AbstractIn a trial of patients with high grade serous ovarian cancer (HGSOC), addition of the ATR inhibitor berzosertib to gemcitabine improved progression free survival (PFS) compared to gemcitabine alone but biomarkers predictive of treatment are lacking. Here we report a candidate biomarker of response to gemcitabine versus combined gemcitabine and ATR inhibitor therapy in HGSOC ovarian cancer. Patients with replication stress (RS)-high tumors (n = 27), defined as harboring at least one genomic RS alteration related to loss of RB pathway regulation and/or oncogene-induced replication stress achieve significantly prolonged PFS (HR = 0.38, 90% CI, 0.17–0.86) on gemcitabine monotherapy compared to those with tumors without such alterations (defined as RS-low, n = 30). However, addition of berzosertib to gemcitabine benefits only patients with RS-low tumors (gemcitabine/berzosertib HR 0.34, 90% CI, 0.13–0.86) and not patients with RS-high tumors (HR 1.11, 90% CI, 0.47–2.62). Our findings support the notion that the exacerbation of RS by gemcitabine monotherapy is adequate for lethality in RS-high tumors. Conversely, for RS-low tumors addition of berzosertib-mediated ATR inhibition to gemcitabine is necessary for lethality to occur. Independent prospective validation of this biomarker is required.