Abstract
BACKGROUND: Targeting clonal hematopoiesis (CH) in an interventional clinical study involves significant challenges including: a lack of validated endpoints, long follow-up time, uncertainty about baseline risk, intolerance for drug-related adverse events, and limitations of genetic tests to identify and track mutant clones. For these reasons, several studies have focused on well-tolerated drugs as potential early interventions to mitigate the malignant and non-malignant risks associated with CH, largely in persons considered to be at higher risk and with readily detectable CH. For example, metformin has been proposed as a candidate therapy to treat CH based on its ability to inhibit mitochondrial Complex 1 (C1) where it has been shown to preferentially inhibit the growth of DNMT3A R882 mutant clones. To help determine the potential efficacy of a CH targeted study with metformin, we conducted an ancillary study using archival data and samples from a 6-month metformin intervention trial of breast cancer survivors.
OBJECTIVES: To characterize potential challenges associated with a randomized, prospective study of metformin on CH using highly sensitive error-corrected sequencing (ECS) and to determine the incidence of CH, its associations with participant (pt) characteristics, and the impact of 6-months of metformin on CH variant allele frequency (VAF).
METHODS: Blood buffy coat DNA samples from 332 non-diabetic female breast cancer survivors enrolled in the Reach for Health (RfH) randomized, placebo-controlled trial with metformin (titrated up to 1500 mg daily) were subjected to targeted ECS to identify somatic mutations of myeloid-malignancy associated genes. Samples were collected at enrollment and again 6 months later after the metformin or placebo treatment. Mutations were detected with max sensitivity of 0.1-0.2% VAF requiring a minimum of 3 consensus reads. Mutations were correlated with data from the RfH trial including demographics, hormonal profiles, inflammatory markers, prior cancer therapies, and clinical features.
RESULTS: CH was detected in 285 out of 332 pts (85.8%) with 1028 unique CH mutations identified in 37 genes. At baseline and after 6-months, 263 of 326 (80.7%) and 273 of 313 (87.2%) pts with evaluable samples had evidence of CH, respectively, with 263 of 285 (92.3%) harboring detectable CH mutations at both time points. VAFs for CH mutations assessed at time baseline and after 6-months were highly correlated (R2=0.971). The median maximum VAF in pts with CH was 0.69% and 27.7% of pts had VAF ≥ 2%, consistent with clonal hematopoiesis of indeterminate potential (CHIP). Mutations of DNMT3A were most frequently detected (in 68.6% of pts), followed by TET2 (39.2%), PPM1D (14.2%), and TP53 (13.6%). 56.3% of pts had CH in two or more genes with 137 out of 228 (60.1%) having two or more different mutations in DNMT3A. The mean number of mutations per subject with CH was 3.1 (median 3, range 1-20) and increased with age through age 75 years. Pts who received chemotherapy had significantly more CH mutations than pts who did not (mean 3.91 vs 3.28, p=0.033). Mutations of ATM co-occurred more often than expected with mutations of PPM1D (p=0.042, OR=2.72) and TP53 (p= 0.007, OR=3.56). TP53 (p=0.035) and PPM1D (p<0.001) mutations occurred more frequently in pts who received chemotherapy for breast cancer treatment. Overweight (BMI 25-30 kg/m2) pts were more likely to have CH mutations compared to obese (BMI >30 kg/m2) pts (p=0.007). Metformin treatment for six months did not appear to significantly alter CH VAF for any gene compared to placebo, even among the 30 pts with DNMT3A R882 mutations, 19 of which received metformin.
CONCLUSIONS: CH can be identified in the vast majority of breast cancer survivors using highly sensitive sequencing techniques and is associated with age and history of cytotoxic therapy. The variation in CH VAF over a 6-month period is low and does not appear to be affected by treatment with standard doses of metformin. This ancillary study demonstrates the utility of ECS for the detection and quantification of CH over time. This study highlights how key variables, such as the duration of treatment, length of follow up, and endpoint selection are critical to the successful interpretation of a randomized prospective CH interventional study.
