Clinical Feature: Translating Clinical Trial Evidence into Practical Medicine

Over the course of my 36-year career, the practice of medicine has changed dramatically. When I was an intern at Georgetown University Hospital, we treated heart attacks with oxygen, morphine, and sometimes not much more than prayers. Barely anyone at that time could conceive of doing a cardiac catheterization on an acute myocardial infarction. The progress we all enjoy in medicine is based on two major factors: the rapid evolution of medical technology, and scientific evidence gleaned from clinical trials, particularly those trials powered adequately enough to reduce cardiovascular morbidity and mortality.¹

Outcome-based clinical research in cardiovascular medicine has given birth to innumerable treatments that have been proven to reduce morbidity and, in some cases, total mortality. The Scandinavian Simvastatin Survival Study (4S)² was the “grandfather study in lipidology” that was shown to decrease overall mortality. It is this type of research that has driven us to utilize pharmaceuticals and technology to ultimately benefit our patients. Our best clinical pathways and treatment guidelines are derived from a generation of formidable clinical research. For instance, as of 2014, who would not consider using aspirin, beta blockers, and statins for patients with a recent myocardial infarction? From the Second International Study of Infarct Survival (ISIS-2)³ to the Beta-Blocker Heart Attack Trial (BHAT)⁴ and on to 4S, to mention a few, we have established clinical guidelines that benefit patient outcomes.

Having been a clinical trialist and clinician as well, I often ask myself if the patient before me whom I am asked to manage would have been included in one or more of the outcome studies based on the specific recruitment criteria.

Often, because of patient diversity, specific clinical trial data is not directly applicable. We are obliged to make clinical decisions for the patients’ maximum benefit based on experience and/or extrapolation of scientific outcome results. It is only with clear knowledge of specific trial inclusion and exclusion criteria, and endpoint results that we make clear evidence-based decisions.

For example, in 1995, the 4S trial randomized 4,444 male patients with coronary artery disease to receive an average dose of 33 mg of simvastatin versus placebo. The average pre-study low-density lipoprotein cholesterol (LDL-C) level was 188 mg/dl. So, going back to 1995, how should we have best managed a female coronary patient with an LDL-C level of 140? At that time, we were forced to make a decision that was not guided directly by patient-specific outcome data. Clinical judgment infused by clinical trial data guided our decision to use lipid- lowering therapy in this atypical patient. Clinical trial data often lags behind, or never will exactly answer how best to treat the vast majority of patients we treat. Fortunately, many statin trials since 1995 added diverse patients as compared to the 4S population, which make our decisions easier. In addition, safety data from many sources about statins in clinical trials and in practice makes us very comfortable with their use.⁵

We continually search to further reduce risk by combining statins with other medications both in practice and in randomized clinical trials. There have been four recent studies combining statins with other pharmaceutical agents to treat metabolic dyslipidemia.6-9 Two of the studies combined statins with fibrates and two combined statins with long-acting niacin.^6,7 Intuitively, as lipidologists, we would like to believe that treatable risk exists for hypertriglyceridemia and low high-density lipoprotein cholesterol (HDL-C), and yet all of these RCT studies seemingly refute this belief. No study and no clinical trials are perfect both in design and/or implementation. In the following, I will dissect the recent studies that combine statins and fibrates/niacin for purposes of attempting to help the practicing clinician make rational decisions about the use of combination therapy in patients with metabolic dyslipidemia.

The Fenofibrate Intervention and Event Lowering in Diabetes (FIELD) study6 randomized nearly 10,000 diabetic patients to receive fenofibrate versus placebo, usually in combination with statins. Of the randomized patients in the FIELD study, 80 percent had the metabolic syndrome, but only 38 percent had metabolic dyslipidemia, defined as a triglyceride level > 150 mg/dl, with a HDL-C level < 40mg/dl in males and < 50mg/dl in females. Inclusion criteria for participation in the FIELD study were a total cholesterol level from 116 mg/dl to 251mg/dl, a total cholesterol to HDL-C ratio (Chol/HDLc)  > 4, and a triglyceride level > 89mg/dl. The mean LDL-C level prior to randomization was 119mg/dl. The mean HDL-C and triglyceride levels were 42mg/dl and 153mg/dl, respectively. It is well known that FIELD investigators were allowed to titrate statin doses in accordance with their clinical judgment based on unblinded LDL-C levels. Unfortunately, 17 percent of the placebo group in this trial were uptitrated on their statin therapy; whereas only 9 percent of the fenofibrate group received additive statin treatment. How much this influenced the negative trial outcome is unclear.

The outcomes of FIELD were somewhat disappointing for many of us. It showed no statistically significant reduction in the primary endpoint of cardiovascular morbidity and mortality. Overall mortality was higher in the fenofibrate group, but this was not statistically significant. The major benefit of fenofibrate in FIELD was on diabetic microvascular disease progression. This was seen in the retinal subpopulation group as well as in those with peripheral vascular disease and microalbuminuria.  A clinical question that arises, based on the design flaws in FIELD, is what we should do with a statin-treated diabetic patient with coronary artery disease who has an HDL of 30 mg/dl and has triglycerides of 300 mg/dl?

In the Helsinki Heart Study (HHS), there was a dramatic benefit seen in patients on a fibrate who had triglyceride levels greater than 204 mg/dl and an LDL/HDL ratio of >5.10 In the subpopulation of FIELD with true metabolic dyslipidemia, the outcomes were very favorable, with a significant reduction in cardiovascular morbidity and mortality. The pertinent clinical issue is: Can we rely on subpopulation data to make patient decisions? The statistical answer to this question is no; yet the majority of patients randomized to FIELD would not have been placed on combination therapy in my practice! Sometimes we are forced to rely on subpopulation results.

As clinicians, we are confronted with a quandary based on study design and overall outcomes in a diabetic population. Here is where we must, in my view, avoid generalization and treat the patient at hand, even without direct outcome data to perfectly support our best judgment. Would I treat a diabetic with coronary disease who has an HDL of 30mg/dl and triglycerides of 300 mg/dl with fenofibrate when on statin therapy? For me, the answer is yes. And for those who would not treat this patient because the FIELD trial did not reach clear statistical differences in its primary outcome, a valid approach would be to consider the benefits of treatment on the outcome of microvascular disease alone. Many diabetics develop complications from retinopathy, nephropathy, or peripheral vascular disease, and this often results in blindness, dialysis, or amputation. While morbidity and mortality are the strongest outcomes to base clinical judgment; we also must have regard for a patient’s pain and suffering. Often, these components of patient care and judgment are not readily appreciated as critical components of evidence-based medicine.

The Action to Control Cardiovascular Risk in Diabetes (ACCORD) lipid trial7 was sponsored by the National Institutes of Health’s Heart, Lung and Blood Institute (NHLBI) to answer the question of whether adding a fenofibrate to a statin in diabetic patients influences cardiovascular outcomes. The diabetic population in ACCORD was at more risk for CVD than those entered into the FIELD study. Similar results were found however. After nearly five years of combined therapy, no significant differences in the primary outcome of myocardial infarction, ischemic stroke, and cardiovascular death were found in the comparator arms. Mean entry lipid levels of LDL-C in the ACCORD trial were 100 mg/dl, triglycerides of 162 mg/dl, and HDL-C of 38 mg/dl. These entry lipid values are not characteristic of the highest- risk group of diabetics with profound metabolic dyslipidemia. Microvascular outcomes, nevertheless, were favorably affected. Retinopathy progression was 6.5 percent of the fenofibrate group and 10.2 percent in the placebo group (P = .003). Progression of microalbuminuria also was reduced with combination statin-fibrate treatment. And finally, a continuing theme is that in the ACCORD trial for participants with triglycerides >204 mg/dL and who also had an HDL-C<39 mg/dL, there was a 31-percent decrease in the primary combined endpoint.

The take-home message here is that, if we had chosen a higher-risk population to study in the first place, outcomes may have been different. Also, in my view, only a therapeutic nihilist would disregard the beneficial effects of fenofibrate on the progression of microvascular disease in diabetics based on combined FIELD/ ACCORD results!

How about the recent niacin outcome studies — from trial data to practical use in clinical practice? The Atherothrombosis Intervention in Metabolic Syndrome with Low HDL/High Triglyceride and Impact on Global Health Outcomes (AIM-HIGH) study⁸ was a prospective randomized clinical trial adding niacin ER therapy to high-risk statin-treated patients. Participants had mean entry LDL-C levels of 71mg/dl, mean baseline triglyceride levels of 161 mg/dl, and mean HDL levels of 35 mg/dl. Also, 1,500-2,000 mg of extended- release niacin was added to simvastatin based on pre-randomization tolerance and compared to no addition (placebo). The placebo group was asked to take 250 mg of crystalline niacin to help maintain the blind due to flushing.

During the course of the AIM-HIGH study, investigators were allowed to manipulate statin doses to maintain LDL-C levels less than 80. There was considerably more statin-added therapy on trial in the placebo group. This could have influenced the trial outcome. The study (n = 3,414) was not large and a 25-percent dropout and down titration rate in the active treatment group occurred.

The results of AIM-HIGH showed no significant difference in cardiovascular morbidity or mortality at a mean follow-up time of 36 months in the niacin-treated patients. There was a tendency to more adverse events in the treatment group and a disturbing increase in stroke rate. Based on these findings the AIM-HIGH data and safety monitoring board (DSMB) discontinued the trial prematurely, with only 50 percent of predicted outcomes events recorded.

For most clinicians and trialists,^7,8 the AIM-HIGH study results came as a surprise. We have all seen progression of coronary and vascular disease in patients with optimal LDL-C levels and significant metabolic dyslipidemia. These results left us with high hopes of positive results from the other niacin-statin mega trial Heart Protection Study 2 — Treatment of HDL to Reduce the Incidence of Vascular Events (HPS2-THRIVE) trial.⁹

HPS-2 THRIVE randomized more than 25,000 high-risk vascular patients worldwide to compare niacin extended release (ER) plus laropiprant, an antiflushing agent, added to optimal statin therapy versus placebo. There were no specific HDL-C inclusion criteria in this study. Baseline mean lipid values in the HPS-2 THRIVE on simvastatin, with or without Ezetimibe, were an LDL-C of 63 mg/dl, HDL-C of 44 mg/dl, and triglycerides of 125 mg/dl. There were 10,000 Chinese patients randomized to HPS-2 THRIVE.

The DSMB of HPS-2 STRIVE stopped the study prematurely because of a lack of benefit on the primary cardiovascular morbidity/mortality endpoint and an increase in adverse events. The majority of these adverse events were related to “minor hyperglycemic problems.” Specifically, the skin reaction rate was from four to five times as high and the rate of musculoskeletal events was nearly twice as high in the treatment (added niacin ER) group. The rate of serious myopathy was six-fold higher in the treatment group. Gastrointestinal side effects, overall bleeding and infection rates also were increased in the study patients on niacin/laropiprant. A disproportionate number of adverse events occurred in the Chinese cohort.

So, as trialist and clinicians scratch their heads about these results, what should we do practically for our high-risk patients with metabolic dyslipidemia and well-controlled LDL-C levels on statin therapy?

My own clinical judgment is as follows: (Note that most of my patients on statin therapy with metabolic dyslipidemia have demonstrated vascular disease and are, therefore, at high or highest NCEP risk.)¹¹

1. Particularly in patients with mild metabolic dyslipidemia on statin therapy, diet, exercise, and weight loss are almost always effective at normalizing the lipid panel. Getting patients to consume less carbohydrate and fat calories and to participate in a programmed exercise regimen is extremely effective.
2. The majority of statin patients I have put on fibrates or niacin in the past 30 years have lipid profiles before treatment that are much worse than mean HDL and triglyceride levels in all of the aforementioned studies. I am not taking any of these patients off therapy unless they experience side effects. How do I rationalize this? There are four prior fibrate trials (HHS,¹⁰ Veterans’ Affairs — High-Density Lipoprotein Intervention Trial [VA-HIT],¹²  FIELD, and ACCORD), one prescription omega-3 study (Japan eicosapentaenoic acid Lipid Intervention Study [JELIS]),¹³ and two niacin trials (Coronary Drug Project [CDP]¹⁴  and AIM- HIGH) that show benefits in those patients in the lowest HDL and highest triglyceride groups.
3. In those high-risk patients with significant metabolic dyslipidemia (triglycerides>200mg/dl, HDL <35mg/dl in males, <45mg/dl in females), I am still prescribing both fibrates and niacin in combination with statins. I am particularly careful to monitor the use of niacin ER, especially during the titration phase. I monitor my patients’ fasting glucoses with each titration of niacin if they are diabetic and every other titration if they are non-diabetic.
4. I am diligent in educating the niacin ER-treated patients about side effects. With this caution, I am able to get more than 70 percent of patients on 1,000 mg or more.
5. I use fish oil (1,000 mg/day) in all patients with vascular disease, regardless of their lipid profile.
6. I believe it is reasonable to use fibrates in high-risk diabetics, particularly those with metabolic dyslipidemia, to reduce microvascular disease.

The practice of medicine should always be based on best evidence when there is proven outcome data relevant to patients meeting the inclusion and exclusion criteria of the trials. However, with any individual patient, certainty from existing trial data does not always directly apply to that individual. This is where the art of medicine and clinical experience are necessary to make therapeutic decisions. The hard endpoints of most CVD clinical trials are not all that should guide us in our treatment strategies. Pain and suffering are important in patient care, but are not common endpoints in cardiovascular trials. As a trialist and clinician, I believe the marriage of scientific knowledge and practical experience should guide our best decisions for patient care. 

Disclosure statement: Dr. Vicari has no disclosures to report.

References are listed on page 35 of the PDF.