Manual review of select hotspots may unmask clonal hematopoiesis variants at low levels in at-risk pediatric patients undergoing initial bone marrow sequencing

Abstract Background: In hematopoietic stem cells, the acquisition of recurrent somatic mutations in select genes is linked to the development of clonal hematopoiesis (CH). CH can lead to complications later in life resulting in both malignant (e.g. increased risk of myelodysplastic syndromes and leukemias) and non-malignant sequelae (e.g. cardiovascular disease and thrombosis). Various exogenous stresses can precipitate CH including chronic infection/inflammation and chemotherapeutics. Moreover, the incidence of CH is increased in at-risk pediatric populations including patients with aplastic anemia, bone marrow failure syndromes, or other inherited disorders that induce proliferative stresses on the marrow elements. The incidence of pediatric CH depends on the time at which samples are assessed and the variant allele frequency (VAF) cut-offs employed. A next-generation sequencing (NGS) VAF cut-off point of ≥2% has been proposed as a criterion for CHIP per the World Health Organization although VAFs of ≥0.5% have also been proposed (PMID: 31672865). In this study, we sought to determine the frequency of CH-related hotspot mutations in at-risk pediatric patients at our tertiary care academic pediatric hospital at their first bone marrow assessment using various VAF cut-offs. Methods: Sequencing data of 465 patient samples that underwent initial targeted bone marrow NGS analysis for non-leukemic indications from 2019-2025 were included in the analysis. Based on published gene lists and CH mutation frequency data in children, 67 CH-related codon hotspot gene alterations in ASXL1 (4), CBL (3), DNMT3A (1), GNAS (1), IDH1 (1), IDH2 (2), JAK2 (2), KRAS (7), NRAS (6), SF3B1 (5), SRSF2 (1), STAT3 (5), TET2 (6), TP53 (20), and U2AF1 (3) were manually reviewed for sequence variants. A minimum of ≥3 variant reads lacking strand bias at ≥1.0% VAF with at least 200x coverage was considered a positive variant call. Results: Samples from 465 patients (0.1-25.5 years; mean: 10.3; median: 10.7) were included in the study. Using a VAF cut-off of ≥1.0%, we identified 62 CH variants in 50 patient samples (10.7%). Mutations were detected in TP53 (26), NRAS (9), DNMT3A (5), KRAS (4), CBL (4), SF3B1 (3), STAT3 (2), TET2 (2), ASXL1 (2), IDH1 (1), IDH2 (1), and SRSF2 (1). Ten patients (2.2%) harbored multiple CH variants; two patients with 3 and eight patients with 2 CH variants. CH variants were found frequently in marrow assessments for cytopenias (16), severe aplastic anemia (SAA) (9), and Shwachman-Diamond syndrome (SDS) (8). Nearly all CH variants detected in SDS marrows were TP53 variants (9/10, 90.0%) with two distinct TP53 variants detected in two patients. Using a VAF cut-off of ≥1.5%, the positivity rate decreased to 2.8% (13) with 5 patients harboring multiple CH variants; one patient with 3 and four patients with 2 CH variants. CH variants were detected in patients with SDS (7), SAA (3), dyskeratosis congenita (1), SAMD9-related bone marrow failure (1), and cytopenias (1). Using a VAF cut-off of ≥2.0%, the positivity rate decreased further to 1.5% (n=7); 6 SDS patients, 2 with multiple CH variants, and one patient with cytopenias and multiple variants. A majority of patients (415, 89.2%) harbored no detectable sequence variants in the positions assessed. Conclusions: Select hotspot variants in CH genes were manually reviewed to unmask variants at low levels in approximately 1.5% to 10.7% of patients depending on the VAF cut-off employed. TP53, DNMT3A, KRAS, NRAS, and CBL variants were the most frequently detected, although the frequency of TP53 variants may have been influenced partially by inclusion of patients with SDS where low-level clonal TP53 alterations are commonly observed. CH variants were enriched in patients with SDS, SAA, and cytopenias. By lowering the VAF cut-off from ≥2.0% to ≥1.0%, the detection rates increased nearly 5-fold. Likewise, the incidence of CH variants in pediatric patients may be higher than previously recognized, as this study was limited to 67 codons and a subset of genes/variants reported in CH. Thus, pediatric patients with these diagnoses who are at-risk for developing clonal malignancies may benefit from more sensitive screening methods at the time of their initial marrow assessments. Continued screening for additional CH variants in this cohort, confirmation of low-level variants to exclude false-positives, and correlation with serial samples and clinical outcome data is ongoing.