The ability to better predict response to cancer therapy would substantially improve the lives of patients. For high-grade serous ovarian cancer (HGSOC) tumors, their DNA repair capacity -via the homologous recombination (HR) pathway -is the most important molecular determinant of sensitivity to platinum compounds, the DNA-damaging chemotherapy used in first-line treatment [1] . Namely, HR-deficient tumors are platinum sensitive. To date, approaches to identify HGSOC patients with HR-deficient

more »

... umors have relied largely on DNA sequencing. They are based on scoring deleterious mutations in BRCA1, BRCA2 and other known HR genes and/or identifying features of the genomic landscape, that is, so-called genomic scars and mutational signatures associated with HR deficiency. These methods carry limitations, however. First, HR deficiency caused by epigenetic alterations (e.g. BRCA1 promoter methylation) is not detected by mutation screening. Second, mutations of uncertain significance may or may not confer HR deficiency. Third, genomic scars and mutational signatures associated with HR deficiency do not necessarily correspond to functional HR deficiency -HR-reversion mutations have been shown to restore HR functionality in tumor cells, while the "historical" genomic signature of HR deficiency is retained. Moreover, high-throughput sequencing remains comparatively expensive and necessitates heavy downstream bioinformatic analyses. Thus, it is not suited for determining tumors' HR capacity at the time when the sample is collected in surgery, nor in a routine clinical laboratory setting with limited resources. Limitations of DNA sequence-based testing can be overcome by directly assessing whether the HR pathway is functional. To this end, detecting nuclear foci of RAD51, the key HR recombinase, has proven a powerful technique in classifying HR-deficient versus HR-proficient breast cancer samples [2, 3] . We set out to adapt this approach for HGSOC samples, and to validate its utility in predicting clinical platinum sensitivity. We obtained an HR score for each HGSOC patient sample by quantifying RAD51mediated repair after ionizing radiation (IR) induced DNA damage in primary cell culture [4] . HGSOC samples clustered into three functional groups: HR-deficient, HR-low and HR-proficient ( Figure 1A) ; genomic characteristics of HR deficiency were absent in one quarter of HR-deficient and HR-low samples. Low HR scores significantly correlated with platinum sensitivity, longer platinum-free interval and improved overall survival. Editorial Figure 1: HR scores, as assessed by functional ex vivo testing, from 21 HGSOC patients (A, data from ref. 4) and 17 TNBC patients (B, data from ref. 5). Dashed lines indicate thresholds for the three HR capacity categories, as defined in the two studies. Asterisks in B mark the position of two TNBC samples with an HR score of zero [5].