The study, published in Nature Genetics by a team led by Kelly L. Bolton of Washington University School of Medicine together with collaborators at Memorial Sloan Kettering Cancer Center and other institutions, addresses a pressing clinical problem. PARP inhibitors have transformed treatment for ovarian, breast, pancreatic and prostate cancers driven by defects in homologous recombination repair, and their use is expanding into frontline settings. But these drugs have been associated with an increased risk of therapy-related myeloid neoplasms — lethal cancers of the blood-forming system with a five-year survival of less than 10 percent. Reported five-year risks of these neoplasms in clinical trials and claims-based studies have varied wildly, from under 1 percent to more than 10 percent, and the reasons for this variation have remained unclear.

The origin of most therapy-related myeloid neoplasms is clonal hematopoiesis, an age-related phenomenon in which hematopoietic stem and progenitor cells acquire somatic mutations that allow them to outcompete their normal neighbors. Mutations in TP53, the most commonly mutated gene in therapy-related myeloid neoplasms, are considered a hallmark of the disease. Previous work by this group and others showed that TP53-mutant clones are often present in the blood before cancer therapy begins, and that exposure to cytotoxic treatment gives these mutants a fitness advantage over wild-type stem cells, driving their expansion. In a subset of individuals, these clones acquire additional hits — loss of the second TP53 allele and copy number alterations — marking the transition to full-blown malignancy.

To dissect the contribution of PARP inhibitors, the team first reanalyzed data from 10,156 patients sequenced with the MSK-IMPACT targeted panel at Memorial Sloan Kettering. They found that 18.5 percent of patients exposed to PARP inhibitors carried detectable mutations in DNA damage response genes as clonal hematopoiesis, compared with only 4.5 percent of patients exposed to other therapies and 1.4 percent of untreated individuals. PARP inhibitor exposure showed the strongest association with DNA damage response clonal hematopoiesis of any treatment class, with an odds ratio of 4.4. But a confounding problem loomed: every PARP inhibitor-treated patient in the cohort had also received other systemic therapies, particularly carboplatin, which the same group had previously linked strongly to clonal expansion. When the researchers adjusted statistically for exposure to cytotoxic therapy classes and their cumulative doses, the association between PARP inhibitors and clonal hematopoiesis shrank steadily, from an odds ratio of 4.4 down to 2.9 and finally to 1.5, no longer statistically significant.

To overcome this confounding directly, the investigators turned to prospective serial sampling. They collected blood from 250 patients before and after treatment with either a PARP inhibitor (100 patients) or carboplatin (150 patients), and compared the results with serial samples from 176 age-matched healthy individuals. Using an ultra-deep error-corrected sequencing approach with the ArCH variant calling pipeline at roughly 18,000-fold coverage, they detected 921 putative driver mutations down to a variant allele fraction of just 0.2 percent — half of which were tiny clones undetectable by standard methods. After treatment, 51 percent of patients carried a DNA damage response clonal hematopoiesis mutation, with an average of two mutations per person, compared with 22.7 percent of healthy controls.

The crucial comparison came from modeling the exponential growth rate of each clone between the two time points. Both carboplatin and PARP inhibitor exposure accelerated the growth of clones carrying mutations in the DNA damage response genes PPM1D, TP53 and CHEK2, while clones carrying mutations in the common epigenetic regulator genes DNMT3A, TET2 and ASXL1 remained unchanged. But carboplatin drove significantly faster expansion than PARP inhibitors across all three genes, a difference that held even when the analysis was restricted to patients carrying only a single TP53 or PPM1D mutation. Within the PARP inhibitor class, the drug with the strongest PARP-trapping activity, talazoparib, was a stronger driver of clone expansion than olaparib, consistent with its greater capacity to inflict DNA damage. Interestingly, patients who had received prior cytotoxic or radiation therapy showed reduced clone growth rates during subsequent PARP inhibitor or carboplatin treatment, suggesting that the most expansion-prone clones may have already been depleted or that the cellular landscape changes with repeated exposure.

The team also examined mosaic chromosomal alterations — acquired copy number changes detectable in blood DNA — using SNP array profiling in 347 patients. Nearly a quarter of individuals carried at least one such event at some point. PARP inhibitor exposure significantly increased the expansion of autosomal copy number alterations, driven largely by newly detected events, with a similar but non-significant trend after carboplatin. Most tellingly, patients with pre-existing TP53 clonal hematopoiesis were 6.4-fold more likely to acquire a detectable mosaic chromosomal alteration during therapy, and the risk rose with the size of the pre-existing clone. This finding fits the established narrative that TP53 loss paves the way for the complex aneuploidy characteristic of therapy-related myeloid neoplasms.

The most surprising result concerned inherited genetics. Because many patients receiving PARP inhibitors carry germline pathogenic variants in homologous recombination genes such as BRCA1 and BRCA2, the researchers hypothesized that a single defective copy of these genes in blood stem cells would amplify the fitness advantage of DNA damage response mutations under treatment pressure. The opposite occurred. Clone growth was significantly reduced in germline HRD carriers after both PARP inhibitor and carboplatin treatment, a pattern specific to DNA damage response genes and absent among the epigenetic regulator clones. Even among HRD carriers, carboplatin still outpaced PARP inhibitors in driving expansion. The pattern held for both BRCA1/2 carriers and carriers of mutations in other homologous recombination genes.

Mouse experiments validated and mechanistically explained the human findings. In competitive bone marrow transplantation models, Trp53-mutant hematopoietic cells carrying the murine equivalent of a common human TP53 hotspot mutation expanded markedly after cisplatin treatment and modestly after talazoparib, but showed no advantage with olaparib — mirroring the patient data. When the researchers introduced a heterozygous Brca1 mutation into both the mutant and competitor cells, the picture changed dramatically. Double-mutant cells showed a severe engraftment defect even without treatment, and after cisplatin or talazoparib they failed to expand, with their contribution to blood and bone marrow significantly reduced compared with Brca1-wild-type controls. Notably, mice with heterozygous Brca1-mutant marrow treated with talazoparib developed severe anemia, and 9 of 34 such mice died during the four-week treatment.

Single-cell analyses revealed the mechanism. A single dose of talazoparib increased double-stranded DNA breaks in all genotypes, measured by neutral comet assay and γH2AX staining, but the damage was greatest in Brca1-haploinsufficient cells. Critically, cells carrying both heterozygous Brca1 and Trp53 mutations showed significantly greater apoptosis — measured by cleaved caspase 3 — than cells with either mutation alone, even at baseline. The interpretation is that Brca1 haploinsufficiency renders blood stem cells vulnerable to PARP inhibitor-induced DNA damage, while the concurrent TP53 mutation cripples their ability to resolve that stress, triggering cell death rather than clonal expansion. This challenges the prevailing view, based largely on epithelial tumor cells, that a single functional copy of BRCA1 is sufficient for DNA repair, and suggests blood stem cells are unusually sensitive to BRCA1 dosage. The authors caution that whether these clonal dynamics translate directly into myeloid neoplasm risk requires long-term follow-up, ideally in prospective PARP inhibitor trials. They also note that analyses of the ARIEL2 and ARIEL3 rucaparib trials found that duration of prior platinum exposure, not inherited predisposition, was the greatest predictor of myeloid neoplasm risk. If validated, serial clonal hematopoiesis profiling before and during treatment could become a practical tool for identifying which patients need heightened surveillance — and reinforce the message that the drugs patients received before a PARP inhibitor may matter as much as the inhibitor itself.

Subject of Research: How germline homologous recombination deficiency and platinum or PARP inhibitor therapy shape TP53-mutant clonal hematopoiesis and therapy-related myeloid neoplasm risk

Article Title: Germline homologous recombination deficiency influences TP53-mutant clonal hematopoiesis fitness during platinum and PARP inhibitor treatment

Article References: Baeten, J. T., Chan, I. C. C., Moukarzel, L., Petrone, G. E. M., Liu, J., Tran, D., Tabs, J., Tabet, I., Agashe, S., Nasrollahzadeh, E., Vasireddy, A., Carter, A., Patel, M., Stopsack, K. H., Kantoff, P. W., Zhou, W., Batchi-Bouyou, A. L., Beeler, J. S., Mustion, G., … Bolton, K. L. (2026). Germline homologous recombination deficiency influences TP53-mutant clonal hematopoiesis fitness during platinum and PARP inhibitor treatment. Nature Genetics. https://doi.org/10.1038/s41588-026-02752-2

Image Credits: AI Generated

DOI: 10.1038/s41588-026-02752-2

Keywords: clonal hematopoiesis, PARP inhibitors, carboplatin, TP53, BRCA1, homologous recombination deficiency, therapy-related myeloid neoplasms, PPM1D, mosaic chromosomal alterations, hematopoietic stem cells, DNA damage response, Nature Genetics