An analysis of 13 independently performed assays to measure homologous recombination deficiency using 90 freshly extracted high grade serous ovarian tumors: findings from the friends of cancer research hrd harmonization project

AACR Ovarian Cancer Special Session 2023 PRESENTATION
Authors Hillary Stires, Lisa M McShane, Rebecca C Arend, Alyssa K Chapman, Li Chen, Tommaso Coletta, Yuan Ding, Mohit Gupta, Nikita Kotlov, Alexander J Lazar, Ming-Chung Li, Yi-Hsuan Lucy Lai, Wenjie Li, Brittany A McKelvey, Jerod R Parsons, Ethan S Sokol, Elizabeth R Starks, Mark D Stewart, Peihua Wang, Zhiwei Zhang, Yingdong Zhao, ShiPing Zou, Jeff Allen

Homologous recombination deficiency (HRD) assays determine patient eligibility for treatment with PARP inhibitors and other DNA repair targeting drugs; therefore, understanding variability in how these assays measure and report HRD is critical for patients and providers. HRD assays measure various factors to determine HRD status including causes (i.e., inactivation in HR pathway genes) and consequences (i.e., genomic scarring). Methodological variability across HRD assays has led to a suggestion that the assays may not agree on a per-patient basis. An empirical assessment of assay variability may guide our understanding of how to implement “HRD status” as a biomarker.

Friends of Cancer Research (Friends) initiated a unique research partnership to assess levels of agreement and variability across HRD assays. We previously presented an analysis of the HRD status of 348 TCGA ovarian cancer samples across 11 assays. Concordance across assays was analyzed by measuring all possible pairings of samples and assays leading to a median (and IQR) positive percent agreement (PPA) of 74% (51-89%) and negative percent agreement (NPA) of 81% (64-92%). The median percent positivity (percent of patients with HRD) was 49% (range 9-67%). However, some groups modified their HRD pipelines to analyze the in silico data and we were unable to interrogate the influence of patient and sample characteristics on HRD calls.

To establish a more comprehensive dataset, we identified 142 archival specimens from the University of Alabama at Birmingham from patients with stage III-IV high-grade serous ovarian cancer diagnosed between 2011 and 2022. Full clinical information is available including response to platinum therapy. FFPE tumor from debulking surgery was sectioned for 99
samples with adequate tissue and underwent DNA and RNA extraction by the Molecular Characterization Laboratory (MoCha) at the NCI Frederick National Laboratory. MoCha shipped identical aliquots of DNA and/or RNA from the 90 samples that passed QC for independent sequencing and HRD measurement by the 13 different assays. Statisticians from the NCI performed statistical analyses to assess concordance across assays.

Three of the 13 assays considered mutations in non-BRCA1/2 HR pathway genes and 7 measured gLOH as determinants for HRD status among other factors. HRD calls resulted in a median pairwise PPA of 81% (69-91%) and a median pairwise NPA of 74% (61-89%). The median percent positivity was 53% (range 23-74%). Ongoing analyses will consider how each of the HRD assays predicts responsiveness to platinum-based chemotherapy. Additional analyses will consider assay, patient, and sample characteristics that may drive variation in HRD calls. Preliminary findings demonstrate variability in HRD calls across HRD assays, similar to the in silico analysis. These findings will help characterize the variability of HRD assays and inform best practices for measuring and reporting HRD.