Predisposition to hematopoietic malignancies by deleterious germline CHEK2 variants

Abstract

The role of germline CHEK2 variants in hematopoietic malignancies (HMs) is poorly understood. We examined pathogenic/likely pathogenic (P/LP) CHEK2 variants in patients with hereditary HMs (HHMs), a solid tumor risk cohort, public datasets, and a knock-in mouse model. In the HHM cohort, 57 probands had germline P/LP CHEK2 variants, mostly p.I157T (53%, 30/57). Among CHEK2 p.I157T carriers, 43% (19/44) had myeloid malignancies, 32% (14/44) had lymphoid malignancies, and 2% (1/44) had both. Among those with other germline P/LP CHEK2 alleles, 36% (13/36) had myeloid malignancies, 28% (10/36) had lymphoid malignancies, and 6% (2/36) had both. CHEK2 p.I157T was enriched in HM patients (OR 6.44, 95%CI 3.68-10.73, P < 0.001). In a solid tumor risk cohort, 36% (15/42) of CHEK2 p.I157T patients had a HM family history. A genome wide association study showed enrichment of CHEK2 loss-of-function variants with myeloid leukemia (P = 5.78e^-7). In public acute myeloid leukemia (AML) datasets, 1% (16/1348) of patients had P/LP CHEK2 variants. In a public myelodysplastic neoplasms (MDS) dataset, 2% (5/214) had P/LP CHEK2 variants. Chek2 p.I161T mice, homologous to human p.I157T, had worse survival as heterozygotes (P = 0.037) or homozygotes (P = 0.005), with fewer Lin-CD34+ and Lin-cKit+ cells. Our data suggest P/LP CHEK2 variants are HHM risk alleles.

Fig. 1. Breakdown of patients with CHEK2 variants in the University of Chicago (UC) cohort.

A A total of 80 patients from 57 families (unique probands) with pathogenic CHEK2 variants (CHEK2-path) were identified from the UC HHM cohort. Amongst these 80 patients, 59 had a HM. In the subset of patients with the CHEK2 p.I157T variant, 34 had a HM. In the subset of patients who were included in the allele burden calculation who had sequential, clinical testing (N = 1085), 851 had a HM and 30 of these had CHEK2-path, with the most common variant being CHEK2 p.I157T (N = 15).

Fig. 2. Germline CHEK2 variants identified in a cohort of patients tested for HHM.

A CHK2 protein schematic showing the locations of the encoded protein variants identified among 69 probands tested for HHMs. CHK2 protein domains are outlined: SQ/TQ (green bar), forkhead-associated (FHA) domain (red bar), and the kinase domain (black bar). Diamonds depict P/LP variants, and circles denote VUS. Red, missense variants; green, intronic and UTR variants; blue, frameshift variants; yellow, nonsense variants; and purple, deletion variants. N gives the number of independent probands/families with a particular allele. B The frequency with which individual alleles were identified amongst probands. Self-reported ethnicities were given by color: Western Europe, red; Polish, blue; Unknown, green; Ashkenazi, purple; Mixed, orange; Eastern Europe, yellow. C Distribution of HMs in those with the CHEK2 p.I157T allele. Color indicates the order of the malignancy for those diagnosed with multiple cancers: first cancer, red; second cancer, blue; third cancer, green. D Cytogenetic and molecular profiles of the myeloid malignancies that developed in those with germline P/LP CHEK2 variants. E–G Representative pedigrees from probands with P/LP germline CHEK2 variants. The proband is indicated by the triangle. The germline CHEK2 variant is indicated for those family members who were genotyped. Cancer diagnoses are given below each pedigree member along with the age at diagnosis (dx.) or death (dc.) when known. Circles indicate women, and squares, men. The generation number is given to the left in Roman numerals. (ns not significant. * P < 0.05, ** P < 0.01, *** P < 0.001).

Fig. 3. HM phenotype in patients with P/LP germline CHEK2 variants tested for solid tumor cancer risk.

Characteristics of patients tested for solid tumor risk indications at the University of Utah who have the CHEK2 p.I157T variant. A Personal history of cancer for those with the CHEK2 p.I157T allele. B Primary malignancy type for those with the CHEK2 p.I157T allele who were diagnosed with cancer. C Family HM history for those with the CHEK2 p.I157T allele. D Type of HM within the family for those with the CHEK2 p.I157T allele. E–G Representative pedigrees of patients with the germline CHEK2 p.I157T allele from the University of Utah cohort. The proband is indicated by the triangle. The germline CHEK2 variant is indicated for those family members who were genotyped. Cancer diagnoses are given below each pedigree member along with the age at diagnosis (dx.) or deceased (dc.) when known. Circles indicate women, and squares, men. The generation number is given to the left in Roman numerals.

Fig. 4. Phenome-wide association study (PheWAS) for CHEK2 predicted loss of function (pLoF) variants and genome-wide association study (GWAS) for myeloid malignancy.

A Phenome wide association study (PheWAS) for CHEK2 predicted loss of function (pLoF) variants from the UK Biobank (UKBB). Hematologic phenome terms are indicated by dark red circles and malignancy diagnoses by light green circles. B Genome-wide association study (GWAS) from the UK Biobank for ICD-10 code C92 myeloid malignancy from the UK Biobank, significance lines indicated at -log₁₀(5e^-08) and -log₁₀(5e^-04) levels. C Cumulative proportion of participants from UKBB with malignant cancer who are CHEK2-wild type (wt) or carry pathogenic/likely pathogenic variants (CHEK2-path) by age of onset. D Cumulative proportion of CHEK2-path UKBB participants with HM diagnosis by age of onset.

Fig. 5. Characteristics of MDS and AML patients with P/LP CHEK2 variants called from public RNA-seq datasets.

A Somatic mutational landscape and disease characteristics of patients with AML with identified P/LP CHEK2 variants. B Median age at time of disease sample in AML patients who are CHEK2-wt (blue, left) vs. those with CHEK2-path variants (red, right). C Burden of bone marrow blasts at time of disease sample in AML patients who are CHEK2-wt (blue, left) vs. CHEK2-path (red, right). D Frequency of CHEK2-path variants in a cohort of MDS patients with bulk RNA-seq from bone marrow mononuclear cells or CD34+ cells. (ns not significant, * P < 0.05, ** P < 0.01, *** P < 0.001).

Fig. 6. Hematopoietic stem and progenitor (HSPC) features of a Chek2 p.I161T mouse model.

A Survival in Chek2^wt/wt (blue, N = 13), Chek2^p.I161T/wt (red, N = 16), vs. Chek2^{p.I161T/p.I161T} (green, N = 11) mice. B Representative flow plots from lineage depleted marrow demonstrating the Lin-CD34+ and Lin-Sca1+cKit+ compartments in young (2-6 month) Chek2 wt (Chek2^wt/wt), heterozygous (Chek2^p.I161T/wt), and homozygous (Chek2^{p.I161T/p.I161T}) mice. C Comparison of Lin-CD34+ compartment in young Chek2 mice. D Comparison of Lin-cKit+ compartment in young Chek2 mice. E Comparison of the Lin-Sca1+cKit+ compartment in young Chek2 mice. F Representative flow plots from whole bone marrow nucleated cells (BMNC) demonstrating the CD34+ and cKit+ in aged (24 month) Chek2 wt (Chek2^wt/wt), heterozygous (Chek2^p.I161T/wt), and homozygous (Chek2^{p.I161T/p.I161T}) mice without pathologic evidence of malignancy. G Comparison of cKit+ compartment in aged Chek2 mice. H Comparison of CD34+ compartment in aged Chek2 mice. I Pre-ranked gene set enrichment analysis (GSEA) for differentially expressed genes from Lin-CD34+ cells from young (aged 5-6 month) Chek2 mice. Highlighted pathways are significant at a false discovery rate (FDR) of <0.25. Pathways from MSigDb m2.all gene set with some titles abbreviated. Full pathway titles and details are in Supplemental Table 7. J–L Summary of consensus diagnosis obtained by analysis of gross features, histology, IHC, and multicolor flow cytometry data from wt (Chek2^wt/wt), heterozygous (Chek2^p.I161T/wt), or homozygous (Chek2^{p.I161T/p.I161T}) mice. M Comparison of mice with identified pathologic abnormalities at endpoint for Chek2-wt (left) vs. Chek2-mt (right), including heterozygous or homozygous mice. (ns, not significant, * P < 0.05, ** P < 0.01, *** P < 0.001).