Commentary

The first part of our discussion focuses on the issue of guidelines and guideline proliferation. Over the last several years, multiple professional groups and health-associated agencies have issued guidelines for the diagnosis and treatment of hypertension. It is not clear to me that these competing documents provide a rational platform for improved care. Despite some recent claims, no guideline can be fully “evidence-based” because we cannot afford clinical trials for each relevant clinical question. Furthermore, the current batch of new guidelines has yielded conflicting recommendations. For example, the definition of “elderly” as age 60 in the Eighth Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 8) and age 80 in the recent joint guidance from the International and American Societies of Hypertension. I fear that the guideline process has essentially broken down—this benefits no one.

First, a little history. Prior JNC documents were produced under the auspices of the National High Blood Pressure Education Program, a consortium of about 3 dozen healthcare professional organizations. As a participant in JNC VI and JNC 7, I can assure you that all participants were dedicated scientists and clinicians whose common goal was to synthesize the best and most robust information into a document that provided practitioners with the most reliable guidance possible. JNC 7 was released in 2003 and was soon widely adopted; it truly stood the test of time. After that, everything changed. Dr. Basile, what has been happening since?

DR. BASILE: Thanks, Dr. Izzo. I agree with you about your concerns with guidelines. This is a very difficult issue for the practicing clinician.

With regard to JNC 8, in 2008, the National Heart, Lung, and Blood Institute (NHLBI) decided to develop its own comprehensive “evidence-based” disease prevention portfolio that included a coordinated cholesterol guideline, obesity guideline, hypertension guideline, as well as a lifestyle/risk factor modification guideline. Following the recommendations put forth by the Institute of Medicine on guidelines, the NHLBI mandated that strict rules of evidence be used for its guideline development. After the JNC 8 committee established the clinical trial criteria for use in their hypertension guideline, the NHLBI hired a professional vendor to vet the literature. The process was essentially going nowhere so a second vendor for hypertension had to be hired. In the end, few clinical trials passed the criteria used to form the basis of JNC 8. And near its completion, after the JNC 8 group toiled for over 5 years putting their report together, they learned in 2013 that the NHLBI withdrew its support for the guideline and asked the American College of Cardiology/American Heart Association to establish a joint task force (to include the JNC 8 group) to assume responsibility for the guideline process. Declining this collaborative opportunity, the JNC 8 group decided on their own to have their work submitted and it was published without organizational support in JAMA.¹

Meanwhile, the Centers for Disease Control published an overall systems approach for primary care clinicians in which they released the Kaiser Permanente algorithm for effective blood pressure (BP) control.²

Of note, there are several other competing guidelines: one which is issued annually by the Canadian Hypertension Education Program. This approach is now extremely well developed and comprehensive with multiple web-based programs and information sources (www.hypertension.ca/en/chep). In addition, there is the recent effort by the European Society of Hypertension/European Society of Cardiology in 2013,³ and a newer, less formal document from the International Society of Hypertension and supported by a working group comprised of members of the American Society of Hypertension.⁴ Finally, there are other national guidelines around the world, perhaps most prominent among these being the British Hypertension Society National Institute for Clinical Excellence or the NICE guidelines.

DR. IZZO: That’s a good summary, Dr. Basile. What constitutes a useful guideline? Who is the audience? I don’t believe that a good guideline should be hidebound by unduly restrictive ‘evidence rules’ that ignore all but a tiny fraction of the available scientific information. A good guideline must synthesize complex data in a clear and concise way and must address the kinds of questions that real doctors need answered every day. Dr. Gradman, do you have an opinion on guidelines?

DR. GRADMAN: Part of what is going on in the revised hypertension guidelines is a worldwide shift from the treatment of individual patients to the treatment of patient populations. Such guidance for treating large groups is of great interest to governments and insurance companies but is less useful for making treatment decisions for individual patients with individual problems.

In fact, the JNC 8 guidelines specifically say that they are aimed at primary care physicians, not specialists or people who have a lot of experience with the treatment of hypertension. I think JNC 8 is to be commended for trying to make the fundamental concepts of the guidelines completely evidence-based. The guidelines delineate general targets and thresholds for treatment and also suggest the best drugs to use for most patients. Unfortunately, the exercise showed that it is not really possible to answer a lot of specific treatment questions using a strict evidence-based approach. For these we must rely on less definitive evidence and expert opinion.

I think there are different audiences; a good general guideline today should be aimed at primary care physicians but I think specialists need their own guidelines. As a cardiologist, I would want a very different set of guidelines than is presented in JNC 8.

DR. IZZO: Dr. Moser, given your long association with the highly successful parent of the JNC process, the National High Blood Pressure Education Program, I think your perspective is invaluable. Do you think any of the current guidelines hit the mark or what do you think should have been done differently?

DR. MOSER: Well, that’s a broad question. JNC 8 doesn’t differ much from JNC 7 except for one or 2 items. JNC 8 set out to answer several questions with a strict evidence-based methodology, and when they were finished, after a long period of time and substantial expenditure of money, they only ended up addressing 3 questions. The first question was when to initiate treatment, the second was related to the target or goal BPs, and the third was how to achieve goal pressures. They did address the question of whether people with comorbidities, especially diabetes and kidney disease, needed to have lower goal BPs.

On the issue of targets for BP, it was decided that 140/90 mm Hg was good enough for most people and that there was no compelling evidence that BP should be lowered below 130/80 mm Hg in diabetics or people with kidney disease. But we all know that in practice, you can’t adjust BPs to exact numbers. JNC 8 also said that the threshold for a diagnosis and target BP in people over the age of 60 was 150/90 mm Hg. This is quite different from other guidelines.

The guidelines also point out a difference in black and white patients. We’ve known for years that calcium-channel blockers (CCBs) and thiazide-type diuretics are more effective in blacks and angiotensin-converting enzyme (ACE) inhibitors or beta blockers are more effective in whites, so I don’t think JNC 8 provided any new information on this. Otherwise, you’ll find very little difference between JNC 8 and 7. What they didn’t address were possible changes in the diagnostic evaluation or nonpharmacologic therapy and they failed to discuss issues related to the delivery of care and adherence, etc. Some physicians have said that the changes are important and different, but when you look at it very carefully they’re not substantive compared to other reports.

Dr. IZZO: Is that a result of their strict rules of evidence?

Dr. MOSER: The report was supposed to be all evidence-based. They hired an outside vendor that reviewed about 6100 English-language papers. The experts on that panel could have picked up the 40 papers that they finally rated as acceptable within a few days. It took the vendors several years. As everyone has mentioned, they used the data from evidence-based trials and reached conclusions, but they also had an “out” on all of them, including the 150 mm Hg BP target in the elderly, which was arbitrarily defined as age 60. As in their other recommendations, they then said that you could rely on “expert opinion,” meaning clinical judgment. This is the basis of the major recommendation: “If you treat below 140/90 mm Hg and everybody’s happy and there are no side effects, then just continue treatment.”

Regarding their approach to the treatment of resistant patients, the proposed algorithm is almost the same as it has been for many, many years. So I’m very disappointed in what they have accomplished, especially since it took such a long period of time and cost so much money.

DR. IZZO: I echo your disappointment, especially on very fundamental grounds. First, a guideline constrained by the suffocating rules applied by the JNC 8 group is enormously biased, especially when over 99% of the potentially relevant information was ignored because it did not come from a randomized trial. Did they not understand just how biased randomized clinical trials really are? A true expert does better than that, prioritizing the value and relevance of all evidence, from bench science to randomized trials. The real test for a guideline is then in the projection of that synthesis, that is translating expert interpretation into understandable and workable recommendations. One other issue: absence of evidence should not imply a negative recommendation. If there is no definitive study proving that a lower BP target is warranted for people with kidney disease, it should also be very clear that there is existing evidence that lowering BP to 130/80 mm Hg instead of 140/90 mm Hg may be a good idea. JNC 8 chose to ignore studies of achieved (not intention-to-treat) BP values that identified better renal outcomes in diabetics⁵ and fewer recurrent strokes⁶ in trial participants with the lowest BPs. While it is true that such benefits are not achieved by all trial participants, it should be recognized that “lower is better” is probably true in at least some people.

Are there any comments or concerns that the narrow definition of evidence can actually lead us astray?

The first part of our discussion focuses on the issue of guidelines and guideline proliferation. Over the last several years, multiple professional groups and health-associated agencies have issued guidelines for the diagnosis and treatment of hypertension. It is not clear to me that these competing documents provide a rational platform for improved care. Despite some recent claims, no guideline can be fully “evidence-based” because we cannot afford clinical trials for each relevant clinical question. Furthermore, the current batch of new guidelines has yielded conflicting recommendations. For example, the definition of “elderly” as age 60 in the Eighth Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 8) and age 80 in the recent joint guidance from the International and American Societies of Hypertension. I fear that the guideline process has essentially broken down—this benefits no one.

First, a little history. Prior JNC documents were produced under the auspices of the National High Blood Pressure Education Program, a consortium of about 3 dozen healthcare professional organizations. As a participant in JNC VI and JNC 7, I can assure you that all participants were dedicated scientists and clinicians whose common goal was to synthesize the best and most robust information into a document that provided practitioners with the most reliable guidance possible. JNC 7 was released in 2003 and was soon widely adopted; it truly stood the test of time. After that, everything changed. Dr. Basile, what has been happening since?

DR. BASILE: Thanks, Dr. Izzo. I agree with you about your concerns with guidelines. This is a very difficult issue for the practicing clinician.

With regard to JNC 8, in 2008, the National Heart, Lung, and Blood Institute (NHLBI) decided to develop its own comprehensive “evidence-based” disease prevention portfolio that included a coordinated cholesterol guideline, obesity guideline, hypertension guideline, as well as a lifestyle/risk factor modification guideline. Following the recommendations put forth by the Institute of Medicine on guidelines, the NHLBI mandated that strict rules of evidence be used for its guideline development. After the JNC 8 committee established the clinical trial criteria for use in their hypertension guideline, the NHLBI hired a professional vendor to vet the literature. The process was essentially going nowhere so a second vendor for hypertension had to be hired. In the end, few clinical trials passed the criteria used to form the basis of JNC 8. And near its completion, after the JNC 8 group toiled for over 5 years putting their report together, they learned in 2013 that the NHLBI withdrew its support for the guideline and asked the American College of Cardiology/American Heart Association to establish a joint task force (to include the JNC 8 group) to assume responsibility for the guideline process. Declining this collaborative opportunity, the JNC 8 group decided on their own to have their work submitted and it was published without organizational support in JAMA.¹

Meanwhile, the Centers for Disease Control published an overall systems approach for primary care clinicians in which they released the Kaiser Permanente algorithm for effective blood pressure (BP) control.²

Of note, there are several other competing guidelines: one which is issued annually by the Canadian Hypertension Education Program. This approach is now extremely well developed and comprehensive with multiple web-based programs and information sources (www.hypertension.ca/en/chep). In addition, there is the recent effort by the European Society of Hypertension/European Society of Cardiology in 2013,³ and a newer, less formal document from the International Society of Hypertension and supported by a working group comprised of members of the American Society of Hypertension.⁴ Finally, there are other national guidelines around the world, perhaps most prominent among these being the British Hypertension Society National Institute for Clinical Excellence or the NICE guidelines.

DR. IZZO: That’s a good summary, Dr. Basile. What constitutes a useful guideline? Who is the audience? I don’t believe that a good guideline should be hidebound by unduly restrictive ‘evidence rules’ that ignore all but a tiny fraction of the available scientific information. A good guideline must synthesize complex data in a clear and concise way and must address the kinds of questions that real doctors need answered every day. Dr. Gradman, do you have an opinion on guidelines?

DR. GRADMAN: Part of what is going on in the revised hypertension guidelines is a worldwide shift from the treatment of individual patients to the treatment of patient populations. Such guidance for treating large groups is of great interest to governments and insurance companies but is less useful for making treatment decisions for individual patients with individual problems.

In fact, the JNC 8 guidelines specifically say that they are aimed at primary care physicians, not specialists or people who have a lot of experience with the treatment of hypertension. I think JNC 8 is to be commended for trying to make the fundamental concepts of the guidelines completely evidence-based. The guidelines delineate general targets and thresholds for treatment and also suggest the best drugs to use for most patients. Unfortunately, the exercise showed that it is not really possible to answer a lot of specific treatment questions using a strict evidence-based approach. For these we must rely on less definitive evidence and expert opinion.

I think there are different audiences; a good general guideline today should be aimed at primary care physicians but I think specialists need their own guidelines. As a cardiologist, I would want a very different set of guidelines than is presented in JNC 8.

DR. IZZO: Dr. Moser, given your long association with the highly successful parent of the JNC process, the National High Blood Pressure Education Program, I think your perspective is invaluable. Do you think any of the current guidelines hit the mark or what do you think should have been done differently?

DR. MOSER: Well, that’s a broad question. JNC 8 doesn’t differ much from JNC 7 except for one or 2 items. JNC 8 set out to answer several questions with a strict evidence-based methodology, and when they were finished, after a long period of time and substantial expenditure of money, they only ended up addressing 3 questions. The first question was when to initiate treatment, the second was related to the target or goal BPs, and the third was how to achieve goal pressures. They did address the question of whether people with comorbidities, especially diabetes and kidney disease, needed to have lower goal BPs.

On the issue of targets for BP, it was decided that 140/90 mm Hg was good enough for most people and that there was no compelling evidence that BP should be lowered below 130/80 mm Hg in diabetics or people with kidney disease. But we all know that in practice, you can’t adjust BPs to exact numbers. JNC 8 also said that the threshold for a diagnosis and target BP in people over the age of 60 was 150/90 mm Hg. This is quite different from other guidelines.

The guidelines also point out a difference in black and white patients. We’ve known for years that calcium-channel blockers (CCBs) and thiazide-type diuretics are more effective in blacks and angiotensin-converting enzyme (ACE) inhibitors or beta blockers are more effective in whites, so I don’t think JNC 8 provided any new information on this. Otherwise, you’ll find very little difference between JNC 8 and 7. What they didn’t address were possible changes in the diagnostic evaluation or nonpharmacologic therapy and they failed to discuss issues related to the delivery of care and adherence, etc. Some physicians have said that the changes are important and different, but when you look at it very carefully they’re not substantive compared to other reports.

Dr. IZZO: Is that a result of their strict rules of evidence?

Dr. MOSER: The report was supposed to be all evidence-based. They hired an outside vendor that reviewed about 6100 English-language papers. The experts on that panel could have picked up the 40 papers that they finally rated as acceptable within a few days. It took the vendors several years. As everyone has mentioned, they used the data from evidence-based trials and reached conclusions, but they also had an “out” on all of them, including the 150 mm Hg BP target in the elderly, which was arbitrarily defined as age 60. As in their other recommendations, they then said that you could rely on “expert opinion,” meaning clinical judgment. This is the basis of the major recommendation: “If you treat below 140/90 mm Hg and everybody’s happy and there are no side effects, then just continue treatment.”

Regarding their approach to the treatment of resistant patients, the proposed algorithm is almost the same as it has been for many, many years. So I’m very disappointed in what they have accomplished, especially since it took such a long period of time and cost so much money.

DR. IZZO: I echo your disappointment, especially on very fundamental grounds. First, a guideline constrained by the suffocating rules applied by the JNC 8 group is enormously biased, especially when over 99% of the potentially relevant information was ignored because it did not come from a randomized trial. Did they not understand just how biased randomized clinical trials really are? A true expert does better than that, prioritizing the value and relevance of all evidence, from bench science to randomized trials. The real test for a guideline is then in the projection of that synthesis, that is translating expert interpretation into understandable and workable recommendations. One other issue: absence of evidence should not imply a negative recommendation. If there is no definitive study proving that a lower BP target is warranted for people with kidney disease, it should also be very clear that there is existing evidence that lowering BP to 130/80 mm Hg instead of 140/90 mm Hg may be a good idea. JNC 8 chose to ignore studies of achieved (not intention-to-treat) BP values that identified better renal outcomes in diabetics⁵ and fewer recurrent strokes⁶ in trial participants with the lowest BPs. While it is true that such benefits are not achieved by all trial participants, it should be recognized that “lower is better” is probably true in at least some people.

Are there any comments or concerns that the narrow definition of evidence can actually lead us astray?

The first part of our discussion focuses on the issue of guidelines and guideline proliferation. Over the last several years, multiple professional groups and health-associated agencies have issued guidelines for the diagnosis and treatment of hypertension. It is not clear to me that these competing documents provide a rational platform for improved care. Despite some recent claims, no guideline can be fully “evidence-based” because we cannot afford clinical trials for each relevant clinical question. Furthermore, the current batch of new guidelines has yielded conflicting recommendations. For example, the definition of “elderly” as age 60 in the Eighth Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 8) and age 80 in the recent joint guidance from the International and American Societies of Hypertension. I fear that the guideline process has essentially broken down—this benefits no one.

First, a little history. Prior JNC documents were produced under the auspices of the National High Blood Pressure Education Program, a consortium of about 3 dozen healthcare professional organizations. As a participant in JNC VI and JNC 7, I can assure you that all participants were dedicated scientists and clinicians whose common goal was to synthesize the best and most robust information into a document that provided practitioners with the most reliable guidance possible. JNC 7 was released in 2003 and was soon widely adopted; it truly stood the test of time. After that, everything changed. Dr. Basile, what has been happening since?

DR. BASILE: Thanks, Dr. Izzo. I agree with you about your concerns with guidelines. This is a very difficult issue for the practicing clinician.

With regard to JNC 8, in 2008, the National Heart, Lung, and Blood Institute (NHLBI) decided to develop its own comprehensive “evidence-based” disease prevention portfolio that included a coordinated cholesterol guideline, obesity guideline, hypertension guideline, as well as a lifestyle/risk factor modification guideline. Following the recommendations put forth by the Institute of Medicine on guidelines, the NHLBI mandated that strict rules of evidence be used for its guideline development. After the JNC 8 committee established the clinical trial criteria for use in their hypertension guideline, the NHLBI hired a professional vendor to vet the literature. The process was essentially going nowhere so a second vendor for hypertension had to be hired. In the end, few clinical trials passed the criteria used to form the basis of JNC 8. And near its completion, after the JNC 8 group toiled for over 5 years putting their report together, they learned in 2013 that the NHLBI withdrew its support for the guideline and asked the American College of Cardiology/American Heart Association to establish a joint task force (to include the JNC 8 group) to assume responsibility for the guideline process. Declining this collaborative opportunity, the JNC 8 group decided on their own to have their work submitted and it was published without organizational support in JAMA.¹

Meanwhile, the Centers for Disease Control published an overall systems approach for primary care clinicians in which they released the Kaiser Permanente algorithm for effective blood pressure (BP) control.²

Of note, there are several other competing guidelines: one which is issued annually by the Canadian Hypertension Education Program. This approach is now extremely well developed and comprehensive with multiple web-based programs and information sources (www.hypertension.ca/en/chep). In addition, there is the recent effort by the European Society of Hypertension/European Society of Cardiology in 2013,³ and a newer, less formal document from the International Society of Hypertension and supported by a working group comprised of members of the American Society of Hypertension.⁴ Finally, there are other national guidelines around the world, perhaps most prominent among these being the British Hypertension Society National Institute for Clinical Excellence or the NICE guidelines.

DR. IZZO: That’s a good summary, Dr. Basile. What constitutes a useful guideline? Who is the audience? I don’t believe that a good guideline should be hidebound by unduly restrictive ‘evidence rules’ that ignore all but a tiny fraction of the available scientific information. A good guideline must synthesize complex data in a clear and concise way and must address the kinds of questions that real doctors need answered every day. Dr. Gradman, do you have an opinion on guidelines?

DR. GRADMAN: Part of what is going on in the revised hypertension guidelines is a worldwide shift from the treatment of individual patients to the treatment of patient populations. Such guidance for treating large groups is of great interest to governments and insurance companies but is less useful for making treatment decisions for individual patients with individual problems.

In fact, the JNC 8 guidelines specifically say that they are aimed at primary care physicians, not specialists or people who have a lot of experience with the treatment of hypertension. I think JNC 8 is to be commended for trying to make the fundamental concepts of the guidelines completely evidence-based. The guidelines delineate general targets and thresholds for treatment and also suggest the best drugs to use for most patients. Unfortunately, the exercise showed that it is not really possible to answer a lot of specific treatment questions using a strict evidence-based approach. For these we must rely on less definitive evidence and expert opinion.

I think there are different audiences; a good general guideline today should be aimed at primary care physicians but I think specialists need their own guidelines. As a cardiologist, I would want a very different set of guidelines than is presented in JNC 8.

DR. IZZO: Dr. Moser, given your long association with the highly successful parent of the JNC process, the National High Blood Pressure Education Program, I think your perspective is invaluable. Do you think any of the current guidelines hit the mark or what do you think should have been done differently?

DR. MOSER: Well, that’s a broad question. JNC 8 doesn’t differ much from JNC 7 except for one or 2 items. JNC 8 set out to answer several questions with a strict evidence-based methodology, and when they were finished, after a long period of time and substantial expenditure of money, they only ended up addressing 3 questions. The first question was when to initiate treatment, the second was related to the target or goal BPs, and the third was how to achieve goal pressures. They did address the question of whether people with comorbidities, especially diabetes and kidney disease, needed to have lower goal BPs.

On the issue of targets for BP, it was decided that 140/90 mm Hg was good enough for most people and that there was no compelling evidence that BP should be lowered below 130/80 mm Hg in diabetics or people with kidney disease. But we all know that in practice, you can’t adjust BPs to exact numbers. JNC 8 also said that the threshold for a diagnosis and target BP in people over the age of 60 was 150/90 mm Hg. This is quite different from other guidelines.

The guidelines also point out a difference in black and white patients. We’ve known for years that calcium-channel blockers (CCBs) and thiazide-type diuretics are more effective in blacks and angiotensin-converting enzyme (ACE) inhibitors or beta blockers are more effective in whites, so I don’t think JNC 8 provided any new information on this. Otherwise, you’ll find very little difference between JNC 8 and 7. What they didn’t address were possible changes in the diagnostic evaluation or nonpharmacologic therapy and they failed to discuss issues related to the delivery of care and adherence, etc. Some physicians have said that the changes are important and different, but when you look at it very carefully they’re not substantive compared to other reports.

Dr. IZZO: Is that a result of their strict rules of evidence?

Dr. MOSER: The report was supposed to be all evidence-based. They hired an outside vendor that reviewed about 6100 English-language papers. The experts on that panel could have picked up the 40 papers that they finally rated as acceptable within a few days. It took the vendors several years. As everyone has mentioned, they used the data from evidence-based trials and reached conclusions, but they also had an “out” on all of them, including the 150 mm Hg BP target in the elderly, which was arbitrarily defined as age 60. As in their other recommendations, they then said that you could rely on “expert opinion,” meaning clinical judgment. This is the basis of the major recommendation: “If you treat below 140/90 mm Hg and everybody’s happy and there are no side effects, then just continue treatment.”

Regarding their approach to the treatment of resistant patients, the proposed algorithm is almost the same as it has been for many, many years. So I’m very disappointed in what they have accomplished, especially since it took such a long period of time and cost so much money.

DR. IZZO: I echo your disappointment, especially on very fundamental grounds. First, a guideline constrained by the suffocating rules applied by the JNC 8 group is enormously biased, especially when over 99% of the potentially relevant information was ignored because it did not come from a randomized trial. Did they not understand just how biased randomized clinical trials really are? A true expert does better than that, prioritizing the value and relevance of all evidence, from bench science to randomized trials. The real test for a guideline is then in the projection of that synthesis, that is translating expert interpretation into understandable and workable recommendations. One other issue: absence of evidence should not imply a negative recommendation. If there is no definitive study proving that a lower BP target is warranted for people with kidney disease, it should also be very clear that there is existing evidence that lowering BP to 130/80 mm Hg instead of 140/90 mm Hg may be a good idea. JNC 8 chose to ignore studies of achieved (not intention-to-treat) BP values that identified better renal outcomes in diabetics⁵ and fewer recurrent strokes⁶ in trial participants with the lowest BPs. While it is true that such benefits are not achieved by all trial participants, it should be recognized that “lower is better” is probably true in at least some people.

Are there any comments or concerns that the narrow definition of evidence can actually lead us astray?

I’m Barry Massie, Professor of Medicine at the University of California in San Francisco, and I’ll be moderating the call. I’m delighted to be joined by 2 colleagues who have devoted a great deal of effort to the pathophysiology of this syndrome and potential approaches to its treatment: Michael Zile, Professor of Medicine at the Medical University of South Carolina, Principal Investigator at the Gazes Cardiac Research Institute, and Director of the Medical Intensive Care Unit at the Ralph H. Johnson Department of Veterans Affairs Medical Center, and Sanjiv Shah, Associate Professor of Medicine–Cardiology at Northwestern University.

Dr. Zile, our first presenter, will discuss the epidemiology and pathophysiology of HFpEF. I’m sure you will quickly recognize his extensive knowledge and passion for the problem. After that, Dr. Shah, a former chief resident and cardiology fellow from our program at the University of California in San Francisco, who’s established a large program devoted to the study and treatment of HFpEF patients, will discuss his approach to the diagnosis and management of these patients.

DR. ZILE: I would like to start with a very short historical perspective that puts into context some of the terminology that’s already been used by Dr. Massie. Before 2006, EF < 35% was an inclusion criterion in virtually all the HF studies conducted. Therefore, all new medical and device-based management strategies were originally applied selectively to patients with clinical HF with a reduced EF (HFrEF), which was then called systolic HF.

However, as early as the 1970s, it was recognized that there were, in fact, a group of patients who had HF but an EF substantially higher than the 35% cutoff, ie, an EF > 50%. Kenneth Kessler was the first to publish the term “diastolic heart failure” in 1988 to describe such a group of patients.¹

Until 2006, almost no reports of large randomized clinical trials that had examined this group of patients were published. Fortunately, between 2006 and 2009, the results of 5 studies that had examined patients with variable EF cutoffs—35%, 40%, or 45%—were published. Therefore, the term “heart failure with preserved EF” was suggested. For symmetry, a complementary term “heart failure with reduced ejection fraction” was proposed.

The diagnostic criteria commonly used for patients with HFpEF were recently published in a comprehensive review published by the Heart Failure Society of America in Journal of Cardiac Failure.² Diagnosis requires clinical evidence of HF, meaning evidence of both symptoms and signs of HF, often supported by findings on chest radiography, cardiopulmonary exercise testing, or assessment of natriuretic peptides.

In addition to the signs and symptoms of HF, there’s a required EF cutoff or partition value. For the sake this presentation, let’s say that cutoff value is 50%. In addition, patients should have a normal left ventricular (LV) end-diastolic volume.

Supportive evidence for the presence of this clinical syndrome is found echocardiographically in structural or functional changes. One should be cognizant of ruling out nonmyocardial disease.

Patients with HFpEF have clinical characteristics that differ from those of patients with HFrEF. Patients with HFpEF tend to be older. The average age of patients approaches 70 to 75 years. They’re typically women and more often have hypertension and less often coronary artery disease (CAD).

The prevalence of this clinical syndrome exceeds 50% and appears to be increasing over time.^3,4 While the rate of HF hospitalization of HFpEF patients continues to increase, the rate of HF hospitalization for HFrEF patients is actually decreasing. Data from epidemiological studies suggest that the survival rate of both HFrEF patients and HFpEF patients at 5 years posthospitalization was approximately 60%.⁵ In contrast, data from randomized clinical trials suggest significant differences in both morbidity and mortality between these 2 groups of patients. In a recent study by Campbell and McMurray, the clinical trials that randomized both HFrEF and HFpEF patients found different rates of overall mortality and overall hospitalization for these 2 groups, specifically higher rates for patients with HFrEF.⁶ However, they found that the functional impairment experienced by HFrEF and HFpEF patients in terms of exercise intolerance is basically equivalent. One example to support this finding is that cardiopulmonary exercise data and peak oxygen consumption (VO₂) values are roughly equivalent in both groups^7,8 of patients and approach the peak VO₂ value of ≤14 mL/kg/min, a criterion for HF transplantation.⁹

HFpEF is commonly associated with clear abnormalities in both structure and function at the organ and ultrastructure levels. In at least one study published in Circulation, it was demonstrated that even patients with chronic compensated HFpEF have elevated pulmonary artery diastolic pressure or pulmonary capillary wedge pressure (PCWP).¹⁰ Those who then go on to develop decompensated HFpEF have further increases in diastolic pressure. The overall relationship between LV diastolic pressure and volume in patients with HFpEF is pushed up and to the left to a steeper curve such that the LV diastolic pressure is higher in HFpEF patients as compared to normal patients at any common volume.

Echocardiographic findings demonstrate that long-term increases in diastolic pressure are evidenced by an increase in the left atrial volume. In 2 recent echocardiographic substudies of Irbesartan in Heart Failure With Preserved Systolic Function (I-PRESERVE) and Candesartan in Heart Failure Assessment of Reduction in Mortality and Morbidity (CHARM-Preserved), 60% to 70% of these patients experienced increase in left atrial volume.^11,12 This increase in left atrial volume predicted a marked increase in cumulative event rates in terms of primary both end points and HF end points in both studies.

HFpEF patients have significant increases in fibrillar collagen, which are under the control of the cardiac fibroblast. Abnormalities can be seen in at least 4 specific mechanisms controlling collagen homeostasis, including synthesis, postsynthetic processing, posttranslational modification, and degradation.

In one exemplary study by Westermann and colleagues published in 2011, LV biopsy of patients with HFpEF revealed a 3-fold increase in collagen volume fraction and a 4-fold increase in collagen type 1 volume fraction compared to that in the control subjects.¹³

If you look at the evidence of abnormalities and extracellular matrix homeostasis in the plasma, you will find changes in collagen pro-peptides such as the N-terminal pro-peptide of collagen 3; in matrix metalloproteinases (MMPs), specifically MMP2 and MMP8; and in the tissue inhibitors of MMP (TIMPs), specifically TIMP4. Abnormalities in these 4 plasma biomarkers taken in aggregate as a biomarker panel are evidence of abnormalities in extracellular matrix homeostasis and can be used diagnostically to detect the presence of HFpEF.

There are, however, some issues that have been raised regarding the natural history and the underlying pathophysiologic mechanisms in HFpEF. I’ll give you 2 examples.

Some scientists and clinicians have argued that HFpEF is not a clinical syndrome, but rather a group of symptoms and signs caused by comorbidities and associated with specific epidemiologic factors such as advanced age, female sex, CAD, diabetes, and hypertension. However, if this were true, then the morbidity and mortality rates should be comparable between patients who have only these underlying comorbidities and patients who have these comorbidities and HFpEF.

Campbell and McMurray compared the overall mortality and HF hospitalization rates of studies that examined patients with underlying comorbidities such as diabetes, hypertension, or CAD and no HFpEF, versus those patients studied in trials with HFpEF.¹⁴ For example, in studies patients with hypertension, diabetes, or CAD but no HF, the HF hospitalization rate was <12/100 patient-years, whereas in patients with these comorbidities and HFpEF, the rate was >40/100 patient-years.

The second challenge put forth by some scientists and clinicians is that HF is a continuum and HFpEF and HFrEF are, in fact, expressions of one disease process.¹⁵ However, there are clear structural and functional differences between these 2 clinical syndromes and that patients with HFpEF do not, in fact, progress to become patients with HFrEF. I’ll give you 2 sets of evidence to support that.

Patients with HFpEF, in general, have a normal ventricular volume but experience increases in LV mass, which supports the existence of a type of structural abnormality summarized by the term “concentric remodeling” or “concentric hypertrophy.” In contrast, patients with HFrEF experience marked increases in volumes without a proportional increase in LV mass, and thus experience eccentric remodeling or eccentric hypertrophy. These chamber abnormalities are mirrored by ultrastructural abnormalities such as those seen in the cardiomyocyte length, width, and area in animal models of eccentric versus concentric hypertrophy. In addition, there are significant differences in extracellular matrix remodeling.¹⁶

Finally, at least 4 studies demonstrate that over variable periods of time and in the absence of an intercurrent myocardial infarction, patients with concentric remodeling, a common structural manifestation of HFpEF, do not frequently progress to eccentric remodeling, a common structural manifestation of HFrEF. This conversion rate is less than 1% per year.^17–20

I think I’ll stop my comments there and let Dr. Massie talk about the clinical trials that have been performed over the past 5 years.

DR. MASSIE: In terms of treating these patients, I think Dr. Zile already alluded to the fact that it’s challenging, but we’re learning more and getting better at it. Compared with HFrEF, the most commonly studied variety of HF, we have fewer data for HFpEF, but our knowledge is growing. This subject is currently gaining importance.

Among the relatively few randomized trials, I’ll review and comment on 4 of them. Guidelines don’t have much to say because, frankly, we haven’t found much for them to say. Treatment remains relatively empiric, and different clinicians have different biases that affect the treatment that they provide; some, for instance, simply treat HFpEF patients with diuretics when they become edematous.

Several small studies have shown that some agents seem to work. Certainly, diuretics, while not truly addressing the underlying mechanism of HFpEF, make patients feel better when they are volume overloaded. Some have tried calcium blockers, although I don’t think they have much clinical benefit. Beta blockers, angiotensin-converting enzyme (ACE) inhibitors, angiotensin-receptor blockers, and now aldosterone blockers. Most recent trials have focused on inhibitors of the renin angiotensin system. Frankly, there hasn’t been a lot of beneficial evidence.

It’s sort of paradoxical, but the first trial of HFpEF ever conducted on a moderate scale was with digoxin in the Digitalis Investigation Group (DIG) trial.²¹ Given digoxin’s utility as an inotropic agent, studying it in patients with HFpEF is, of course, somewhat counterintuitive. Nevertheless, I think studying these patients was a wonderful idea by Richard Gorlin, who was running the large DIG trial of HFrEF and thought it was about time to look at HFpEF as well.

The DIG study enrolled patients aged 67 years and older and included a larger proportion of women than you’d generally see in trials of HFrEF. The criterion for enrollment was an EF > 45%. The average EF was 55%, but 27% of patients had an EF < 50%. I will talk about that in more detail further, but this is where the 2 types of HF sometimes overlap in these trials.

Only half of the patients had a history of myocardial infarction, which is not surprising given the physiology and pathophysiology that Dr. Zile talked about, but hypertension was the dominant finding. When asked to speculate on the cause, the investigators stated hypertension for 23% of the patients, but we’ll see in other trials that hypertension is much more prevalent. In the Digitalis Investigation Group-Preserved Ejection Fraction (DIG-PEF), 56% of the patients were deemed to have an ischemic etiology.

The CHARM study,²² an angiotensin receptor blocker trial, that also conducted parallel examination of groups of HFrEF and HFpEF patients, observed the same phenomenon observed in the DIG trial: When you examine heterogeneous populations, you ultimately examine people who are more likely to reflect a dominant population, which in this case were HFrEF patients.

The Perindopril in Elderly Persons (PEP) trial,²³ in which the mean age was 75 years and included more women than men, reported a higher mean EF than the previous 2 trials. We don’t know how many patients had EF < 50% because the PEP trial used a wall motion score index rather than the more typical EF value. As in the previous trials, relatively low rates of prior myocardial infarction were observed, and hypertension was thought to be the major player for a very large number of patients.

The last of the 4 investigations is the I-PRESERVE trial,²⁴ the largest of these trials. As in the previous studies, the trial included patients of advanced age with a normal EF and a high percentage of women. Among the 88% of patients who had hypertension, the investigators indicated that the hypertension was responsible for the HFpEF in about two-thirds of the population, although it’s not entirely clear how they could make that judgment. Ischemic etiology was less common.

In the DIG-Preserved trial, there was early separation in the event-free survival curves for digoxin and placebo, which is perhaps a little difficult to explain, but probably one of the few positive findings at any point in any of these trials.

I wanted to elaborate a little more on I-PRESERVE, a trial that included 360 centers in 29 countries, for several reasons. One is that many of us were involved in it, and we believe that it probably provided more data than any other single trial. It used an angiotensin receptor blocker like CHARM did, but clearly separated the HFrEF and HFpEF patients. Over 4000 HFpEF patients developed 140 events.

The inclusion criteria for I-PRESERVE were EF < 45% with symptomatic HF and either recent hospitalization for HF or findings that fit the profile of diastolic HF or HFpEF. Use of an ACE inhibitor was not a contraindication, largely because ACE inhibitors were becoming the standard therapy for diabetics who comprised a fair number of these patients. We also didn’t feel that we could ethically withhold an ACE inhibitor from them.

The primary end point was a composite outcome of all-cause mortality or cardiovascular hospitalization. An echocardiography substudy that Dr. Zile published gives much insight, more than we had in the past, on the pathophysiology of HFpEF.²⁵

There were no patients in whom EF was <45%. In the 2 trials that did not allow patients with EF < 45% (PEP and I-PRESERVE), the Kaplan–Meier event curves did not separate much (ie, there were no major differences in outcomes).

Those are the trials that have been completed, and from which we’ve derived most of our current data. Trial of Aldosterone Antagonist Therapy in Adults With Preserved Ejection Fraction Congestive Heart Failure (TOPCAT), a National Institute of Health-sponsored randomized controlled trial of spironolactone versus placebo in HFpEF,²⁶ has clearly finished enrollment, but the results are not yet available. TOPCAT was originally designed to enroll 4500 patients, but could not reach that total because of challenges in the identification and recruitment of patients with HFpEF. The criteria for TOPCAT enrollment were age ≥50 years; EF ≥45%, as measured within 6 months of enrollment; signs and symptoms of HF; control of blood pressure at the time of study enrollment, despite the fact that the vast majority of patients had a history of hypertension and most other trials did not share this inclusion criterion; and, to ensure adequate event rates, at least one other high-risk HF feature, including hospitalization for HF within the past year and/or elevated levels of brain natriuretic peptides, either B-type natriuretic peptide (BNP) or N-terminal BNP.

Many of us thought that spironolactone might be an ideal drug because fibrosis, which spironolactone can at least reduce, if not prevent, is a major problem in some HFpEF patients.²⁷ The composite primary end point in TOPCAT was cardiovascular death, aborted cardiac arrest, and hospitalization for HF management. The TOPCAT results are projected to be available in mid-2013.

So, what’s the conclusion? Do we know how to treat this entity? The answer in my mind is not yet, but we can prevent it, which would be my approach until we come up with a magic bullet to treat it. Most of these patients have hypertension, and many binge on salt, which is one of the reasons they have hypertension; obesity and diabetes seem to play a role to some degree as well. At this point, we don’t have any definitive answers about therapy.

I’m Barry Massie, Professor of Medicine at the University of California in San Francisco, and I’ll be moderating the call. I’m delighted to be joined by 2 colleagues who have devoted a great deal of effort to the pathophysiology of this syndrome and potential approaches to its treatment: Michael Zile, Professor of Medicine at the Medical University of South Carolina, Principal Investigator at the Gazes Cardiac Research Institute, and Director of the Medical Intensive Care Unit at the Ralph H. Johnson Department of Veterans Affairs Medical Center, and Sanjiv Shah, Associate Professor of Medicine–Cardiology at Northwestern University.

Dr. Zile, our first presenter, will discuss the epidemiology and pathophysiology of HFpEF. I’m sure you will quickly recognize his extensive knowledge and passion for the problem. After that, Dr. Shah, a former chief resident and cardiology fellow from our program at the University of California in San Francisco, who’s established a large program devoted to the study and treatment of HFpEF patients, will discuss his approach to the diagnosis and management of these patients.

DR. ZILE: I would like to start with a very short historical perspective that puts into context some of the terminology that’s already been used by Dr. Massie. Before 2006, EF < 35% was an inclusion criterion in virtually all the HF studies conducted. Therefore, all new medical and device-based management strategies were originally applied selectively to patients with clinical HF with a reduced EF (HFrEF), which was then called systolic HF.

However, as early as the 1970s, it was recognized that there were, in fact, a group of patients who had HF but an EF substantially higher than the 35% cutoff, ie, an EF > 50%. Kenneth Kessler was the first to publish the term “diastolic heart failure” in 1988 to describe such a group of patients.¹

Until 2006, almost no reports of large randomized clinical trials that had examined this group of patients were published. Fortunately, between 2006 and 2009, the results of 5 studies that had examined patients with variable EF cutoffs—35%, 40%, or 45%—were published. Therefore, the term “heart failure with preserved EF” was suggested. For symmetry, a complementary term “heart failure with reduced ejection fraction” was proposed.

The diagnostic criteria commonly used for patients with HFpEF were recently published in a comprehensive review published by the Heart Failure Society of America in Journal of Cardiac Failure.² Diagnosis requires clinical evidence of HF, meaning evidence of both symptoms and signs of HF, often supported by findings on chest radiography, cardiopulmonary exercise testing, or assessment of natriuretic peptides.

In addition to the signs and symptoms of HF, there’s a required EF cutoff or partition value. For the sake this presentation, let’s say that cutoff value is 50%. In addition, patients should have a normal left ventricular (LV) end-diastolic volume.

Supportive evidence for the presence of this clinical syndrome is found echocardiographically in structural or functional changes. One should be cognizant of ruling out nonmyocardial disease.

Patients with HFpEF have clinical characteristics that differ from those of patients with HFrEF. Patients with HFpEF tend to be older. The average age of patients approaches 70 to 75 years. They’re typically women and more often have hypertension and less often coronary artery disease (CAD).

The prevalence of this clinical syndrome exceeds 50% and appears to be increasing over time.^3,4 While the rate of HF hospitalization of HFpEF patients continues to increase, the rate of HF hospitalization for HFrEF patients is actually decreasing. Data from epidemiological studies suggest that the survival rate of both HFrEF patients and HFpEF patients at 5 years posthospitalization was approximately 60%.⁵ In contrast, data from randomized clinical trials suggest significant differences in both morbidity and mortality between these 2 groups of patients. In a recent study by Campbell and McMurray, the clinical trials that randomized both HFrEF and HFpEF patients found different rates of overall mortality and overall hospitalization for these 2 groups, specifically higher rates for patients with HFrEF.⁶ However, they found that the functional impairment experienced by HFrEF and HFpEF patients in terms of exercise intolerance is basically equivalent. One example to support this finding is that cardiopulmonary exercise data and peak oxygen consumption (VO₂) values are roughly equivalent in both groups^7,8 of patients and approach the peak VO₂ value of ≤14 mL/kg/min, a criterion for HF transplantation.⁹

HFpEF is commonly associated with clear abnormalities in both structure and function at the organ and ultrastructure levels. In at least one study published in Circulation, it was demonstrated that even patients with chronic compensated HFpEF have elevated pulmonary artery diastolic pressure or pulmonary capillary wedge pressure (PCWP).¹⁰ Those who then go on to develop decompensated HFpEF have further increases in diastolic pressure. The overall relationship between LV diastolic pressure and volume in patients with HFpEF is pushed up and to the left to a steeper curve such that the LV diastolic pressure is higher in HFpEF patients as compared to normal patients at any common volume.

Echocardiographic findings demonstrate that long-term increases in diastolic pressure are evidenced by an increase in the left atrial volume. In 2 recent echocardiographic substudies of Irbesartan in Heart Failure With Preserved Systolic Function (I-PRESERVE) and Candesartan in Heart Failure Assessment of Reduction in Mortality and Morbidity (CHARM-Preserved), 60% to 70% of these patients experienced increase in left atrial volume.^11,12 This increase in left atrial volume predicted a marked increase in cumulative event rates in terms of primary both end points and HF end points in both studies.

HFpEF patients have significant increases in fibrillar collagen, which are under the control of the cardiac fibroblast. Abnormalities can be seen in at least 4 specific mechanisms controlling collagen homeostasis, including synthesis, postsynthetic processing, posttranslational modification, and degradation.

In one exemplary study by Westermann and colleagues published in 2011, LV biopsy of patients with HFpEF revealed a 3-fold increase in collagen volume fraction and a 4-fold increase in collagen type 1 volume fraction compared to that in the control subjects.¹³

If you look at the evidence of abnormalities and extracellular matrix homeostasis in the plasma, you will find changes in collagen pro-peptides such as the N-terminal pro-peptide of collagen 3; in matrix metalloproteinases (MMPs), specifically MMP2 and MMP8; and in the tissue inhibitors of MMP (TIMPs), specifically TIMP4. Abnormalities in these 4 plasma biomarkers taken in aggregate as a biomarker panel are evidence of abnormalities in extracellular matrix homeostasis and can be used diagnostically to detect the presence of HFpEF.

There are, however, some issues that have been raised regarding the natural history and the underlying pathophysiologic mechanisms in HFpEF. I’ll give you 2 examples.

Some scientists and clinicians have argued that HFpEF is not a clinical syndrome, but rather a group of symptoms and signs caused by comorbidities and associated with specific epidemiologic factors such as advanced age, female sex, CAD, diabetes, and hypertension. However, if this were true, then the morbidity and mortality rates should be comparable between patients who have only these underlying comorbidities and patients who have these comorbidities and HFpEF.

Campbell and McMurray compared the overall mortality and HF hospitalization rates of studies that examined patients with underlying comorbidities such as diabetes, hypertension, or CAD and no HFpEF, versus those patients studied in trials with HFpEF.¹⁴ For example, in studies patients with hypertension, diabetes, or CAD but no HF, the HF hospitalization rate was <12/100 patient-years, whereas in patients with these comorbidities and HFpEF, the rate was >40/100 patient-years.

The second challenge put forth by some scientists and clinicians is that HF is a continuum and HFpEF and HFrEF are, in fact, expressions of one disease process.¹⁵ However, there are clear structural and functional differences between these 2 clinical syndromes and that patients with HFpEF do not, in fact, progress to become patients with HFrEF. I’ll give you 2 sets of evidence to support that.

Patients with HFpEF, in general, have a normal ventricular volume but experience increases in LV mass, which supports the existence of a type of structural abnormality summarized by the term “concentric remodeling” or “concentric hypertrophy.” In contrast, patients with HFrEF experience marked increases in volumes without a proportional increase in LV mass, and thus experience eccentric remodeling or eccentric hypertrophy. These chamber abnormalities are mirrored by ultrastructural abnormalities such as those seen in the cardiomyocyte length, width, and area in animal models of eccentric versus concentric hypertrophy. In addition, there are significant differences in extracellular matrix remodeling.¹⁶

Finally, at least 4 studies demonstrate that over variable periods of time and in the absence of an intercurrent myocardial infarction, patients with concentric remodeling, a common structural manifestation of HFpEF, do not frequently progress to eccentric remodeling, a common structural manifestation of HFrEF. This conversion rate is less than 1% per year.^17–20

I think I’ll stop my comments there and let Dr. Massie talk about the clinical trials that have been performed over the past 5 years.

DR. MASSIE: In terms of treating these patients, I think Dr. Zile already alluded to the fact that it’s challenging, but we’re learning more and getting better at it. Compared with HFrEF, the most commonly studied variety of HF, we have fewer data for HFpEF, but our knowledge is growing. This subject is currently gaining importance.

Among the relatively few randomized trials, I’ll review and comment on 4 of them. Guidelines don’t have much to say because, frankly, we haven’t found much for them to say. Treatment remains relatively empiric, and different clinicians have different biases that affect the treatment that they provide; some, for instance, simply treat HFpEF patients with diuretics when they become edematous.

Several small studies have shown that some agents seem to work. Certainly, diuretics, while not truly addressing the underlying mechanism of HFpEF, make patients feel better when they are volume overloaded. Some have tried calcium blockers, although I don’t think they have much clinical benefit. Beta blockers, angiotensin-converting enzyme (ACE) inhibitors, angiotensin-receptor blockers, and now aldosterone blockers. Most recent trials have focused on inhibitors of the renin angiotensin system. Frankly, there hasn’t been a lot of beneficial evidence.

It’s sort of paradoxical, but the first trial of HFpEF ever conducted on a moderate scale was with digoxin in the Digitalis Investigation Group (DIG) trial.²¹ Given digoxin’s utility as an inotropic agent, studying it in patients with HFpEF is, of course, somewhat counterintuitive. Nevertheless, I think studying these patients was a wonderful idea by Richard Gorlin, who was running the large DIG trial of HFrEF and thought it was about time to look at HFpEF as well.

The DIG study enrolled patients aged 67 years and older and included a larger proportion of women than you’d generally see in trials of HFrEF. The criterion for enrollment was an EF > 45%. The average EF was 55%, but 27% of patients had an EF < 50%. I will talk about that in more detail further, but this is where the 2 types of HF sometimes overlap in these trials.

Only half of the patients had a history of myocardial infarction, which is not surprising given the physiology and pathophysiology that Dr. Zile talked about, but hypertension was the dominant finding. When asked to speculate on the cause, the investigators stated hypertension for 23% of the patients, but we’ll see in other trials that hypertension is much more prevalent. In the Digitalis Investigation Group-Preserved Ejection Fraction (DIG-PEF), 56% of the patients were deemed to have an ischemic etiology.

The CHARM study,²² an angiotensin receptor blocker trial, that also conducted parallel examination of groups of HFrEF and HFpEF patients, observed the same phenomenon observed in the DIG trial: When you examine heterogeneous populations, you ultimately examine people who are more likely to reflect a dominant population, which in this case were HFrEF patients.

The Perindopril in Elderly Persons (PEP) trial,²³ in which the mean age was 75 years and included more women than men, reported a higher mean EF than the previous 2 trials. We don’t know how many patients had EF < 50% because the PEP trial used a wall motion score index rather than the more typical EF value. As in the previous trials, relatively low rates of prior myocardial infarction were observed, and hypertension was thought to be the major player for a very large number of patients.

The last of the 4 investigations is the I-PRESERVE trial,²⁴ the largest of these trials. As in the previous studies, the trial included patients of advanced age with a normal EF and a high percentage of women. Among the 88% of patients who had hypertension, the investigators indicated that the hypertension was responsible for the HFpEF in about two-thirds of the population, although it’s not entirely clear how they could make that judgment. Ischemic etiology was less common.

In the DIG-Preserved trial, there was early separation in the event-free survival curves for digoxin and placebo, which is perhaps a little difficult to explain, but probably one of the few positive findings at any point in any of these trials.

I wanted to elaborate a little more on I-PRESERVE, a trial that included 360 centers in 29 countries, for several reasons. One is that many of us were involved in it, and we believe that it probably provided more data than any other single trial. It used an angiotensin receptor blocker like CHARM did, but clearly separated the HFrEF and HFpEF patients. Over 4000 HFpEF patients developed 140 events.

The inclusion criteria for I-PRESERVE were EF < 45% with symptomatic HF and either recent hospitalization for HF or findings that fit the profile of diastolic HF or HFpEF. Use of an ACE inhibitor was not a contraindication, largely because ACE inhibitors were becoming the standard therapy for diabetics who comprised a fair number of these patients. We also didn’t feel that we could ethically withhold an ACE inhibitor from them.

The primary end point was a composite outcome of all-cause mortality or cardiovascular hospitalization. An echocardiography substudy that Dr. Zile published gives much insight, more than we had in the past, on the pathophysiology of HFpEF.²⁵

There were no patients in whom EF was <45%. In the 2 trials that did not allow patients with EF < 45% (PEP and I-PRESERVE), the Kaplan–Meier event curves did not separate much (ie, there were no major differences in outcomes).

Those are the trials that have been completed, and from which we’ve derived most of our current data. Trial of Aldosterone Antagonist Therapy in Adults With Preserved Ejection Fraction Congestive Heart Failure (TOPCAT), a National Institute of Health-sponsored randomized controlled trial of spironolactone versus placebo in HFpEF,²⁶ has clearly finished enrollment, but the results are not yet available. TOPCAT was originally designed to enroll 4500 patients, but could not reach that total because of challenges in the identification and recruitment of patients with HFpEF. The criteria for TOPCAT enrollment were age ≥50 years; EF ≥45%, as measured within 6 months of enrollment; signs and symptoms of HF; control of blood pressure at the time of study enrollment, despite the fact that the vast majority of patients had a history of hypertension and most other trials did not share this inclusion criterion; and, to ensure adequate event rates, at least one other high-risk HF feature, including hospitalization for HF within the past year and/or elevated levels of brain natriuretic peptides, either B-type natriuretic peptide (BNP) or N-terminal BNP.

Many of us thought that spironolactone might be an ideal drug because fibrosis, which spironolactone can at least reduce, if not prevent, is a major problem in some HFpEF patients.²⁷ The composite primary end point in TOPCAT was cardiovascular death, aborted cardiac arrest, and hospitalization for HF management. The TOPCAT results are projected to be available in mid-2013.

So, what’s the conclusion? Do we know how to treat this entity? The answer in my mind is not yet, but we can prevent it, which would be my approach until we come up with a magic bullet to treat it. Most of these patients have hypertension, and many binge on salt, which is one of the reasons they have hypertension; obesity and diabetes seem to play a role to some degree as well. At this point, we don’t have any definitive answers about therapy.

I’m Barry Massie, Professor of Medicine at the University of California in San Francisco, and I’ll be moderating the call. I’m delighted to be joined by 2 colleagues who have devoted a great deal of effort to the pathophysiology of this syndrome and potential approaches to its treatment: Michael Zile, Professor of Medicine at the Medical University of South Carolina, Principal Investigator at the Gazes Cardiac Research Institute, and Director of the Medical Intensive Care Unit at the Ralph H. Johnson Department of Veterans Affairs Medical Center, and Sanjiv Shah, Associate Professor of Medicine–Cardiology at Northwestern University.

Dr. Zile, our first presenter, will discuss the epidemiology and pathophysiology of HFpEF. I’m sure you will quickly recognize his extensive knowledge and passion for the problem. After that, Dr. Shah, a former chief resident and cardiology fellow from our program at the University of California in San Francisco, who’s established a large program devoted to the study and treatment of HFpEF patients, will discuss his approach to the diagnosis and management of these patients.

DR. ZILE: I would like to start with a very short historical perspective that puts into context some of the terminology that’s already been used by Dr. Massie. Before 2006, EF < 35% was an inclusion criterion in virtually all the HF studies conducted. Therefore, all new medical and device-based management strategies were originally applied selectively to patients with clinical HF with a reduced EF (HFrEF), which was then called systolic HF.

However, as early as the 1970s, it was recognized that there were, in fact, a group of patients who had HF but an EF substantially higher than the 35% cutoff, ie, an EF > 50%. Kenneth Kessler was the first to publish the term “diastolic heart failure” in 1988 to describe such a group of patients.¹

Until 2006, almost no reports of large randomized clinical trials that had examined this group of patients were published. Fortunately, between 2006 and 2009, the results of 5 studies that had examined patients with variable EF cutoffs—35%, 40%, or 45%—were published. Therefore, the term “heart failure with preserved EF” was suggested. For symmetry, a complementary term “heart failure with reduced ejection fraction” was proposed.

The diagnostic criteria commonly used for patients with HFpEF were recently published in a comprehensive review published by the Heart Failure Society of America in Journal of Cardiac Failure.² Diagnosis requires clinical evidence of HF, meaning evidence of both symptoms and signs of HF, often supported by findings on chest radiography, cardiopulmonary exercise testing, or assessment of natriuretic peptides.

In addition to the signs and symptoms of HF, there’s a required EF cutoff or partition value. For the sake this presentation, let’s say that cutoff value is 50%. In addition, patients should have a normal left ventricular (LV) end-diastolic volume.

Supportive evidence for the presence of this clinical syndrome is found echocardiographically in structural or functional changes. One should be cognizant of ruling out nonmyocardial disease.

Patients with HFpEF have clinical characteristics that differ from those of patients with HFrEF. Patients with HFpEF tend to be older. The average age of patients approaches 70 to 75 years. They’re typically women and more often have hypertension and less often coronary artery disease (CAD).

The prevalence of this clinical syndrome exceeds 50% and appears to be increasing over time.^3,4 While the rate of HF hospitalization of HFpEF patients continues to increase, the rate of HF hospitalization for HFrEF patients is actually decreasing. Data from epidemiological studies suggest that the survival rate of both HFrEF patients and HFpEF patients at 5 years posthospitalization was approximately 60%.⁵ In contrast, data from randomized clinical trials suggest significant differences in both morbidity and mortality between these 2 groups of patients. In a recent study by Campbell and McMurray, the clinical trials that randomized both HFrEF and HFpEF patients found different rates of overall mortality and overall hospitalization for these 2 groups, specifically higher rates for patients with HFrEF.⁶ However, they found that the functional impairment experienced by HFrEF and HFpEF patients in terms of exercise intolerance is basically equivalent. One example to support this finding is that cardiopulmonary exercise data and peak oxygen consumption (VO₂) values are roughly equivalent in both groups^7,8 of patients and approach the peak VO₂ value of ≤14 mL/kg/min, a criterion for HF transplantation.⁹

HFpEF is commonly associated with clear abnormalities in both structure and function at the organ and ultrastructure levels. In at least one study published in Circulation, it was demonstrated that even patients with chronic compensated HFpEF have elevated pulmonary artery diastolic pressure or pulmonary capillary wedge pressure (PCWP).¹⁰ Those who then go on to develop decompensated HFpEF have further increases in diastolic pressure. The overall relationship between LV diastolic pressure and volume in patients with HFpEF is pushed up and to the left to a steeper curve such that the LV diastolic pressure is higher in HFpEF patients as compared to normal patients at any common volume.

Echocardiographic findings demonstrate that long-term increases in diastolic pressure are evidenced by an increase in the left atrial volume. In 2 recent echocardiographic substudies of Irbesartan in Heart Failure With Preserved Systolic Function (I-PRESERVE) and Candesartan in Heart Failure Assessment of Reduction in Mortality and Morbidity (CHARM-Preserved), 60% to 70% of these patients experienced increase in left atrial volume.^11,12 This increase in left atrial volume predicted a marked increase in cumulative event rates in terms of primary both end points and HF end points in both studies.

HFpEF patients have significant increases in fibrillar collagen, which are under the control of the cardiac fibroblast. Abnormalities can be seen in at least 4 specific mechanisms controlling collagen homeostasis, including synthesis, postsynthetic processing, posttranslational modification, and degradation.

In one exemplary study by Westermann and colleagues published in 2011, LV biopsy of patients with HFpEF revealed a 3-fold increase in collagen volume fraction and a 4-fold increase in collagen type 1 volume fraction compared to that in the control subjects.¹³

If you look at the evidence of abnormalities and extracellular matrix homeostasis in the plasma, you will find changes in collagen pro-peptides such as the N-terminal pro-peptide of collagen 3; in matrix metalloproteinases (MMPs), specifically MMP2 and MMP8; and in the tissue inhibitors of MMP (TIMPs), specifically TIMP4. Abnormalities in these 4 plasma biomarkers taken in aggregate as a biomarker panel are evidence of abnormalities in extracellular matrix homeostasis and can be used diagnostically to detect the presence of HFpEF.

There are, however, some issues that have been raised regarding the natural history and the underlying pathophysiologic mechanisms in HFpEF. I’ll give you 2 examples.

Some scientists and clinicians have argued that HFpEF is not a clinical syndrome, but rather a group of symptoms and signs caused by comorbidities and associated with specific epidemiologic factors such as advanced age, female sex, CAD, diabetes, and hypertension. However, if this were true, then the morbidity and mortality rates should be comparable between patients who have only these underlying comorbidities and patients who have these comorbidities and HFpEF.

Campbell and McMurray compared the overall mortality and HF hospitalization rates of studies that examined patients with underlying comorbidities such as diabetes, hypertension, or CAD and no HFpEF, versus those patients studied in trials with HFpEF.¹⁴ For example, in studies patients with hypertension, diabetes, or CAD but no HF, the HF hospitalization rate was <12/100 patient-years, whereas in patients with these comorbidities and HFpEF, the rate was >40/100 patient-years.

The second challenge put forth by some scientists and clinicians is that HF is a continuum and HFpEF and HFrEF are, in fact, expressions of one disease process.¹⁵ However, there are clear structural and functional differences between these 2 clinical syndromes and that patients with HFpEF do not, in fact, progress to become patients with HFrEF. I’ll give you 2 sets of evidence to support that.

Patients with HFpEF, in general, have a normal ventricular volume but experience increases in LV mass, which supports the existence of a type of structural abnormality summarized by the term “concentric remodeling” or “concentric hypertrophy.” In contrast, patients with HFrEF experience marked increases in volumes without a proportional increase in LV mass, and thus experience eccentric remodeling or eccentric hypertrophy. These chamber abnormalities are mirrored by ultrastructural abnormalities such as those seen in the cardiomyocyte length, width, and area in animal models of eccentric versus concentric hypertrophy. In addition, there are significant differences in extracellular matrix remodeling.¹⁶

Finally, at least 4 studies demonstrate that over variable periods of time and in the absence of an intercurrent myocardial infarction, patients with concentric remodeling, a common structural manifestation of HFpEF, do not frequently progress to eccentric remodeling, a common structural manifestation of HFrEF. This conversion rate is less than 1% per year.^17–20

I think I’ll stop my comments there and let Dr. Massie talk about the clinical trials that have been performed over the past 5 years.

DR. MASSIE: In terms of treating these patients, I think Dr. Zile already alluded to the fact that it’s challenging, but we’re learning more and getting better at it. Compared with HFrEF, the most commonly studied variety of HF, we have fewer data for HFpEF, but our knowledge is growing. This subject is currently gaining importance.

Among the relatively few randomized trials, I’ll review and comment on 4 of them. Guidelines don’t have much to say because, frankly, we haven’t found much for them to say. Treatment remains relatively empiric, and different clinicians have different biases that affect the treatment that they provide; some, for instance, simply treat HFpEF patients with diuretics when they become edematous.

Several small studies have shown that some agents seem to work. Certainly, diuretics, while not truly addressing the underlying mechanism of HFpEF, make patients feel better when they are volume overloaded. Some have tried calcium blockers, although I don’t think they have much clinical benefit. Beta blockers, angiotensin-converting enzyme (ACE) inhibitors, angiotensin-receptor blockers, and now aldosterone blockers. Most recent trials have focused on inhibitors of the renin angiotensin system. Frankly, there hasn’t been a lot of beneficial evidence.

It’s sort of paradoxical, but the first trial of HFpEF ever conducted on a moderate scale was with digoxin in the Digitalis Investigation Group (DIG) trial.²¹ Given digoxin’s utility as an inotropic agent, studying it in patients with HFpEF is, of course, somewhat counterintuitive. Nevertheless, I think studying these patients was a wonderful idea by Richard Gorlin, who was running the large DIG trial of HFrEF and thought it was about time to look at HFpEF as well.

The DIG study enrolled patients aged 67 years and older and included a larger proportion of women than you’d generally see in trials of HFrEF. The criterion for enrollment was an EF > 45%. The average EF was 55%, but 27% of patients had an EF < 50%. I will talk about that in more detail further, but this is where the 2 types of HF sometimes overlap in these trials.

Only half of the patients had a history of myocardial infarction, which is not surprising given the physiology and pathophysiology that Dr. Zile talked about, but hypertension was the dominant finding. When asked to speculate on the cause, the investigators stated hypertension for 23% of the patients, but we’ll see in other trials that hypertension is much more prevalent. In the Digitalis Investigation Group-Preserved Ejection Fraction (DIG-PEF), 56% of the patients were deemed to have an ischemic etiology.

The CHARM study,²² an angiotensin receptor blocker trial, that also conducted parallel examination of groups of HFrEF and HFpEF patients, observed the same phenomenon observed in the DIG trial: When you examine heterogeneous populations, you ultimately examine people who are more likely to reflect a dominant population, which in this case were HFrEF patients.

The Perindopril in Elderly Persons (PEP) trial,²³ in which the mean age was 75 years and included more women than men, reported a higher mean EF than the previous 2 trials. We don’t know how many patients had EF < 50% because the PEP trial used a wall motion score index rather than the more typical EF value. As in the previous trials, relatively low rates of prior myocardial infarction were observed, and hypertension was thought to be the major player for a very large number of patients.

The last of the 4 investigations is the I-PRESERVE trial,²⁴ the largest of these trials. As in the previous studies, the trial included patients of advanced age with a normal EF and a high percentage of women. Among the 88% of patients who had hypertension, the investigators indicated that the hypertension was responsible for the HFpEF in about two-thirds of the population, although it’s not entirely clear how they could make that judgment. Ischemic etiology was less common.

In the DIG-Preserved trial, there was early separation in the event-free survival curves for digoxin and placebo, which is perhaps a little difficult to explain, but probably one of the few positive findings at any point in any of these trials.

I wanted to elaborate a little more on I-PRESERVE, a trial that included 360 centers in 29 countries, for several reasons. One is that many of us were involved in it, and we believe that it probably provided more data than any other single trial. It used an angiotensin receptor blocker like CHARM did, but clearly separated the HFrEF and HFpEF patients. Over 4000 HFpEF patients developed 140 events.

The inclusion criteria for I-PRESERVE were EF < 45% with symptomatic HF and either recent hospitalization for HF or findings that fit the profile of diastolic HF or HFpEF. Use of an ACE inhibitor was not a contraindication, largely because ACE inhibitors were becoming the standard therapy for diabetics who comprised a fair number of these patients. We also didn’t feel that we could ethically withhold an ACE inhibitor from them.

The primary end point was a composite outcome of all-cause mortality or cardiovascular hospitalization. An echocardiography substudy that Dr. Zile published gives much insight, more than we had in the past, on the pathophysiology of HFpEF.²⁵

There were no patients in whom EF was <45%. In the 2 trials that did not allow patients with EF < 45% (PEP and I-PRESERVE), the Kaplan–Meier event curves did not separate much (ie, there were no major differences in outcomes).

Those are the trials that have been completed, and from which we’ve derived most of our current data. Trial of Aldosterone Antagonist Therapy in Adults With Preserved Ejection Fraction Congestive Heart Failure (TOPCAT), a National Institute of Health-sponsored randomized controlled trial of spironolactone versus placebo in HFpEF,²⁶ has clearly finished enrollment, but the results are not yet available. TOPCAT was originally designed to enroll 4500 patients, but could not reach that total because of challenges in the identification and recruitment of patients with HFpEF. The criteria for TOPCAT enrollment were age ≥50 years; EF ≥45%, as measured within 6 months of enrollment; signs and symptoms of HF; control of blood pressure at the time of study enrollment, despite the fact that the vast majority of patients had a history of hypertension and most other trials did not share this inclusion criterion; and, to ensure adequate event rates, at least one other high-risk HF feature, including hospitalization for HF within the past year and/or elevated levels of brain natriuretic peptides, either B-type natriuretic peptide (BNP) or N-terminal BNP.

Many of us thought that spironolactone might be an ideal drug because fibrosis, which spironolactone can at least reduce, if not prevent, is a major problem in some HFpEF patients.²⁷ The composite primary end point in TOPCAT was cardiovascular death, aborted cardiac arrest, and hospitalization for HF management. The TOPCAT results are projected to be available in mid-2013.

So, what’s the conclusion? Do we know how to treat this entity? The answer in my mind is not yet, but we can prevent it, which would be my approach until we come up with a magic bullet to treat it. Most of these patients have hypertension, and many binge on salt, which is one of the reasons they have hypertension; obesity and diabetes seem to play a role to some degree as well. At this point, we don’t have any definitive answers about therapy.