Germline predisposition to myeloid neoplasms: characteristics and management of high versus variable penetrance disorders

Abstract

Myeloid neoplasms with germline predisposition have been recognized increasingly over the past decade with numerous newly described disorders. Penetrance, age of onset, phenotypic heterogeneity, and somatic driver events differ widely among these conditions and sometimes even within family members with the same variant, making risk assessment and counseling of these individuals inherently difficult. In this review, we will shed light on high malignant penetrance (e.g., CEBPA, GATA2, SAMD9/SAMD9L, and TP53) versus variable malignant penetrance syndromes (e.g., ANKRD26, DDX41, ETV6, RUNX1, and various bone marrow failure syndromes) and their clinical features, such as variant type and location, course of disease, and prognostic markers. We further discuss the recommended management of these syndromes based on penetrance with an emphasis on somatic aberrations consistent with disease progression/transformation and suggested timing of allogeneic hematopoietic stem cell transplant. This review will thereby provide important data that can help to individualize and improve the management for these patients.

Introduction

As discovery and understanding of the numerous acquired genetic aberrations that drive development of myeloid neoplasms, such as myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML), have expanded over the past decade, so too have the discovery and characterization of germline predisposition to these same and other hematopoietic malignancies (HMs). Since the initial discovery of familial platelet disorder with propensity to myeloid malignancies (FPD/MM) due to pathogenic (P) or likely pathogenic (LP) germline RUNX1 variants in 1999 [1], there has been an exponential rise in the discovery, academic and clinical interest, and recognition of other germline predisposition disorders associated with heightened risk of HMs, due in large part to the advent of widely available next-generation sequencing (NGS) technologies.

Prior to undertaking NGS, appropriate tissue sample selection for the acquisition of germline DNA is important to ensure variants detected are truly germline in origin. In myeloid malignancies, such as MDS and AML the affected patients’ blood and bone marrow contain malignant cells and therefore may contain a mix of both somatically altered DNA in addition to germline DNA. To avoid misinterpreting a somatic variant as a germline variant, it is recommended to obtain tissue free from hematopoietic cell contamination, such as cultured skin fibroblasts or cultured mesenchymal stromal cells (MSCs). MSCs are adherent cells and can be obtained through tissue culture of whole bone marrow or the leftover cells within the bone marrow cytogenic tissue culture flask.

The importance and increased recognition of these syndromes is underscored by the first appearance of a “Myeloid Neoplasms with Germline Predisposition” chapter in the revised 2016 WHO classification of hematopoietic tumors [2]. Since the publication of this chapter, much progress has been made is this field regarding better phenotypic and molecular characterization of known disorders as well as the discovery of new “emerging” conditions. The expanded list and focus on myeloid neoplasms with germline predisposition in the 2022 National Comprehensive Cancer Network MDS guidelines, the recently published 2022 European Leukemia Net (ELN) AML recommendations, and others are examples of this greater understanding and recognition [3–7]. Germline predisposition variants to lymphoid malignancies and plasma cell dyscrasias have also been discovered and are areas of active research [8–10]. However, this review will focus on those which most strongly predispose to myeloid malignancies. It is important to recognize that germline predisposition is not lineage exclusive, and many syndromes can predispose to both myeloid and lymphoid malignancies as well as to solid tumors [11–13].

There is much heterogeneity among the various germline predisposition conditions in terms of clinical phenotypes, age of onset of disease, penetrance, expression, and the acquisition of somatic cytogenetic and/or molecular aberrations prior to or at the time of malignant transformation [14,15]. Deleterious germline variants in some genes, like those at the 5’-end of CEBPA, are nearly fully penetrant for development of AML. However others, such as those in RUNX1, have an incomplete malignant penetrance of ~44%, but near-complete penetrance for thrombocytopenia and platelet aggregation defects [16–18]. In the case of many other germline predisposition disorders, such as those associated with bone marrow failure (BMF) syndromes, telomere biology disorders, and deleterious variants in genes like ETV6, the precise risk for malignancy is often unknown, largely due to a scarcity of detailed published data, but is likely incomplete and variable across individuals [19]. The reason(s) for variability in penetrance among the different germline predisposition disorders is not fully elucidated. However, the necessity and propensity for acquisition of cooperating and myeloid malignancy-driving somatic variants prior to malignant transformation are likely key factors. Better understanding of the clinical phenotype, malignant penetrance, and somatic molecular landscape driving HM development in these conditions is necessary for proper genetic counselling, surveillance, and treatment recommendations. This review will focus on the differences in penetrance and somatic landscape for some of the better-known germline syndromes that predispose to myeloid neoplasms, such as MDS and/or AML, and provide recommendations surrounding molecular monitoring and timing of treatment for these disorders.

A. High penetrance disorders

Deleterious germline variants in particular genes, such as CEBPA and GATA2, are known to have an extremely high malignant penetrance with the vast majority of affected individuals ultimately developing a myeloid malignancy (Table 1) [20–22].

Table 1.

Characteristics of high penetrance germline predisposition disorders

Gene name	Inheritance	Malignant Phenotype	Malignant Penetrance	Median age of malignancy onset (yrs)	Common associated somatic alterations	Consideration for pre-emptive HSCT recommended?
CEBPA	AD	AML	90% (N-terminal variants); 46% (C-terminal variants)	25	CEBPA, GATA2, TET2	Yes
GATA2	AD	MDS/AML, HPV and EBV-associated tumors	75%	20	-7, +8, der(1;7), +21, STAG2, ASXL1, SETBP1, RUNX1, EZH2	Yes
SAMD9/9L	AD	MDS/AML	Near complete (SAMD9), 70% (SAMD9L)	9.6	-7, SETBP1, ASXL1, RUNX1, PTPN11, KRAS, CBL, EZH2, ETV6	No
TP53	AD	Adrenocortical carcinoma, breast cancer, central nervous system tumors, soft tissue and osteosarcomas, ALL, MDS/t-AML, many others	70 - ≥90% (any malignancy), 2–4% (hematopoietic malignancy)	25 (for any malignancy) [131]	Complex karyotype	No

Abbreviations used: AD, autosomal dominant; ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; EBV, Epstein-Barr virus; HPV, human papilloma virus; HSCT, hematopoietic stem cell transplant; MDS, myelodysplastic syndrome; t-AML, therapy-related AML.

CEBPA

CEBPA is a single exon gene that encodes CCAAT/enhancer-binding protein alpha, a myeloid transcription factor, which contains two N-terminal transactivating domains (TAD1 and TAD2) and a C-terminal DNA-binding region and basic leucine zipper (bZIP) domain [23]. The latter recognizes and binds to the CCAAT motif and acts as an activator on target genes. Germline predisposition to AML due to P/LP CEBPA variants was first described in 2004 [24]. We now recognize that in 7 – 15% of cases of AML with biallelic CEBPA variants, one of the two variants is germline in origin [20,25,26], typically the variant at the 5’-end of the gene, disrupting the N-terminal region of the protein. The second CEBPA variant is an acquired variant on the 3’-end of the wild-type allele, typically in-frame, within a region encoding the bZIP domain [20,26,27]. Although less common, there have also been several reports of deleterious 3’-end germline variants that alter the protein’s C-terminus and are typically less penetrant [28–31].

In addition to acquired CEBPA variants, acquired GATA2 variants are also common, found in approximately 30% of cases, followed by TET2 variants in nearly 20% of patients [27]. It is unclear whether this somatic profile differs from that of somatic biallelic CEBPA-mutated AML without a germline variant, as the origin (germline or somatic) of the CEBPA variants was not determined in the majority of reported cases that report on co-occurring variants.

5’-end germline CEBPA variants are associated with near complete (90%) penetrance for AML [20]. The development of AML typically occurs in young adulthood with a median age of 25 years, but can occur as early as infancy [20]. Unlike some other predisposition syndromes, there is currently no known association with MDS, pre-existing cytopenias, or other organ manifestations. The variants are typically frameshift or nonsense variants that disrupt the protein prior to the p30 isoform (NM_001285829) start codon [20,32]. Although the prognosis for these patients is favorable with an estimated 10-year overall survival of 67%, without allogeneic hematopoietic stem cell transplant (HSCT) AML ‘recurrence’ is very common [16,20]. In the majority of reported cases, AML ‘recurrence’ actually represents development of a new de novo leukemic clone with a distinct somatic CEBPA variant profile, rather than a true relapse of the original disease [16]. Consequently, the prognosis for these patients remains relatively good, with high rates of achieving a second complete remission in comparison to the poor prognosis for patients with truly relapsed disease [16].

In comparison to 5’-end variants, germline 3’-end variants have a lower penetrance, of approximately 46%, when pooling all currently published cases [31]. Variants are typically missense variants encoding amino acid substitutions within the bZIP domain. Like the 5’-end germline cases, patients typically present at a young age with a median age of 30 years (range 6 – 60 years), prognosis appears favorable, and AML ‘recurrences’ after chemotherapy alone, including development of second or even third de novo leukemias [28–31]. New de novo leukemias tend to occur sooner in germline 3’-end cases with a median time to first ‘recurrence’ of nine months vs 27 months for 5’-end cases [16,31]. Due to the small number of cases, the somatic landscape of germline 3’-end AML is less well characterized. However, among the reported cases, including ten individuals from eight different families for which somatic data were available; five had a CEBPA variant encoding changes in the bZIP domain, of which the majority were frameshift variants; three had a CEBPA variant encoding changes in the N-terminus; and three had a GATA2 variant [25,27,30,33,34]. Interestingly, the favorable outcome is associated with indel bZIP variants, whereas other somatic variants are no longer linked to a favorable prognosis according to the 2022 ELN guidelines [4].

GATA2

GATA2, located on the long arm of chromosome 3, encodes GATA2, a zinc-finger transcription factor that is essential for regulation of normal hematopoiesis [22]. GATA2 contains two zinc finger domains, the N-terminal ZF1 and the C-terminal ZF2. ZF2 binds DNA at GATA consensus binding motifs within regulatory regions of target genes. Initially described in 2011, heterozygous deleterious germline GATA2 variants are associated with a range of different phenotypes, including predisposition to myeloid malignancies, immunodeficiency, and lymphedema, now known collectively as GATA2 deficiency syndrome [35,36].

GATA2 deficiency syndrome often arises de novo (up to 80% of cases) [37], and thus, lack of a striking family history is common. The clinical phenotype can vary significantly between affected patients, including a variety of hematologic (cytopenias and myeloid malignancies), infectious (viral, fungal, bacterial, non-tuberculous mycobacterial), autoimmune, lymphedema, pulmonary, genitourinary, and neurologic phenotypes. Among these, myeloid neoplasms are very common, with a penetrance of approximately 75% in GATA2 variant carriers [37]. MDS is the most common myeloid neoplasm, although AML (second most common), chronic myelomonocytic leukemia (CMML), and myeloproliferative neoplasms can also occur. The median age for development of myeloid neoplasm is 20 years [37]. GATA2 deficiency syndrome accounts for a large proportion of pediatric and adolescent cases of MDS, previously identified in 7% of all pediatric MDS cases and 37% of those with MDS and monosomy 7 [37,38], whereas GATA2 deficiency syndrome in adult MDS is less common and associated with a younger age of onset [39]. Immunodeficiency and consequent propensity towards particular types of infections, which can precede MDS/AML development, is characterized by monocytopenia, low B and NK cells, inversion of the normal CD4:CD8 ratio, and neutropenia [37,38].

Germline variants identified resulting in GATA2 deficiency syndrome are loss of function (LOF) variants that commonly consist of null variants in regions encoding stop codons within or prior to the ZF2 domain leading to haploinsufficiency; missense LOF variants within the ZF2 domain; or variants located in an enhancer region of intron 4 (NM_032638.5) that disrupt the protein’s EBOX-GATA2-ETS domain [22]. Additional rare variants outside of these regions have also been described [40]. In contrast, somatic GATA2 variants found in AML (often in cases of biallelic CEBPA variant and or germline CEBPA variant) are mostly located within regions encoding the ZF1 domain and can be either loss (LOF) or gain of function (GOF) variants [35,41,42].

Development of MDS or AML in patients with GATA2 deficiency syndrome is commonly associated with acquisition of karyotypic abnormalities. Monosomy 7 is the most common, occurring in approximately 30–40% of cases, followed by trisomy 8 in 15–35% of cases, and der(1;7) in 9–20% of cases [27,37,43]. Other karyotypic abnormalities such as trisomy 21 are also recurrently identified, whereas the finding of a complex karyotype is very rare [27]. In addition to karyotypic abnormalities, somatic variants occur recurrently in known MDS/AML driver genes, including STAG2, ASXL1, SETBP1, and RUNX1 [27,37,43]. Somatic variants of the wild-type GATA2 allele are uncommon [44]. GATA2 deficiency syndrome patients can be divided into three groups based on their bone marrow findings: normal; hypoplastic and/or low-grade MDS; versus MDS with excess blasts, AML, or CMML [43]. No acquired molecular or karyotypic aberrations were found in those with a normal bone marrow, but somatic variants in SETBP1, RAS pathway genes, and RUNX1 were more frequent in those with MDS with excess blasts, AML, or CMML, whereas STAG2 variants were found primarily in patients with hypoplastic marrows or low-grade MDS and were not associated with progression to higher risk disease or AML [43]. In contrast, in sporadic MDS, STAG2 variants are associated with shorter overall and leukemia-free survival and higher risk of leukemic transformation [45].

SAMD9 and SAMD9L

SAMD9 and SAMD9L are interferon inducible genes located adjacent to one another on the long arm of chromosome 7. The function of SAMD9 and SAMD9L in hematopoietic cells is not entirely clear, but the proteins may have an important role in cell cycling, cellular proliferation, and protein translation in hematopoietic stem and progenitor cells with deleterious variants resulting in DNA damage repair defects and apoptosis [46]. These proteins also play a role in antiviral responses, inflammation, and development [47,48]. In 2016, P/LP germline variants in SAMD9 were associated with an early-onset multisystem disorder characterized by MDS, infections, restriction of growth, adrenal hypoplasia, and enteropathy (MIRAGE syndrome) [48]. Simultaneously, P/LP germline variants in SAMD9L were linked to a progressive neurologic syndrome with cerebellar dysfunction as well as pancytopenia with a hypocellular bone marrow (Ataxia Pancytopenia syndrome) [47]. Patients with either of these syndromes were also found to have a high propensity towards young-onset MDS with monosomy 7, with near complete penetrance for patients with a SAMD9 variant and approximately 70% for patients with a SAMD9L variant [48,49]. P/LP germline variants are typically de novo GOF alleles encoding amino acid substitutions within the C-terminus of the protein near or within the P-loop nucleoside triphosphatase domain [47,48]. P/LP germline variants in these two genes account for 8–17% of childhood MDS with a median age at MDS diagnosis of 9.6 years (range: 0.2 – 17.6) [50,51]. The penetrance and severity of immune and/or neurologic dysfunction associated with SAMD9/9L variants are variable with some patients developing cytopenias and/or MDS without other syndromic features. Rare cases of entirely unaffected carriers have also been described [49].

Patients with P/LP germline SAMD9/9L variants display a unique disease mechanism whereby clonal evolution results in loss of the mutated copy of chromosome 7 prior to or at the time of development of MDS [44]. This loss of the variant allele makes detection of P/LP germline SAMD9/9L variants very difficult if testing is performed on hematopoietic tissue, but testing true germline tissue, such as cultured skin fibroblasts, is not subject to this limitation. Progression to advanced MDS or AML has also been associated with additional acquisition of variants in known driver genes such as SETBP1, ASXL1, RUNX1, and ETV6 [47,51]. Pediatric MDS patients with P/LP germline SAMD9/9L variants have favorable outcomes after HSCT with a 5-year overall survival of 85% [50]. However, for those with severe multi-organ involvement as a result of SAMD9 variants, HSCT outcomes are poor as a result of syndrome-associated comorbidities [52,53].

Somatic revertant mosaicism, protecting against MDS, has also been described in SAMD9/9L patients as a result of acquisition of LOF truncating variants on the affected allele or duplication of the wild-type allele through uniparental disomy [44]. Although rare, cases of spontaneous and sustained remission of affected patients via uniparental disomy of 7q have also been reported [49].

TP53

TP53 is an important tumor suppressor gene with vital roles in DNA damage repair, cell cycling, apoptosis, senescence, and differentiation [54]. It is one of the most commonly somatically mutated genes in human cancer and is typically associated with inferior outcomes [55]. P/LP germline variants in TP53 result in a cancer predisposition syndrome known as Li-Fraumeni syndrome (LFS). Patients with LFS have a very high life-time risk of cancer development with an estimated ≥ 70% risk for men and ≥ 90% risk for women [54]. The most common types of cancers in these patients include: adrenocortical carcinomas, breast cancer, central nervous system tumors, soft tissue sarcomas, and osteosarcomas. Patients are also at increased risk of numerous other malignancies, including acute lymphoblastic leukemia (ALL), MDS, and AML, however these account for only 2–4% of LFS-associated cancers [54,56]. LFS patients are particularly sensitive to the carcinogenic effects of chemotherapy and radiation as the rates of second primary malignancies as well as therapy-related malignancies is high (~45% risk of developing a second cancer) [11,54,57]. In several studies, P/LP germline TP53 variants have been identified in approximately 6% of cases of therapy-related AML (t-AML), but were less common in de novo AML [11,58]. Complex karyotype and loss of the wild-type TP53 allele in leukemic samples was found frequently in LFS patients with t-AML [11,58].

B. Management recommendations for those with high penetrance disorders

Patients with deleterious germline CEBPA, GATA2, or SAMD9/9L variants all have a high penetrance for myeloid malignancy, which tends to occur at atypically young ages and is associated with acquisition of variants in known MDS/AML driver genes at the time of malignant transformation. However, the existence and penetrance of other organ/systemic manifestations differ. These differences have important implications for patient management and treatment tolerability. Unfortunately, there are no high-quality evidence-based guidelines to follow for the surveillance and management of germline predisposition syndromes with high myeloid malignant penetrance. However, there are gene-specific recommendations based on expert opinion.

LFS, which has a high malignant penetrance, but less so for myeloid malignancies specifically, has more well-established guidelines for surveillance and cancer-preventative measures to which readers are referred [59,60]. Of note, there are no specific surveillance or management recommendations regarding myeloid malignancies for LFS patients as these are less commonly encountered than solid tumors, but a complete blood count (CBC) every 6–12 months, especially in those with a history of prior exposure to chemotherapy and/or radiation would be reasonable. For LFS patients with breast cancer, bilateral mastectomy is preferred over lumpectomy and radiation to reduce the risk of radiation-induced secondary malignancies, including both hematopoietic and solid tumors. The optimal HSCT conditioning regimen for LFS patients with myeloid malignancy remains unknown with generally poor outcomes including high rates of relapse [56].

For patients with CEBPA-associated familial AML, treatment is similar to that recommended for standard forms of AML, including high intensity induction chemotherapy for fit patients. As the risk of developing subsequent independent additional AMLs additional is particularly high in these patients, given the persistence of the germline predisposition allele following treatment with chemotherapy alone, HSCT in first complete remission should be strongly considered regardless of the traditional molecular and cytogenetic AML risk stratification. Donor selection is of critical importance with selection of a related donor lacking the same CEBPA predisposition allele or use of an unrelated donor strongly recommended to avoid “giving back” the same high penetrance predisposing variant. Although controversial, pre-emptive HSCT for asymptomatic individuals with P/LP 5’-end CEBPA germline variants and a human leukocyte antigen (HLA) 10/10 donor should be considered, given the nearly complete penetrance of this disorder (Figure 1A). It is known that younger fit patients tolerate HSCT with less transplant associated morbidity and mortality than older patients or patients who may have complications, such as severe cytopenias, infections, or organ dysfunction as a result of active AML or recent chemotherapy [61]. As the penetrance of P/LP 3’-end CEBPA germline variants appears to be incomplete (~46%), pre-emptive HSCT is not recommended for these patients. In the absence of a pre-emptive HSCT, asymptomatic carriers should be monitored with CBCs every 6–12 months and have a bone marrow examination with molecular testing performed if any CBC abnormalities are encountered. Early detection, however, is challenging as there are no known pre-leukemic cytopenias or dysplasia that otherwise would warn of impending leukemia development.

Figure 1. Visual summary of the recommendations for timing of hematopoietic stem cell transplant (HSCT) for germline predisposition syndromes.

(A) For CEBPA and GATA2 germline variant carriers that alter the proteins’ N-termini, penetrance for myelodysplastic syndrome/acute myeloid leukemia is complete or near-complete, and as such, HSCT is recommended early, including prior to the acquisition of additional somatic variants and/or development of malignancy. * Note that P/LP germline CEBPA variants that alter the C-terminus have variable penetrance. (B) For syndromes with variable penetrance, such as RUNX1, ANKRD26, ETV6, and DDX41 germline variant carriers as well as bone marrow failure syndrome patients, HSCT prior to hematopoietic malignancy development is generally not recommended. However, acquisition of second-hit variants in DDX41 and RUNX1 germline variant carriers as well as multi-hit TP53 variants or other high-risk somatic changes in patients with bone marrow failure syndromes should prompt consideration for HSCT at the time of detection of these variants given their associated high-risk for impending malignant transformation. Created with BioRender.com.

Patients with GATA2 germline variants have a high penetrance of MDS/AML as well as infections and organ dysfunction, which if severe, can complicate HSCT and increase risk of adverse effects. Experts recommend vaccinating children against human papilloma virus and using azithromycin prophylaxis to prevent mycobacterial infections [62]. Baseline HLA-typing and monitoring for early development of dysplasia or cytopenias with CBCs every 3–6 months and bone marrow biopsy with molecular and cytogenetics every 1–2 years is recommended [62]. HSCT should be performed before the development of higher-risk MDS, AML, severe systemic infections, or severe pulmonary disease and use of intensive chemotherapy prior to HSCT should be avoided or limited due to high risk of infection [22,37]. There is no gold-standard timing for HSCT in GATA2 deficiency patients. However, pre-emptive HSCT should be considered and discussed with patients upon initial consultation to avoid the severe infections that often develop in these individuals (Figure 1A). In the absence of a pre-emptive HSCT, HSCT is recommended once bone marrow hypocellularity is identified, ideally prior to acquisition of monosomy 7, which has been found to be an early somatic event associated with development of MDS/AML, or additional somatic aberrations in genes associated with high-risk disease such as ASXL1, SETBP1, and RUNX1 [22,37,44].

Similar to GATA2 deficiency but to a greater extent, the hematologic management of patients with SAMD9/SAMD9L germline variants is complicated by the common presence of pre-existing organ dysfunction. For patients with P/LP germline SAMD9/9L variants, surveillance with CBCs every 4–6 months and annual bone marrow biopsies with molecular and cytogenetic testing for identification of high-risk somatic alterations (e.g., monosomy 7, associated driver gene variants) is recommended [63,64]. Careful consideration for HSCT in pediatric patients with SAMD9/9L variants and monosomy 7 must be given and is generally recommended in the absence of other severe organ dysfunctions [44]. For patients with MIRAGE syndrome and MDS or bone marrow failure, however, the majority of patients reported in the literature who underwent HSCT had severe complications, such as adrenal crisis, severe infections, veno-occlusive disease, and respiratory failure [52,53]. Due to these risks and the rare, but observed possibility of, spontaneous remission as a result of uniparental disomy of 7q, pre-emptive HSCT is not recommended, and HSCT decisions should be made on a case-by-case basis for patients with severe cytopenias, BMF, MDS, or AML. The use of high-dose cytarabine should also be limited in MDS/AML patients with P/LP germline SAMD9L variants due to the risk of causing severe cerebellar dysfunction or worsening pre-existing ataxia.

C. Variable penetrance disorders

See Table 2 for a summary of the characteristics of variable penetrance germline predisposition syndromes.

Table 2.

Characteristics of variable penetrance germline predisposition disorders

Gene name	Inheritance	Common Malignant Phenotype	Malignant Penetrance	Median age of malignancy onset (yrs)	Common associated somatic alterations	When to consider HSCT?
RUNX1	AD	MDS/AML, T-cell ALL	30–50%	29–33	RUNX1, BCOR, PHF6	At time of HM development*
ETV6	AD	B-cell ALL, MDS/AML	30%	Highly variable	unknown	At time of HM development
ANKRD26	AD	MDS/AML, CML, CMML	10%	Range: 30–70 years	unknown	At time of HM development
DDX41	AD	MDS/AML	50% (MDS/AML)	66–68	DDX41, ASXL1, DNMT3A, TP53	At time of HM development*
BMF syndrome genes	AD/AR/X-linked	MDS/AML	Variable, may range from <10% to >50%	Variable, BMF usually occurs earlier	TP53 alterations (multi-hit), monosomy 7/del(7q), 1q+, EVI1 overexpression, RUNX1 alterations, CSF3R variants#	Severe BMF, at time of HM development

Abbreviations used: AD, autosomal dominant; ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; HM, hematopoietic malignancy; HSCT, hematopoietic stem cell transplant; MDS, myelodysplastic syndrome.

^*Consider initiating work-up if an acquired second hit in the unaffected allele is identified.

^#In patients with congenital neutropenia.

RUNX1, ETV6, and ANKRD26

Within the 2022 ELN AML guidelines, P/LP germline RUNX1, ETV6, and ANKRD26 variants belong to the category, “myeloid neoplasms with germline predisposition and pre-existing platelet disorders” due to their shared high penetrance for quantitative and/or functional platelet abnormalities in addition to a predisposition to MDS/AML and/or other HMs [4]. All three genes have an autosomal dominant mode of inheritance. Carriers of a deleterious germline RUNX1 or ETV6 variant have a penetrance of ≥ 90% for life-long mild to moderate thrombocytopenia (50–150 × 10⁹/L), and those with similar variants in ANKRD26 have the lowest average platelet count of 48 × 10⁹/L with complete penetrance [65–69]. In all three disorders, platelet size is normal, and bleeding phenotype can vary, with P/LP germline RUNX1 variant carriers typically manifesting a mild to moderate bleeding diathesis as a result of decreased dense granules and demonstration of an aspirin-like defect on platelet aggregation studies [17,69]. In contrast, those with P/LP germline ETV6 or ANKRD26 variants may have absent to mild propensity towards mucocutaneous bleeding [65,67].

Although these disorders show complete or near-complete penetrance for platelet abnormalities, their malignant penetrance is variable and incomplete. Individuals with P/LP germline RUNX1 variants have the highest penetrance for developing a HM among these three conditions, up to 50%, with MDS/AML being the most common, followed by T-cell ALL [17,32]. The median age of onset for these malignancies varies among studies, but is estimated as 29–33 years [32]. Carriers of P/LP germline ETV6 variants have a slightly lower life-time malignant penetrance of approximately 30%, with a highly variable age at onset ranging from 2 – 82 years [19]. Two-thirds of the malignancies that develop are B-cell ALLs followed by MDS/AML. Those with P/LP germline ANKRD26 variants have the lowest malignant penetrance among these three disorders at approximately 8%, with MDS/AML being the most frequently observed, followed by chronic myeloid leukemia in about 1% of cases [65,66,70].

P/LP germline RUNX1 variants tend to be unique to each FPD/MM family with few recurrent variants identified [70]. A broad spectrum of variants, including missense, frameshift, stop-gain, as well as partial and whole gene deletions have been reported [32]. Missense variants are frequently located in the exons encoding the Runt homology domain with some acting in a dominant-negative fashion [71], whereas deletions and frameshift variants are found throughout the gene and typically lead to haploinsufficiency [32]. Therefore, sequencing of all exons with incorporation of copy number analysis when conducting genetic testing for deleterious germline RUNX1 variants is recommended.

P/LP germline ETV6 variants, including LOF missense, nonsense, and frameshift variants, have been identified throughout the gene, the majority being missense variants clustering in exons encoding the C-terminal DNA binding ETS domain [19,70]. A large deletion including exon 2 of ETV6 has also been described as in inherited variant in a child with ALL [72].

Virtually all P/LP germline ANKRD26 variants described to date are single nucleotide substitutions or small deletions located in a short 19-basepair stretch within the gene’s 5’-UTR, which contains binding sites for RUNX1 and FLI1 [65,70]. In normal hematopoiesis, transcription factors RUNX1 and FLI1 bind to the 5’-UTR of ANKRD26, down-regulating the gene’s expression. P/LP germline variants in the 5’-UTR of ANKRD26 disrupt one/both of these binding sites, resulting in persistent expression of ANKRD26 in megakaryocytes, and consequent defective pro-platelet formation as well as increased downstream signaling through the JAK/STAT, PI3K, and MAPK/ERK pathways [70].

The delayed onset and incomplete penetrance of malignancy for these syndromes suggest that the germline variant itself is insufficient for leukemogenesis. In the case of P/LP germline RUNX1 variant carriers, BCOR was found to be the most commonly mutated gene, with multiple BCOR variants identified in some patients [73]. However, the presence of BCOR variants, regardless of their number or variant allele frequency, do not correlate with risk of MDS/AML development [73]. MDS/AML variants in genes frequently associated with clonal hematopoiesis, such as TET2, DNMT3A, and KRAS, have also been found without MDS/AML, however, somatic variants in RUNX1 have only been found in malignant clones [74,75]. Malignant transformation is most often accompanied by the accumulation of RUNX1 LOF variants, which occur in over 30% of MDS/AML cases [27]. Somatic variants in other genes including BCOR, PHF6, TET2, and WT1 have also been identified in patients at the time of diagnosis of MDS/AML [27,32]. The mechanism of leukemogenesis in germline ETV6 or ANKRD26 variant carriers is unknown. A recent cohort study of seven ANKRD26 and three ETV6 germline variants identified clonal hematopoiesis in only one of seven ANKRD26 germline variant carriers (an SF3B1 missense variant) and none of the ETV6 patients [76]. Additional somatic variants in ETV6 at the time of HM development have not commonly been reported, and although somatic variants in BCOR, RUNX1, ASXL1, and other genes have been identified, it is not clear whether these occur more frequently in ETV6 germline-mutated compared to sporadic cases of HMs [27,68]. Given the lower malignant penetrance and rarity of germline ANKRD26-associated myeloid malignancies, even less is known about clonal evolution and leukemogenesis in these patients, although ASXL1 and KRAS variants have been identified in a patient at the time of CMML diagnosis [77].

DDX41

DDX41 is located on the long arm of chromosome 5 and encodes a multifunctional DEAD-box helicase protein that has a role in mRNA splicing, R-loop accumulation, innate immune response, and is believed to be a tumor suppressor [78]. P/LP germline variants in DDX41 are known to predispose to MDS/AML with a median age at disease onset of 66–68 years [79,80], no different from that of sporadic MDS/AML, and in stark contrast to the early-onset observed for the aforementioned predisposition syndromes. As interest and studies in the gene have increased over the past few years, it has now been identified in approximately 2–6% of all adult MDS/AML patients within multiple independent cohorts, making it the most common MDS/AML predisposition gene identified in malignancies diagnosed in adults and rarely causing malignancies prior to age 40 [79,80]. The penetrance for developing myeloid malignancies in those with P/LP germline DDX41 variants is incomplete, but estimated to be approximately 50% [81], with progression occurring three times more often in men compared to women [79–81]. Other consistent features of germline DDX41 mutated MDS/AML cases include an association with hypocellular bone marrow and normal karyotype MDS/AML [78,79]. P/LP germline DDX41 variants are associated with myeloid malignancies with favorable prognoses compared to age-matched sporadic cases, independent of traditional risk stratification systems [79–81]. Outside of a predisposition towards MDS/AML, P/LP germline DDX41 variants have also been identified in patients with aplastic anemia, myeloproliferative neoplasms, lymphoid malignancies, and solid tumors, however given the lower frequency of occurrence of these malignancies, the causal relationship of the germline DDX41 variant and the malignancies remains unclear [79–81].

P/LP germline DDX41 variants include start-loss, frameshift, nonsense, splicing, missense, and deletions, several of which are relatively frequent in certain populations and geographic distributions [81]. DDX41 variants (NM_016222) c.3G>A, p.Met1? and c.415_418dup, p.Asp140delinsGlyTer are the most common germline variants found in North America, the United Kingdom, and France, whereas the c.1496dup, p.Ala500CysfsTer9 is the most common germline DDX41 variant in Japan, Korea, and China [80,81]. Other recurrent variants, namely c.121C>T, p.Gln41Ter and c.653G>A, p.Gly218Asp were primarily found in patients from Germany and Italy, respectively [80]. Uncovery of additional population-specific variants is expected as sequencing of patients from additional ethnicities is underway. To date, nearly all truncating DDX41 variants identified are germline in origin [81].

Development of MDS/AML in P/LP germline DDX41 variant carriers is commonly associated with the acquisition of a somatic DDX41 variant on the trans allele, occurring in 55–70% of cases [79–81]. Among these acquired DDX41 variants, the c.1574G>A, p.Arg525His is the most common, seen across all ethnic groups and accounting for ~65% of all acquired DDX41 variants [79–81]. The majority of other acquired DDX41 variants are also missense variants located within the DEAD-box helicase domain, including: c.680C>T, p.Thr227Met; c.962C>T, p.Pro321Leu; c.1589G>A, p.Gly530Asp; c.1588G>A, p.Gly530Ser [80,81]. The frequency of other somatic driver variants varies across studies, ranging from rarely being identified to observed in 46% of patients [27,79–81]. Among those studies for which aquired variants were identified, the most frequently mutated genes were ASXL1 (in 10–20% of cases) [27,81], followed by DNMT3A, TP53, and CUX1. In contrast to the other somatic variants, which were found to occur at a similar frequency in sporadic MDS, CUX1 variants were found to be over-represented in DDX41-mutated cases cases [81]. Although the timing of acquisition of additional DDX41 and/or other somatic driver variants and onset of MDS/AML remains unknown, some individuals with P/LP germline DDX41 variants develop cytopenias months to years prior to disease onset [79]. Observations that the variant allele frequencies of the DDX41 somatic mutant clone is greater than those of other driver variants suggest that acquisition of the somatic DDX41 variant may occur early during malignant transformation [81].

BMF Syndromes

The most common BMF syndromes include disorders caused by deficient DNA repair, such as Fanconi anemia, telomeropathies, and defects in ribosome biogenesis, like Diamond-Blackfan anemia. Cytopenias/overt BMF are the hallmark of these syndromes. Other phenotypic or laboratory features may hint at a specific underlying genetic cause. For example, exocrine pancreatic insufficiency is linked to Shwachman-Diamond-syndrome, whereas short stature, radial ray abnormalities and increased chromosomal breakage are observed frequently in patients with Fanconi anemia. The classic triad of dyskeratosis congenita: nail dystrophy, oral leukoplakia, and lacy reticular skin pigmentation may only occur in a subset of those with P/LP germline variants in BMF genes. However, at least one of these features in addition to the presence of interstitial lung disease and early-onset head and neck squamous cell cancer is common and may be accompanied with short telomeres [82].

Some of the more recently described syndromes blur the lines between these categories. Namely, P/LP germline variants in DNAJC21, encoding a ribosome maturation factor, present with a Shwachman-Diamond-like syndrome based on the frequent observation of pancreas lipomatosis and exocrine pancreatic dysfunction [83,84], whereas abnormal skin pigmentation, dental and retinal abnormalities seem to resemble characteristics of telomeropathies [83]. The microcephaly and short stature observed in patients with combined ADH5/ALDH2 deficiency mimic some features seen in Fanconi anemia patients. Crucially, radial ray abnormalities and increased chromosomal breakage are absent in these patients, although the phenotype seems to be caused by deficient formaldehyde detoxification causing increased DNA damage, thereby overwhelming the DNA repair capacity [85,86].

The overall penetrance of cytopenias/overt BMF is high, in some syndromes even complete [87]. Often, the first hematopoietic abnormalities are diagnosed at birth or in infancy, although later occurrences as well as spontaneous, intermittent, or prolonged improvement of cytopenias have been described [87]. The risk of developing HMs as well as the type of malignancy and its prognosis varies greatly depending on the underlying cause. Some syndromes go along with a high(er) likelihood of developing MDS/AML, e.g., approximately 80% of patients with digenic ADH5/ALDH2 deficiency [85,86,88], 33% of patients with bi-allelic ERCC6L2 variants [39,89–91], and 31% of patients with severe congenital neutropenia [92]. A moderate risk of AML/MDS in the range of 10% to 30% has been reported for patients with telomeropathies (most commonly affected genes include TERT, RTEL1, TERC, and DKC1), Shwachman-Diamond syndrome, Fanconi anemia, and bi-allelic DNAJC21 variants [82,93,94]. In patients with Diamond-Blackfan anemia and P/LP germline MECOM variants, the overall risk is below 10%, however, penetrance of cytopenias/BMF is near-complete [95–97]. Of note, early HSCT due to severe BMF may confound the reported risk of AML/MDS development in some of these syndromes.

Factors predicting disease progression, including development of clonal evolution, bone marrow dysplasia, and/or increase in blast count, are vital to identify the ideal time point for a (pre-emptive) HSCT, given that HSCT-related morbidity and mortality are higher in this subset of patients [98]. Studies in patients with Shwachman-Diamond syndrome have revealed that somatic EIF6 and TP53 variants are most common [99]. However, the vast majority of patients acquire at least one TP53 variant over time, which is not a sign of leukemic transformation per se, whereas clones with multi-hit TP53 variants, defined as ≥2 distinct TP53 variants or a single TP53 variant with concurrent deletion of the other allele [3], may be more prone to leukemogenesis [99]. Likewise, multi-hit TP53 variants in patients with P/LP germline ERCC6L2 variants may be indicative of disease progression and given the poor prognosis of MDS/AML in these patients, early HSCT should be considered [39,90]. Patients with Fanconi anemia often develop gains of the long arm of chromosome 1, which are characteristic of the disease, but do not correlate with leukemic transformation [100]. In contrast, monosomy 7/del(7q), gains of 3q26 with EVI1 overexpression and RUNX1 alterations, indicate a higher risk of transformation [100–102]. CSF3R variants, providing a growth advantage in the presence of granulocyte colony-stimulating factor, occur in patients with severe congenital neutropenia and are a marker of impending AML/MDS development [103]. Most severe congenital neutropenia patients also acquire RUNX1 variants later in the process of disease progression [103].

D. Management recommendations for those with variable penetrance disorders

The recommendations for management and surveillance of individuals with P/LP germline variants associated with a variable penetrance of myeloid malignancies are based on expert opinion rather than rigorous evidence-based clinical practice guidelines.

For patients with such variants in genes associated with pre-existing platelet dysfunction (i.e., RUNX1, ETV6, or ANKRD26), but without myeloid malignancy, performing platelet aggregation studies at baseline, particularly if there is a clinical history of easy bruising or bleeding is recommended. For patients with a history of mild to moderate bleeding or for whom a surgical procedure associated with moderate to high bleeding risk is required, or in anticipation of childbirth, use of antifibrinolytic agents and/or desmopressin is commonly recommended [15]. The use of platelet transfusions is typically reserved for those with severe bleeding to minimize the risk of alloimmunization.

Baseline assessment of CBC with white blood cell differential, repeating the CBC/differential several times per year (every 3–4 months for those with P/LP germline variants in RUNX1 given the higher malignant penetrance; and every 6–12 months for those with such alleles in ETV6 or ANKRD26) should be conducted. Baseline bone marrow biopsy/aspiration including cytogenetic analysis and molecular profiling, with a repeat bone marrow biopsy/aspiration at the time of any persistent unexplained abnormalities on the CBC or differential is recommended [15]. Recommendations for performing annual bone marrow examinations are inconsistent and should be considered on a case-by-case basis taking patient preference into account. HSCT with standard conditioning is recommended upon the development of a HM after remission induction, but given the incomplete penetrance of these disorders, pre-emptive HSCT is not recommended. For those with P/LP germline RUNX1 variants, the acquisition of somatic RUNX1 variants should raise concern for early or overt MDS/AML development and prompt HSCT work-up prior to further progression (Figure 1B).

Similar recommendations are prudent for asymptomatic DDX41 P/LP germline-variant carriers, however given the late median age at onset of MDS/AML, the age at which to begin this screening is not established. As more than 90% of MDS/AML cases reported thus far occur after the age of 40 years, we recommend starting at age 40, or 10 years younger than the earliest case of MDS/AML onset in the family, whichever occurs sooner. A baseline bone marrow biopsy and aspirate with cytogenetic analysis and molecular testing is generally recommended as dysplastic features in the absence of overt MDS can been seen at baseline in this and other germline predisposition disorders, including those associated with platelet dysfunction and BMF. Interpretation by experienced pathologists is essential to avoid over-calling MDS in these patients. Most experts also agree on performing a bone marrow aspirate and biopsy with cytogenetic analysis and molecular testing with the development of persistent or progressive cytopenias or macrocytosis. Performing annual bone marrow examinations in the presence of a normal CBC may be recommended in particular cases after discussion with individuals on a case-by-case basis. If molecular testing is performed, acquisition of somatic DDX41 variants may warrant closer monitoring and initiation of HSCT, as this is known to be an early event in malignant transformation. Pre-emptive HSCT is not recommended for DDX41 germline mutated patients given the variable and incomplete penetrance as well as the older age at presentation. HSCT should be considered upon the development of MDS/AML and take into account the patient’s biologic fitness. High rates of severe acute graft versus host disease (GVHD) have been observed for patients with P/LP germline DDX41 variants (38% stages 3–4) when wild-type donors were used [104]. Compared to other GVHD prophylaxis regimens, post-transplant cyclophosphamide was associated with reduced risk of severe acute GVHD in this patient population and is therefore recommended when P/LP germline DDX41 variant carriers undergo HSCT [104].

If an individual with a known BMF syndrome-associated P/LP germline variant has a normal CBC, monitoring the peripheral blood cell counts at least every 6 months is advised. HSCT is currently recommended for those with BMF syndromes when the individual develops severe BMF with transfusion-dependence, recurrent life-threatening infections, bleeding complications, and/or MDS/AML. HSCT can also be considered for individuals with multi-hit TP53 alterations, which are generally considered signs of impending transformation, whereas the significance of monoallelic TP53 alterations is less clear and may depend on the underlying syndrome [39,90,99]. Many BMF syndromes do not have sufficiently established somatic markers to provide clinicians with enough evidence to identify impending clonal evolution and leukemic transformation. More studies, specifically on larger numbers of patients who develop AML/MDS, are needed to improve the recommendations for timing of HSCT. Spontaneous or intermittent improvement of cytopenias, with or without somatic reversion, has been described in a variety of BMF syndromes and is particularly common in patients with Diamond-Blackfan anemia, which might affect consideration of HSCT in these individuals [105]. Since most BMF syndromes manifest in infancy/early childhood, HSCT may occur early, which is associated with a lower risk of HSCT-related morbidity or mortality [106]. Tailored conditioning regimens have been recommended to reduce non-relapse mortality and morbidity. For example, conditioning-regimens without whole-body irradiation have been recommended in patients with Fanconi anemia due to the increased risk of secondary solid tumors with ionizing radiation exposure [106,107], and reduced-intensity regimens seem to improve outcome after HSCT for patients with telomere biology disorders [107,108]. Irrespective of the underlying genetic cause, patients with unrecognized inherited BMF have an inferior HSCT outcomes, emphasizing the importance of establishing a genetic diagnosis prior to HSCT [109].

As with all autosomal dominant myeloid malignancy germline predisposition disorders, if an HSCT is being undertaken, selection of a donor that lacks the same predisposition variant is strongly recommended. There have been reports of donor-derived leukemia with the use of related donors that shared the same familial predisposition variant [110–113]. Therefore, it is advised that any potential related donors undergo genetic testing as part of their screening for eligibility. Additionally, the use of post-transplant cyclophosphamide for those with a P/LP germline DDX41 variant is recommended even when wild-type donors are used.

Summary

Penetrance for development of HMs, phenotypic heterogeneity, age of onset and clinical course of disease varies greatly across the different inherited syndromes. Deleterious variants in CEBPA, GATA2, SAMD9/SAMD9L, and TP53 are associated with high-penetrance disorders that require more rigorous surveillance. Management plans should consider the use of pre-emptive HSCT for those with 5’-end CEBPA variants or those with deleterious GATA2 variants with near-complete malignant penetrance. Specific guidelines for variable penetrance syndromes such as ANKRD26, DDX41, ETV6, RUNX1, and bone marrow failure syndromes are based on expert opinion, and HSCT is usually recommended at the time of malignancy development. Acquired second hits of the unaffected allele (common in patients with deleterious germline CEBPA, DDX41, or RUNX1 variants) as well as other specific cytogenetic and molecular events (e.g., monosomy 7, ASXL1, or multi-hit TP53 variants) may lead to disease progression/transformation, which should prompt early consideration of HSCT depending on the type of event and time of detection. The average age of onset, type of malignancy, family history, and estimated HSCT-related mortality and morbidity as well as specific conditioning regimens should also be taken into account. For most disorders, somatic markers still need to be identified and validated, emphasizing the need for sequential molecular monitoring of these patients to enable future correlations and improve the care and outcomes of individuals and families.

Practice Points

• The penetrance and clinical manifestation of each germline predisposition disorder vary greatly, even among family members with the same germline allele.

• Care should be taken at the time of HSCT to avoid using donors with deleterious germline predisposition alleles.

• Carriers of deleterious germline DDX41 variants benefit from post-transplant cyclophosphamide after HSCT even when wild-type donors are used.

Research Agenda

• Additional germline predisposition disorders to HMs are likely to be defined.

• Markers of clinical progression to HMs will be important to define to assist in prognostication.

• Determining the outcomes of individuals transplanted with HSCT donors who have deleterious germline variants will be important.