OS a reasonable primary endpoint when median survival post-progression is short

A pair of statisticians working at Texas A&M University and The University of Texas M.D. Anderson Cancer Center concluded that OS was a reasonable primary endpoint for trials involving metastatic cancer but only when median survival post-progression was less than six months.

When median survival post-progression went beyond 12 months, Kristine R. Broglio, PhD, and Donald A. Berry, PhD, said OS as a primary endpoint was “too high a hurdle.”

Broglio and Berry simulated clinical trials with equal numbers of patients randomly assigned to two arms — one with median PFS of six months and one with median PFS of nine months. They then calculated OS for each patient as the sum of PFS plus survival post-progression.

They found that, for a trial with 80% power for detecting PFS, the median P value for OS was .013 when median survival post-progression was two months. When median survival post-progression was 12 months, the median P value was .199 and .322 at 24 months. Broglio and Berry determined that as median survival post-progression increases, the power for detecting significant differences in OS decreases.

With a median survival post-progression of two months, a study designed with 80% power to detect the difference in PFS had a 70% probability of detecting significant difference in OS. Conversely, when median survival post-progression was 12 months, probability of detecting significant difference in OS dropped to 24%.

Broglio and Berry concluded that the significant benefit of PFS is likely to disappear with a moderate median survival post-progression and will probably disappear when the median survival post-progression is large.

Broglio KR. J Natl Cancer Inst. 2009;101:1642-1649.