Combination therapies and the theoretical limits of evidence-based medicine

Abstract

Background: Advances in molecular pharmacology and surgical, endovascular, and radiation techniques have yielded multiple effective or promising, and potentially complementary, classes of treatments for virtually every major medical disorder. Consequently, determining the optimum combination of therapies for a condition is a burgeoning challenge to clinical trialists and practicing physicians.

Methods: General phase III trial strategies for testing combination regimens are described, and then applied to two illustrative conditions, Alzheimer disease and ischemic stroke.

Results: Strategies for testing combination regimens include: head to head trials of all combinations, which lead to unwieldy trial numbers; very large multi-arm trials, which impractically delay interval information on regimen utility; and hierarchical, serial clinical trials. Systematic literature review revealed seven classes of agents already approved or in late phase III testing for preventing the development or slowing the progression of Alzheimer disease and five for ischemic stroke prevention. Possible combination regimens number 128 (2(7)) for Alzheimer disease and 32 (2(5)) for ischemic stroke. Hierarchical, serial clinical trials would permit identification of the optimum combination of these agent classes for Alzheimer disease through 127 trials, enrolling 63,500 patients, requiring 286 years; for ischemic stroke through 31 trials, enrolling 186,000 patients, requiring 155 years.

Conclusions: Marked limitations in the ability of clinical trials to interrogate varied treatment combinations to determine the most effective ensemble exist, and their scope is widely underappreciated. Steps that may attenuate, though not eliminate, the challenge of a surfeit of combination treatment regimens include preclinical testing to identify the most promising regimens, use of surrogate outcome measures in exploratory clinical trials, and use of hierarchical, serial and factorial phase III trials.