Influence of Genomic Landscape on Cancer Immunotherapy for Newly Diagnosed Ovarian Cancer: Biomarker Analyses from the IMagyn050 Randomized Clinical Trial

Abstract

Purpose: To explore whether patients with BRCA1/2-mutated or homologous recombination deficient (HRD) ovarian cancers benefitted from atezolizumab in the phase III IMagyn050 (NCT03038100) trial.

Patients and methods: Patients with newly diagnosed ovarian cancer were randomized to either atezolizumab or placebo with standard chemotherapy and bevacizumab. Programmed death-ligand 1 (PD-L1) status of tumor-infiltrating immune cells (IC) was determined centrally (VENTANA SP142 assay). Genomic alterations, including deleterious BRCA1/2 alterations, genomic loss of heterozygosity (gLOH), tumor mutation burden (TMB), and microsatellite instability (MSI), were evaluated using the FoundationOne assay. HRD was defined as gLOH ≥ 16%, regardless of BRCA1/2 mutation status. Potential associations between progression-free survival (PFS) and genomic biomarkers were evaluated using standard correlation analyses and log-rank of Kaplan-Meier estimates.

Results: Among biomarker-evaluable samples, 22% (234/1,050) harbored BRCA1/2 mutations and 46% (446/980) were HRD. Median TMB was low irrespective of BRCA1/2 or HRD. Only 3% (29/1,024) had TMB ≥10 mut/Mb, and 0.3% (3/1,022) were MSI-high. PFS was better in BRCA2-mutated versus BRCA2-non-mutated tumors and in HRD versus proficient tumors. PD-L1 positivity (≥1% expression on ICs) was associated with HRD but not BRCA1/2 mutations. PFS was not improved by adding atezolizumab in BRCA2-mutated or HRD tumors; there was a trend toward enhanced PFS with atezolizumab in BRCA1-mutated tumors.

Conclusions: Most ovarian tumors have low TMB despite BRCA1/2 mutations or HRD. Neither BRCA1/2 mutation nor HRD predicted enhanced benefit from atezolizumab. This is the first randomized double-blind trial in ovarian cancer demonstrating that genomic instability triggered by BRCA1/2 mutation or HRD is not associated with improved sensitivity to immune checkpoint inhibitors. See related commentary by Al-Rawi et al., p. 1645.

Figure 1.

A, Genomic landscape of biomarker-evaluable population from IMagyn050 (pooled treatment arms) according to FoundationOne® CDx assay. B, Relationships between TMB, BRCA1/2 mutation status, and HR status. C, Prevalence of BRCA1/2 mutation by PD-L1 status. D, Prevalence of HRD by PD-L1 status. ^aBRCA1/2 mutation: known and likely deleterious tumor germline/somatic BRCA1/2 mutations; variants of unknown significance excluded. ^bHRD: gLOH ≥ 16%; HRP: gLOH < 16%, regardless of BRCA1/2 mutation status. For visualization purposes, patients with TMB = 0 were set to TMB = 0.01 and those with gLOH = 0 were set to gLOH = 0.1. Patients with no data are blank. HR, homologous recombination; LGSOC, low-grade serous ovarian cancer.

Figure 2.

A, Gene mutations associated with PFS (univariate analysis). B, PFS according to BRCA1/2 mutation status in the placebo, chemotherapy, and bevacizumab control arm and the atezolizumab, chemotherapy, and bevacizumab arm. C, PFS according to homologous recombination status in the placebo-containing control arm and the atezolizumab-containing arm.

Figure 3.

A, Association between PFS outcome, BRCA1/2 mutation status, and PD-L1 status. B, PFS according to treatment arm and BRCA1 versus BRCA2 status. C, Forest plot of PFS according to treatment arm, PD-L1 status, and BRCA mutation status. Four patients with both BRCA1 and BRCA2 mutations are excluded from panel B (1 patient in the placebo arm with PFS of 12.5+ months; 3 in the atezolizumab-containing arm with PFS of 17.1, 18.1+, and 12.7+ months). CPB, paclitaxel, carboplatin, and bevacizumab; NE, not estimable.

Figure 4.

Association between PFS outcome, homologous recombination status (HRD: gLOH ≥16%; HRP: gLOH <16%), and PD-L1 status. CPB, paclitaxel, carboplatin, and bevacizumab.