Are the current guidelines for identification of myelodysplastic syndrome with germline predisposition strong enough?

Abstract

Myelodysplastic syndrome (MDS) is a malignant condition characterized by ineffective haematopoiesis that typically presents in older adults (73.5 years on average) with a stepwise process driven by the acquisition of age-related somatic mutations. Conversely, MDS with germline genetic predisposition is more frequently encountered among children and younger adults. Recently, it has been estimated that germline mutations could underlie at least 15% of both adult and paediatric MDS cases.1 In recognition of such a growing number of MDS associated with germline predisposition or familial cases, the classification of the World Health Organization (WHO) included the entity of myeloid neoplasms with germline predisposition,2 now endorsed by both the revised 5th WHO3 and the International Consensus Classification of Myeloid Neoplasms and Acute Leukaemias (ICC).4 Mutations in several myeloid genes have been described as predisposing factors to MDS.5 For instance, germline hits of TP53, SAMD9, SAMD9L, SRP72 and TERT along with other myeloid genes such as DDX41, GATA2, ANKRD26, ETV6, CEBPA, ASXL1 and RUNX1 have been associated with MDS development,6 which is also described in certain inherited bone marrow failure syndromes (BMF).7 Accurate identification of patients with germline predisposition to MDS is mandatory for an appropriate medical management. This is consequential to prevent unnecessary or even harmful medication, but also for genetic counselling of family members at risk or the selection of donors for haematopoietic stem cell transplantation (HSCT) purposes.8 However, there is no consensus for the identification of these patients in the clinical setting. To this end, several groups are proposing clinical guidelines for the evaluation and early detection of MDS patients carrying germline mutations.9 The most established consensus for diagnostics, surveillance and management (including considerations for HSCT) for MDS patients with germline predisposition was described by the Nordic guidelines.10 In general, germline predisposition to MDS can be suspected when patients younger than 50 years harbour mutations in known predisposition genes detected at a high clonal burden (variant allele frequency, VAF > 30%). These patients also appear to be more frequently affected by somatic monosomy 7. More controversial is the assignment of biologic importance to variants of unclear significance, particularly in special circumstances such as compound heterozygous or lesions in genes usually affected by dominant alterations. However, the increasing application of genome scanning approaches led to the discovery of a high frequency of pathogenic/likely pathogenic (P/LP) germline variants also in MDS patients diagnosed in older adulthood. Recently, two studies, Feurstein et al.11 and Kubota et al.12 identified P/LP germline variants in 5%–12% of the studied MDS patients within all age deciles including cases aged older than 50 or even 60 years, thereby advocating for a comprehensive germline genetic testing in all patients regardless of their age at presentation.11 While clinical outcomes did not correlate with the presence of P/LP germline variants in the first study, the larger study by Kubota et al. (N = 2658) found a negative impact of these alterations on patient outcomes.12 Not only heterozygous carriers of recessive traits were found more frequently among MDS patients, but also the frequencies of compound heterozygous alterations appeared to be increased in Kubota et al.'s study, suggesting that complex genetic traits may play a role. In light of the conclusions suggested by these two recent studies11, 12 we investigated the germline status of candidate variants found in bone marrow (BM) samples in our MDS patient cohort using paired non-clonal DNA. A series of BM samples from 168 consecutive patients with primary MDS and available data from Targeted Next- Generation Sequencing (tNGS) were selected for this study. Patients were previously sequenced by using a custom gene panel of 50 genes reported in myeloid neoplasms13 (Figure 1A). Patients with variants in genes associated with germline MDS predisposition (ASXL1, ANKRD26, CEBPA, DDX41, ETV6, GATA2, RUNX1 and TP53) with a VAF greater than 30% and minor allele frequency (MAF) lower than 0.02% were selected. Synonymous and intronic variants, as well as those annotated as polymorphisms in the single nucleotide polymorphism database (dbSNP), were excluded. Considering the results shown by the afore-mentioned reports,11, 12 age at onset or family history criteria were not taken into account for the purpose of this study. CD3+ lymphocytes from blood were used as a reliable source of germline control as previously demonstrated.14 Overall, 21 variants from 18 patients were selected and validated in CD3+ lymphocytes. Eleven variants were studied by tNGS (tNGS series), while 10 variants were validated by Sanger sequencing (Sanger series). In addition, 31 variants from 21 patients studied by paired whole-exome sequencing (WES) (both BM and CD3+ samples) were also included in this study and fulfilled the same inclusion criteria (Figure 1A). In summary, 52 variants found in BM samples from 39 patients were selected (Figure 1A). Of these, 34 were confirmed in germline tissue and 18 were not found in CD3+ lymphocytes and were thus considered somatic (Table 1). Average VAF of germline and somatic variants in BM samples were 50.30% and 47.23% respectively (Figure 1B) with no significant differences (p = 0.35). Therefore, variants with VAF higher than 30% not always a germline origin and, significantly, are indistinguishable from the somatic counterparts. Sanger tNGS IJC IJC We then compared the number of germline and somatic variants per 5% intervals of VAF and found an opposite trend (Figure 1C). Indeed, the number of germline variants increased at higher VAF intervals, whereas the number of somatic variants had a negative correlation with VAF increase. Notwithstanding, germline and somatic variants were observed throughout all VAF ranges (Figure 1C). Thus, despite opposite trends, both types of variants overlapped in all intervals of VAF higher than 30%. Even more, the number of variants was only found to be significantly different when the VAF was greater than 50% (Figure 1B). When comparing the percentage of somatic or germline variants annotated per gene, no differences were found for ASXL1 (Figure 1D). Interestingly, 11 out of the 18 somatic variants (61.11%) involved RUNX1 and TP53 genes versus only three out of the 34 germline variants (8.82%). Furthermore, the 16 variants detected in the ANKRD26 gene were confirmed as germline, accounting for almost half (47.06%) of the entirety of germline alterations. Somatic variants were neither found in GATA2 nor ETV6 genes, whereas only three germline variants were identified in these genes. No variants were found in DDX41. We then sought to explore whether patients with somatic or germline variants had differences as to phenotype, outcome, or predictive factors (Table 1). The MDS subtypes were categorized into three risk groups on the basis of the survival time and incidence of evolution to AML according to the revised international prognostic scoring system (IPSS-R).3 However, no differences were observed (Figure 1E). Remarkably, patients with germline or somatic variants did not differ with regard to age at disease onset (Figure 1F). Finally, IPSS risk stratification (IPSS-R) and cytogenetic risk were also compared within patients following the standard stratification scheme and assigning a karyotype score according to the number of cytogenetic alterations.15 However, no differences were found between patients harbouring germline or somatic variants (Figure 1F). In light of recent contributions, current boundaries for germline MDS are suggested to be unsuitable for the clinical practice. Such publications described that deleterious germline predisposition variants are shared by MDS patients of all ages11, 16 in contrast to the current guidelines, which consider that MDS arising from germline predisposition occurs only in younger patients. A similar observation was found for our series of patients, where no difference was found as to the age of onset of patients carrying germline versus somatic variants (Figure 1F). Thus, older age at presentation is a clinical criterion insufficient to rule out the presence of an underlying germline predisposition disorder. Furthermore, considering our findings we can conclude that a VAF higher than 30% threshold should neither be considered as irrefutable to identify germline predisposition either. Indeed, we found that 35% of the variants with VAF higher than 30% found in BM samples were somatic after validation in CD3+ lymphocytes (Table 1). Although the higher the VAF is, the higher the possibility to find a germline variant, overlapping VAFs between somatic and germline variants are expected in MDS patients (Figure 1B,C). In addition, the lack of knowledge regarding the complete genetic landscape driving germline MDS, and other genetic conditions such as low penetrance or polygenic models for described involved genes, complicates the strong determination and precludes the establishment of concise guidelines for diagnostics of germline predisposition disorders. Finally, somatic and germline variants were identified in different genes but no differences in prognostic or predictive factors were observed between patients carrying such alterations. In summary, genetic testing must be recommended for all MDS patients and variant validation in germline tissue should always be conducted at diagnosis to shed light on variant calling and determination of mutational ontogenesis. Our results together with other recent observations suggest that boundaries of age and VAF must be revised in order to better standardize guidelines for MDS with germline predisposition. Oriol Calvete and Francesc Solé designed the study and wrote the paper; Julia Mestre, Arda Durmaz, Carmelo Gurnari and Jaroslaw P. Maciejewski collected bioinformatic and clinical data; Oriol Calvete and Julia Mestre analysed data and performed statistical analysis; Julia Mestre, Carmelo Gurnari and Jaroslaw P. Maciejewski critically edited the manuscript. This work was supported in part by a grant from the Instituto de Salud Carlos III, Ministerio de Economia y Competitividad, Spain (PI 20/000531); TRANSCAN (AECC AC 18/000002 and ISCIII), Stiftung Carreras Foundation (2020), 2017 SGR288 (GRC) Generalitat de Catalunya; economical support from CERCA Programme/Generalitat de Catalunya, Fundació Adey and Fundació Internacional Josep Carreras. Carmelo Gurnari was supported by a grant from the Edward P. Evans Foundation. The authors declare no conflict of interest.