Median progression-free survival (PFS) based on traditional RECIST criteria did not correlate with overall survival (OS) in a meta-analysis of 12 randomized controlled trials of nivolumab or pembrolizumab monotherapy.
There was no correlation in terms of medians or gains in medians, although hazard ratios for OS and PFS did correlate significantly, said Bishal Gyawali, MD, PhD, of Nagoya (Japan) University Hospital with his associates. “The protective effects of treatment were greater for OS than for PFS,” they concluded in JAMA Network Open. “Progression-free survival cannot adequately capture the benefit of PD-1 inhibitors; thus, OS should remain the gold standard end point for trials of PD-1 inhibitors.”
Progression-free survival often has been used as a surrogate for OS because the latter takes time to ascertain and can be contaminated by crossover or postprogression treatment. However, it can be problematic to assume that the two outcomes correlate. Progression “is defined as an increase in tumor size beyond an arbitrary cutoff and is prone to bias, particularly when the investigators are not blinded,” the researchers noted. Furthermore, PD-1 inhibitors show an “atypical response pattern,” including long durations of response, responses after initial progression (known as pseudoprogression), and even response after treatment cessation.
The analysis, the first to formally compare PFS and OS across PD-1 inhibitors, included 10 randomized, controlled trials comparing nivolumab or pembrolizumab with nonimmunotherapy in adults with solid tumors. Two additional trials evaluated pembrolizumab or nivolumab following treatment with ipilimumab. In all, the studies included 5,417 patients. There was no significant heterogeneity among studies, the researchers said.
Median PFS and median OS correlated poorly, with an R2 value of 0.46 (P = .09). Change in PFS also did not correlate with change in OS (R2 = 0.23; P = .28). In contrast, hazard ratios for PFS and OS correlated significantly (R2 = 0.41; P = .048). The protective effects of treatment were higher for OS than for PFS (pooled HR, 1.2; 95% confidence interval, 1.1-1.3; P = .002).
This might be because traditional RECIST (response evaluation criteria in solid tumors) criteria predate the era of immunotherapy and do not accurately capture disease progression when patients are on immuno-oncologics. For example, pseudoprogression (in which T-cell infiltrates cause the tumor to grow before it shrinks) could be misconstrued as progression. Also, PD-1 inhibitors can continue working even after treatment cessation, which could affect OS more than PFS, the researchers noted.
Regardless, “PD-1 inhibitors may have larger effects on OS than on PFS, which would be unprecedented in oncology therapeutics,” they concluded. “These results support the rationale of using OS as the primary end point of future phase 3 trials of PD-1 inhibitors and discourage the use of PFS as a sole primary end point as the latter may provide misleading information about the efficacy of these drugs.”
Funders included the Laura and John Arnold Foundation, the Harvard Program in Therapeutic Science, and the Engelberg Foundation. The investigators reported having no relevant conflicts of interest. One coinvestigator reported research support from the Laura and John Arnold Foundation. The other investigators had no conflicts.
SOURCE: Gyawali B et al. JAMA Network Open. 2018 June 22. doi: 10.1001/jamanetworkopen.2018.0416.
