Mutational Analysis of Clonal Hematopoiesis in Solid Tumor Patients Illustrates the Critical Role of Systemic Anti-Cancer Therapies in the Evolution of Somatic Leukemia Disease Alleles

Background: Clonal hematopoiesis (CH) as evidenced by recurrent somatic mutations in leukemia-associated genes has been demonstrated to predict increased risk for subsequent hematologic cancers and decreased overall survival (OS) in large, population-based studies. We examined the incidence and clinical implications of CH in a large cohort of patients with advanced, non-hematologic cancers to examine whether individual mutations carried unique phenotypes with respect to effect on hematologic parameters and association with prior toxic exposures. Methods: We analyzed deep coverage targeted next-generation sequencing data of 410 cancer genes in paired tumor and blood samples from 7,200 patients with non-hematologic cancers. We called mutations in the blood and genotyped them in the matched tumor and required them to be supported by at least 8 reads and present at twice the variant allele frequency (VAF) or greater in blood cells compared to the tumor. Mutations were scored as hematopoietic in origin if they were detected with a VAF greater than 2% for a selected panel of leukemia genes and 5% for other genes. When at least one mutation exceeded these thresholds, we reduced the VAF threshold in blood to 1% to detect subsequent events for leukemia genes and 3% for non-leukemia genes. Mutations where VAF in blood and tumor were both greater than 35% were excluded as likely germline events except in cases where the blood-associated variants were seen in the blood at >35% for mutations reported in COSMIC on at least 10 occasions.Common SNPs were excluded. We excluded patients with concurrent hematologic malignancies.We examined whether different mutations harbored unique phenotypes in regard to their effect on hematologic parameters and associations with prior toxic exposuresin a subset of 5,145 patients. Results: CH was identified in 23.4% of patients. CH was associated with increased age (p<0.001), prior radiation therapy (p<0.001) and tobacco use (p<0.001), but not with prior chemotherapy (p=0.85). The most commonly mutated genes were DNMT3A , TET2 , PPM1D , ASXL1 , TP53 , ATM , CHEK2 , CBL , SF3B1 , and JAK2 (see table). All genes were associated with increased age, except for mutations in JAK2 . DNMT3A mutations were associated with increased WBC count, ANC, ALC, and AMC. TET2 mutations were associated with increased WBC count, AMC, and MCV. TET2 mutations were also associated with increased neutrophil to lymphocyte ratio (NLR), suggestive of myeloid bias. JAK2 mutations were associated with an increased platelet count. Of the 35 JAK2 mutations identified in entire cohort, 13 occurred in patients with a known diagnosis of hematologic malignancies and were excluded from our analyses. Median VAF of JAK2 mutations in patients with known malignancies was 26% compared the median VAF of 3.1% in patients without a known hematopoietic malignancy. All 13 patients with known hematologic malignancies harbored the V617F mutation; however, this mutation was observed only in 14 of the 22 patients without known hematologic malignancies. DNMT3A , TET2 , ASXL1 , and TP53 mutations were all associated with current/former tobacco use, consistent with an important role for smoking in selecting for mutation-driven CH. DNMT3A , TET2 , ASXL1 , ATM, were not significantly associated with prior chemotherapy or radiation therapy (RT). CHEK2 mutations were associated with prior chemotherapy but not with prior RT. PPM1D and TP53 mutations were both associated with prior chemotherapy and RT. We examined a subset of CH mutations in presumptive driver genes (CH-PD) occurring at VAF>10%. CH-PD led to an increased risk for subsequent hematologic cancers (p<0.001) in all patients and predicted for inferior OS (HR= 1.38, p=0.025) in patients aged 65 and older. No difference in OS was seen in patients under 65. Conclusions: Individual CH mutations exhibit unique phenotypes in regard to their effect on hematologic parameters and associations with prior toxic exposures. JAK2 mutations were associated with increased platelet counts even after excluding patients with a known diagnosis of hematologic malignancies. TP53 and PPM1D mutations demonstrated the strongest association with prior chemotherapy and radiation exposure. CH-PD is associated with increased risk for subsequent hematologic cancers and inferior OS in patients 65 and older. ![Figure][1] Disclosures Levine: Qiagen: Membership on an entity's Board of Directors or advisory committees; Novartis: Consultancy. [1]: pending:yes