Association of PD-L1 Expression with Clinical Response to Pembrolizumab in Patients with Advanced Recurrent Ovarian Cancer

Conference Correspondent

Previously reported results from the phase 2 KEYNOTE-100 study showed that pembrolizumab has clinical activity in patients with advanced ovarian cancer, and that tumor expression of programmed cell death-ligand 1 (PD-L1; combined positive score 10) was associated with response.1 Here, Ledermann and colleagues presented the results of other potential response-associated biomarkers that were evaluated in that study.2

KEYNOTE-100 included women with epithelial ovarian, fallopian tube, or primary peritoneal cancer, who had confirmed recurrence following frontline platinum-based therapy. All patients had a European Cooperative Oncology Group performance score of 1 and an available tumor sample. During the study, patients received 200 mg pembrolizumab intravenously once every 3 weeks for 2 years or until progression, death, unacceptable toxicity, or withdrawal of consent. Whole exome sequencing of paired tumor and normal samples was used to determine homologous recombination deficiency (HRD) genomic scar and BRCA1/2 mutation status (BRCA) using standard algorithms. Associations of response with T-cell–inflamed 18-gene expression profile (T-cell-GEP) score, HRD, BRCA, and microsatellite instability-high (MSI-H) were evaluated.

Data on T-cell-GEP, BRCA, and HRD were available from the first 100 patients enrolled, whereas MSI-H data were available from the entire study population (n = 319). Of the patients with available T-cell-GEP, the distribution of GEP scores was significantly higher in responders than in nonresponders (1-sided P = .03 from Wilcoxon rank sum test; n = 83). Seven of these 83 patients (8.4%) had a response. In patients with available PD-L1 combined positive score and GEP score (n = 79), the area under the receiver characteristic curves were numerically similar (0.73 vs 0.72, respectively; Spearman’s correlation, P = .57). Likewise, no statistically significant differences were observed with HRD values among responders and nonresponders (1-sided P = .29; n = 71), nor were associations between BRCA status and response observed (n = 11 mutant; n = 60 wild type; 1-sided P = 0.65). Finally, of 319 paired samples tested for MSI-H, all were microsatellite stable.

Based on these and previous biomarker data, in addition to PD-L1, combined positive scores and T-cell-GEP are associated with a response to pembrolizumab monotherapy for treatment of advanced ovarian cancer in a single-arm setting. Notably, biomarkers for HRD and BRCA were not found to be associated with response. Moreover, CA-125 levels were largely concordant with radiographic tumor response.


  1. Matulonis UA, et al. ASCO 2018. Abstract 5511.
  2. Ledermann JA, et al. ESMO 2018. Abstract LBA36.

Related Articles