Ovarian cancer tumor microenvironment and atezolizumab (atezo) clinical activity: IMagyn050 sub-study

Abstract Background: Tumor biomarkers such as CD8 density and location (i.e., immune inflamed phenotype) and immune rich molecular subtype have been linked to immune checkpoint blockade (ICB) overall survival (OS) in different cancers. The IMagyn050 trial (NCT03038100), which evaluated the efficacy of Atezo vs placebo (Pla) with carboplatin, paclitaxel and bevacizumab (CPB) in front line ovarian cancer patient (pts), did not meet its co-primary endpoints of PFS in ITT or PD-L1+ (Moore et al. JCO 2021). In the current IMagyn050 substudy we assessed potential predictive tumor immune biomarkers for Atezo clinical benefit. Methods: FFPE tumors from the biomarker evaluable population were tested for PD-L1 IHC, CD8/PanCK IHC (total CD8 T cells and immune location phenotypes [inflamed, excluded, desert]) and RNA-seq (to derive molecular subtypes, biological pathways and cellular components [xCELL]) in tissue from baseline (n=860), on-treatment (OT, 9 weeks, n=233), intra- (n=8) and inter-lesion (n=12) matched samples. Hazard ratio (HR) interaction test from multivariate adjusted Cox-regression analysis for PFS and OS was performed to test predictive biomarkers. Results: While tumors with CD8 T cells, immunoreactive molecular subtypes or immune inflamed phenotype were enriched for PD-L1+, only pts with immune inflamed tumors showed improved OS Atezo benefit (HR 0.67). Improved Atezo PFS/OS benefit was also observed in pts with whose tumors had high oxidative phosphorylation (OXPHOS, HR: 0.72/0.65) and UV Response (UV, HR: 0.64/0.58) but not IFNγ response. Plasma B cells were linked to improved OS Atezo benefit vs Pla (HR 0.53). We leveraged OT samples from pts in the neoadjuvant cohort to assess treatment effect on the tumor microenvironment. Analyses showed that CPB reduced tumor proliferation and increased tumor immune inflammation (CD8 T cells, PD-L1 and IFNα/IFNγ response), further increased by Atezo. Immune inflammation is challenging in ovarian cancer due to extensive tumor heterogeneity. Prevalence of tumor biomarkers varied by anatomic locations: total CD8, CD8 localization and molecular subtypes. Inter- and intra-lesion biomarker status within the same pt showed PD-L1 and plasma B cells as most consistent. Molecular subtypes and immune phenotypes had moderate intra-lesion agreement but discordant between lesions. PD-L1 and OXPHOS were the only biomarkers linked to Atezo benefit regardless of anatomical location. Conclusion: This comprehensive exploratory study suggests that DNA damage, OXPHOS, plasma B cells and immune inflamed tumors, but not molecular subtypes or total CD8 T cells, may predict Atezo + CPB OS. This treatment promotes immune inflammation in OC tumors. Notably, we found that several biomarkers are highly heterogeneous. Our findings highlight the challenges of achieving durable clinical benefit from targeted immunotherapy in ovarian cancer pts. Citation Format: Venkatesh Krishnan, Ching-Wei Chang, Habib Hamidi, Michael A. Bookman, Charles Landen, Tashanna Myers, Hiroaki Kajiyama, Sakari Hietanen, Lyndsay Willmott, Premal Thaker, Cagatay Taskiran, Jalid Sehouli, Victor Khor, Yvonne Lin Liu, Sandro Pignata, Kathleen Moore, Luciana Molinero. Ovarian cancer tumor microenvironment and atezolizumab (atezo) clinical activity: IMagyn050 sub-study. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 5702.