Characterization of the germline and somatic mutation profile of diamond-blackfan anemia

Abstract Diamond-Blackfan Anemia (DBA) is a ribosomopathy characterized by isolated hypoplastic anemia, phenotypic abnormalities and cancer predisposition. In most cases, DBA is caused by heterozygous germline pathogenic variants (PGV) in small (RPS) or large (RPL) ribosomal subunits; however, its germline etiology is unknown in approximately 30% of patients. Compared with other inherited bone marrow failure syndromes (IBMFS), clonal hematopoiesis (CH) is not well characterized in DBA. We assessed 127 DBA patients from 3 centers in USA and Brazil with chronic anemia, reticulocytopenia, reduced/absent bone marrow erythroid precursors and at least one of: 1) presentation by the third year of life, 2) known DBA-associated PGV, and/or 3) unknown PGV but family history consistent with DBA. We examined complete blood counts, iron parameters and bone marrow morphology. We designed a comprehensive error-corrected sequencing panel with integrated copy number variant (CNV) detection (sensitivity of 0.5%) targeting 181 genes related to DBA, myeloid neoplasms (MN), erythropoiesis and ribosome biology. CH was screened in patients' granulocytes (NHLBI) or total leukocytes (NCI and USP) at last visit. Median age was 18 years (0-69), and 43% of patients were female. Median age (range) varied by institution (p<0.001): USP patients (6 y [0-40y]) were younger than those from NCI (25 y [6-69y], p<0.0001) and NHLBI (27 y [2-58y], p<0.001). These differences likely reflect disparities in healthcare infrastructure and timing of implementation of iron chelation therapy in the NCI/NHLBI cohorts. Phenotypic abnormalities were present in 39% of patients (cardiac [21%], limb/thumb [20%] and genitourinary [9%]). Neutropenia occurred in 19%, thrombocytopenia in 4%, iron overload in 46%, and 52% were on chelation therapy. Germline RPS/RPL variants were detected in 85.4% of patients, most commonly in RPS19 (29.6%), RPS26 (10.4%), RPS29 (8.0%), RPL11 (7.2%) and RPL5 (6.4%). Novel PGVs or gene deletions were detected in 11 patients with a previous unknown genotype, most commonly in RPS26, RPS17, and RPS19. Two patients meeting DBA clinical criteria had PGVs in ADA2 (CECR1): 1 with a homozygous splice-site variant, the other with a heterozygous duplication of exons 2–8 with absent plasma ADA2 levels. Leukocyte, neutrophil and platelet counts correlated with prevalent genotypes (RPS19, RPS26, and RPS29): patients with “other variants” had lower leukocytes than RPS19 and RPS26 and lower neutrophils than RPS29, whereas RPS26 had higher platelets than both RPS19 and RPS29. CH was detected in 9/109 (8%) of NCI/USP patients (median age of 16y) at median variant allele frequencies (VAF) of 2% (range: 0.6-25%); no recurrently mutated gene was found. TP53 mutations were seen in two patients at VAFs 3%. CH associated with aging was detected in patients >17 y (median [range]: 35 [1-50]). In the NHLBI cohort, CH frequency was higher (4/25; 16%), likely due to use of granulocytes for screening and was detected in patients with median age of 35y [27-42y]. These findings are significantly different from Shwachman-Diamond syndrome, another ribosomopathy, where CH was detected in >70% of patients at median age of 10y (Kennedy et al, 2021). Additionally, 2 patients were detected with RPS19 p.E32X (VAF 29%) and RPL26 c.310-16_353del (VAF 30%), suggestive of possible somatic reversion (germline testing is underway for confirmation); the RPL26 patient had spontaneous remission (steroid and transfusion independent) during follow-up (age 5y). Somatic mutations were also found in non-classical myeloid cancer genes, such as HSPA1A, TFRC, BAX, and EIF5B. CH did not correlate with signs of myeloid neoplasia (bone marrow dysplasia or increased blast counts) or cytopenia at last follow-up. Unlike other IBMFS, CH frequency in DBA is low, even when enriched granulocytes are used for screening; CH was more common in older patients. These findings highlight the unique DBA pathophysiology and may explain the low MN incidence in these patients. Integration of CNV analysis in our panel facilitated molecular diagnosis of 6% of patients with large RPS/RPL gene deletions. Although rare, revertant CH in DBA may correlate with clinical presentations. Ongoing longitudinal follow-up and functional studies will be essential to elucidate the clinical impact of these somatic events in DBA.