DISCUSSION
We report the results of detailed genomics studies on four families in which pairs of siblings developed myeloid neoplasms and, in particular, AML in 7 out of 8 cases. According to the current World Health Organization (WHO) classification of hematological neoplasms with germline predisposition and the last published recommendations from ELN for diagnosis and management of AML, we identified germline mutations both in genes with a well-established role in predisposing to the development of MDS/AML and genes with no defined role in this context. Indeed, CEBPA and DDX41 are included in myeloid neoplasms with germline predisposition without a pre-existing platelet disorder;FANCA is listed in the category of genes mutated in myeloid neoplasms associated with bone marrow failure syndromes; JAK2 is included in emerging disorders with germline predisposition; and, finally, LYST and ERBB4 have not been reported yet in such classifications. None of the identified mutations is currently annotated as pathogenic in cancer databases; however, they could all play an important role in hematological diseases development.
In particular, the variant identified in CEBPA is a truncating mutation and, importantly, it is located at the 5’-end of the gene (c.62, p.S21). N-terminal mutations in CEBPA are known to have a penetrance close to 100% of leukemia development; however, they correlate with a favorable prognosis. Due to the unavailability of AML cells, we could not assess if in our patients there was acquisition of a somatic CEPBA mutation; nonetheless, both siblings are currently in CR and under clinical monitoring. The frameshift indel identified inDDX41 tumor suppressor gene is currently annotated as VUS and it is not listed either in COSMIC or cBioportal. This mutation is expected to truncate the protein before its functional helicase domain and likely causes a loss of function. Truncating mutations in DDX41 have been shown to increase the risk of developing myeloid neoplasms and are associated with faster progression to leukaemia. Both our patients are males who, in presence of DDX41 mutations, are expected to develop myeloid malignancies more frequently than females. Moreover, patient ID3 harbored the hotspot mutation p.R525H in DDX41 , frequently acquired as somatic mutation in carriers of germlineDDX41 variants. No pathological DNA was available for the other sibling, who however developed a Ph+ CML.
The LYST gene is mutated in autosomal recessive mode in Inborn Errors of Immunity syndromes and, in particular, in Familial Hemophagocytic Lymphohistiocytosis (FHL) syndromes with Hypopigmentation: Chediak-Higashi syndrome and Hemophagocytic Lymphohistiocytosis (HLH). LYST is a lysosomal trafficking regulator and a key effector of cytotoxic granules’ biogenesis. It is involved in the modulation of cytotoxic T-lymphocytes (CTL) and natural killer (NK)-cell functions by regulating degranulation. LYST-deficient CTLs and NK-cells display impaired ability to kill target cells and accumulate giant cytotoxic granules. Patients with Chediak-Higashi syndrome display oculocutaneous albinism, easy bruising, recurrent pyogenic infections and exhibit abnormal functions of NK-cells and alterations in neutrophils, leading to neutropenia. We have no evidence to infer a direct role for the identified LYST mutation in predisposing to development of myeloid neoplasms; however, we can envision a possible role for a deregulated immune system in controlling the homeostasis of hematopoietic differentiation.
In the fourth family we scored a more complex landscape with three germline VUS shared by both siblings. The most noteworthy is mutation p.S858R in FANCA gene. It is annotated as VUS in cancer databases; however, notably, it is annotated as pathogenic in FAMutdb, a database of variants identified in Fanconi Anemia (FA) (http://www2.rockefeller.edu/fanconi/) and it is reported in several FA patients of different origins (Italian, German, Indian-Jewish). FA is an autosomal recessive disease usually associated to mutations of other FANC members for its manifestation. The penetrance and phenotypic manifestations of the syndrome are highly variable. Our patients had no signs of FA and we did not identify pathogenic mutations in other sequenced members of FANC gene family. However, carriers of heterozygous FA mutations present increased risk for development of MDS and AML. Therefore, we can envision an incompletely penetrant phenotype imposed by p.S858R FANCAmutation, which requires cooperation with other germline lesions.
Germline predisposition to myeloid neoplasms due to pathogenic or likely pathogenic variants of JAK2 gene are emerging as new disorders; however, in both siblings of Family4, we identified a germline variant currently annotated as VUS: p.G571S. This mutation is located in exon 13 and, as the most common oncogenic variant in JAK2 , p.V617F, within the region encoding for the autoinhibitory JH2 pseudokinase domain of the protein. Although the biological significance of this variant is not well established yet, in vitro assays in Ba/F3 cells suggest no significant impact on the JAK2 protein functions. This variant has been reported associated to MPN with a frequency of around 0.01% and as germline both in sporadic and in familial cases of Essential Thrombocythemia (ET),. Moreover, it is listed in COSMIC (COSM29107, COSM142855: 10 mutations, 7 of which in the Haematopoietic and lymphoid category).
Finally, the missense variant in the receptor tyrosine kinaseERBB4 has an unknown biological significance and it has not been reported in COSMIC. We currently have no clues on its possible role in our clinical context.
Although functional characterization of these mutations is still required, we speculate that in Family4 the full MDS/AML phenotype may result from the cooperation between the FANCA and JAK2missense mutations. Interestingly, in a WES analysis of mutations in a cohort of patients with BM failure syndromes of suspected inherited origin, 11.6% of patients carried 2 co-occuring potential alterations.
In Family4, in siblings sharing this same complex germline landscape, the clinical history of the disease followed independent pathways, with acquisition of independent somatic mutations except for a common prevalent clone (VAF>30% in both siblings) with p.K700E mutation in SF3B1, suggesting a selective pressure imposed by the germline variants on acquisition of this somatic mutation. In a cohort of 16 FA patients, this same mutation in SF3B1 was identified in a patient with a germline FANCA mutation, who developed refractory anemia with ring sideroblasts and mutations in SF3B1and JAK2 seem to co-occur in myelodysplastic/myeloproliferative neoplasms with ring sideroblasts and thrombocytosis ,. Notably, clonal evolution reconstruction by single cell sequencing in a FA patient harboring a mutation in FANCA , showed appearance, in the early stage of MDS, of a clone with the SF3B1 p.K700E mutation, together with a mutation in RUNX1 , which expanded to become dominant with progression of the disease to AML.
Currently, a field of intense investigation is the study of clonal hematopoiesis (CH), a premalignant state in which hematopoietic stem and progenitor cells clonally expand due to acquisition of somatic mutations in genes that confer selective growth advantages. CH has been associated to increase risk of development of a number of diseases that include both hematological tumors and non-malignant conditions such as ischemic cardiovascular, inflammatory and autoimmune diseases. The analysis of CH is performed on peripheral blood in several clinical contexts, independent from hematological disease, by NGS that include both WES and targeted gene panels. The search for CH in an ever-growing number of non-hematological patients will generate a critical mass of genomic data on the hematopoietic system that could prove extremely helpful in the field of germline predisposition to myeloid neoplasms, significantly increasing the wealth of information available for this traditionally under investigated cancer type.
In conclusion, we show that thorough genomic analysis using large gene panels appositively targeted on cancer predisposing genes allow identification of novel germline variants. Indeed, in each family of our cohort, we identified at least one novel variant, affecting also genes not included in the current ENL guidelines for AML management. According to The American College of Medical Genetics and Genomics (ACMG) guidelines for interpretation of sequencing variants, a critical component for understanding of significance of a VUS is the observed clinical phenotype. On these basis, we suggest reclassification as pathogenic of the likely pathogenic p.S21Tfs*139 in CEPBA and the VUS p.K392Afs*66 in DDX41 . Finally, considering that myeloid neoplasms with germline predisposition often display clinical and morphological characteristics similar to sporadic cases and that age at diagnosis in the two groups often overlap, it is really challenging to suspect familiarity in the context of hematological tumours. Indeed, based exclusively on their clinical parameters, none of the cases of our cohort would have been suspected of familiarity. However, studying affected siblings, we identified potential germline predisposition in each family. Our data underline how current clinical practice underestimate familial cases within hematological neoplasms and calls for implementation of novel clinical practices that should include thorough reconstruction of personal and family history and genetic testing with large gene panels targeted for predisposing genes.