Clonal Hematopoiesis Dynamics Under Systemic Therapy for Breast Cancer

Introduction: Clonal hematopoiesis (CH) is caused by the expansion of hematopoietic stem cells that carry somatic alterations in genes that are recurrently mutated in myeloid neoplasms. CH with a variant allele frequency (VAF) of 2% or higher is classified as clonal hematopoiesis of indeterminate potential (CHIP) and is associated with an elevated risk of overall mortality, especially from cardiovascular causes and progression to neoplasia. CH is more common in patients (pts) with solid tumors compared to healthy population-based cohorts. Retrospective cross-sectional studies have demonstrated a higher prevalence of CH/CHIP after exposure to cancer treatment, but the dynamics and treatment-specific effects on CH are poorly understood. Here, we conducted a systematic, longitudinal study on the impact of therapeutics on evolution of CH.

Methods: We analyzed data from 179 treatment-naïve pts with breast cancer for whom prospectively collected pre- and post-treatment peripheral blood samples were available under the Total Cancer Care institutional biorepository. Samples were collected before and shortly after first-course treatment for breast cancer. To detect CH, we targeted the exons of 81 hematopoietic disorder-related genes using error-corrected, ultra-deep DNA sequencing (mean depth 5,820). We used germline single nucleotide polymorphism loci linkage to confirm correct sample pairing and identified sites different from the reference genome. We excluded variants in germline calls from population cohorts and those with VAF >40%; we retained good quality mutations with VAF ≥0.1%. We estimated clonal growth rates using the maximal mutation frequency slope based on the Z-score of changes in VAF in serial samples.

Results: The median age of pts at breast cancer diagnosis was 58 years old (range 24-80). For breast cancer treatment, pts received only chemotherapy (n = 74; 43%), only radiation (n = 56; 32%), chemotherapy followed by radiation (n = 8; 5%), or only hormonal therapy (n = 36; 21%).

Blood samples were collected before systemic treatment for breast cancer or at a median of 6.1 (range 2.1-17.7) months after treatment start (n=406; 174 pre-, 168 post-, 3 during treatment; 61 additional serial). We detected CH in 52% of the pts (n = 93) prior to treatment, dominated by mutations in DNMT3A (n = 59/155), YLPM1 (n = 9), TET2 (n = 6), and PPM1D (n = 6). Most pts (58%, 54/93) had 1 CH mutation pre-treatment (range 1-5). In 18% (32/174) of pre-treatment samples, CH VAFs were sufficiently high (≥2%) to characterize CHIP. The prevalence of CH and the number of CH mutations pre-treatment were associated with age (P <0.001 and P=0.04, respectively). Hispanic pts (26%, 6/23) were less likely to have CH than Non-Hispanic pts (58%, 87/151; p=0.005). Pts with CH had a trend toward higher white blood cell count than those without CH (6.93 v 6.25 x10^9/L, p=0.09). There were no differences in pre-treatment CH by breast cancer subtypes or treatments received.

Of the CH mutations detected in post-treatment samples, 92% (153/166) were also called in pre-treatment samples; however, all were present at low VAF (range 0.01-0.08%) by manual review of pre-treatment data. Pts treated with chemotherapy were more likely to gain new CH mutations compared to those treated with radiation or hormonal therapy (q=0.004). Most CH mutations remained below the CHIP threshold after treatment (VAF <2%; 71%, 110/155), but some expanded to CHIP after treatment (n=7); most CHIP mutations remained at VAF ≥2% after treatment (92%, 35/38).

There were significant increases in gene-specific and cross-sectional growth rates for CH in pts treated with chemotherapy and/or radiation (q<0.05). In contrast, CH remained statistically unchanged after hormonal therapy. CH mutations in YLPM1 showed the most significant increase in growth rate under chemotherapy followed by those in SH2B3, TP53, GNAS, PPM1D, and DNMTA.

Conclusion: CH is common in pts with breast cancer. While CH evolutionary dynamics and degree of associations with clinical outcomes in cancer pts may be gene dependent and highly varied, our results provide strong evidence for acute changes in CH during the course of genotoxic treatments with implications for all solid tumors. Additional work that can be presented is focused on understanding the impact of variable CH dynamics on clinical outcomes.

Khiabanian:Regeneron Pharmaceuticals: Current Employment. Ganesan:KayoThera: Consultancy; Lunit: Consultancy; Gandeeva: Research Funding; Ipsen: Consultancy; Roche: Consultancy; Merck: Consultancy; Foghorn Therapeutics: Consultancy; EQRX: Consultancy; M2GEN: Research Funding.