News Headline: Lung Cancer Trials Soften Endpoint
Outlet Full Name: Med Page Today
Author: Charles Bankhead
Use of overall survival as the primary endpoint has decreased significantly in clinical trials of advanced lung cancer, as has the magnitude of clinically relevant improvement, a review of 200 studies showed.
Before 2000, virtually all phase III trials in advanced non-small cell lung cancer (NSCLC) had survival as the primary endpoint. But from 2001 to 2010, 19% of phase III trials had primary endpoints other than overall survival (P<0.001 versus prior decades), most often progression-free survival (13%), Natasha B. Leighl, MD, of the University of Toronto, and co-authors reported online in the Journal of Clinical Oncology. From 1990 to 2010, the proportion of trials that had progression-free survival (PFS) as the primary endpoint increased from 0% to 13%, they noted. The proportion of trials meeting the specified primary endpoint remained stable, but the percentage of trials reporting positive results increased. The magnitude of the survival benefit decreased over time, but the change did not achieve statistical significance, the authors wrote. "These findings raise questions regarding the design and interpretation of phase III trials in advanced non-small cell lung cancer," they concluded. "The intention of multiphase clinical trials is to focus resources on promising agents, with each phase eliminating either toxic or minimally active agents. "Our results suggest that we are progressively less efficient in accomplishing this task in advanced NSCLC trials." Clinical trials in advanced NSCLC have produced first- and second-line chemotherapy and targeted agents for use in selected patients after failure of chemotherapy. However, the benefits of the treatments have been modest, the authors noted. Growing interest in NSCLC within the oncology community and pharmaceutical industry has been accompanied by changes in trial design and interpretation, they continued. The changes have encompassed selection of primary and secondary endpoints, statistical power, and sample size. Use of PFS as a primary endpoint has been advocated because of the growing number of patients who cross over to a different therapy after disease progression and because of the increased number of therapeutic options. Although the FDA has accepted PFS as a potential primary endpoint for drug approval in advanced NSCLC, the agency expressed reservations in doing so. Opponents of PFS as a primary endpoint have questioned the potential for subjectivity in evaluation of the endpoint and its relationship to clinically relevant outcomes. PFS and other emerging endpoints have sparked debates about the definition of clinically meaningful benefit. "Apart from changes in the design of clinical trials, the manner in which these trials are interpreted may be changing as well," the authors said. "An increasing willingness to interpret the results of clinical trials as positive despite extremely modest or questionable benefit of a new treatment has been previously reported." To inform the discussion, Leighl and colleagues performed a literature review to examine changes in the design and interpretation of phase III trials in advanced NSCLC over the past 3 decades. Initially, the authors identified 248 trials published from 1980 to 2010. After exclusion of trials that did not meet criteria of systemic therapy for advanced NSCLC, the analysis comprised 203 unique studies: 32 conducted in the 1980s, 53 in the 1990s, and 118 in the most recent decade. Trial size increased from a median of 152 patients during the first decade of the review to 413 in trials conducted from 2001 to 2010 (P<0.001). Three-drug combinations predominated in the 1980s and doublets in the 1990s, whereas evaluation of targeted therapy emerged in the last decade. The authors assessed change in trial interpretation by evaluating statistical significance and the characterization of a statistically significant outcome as positive or negative. The proportion of trials demonstrating statistically significant improvement remained stable at about 30% across the 3 decades. However, the percentage of trials characterized as positive -- despite not demonstrating a significant improvement in the primary endpoint -- increased from 30% in the first decade to 53% of trials reported from 2001 to 2010 (P<0.001). Trials were deemed positive because of improvements in secondary endpoints (N=24), asserting noninferiority in the absence of noninferiority trial design (N=26), or calling for additional studies on the basis of a numerical trend in the primary outcome (N=9). Finally, the authors examined the magnitude of benefit that met the definition of statistically significant improvement in survival. In the subgroup of 60 trials that reported significant survival improvement, the median benefit was 3.9 months during 1981 to 1990, 2.4 months during 1991 to 2000, and 2.5 months in the most recent decade. In an analysis of benefit that included all trials characterized as "positive," the survival benefit decreased from 3.9 months in the first decade to 2.0 months in the second and to 0.9 months in trials conducted from 2001 to 2010. "Our findings confirm a disquieting shift in phase III trial design in advanced NSCLC," the authors concluded. "Despite the advances in design through large size and statistical power over the past 3 decades, the bar for judging clinical benefit of new agents in lung cancer appears to be falling."