Germline CHEK2 and ATM Variants in Myeloid and Other Hematopoietic Malignancies

An intact DNA damage response is crucial to preventing cancer development, including in myeloid and lymphoid malignancies. Deficiencies in the homologous recombination (HR) pathway can lead to defective DNA damage responses, and this can occur through inherited germline mutations in HR pathway genes, such as CHEK2 and ATM. We now understand that germline mutations can be identified frequently (~ 5–10%) in patients with myeloid and lymphoid malignancies, and among the most common of these are CHEK2 and ATM. We review the role that deleterious germline CHEK2 and ATM variants play in the development of hematopoietic malignancies, and how this influences clinical practice, including cancer screening, hematopoietic stem cell transplantation, and therapy choice. In recent large cohorts of patients diagnosed with myeloid or lymphoid malignancies, deleterious germline loss of function variants in CHEK2 and ATM are among the most common identified. Germline CHEK2 variants predispose to a range of myeloid malignancies, most prominently myeloproliferative neoplasms and myelodysplastic syndromes (odds ratio range: 2.1–12.3), and chronic lymphocytic leukemia (odds ratio 14.83). Deleterious germline ATM variants have been shown to predispose to chronic lymphocytic leukemia (odds ratio range: 1.7–10.1), although additional studies are needed to demonstrate the risk they confer for myeloid malignancies. Early studies suggest there may also be associations between deleterious germline CHEK2 and ATM variants and development of clonal hematopoiesis. Identifying CHEK2 and ATM variants is crucial for the optimal management of patients and families affected by hematopoietic malignancies. “A 45 year-old woman presents to your clinic with a history of triple-negative breast cancer diagnosed five years ago, treated with surgery, radiation, and chemotherapy. About six months ago, she developed cervical lymphadenopathy, and a biopsy demonstrated small lymphocytic leukemia. Peripheral blood shows a small population of lymphocytes with a chronic lymphocytic leukemia immunophenotype, and FISH demonstrates a complex karyotype: gain of one to two copies of IGH and FGFR3; gain of two copies of CDKN2C at 1p32.3; gain of two copies of CKS1B at 1q21; tetrasomy for chromosome 3; trisomy and tetrasomy for chromosome 7; tetrasomy for chromosome 9; tetrasomy for chromosome 12; gain of one to two copies of ATM at 11q22.3; deletion of chromosome 13 deletion positive; gain of one to two copies of TP53 at 17p13.1). Given her history of two cancers, you arrange for germline genetic testing using DNA from cultured skin fibroblasts, which demonstrates pathogenic variants in ATM [c.1898 + 2 T > G] and CHEK2 [p.T367Metfs]. Her family history is significant for multiple cancers. (Fig. 1).”Fig. 1 Representative pedigree from a patient with germline pathogenic ATM and CHEK2 variants who was affected by early onset breast cancer and chronic lymphocytic leukemia. Arrow indicates proband. Colors indicate cancer type/disease: purple, breast cancer; blue, lymphoma; brown, melanoma; yellow, colon cancer; and green, autoimmune disease Representative pedigree from a patient with germline pathogenic ATM and CHEK2 variants who was affected by early onset breast cancer and chronic lymphocytic leukemia. Arrow indicates proband. Colors indicate cancer type/disease: purple, breast cancer; blue, lymphoma; brown, melanoma; yellow, colon cancer; and green, autoimmune disease


Introduction
DNA repair and maintenance are crucial homeostatic mechanisms for all human cells and are particularly relevant to cells that must divide continually throughout an individual's lifetime, such as hematopoietic stem and progenitor cells (HSPCs) [1]. Proper repair of DNA double-strand breaks (DSB) is a core element of the maintenance of genomic stability, directed through three pathways active in most human cells: (1) Homologous recombination (HR); (2) Canonical non-homologous end joining (NHEJ); and (3) Alternative NHEJ [2]. Canonical NHEJ is the simplest DNA repair mechanism and involves directly adjoining DSBs through the binding of the Ku-80-Ku7p proteins to the fragmented DNA ends, followed by recruitment of DNA-dependent protein kinases (DNA-PKcs), which then activate ligase IV and co-factors which seal the DNA break. The alternative NHEJ mechanism involves recruitment of PARP to the DNA ends, ending in the DNA DSB being sealed by Ligase I and III [3]. Although NHEJ effectively repairs DNA DSBs, it does not involve the usage of a complementary DNA template and, as such, is error-prone, inducing chromosomal abnormalities and chromothripsis [3].
In contrast, HR is the most error-free of the DNA repair pathways, since it uses a complementary DNA template available during S-phase to correct the detected DNA lesion [4]. The HR pathway is engaged (Fig. 2) when the MRE11-RAD50-NBS1 (MRN) protein complex is recruited to the fragmented DNA ends, which subsequently recruits the ataxia telangiectasia mutated (ATM) serine/threonine kinase. Activated ATM then phosphorylates the CHK2 protein, resulting in the downstream activation of a series of proteins, including CDC25C, p53, BRCA1/2, and cyclin-D kinases, which coordinate template-based DSB repair, cell-cycle arrest, and potentially apoptosis [2,3]. The ATM protein can also be activated in response to oxidative stress, nutrient deprivation, and hypoxia [5][6][7][8][9].
The cell-cycle phase is an important determinant of whether HR or NHEJ is chosen by the cell. NHEJ is more active in G1, when the sister chromatid template for HR is absent, whereas HR is restricted to S through G2 phases when a complementary DNA strand is available [1,3]. Another important feature of the DSB response is the induction of cell cycle checkpoint arrest, Fig. 1 Representative pedigree from a patient with germline pathogenic ATM and CHEK2 variants who was affected by early onset breast cancer and chronic lymphocytic leukemia. Arrow indicates proband. Colors indicate cancer type/disease: purple, breast cancer; blue, lymphoma; brown, melanoma; yellow, colon cancer; and green, autoimmune disease Fig. 2 Schematic diagram of the Homologous Repair (HR) response to DNA double strand breaks (DSB) in humans. Upon sensing a double strand break, the MRE11-RAD50-NBS1 (MRN) complex recruits and phosphorylates the ataxia telangiectasia mutated (ATM) serine/threonine kinase. Activated ATM then phosphorylates several downstream targets, including γH 2 AX, p53, and CHK2, among others. Phosphorylated CHK2 then activates several downstream targets, including p53, BRCA1, BRCA2, and cyclin D kinases (CDC). The downstream effectors then modulate apoptosis, cell cycle (G1/S checkpoint), and DNA repair mediated during the S or G2 phase by ATR and ATM, and is essential for allowing the cell to re-enter mitosis after successful DSB repair [3]. There are several other important regulatory proteins and co-factors involved in this process, such as γH2AX, POLQ, PALB2, RAD51, and others [2].
Germline pathogenic and likely pathogenic (P/LP) gene variants that result in loss of function (LoF) have been identified and characterized at several levels in the HR pathway. The best characterized of these are mutations in ATM, CHEK2, BRCA1, BRCA2, and the Fanconi gene family, among others. The broad consequence of a mutation in one of these genes is a defective HR pathway, with consequent reliance on error-prone NHEJ mechanisms for DNA repair. The downstream result of using error-prone DNA repair pathways is an accumulation of somatic chromosomal abnormalities and DNA changes, particularly within rapidly dividing cells (e.g., epithelial, mammary, hematopoietic), with an increased risk for the development of overt malignancy. Germline mutations in these genes have been well-characterized as risk factors for breast [10], prostate [11,12], and pancreatic [13] cancers. Increasingly, we appreciate that germline mutations in several of these genes, particularly ATM, CHEK2, and BRCA1/2, are risk factors for hematopoietic malignancies as well. In this review, we outline how germline LoF CHEK2 and ATM mutations predispose to myeloid and other hematopoietic malignancies and examine the clinical and therapeutic implications of identifying patients who are carriers of these mutations.

Germline Mutations in CHEckpoint Kinase 2 (CHEK2)
As noted above, the CHK2 protein is essential to the transmission of the DSB signal from ATM to downstream effectors CDC25C, p53, BRCA1/2, cyclin-D kinases, and others via phosphorylation. CHK2 protein domains include a S/TQ domain for ATM binding [14,15], a Forkhead-associated domain (FHA), and a kinase domain [14] (Fig. 3A). Germline LoF mutations in the CHEK2 gene occur at population allele frequencies of up to 0.0132 in population databases (gnomAD v2) and, thus, are relatively common in some populations. A variety of mutation types has been identified, including splice site, missense, and frameshift, without a predisposition towards mutational hotspots [14]. The majority of LoF variants identified affect either the FHA or kinase domains [14] ( Table 1). Several of these variants occur as either founder mutations (e.g., p.I200T in Eastern European populations) or as population-specific variants (e.g., p.S428F in Ashkenazi Jewish populations) as well as recurrent non-founder mutations (e.g., p.T476M) [16]. The majority of these mutations have been nearly exclusively identified as germline, with somatic CHEK2 variants being very rare events [17]. Although the majority of patients carrying CHEK2 mutations are in the heterozygous state, individuals with homozygous LoF CHEK2 do occur and have a Li-Fraumeni like phenotype [18]
The literature addressing the role for CHEK2 mutations in the development of AML remains in evolution. One study that demonstrated the role for CHEK2 in the development of MDS or AML which did identify CHEK2 as a risk allele did not identify a similar risk for de novo AML when considering total LoF CHEK2 mutations (OR 1.03 [95% CI 0.4-2.4], P = 0.831), although few cases (N = 6) were included in this analysis [29]. However, the presence of CHEK2 mutations has been identified in several AML patients [17,37,38]: one study found LoF CHEK2 variants in 7% of AML samples tested (N = 3/41) [17] with another study identifying CHEK2 variants in several newly diagnosed AML patients [39]. There is also in vitro evidence demonstrating that the promyelocytic leukemia gene (PML), classically rearranged in acute promyelocytic leukemia (PML-RARα), is involved in regulating phosphorylation and activity of the CHK2 protein [40][41][42], and arsenic trioxide can augment CHK2/p53 mediated apoptosis [43]. Another in vitro study also suggested that heterogeneity in CHK1 and CHK2 phosphorylation and downstream activation of caspase 3 is involved in responsiveness of AML cells to the anti-CD33 antibody-drug conjugate gemtuzumab ozogamicin [44].

CHEK2 Mutations in Other Hematopoietic Malignancies
Germline CHEK2 mutations have also been identified as risk factors for lymphoid malignancies. One study identified carrier status for a LoF CHEK2 variant as a risk factor for non-Hodgkin lymphoma (NHL) (OR 2.86 [95% CI 1.42-5.79], P = 0.02) and was associated with a worse progression-free survival (HR 2.1 [95% CI 1.12-4.05] P = 0.02) [45]. Another study identified carrier status for CHEK2 p.I200T as a strong risk factor for the development of chronic lymphocytic leukemia (CLL) (OR 14.83 [95% CI 1.85-∞], P = 0.0008) [46], and this variant has also been identified in other studies [47]. However, the CHEK2 c.1100delC allele was not found to be associated with CLL in another study (N = 973) from the UK (OR 0.74 [95% CI 0.32-1.70], P = 0.47) [48]. Experimental evidence though suggests a role for the CHEK2 c.1100delC allele in development of hematopoietic malignancies: Mice with this allele frequently develop multiple malignancies, including mammary, hematopoietic, lung, granulosa, and other epithelial and mesenchymal tumors [49]. Moreover, in vitro data demonstrate that phosphorylated CHK2 interacts with ERK in diffuse large B-cell lymphomas (DLBCL) [50] and that decreased CHK2 expression may correlate with aggressive DLBCL subtypes [51]. Mutations in CHEK2 affecting the FHA domain (e.g., p.I200T) have also been shown to increase the risk for Hodgkin lymphoma (HL) (OR 2.11 [95% CI 1.08-4.13], P = 0.04) [52], and CHEK2 expression is epigenetically downregulated in HL cell lines [53]. There are additional in vitro data suggesting a role for CHK2 function in the pathogenesis of NK/T-cell lymphomas [54,55].

Germline Mutations in ATM
Germline LoF mutations in the ATM gene have long been associated with early-onset myeloid malignancies, in addition to solid tumors such as breast and pancreatic cancers [56]. Loss of ATM function generates a greater risk of chromosomal translocations and other deleterious mutations associated with myeloid leukemia development [57]. Germline LoF ATM mutations are relatively common in the population, with 0.44% of individuals having an identifiable mutation in large cohorts [58].
Patients carrying LoF ATM mutations in the homozygous or compound heterozygous states present with Ataxia Telangiectasia (A-T), an autosomal recessive disorder characterized by a 50 to 150-fold increased risk of cancer development, and also cerebellar degeneration, telangiectasia, immunodeficiency, and radiation sensitivity [59,60]. The clinical presentation of A-T can be heterogeneous as some pathogenic mutations have a milder effect on protein activity than others, and a diagnosis of A-T is sometimes preceded by a cancer diagnosis in patients presenting with mild symptoms [61]. However, most individuals with germline deleterious ATM mutations are heterozygous carriers with a two to 13-fold increased risk for early-onset cancer development but do not have other features of A-T [62]. Several solid organ malignancy types have been linked to LoF ATM mutations, including prostate, breast, stomach, and lung [63].
The ATM protein is 350 kDa and contains several functional domains, including the TAN, kinase domain, FAT (FRAP-ATM-TRRAP) domain, and the FATC (FAT C-terminal) domains [8,64]. (Fig. 3B) There are limited functional data on the impact of ATM variants, and consequently, many are classified as variants of uncertain significance (VUSs). Some of these ATM VUSs are likely to be pathogenic. Although ATM P/LP and VUSs occur in all functional domains, many cluster in the PI3K domain as well as an undefined region between the TAN and PI3K domains (aa 702-1611). Damaging mutations exons encoding the PI3K domain are clinically impactful due to disruption of kinase function. Mutations in exons encoding the undefined region may induce structural changes that disrupt dimerization (Fig. 3). Somatic mutations in ATM are also common in multiple cancer types, in contrast to their rarity of somatic CHEK2 mutations.

ATM Mutations in Myeloid Malignanies
The role for ATM mutations in the myeloid malignancies remains in evolution and is less well characterized than for CHEK2. However, pathogenic ATM mutations have been identified at diagnosis in several patients with de novo AML [39]. However, germline ATM variants were not found in patients with germline syndromes or inherited bone marrow failure who had myeloid disease in another cohort (N = 144) [65]. There have also been several case reports of patients with A-T who developed AML, although few systematic studies examining this association [66][67][68]. One study suggested a possible association between the c.5144A > T and c.4138 T ATM variants and risk for chronic myeloid leukemia [69]. One paper reported an association between an intronic ATM polymorphism (rs228593) and MDS, although the implications of this are not clear [70]. However, intact ATM function has been well established as being critical for hematopoietic stem cell function [71], and ATM function and the associated signaling axis have been shown to have shown in vitro to modulate pathogenesis in AML [72][73][74][75]. Given this, it may be that there are more roles for ATM in the pathogenesis of myeloid malignancies that remain to be uncovered by future studies.

Germline CHEK2 and ATM Mutations and Clonal Hematopoiesis
In recent years, there is an increasing awareness for the role of clonal hematopoiesis (CH) in the development of hematopoietic malignancies, cardiovascular diseases, and other conditions. [87,88] CH is an acquired disorder wherein a subset of hematopoietic cells develops a single-nucleotide variant (SNV), insertion/deletion (indel), copy number variant (CNV), or chromosome level changes, although peripheral blood cell counts are normal. Among studies that examine germline DNA variants associated with the development of CH, [89] one large scale (N = 97,691) GWAS identified deleterious CHEK2 variants as risk factors for DNMT3Aand TET2-mutated CH (OR 1.7 P = 1.3 × 10 −5 ) [90]. Two other studies have also identified CHEK2 mutations as risk factors for the development of JAK2 V617F CH (OR 4.07, P = 0.0015) [27,28]. However, most of the studies examining the interaction of CHEK2 and CH development have been associational in nature, though one study suggested cord blood HSPCs have enhanced proliferative capacity upon CHEK2 knockdown [90]. Additional studies examining the mechanisms by which heterozygous CHEK2 mutations lead to CH are ongoing.
As with the literature surrounding myeloid malignancy, our understanding of the role for ATM in CH remains in evolution. One GWAS identified a single-nucleotide polymorphism associated with ATM (rs4754301) that was associated with somatic mosaicism manifesting as loss of chromosome Y (P = 1.3 × 10 −9 ) [91]. Another study of heterozygous germline ATM mutation carriers (N = 34) demonstrated a significant number of peripheral blood CH versus controls (N = 22), including somatic mutations in NF1, BCORL1, and DNMT3a (P = 0.00003); interestingly, there was also a high proportion of patients with germline ATM mutations in this study with acquired somatic ATM-mutated CH at 31% (N = 14/34), which was entirely absent in controls [92]. Future studies examining the development of CH in carriers of germline ATM variants will be of great interest.

Clinical Implications: Screening and Therapy Choice
Timely and comprehensive identification of germline predisposition mutations, of which CHEK2 and ATM are among the most common, is crucial to ensuring optimal care for patients diagnosed with myeloid malignancies, lymphoid malignancies, or CH. The first step upon suspicion of a patient with a hereditary cancer syndrome, either first identified on tumor-focused sequencing or through a suggestive clinical history, is confirmation of the presence of a germline variant. Although testing of peripheral blood is often utilized for patients with solid organ malignancies, this is usually inappropriate for patients with hematopoietic malignancies as peripheral blood undergoes somatic reversion easily leading to false negative results and is often contaminated with tumor cells. We suggest that patients suspected of having a hereditary cancer syndrome with an associated hematopoietic malignancy should have germline testing using DNA derived from cultured skin fibroblasts.
Cultured skin fibroblasts are considered the gold standard for germline tissue testing, as this tissue is derived from the ectoderm (rather than the mesoderm, which gives rise to the hematopoietic system), and the culturing process removes all blood contamination. In this way, one can be sure that the observed variant is constitutional in nature. Further, this method reliably produces large amounts of DNA for analysis. However, culture failures occur in about 5% of cultures, often associated with delayed culture initiation (OR 4.3, P < 0.01) or a pathogenic variant in a gene associated with telomere maintenance (OR 42.6, P < 0.01) [93]. Other germline tissue sources have been utilized for germline testing, including hair follicles, nail clippings, buccal swabs, and sorted lymphocytes, although each of these comes with notable limitations. The use of hair follicles and nail clippings results in low DNA recovery, which can complicate assays that require at least hundreds of nanograms of DNA. Moreover, hair follicles and nail clippings can be contaminated with hematopoietic cells, especially monocytes in nails, which can lead to false positives arising from the identification of somatic variants, which has been reported in up to 48% of hair follicle samples [94]. Buccal swabs have similar issues, with low DNA recovery, and false-positive and negative rates of 3.9% and 3.6%, respectively [94]. Sorted peripheral blood lymphocytes also frequently carry falsepositive somatic mutations, at rates of up to 48% [94]. Given this, there remains no substitute for germline genetic testing using DNA derived from cultured skin fibroblasts, and efforts should be made to expand access to this methodology at more centers.
All patients with an identified LoF germline variant in CHEK2 or ATM should be offered an appointment with a certified clinical genetic counselor [95]. In this visit, a comprehensive family history should be obtained, the patient counseled on the implications of the variant identified, and cascade testing of potentially affected family members should then be offered [95]. Recommendations around cancer screening vary by institution; however, female carriers of pathogenic CHEK2 mutations will typically be offered annual mammography with annual breast magnetic resonance imaging (MRI) starting at age 40 years, consideration given to a prophylactic bilateral salpingo-oophrectomy in patients with a family history of ovarian cancer, and every 5-year screening colonoscopy starting at either age 40 or 10 years prior to the youngest first-degree relative who developed colon cancer [96,97]. Similarly, carriers of pathogenic ATM mutations are recommended to undergo annual mammography with annual breast MRI starting at age 40 years [96]. For patients with a family history of pancreatic cancer, screening is offered with either endoscopic ultrasound or MRI [96]. Typically, the presence of a heterozygous ATM variant will not alter the approach to radiation or chemotherapy for patients with an established cancer, given that these patients are less sensitive to ionizing radiation than true A-T patients [98]. Generally, prophylactic mastectomy is not recommended for either CHEK2 or ATM carriers, although this remains an individualized decision [96,97]. Similarly, prophylactic bilateral salpingo-oophrectomy is not usually recommended for ATM carriers [96,97]. For hematopoietic malignancy risk mitigation in CHEK2 or ATM carriers, depending on the variant identified and the family history, appropriate measures can include a baseline bone marrow biopsy, regular peripheral blood counts, and regular lymph node examination. In the future, this may also include regular monitoring by next-generation sequencing or circulating tumor DNA for the development and evolution of CH.
Another significant aspect of identifying germline variants in myeloid and lymphoid malignancies, including CHEK2 and ATM mutations, is the prominent role for allogeneic hematopoietic stem cell transplantation (HSCT) for many of these disorders. Many patients undergoing allogeneic HSCT will utilize a related stem cell donor, either sibling or haploidentical. Given the rapid increase in the number of haploidentical transplants being performed, identification of germline variants in related donors will affect increasingly more patients [99]. Although there is not yet definitive evidence on the impact of allogeneic HSCT with a related donor with germline CHEK2 or ATM mutation, there are several potential adverse effects, including increased risk of post-transplant relapse, graft failure, and exposure of the donor to granulocyte colony stimulating factor. As such, we recommend testing of potential related donors in patients confirmed or suspected of having a germline variant and avoiding the use of a donor carrying a germline variant, including CHEK2 or ATM, unless no other options exist. There are also new therapeutic options arriving for patients with myeloid or lymphoid malignancies with deficiencies in HR, including germline CHEK2 and ATM mutations. The use of PARP inhibitors, which create synthetic lethality by blocking NHEJ in HR-deficient malignant cells, has entered into clinical trials for both myeloid and lymphoid malignancies [100]. Appropriate patients with germline CHEK2 or ATM mutations and hematopoietic malignancy should be considered for such trials, if appropriate.
Return to the clinical case: "The patient is initiated on ibrutinib and tolerates this well. You recognize that this individual has two germline mutations in the HR DNA repair pathway. You arrange for cascade testing of family members, genetic counselling, and create an individualized cancer risk mitigation plan. In addition, you make an early transplant referral so a donor without these mutations can be identified in the event allogeneic HSCT is eventually required."

Conclusion and Future Directions
Germline LoF mutations in CHEK2 and ATM are common and predispose to hematopoietic malignancies. The current evidence strongly supports a role for CHEK2 as a risk allele for myeloid malignancies, while the evidence for ATM LoF mutations primarily demonstrates an increased risk for lymphoid malignancies. It is important to include testing for both CHEK2 and ATM as a part of comprehensive screening for germline mutations in patients with myeloid and lymphoid malignancies. Identification of these variants is crucial to ensuring appropriate patient and family counseling, cascade testing, screening for malignancy prevention, allogeneic HSCT donor selection, and potentially therapy choice (e.g., PARP inhibitors). There is also early and evolving evidence that germline CHEK2 and ATM variants may be important to the development of CH, and further investigation into this interaction will be a crucial future area of study.

Conflicts of Interest
The authors declare no competing interests.

Human and Animal Rights and Informed Consent
This article does not contain any studies with human or animal subjects performed by the authors.