In submitting Provenge vaccine for the treatment of advanced prostate cancer for fast-tracked FDA approval despite aborting one clinical trial due to failure to meet the primary endpoint, Dendreon relied on meta-analysis of combined data from two parts of a phase III study, D9901. In fact, after the trials failed to meet their primary endpoint, time to progression, they were analysed twice over. Dr. Eric Small presented meta-analyses of impact on overall survival and prostate cancer specific survival and Dr. Daniel Petrylak presented a further analysis or suvival focused on a subset of patients who, when they progressed on Provenge, took Taxotere.

Two oncologists specialiizing in prostate cancer, Howard I Scher M.D. and Maha Hussain, M.D., raised questions both during and after the FDA meeting about the robustness of survival data based on post-hoc analysis from a trial which in their view was neither designed nor powered to answer the question of the effect of Provenge on survival.

Now we have seen the same case made from virtually the opposite perspective (and in relation to safety, not efficacy). On June 6 the Oversight and Government Reform Committee, chaired by Henry Waxman, held a hearing into FDA’s role in evaluating the safety of the diabetes drug Avandia. During the hearing an officer of the major pharmaceutical company under the spotlight, GlaxoSmithKline, made arguments similar to those put by Dr. Scher, Dr. Hussain, and others who are not yet satisfied that Provenge works.

But in this case roles were reversed. In the case of Avandia, Steven Nissen, M.D., Chairman, Department of Cardiovascular Medicine, Cleveland Clinic did a meta-analysis. Bruce M. Psaty, M.D., Ph.D., Professor of Medicine, Epidemiology and Health Services; Co-director, Cardiovascular Health Research Unit, University of Washington, supported Nissen’s results at the hearing as did John B. Buse, M.D., Ph.D., Professor of Medicine and Chief, Division of Endocrinology, University of North Carolina School of Medicine, who alleged that threats had been made against him by a company executive (see New York Times coverage with links to NEJM editorials:
http://tinyurl.com/387wkc; http://tinyurl.com/3dzytz).

One of Glaxo’s Congressional supporters, Rep. Issa (San Diego, Rep) ridiculed Dr. Nissen’s analysis (he said it “looks like an anecdotal concoction” intended to “go after a company”).

Moncef Slaoui, Ph.D., Chairman, Research and Development, GlaxoSmithKline said “metaanalyses are only as good as the studies that go into them.” He said the right way to study drugs for diabetes, “a very long term disease.” is “prospective large scale studies.” Metaanalyses “generate hypotheses,” he said, “they do not provide answers.”

Guess it all depends on whose ox is gored.

Waxman asked for “patient level data” and Moncef Slaoui said yes. This is what is needed for Provenge.