Impact of PARP Inhibitor and Platinum Therapy on Clonal Hematopoiesis

Poly (ADP-ribose) polymerase inhibitors (PARPi) are a promising new class of targeted therapy used in a variety of solid tumors with homologous recombination deficiencies (HRD). Data from clinical trials and observational studies have linked PARPi and platinum-based agents to therapy-related myeloid neoplasia (tMN). However, interpretation of these findings is limited by confounding variables including germline mutation status and prior therapeutic exposure. We and others have shown that the oncologic therapies known to confer an elevated risk of leukemia also promote the expansion of clonal hematopoiesis (CH) due to mutations of genes in the DNA damage response (DDR) pathway, including TP53, PPM1D, and CHEK2. However, the extent to which CH is associated with targeted therapies such as PARPi is not clear. We hypothesized that exposure to PARPi confers a competitive advantage to hematopoietic stem/progenitor cells (HSPCs) containing mutations in DDR pathway genes. Moreover, we hypothesized that in patients with germline mutations in HRD genes, such as BRCA1/BRCA2, genotoxic stress induced by PARPi or cytotoxic agents is increased, further enhancing the fitness of HSPCs carrying DDR gene mutations.

To test these hypotheses, we performed longitudinal studies on 942 peripheral blood samples collected from 418 patients (ages 32 to 94) before and after treatment with PARPi (n = 100), carboplatin (n = 142), and untreated individuals (n = 176). Patients with lung, ovarian, uterine, prostate, and breast cancer who received PARPi or carboplatin on clinical trials or through standard of care were included. Patients were treated with a variety of PARPi including Olaparib (n=48), Talazoparib (n=33), Rucaparib (n = 13) and Niraparib (n = 6.) Of the 418 patients, 45 individuals harbored an HRD germline mutation. Samples were sequenced at a depth of 18,946x using a UMI-based panel including 9 common CH genes (DNMT3A, TET2, ASXL1, TP53, PPM1D, SRSF2, SF3B1 and JAK2). Both carboplatin and PARPi treatment significantly increased the growth rate of CH mutations, driven almost entirely by mutations in DDR genes (Figure 1). Patients undergoing carboplatin treatment had significantly higher rates of mutational growth compared to those undergoing PARPi therapy (p-value = 5e^-17). The strongest PARP-trapping member, Talazoparib, had the largest effect on the growth rate of CH mutations in DDR genes (p-value = 3e^-9) but still lower compared to carboplatin. Surprisingly, DDR CH expansion was largely abrogated in patients harboring a pathogenic HRD germline variant for both carboplatin (p-value = 3e^-5) and PARPi (p-value = 7e^-10).

To validate these findings, we developed a mouse model of TP53-mutated CH by transplanting a limited number of Trp53 R172H bone marrow cells along with congenic wildtype bone marrow into irradiated recipients. The percentage of Trp53 R172H cells in the blood was measured by flow cytometry at baseline (6 weeks after transplantation) and after treatment with vehicle alone, Talazoparib or Olaparib daily, or cisplatin weekly, for 4 weeks. As expected, treatment with cisplatin resulted in a significant expansion of TP53-mutated cells (Figure 2). A non-significant trend to expansion of TP53-mutated cells also was observed after talazoparib, but not olaparib. We next modeled the effect of germline HRD mutations by generating bone marrow chimeras containing heterozygous Brca1-mutated HSPCs competed with Brca1-mutated HSPCs also carrying Trp53 R172H. Consistent with the human CH data, no expansion of TP53-mutated cells was observed after treatment with either cisplatin or Talazoparib, indeed there was a small, but significant decrease (Figure 2).

Collectively, these data support the hypothesis that mutations in DDR genes provide a fitness advantage to HSPCs following PARP inhibitor and platinum therapy. However, compared to platinum therapy, PARPi shows less selective pressure on DDR CH. Importantly, these data strongly argue against our initial hypothesis that germline HRD variants enhance this fitness advantage. Indeed, our data suggest the opposite, with the fitness advantage of DDR gene-mutated HSPCs lost in cells with a heterozygous Brca1-mutations. These data suggest that PARPi therapy may have less of an impact on leukemia risk compared to carboplatin and in fact may show synergistic effects with HRD in blocking the competitive advantage of DDR CH during genotoxic stress.