PARP Inhibitor Treatment Increases DNA Damage in Bone Marrow and Drives Compensatory Changes in Multiple Hematopoietic Stem and Progenitor Cell Populations Contingent on Brca1 Genotype

Poly(ADP-ribose) polymerases (PARPs) are a family of 17 proteins crucial for DNA repair. The most abundant members, PARP1 and PARP2, can detect single-strand DNA breaks (SSBs) and initiate poly(ADP-ribosyl)ation (PARylation) to recruit additional repair proteins. Both PARP1 and PARP2 are targeted by a class of cancer therapeutics called PARP inhibitors (PARPi). PARPi are most effective in cancers with defects in the double-strand break repair pathway, primarily through loss of homologous recombination (HR) genes like BRCA1 and BRCA2, resulting in cell death through synthetic lethality. There are four FDA-approved PARPi with efficacy in delaying disease relapse or progression in several advanced cancers. However, severe hematologic adverse events are the most common and serious toxicities including severe anemia and often lethal myelodysplastic syndrome and acute myeloid leukemia. Since these drugs are commonly given to individuals with heterozygous germline mutations, such as women with ovarian cancer due to a germline BRCA1 mutation, it is critical to understand whether their normal tissues are more susceptible to these serious off-target hematologic adverse events.

Previously, we demonstrated that Brca1 deficiency impairs hematopoiesis in mice. To test whether PARPi cause more hematologic toxicities in those with heterozygous Brca1 deficiency versus wild-type, we exposed Mx1-Cre⁺Brca1^+/- mice to 30 days of olaparib after conditional deletion of one allele via pIpC. Even with this short duration of exposure, weekly blood count monitoring revealed a downward trend in white blood cells (Brca1^+/+, p = 0.13; Brca1^+/-, p= 0.01) and hemoglobin (Brca1^+/+ F, p = 0.03; Brca1^+/+ M, p = 0.56; Brca1^+/- F, p = 0.74; Brca1^+/- M, p = 0.065) as well as an upward trend in reticulocytes (Brca1^+/+, p = 0.21; Brca1^+/-, p= 0.05) of both genotypes treated with olaparib as compared to vehicle. However, the alterations were more pronounced in Brca1^+/- mice. The reticulocytosis was of particular interest due to data showing Parp2^-/- mice develop anemia secondary to hemolysis. Therefore, we investigated potential mechanisms of anemia with reticulocytosis. We did not find evidence of bleeding on necropsy or intravascular hemolysis as measured by serum haptoglobin. However, we did detect a decrease in average RBC survival in the peripheral blood of mice treated with olaparib versus vehicle as measured by in vivo RBC biotinylation. We also detected a significant increase in the number of Howell-Jolly bodies (HJB) in RBCs of olaparib treated mice, a marker of mutagenesis, in both genotypes.

Given the multilineage consequences, we assessed the impact of olaparib on hematopoietic stem and progenitor cell (HSPC) populations, function, and DNA damage in the bone marrow (BM). Unexpectedly, vehicle-treated Brca1^+/- mice had significantly more MPP2, MPP3, and GMP (p = 0.01; p = 0.03; p<0.01) and significantly fewer MPP4 progenitors (p=0.02) than Brca1^+/+ mice, suggesting a myeloid bias at baseline. Upon olaparib treatment, compensatory changes in specific HSPC populations were often in opposite directions in the two genotypes. In Brca1^+/- mice, olaparib treatment caused decreases in almost all HSPC populations, which was accompanied by an increase in whole BM apoptosis. Despite this, LT-HSCs increased and were functional in a competitive transplantation assay in which they outcompeted cells from vehicle-treated Brca1^+/- mice. To analyze the DNA integrity of key HSPC subsets, we utilized FACS and a COMET assay to quantify DNA damage. We detected increased DNA damage in MEPs and LT-HSCs in PARPi-treated mice (p <.01; p = 0.10).

Our findings suggest that PARPi treatment, particularly in Brca1^+/- mice, causes DNA damage and compromises multiple HSPC and mature cell populations. This leads to expansion of LT-HSCs which accumulate DNA damage, potentially selecting subclones with increased fitness. Ongoing work is assessing the molecular changes and additional Brca1 genotype by exposure fitness differences. Our data are critical for understanding the safety of PARPi use in the clinic and imply that germline genotype may impact risk for cytopenias and, potentially, therapy-related myeloid neoplasms.