Abstract
Currently, there are at least a dozen recognized hereditary hematopoietic malignancies (HHMs), some of which phenocopy others. Among these, three HHMs driven by germline mutations in ANKRD26, ETV6, or RUNX1 share a phenotype of thrombocytopenia, qualitative platelet defects, and an increased lifetime risk of hematopoietic malignancies (HMs). Prior work has demonstrated that RUNX1 germline mutation carriers experience an elevated lifetime risk (66%) for developing clonal hematopoiesis (CH) prior to age 50. Germline mutations in ANKRD26 or ETV6 phenocopy RUNX1 germline mutations, but no studies have focused on the risk of CH in individuals with germline mutations in ANKRD26 or ETV6.
To determine the prevalence of CH in individuals with germline mutations in ANKRD26 or ETV6, we performed next generation sequencing on hematopoietic tissue from twelve individuals with either germline ANKRD26 or germline ETV6 mutations. Each patient had thrombocytopenia but had not developed HMs. Among the seven individuals with germline ANKRD26 mutations, one patient had a CH clone driven by a somatic SF3B1 mutation (p.Lys700Glu). This mutation increased from a variant allele frequency (VAF) of 9.4% at age 56 to 17.4% at age 60. None of the germline ETV6 mutation carriers had evidence of CH at the limits of detection of the NGS assay (5% VAF). Unlike individuals with germline mutations in RUNX1, no individuals under the age of 50 with germline mutations in ANKRD26 or ETV6 had detectable CH. This work demonstrates that ANKRD26 germline mutation carriers, but not ETV6 mutation carriers, experience elevated risk for CH.
To the Editor
Currently, there are at least a dozen recognized hereditary hematopoietic malignancies (HHMs), some of which phenocopy others. Among these, three HHMs driven by germline mutations in ANKRD26, ETV6, or RUNX1 share a phenotype of thrombocytopenia, qualitative platelet defects, and an increased lifetime risk of hematopoietic malignancies (HMs).1 Individuals with germline mutations in these hereditary thrombocytopenia/hereditary hematopoietic malignancy (HT/HHM) associated genes experience a lifetime risk for HMs of approximately 8% (ANKRD26), 33% (ETV6), or 44% (RUNX1).1
Nine unaffected RUNX1 germline mutation carriers with thrombocytopenia, but no HMs, were sequenced in a previous study. This demonstrated that 66% of these individuals had clonal hematopoiesis (CH) prior to age 50, an elevated CH risk as compared to population controls.2, 3 A subsequent study of four RUNX1 germline mutation carriers (age 49, 53, 56, and 71 years) with thrombocytopenia, but no HMs, demonstrated CH in three of these individuals.4 Germline mutations in ANKRD26 or ETV6 phenocopy RUNX1 germline mutations, but no studies have focused on the risk of CH in individuals with germline mutations in ANKRD26 or ETV6.
To address this knowledge gap, we performed a cross sectional study of twelve individuals with either germline ANKRD26 or germline ETV6 mutations who had thrombocytopenia but who had not developed HMs. We determined if ANKRD26 or ETV6 germline mutations lead to increased rates of CH, as is observed in RUNX1 mutation carriers.2, 4 Given that the penetrance of HMs is lower in ANKRD26 and ETV6 germline mutation carriers than in RUNX1 mutation carriers, we hypothesized that germline ANKRD26 or ETV6 mutation carriers would experience lower rates of CH relative to germline RUNX1 mutation carriers of similar ages.1 Additionally, all pathogenic/likely pathogenic ANKRD26 variants with supporting evidence in ClinVar are located in a regulatory domain, the 5’ untranslated region (UTR).5 RUNX1 encodes for a transcription factor (RUNX1) that binds to the ANKRD26 5’ UTR and suppresses ANKRD26 expression (Supplementary Figure 1).5 Therefore, we hypothesized that ANKRD26 and RUNX1 germline mutation carriers would experience somatic mutations in a similar set of genes as compared to ETV6 germline mutation carriers.
We enrolled twelve patients from unrelated families on Institutional Review Board-approved protocols at the University of Chicago (UChicago) or the Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo, Brazil (Supplementary Table 1). Seven unaffected ANKRD26 mutation carriers and four unaffected ETV6 mutation carriers were enrolled. Patient ages ranged from 8 to 63 years. Additionally, we included one affected ANKRD26 mutation carrier with acute myeloid leukemia (AML) with 23% blasts in order to compare CH-related mutations to an HM in this syndrome (Table 1, Supplementary Table 1). Supplementary Figure 2 shows a pedigree for each family. Each HT/HHM-related germline variant was classified using Association for Molecular Pathology and American College of Human Genetics and Genomics criteria.6 Patient samples included germline tissue (cultured skin fibroblasts) or hematopoietic tissue equivalents (peripheral blood, bone marrow, or saliva). Panel-based sequencing was performed at UChicago as described previously.7 Details regarding sample processing and sequencing are in the Supplementary Methods section.
The ages of individuals with germline ANKRD26 or ETV6 mutations at the time of sample collection are shown in Table 1. The median age at sample collection was 43 years for unaffected ANKRD26 germline mutation carriers, with serial samples from one individual. The median age for unaffected ETV6 germline mutation carriers was 59 years, with serial samples from two individuals.
Among the seven individuals with germline ANKRD26 mutations, one patient with a germline ANKRD26 mutation had a CH clone driven by a somatic SF3B1 mutation (p.Lys700Glu). This mutation increased from a variant allele frequency (VAF) of 9.4% at age 56 to 17.4% at age 60 (Table 1, Figure 1). SF3B1 p.Lys700Glu is a recognized somatic hotspot mutation that is observed in 2.1% of COSMIC HMs (n = 525/25028).8 Of note, the only patient with CH in the ANKRD26 cohort was also the oldest individual in that cohort. None of the germline ETV6 mutation carriers (n = 4) had evidence of CH at the limits of detection of the NGS assay (5% VAF). Unlike individuals with germline mutations in RUNX1, no individuals under the age of 50 with germline mutations in ANKRD26 (n=6) or ETV6 (n=3) had detectable CH despite nearly half of the unaffected samples being collected from individuals in this age group (Figure 1).2, 4
The only patient with a germline ANKRD26 mutation and a malignancy (AML) had a leukemic clone with both typical and atypical driver mutations: CUX1 (p. Phe472GlnfsX105, VAF 92.4%), RUNX1 (p.Arg320X, VAF 55.8%), TET2 (p.Phe1309LeufsX54, VAF 40.0%), FLT3 (p.Asp835His, second tyrosine kinase domain (TKD), VAF 18.3%), and SAMD9 (p.Val798GlyfsX7, VAF 13.3%) (Table 1). FLT3 is mutated in 18.8% of COSMIC HMs,8 and RUNX1 somatic mutations are the most common second hit in RUNX1 germline mutation carriers who have developed HMs.8, 9 Given the role of RUNX1 in regulating ANKRD26 expression, the RUNX1 mutation in this ANKRD26 germline mutation carrier may effectively represent a second hit that serves as a late leukemogenic event.5 The CUX1 and SAMD9 mutations were not described previously in COSMIC.8 The leukemic karyotype was 46, XX, -6, del(7)(q11.2),+mar[20].
The total observation time for the ANKRD26 cohort was 275 years. The incidence of CH in the ANKRD26 cohort was 4.5×10−3 CH cases/observation year (4.5 CH cases per 1000 observation years). The incidence rate of HMs in the ANKRD26 cohort was 3.6×10−3 malignancies/observation year (3.6 malignancies per 1000 observation years). This HM incidence rate was similar to that seen previously in a cohort of Italian germline ANKRD26 mutation carriers (2.13 malignancies per 1000 observation years).10 The observation time for the ETV6 cohort was 258 years, with no diagnoses of CH or HMs.
Among the known HT/HHM phenocopies, only RUNX1 has been systematically evaluated for CH risk. This bias has likely occurred for two reasons. First, RUNX1-driven HT/HHMs were identified 12 years before ANKRD26-driven HT/HHMs and 16 years before ETV6-driven HT/HHMs, which has provided a longer period of time for researchers to identify and study families with germline RUNX1 mutations.1, 11-13 Second, RUNX1-driven HT/HHMs have the highest penetrance for HMs among the HT/HHM phenocopies, with approximately 44% of mutation carriers developing blood cancers. This penetrance is higher than that experienced by ANKRD26 (8%) and ETV6 (33%) germline mutation carriers.1 In our clinical experience, the most “severe” hereditary syndromes are more easily recognized than syndromes with more subtle symptoms and lower penetrance phenotypes. Therefore, it is not surprising that prior work in the HHM field has largely focused on RUNX1-driven HT/HHMs.
To our knowledge, this is the first study examining pre-leukemic states in the HT/HHM phenocopies driven by germline ANKRD26 or ETV6 mutations. In our cohort, CH was detected in 14% of ANKRD26 germline mutation carriers, but no CH was present in ETV6 germline mutation carriers. No ANKRD26 or ETV6 mutation carriers developed malignancies during 533 years of observation time. It is possible the limited number of germline variants and families in this study, with four families carrying three ANKRD26 variants and one family with one ETV6 variant, are not representative of the leukemogenic risk observed in the full spectrum of HT/HHM-related ANKRD26 or ETV6 germline variants. Ultimately, larger numbers of germline ANKRD26 or ETV6 mutation carriers should be studied to better determine the pre-leukemic genetic milieu that exists in these syndromes.
In conclusion, this is the first cross sectional study focused on leukemogenic mechanisms in individuals with ANKRD26-or ETV6-driven HT/HHM phenocopies. We identified CH in 14% of older germline ANKRD26 mutation carriers but did not detect CH in ETV6 germline mutation carriers. We did not detect early-onset CH under the age of 50 in individuals with germline ANKRD26 or ETV6 mutations, as has been observed in RUNX1 mutation carriers.2 We also identified a rare somatic RUNX1 mutation in a germline ANKRD26 mutation carrier with AML, which may effectively represent a second hit event given the role of RUNX1 in regulating ANKRD26 expression.5 Given the relatively small sample size of our cohort and the limited number of pedigrees with germline ANKRD26 mutations worldwide, future studies focused on evaluating leukemogenic mutations before and after the development of HMs in ANKRD26 or ETV6 germline mutation carriers should be performed.
Disclosures
MWD has received consulting fees from Cardinal Health, Inc. HSS has received honoraria from Celgene, Inc. LAG receives royalties from a coauthored article on inherited hematopoietic malignancies in UpToDate, Inc.
Supplementary Files
Supplementary Methods
Sample processing and next generation sequencing methods
Genomic DNA (gDNA) was extracted with the QIAamp DNA Blood Mini Kit (Qiagen) following the manufacturer’s instructions. DNA concentrations were measured via a Nanodrop (Thermo Scientific) and/or Qubit fluorometer (Life Technologies). At least 100 ng of genomic DNA from each sample was sheared, selected by size, ligated to adapters, and standard sequencing libraries were generated via PCR amplification. Following library generation, genomic capture was performed using a custom SeqCap EZ capture panel that covered 1212 genes (Roche), and an additional PCR amplification with real-time quantitative PCR quantification was performed. An Illumina HiSeq was used to sequence the pooled capture libraries. Sequencing data were stored on a protected high-performance computing system at UChicago that exceeds requirements for the Health Insurance Portability and Accountability Act. The data were initially analyzed via a bioinformatics pipeline that melded publicly available packages built off of the GATK package and a custom bioinformatics pipeline developed at UChicago. These data were initially reviewed by MWD and KY for driver mutations.14 Following an initial round of review, the raw FASTQ files were then transferred to the University of South Australia in order to analyze the data using the freebayes-based RUNX1db bioinformatics pipeline.15 The data were filtered for read quality and depth as previously described, with thresholds as follows: variant allelic depth >= 5, read depth >= 20, population prevalence (variants at 0.1% or higher in any population database were removed), pathogenicity (missense variants that were not predicted to be damaging in 2 or more in silico predictors were removed; CADD scores with values less than 20 or higher were removed), and oncogenicity (variants not in genes with known roles as drivers in myeloid malignancies, not in COSMIC, or RUNX1 variants were removed). We analyzed the subsequent list of candidate variants and used IGV to review each variant of interest manually in the individual BAM files. We removed any variants labeled as artifacts after the aforementioned steps.
We then analyzed the remaining IGV-confirmed variants to label each variant as germline or somatic in origin. For individuals with sequencing data from cultured skin fibroblasts, we compared variants identified in hematopoietic tissue equivalents directly to data obtained from cultured skin fibroblasts. Samples without paired germline tissue were analyzed using a combination of population allelic frequency (minor allele threshold of 0.01% or lower), VAF (with likely germline VAFs considered to be between 30 and 60% for genes on autosomal chromosomes and 80% or higher for genes on the X chromosome), and the frequency of the variant in question in tumor databases such as COSMIC. Any variant passing the above population filters, but which still occurred more than twice in COSMIC, was considered to not be of germline origin.
This filtering process produced a list of variants of likely somatic or definitive somatic origin which we reviewed manually for clinical and biological relevance. The determination of “likely somatic” or “somatic” origin adhered to criteria defined in the original RUNX1 database manuscript.15 “Clinically relevant” variants were known pathogenic germline variants in leukemia or variants that were present more than twice in COSMIC in hematopoietic and lymphoid samples (H&L samples). Novel driver variants were clinically relevant if they were present in a gene known to be recurrently mutated in COSMIC H&L samples, were a truncating variant (nonsense, frameshift indels, essential splice site variants), were in the same domain as known pathogenic variants (for example, the RUNT domain in RUNX1), or were a deletion in a gene where deletion is a known mechanism of disease. Missense variants were considered to be clinically relevant if they were damaging in at least 3 in silico algorithms and were highly conserved via GERP and Phylop scores. All somatic and likely somatic variants that did not meet criteria for clinical relevance were categorized as “possibly relevant” or “of unknown relevance”.
Acknowledgements
We thank the patients and their families for their participation in this research program and for providing samples. This work was supported by the Damon Runyon Cancer Research Foundation Physician-Scientist Training Award, the Edward P. Evans Foundation Young Investigator Award, the Cancer Research Foundation Young Investigator Award, and the NIH Paul Calabresi K12 Program in Oncology (MWD). KY, EK, RS, and PPL are supported by the Intramural Research Program at the National Human Genome Research Institute, NIH.