OBJECTIVES:

Single-arm basket trials represent significant challenges for assessing comparative effectiveness. We compared progression-free survival (PFS) and overall survival (OS) of patients with eight metastatic cancers who received pembrolizumab versus patients receiving NCCN-guideline-based chemotherapies.

METHODS:

We applied inclusion/exclusion criteria from microsatellite instability–high (MSI-H) trials (Keynote-164 and Keynote-158) to TriNetx electronic health record database to derive chemotherapy control cohorts. Gaussian copulas were fitted to extract the dependence structure from the real-world controls, and trial samples were simulated by combining with marginal distributions of trial baseline covariates. Imbalances of known covariates were adjusted with inverse odds weighting. Median OS and PFS were compared between trial and real-world cohorts for all tumor types. Weighted Cox proportional hazard regressions were also conducted for colorectal cancer (CRC) and endometrial cancer (EC) given available Kaplan-Meier data.

RESULTS:

Heterogenous median real-world OS (mrwOS, in months) and median real-world PFS (mrwPFS, in months) were observed among 9267 patients with metastatic cancers who progressed on ≥1 line of systemic therapies. Compared to patients receiving pembrolizumab, patients receiving chemotherapy for colorectal (39.3+ vs. 34.3), endometrial (56+ vs. 36.3), cholangiocarcinoma (19.4 vs. 11.7), small intestine (56+ vs. 21.8), and ovarian cancers (33.6 vs. 28.5) experienced shorter mrwOS, whereas patients with gastric (11 vs. 16.5), pancreatic (3.7 vs. 10), and brain cancers (5.6 vs. 10.3) had longer mrwOS. Similar patterns were observed for mrwPFS, which ranged from 2.8 to 4.7 months. Pembrolizumab was associated with more favorable PFS than chemotherapy for treating CRC and EC, respectively (CRC: Hazard Ratio [HR], 0.47; 95%CI, 0.34-0.66; EC: HR, 0.34; 95%CI, 0.26-0.44). However, MSI-H patients with CRC and EC receiving pembrolizumab did not demonstrate significantly better overall survival.

CONCLUSIONS:

This study provided external comparators for efficacy evidence from single-arm tumor-agnostic trials. Pembrolizumab was associated with significant prolonged PFS but not OS versus chemotherapy in patients with metastatic CRC or EC.