Author: Halford, Zan ; Bi, Shulin ; Zhang, Zhiwei ; Allen, Jeff D. ; Brunner, Lauren ; Li, Wenjie ; Kahn, Andrea G. ; Zou, ShiPing ; McShane, Lisa M. ; Arend, Rebecca ; Oliver, Bailee D. ; McKelvey, Brittany A. ; Zhao, Yingdong ; Kincaid, Kaitlyn ; Kotlov, Nikita ; Stenzinger, Albrecht ; Nam, Hyunjun ; Foxall, McKenzie ; Saul, Daniel ; Starks, Elizabeth R. ; Gupta, Mohit ; Lai, Yi-Hsuan Lucy ; Kohn, Elise C. ; Chapman, Alyssa ; Li, Ming-Chung ; Lazar, Alexander J. ; Karlovich, Chris ; Coletta, Tommaso ; Laird, A. Douglas ; Andrews, Hillary S. ; Safabakhsh, Pegah ; Chen, Li ; Pabla, Sarabjot ; Sokol, Ethan S. ; Taxter, Timothy ; Ding, Yuan ; Stewart, Mark D.
PURPOSEHomologous recombination deficiency (HRD) assays measure DNA damage repair dysfunction to identify patients with high-grade serous ovarian cancer (HGSOC) who may benefit from poly ADP-ribose polymerase inhibitors (PARPis). Numerous assays are available, but only two have undergone prospective clinical validation. Assay variability can affect patient and provider treatment choices; however, the level of assay variability across laboratory developed tests is unknown.METHODSFriends of Cancer Research initiated a research partnership, inviting HRD assay developers to participate in two blinded analyses. In the first, 11 assay developers reported HRD status for the Cancer Genome Atlas HGSOC data set (In Silico; n = 348) and then 17 assay developers reported HRD status for nucleic acids freshly extracted from archival specimens (n = 90) from patients with advanced HGSOC (clinical). HRD status was compared for each analysis.RESULTS
The median (IQR) pairwise positive percent agreement (PPA) for the in silico analysis was 74% (51%-89%) and 81% (64%-92%) for pairwise negative percent agreement (NPA); for the clinical analysis PPA was 83% (70%-91%) and NPA was 80% (62%-91%). There was higher positive agreement on HRD status calls among those with a
BRCA1
or
BRCA2
mutation and a higher negative agreement in
CCNE1
-amplified cases. Sample characteristics like tissue block age were not observed to be associated with agreement. A subgroup of tumors largely called HRD across assays with no
BRCA1
or
BRCA2
mutations was associated with better outcomes on standard platinum-based therapy compared with not HRD; however, the subgroup was small, and further research is warranted.
CONCLUSIONThis analysis demonstrates how results from 20 HRD assays compare when assessing HGSOC. The results set the stage to improve alignment and establish standards for acceptable levels of agreement moving forward.