Germline variants in DNA repair genes, including BRCA1/2, may cause familial myeloproliferative neoplasms

Myeloproliferative neoplasms (MPNs) typically occur sporadically and are caused by somatic driver mutations in the JAK2 (Janus kinase 2), CALR (calreticulin), and MPL (thrombopoietin receptor) genes. However, familial clustering occurs in 7% to 8% of cases, with relatives of MPN patients exhibiting a five- to sevenfold MPN risk.^1-3 Interestingly, driver mutations and MPN phenotypes vary between family members,³ suggesting a common germline predisposition for the subsequent driver mutations . Known predisposing germline genetic variations and associated MPN risks include: JAK2 46/1 haplotype (up to fourfold)^4-6; telomerase reverse transcriptase TERT promoter variant (rs2736100) (2-fold),⁷ duplication of ATG2B (autophagy related 2B) and GSKIP (GSK3B interacting protein; high risk of progression from essential thrombocythemia [ET] to myelofibrosis or leukemia),⁸ and germline RBBP6 (retinoblastoma binding protein 6) mutations (∼5% of familial MPN cases).⁹ 

Our study was approved by the ethics committee of the Medical Faculty of RWTH Aachen University (EK306) and was conducted in accordance with the Declaration of Helsinki. To identify additional predisposing germline factors in familial MPNs, we performed whole-exome sequencing (WES) in 5 families affected by MPNs and carrying a somatic JAK2 V617F (c.1849G>T) mutation (Figure 1). WES was carried out by analyzing the DNA of peripheral blood samples on a NextSeq500 Sequencer with 2 × 75 cycles on a high-output flow cell (Illumina, San Diego, CA). A probe-based capture method was used to enrich the target regions (Integrated DNA Technologies, Coraville, IA, or Nextera Rapid Capture Exome [version 1.2]; Illumina). Alignment to the reference genome (hg19 or hg38) and variant calling were performed using an in-house pipeline based on SeqMule (http://seqmule.openbioinformatics.org/en/latest). Variant analysis and prioritization were performed using Kggseq software (http://pmglab.top/kggseq/). Variants of interest were confirmed by Sanger sequencing on an ABI3500 platform (Applied Biosystems, Foster City, CA); to exclude a somatic origin, DNA from hair roots and/or finger nails was used for Sanger sequencing.

In 4 of 5 families, we found heterozygous germline variants in DNA repair genes associated with hereditary cancer predisposition syndromes, especially hereditary breast and ovarian cancers (Figure 1).

In family 1, a brother (1.1) and sister (1.2) with prefibrotic myelofibrosis carried a heterozygous loss-of-function mutation in the BRCA1 (breast cancer 1) gene (OMIM*113705, transcript NM_007294.3). The sister (1.2) was first diagnosed with MPN at age 53 years; she later developed uterine cancer at age 61 years. The variant c.2722G>T, p.(Glu908*) leads to a premature translational stop and is not listed in the Genome Aggregation Database (gnomAD). It is reported once (1 in 8600 alleles) in the Exome Sequencing Project database (as of 1 February 2019) and has been described in patients with hereditary breast or ovarian cancer.¹⁰ According to the American College of Medical Genetics and Genomics (ACMG) 2015 guidelines, the variant must be classified as pathogenic. Family history of the patients in family 1 was conspicuous for cancer, with deaths resulting from colorectal cancer in the mother at age 65 years and 1 brother at age 77 years, ovarian cancer in 1 sister at age 52 years, and leukemia in 1 brother at age 5 years.

In family 2, a father (2.3) and daughter (2.4) presented with ET. Subsequently, the father experienced progression to AML. In both individuals, a missense variant in the BRCA2 (breast cancer 2) gene (OMIM*600185, transcript NM_000059.3) was identified. The variant c.4460A>C, p.(Lys1487Thr) has not yet been described in large genome databases (eg, gnomAD as of 23 December 2019) or in the literature. High conservation of the affected amino acid argues for its functional relevance. Bioinformatic prediction programs predicted the effect of the variant inconsistently: SIFT (version 6.2.0), tolerated (score, 0.57; median, 2.68); MutationTaster (version 2013), disease causing (probability, 0.971); and PolyPhen-2 (HumVar), possibly damaging (score, 0796). Formally, the variant must be classified as a variant of uncertain significance according to the ACMG criteria. Family history of the rather small family was unremarkable. Interestingly, patient 2.3 showed an excellent response to decitabine monotherapy for AML.

In a mother (3.5) and daughter (3.6) of family 3 who had ET (although prefibrotic myelofibrosis could not be completely ruled out) and PV, respectively, the pathogenic frameshift variant c.1100del (p. Thr367Metfs*15) in the CHEK2 (checkpoint kinase 2) gene (OMIM*604373, transcript NM_007194.4) was identified, likely leading to loss of protein function. It is listed in gnomAD (as of 3 October 2020) with a frequency of 0.21% and has been described in association with tumor predisposition as the most common pathogenic CHEK2 mutation.¹¹ According to the ACMG guidelines, the variant is classified as pathogenic. A predisposition to MPNs caused by this variant has already been postulated in the literature.¹² The patient’s (3.5) mother (grandmother of patient 3.6) died as a result of leukemia not further specified at age 46 years. The father of patient 3.5 (grandfather of patient 3.6) died as a result of bladder cancer at age 75 years. Her cousin died as a result of ovarian cancer in young adulthood.

In family 4, the father (4.7), affected by prostate carcinoma at age 64 years (treated by surgery only), was diagnosed with PV at age 69 years, which subsequently progressed to AML at age 75 years. His daughter (4.8) suffered from ET, which subsequently progressed to PV. Both (4.7 and 4.8) carried a heterozygous missense variant in the tumor predisposition gene ATM (ataxia-teleangiectasia mutated; OMIM*607585, transcript NM_000051.3). The variant c.1009C>T, p.(Arg337Cys) is listed with a low frequency of 0.0092% (15 December 2020). Bioinformatically, the alteration is classified as likely disease causing (PolyPhen-2, MutationTaster, and SIFT). Formally, the variant must be classified as a variant of uncertain significance according to the ACMG criteria. Interestingly, family history was conspicuous for breast cancer. On the father’s side, 1 sister developed bilateral breast cancer, and another sister died at age 85 years as a result of breast cancer.

In a mother (5.9) and daughter (5.10) age 11 years of family 5, both affected by ET, we did not identify any pathogenic or likely pathogenic variant in the applied WES approach or BRCA1/2 multiplex ligation-dependent probe amplification. Submicroscopic chromosomal imbalances were excluded in the patients of family 5 by single-nucleotide polymorphism array analysis (CytoScan HD array; Affymetrix, Santa Clara, CA). Family history was not unremarkable. The deceased mother of 5.9 (grandmother of 5.10) had been diagnosed with metachronous breast cancer. Further information regarding this was not available.

While an association of germline variants in ATM⁺, CHEK2⁺, and JAK2 V617F⁺ MPNs has already been postulated,¹² such an association, to our knowledge , has not been demonstrated for the high-risk hereditary cancer genes BRCA1 and BRCA2. Nevertheless, germline heterozygous pathogenic variants in genes of the Fanconi anemia–BRCA pathway are known to predispose for hematological malignancies, which may include an increased risk for MPNs.¹³ In addition, we have published data demonstrating that JAK inhibition by ruxolitinib decreases the BRCA protein in JAK2 V617F–mutant MPN cells and renders them more susceptible to PARP inhibitor–induced apoptosis,¹⁴ supporting a functional interaction between the Fanconi anemia–BRCA and JAK-STAT pathways. These results suggest that BRCA haploinsufficiency may predispose patients to an MPN but also render them more susceptible to combined JAK and PARP inhibition. The exact mechanisms are not yet understood but could involve germline predisposition to genomic instability and subsequent acquisition of oncogenic mutations, such as JAK2 V617F, which in turn stimulate increased reactive oxygen species production and again result in genomic instability. Interestingly, mutant CHEK2 has already been demonstrated to enhance stem cell expansion in both the absence and presence of genotoxic stress.¹⁵ 

The identification of an underlying hereditary tumor predisposition has therapeutic consequences for patients as well as their families. The detection of a pathogenic germline variant in the high-penetrance genes BRCA1 and BRCA2 should prompt the discussion of prophylactic surgery (mastectomy or oophorectomy). The fact that 2 patients with ET or PV experienced progression to AML may suggest a high risk of leukemic transformation and the necessity for closer monitoring of clonal evolution. Germline mutations in DNA repair genes may affect the type of cytoreductive therapy (eg, the cytostatic agent hydroxyurea may be more toxic but also more efficacious than other treatments). Moreover, in the case of a planned allogeneic stem cell transplantation, it may be important to exclude the pathogenic germline mutation in suitable family donors and also to adapt conditioning protocols in the respective patient.

In summary, the data suggest that MPNs are more frequently triggered by germline mutations than previously assumed, and conversely, the spectrum of known tumor predisposition syndromes seems to be broader than anticipated.

The authors thank Elvira Golz-Staggemeyer and Stephan Klinkenberg for excellent technical support.

L.L.T. was supported by Deutsche Forschungsgemeinschaft (German Research Foundation) grants TRR237-369799452 and EXC2151-390873048. This work was supported in part by the Interdisciplinary Oncologic Project Groups for MPN/CML and for Hereditary Tumor Syndromes of the Center of Integrated Oncology at Aachen-Bonn-Cologne-Düsseldorf, funded by Deutsche Krebshilfe (German Cancer Aid).

Contribution: The study was designed by M.E., L.L.T., and S.K. Data were collected by M.E., R.M., K.K., D.G., U.G., S.I., and S.K. Data were analyzed by all authors. The paper was written by M.E. and S.K. and edited by all authors. All authors approved the final manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Steffen Koschmieder, Department of Hematology, Oncology, Hemostaseology, and Stem Cell Transplantation, Pauwelsstr 30, 52074 Aachen, Germany; e-mail: skoschmieder@ukaachen.de.