Sickle cell disease is associated with early-onset clonal hematopoiesis involving DNA damage response pathway mutations

Abstract

Abstract Background: Individuals with sickle cell disease (SCD) face an elevated risk of myeloid leukemias. Recently, myelodysplastic syndrome and acute myeloid leukemia have emerged as complications of curative SCD therapies, including gene therapy and allogeneic hematopoietic cell transplantation (HCT). Leukemias arising in SCD have been reported to harbor somatic TP53 mutations, and post-HCT TP53-mutant leukemias have been traced to low-level TP53 clones detectable pre-HCT. These findings suggest that SCD itself may predispose patients to high-risk clonal hematopoiesis (CH). Prior studies of CH in SCD used sequencing methods with limited sensitivity, yielding conflicting conclusions and potentially missing small, clinically relevant clones. In this multinational cohort, we defined CH prevalence, age distribution, and mutational profiles in SCD relative to non-SCD controls and other hemoglobinopathies. Methods: We analyzed archived blood DNA from 7,283 individuals across 17 cohorts in 4 countries: 3,885 with SCD (SS, SC, Sꞵ0, Sꞵ+), 3,398 without SCD (AA, AS, AC), and 188 with beta-thalassemia. Using duplex sequencing, we identified somatic CH variants at ≥0.001 variant allele fraction (VAF), germline variants in leukemia predisposition genes, and HBB genotypes. CH was analyzed by gene and in pre-specified biological groups: DNMT3A/TET2 (DT-CH) and DNA damage response (DDR-CH: TP53, PPM1D, CHEK2, ATM). We used binomial logistic regression (age- and sex-adjusted) to compare the prevalence of CH in SCD vs non-SCD controls. Results: We detected 6,661 CH variants in 2,673 individuals (median VAF=0.002). CH occurred earlier in SCD and was more prevalent in SCD cases compared to non-SCD controls among those aged 0-19 years [10.6% (95% CI: 9.1, 12.2) vs 3.5% (2.4, 5.0); p <0.0001]. This was driven by a selectively increased prevalence of DDR-CH in SCD [SCD: 3.3% (2.5, 4.3) vs non-SCD: 0.6% (0.2, 1.5), p = 0.0012] which extended across adult age groups (20-29 years: 3.5% vs 0.9%; 30-39 years: 7.7% vs 1.8%; 40-49 years: 15.1% vs 5.5%). Within DDR-CH, PPM1D was enriched in SCD compared to controls (36.4% vs 21.8%; p<0.0001); ATM (4.9% vs 9.5%; p=0.016) and CHEK2 (26.7% vs 34.1%; p=0.03) were underrepresented; and TP53 was similar (32.0% vs. 34.7%; p=0.45). To further evaluate the onset of CH in children with SCD, we performed serial sequencing of 148 participants enrolled in the BabyHUG trial (age 0.6 to 1.4 years) with follow-up samples obtained between 3 and 11 years of age. We detected CH, including DDR-CH, in 4.7% of children at baseline, all of which persisted in subsequent samples. Among those without CH at baseline, 3.9% developed incident CH during follow-up. In contrast to DDR-CH, DT-CH prevalence was higher in SCD among the youngest population [0-19: 6.4% (5.2, 7.7) vs 2.5 (1.5, 3.9), p<0.0001] but progressively decreased with advancing age relative to non-SCD controls. Among older individuals, (age ≥50 years), the prevalence of DT-CH was significantly lower in those with SCD than in those without SCD [54.4% (47.7, 61.0) vs 76.9% (74, 79.6), p<0.0001]. To determine whether sickle cell trait also had increased CH prevalence, we compared individuals with AA to those with AS/AC genotypes. CH prevalence was not higher in AS/AC compared to AA: overall CH (OR 1.01, p=0.95), DDR-CH (OR 0.99, p=0.50), DT-CH (OR 0.92, p=0.38). Then, to evaluate whether the association of SCD with early-onset CH was generalizable to other beta hemoglobinopathies, we analyzed the CH prevalence in pediatric beta-thalassemia patients (n=166). CH prevalence in beta-thalassemia was similar to AA controls (OR 1.37, p=0.49) and lower than SCD (OR 0.41, p=0.027). We observed no DDR-CH in beta-thalassemia. The prevalence of pathogenic/likely pathogenic germline variants in leukemia predisposition genes, such as DDX41, TERT, GATA2, and RUNX1 was similar across all evaluated HBB genotypes.Conclusions: Using deep targeted sequencing, we demonstrate that SCD is associated with a predisposition to early onset high-risk CH. Individuals with SCD exhibit markedly increased prevalence of DDR pathway mutations compared to controls, with the earliest clones detectable in infancy. This precocious DDR-CH is specific to SCD and not observed in individuals with sickle cell trait or beta-thalassemia. These findings provide a plausible mechanistic basis for the elevated relative risk of myeloid leukemias in SCD and therapy-related leukemias as a complication of curative therapies.