Genomic variation in <scp><i>DDX41</i></scp> identified through clinical sequencing

To the Editor, Germline pathogenic variants in the DEAD-box helicase 41 (DDX41) gene are the most common inherited cause of predisposition to myeloid neoplasms (MNs) in adults, accounting for approximately 80% of cases where a germline predisposition is identified.1, 2 Myelodysplastic syndrome (MDS) and acute myeloid leukaemia (AML) are the most well-established MNs associated with pathogenic germline DDX41 variants, demonstrating incomplete penetrance with a median age of disease onset of approximately 65 years.2-4 The identification of patients with pathogenic germline DDX41 variants is of critical clinical relevance, with potential impacts on disease prognosis, predictive testing for family members, appropriate donor selection for allogeneic haematopoietic stem cell transplantation (HSCT) and appropriate surveillance for at-risk individuals and those in remission.1, 5 Since the initial descriptions of germline DDX41 pathogenic variants, multiple studies have highlighted regional variation in DDX41 variants.1, 2, 6, 7 In addition, emerging data reveal an increasing diversity of genomic variants causative of DDX41 germline disease including synonymous variants and larger structural variants.6 Taken together, these data suggest that a significant amount of clinically relevant DDX41 genomic variation remains to be discovered. We aimed to describe the comprehensive genomic assessment of DDX41 variants detected through clinical next-generation sequencing (NGS) from a molecular diagnostic laboratory to potentially discover novel variants and expand upon the understanding of DDX41 genomic variation. Three hundred and twelve patients from 257 unrelated kindreds with clinically reported DDX41 variants were identified from samples referred for NGS testing (Supplementary Methods) between November 2021 and November 2024. Patients were referred as part of the investigation or management of suspected or established haematological malignancy or predictive testing of family members. Among the 257 kindreds, 85.6% (n = 220) had confirmed or assumed (variant allele frequency [VAF] >40%) germline DDX41 variants, of which germline origin was confirmed in 37.3% (n = 82) of probands through the analysis of DNA extracted from hair follicle samples, or via the presence of the variant in a relative. 11.7% (n = 30) had confirmed or assumed somatic-only DDX41 variants (VAF <35%), and in 2.7% (n = 7), the likely origin of the variant could not be determined (Figure S1, Table S1). Of the 257 probands, the median age at testing was 69 years with a male predominance (68.9%). The most common indication for testing was an established diagnosis of MDS (33.9%, n = 87) or AML (32.7%, n = 84). Other indications included investigation of non-myeloid haematological malignancies (9.3%, n = 24), clonal cytopenia of undetermined significance (CCUS) (6.6%, n = 17), myeloproliferative neoplasm (MPN) (1.6%, n = 4) and MDS/MPN overlap syndromes (3.1%, n = 8) (Table S1). Germline DDX41 variants were classified according to the American College of Medical Genetics and Genomics (ACMG) guidelines8 with gene-specific modifications (Supplementary Methods). Of the 220 probands identified with an assumed or confirmed germline DDX41 variant, 71.4% (n = 157) had variants classified as pathogenic (P) or likely pathogenic (LP) and the remaining 63 probands had variants classified as variants of uncertain significance (VUS). Within the probands harbouring LP/P variants (n = 157), 43 unique DDX41 variants were observed (Figure 1A; Table S1). The four most frequently observed variants were NM_016222.2:c.3G>A; p.? (n = 32; 20.4%), Asp140Glyfs*2 (n = 29; 18.5%), Met316Aspfs*31 (n = 13; 8.3%) and Gly173Arg (n = 12; 7.6%). Two structural variants were observed: one large deletion with exonic breakpoints in exons 14 and 17 (c.1536_1774del, Glu512Aspfs*3) and one complex structural variant (hg19(chr5):g.176943186_176951249delins[TGG;176958663_176958740inv]) involving the loss of exon 1 to part of exon 4. Among the 43 unique LP/P DDX41 variants, 10 were previously undescribed in the literature or ClinVar (NCBI; National Institute of Health, USA) to our knowledge (Figure 1B; Table S2). Of these 10 novel variants, seven were nonsense/frameshift mutations (including the aforementioned complex structural variant (hg19(chr5):g.176943186_176951249delins [TGG;176958663_176958740inv])), two were splicing variants (c.798+1G>A and c.1302+5G>C) and one was a missense variant occurring in the DEAD-box domain (Asp220Tyr). RNA was available for the patient harbouring the c.798+1G>A variant, and RNA-sequencing demonstrated evidence of intron retention accompanied by nonsense mediated decay (Figure S2). Of the 63 probands with assumed or confirmed germline DDX41 VUSs, the vast majority carried missense variants (n = 57) (Figure S3; Table S1). The most frequently observed VUS was Met155Ile (n = 6), which was observed in conjunction with a somatic Thr227Ala DDX41 variant in two unrelated patients within our cohort. Among the 157 probands with germline LP/P DDX41 variants, 76.4% (n = 120) had either one (n = 109) or two (n = 11) somatic DDX41 co-mutations. All observed somatic co-mutations were missense variants, the most common being Arg525His, identified in 65.8% (n = 79) of patients with somatic DDX41 co-mutations. The most frequently observed non-DDX41 co-mutation was in ASXL1, present in 19.1% of probands with LP/P DDX41 variants, followed by DNMT3A (18.5%), SRSF2 (12.1%), TP53 (9.6%), CBL and EZH2 (7.6%), TET2 (7.0%) and RUNX1 (5.7%) (Figure 1B). Similar to previous studies,9, 10 the allele frequency of the somatic DDX41 variants was noted to be of relatively low VAF (median VAF for all somatic DDX41 variants was 6.31%; range 0.51%–46.27%). Interestingly, we noted differences in the spectrum of VAFs observed across different somatic DDX41 variants. In our cohort, the Pro321Leu variant was consistently observed at a higher VAF (n = 9, median 23.36%; range 0.57%–38.80%) compared to other commonly observed somatic DDX41 variants, including Arg525His (n = 102, median 5.84%; range 0.51%–45.53%) and Gly530Asp (n = 19, median 6.90%; range 0.78%–24.88%) (Figure 2A). All nine patients with Pro321Leu somatic variants had concurrent germline LP/P (n = 8) or VUS (n = 1) DDX41 variants. Interestingly, the relatively higher Pro321Leu VAF observed in our dataset is also observed in other study cohorts6, 7 (Figure 2B). Twenty-five asymptomatic carriers of LP/P germline DDX41 variants without haematological abnormalities (typically identified through predictive testing of family members) underwent longitudinal testing (Figure 2C). The median age of this cohort at first testing was 56 years, and median follow-up duration was 14 months (range 6–32 months). No asymptomatic carriers acquired new somatic DDX41 variants though two acquired somatic non-DDX41 variants, both at a low VAF. Twenty patients with established malignant or premalignant conditions (MDS [n = 14], CCUS [n = 3], AML [n = 1], non-myeloid [n = 1], other [n = 1]) underwent longitudinal testing (Figure 2C). Of these patients, 15% (n = 3) progressed during monitoring with either worsening MDS severity (n = 2) or progression to AML (n = 1). Progression in these three patients was associated with worsening cytopenia and/or morphological changes on histology. No molecular changes in DDX41 or non-DDX41 genes, such as additional somatic hits or expansion of pre-existing clones, were detected in these three patients. One patient with clinically stable MDS demonstrated expansion of a pre-existing DDX41 somatic clone, from 10% to 20% VAF. In one patient with CCUS, a DDX41 Arg525His variant, previously only detectable on bone marrow aspirate, was detected in peripheral blood over an interval of 20 months, at a low VAF of 1%. These data provide multiple important insights into the spectrum of genomic DDX41 variation. We have described 10 novel variants in DDX41, comprising almost 25% of the LP/P DDX41 genomic variants detected in our cohort. These findings highlight the incomplete characterisation of pathogenic variation in DDX41 to date, underscoring the importance of ongoing study of diverse populations to facilitate comprehensive variant identification. The genomic variants discovered in this study include intronic variants outside canonical splice acceptor/donor sites (±1/2) as well as larger structural variants. It is essential that diagnostic laboratories recognise the potential for these types of variants as they may be overlooked with standard analytical/bioinformatic approaches. In our cohort, we have made the observation of a relatively higher VAF of the DDX41 Pro321Leu when occurring as a somatic variant. It has previously been established that the somatic Arg525His variant is predominantly observed at a low subclonal VAF.9, 10 While the precise reason behind this phenomenon is yet to be understood, our data suggest that not all DDX41 somatic variants are biologically equivalent and that certain variants may be associated with distinct biological characteristics. Although working group recommendations exist, there is no widely accepted approach to the surveillance of asymptomatic individuals with germline DDX41 variants.11, 12 In our cohort of asymptomatic family members, we did not observe clinically relevant molecular progression; however, study limitations that preclude definitive conclusions include small sample size, younger age and short follow-up duration of our cohort. Further follow-up of this cohort is necessary to clarify the rate and clinical relevance of molecular progression. Surveillance may also pose psychological and economic challenges, particularly given the relatively high prevalence of germline pathogenic DDX41 variants in the general population (~1/429) and their late and incomplete penetrance.13 The typically low penetrance of DDX41-associated MNs before age 601 combined with the potential psychological burden of carrier status and economic impact of routine screening underscores the need for further research to define optimal monitoring intervals and ensure effective and sustainable surveillance. In summary, we have described the comprehensive genomic analysis of DDX41 in the molecular diagnostic laboratory and have identified multiple novel genomic findings that have expanded our understanding of the spectrum of variation in DDX41. This detailed understanding is crucial for improving accurate variant curation, and subsequently guiding management, such as prognostication, cascade testing and surveillance and safe donor selection for HSCT candidates. PB designed the study, managed patients and wrote the manuscript. CW collected and analysed data and wrote the manuscript. IST analysed data and managed patients. NDE, EG, MT, SW and LCF managed patients and contributed data. SH, YK, KM and ERT contributed data. All authors reviewed and approved the final version of the manuscript. The authors gratefully acknowledge funding sources including the Wilson Centre for Blood Cancer Genomics and the Snowdome Foundation. Open access publishing facilitated by The University of Melbourne, as part of the Wiley - The University of Melbourne agreement via the Council of Australian University Librarians. The authors declare no conflicts of interest. Data S1. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.