Preliminary Observations From Preliminary Trial Results
Have We Finally Had Enough?
Results from the Simvastatin Ezetimibe Aortic Stenosis study (SEAS, registration No. NCT00092677) were prematurely released in a surprise announcement on July 21, 2008. In a carefully scripted media event, including a webcast from the United Kingdom, surprise findings of increased incidence of cancer and cancer death in patients treated with simvastatin-ezetimibe were disclosed and immediately dismissed as “not credible.” To refute the observation of excess cancer events, the SEAS investigators offered a well-intentioned but hastily performed pooled analysis using data derived from 2 prematurely unblinded, ongoing ezetimibe outcomes trials (Study of Heart and Renal Protection [SHARP, registration No. NCT00125593] and IMProved Reduction of Outcomes: Vytorin Efficacy International Trial [IMPROVE-IT, registration No. NCT00202878]). Unrelated to the cancer findings, primary and secondary outcome data were also prematurely disclosed. The purpose of this event was no doubt to dispel cancer concerns for a “blockbuster” drug (ezetimibe) that had already been subjected to considerable criticism and scrutiny.1,2 However, all of this occurred without any peer review of the data, which denied the medical community the opportunity to fully evaluate the benefits and risks of this drug in the usual scientific fashion.
SEAS was a 4-year, prospective, randomized trial that compared simvastatin-ezetimibe with placebo in 1873 adults with mild to moderate aortic stenosis. The primary end point of the SEAS study was “major cardiovascular events,” a composite of events associated with aortic valve and atherosclerotic diseases. The secondary end points were the 2 separate components of the primary end point: “aortic valve disease events” (surgical valve replacement, hospitalization because of heart failure, and cardiovascular death), and “atherosclerotic disease events” (nonfatal myocardial infarction, coronary artery bypass surgery or percutaneous coronary intervention, hospitalization for unstable angina pectoris, nonhemorrhagic stroke, and cardiovascular death). In their haste to report these results at a press conference before peer review, the SEAS investigators reported the pooled cancer-incidence analysis differently than both the original SEAS investigators and a subsequently pooled analysis. We hope that these discrepancies will be resolved during the publication process. In advance of actual publication, 8 points of emphasis emerge with regard to the trial, its results, and the manner of the data release.
1. Does Ezetimibe Cause Cancer?
Although there was no a priori expectation of an increased cancer risk from ezetimibe, the SEAS study revealed troubling cancer signals including, among simvastatin-ezetimibe–treated patients, higher incidences of cancer (93 [9.9%] versus 65 [7.0%] patients in the placebo arm; unadjusted P=0.03) and cancer death (39 [4.1%] versus 23 [2.5%]; unadjusted P=0.05). Then, an analysis was reported by the Oxford University investigators, who are directing the ongoing SHARP trial, of cancer data from the ongoing SHARP and IMPROVE-IT trials, which were prematurely unblinded to enable their analysis. Of note, the Oxford group reported a different number of cancer events for the SEAS trial, namely, 102 versus 67 patients, P=0.006 (Fisher exact t test). The participants in the press conference also reported that the pooled analysis of the 3 trials did not “confirm” an increase in cancer incidence. However, an increase in cancer death was identified in the 2 prematurely unblinded trials (97 versus 72 patients for simvastatin-ezetimibe versus placebo), with borderline statistical significance of P=0.053 using a 2-sided Pearson χ2 test. Combining the 3 trials, the increase in cancer death was highly significant (odds ratio 1.43, 95% confidence interval 1.1 to 1.86, P=0.009). According to the Oxford statistical review, no particular type of cancer was identified, nor did exposure duration factor into the analysis. Although worrisome, the overall trial conclusion was that ezetimibe was “generally well tolerated and safe.”
Pooling data, including ongoing trial data, is not without hazard, owing to variations in exposure duration, study design, and population. Final determination of the strength of the cancer association will not be possible until individual trials are completed and a detailed, adjusted meta-analysis is conducted using time-to-event analysis. Accordingly, the observation of a cancer adverse-event signal must be considered hypothesis generating. These findings may be a statistically spurious association. Prior concerns about statins (including simvastatin) and cancer were subsequently dispelled.3,4 However, the chemical compound studied in SEAS is distinct from statins, and a linkage between ezetimibe and cancer has some biological plausibility. Ezetimibe inhibits absorption of both cholesterol and phytosterols.5 These other sterols, which are found in fruits and vegetables, have been linked to protection from cancer via inhibition of cancer cell proliferation and other mechanisms.6 Thus, judged through this potential biological link, the clinical phenomenon bears close monitoring through ongoing surveillance by the US Food and Drug Administration and data safety monitoring board activity. Furthermore, patients need to be informed about this potential safety concern.
2. Was It Appropriate to Unblind the Ongoing IMPROVE-IT and SHARP Trials to Further Explore the Cancer “Signal” in SEAS?
This was a debatable decision, yet the answer is probably“no.” Unexpected findings are common in completed trials and should be reported along with appropriate cautionary statements when they involve findings not hypothesized or prespecified. Ongoing trials are not a reliable source of additional information, because they represent incomplete experiments that may show transient findings at preliminary study stages that are not evident at study completion. The premature unblinding of 2 ongoing trials appears to be driven by the commercial needs of a sponsor under siege over a potentially ineffective or unsafe product, not good scientific discipline. The preliminary findings reported for unblinded trials may affect physician or patient behaviors and thereby introduce important and irretrievable biases in the ongoing studies. Accordingly, unblinding an ongoing trial has grave consequences and should not be performed for commercial reasons. The investigators of SHARP and IMPROVE-IT should have resisted the pressure to partially unblind their trials. If a decision to unblind was undertaken, the analysis should have been performed by an independent academic group with no stake in the outcome, and not the SHARP trial investigators.
3. Was the Finding of No Benefit for Primary and Secondary Aortic Stenosis End Points Surprising?
Aortic stenosis–related events were not prevented by ezetimibe-simvastatin, which resulted in no benefit for the overall primary end point of the trial. This result is not surprising given the failure of statins in similarly designed trials.7 However, SEAS tested a novel hypothesis with regard to the effects of other sterols on the progression of aortic stenosis. Because plant sterols have been recovered from stenotic aortic valves,8 SEAS investigators hypothesized that inhibition of sterol absorption by ezetimibe would produce more favorable effects for this compound compared with prior statin trials. Despite the pathological similarities between aortic valve stenosis and atherosclerosis, it is now clear that neither therapy for lowering low-density lipoproteins nor inhibition of phytosterol absorption can reduce the progression of aortic stenosis.
4. What Is the Appropriate Interpretation of the Secondary Outcome Findings on Ischemic Cardiovascular Events?
Ischemic cardiovascular events were reduced by combination treatment with simvastatin-ezetimibe. Composite ischemic events were reduced from 20.1% to 15.7%, which corresponds to a relative risk reduction of 22% (95% confidence interval 3% to 37%, P=0.02). This finding is not surprising in a trial that studied an older, high-risk population and included the use of a statin in the active treatment group. Details on the components within this composite end point are crucial toward understanding the effect of the intervention. Potentially troubling is the report that, although specific component events (eg, nonfatal myocardial infarction) have not been revealed in detail, the significant benefit in the secondary end point of major ischemic events was reportedly driven by a reduction in the rate of a relatively soft end point, namely, coronary bypass surgery, and that most of these surgeries were coincidental to aortic valve surgery. This is an area that requires close scrutiny within the data submitted for peer review.
Unfortunately, the design of SEAS does not permit examination of ischemic end points to answer the most critical question, which is whether ezetimibe provides any incremental value beyond that provided by a statin alone. Unfortunately, IMPROVE-IT is the only ongoing clinical trial that will provide the answer to this pivotal question, and it will not be completed until 2012 at the earliest. In the interim, cross-trial comparisons with statin monotherapy studies are the only approach that may yield clues to the potential additive value of ezetimibe. Such approaches are inherently weak. SEAS did reveal a large degree of low-density–lipoprotein cholesterol reduction (61%) in the active treatment group compared with placebo. Nevertheless, the event reduction observed in SEAS is not impressive compared with simvastatin monotherapy outcomes trials conducted in other high-risk populations (Scandinavian Simvastatin Survival Study9 and Heart Protection Study10), which showed relative risk reductions of 34% and 24%, respectively, despite substantially less low-density–lipoprotein cholesterol reduction than observed in SEAS. Noting the difficulties and hazards in cross-trial comparisons, it does focus attention squarely on the need for ezetimibe to demonstrate an incremental benefit of health outcomes. It should be recognized that safety concerns require greater attention when observed for therapies without a proven health-outcomes benefit. Put another way, we can tolerate more uncertainty about safety when there is well-documented evidence of efficacy in reducing serious adverse outcomes.
5. Does the Design of SEAS Move the Field Forward?
SEAS, begun in 2001, studied a reasonable hypothesis about the effects of lipid-lowering therapy on aortic stenosis. However, the design of this study was disappointing because of the absence of a simvastatin control arm. Accordingly, any observed benefit could be attributed to either component of the ezetimibe-simvastatin combination product. More disappointing is the fact that nearly all of the early trials of ezetimibe failed to examine the critical question of whether this agent adds any incremental value beyond that provided by simvastatin alone. In many ways, the absence of these critical data represent misjudgment by the US Food and Drug Administration, which approved ezetimibe on the basis of the ability of this drug to reduce low-density–lipoprotein cholesterol by a modest 15% to 18%. If the ezetimibe approval had been coupled with a mandate to perform appropriate outcome trials, we would not find ourselves in the current dilemma of questioning whether ezetimibe is effective or safe. At best, we will now learn, approximately a decade after introduction of this drug, whether ezetimibe offers any incremental benefits beyond statin monotherapy. For an agent taken by millions of patients, that’s far too long an interval. Consideration of control by the US Food and Drug Administration over biased clinical trial designs leading to de facto positive studies is needed to maximize the value of human volunteer participation and to inform the treated public with meaningful research results.
6. Is the Media the Most Appropriate Venue for Initial Presentation of Scientific Results?
Although the investigators are to be applauded for promptly bringing to public awareness the potential adverse-event signal (no doubt, a lesson learned from other recent delays in public disclosure of trial results), the answer to this question is most certainly “no”. A simple survey of the vastly divergent headlines stemming from the media event, ranging from “Trial Intensifies Concerns About Safety of Vytorin” to “Vytorin Misses Primary End Point in SEAS Study,” illustrates the potential for confusion. Venues exist for rapid, peer-reviewed dissemination of clinical trial results, and these venues should be preferred over carefully controlled and scripted media sessions and press releases. Many of the top-tier medical journals offer “expedited review,” which enables publication within a few weeks after generation of the manuscript. If an “emergency press conference” were required to protect public health, it should not have been orchestrated by the sponsor, and it should definitely not have included other clinical trial results (eg, primary and secondary end point data from the study) beyond those necessary for immediate public disclosure to address safety concerns.
7. Is the Flow of Information Determined by the Needs of the Patients or the Needs of the Financial Markets?
The timing of the release was closely orchestrated to coincide with reports of corporate earnings to address an apparent concern that the news would induce volatility within the financial markets. This process accentuates concerns that the purpose of the premature release of data, and the precise findings that were disclosed (ie, primary and secondary outcome data along with adverse-event data) were directed at financial markets rather than the sanctity of the trial and the traditional process of external peer review. The early release of data in this manner compromised the ability of professional associations to accurately comment on these preliminary trial results. Medical journals, which are ultimately responsible for peer review of scientific findings, should reconsider their approach to subsequent publication of such prematurely released data. Coupling this release to reports of company earnings leaves the scientific community with a sense of unease.
8. What Happened to the Role of the Regulators?
Given the unexpected nature and sensitivity of a possible cancer risk finding, and its large impact on millions of patients that have been prescribed ezetimibe, the exclusion of the US Food and Drug Administration from the review, release, and commentary of the data seems questionable. No doubt these data will be carefully scrutinized by regulators in an ongoing evaluation of ezetimibe. On the basis of the current knowledge, physician investigators must also consider altering the research-consent process for ongoing trials of ezetimibe. Moreover, these findings provide more impetus to reconsider the drug-approval process for novel, first-in-class compounds based on “surrogate end points.” Such accelerated approval often occurs before the balance of safety and efficacy is fully established. Unfortunately, we practice in an era where direct-to-consumer marketing of pharmaceuticals can drive massive overuse of agents before acquisition of robust outcome data for both efficacy and safety.11
The manner of release of the SEAS trial and the unblinding of 2 other ongoing trials raises important questions about the scientific process in an era of mass media communications. Premature release of data via a highly orchestrated media event does not serve the scientific community well. Premature unblinding of ongoing trials for commercial purposes is highly undesirable and compromises the integrity of these studies. Finally, approval of a first-in-class drug on the basis of biochemical surrogates is fraught with hazard. The ultimate price may be widespread utilization of agents that lack a favorable balance between safety and efficacy.
Dr Taylor receives research funding (without salary support) and speaker honoraria from Abbott Laboratories on the topic of HDL cholesterol and the use of niacin.
Dr Nissen reports that Cleveland Clinic Coordinating Center for Clinical Research has received research support to perform clinical trials from Pfizer, AstraZeneca, Novartis, Sankyo, Takeda, Sanofi-Aventis, and Eli Lilly. Dr Nissen does not receive personal remuneration for conducting clinical trials. Dr Nissen consults for many pharmaceutical companies, but requires them to donate all honoraria or consulting fees directly to charity so that he receives neither income nor a tax deduction.
The opinions expressed in this article are not necessarily those of the editors or of the American Heart Association.
The opinions or assertions herein are the private views of the authors and are not to be construed as reflecting the views of the Department of the Army or the Department of Defense.
Kastelein JJ, Akdim F, Stroes ES, Zwinderman AH, Bots ML, Stalenhoef AF, Visseren FL, Sijbrands EJ, Trip MD, Stein EA, Gaudet D, Duivenvoorden R, Veltri EP, Marais AD, de Groot E. Simvastatin with or without ezetimibe in familial hypercholesterolemia. N Engl J Med. 2008; 358: 1431–1443.
Baigent C, Keech A, Kearney PM, Blackwell L, Buck G, Pollicino C, Kirby A, Sourjina T, Peto R, Collins R, Simes R. Efficacy and safety of cholesterol-lowering treatment: prospective meta-analysis of data from 90 056 participants in 14 randomised trials of statins. Lancet. 2005; 366: 1267–1278.
Helske S, Miettinen T, Gylling H, Mäyranpää M, Lommi J, Turto H, Werkkala K, Kuupari M, Kovanen PT. Accumulation of cholesterol precursors and plant sterols in human stenotic aortic valves. J Lipid Res. 2008; 49: 1511–1518.