“Doctor do I really need this drug?” Uncertainty and best judgements in cholesterol drugs
One can only imagine how anxious yesterday’s news of a possible link to cancer must have left people taking cholesterol-lowering drugs. With the publication of the results of the 5-year SEAS trial (Simvastatin and Ezetimibe in Aortic Stenosis), arguments have been vehemently made on both sides of cancer risks, making things even more confusing. But no one is arguing about two things.
No one is debating the null finding of the clinical trial. Vytorin (Zetia and Zocor) was shown to have no effect on the progression of aortic stenosis or to be of any benefit in the primary clinical outcomes — reducing cardiovascular events or all-cause mortality.
And no one is very pleased about the way the cancer findings were released, with a sponsor-funded press conference earlier this summer and, now, the unblinding of two studies that are still ongoing, with the subsequent release of incomplete safety information to the public that could lead doctors and patients to make decisions based on insufficient data. It may simply be a matter of appearances, and a company on the defensive chosing to bypass the medical community and release science to media before publishing it in peer-reviewed journals. But that served to fuel suspicions and concerns that there might be something more to the findings.
It all began when the SEAS trial came up with an unexpected finding of more cases of cancer among those taking the medication compared to those on a placebo (105 versus 70 cases). Was it just a statistical fluke or is there other evidence to suggest a concern?
The sad fact of the matter is that doctors and patients, alike, are both left with continued uncertainty about simvastatin and ezetimibe’s efficacy and safety. There is no clean, concise evidence either way that we can point to with assurance to help guide us. Cancer correlations come from incidental reports from trials’ adverse event reports. After all these years, there has been no study specifically following cancer incidences, provided adequate information on their onset or follow-up of sufficient duration to reliably evaluate the risks.
The ENHANCE trial of Vytorin first brought public attention to debates surrounding this drug as this clinical trial itself became a melodrama, with the ongoing IMPROVE-IT trial likely to follow a similar path of confusion. The ENHANCE trial findings were anticlimactic by the time they were published, reporting no clinically meaningful benefits for the patients on Vytorin compared to those on simvastatin alone on the progression of coronary artery disease, despite significant decreases in their LDL-cholesterol levels. But the controversy did help to call attention to the problem of treating surrogate measures — such as cholesterol numbers — and relying on surrogate endpoints as being indications of health or future health outcomes. More people now understand the term “false surrogate endpoint” and know that health indices are not the same thing as actual clinical measures of health.
Most frustrating, is that some of this entire worry might have been prevented long ago, had the drugs been soundly shown on balance, to have a firm benefit on actual hard clinical outcomes, as in all-cause mortality, over risks. Instead, the drugs were approved and prescribed based only on their ability to lower cholesterol levels, with the belief that that would translate to longer, healthier lives, and millions of people were prescribed them before clinical benefits were known. What has been even more unsettling is that the reports downplaying the risks seem to most come from those with an appearance of bias, whereas the reports heightening concerns often don’t.
Few people are privy to the revealing discussions that must be taking place at the European Society of Cardiology Congress 2008 in Munich, Germany this week. While the cardiologists are reported as backing the drug’s safety, a New England Journal of Medicine editorial cautions it’s too soon to dismiss concerns.
Because of the enormous focus on the results of the SEAS trial published in the New England Journal of Medicine, the full SEAS article is available online to the public for free. This was a randomized, double-blind, placebo-controlled clinical trial on 1,873 patients across seven countries, who were 45-85 years old with asymptomatic aortic stenosis. Half received ezetimibe and simvastatin (Vytorin) and the other half a placebo. The primary outcomes were to see if lowering cholesterol could help to lower risks for major cardiovascular events, as well as deaths from any cause.
The trial had originally been developed and begun to test simvastatin alone but, according to the study authors, 21 months into the trial, “the logistics of the SEAS trial was transferred to the sponsor,” Merck and Schering-Plough Pharmaceuticals, after the company suggested adding ezetimibe to the protocol. The authors said that “scientific responsibility remained with the independent steering committee, which included two nonvoting members of the sponsor.”
After an average of 52.2 months, the LDL-cholesterol levels among the intervention group had decreased by an average of 53.8%, compared to no change in the placebo group. While there was a 4% lower odds ratio for the primary cardiovascular-related outcome measures among the intervention group, there was a 4% higher odds ratio for deaths from all causes [an actual difference in deaths of 11.1% versus 10.8%]. No statistically tenable difference.
This would have just been a major finding of additional support of an ineffective drug in improving clinical outcomes, but it didn’t end there. Unexpectedly, a total of 11.1% of the patients on Vytorin were diagnosed with cancer during those five years compared to 7.5% of the placebo group. And this is the source of the media frenzy. The cancer cases were distributed among diverse types of cancers [the largest numbers of cases were for skin and prostate cancers] but were unrelated to the degree of LDL-cholesterol lowering, according to the authors. That, in addition to the cancers’ fast appearances lend additional support for the authors’ conclusions that “the observed difference in cancer rates in the study may have been the result of chance, but this possibility requires further study.”
The authors were quick to downplay cancer risks, referring to “an independent analysis of cancer data from three trials of ezetimibe,” published in the same issue of NEJM alongside the SEAS trial results. This analysis, led by Richard Peto, of the Clinical Trial Service Unit at Oxford University, UK, was written by the principal investigators of the SHARP and IMPROVE-IT trials. They said they were “doing so independently of the study funders, Merck and Schering-Plough.” The statistician among the authors is also a paid statistician with the IMPROVE-IT trial and another author, Robert Califf, M.D., disclosed additional consulting fees from Merck and Schering-Plough which “are donated to his university and to nonprofit charities.”
Backlash from this paper have not just been because of the appearances of non-neutral authors, but because of the data they selectively chose to use. The authors didn’t examine cancer incidences from large completed trials, despite the fact the stated goal of their analysis was to test the hypothesis of an increased incidence of cancer. [The ENHANCE trial never published the data on the incidences of cancer among that study’s participants, by the way.] Instead, they chose only data from the completed SEAS trial data and compared it to interim data from their own two trials still in progress: the Study of Heart and Renal Protection (SHARP) trial that isn’t estimated to be completed until July 2010; and the Improved Reduction of Outcomes: Vytorin Efficacy International Trial (IMPROVE-IT) which is still recruiting participants and isn’t estimated to be completed until June 2012. The authors said that they got permission from the safety monitoring committees to take the unusual step to unblind the interim data on cancer for this analysis.
While their two studies involved larger population groups [a total of 20,617 people], and could potentially detect adverse effects more readily, the pooled data was on study participants who also haven’t been followed for nearly as long [an average of 1 year for the IMPROVE-IT participants and 2.7 years from the SHARP participants] and cancers may take longer than that to appear and give a full picture of the risk odds among a full cohort. The authors pointed out that their analysis showed no evidence of a trend in the risk of cancer with increasing duration of follow-up in their trials or when the three trials were combined, or difference in incidences of cancer between treatment and control groups. It should be noted, however, that, unlike the SEAS trial, the SHARP and IMPROVE-IT trials are comparing ezetimibe and simvastatin to simvastatin alone, rather than to a placebo, which could also mask any potential heightened links of the statin, itself, to cancer.
So, while the Peto et.al. paper concluded “the available results from these three trials do not provide credible evidence of any adverse effects of ezetimibe on rates of cancer,” and said that a link to cancer is an “implausible hypothesis,” because of the appearances of conflicts,it wasn’t likely viewed by many as altogether reassuring evidence.
The SEAS trial authors acknowledged that the safety of ezetimibe hasn’t been studied extensively, but to reassure patients, they said statins themselves haven’t been associated with an increased risk for cancer. To support that statement, they referenced a 2005 meta-analysis that had been conducted by the Cholesterol Treatment Trialists’ Collaborators [a collaboration of the same CTSU at Oxford and statin investigators at the University of Sydney, Australia] “of data from 14 statin trials involving approximately 90,000 patients [that] showed no evidence of an increased incidence of or death from cancer.” The CTSU authors of that meta-analysis had also reported that with an average of five years of follow-up, “there was no evidence of an excess incidence of cancer emerging with increasing duration of treatment...no apparent excesses among any particular site-specific cancer.” Their data revealed that with 5+ years of therapy, there was only a 5% increased relative risk for cancer incidence per mmol/L cholesterol reduction by year.
By lumping together data from 14 varied studies, this meta-analysis didn’t address specific populations that some studies have appeared to suggest might be at slightly higher risks, such as elderly and women.
The PROSPER trial of 5,804 seniors (70-82 years of age), for example, found a 25% higher incidence of cancers among those on the statin pravastatin compared to the placebo group over the three years. In fact, for decades, it’s been observed that seniors with higher cholesterol levels over the long term live longer and healthier lives, which has fueled concerns of statins used for primary prevention. The 1997 Weverling-Rijnsburger AW study of 724 advanced-age seniors, for example, reported that during ten years of follow-up, mortality from cancer and infection was significantly lower among the participants with the highest total cholesterols, mostly explained by lower all-cause mortality. “Each 1 mmol/L increase in total cholesterol corresponded to a 15% decrease in mortality,” it reported. As Dr. Beatrice A. Golomb, M.D., Ph.D., a geriatric specialist, wrote of the link between statins and cancer in elderly populations seen in the PROSPER trial, “the significant increase in cancer cannot be dismissed as necessarily a fluke.”
Cardiologists at Tufts University School of Medicine in Boston, attempted to assess the relationship between lowering LDL-cholesterol with statins and the effects on elevated liver enzymes, rhabdomyolysis, and cancer. They analyzed 23 statin treatment arms in clinical trials, with 309,506 person-years of follow-up, and published their findings last summer in the Journal of the American College of Cardiology. Their results were considerably different than the 2005 CTSU analysis. They found “no significant relationship between percent LDL-C lowering and rates of elevated liver enzymes or rhabdomyolysis. But their analysis demonstrated a significant inverse association between cancer incidence and achieved LDL-C levels (R2 = 0.43, p=0.009).” Concerning the risk-benefit analysis, they concluded that the cardiovascular benefits of lowering LDL-cholesterols “may in part be offset by an increased risk of cancer.”
With growing lipid researchers suspecting that statins offer cardiovascular benefit, not by lowering cholesterol, but by other mechanisms such as anti-inflammatory and/or anti-clotting effects that might be achieved by aspirin and safer interventions, the possible cancer link becomes more relevant.
Dr. Westby G. Fisher, M.D., FACC is a board certified internist, cardiologist, and cardiac electrophysiologist and is Associate Professor of Medicine at the Feinberg School of Medicine, Northwestern University, Evanston, IL, is another doctor not so ready to dismiss cancer concerns without additional research. He pointed out that “the CARE trial demonstrated a larger incidence of breast cancer diagnosis in the treatment group as compared to the placebo group.”
A few weeks ago, he said he didn’t think statins had been convincingly shown to induce cancers but wondered if “the combination of simvastatin and ezetimibe might have an unexpected synergistic cancer-promoting effect.” But, he wrote, “we really don't know.” Some studies have suggested a possible mechanism for a link, such as animal studies showing that lowering of serum cholesterol by the unabsorbed bile acid–binding resin cholestyramine promotes mammary gland carcinogenesis. As he concluded:
So while I'm willing to accept the premise that the relationship of carcinogenesis is unknown with the simvastatin/ezetimibe combination, given the prior art of research available, I'm not sure I'm willing to write off the potential for a relationship as "implausible."
Even Dr. Timothy Gardner, president of the American Heart Association, was cautious. He was quoted in Reuters as telling reporters: “We need to be very careful and watch this. We are not able to definitively say that there is no risk here with this drug.”
To try and sort out the swirling controversies and uncertainties, the NEJM wrote an editorial in this issue. Dr. Jeffrey M. Drazen, M.D., and colleagues wrote:
The randomized clinical trial is considered to be the most reliable tool to assess the efficacy and safety of new drugs. At times, however, randomized trials detect adverse events that are unanticipated and not easily explained on the basis of current knowledge. An unexpected finding of this kind may ultimately prove to be due to chance, but follow-up studies sometimes confirm the adverse drug effect. Particularly when an unexpected finding raises a safety concern with regard to a drug, physicians face uncertainty about how to act on the information...
They went on to review the SEAS trial and the Peto et.al. analysis. But some have also raised alerts that by looking only at cancer deaths during such short studies could under-reveal the actual cases of cancers diagnosed among patients, most of whom might not die until after the study’s short follow-up period. Even so, they went on to add their own cautions:
It is important to note that none of the three trials were designed primarily to address cancer risk. However, cancer mortality is an end point that would be expected to be reliable. When the cancer mortality data from the SEAS, SHARP, and IMPROVE-IT trials were combined, there was an increase in cancer mortality risk in the combined ezetimibe groups (134 deaths, as compared with 92 deaths in controls; risk ratio, 1.45; 95% CI, 1.02 to 2.05; uncorrected P=0.007). Because this P value was obtained from one of several data-driven analyses rather than from a test of a single prespecified hypothesis, it should be interpreted cautiously. The Oxford researchers believe that this finding is due entirely to the play of chance rather than to a true increase in cancer mortality. They argue that an increase in the risk of cancer death, if real, should be paralleled by an increase in the risk of cancer incidence, which was not found in the combined analysis, and that there is no plausible mechanism for such an effect.
Although the Oxford group may ultimately prove to be correct, it is appropriate to raise a note of caution. Whether the increased mortality risk is due solely to the play of chance is uncertain. Ezetimibe interferes with the gastrointestinal absorption not only of cholesterol, but also of other molecular entities that could conceivably affect the growth of cancer cells. The fact that the combined data from all three trials showed an increase in cancer mortality with ezetimibe should not be assumed to be a chance finding until further data are in. It is appropriate that SHARP and IMPROVE-IT continue. Careful follow-up of the patients in these trials will be essential, and other existing data sets on ezetimibe-treated patients should be analyzed for cancer end points...
Physicians and patients are unfortunately left for now with uncertainty about the efficacy and safety of the drug.
“A broken system”
Coincidentally, today’s issue of the Journal of the American Medical Association featured an article that expressed the frustrations of many medical professionals about controversies like this — where industry-funded research seems to portray positive news on drugs and treatments, that appear to be problematic or incomplete or their conclusions are inconsistent with other sources, leaving medical professionals without information they feel they can trust as objective sound guidance.
This article may help consumers better understand one reason why their doctors can’t always give hard and fast answers and sometimes have to admit: “We just don’t know.”
Dr. Marcia Angell, M.D., former editor-in-chief of the New England Journal of Medicine and currently at the Department of Global Health and Social Medicine at Harvard Medical School, wrote that “over the past two decades, the pharmaceutical industry has gained unprecedented control over the evaluation of its own products.” She said that “drug companies now finance most clinical research on prescription drugs, and there is mounting evidence that they often skew the research they sponsor to make their drugs look better and safer.” She went on to write:
Two recent articles underscore the problem: one showed that many publications concerning Merck’s rofecoxib that were attributed primarily or solely to academic investigators were actually written by Merck employees or medical publishing companies hired by Merck; the other showed that the company manipulated the data analysis in 2 clinical trials to minimize the increased mortality associated with rofecoxib. Bias in the way industry sponsored research is conducted and reported is not unusual and by no means limited to Merck.
The problem is not so much the sponsorship itself but the terms. Before the 1980s, industry grants to academic institutions to fund studies by faculty members gave investigators total responsibility. The investigator designed the studies, analyzed and interpreted the data, wrote the papers, and decided where and how to report the results. Generally, neither the investigators nor their institutions had other financial connections to sponsoring companies. In recent years, however, sponsoring companies have become intimately involved in all aspects of research on their products. They often design the studies; perform the analysis; write the papers; and decide whether, when, and in what form to publish the results. In some multicenter trials, author may not even have access to all their own data.
She went on to outline the various financial ties of researchers to sponsors of research, including being paid consultants, speaker’s bureaus, and advisory boards. She overlooked pharmaceutical and industry funding that comes through nonprofit foundations, but her concerns of bias, regardless of the source, was an important message. Conflicts of interest can bias more than research, she wrote, to influence clinical practice guidelines issued by professional and governmental bodies. She repeated the well-known conflicts in the cholesterol guidelines and the National Cholesterol Education Program, for example, which had been developed by a panel where 8 out of 9 members had financial ties to statin makers.
“Physicians who would be quite skeptical about drug company advertisements and the pitch of sales representatives,” she said, “tend to trust the peer-reviewed medical literature. One result of the bias in this literature is that physicians learn to practice a very drug-intensive style of medicine.” But she also lambasted academic investigators and their institutions who look the other way about these conflicts, because “it is convenient and profitable for them to do so.” She said they should, instead, be at the forefront of efforts to clean up the system of clinical research and not leave it to the government and lawyers. “It is more than a matter of perception or appearances: it is a matter of public health.”
For now, she said:
Looking at this picture altogether, it would be naive to conclude that bias is only a matter of a few isolated instances. It permeates the entire system. Physicians can no longer rely on the medical literature for valid and reliable information. This is the conclusion I reluctantly reached toward the end of my 2 decades as an editor of the New England Journal of Medicine, and it has been reinforced in subsequent years. Clinicians just do not know anymore how safe and effective prescription drugs really are, but these products are probably nowhere near as good as the published literature indicates.
© 2008 Sandy Szwarc