Laboratory cross-comparison of homologous recombination repair mutation analysis in tumor in a multicenter epithelial ovarian cancer series: The BORNEO GEICO 60-0 study.

Abstract

e17550 Background: In epithelial ovarian cancer (EOC), the identification of mutations in homologous recombination repair (HRR) genes on tumor is prognostic, predictive of response to PARP inhibitors, and a tool to identify individuals at genetic cancer risk. The aim of this study is to compare the concordance between two laboratories in identifying and classifying genetic variants in HRR genes. Methods: In a multicentre ambispective series of unselected, non mucinous EOC of all stages formalin-fixed and paraffin embedded tumors were collected. These samples underwent the same mutational analysis of 15 HRR genes ( ATM, BARD1, BRCA1, BRCA2, BRIP1, CDK12, CHEK1, CHEK2, FANCL, PALB2, PPP2R2A, RAD51B, RAD51C, RAD51D, RAD54L) in two different Laboratories (Lab1, Lab2) that used their own validated multi-gene NGS panels. Variant allele frequency (VAF) threshold was 5% for single nucleotide polymorphism and 10% for indels. Large rearrangements were not analyzed. Variants were classified into three categories based on ACMG criteria: non-mutated (class 1-2), Variants of Uncertain Significance (VUS: class 3) and likely pathogenic/pathogenic (class 4-5). Results: A total of 81 cases were sent for the analysis. One had low DNA quality and therefore 80 cases were finally studied (85% high grade serous and 74% FIGO stage III-IV). Results reported by Lab1 and Lab2 (lab1/Lab2) were the following: 21/19 (26%/24%) cases had BRCA1/2 mutations, 7/8 (8.7%/10%) mutations on other HRR genes including two in ATM and RAD51D, one in CHEK1, CHEK2, and FANCL and one RAD51C reported in Lab2 only while the rest were either VUS 23/27 (29%/34%) or non-mutated 29/26 (36%/33%). Concordance between laboratories in classifying patients was 93.75% (kappa coefficient 0.86). Discrepancies (DC) on variants were classified into arbitrary categories, namely 0= complete concordance, category 1 meaning DC in detection assumed to be due to tumor heterogeneity (VAF nearby the threshold) or technique (1A), or caused by laboratories performance and avoidable (1B) and the category 2 identified DC in interpretation without clinical relevance (2A) or clinically relevant (2B), the results of total number of variants are shown in table. Overall, regarding clinically relevant DC in HRR genes, 9 DC in variants were observed including six 2B, two 1A and one 1B and they affect 5 (6.3%) patients since some were overlapping. Conclusions: In our EOC series the concordance of two Laboratories in the identification of clinically relevant HRR mutations on tumor is high but discrepancies in interpretation remain a challenge that needs further harmonization.[Table: see text]