Investigation of clonal hematopoiesis frequency following PARP inhibitor treatment and its association with prior platinum therapy exposure.

Abstract

e15017 Background: Poly (ADP-ribose) polymerase inhibitors (PARPi) have become a critical part of treatment for multiple solid tumors, especially among patients with germline homologous recombination deficiency (HRD). With their significant impact on survival has come a concern for therapy-related myeloid neoplasms (tMN). DNA damage response (DDR) mutations causing clonal hematopoiesis (CH) are the driving force behind the tMNs. We hypothesized that platinum and PARPi therapy have significant variation in their impact on CH, further altered by the presence of HRD. Methods: Serial blood samples from 250 patients with a variety of cancers, primarily prostate, ovarian, and lung cancer treated with PARPi (n = 100) or carboplatin (n = 150) were compared to 176 age-matched patients who were not treated for cancer. Samples were sequenced at an average depth of 19,964x using a sequencing assay that captured the exonic regions of nine common CH genes ( DNMT3A, TET2, ASXL1, TP53, CHEK2 , JAK2, SRSF2, SF3B1, PPM1D ). Mixed bone marrow chimeric mice were created by competitive transplant of Trp53 R172H/+ cells—the murine equivalent of the R175H TP53 hotspot mutation seen in myeloid neoplasms— against wild-type cells. Mice were treated with vehicle (n = 19), cisplatin (n = 16), olaparib (n = 9), or talazoparib (n = 14) for 4 weeks before sequential bleeding and bone marrow collection. Results: The average length of carboplatin treatment was 5.6 months (range 1.15-40.8 months) and 8.76 months (range 1.78-44.4 months) for PARPi treatment. Among patients treated with either carboplatin or PARPi, 51% had on average two DDR CH mutations (range 0-19) versus 23% of untreated patients. Carboplatin induced significantly greater growth of DDR CH compared to PARPi (p = 2.6e-2). Expansion of DDR CH mutations during treatment was significantly reduced in patients with a germline HRD pathogenic variant (N = 44) as compared to those without (N = 276, p < 0.001). No significant expansion of DDR CH mutations was observed after either PARPi (p = 3.9e-2) or carboplatin exposure (p = 1.4e-2) in patients with a germline HRD pathogenic variant. Similarly, within our murine model we observed that cisplatin (p = 5.5e-11) and to a lesser extent talazoparib (p = 0.15) increased Trp53-mutant blood leukocytes. However, the competitive advantage of Trp53-mutant cells during treatment with both cisplatin and talazoparib was lost among mice with hematopoietic-specific Brca1 heterozygote mutations. Conclusions: We observed a high frequency of DDR CH expansion after carboplatin and to a lesser extent PARPi treatment. This expansion was largely abrogated in patients with germline HRD. We validated these findings in a mouse model of Trp53-mutated CH. Our findings suggest that the increased risk of tMN following PARPi is likely influenced by prior exposure to other oncologic therapies, including platinum therapy, and may be modified by HRD status.