THROMBOLYTIC THERAPY CLINICAL TRIALS: TOTAL MORTALITY IS PRIMARY ENDPOINT; PATENCY RATES NECESSARY BUT NOT SUFFICIENT -- FDA CARDIO-RENAL ADVISORY CMTE.

The primary endpoint for studies of new thrombolytic therapies should still be overall mortality, FDA's Cardiovascular and Renal Drugs Advisory Committee concluded at its Dec. 16 meeting. Data packages for thrombolytic therapies also should include information on patency response, the committee decided, although those data alone are not a sufficient determinant alone of safety and efficacy. While mortality is the most important endpoint, new thrombolytic therapies for the treatment of acute myocardial infarction must do more than demonstrate mortality rates superior to placebo to be approvable, the committee concurred. "We already have several drugs and we know what they do, we know they save lives, we know what their safety profile is," committee member Salim Yusuf, National Heart, Lung & Blood Institute, commented. "How much worse can a new drug be and still be acceptable?" The committee met to provide FDA with guidelines on requirements for approval of thrombolytics. The group was asked to respond to a series of 23 questions from the agency and to review a proposed protocol for a 5,000-patient trial by Boehringer- Mannheim for an unnamed thrombolytic. The committee agreed that the Boehringer-Mannheim study would be insufficient to render a new thrombolytic drug approvable. The proposed study would try to establish indirectly through an active-control trial a reduction in mortality rates versus placebo for the thrombolytic drug. Boehringer-Mannheim researchers Daniel Odenheimer, MD, and Wolfgang Meyer-Sabellek, MD, proposed to use 30-day, all-cause mortality as the primary measure of efficacy in a positive- controlled Phase III thrombolytic trial enrolling 2,500 patients in each group. The trial would attempt to show the new agent to be superior to placebo by comparing the difference in mortality rates between the new agent and the control agent to the improvement in mortality rates found for the control agent from placebo in previous trials. The duo performed a meta-analysis of placebo-controlled trials of approved thrombolytics and found a consistent absolute mortality benefit of 2.6 plus or minus .25 percentage points, so that the lower 95% confidence limit of the benefit is 2.1 percentage points. Therefore, the duo maintained, in any non- placebo-controlled trial using one of the approved thrombolytics, the mortality rate that would have occurred with placebo could be assumed with 95% confidence to have been at least 2.1 percentage points higher than that seen with the approved thrombolytic. As long as the difference between the mortality rates of the new agent and the approved agent in such a trial is less than 2.1 percentage points (within a 95% confidence interval), Odenheimer and Meyer-Sabellek argued, the new agent may with 95% confidence be said to be superior to placebo. According to this model, Odenheimer said, 5,000 patients would give at least 80% power to demonstrate equivalency of the mortality endpoint within 2.1 percentage points, assuming an underlying mortality rate of 8%-10%. A study with 2,500 patients per group also would be sufficient, he argued, to "provide a precise estimate of intracranial hemorrhage and stroke rates." Therefore, the Boehringer-Mannheim team asserted: "If we perform such a study and equivalence is demonstrated and the safety profile is acceptable, we will have demonstrated that the new agent is in fact both safe and effective and hence will be approvable." The committee, however, voted unanimously that this approach alone would not be sufficient for approval. Commenting on the trial, NHLBI's Yusuf said the 5,000 patients proposed would not be enough to give a demonstration of reasonable equivalency between the new agent and the approved thrombolytic. "The new thrombolytic agent [could be] only half as effective as an existing thrombolytic agent and you [would be] willing to call them equivalent," he remarked. "That's a value judgment." He added: "You've used assumptions that all tend to lower your sample size. The estimates are all really imprecise, not 'precise.' I would suggest that we need to guarantee at least two-thirds or three- quarters of the efficacy of an existing drug" to deem it approvable. FDA asked the committee if there was some other endpoint or combination of endpoints with a higher incidence rate than mortality that might be used to evaluate efficacy. Equivalency to an approved thrombolytic could be demonstrated with a greater degree of certainty in a smaller trial if the endpoints of the trial occurred with greater frequency. The committee, however, voted that total mortality should always be considered as the primary basis for approval. In particular, the committee voted that patency response, while important, is not, at present, a sufficient surrogate for clinical benefit. It is, however, necessary to establish patency response as a function of dose for approval of a wholly new molecule, a modified version of an old molecule, and a new regimen of an old molecule, the committee unanimously voted. The committee declined to mandate when patency must be measured, but did specify that it should be measured more than once. Given that these surrogate endpoints are unacceptable, Office of Drug Evaluation I Director Robert Temple, MD, asked: "Is there some other package of material, combined with something like this [proposed Boehringer Mannheim study], that would be persuasive? Is there anything short of an unequivocal mortality study that would be satisfying?" "There is going to be some uncertainty," committee member Lloyd Fisher, PhD, University of Washington, said. "It's unrealistic to go for the same order of magnitude as you get from the pooled data [of earlier thrombolytic studies] if everything else looks good." There "has to be some sort of a trade-off between having a little more [flexibility in review of clinical trial data] and drugs that aren't quite as good and slowly inhibiting pharmaceutical development," Fisher continued, adding: "I would like to see the evaluation [of safety and efficacy] combined, and I would like to see a study planned that would look both at the mortality and the mortality plus the intracranial hemorrhage." Other committee members echoed Fisher's reluctance to separate the issues of efficacy and safety. The committee concluded that while mortality is the most important endpoint and a thrombolytic study more on the scale of 20,000 patients would be necessary to establish equivalent efficacy within a reasonable degree of certainty, the consideration of other endpoints showing clinical benefit could weigh heavily. Committee Chairman Craig Pratt, MD, Baylor College of Medicine, explained, "If other trials show there were other benefits [to the patient], I would be more lenient about the numbers [of patients and the degree of certainty]; if not, I would be tighter." The meeting concluded with votes by the committee that if the Boehringer-Mannheim trial were conducted as proposed and that trial or another trial also showed clinical superiority of the new thrombolytic over the control, then the combination would be sufficient for approval. If the other clinical endpoints showed only equivalence, and mortality equivalence was not demonstrated within an appropriate confidence interval, the committee voted it would not find the data sufficient for approval. Before considering the Boehringer-Mannheim clinical trial proposal, the committee first heard a series of nine presentations that provided members with background material and a framework on thrombolytic therapy areas of interest to FDA. These included: endpoints for clinical benefit, endpoints for dose ranging, populations for placebo-controlled trials, criteria for evaluation of intracranial events and factorial trials. In his presentation on positive-controlled clinical trials, Temple addressed the difficulties raised by the ethical impossibility of performing placebo-controlled trials for thrombolytics. "The principal problem is that in an equivalence trial you cannot know for sure what your control is doing because you don't have a placebo to tell you." He added: "You have to have belief that the control agent is reliably distinguishable from placebo." For thrombolytics, Temple noted, "that probably is a reasonable assumption." It is also important that equivalence trials be carried out "under conditions that match very closely to the conditions under which the control agent was effective," Temple said. "You need the same population, you need the same dose, you need the same endpoints [and] the timing of therapy with respect to the cardiac event should be as similar as possible. Once you start changing conditions the reliability is weakened," Temple said. FDA had presented the committee with 23 questions to be addressed at the meeting. The afternoon's general discussion covered only the first nine, omitting those dealing with stroke associated with thrombolytic therapy, factorial trials and the approval of antithrombotics. FDA's questions regarding stroke associated with thrombolytics asked if there should be "an upper bound to the upper confidence limit of the estimated stroke rate of a regimen employing a new molecule...a modified version of an old molecule...and a new regimen of an old molecule, above which the regimen could not be approved?"; how hemorrhagic stroke should fit into evaluations of thrombolytic regimens; and where the assessment of thrombotic stroke fits in. The agency also proposed to ask the committee whether it is reasonable, "when development of therapies for acute myocardial infarction will require trials involving large numbers of patients," to encourage investigators to utilize factorial trials, "so that guidance for concomitant therapy might be obtained at low marginal cost." In discussion earlier in the day, FDA Cardio-Renal Drug Division Director Raymond Lipicky, MD, emphasized that the perception that the agency discourages factorial trials was not true. FDA also raised the issue that there needs to be further guidance on what are the appropriate endpoints for antithrombotics and anti-coagulants. The agency's questions noted that enhanced patency response has not been demonstrated for these drugs, and the beneficial effects of anticoagulants have not been clearly defined. The meeting agenda asks the committee whether the safety evaluation with respect to intracranial hemorrhage must be as vigorous for these agents as for thrombolytics, given that dramatic effects have not yet emerged. The agency also intended to hear from the committee on what situations can be studied in placebo-controlled trials -- a new antithrombotic omitting aspirin; a new dose of aspirin; a new anticoagulant omitting heparin; a randomized switch to aspirin/heparin or placebo at some time interval after acute MI -- and in what situations and for which endpoints it would be sufficient for approval for a new agent to be equivalent to aspirin or heparin.