Prospective Genomic Study for Optimizing Clonal Hematopoiesis Surveillance in Bone Marrow Failure Syndromes and Related Cytopenia

Introduction. Inherited bone marrow failure syndromes (IBMFS) and acquired aplastic anemia (AA) carry variable risks for myeloid neoplasms. Current surveillance protocols utilize serial peripheral blood (PB) counts and yearly bone marrow (BM) analysis, the latter often avoided due to its invasive nature and associated discomfort. This approach lacks standardization and misses acquired molecular lesions preceding overt malignancy, highlighting the need for more sensitive monitoring strategies. Clonal hematopoiesis (CH) commonly arises in IBMFS/AA to alleviate the intrinsic germline defects, serving as an early biomarker of disease progression. However, no consensus exists on implementing CH detection in routine surveillance. This study aims to evaluate the diagnostic utility of PB vs. BM for early CH detection and surveillance in IBMFS/AA, addressing the undetermined comparative efficacy of these analytical sources. Methods. From 05/2019 to 05/2024, we prospectively investigated 147 patients (pts) aged ≤25 years (yrs) with IBMFS, AA or unexplained cytopenia at the St. Jude Children's Hospital BMF referral center. For diagnosis, we used our standard workup including germline genetic testing. All pts underwent somatic analysis by myeloid malignancy NGS (68 gene panel) and single nucleotide polymorphism array (SNP-A) testing (ARUP laboratories), with a total of 313 NGS and 315 SNP-A tests. Paired BM/PB analysis from the same day using NGS/SNP-A (n=123) was performed in 59 pts. Longitudinal NGS/SNP-A was performed in 65 pts (44.2%), with a median interval of 12 (IQR 6-14) months (m) between timepoints. Results. The median age at diagnosis was 10.5 yrs (1 m-25 yrs), with a 0.9 M|F ratio and 661 person-years of longitudinal observations. Germline causative variants across 13 genetic syndromes were detected in 74 (50.3%) pts and somatic alterations were detected in 63 (42.8%) pts. In 59 pts with paired BM/PB tests, NGS results were concordant (BM=PB) in 87.4% and discordant, with detection only in BM in 11.5% and only in PB in 1.1%. For variant allele frequency (VAF) ≥5%, PB NGS analysis showed 92% clinical sensitivity, and 100% specificity compared to BM. Only 2 clones, both with BCORmut, were detected at ≥5% (6.5% and 8.2%) VAF in BM and missed in PB with a predetermined limit of detection of 3% for missense and 1% for indel variants. SNP-A results were largely concordant between BM and PB with two exceptions found only in BM: monosomy 7 (-7) at 5-10% and a uniparental disomy (UPD) 7q at 20% of clone size. SNP-A proved valuable for both diagnosis (identifying disease-specific alterations such as UPD6p in AA) and surveillance (high-risk -7 CH). In one case of unexplained cytopenia where multiple BM evaluations failed, SNP-A in PB revealed isochromosome 6p and trisomy 8. This prompted further testing ultimately leading to the diagnosis of KMT2A::MLLT10 rearranged leukemia. In a subgroup of pts with a confirmed diagnosis of IBMFS and AA (n=101), 77 CH events were detected in 42 pts (41.6%). These were categorized based on their potential impact: 54.2% somatic gene rescue CH, 35.8% indeterminate significance CH and 10% high-risk CH; the respective cumulative incidence by age 20 yrs was 15.3%, 20.4%, and 6.1%, respectively. SAMD9/SAMD9L syndromes exhibited the highest CH frequency (85.7%, 12/14 pts). Novel CH events were detected in Diamond Blackfan anemia, including subclonal UPD2p25 in germline RPS7mut and a somatic CBLmut in germline RPL15mut pts. Investigating longitudinal cohort (n=65), 3 cases (4.6%), with congenital neutropenia, ERCC6L2 syndrome and Shwachman-Diamond syndrome (SDS) progressed to leukemia. Importantly, in the SDS pt, interval PB analysis detected early molecular transformation (TP53mut VAF doubling via UPD17p) emerging within 6 months from previous test, which prompted earlier BM evaluation and timely transplant, highlighting the value of PB surveillance. Conclusions. Our results demonstrate the utility of integrating NGS and SNP-A analysis for diagnosis and disease monitoring in IBMFS/AA. PB-based genomic surveillance effectively complements existing methods, particularly for interval testing between yearly BM and when BM sampling is challenging. The ability to detect and track high-risk lesions like TP53mut, UPD17p, or -7 in PB, underscores its potential to enhance long term surveillance and to guide timely clinical interventions.