Dynamics of clonal hematopoiesis in ovarian cancer survivors: Correlation with carboplatin and PARP inhibitor exposure

Ovarian cancer (OC) survivors may develop therapy-related myeloid neoplasia (TMN), and risk is increased by PARP inhibitor (PARPi) exposure. We characterized clonal hematopoiesis (CH), which may pre-date TMN, in blood from OC survivors, and its relationship to therapeutic exposures in the CHANCES ( C lonal H ematopoiesis in c ANCE r S urvivors) study. We sequenced blood cell DNA using BROCA-MY, our custom next-generation sequencing (NGS) panel with 72 genes associated with OC, bone marrow failure, and leukemia and CH and identified pathogenic variants (CHVs) at ≥1% variant allele frequency (VAF).We sequenced blood from 170 OC patients treated with carboplatin-based chemotherapy (OC chemo), 24 OC patients on a PARPi trial collected prior to subsequent diagnosis of TMN (pre-TMN), and 180 control subjects including OC patients without chemotherapy exposure (N=57) and age-matched cancer-free females (N=123). CHVs were common across all sample groups but were significantly more frequent in OC chemo (88/170, 52%) than controls (68/180, 38%, p=0.01). Within OC chemo cases, PARPi exposure was significantly associated with CHV detection: 39 (34.2%) had 0 CHV, 20 (17.5%) had 1 CHV and 55 (48.2%) had ≥2 CHVs, versus no PARPi exposure: 51 (51%) had 0 CHV, 29 (29%) had 1 CHV, and 20 (20%) had ≥2 CHVs (P<0.0001, chi-squared test for trend). CHVs were significantly more common in OC chemo cases than controls for TP53 (7.6% vs 1.7%, p=0.009), CHEK2 (11.2% vs 0.6%, p<0.0001) and PPM1D (21.8% vs 0.6%, p<0.0001) while CHVs in DNMT3A (21.8% vs 20.6%) and TET2 (10% vs 7.2%) were identified at similar frequency. OC chemo and pre-TMN cases receiving more >1 platinum regimen had greater frequency of TP53 -mutated CH (19/96, 19.8%), compared to ≤1 platinum regimens (4/81, 4.9%, p=0.003) We calculated the CH risk score (CHRS) for samples with available complete blood count results; the CHRS stratifies the risk of progression from CH to overt malignancy. OC chemo cases with CH were more likely to be categorized as high risk (7/23, 30.4%, CHRS≥1.0 predicting a >50% 10-year risk of progression to myeloid neoplasia) versus controls with CH (2/43, 4.7%, p=0.007). In serial samples collected from 41 participants in the CHANCES study, 83/88 (94.4%) CHVs were detectable in all samples. CH is frequent in both OC chemo cases and controls, but the number of mutations and involved genes vary by exposures, with chemotherapy selecting CHV in CHEK2, PPM1D, and TP53 . DNMT3A and TET2 CHVs, known to be associated with aging, were not impacted by chemotherapy exposure, while TP53 -mutated CH was directly associated with platinum exposure, consistent with an increased risk for TMN development. We continue to collect serial samples via the CHANCES study to characterize CH dynamics and TMN risks in cancer survivors with the goal to generate a risk calculator specific for TMN and establish recommendations for surveillance of CH in cancer survivors.