Analysis Validates Use of PFS As Surrogate Endpoint in Sarcoma Trials

An analysis of randomized controlled trials (RCTs) in soft-tissue sarcoma (STS) over 40 years found that progression-free survival (PFS) and response rate (RR) are reasonably well correlated with overall survival (OS), and are thus acceptable surrogates to use. Toxicity reporting in these trials is often less than optimal, however.

“In STS, trial design is particularly challenging, owing to the rarity and heterogeneity of this disease,” wrote study authors led by Alona Zer, MD, of Princess Margaret Cancer Centre in Toronto. They undertook an analysis of RCTs in STS that were published between 1974 and 2014, to determine if surrogate endpoints are clinically relevant and to investigate time trends in these trials.

In total, the study included 52 RCTs including 9,762 patients; 19 of these were published between 1974 and 1993, while the other 33 were published in the last 20 years of the study period. Most of the trials (45 of 52) investigated cytotoxic agents; some had three total patient groups, leaving 63 total comparisons in the analysis. The results were published online ahead of print in the Journal of Clinical Oncology.

OS was used as the primary endpoint in only two of the trials (4%). Twenty-three (44%) others used time-to-event primary endpoints; 3-month PFS and 6-month PFS was used in 5 of 22 phase II RCTs that had clearly defined endpoints. The use of PFS in general increased over time, with an odds ratio (OR) of 1.28 (95% CI, 1.09–1.50; P for trend = .002); this was mirrored by a reduction in the use of response rate (RR) over time, with an OR of 0.19 (95% CI, 0.07–0.53; P for trend = .001). Most of the studies (75%) used OS as a secondary endpoint.

In an analysis of surrogacy, the researchers found a highly significant correlation between OS and PFS (R = 0.61), as well as between RR and OS (R = 0.51). There was no significant correlation between 3-month and 6-month PFS, suggesting these endpoints should be used with caution.

Toxicity was reported “comprehensively” in only 20 studies (47% of those with available analysis), and it was reported “poorly” in six studies (14%). There was no change in toxicity reporting over time.

Four of the studies with defined endpoints reported positive results even though the primary endpoints were not reached. The researchers noted that three of these four trials were supported by industry.

“There has been an encouraging improvement over time in trial design, manifested by increased use of intent-to-treat analysis and prespecification of primary endpoints,” the authors concluded, adding that these data support the use of PFS as a surrogate for OS.

In an accompanying editorial, Fengmin Zhao, MHS, PhD, of Dana-Farber Cancer Institute in Boston, wrote that the limited number of trials available to assess surrogacy in this field makes the analysis problematic. Still, she wrote that validation of surrogate endpoints remains important for a variety of reasons, including the fact that “it is difficult, if not impossible, to conduct a phase III trial with OS as the primary endpoint for a rare disease” like STS.