.84, 95% CI 0.74–0.95, P = .004).

The rates of all TIMI major and minor bleeding, as well as bleeding requiring transfusion or discontinuation of the study drug, were significantly higher in both ticagrelor dosing groups than in the placebo group (P < .01 for both groups vs placebo). The rates of fatal bleeding and nonfatal intracranial hemorrhage were not significantly higher. Although there was an overall reduction in ischemic end points with the addition of ticagrelor, there was also a significantly higher incidence of bleeding in this group.

Comment. Thus, with or without percutaneous coronary intervention in acute coronary syndrome as well as in stable coronary artery disease, dual antiplatelet therapy was shown to improve outcomes and decrease ischemic complications compared with aspirin alone. It provided benefit in the setting of acute coronary syndrome (in the CURE trial) and percutaneous coronary intervention (in the CREDO trial) for up to 1 year.

Major questions remained to be addressed:

• Do the results of CREDO, which was performed before the current interventional era and the use of drug-eluting stents, reflect outcomes after current interventional practice?
• Could shorter periods of dual antiplatelet therapy be sufficient, especially with newer stents with less risk of late thrombosis?
• Does the benefit of dual antiplatelet therapy extend beyond the 1-year time period tested in those trials to date?

RECOMMENDATIONS FOR DOSING

The American College of Cardiology Foundation/American Heart Association guidelines for dosing of antiplatelet agents for non-ST-elevation myocardial infarction are summarized in Table 2, and those for ST-elevation myocardial infarction are summarized in Table 3.^1,2

WOULD SHORTER THERAPY AFTER STENTING WORK AS WELL?

The American College of Cardiology Foundation/American Heart Association currently recommend dual antiplatelet therapy for at least 12 months after drug-eluting stent placement, with shorter courses appropriate for patients who develop excessive bleeding complications or who are at high risk of bleeding.

Four trials (Table 4) evaluated whether shorter durations of dual antiplatelet therapy would suffice: SECURITY,⁴⁰ EXCELLENT,⁴¹ OPTIMIZE,⁴² and RESET.⁴³ All of them showed that short-duration therapy was not inferior to standard-duration therapy.⁴⁴ These studies were comparable in that:

• Patients were randomized at the time of percutaneous coronary intervention or within 24 hours of it.
• Most patients received a second-generation drug-eluting stent, with the following exceptions: in EXCELLENT,⁴¹ one-fourth of patients received a Cypher first-generation drug-eluting stent, and in RESET,⁴³ approximately one-fourth of the patients received a sirolimus-eluting stent in the standard-duration group for short lesions. Those patients with longer lesions in the RESET standard-duration group received an evero­limus drug-eluting stent.
• The second antiplatelet added to aspirin in all studies was clopidogrel, with the exception of the SECURITY trial, in which fewer than 2% of patients received ticagrelor or prasugrel.⁴⁰
• All the trials except RESET excluded patients who had had a myocardial infarction within 72 hours, and thus most patients studied had a lower risk profile.
• All of the trials sought to study noninferiority of short- vs standard-duration dual antiplatelet therapy, defined as the occurrence of a primary end point at 1 year (a composite of cardiovascular death, myocardial infarction, stroke, stent thrombosis, target vessel failure or revascularization, or bleeding).

Their low-risk patient populations and infrequent end points rendered these studies underpowered to make definitive conclusions about the relative efficacy of 6-months vs 12-months of dual antiplatelet therapy.

WOULD LONGER THERAPY BE BETTER?

The PRODIGY trial⁴⁵ assessed durations of dual antiplatelet therapy both shorter and longer than the conventional 1 year, randomizing patients undergoing placement of a bare-metal stent, first-generation drug-eluting stent, or second-generation drug-eluting stent to receive aspirin and clopidogrel for either 6 months or 24 months. The study showed no significant difference in primary outcomes in the short- or long-duration groups.

Other trials that compared the standard 12 months of dual antiplatelet therapy with extended duration beyond 12 months were DAPT,³ ARCTIC-Interruption,⁴⁶ and DES-LATE.⁴⁷ The trials were comparable in that:

• All patients were randomized after completing 12 months of dual antiplatelet therapy following drug-eluting stent placement.
• All patients who were included had been free of major cardiac ischemic events or bleeding during the 12 months following stent placement.
• The primary aim of all three studies was to compare primary end points in groups receiving aspirin alone vs extended dual antiplatelet therapy. The primary end point was a composite of death due to a cardiovascular cause, nonfatal myocardial infarction, stroke, or stent thrombosis.
• The principal safety end point was bleeding.

Although the two earlier studies (ARCTIC-Interruption and DES-LATE) did not show any benefit of extended dual antiplatelet therapy compared with the standard 12-month duration, the recent DAPT study did.

The DAPT study

The DAPT study³ was an international, multicenter, placebo-controlled, double-blind randomized trial designed to examine the benefit of dual antiplatelet therapy beyond 1 year in a patient population large enough to provide definitive assessment of benefit and risk.

A total of 9,961 patients who received drug-eluting stents were randomized after 12 months of dual antiplatelet therapy to receive either a thienopyridine (clopidogrel or prasugrel) plus aspirin or placebo plus aspirin. They were followed for an additional 18 months. The coprimary efficacy end points were stent thrombosis and a composite of death, myocardial infarction, or stroke, while the primary safety end point was moderate or severe bleeding. The patients were also observed from months 30 to 33 on aspirin alone after stopping the thienopyridine.

Results. Longer therapy substantially reduced the risks of stent thrombosis (hazard ratio [HR] 0.29, 95% confidence interval [CI] 0.17–0.48) and the composite ischemic end point (HR 0.71, 95% CI 0.59–0.85). Follow-up during the 3-month thienopyridine discontinuation phase starting at 30 months revealed convergence of the ischemic event-rate curves in the two groups, which suggested that continuing dual antiplatelet therapy beyond 30 months might have been beneficial. Myocardial infarction unrelated to stent thrombosis accounted for 55% of the treatment benefit of dual antiplatelet therapy.

The risk of bleeding was higher in the thienopyridine group during the treatment period (2.5% vs 1.6%, P = .001). There was also a higher rate of noncardiovascular mortality in the thienopyridine group, although this difference may have been due to chance.^3,48

Why were the results different?

All three trials included first- and second-generation drug-eluting stents, with different proportions in different trials. In ARCTIC-Interruption,⁴⁶ 43% of the patients in the continuation group had a first-generation stent, as did 64% of the patients in the dual antiplatelet group of DES-LATE.⁴⁷ In the DAPT trial,³ 38% of the patients in the longer-duration arm had a first-generation stent, and in 26% of cases it was a paclitaxel-eluting stent.

Only clopidogrel was used as the second antiplatelet agent in DES-LATE, whereas prasugrel was used in 10% of patients in ARCTIC-Interruption and 35% in DAPT.

Yet none of these differences seem to explain the differences in outcome among the studies. ARCTIC-Interruption and DES-LATE did not show any benefit of continued dual antiplatelet therapy beyond 12 months. DAPT showed benefit of extended therapy with prasugrel or with clopidogrel, and with first-generation or second-generation drug-eluting stents. The most likely explanation for the different results was that DAPT was the only trial sufficiently powered to definitively assess the end points, including stent thrombosis.

A balance between ischemic efficacy and bleeding risk is the major consideration with any antithrombotic and antiplatelet therapy. In the three largest trials we discussed (the vascular disease subgroups of CHARISMA,³⁸ PEGASUS,³⁹ and DAPT³), comparison of the prespecified efficacy and safety end points of each trial suggests that dual antiplatelet therapy has a net benefit, particularly given the irreversible nature of ischemic end points.

In CHARISMA,³⁸ 60 cardiovascular deaths, myocardial infarctions, or strokes were prevented per year per 10,000 patients treated, at the cost of 28 excess moderate bleeding events.

In PEGASUS,³⁹ 42 cardiovascular deaths, myocardial infarctions, or strokes were prevented, at the cost of 79 excess bleeding events requiring transfusion.

In DAPT (a selected population who had tolerated dual antiplatelet therapy for 1 year), 106 deaths, myocardial infarctions, or stroke events were prevented, at the cost of 47 excess moderate bleeding events^.3

Indirect comparisons between trials are problematic, given different end point definitions, populations, and background therapies. But their results suggest that less-intensive inhibition with clopidogrel as the second antiplatelet long-term (as in CHARISMA) may provide the best balance of benefit vs risk.

BALANCING RISK AND BENEFIT

The evidence is unequivocal that dual antiplatelet therapy suppresses coronary ischemic complications resulting from thrombosis at sites of spontaneous plaque rupture following acute coronary syndromes or mechanical plaque disruption and foreign body implantation associated with percutaneous coronary intervention.

Three large-scale trials (DAPT,³ PEGASUS,³⁹ and the secondary prevention subgroup of CHARISMA³⁸) showed that the protective effect of dual antiplatelet therapy continues with prolonged therapy in patients who have experienced an acute coronary syndrome event or have received a drug-eluting stent. That benefit seems to be due to the action of these therapies on the culprit vessel (the one that caused the acute coronary syndrome or the site of stenting), as well as nonculprit arteries, emphasizing that dual antiplatelet therapy protects against atherosclerosis progression and future plaque rupture events.

For the durations studied in the longest trials thus far, 30 months (DAPT³) and 36 months (PEGASUS³⁹), event curves continue to diverge, indicating that the advantage of dual antiplatelet therapy may persist for an indefinite period of time. Thus, indefinite therapy with dual antiplatelet agents can be supported, particularly in patients with advanced coronary artery disease or those who have had multiple coronary events.

We believe that the balance of evidence suggests that smaller studies that failed to show a benefit of longer-term therapy were underpowered to do so.

The ischemic protection is associated with the adverse effect of increased bleeding risk. Unfortunately, there has been little success in guiding dual antiplatelet therapy based on ischemic vs bleeding risk, in part because the same factors that predict risk of ischemic complications seem to predict increased susceptibility to bleeding. Nevertheless, indirect comparisons between studies suggest that for longer-term therapy clopidogrel may be superior to ticagrelor or prasugrel: the absolute excess bleeding risk with dual antiplatelet therapy vs aspirin in the CHARISMA secondary prevention subgroup was less than that in PEGASUS, with similar absolute reductions in ischemic events. So while the TRITON-TIMI 38²² and PLATO²³ trials support the superiority of prasugrel or ticagrelor over clopidogrel for the first year after acute coronary syndrome, subsequent years of therapy may best be provided with clopidogrel.

Some patients may have identifiable factors that place them at very high risk of bleeding—need for surgical procedures, need for anticoagulation, or occurrence of bleeding complications or excessive “nuisance bleeding.” In those patients, the data suggest that dual antiplatelet therapy could be discontinued after 6 months, or perhaps even 3 months in the highest bleeding risk circumstances after second-generation drug-eluting stent placement.

WOEST⁴⁹ was an open-label randomized controlled trial that studied the safety of antiplatelet regimens in patients on anticoagulation requiring percutaneous coronary interventions. Patients were randomized to double therapy with anticoagulant and clopidogrel vs triple therapy with additional aspirin following percutaneous coronary intervention. The primary end point was bleeding events within 1 year. Clopidogrel without aspirin was associated with significantly fewer bleeding events compared with triple therapy, with no increase in adverse ischemic events. The strategy tested in the WOEST trial seems reasonable in the specific group of patients who require ongoing anticoagulant therapy after drug-eluting stent placement, recognizing that the trial was somewhat underpowered to make definitive conclusions, particularly in patients at high risk for stent thrombosis.

Based on the results of PEGASUS and the CHARISMA subgroup with established ischemic burden, in which dual antiplatelet therapy was started after an interruption following the index coronary event, it is also reasonable to restart long-term dual antiplatelet therapy in patients who require interruption for short-term indications such as a surgical procedure.