Pathogenicity evaluation of coding germline variants identifies rare alleles enriched in hematological patients of a founder population

Jessica R. Koski; Laura Langohr; Tuulia Räisänen; Atte K. Lahtinen; Marja Hakkarainen; Caroline A. Heckman; Ulla Wartiovaara-Kautto; Esa Pitkänen; Outi Kilpivaara

doi:10.1101/2024.10.23.24315723

Abstract

Background The clinical significance of most germline variants in hematological malignancies (HMs) remains unknown. This presents a challenge in the clinical setting, as the inability to accurately detect pathogenic variants can influence therapeutic decisions. Population isolates have been shown to be beneficial in pathogenic variant discovery due to presence of rare deleterious variants in relatively high frequencies.

Methods We developed and applied PaVaDi, a computational pipeline that follows American College of Medical Genetics and Genomics (ACMG) guidelines, to evaluate the pathogenicity of germline variants in 511 HM patients from the Finnish founder population. We conducted an exome-wide burden analysis to assess the overall contribution of pathogenic variants to HMs and identified significant gene associations. We also examined genes previously associated with hematological diseases and DNA repair in more detail, and performed protein stability analyses to resolve variants of unknown significance (VUS).

Results The exome-wide burden analysis revealed potential pathogenic alleles in CUX2, RNPC3, and MFSD2A that have not previously been linked to HM predisposition. We also identified the largest series of CHEK2 variant carriers reported in hematological diseases, including pathogenic/likely pathogenic (P/LP) variants (n=19), Ile200Thr (i.e., Ile157Thr) (n=49), and other variants of uncertain significance (n=3). CHEK2 variants were 1.7-fold enriched in patients compared to controls (13.9% vs 8.3%, p=2×10⁻⁵). Strikingly, Ile200Thr was enriched over four-fold in acute lymphoblastic leukemia patients. Finally, protein structure stability analyses suggested novel MPO variants to be potentially highly deleterious.

Conclusions This study highlights the importance of germline testing in hematological malignancies and demonstrates the utility of population isolates for pathogenic variant discovery. Our findings identify a significant burden of deleterious variants in HM patients, particularly in CHEK2, and underscore the potential of multi-disease joint analyses in revealing germline contributions to hematological diseases.

Background

Interpretation of germline genetic analyses challenges modern hematological disease diagnostics and care. In clinical practice, the quality of pathogenicity evaluation can vary substantially depending on evaluation procedures in place and training of clinical geneticists (1,2). Guidance established by ACMG is being used to combine evidence supporting variant pathogenicity, providing a rule-based framework to classify germline variant pathogenicity into five categories (3). The ACMG framework integrates functional evidence, in silico pathogenicity predictions, population allele frequency information and other sources of evidence. Many computational tools implementing the ACMG guidelines have been developed (4–7). However, the degrees of subjectivity and uncertainty allowed in the guidelines have been found to lead to inconsistent and partly incompatible results (8).

Many studies and also clinical practice use gene panels of known predisposing genes to circumvent the challenges in variant interpretation. This approach does not allow identification of novel predisposing genes and restricts the detection of variants of unknown significance (VUS). Furthermore, the lack of well-matched population controls complicates high-quality evaluation of rare germline variants. This is exemplified when analyzing findings in populations underrepresented in genetic variant repositories such as gnomAD (i.e., non-Europeans).

Despite successes in revealing the role of somatic genetic changes, hereditary predisposition in hematological malignancies has been acknowledged only recently. Reports typically focus on just one single hematological disease/malignancy or a gene (9). However, identification of pathogenic germline variants is crucial in hematology, as if timely recognized, they can direct therapeutic decisions, such as conditioning modalities or donor selection in allogeneic hematopoietic stem cell transplantation (HSCT) (10,11).

The Finnish population has been isolated, and has undergone genetic bottlenecks which left the population with a higher number of deleterious low-frequency variants compared to larger, mixed populations (12). This enrichment of variants allows powerful detection of genotype-phenotype associations of rare deleterious variants. In this study, we report a comprehensive pathogenicity evaluation of germline variants discovered in exome sequence data of 511 Finnish adult hematological patients (Fig. 1). To this aim, we developed PaVaDi, an automated pipeline implementing ACMG guidelines for variant pathogenicity classification. In gene panel level analysis we focused on genes associated with hematological diseases (HemMut) or DNA repair (DDR) and identified multiple P and LP variants. In an exome-wide burden analysis we identify three novel hematological malignancy predisposing gene candidates (CUX2, RNPC3, and MFSD2A) and affirm the role of POT1. We also demonstrate evidence for pathogenicity of certain VUSes in MPO. Finally, we solidify the significance of germline CHEK2 variants doubling the risk for hematological malignancies.

Figure 1.

Overview of the sample set, study design and analysis workflow. AML, acute myeloid leukemia; ALL, acute lymphoid leukemia; BM, bone marrow; BMF, bone marrow failure; CML, chronic myeloid leukemia; MDS, myelodysplastic syndrome; MM, multiple myeloma; MPAL, mixed phenotype acute leukemia; MPN, myeloproliferative neoplasm; LOH, loss-of-heterozygosity; LP, chronic lymphoproliferative disease; Mb Waldenström, Waldenström’s macroglobulinemia.

Methods

Patients, samples, and exome sequencing

The study set consisted of 511 adult patients diagnosed with various hematological diseases, mostly malignancies in Finland. Patient characteristics were retrieved from hospital records and the Finnish Hematological Registry, and are summarized in Table 1 and in Supplementary Table 15. Samples were collected between 2011 and 2022 at the Department of Hematology of Helsinki University Comprehensive Cancer Center in (HUH CCC) and the Finnish Hematology Registry and Clinical Biobank (FHRB Biobank). Whole exome sequencing (WES) data was available for all 511 patients. Germline sequencing was performed on skin biopsy samples except for one remission on blood sample as described previously (13) with Nimblegen SeqCap EZ exome v2.0 (Roche, Basel, Switzerland), The SeqCap EZ MedExome (Roche, Basel, Switzerland), Twist Core Exome (Twist Bioscience, SF, USA) or Agilent Clinical Research Exome (Agilent, Santa Clara, CA, USA). In addition to the 511 germline WES, tumor WES was available for 460 patients (88.8%). Tumor samples were obtained from bone marrow (N=297), peripheral blood (N=31), CD138+ cells (N=121), CD3+; CD34+; CD4+/CD8+; CD8+ (N=1 in each), cerebrospinal fluid (N=1) or unknown primary site (N=6) at diagnosis, remission or relapse stage of the disease. Tumor WES had been performed as described previously (14). Both normal and tumor sequencing was performed at the Institute for Molecular Medicine Finland (FIMM) Genomics unit supported by HiLIFE and Biocenter Finland or at the Biomedicum Functional Genomics Unit (FuGU) at the Helsinki Institute of Life Science and Biocenter Finland at the University of Helsinki.

View this table:

Table 1. Patient characteristics.

Population controls

GnomAD (v.2.1.1) global (n=125,748) and non-cancer Finns (n=10,816) (15) were used as population controls in variant level Fisher’s exact tests and gene level burden tests.

Germline variant calling

To detect germline single-nucleotide variants (SNVs), multiple-nucleotide variants (MNVs) and short insertions or deletions (indels), variant calling was performed on normal tissue WES data using Freebayes (v.1.3.2) (16) in population sample calling mode with parameters -min-repeat-entropy 1, report-genotype-likelihood-max, -alternate-fraction 0.2, and -no-partial-observations. Raw variant calls were filtered to include only good quality variants in the set (filters for quality QUAL>20, QUAL/AO>2; strand bias artifacts SAF>1,SAR>1; read position artifacts RPR>1,RPL>1). Variants were normalized using bcftools (version 1.12) (17) and variants with multiple alleles were decomposed using vt (version 0.57) (18). Variants were further filtered on a sample level by including only calls with sample genotype quality (GQ) ≥ 20, coverage > 7 and allelic fraction (AF) ≥ 0.3.

Somatic mutation calling

Somatic mutations were detected using Mutect2 (included in GATK (v4.2.0) (19)) in tumor-normal mode. A panel of normals was constructed for each sequencing kit used and gnomAD v.2.1.1 (15) was used as a germline variant resource. Somatic calls were filtered with GATK FilterMutectCalls. Only mutations with >4 supporting reads were included in the analysis.

Evaluation of genetic ancestry

Genetic ancestry of the patients were evaluated by computing the first four principal components (PCs) of genotypes. Principal component analysis (PCA) was performed using the prcomp function in R (version 4.3.2). Prior to PCA, the variant data were pruned for linkage disequilibrium (LD) using PLINK software (version 2.0). LD pruning was performed with a window size of 50 single nucleotide polymorphisms (SNPs), a step of five SNPs and an r² threshold of 0.2. PCs were visualized to identify clusters corresponding to different ancestral populations. Eight patients were found to have non-Finnish ancestry and were removed from the exome-wide variant burden analysis.

Gene panels

To examine genes previously associated with hematological disorders and DNA repair, we composed two panels of genes. The first panel consisted of 297 genes associated with inherited hematological disorders (HemMut; a panel developed for diagnostics at HUSLAB Laboratory of Genetics, HUS Diagnostic Center, Helsinki University Hospital) (Supplementary Table 1). The second panel consisted of 276 DNA damage repair (DDR) genes (Supplementary Table 2) (20).

Inheritance patterns

Inheritance patterns were retrieved from Clinical Genomic Database (version December 2023) except for SBDS, MYSM1 and MPO for which we annotated the inheritance patterns as AR, AR and AD/AR respectively.

PaVaDi, a variant pathogenicity classification pipeline

The ACMG has published guidelines including 28 criteria for the clinical interpretation of germline variants with respect to human diseases (3). Following a rule-based strategy outlined in the guidelines, variants can be classified as pathogenic (P), likely pathogenic (LP), benign (B), likely benign (LB) or as variants of uncertain significance (VUS). There are different levels of pathogenicity supporting criteria: very strong (PVS1), strong (PS1-PS4), moderate (PM1-PM6) and supporting (PP1-PP5). Criteria supporting benign classification are stand-alone (BA1), strong (BS1-BS4) and supporting (BP1-BP7). If no criteria is met, or B/LB and P/LP evidence is contradictory, the variant is classified as VUS.

We developed an automated pipeline PAthogenic VAriant DIscovery (PaVaDi) that implements 17 of the criteria (PVS1, PS1, PM1, PM2, PM4, PM5, PP2, PP3, PP5, BA1, BS1, BS2, BP1, BP3, BP4, BP6, BP7) in addition to three strengthened criteria PM2_s, PP5_s, BP6_s that could be automatized with the data available to us (Supplementary Table 3; Supplementary Text). The pipeline consists of an annotation module scoring each variant with a set of rules, each 1 or 0, based on the available evidence, and a classification module assigning a pathogenicity level (B, LB, VUS, LP or P) based on the scored rules (Supplementary Figure S2).

We integrated multiple tools and databases that can be used to assess the significance of genetic variants to disease into the pipeline (Supplementary Table 4). They provide evidence on variant pathogenicity for the criteria, upon which the pipeline is assigning the variants a pathogenicity level (B, LB, VUS, LP or P). We used variant and gene lists from the InterVar package for evidence in the scoring of criteria (21).

Exome-wide variant burden analysis

We studied the enrichment of rare variants in the patient set compared to gnomAD Finns without a cancer diagnosis. We performed a burden test with the Total Frequency Test (TFT) (22), collating rare P+LP, B+LB, and VUS variants at the gene level in each pathogenicity class separately. To alleviate potential detection bias due to differences in sequencing technologies and data analysis methods between the patient set and the population controls, we included only non-synonymous variants which were present in both the patient set and controls, and altogether had at least three variant alleles in the patient set and controls combined. Genomic inflation in the data set was corrected by dividing χ² statistics with λ using P/LP, B/LB and VUS burden per gene, where λ is the observed median χ² statistic divided by the expected median χ² statistic (Supplementary Figure S3-5). Benjamini-Hochberg procedure was used for multiple testing correction with a false discovery rate (FDR) threshold at 10%. All variants with FDR less than 10% were visually confirmed in sequencing data with BasePlayer (23).

We studied the enrichment of variants across different pathogenicity classes by comparing the minor allele frequency (MAF) in the study set to the MAF in gnomAD non-cancer Finns or gnomAD non-cancer total, selecting the higher MAF of these two as the frequency in controls. Only variants with more than zero alleles in both cases and control were included to alleviate potential detection bias. Fisher’s exact test (two-sided) was used to determine how likely it is that the variant enriched in cases (over twofold enrichment in cases compared to control) is P/LP.

Linear regression analysis

We employed ordinary linear regression models, with intercept, using Python package statsmodels to predict 1) AlphaMissense variant pathogenicity score and 2) variant enrichment in the patient set (log₂(MAF(case) / MAF(control)) with ACMG rules as predictors.

In silico protein stability prediction

We used DynaMut and mCSM (24) to assess the impact of missense mutations on protein stability. Protein stability alterations were determined as differences in free energy (ddG). DdG ≥ 0 was defined as stabilizing and ddG < 0 as destabilizing. In addition, we used AlphaMissense (25) to assess whether pathogenicity predictions support the protein stability analysis results. Protein structures and mutations were visualized with PyMol (The PyMOL Molecular Graphics System, Version 2.5.0 Schrödinger, LLC).

In silico splicing prediction

We used SpliceAI (26) to predict the probability that CUX2 splice variant c.1259-2A>G disrupts splicing. SpliceAI outputs delta-scores (values between 0 and 1) for the probability. Cutoff 0.8 is considered high precision.

Loss of heterozygosity and biallelic events

We estimated loss-of-heterozygosity (LOH) at variant loci by testing whether the variant allelic fraction (VAF) was significantly higher in tumor than in normal using Fisher’s exact test for the ratio VAF^tumor / VAF^germline for each variant. The resulting p-values were adjusted with the Benjamini-Hochberg procedure, and events with FDR<1% were considered LOH.

For biallelic events, we examined genes carrying both a P/LP germline variant and either a missense or a truncating somatic mutation.

Post-translational modification analysis

We mapped germline missense variants to post-translational modification sites (PTM) using ActiveDriverDB (27). PTM site substitutions were divided into four categories: direct substitutions replaced the amino acid affected by post-translational modification, proximal substitutions replaced amino acids within ±2 and distal substitutions replaced amino acids within ±7 amino acids of the PTM site, and network-rewiring substitutions replaced amino acids near PTM site in kinase binding motif.

Results

Evaluation of germline variant pathogenicity in 511 patients with a hematological disease

In order to assess the pathogenicity of a large number of germline variants in whole exome data, we developed PaVaDi, an automated pipeline integrating evidence available in databases such as ClinVar, and allele frequencies (AF) in population databases to automatically assign each variant with one of five pathogenicity classes (B, LB, VUS, LP, P) based on 17 of the rules described in the ACMG guidelines (3) (Methods; Supplementary Table 3; Supplementary Figure S2). PaVaDi is freely available at https://github.com/jezkoski/PaVaDi.

We discovered 156,245 germline variants (Fig. 2 a-c) in our study set of 511 patients with hematological disease diagnosed in Finland (Methods). We assigned each variant a pathogenicity class (“ACMG verdict”) with PaVaDi, resulting in 54,381 benign (B, 34.8%), 54,920 likely benign (LB, 35.1%), 45,289 (VUS, 29.0%), 1,224 likely pathogenic (LP, 0.8%) and 431 pathogenic variants (P, 0.3%) (Fig. 2 a). As expected, B/LB variants showed higher allele frequencies than P/LP variants in both cases and controls (Fig. 2 b-c), and variants rare in population showing higher frequencies in cases (Fig. 2 d). The latter skew was partly (34%) attributable to case-control imbalance (Supplementary Text, Supplementary Figure S1), with the remainder likely stemming from differences in variant analysis workflows and presence of true pathogenic variants. Moreover, the variants which were more than two-fold enriched in our study set were more frequently P/LP/VUS variants than in the Finnish population (5.4-fold, p<2.2×10⁻¹⁶, 95% CI 5.1–5.8; Fig. 2 f; Supplementary Text). Multiple ACMG rules were found to correlate with enrichment (Fig. 2 h; Methods). As expected, these included rules for higher (PM2) or lower frequency in controls (BS1, BA1), but also rules denoting a known mechanism or mutation hotspots (PVS1, PM1), and evidence of (non-)pathogenicity (BP6, PP3). The P/LP/VUS variants were also enriched in the Finnish population over the global population (5.4-fold, p<2.2×10⁻¹⁶, 95% confidence interval 5.2–5.6; Fig. 2 e; Supplementary Text), compatible with previous studies (12).

Figure 2.

Germline variants (n=156,245) discovered in the hematological patient cohort. Distribution of germline variants by variant pathogenicity class assigned by PaVaDi, and a) variant type, b) MAF in cases and c) MAF in controls. d) A scatterplot of MAF in controls (X-axis) and cases (Y-axis) stratified by pathogenicity class. Solid line shows a local regression (LOESS) fit. e) Variant allele frequency enrichment stratified by pathogenicity class. Ratio of MAFs in gnomAD Finns (FinnsAF) vs gnomAD total (n=92,855). f) Ratio of MAFs in the hematological patient cohort (CaseAF) vs maximum MAF in gnomAD Finns and gnomAD total (n=101,225) stratified by pathogenicity class. g) AlphaMissense pathogenicity predictions stratified by pathogenicity class. Light blue line denotes the threshold for benign and pink for pathogenic prediction. h) Coefficients from a linear regression model explaining variant enrichment in cases with ACMG rules. Pathogenic and benign rules denoted by red and blue colors, respectively. i) Coefficients from a linear regression model of P/LP variants explaining AlphaMissense score with ACMG rules. j) Volcano plot of exome-wide variant burden analysis showing variant enrichment (log₂(OR); X-axis) and significance (−-log₁₀(p); Y-axis) stratified by pathogenicity class. Genes with significant P/LP variant burden at (FDR<11%) are labeled.

We then investigated the correlation of our pathogenicity classification with protein folding based predictions of functional effects for missense variants by AlphaMissense (25) in our study set. For P and LP variants, we observed a bimodal distribution of AlphaMissense scores, with 70% of P and 48% of LP variants being classified as pathogenic, whereas a majority of B/LB/VUS variants were classified as benign (Fig. 2 g). High AlphaMissense scores associated with pathogenic ACMG rules PP3, PS1, PM2s, PP5 and PM1, but interestingly not with PM5 denoting a novel missense at an amino acid residue where another missense has been previously determined pathogenic (Fig. 2 i; Methods).

Gene panels highlight CHEK2 in hematological malignancy predisposition

We performed an inspection of P/LP variants in genes that have previously been associated with predisposition to hematological disorders (HemMut panel, 297 genes), or DNA repair and replication (276 genes; Methods).

Examining the genes in the HemMut panel (Supplementary Table 1, Methods), we found 274 P/LP variants out of a total of 2,392 variants. In total, 115/511 (22.5%) individuals carried P/LP variants in AD, AD/AR, biallelic AR, or XL (males) genes (Fig. 3; Supplementary Table 5-7). Five individuals carried biallelic mutations in recessively inherited ERCC6L2 (n=4, 0.7%) and CTC1 (n=1, 0.2%) genes. The most frequently mutated genes were CHEK2 (n=19 patients, 3.7% of all patients), FANCM (n=14, 2.7%), CTC1 (n=13, 2.5%), TRNT1 (n=12, 2.3%), ANO6 (n=9, 1.8%), GP6 (n=9, 1.8%), and VWF (n=9, 1.8%). Seven female individuals were identified with a hemizygous variant in three X-linked genes ALAS2 (n=3), ATRX (n=1), and F8 (n=3). Two of the ALAS2 variant carriers had previously been diagnosed with congenital dyserythropoietic anemia. Moreover, we observed six P/LP variants enriched in the Finnish population (2-12 fold) to be more prevalent among patients than expected (Table 2).

Figure 3.

Individuals with the most frequent heterozygous AD and XL, and homozygous/compound heterozygous AR P/LP variants in 20 genes of the HemMut panel (Y-axis) for 98 carriers (X-axis). Homozygous AR gene variants marked with *. Columns on the right indicate the mode of inheritance and the total number of variants colored by variant type. Rows on the bottom indicate diagnosis group and sex for each patient.

View this table:

Table 2. Finnish enriched P/LP variants in the HemMut panel.

Inspecting the DDR genes (DDR panel, Methods), we found 12.3% (63/511) of patients with at least one rare P or LP heterozygous AD, AD/AR, or biallelic autosomal AR variant in 21 genes (Fig. 4; Supplementary Table 7-8; Supplementary Table 5). CHEK2 variants were the most frequent also in this panel. We then studied heterozygous variants in AR genes of the DDR panel as some of them have been shown to confer risk even at heterozygous state (12,28) (Supplementary Table 7). The most frequently mutated genes were FANCM (n=14), ERCC8 (n=10), MSH5 (n=6), MUTYH (n=6) and RNF168 (n=5). Analysis of somatic secondary events revealed four genes (FANCM, ATM, LPIN2 and DDX41) where a P/LP germline variant was coupled with a missense or truncating somatic mutation in the same gene (Supplementary Text; Supplementary Table 14).

Figure 4.

Heterozygous AD and homozygous AR P/LP variants in 14 genes of the DDR panels (Y-axis) for 57 carriers (X-axis). Only genes with two or more variants are shown. Homozygous AR gene variants marked with *. Columns on the right indicate the mode of inheritance and the total number of variants colored by variant type. Rows on the bottom indicate diagnosis group and sex for each patient.

Germline CHEK2 variants are enriched in both myeloid and lymphoid hematological malignancies

We then examined CHEK2 variants in more detail, and identified 19 P/LP variants (c.1497dup, c.1229del, c.319+2T>A) and four VUSes (Asp481Tyr, Glu282Lys). The known founder variant Ile200Thr was classified as LB in our pipeline due to the high MAF among Finns (Supplementary Table 10). However, we included it in our analyses given the evidence of a predisposing role in e.g. MPNs (29). In total, 14% (71/511) of patients were carriers of one of six unique CHEK2 variants (OR=1.78, CI 95% 1.35-2.30, p=4.5×10⁻⁵) (Fig. 5; Supplementary Table 11). The most prevalent variant Ile200Thr was enriched at the whole cohort level (n=49 carriers, 10%, OR=2.04, CI 95% 1.47–2.78, p=2.0×10⁻⁵) and, more specifically, in both ALL (19%, 11/58, OR=4.5, CI 95% 2.09–8.86, p=0.00012) and AML patients (9%, 16/180, OR=1.87, CI 95% 1.04–3.17, p=0.024). The second most frequent variant was c.1229del, which was found in 3% of patients (16/511, OR=1.81, CI 95% 1.01–3.05, p=0.039). We also identified one myeloma patient (JK202) with LOH (FDR=0.002) at this locus. One patient (JK184) carried two variants in CHEK2: a pathogenic frameshift variant c.1497dup and a VUS missense variant Glu282Lys.

Figure 5.

A forest plot showing ORs and 95% confidence intervals of P/LP (c.1497dup, c.1229del, c.319+2T>A), VUS (Asp481Tyr, Glu282Lys) and Ile200Thr CHEK2 variants in the whole patient cohort and subgroups compared to population control gnomAD Finns. Only groups with five or more variants are shown. Lymph Malign patient set includes ALL, MM and LP. Myel Malign patient set includes AML, MPN, MDS, MDS/MPN and CML.

Exome-wide germline variant burden analysis suggests novel predisposing genes CUX2, RNPC3, and MFSD2A

We next asked whether the study set harbored previously uncharacterized risk alleles missed in the panel analyses. To this end, we performed an exome-wide burden test to detect genes with more P/LP germline variants than expected in the patient study set with 503 patients of Finnish ancestry (Methods). Four genes, CUX2, RNPC3, MFSD2A and POT1, harbored more P/LP variants in the patient set compared to population controls (OR>2.7, gnomAD Finns non-cancer, FDR<10%; Fig. 2 j; Table 3; Supplementary Table 9; Supplementary Figure S3-4). The most frequent variant observed in CUX2 was splice variant c.1259-2A>G (n=6). One of the patients was homozygous for this variant. Four out of seven CUX2 variant carriers with known family history had first-degree relatives with cancer (Supplementary Table 9). SpliceAI predicted the variant to be splicing disturbing (probability 0.97; Methods).

View this table:

Table 3. P/LP variant burden test.

Two other variants observed in the gene were Arg50Gln (n=1) and Asp307Tyr (n=1). We also identified five patients with three distinct RNPC3 variants: Asp250Gly (n=3), Gln432Arg (n=1), and Glu229Val (n=1) and seven patients with three unique MFSD2A variants: Arg366Trp (n=3), Tyr444Cys (n=2), and Pro476Ala (n=1). We have previously reported the four POT1 variants in association with myeloma predisposition (30).

Novel germline variants in MPO are predicted to destabilize the protein

MPO has only recently been proposed as a predisposing gene to myeloid malignancies in heterozygous state (31)). We thus examined also VUSes in MPO aiming to identify novel potentially pathogenic variants. In our study set three patients carried a pathogenic splice variant (c.2031-2A>C; Supplementary Table 12) previously reported in MPO deficiency syndrome (32), and as a predisposing variant to myeloid malignancies (31). Furthermore, eight patients harbored heterozygous VUS variants. Four of the patients carried missense variants (Arg548Trp in two patients, Arg590Cys, Arg460Gln) previously reported as LP in (31,33) and four patients novel VUSes (Supplementary Table 12). Of these novel variants, Ile640Phe showed LOH (FDR=0.5%) while Asn716Ile is near a phosphorylation site (PTM analysis, Methods). Both of these and the other two VUSes (Thr417Ile and Trp255Arg) are located in conserved positions (phyloP100way_vertebrate>6 and >9, respectively).

In our patient set, the detected MPO variants were not limited to patients with myeloid malignancies, but occurred also in patients diagnosed with multiple myeloma, T-cell large granular lymphocytosis, and bone marrow failure (BMF) (Supplementary Table 12). All but one of the nine distinct MPO variants identified (P/LP n=3; VUS n=8) are located in the myeloperoxidase-like domain (Fig. 6 e).

Figure 6.

Protein stability analysis of MPO variants. Amino acid structures and bonds with neighboring residues of wild type Trp255 (a) and Trp255Arg (b) predicted with DynaMut (24). c) Trp255Arg (turquoise) and a catalytic residue Gln257 (pink) are located in the same alpha helix. d) Protein stability and pathogenicity predictions for MPO variants identified in our analysis (cases) as well as for population variants (gnomAD) that were classified as B/LB or P/LP in ClinVar. Negative and positive ddG predicted by DynaMut and mCSM for the missense mutation imply destabilizing and stabilizing effects, respectively, on the protein structure stability. MPO variants are grouped as follows: cases, novel=Novel variants in cases; cases, reported LP=Variants reported earlier to be LP; control B=gnomAD variants labeled as benign in ClinVar; control P=gnomAD variants labeled as pathogenic in ClinVar. AlphaMissense pathogenicity predictions are denoted by shape. Threshold (ddG<0) for destabilizing prediction for both DynaMut and mCSM in gray dotted line. e) Variants found in the patient cohort distributed across MPO protein.

We then evaluated the effect of these variants on MPO stability with DynaMut and mCSM (24). DynaMut and mCSM scores showed high correlation (r=0.86, p=9.48×10⁻⁵). mCSM predicted all variants to be destabilizing (Supplementary Table 13, Fig. 6 d). However, DynaMut predicted all variants except Trp255Arg to be stabilizing. DynaMut predicted Trp255Arg to lead to substantially fewer bonds to the nearby residues (Fig. 6 a, b), with a potential effect on the catalytic residue Gln257 in the same alpha helix (34) (Fig. 6 c). Finally, analysis with AlphaMissense (Fig. 6 d) supported the pathogenicity of the novel variants discovered here.

Discussion

Founder populations such as the Finnish population offer unique advantages to discover pathogenic germline variants (12). Here, we performed a pan-hematological analysis to discover and characterize possible hereditary risk variants in known and unknown predisposition genes. The strength of this study is that our analyses are all based on matched population controls enabling accurate risk stratification.

We developed a computational pipeline PaVaDi that allowed us to classify all 156,245 germline variants discovered in the study cohort by pathogenicity, resulting in 1% of variants being classified either pathogenic or likely pathogenic. To extract the clinically important and validated germline variant incidence in hematological patients, we first focused on variants in genes with a known role in hematological diseases or DNA damage repair. We demonstrate that one fifth of the study patients carried a P/LP germline variant. This is in line with our expectations considering the spectrum of the genes and diagnoses of study patients analyzed.

Strikingly, in both panels analyzed CHEK2 emerged as the gene with the most variants. Powered by the high frequency of CHEK2 variants (n=71) among our study patients we were able, for the first time, to demonstrate a well-quantified enrichment of these variants in a pan-hematological setting. CHEK2 variants conferred a two-fold risk for HMs as a whole which is compatible with previous findings examining myeloid cancers (29,35). In our study, the risk for myeloid malignancies was approximately two-fold. Intriguingly, CHEK2 Ile200Thr associated with 4.5-fold risk for ALL. This exceeds the risk for solid cancers previously reported for this or any CHEK2 germline variant (36,37).

As predisposition genes for ALL are much less studied and acknowledged in both scientific literature and clinical practice than myeloid malignancies, our discovery has implications to comprehensive ALL care. Our ALL study population consisted only of adult ALL patients further increasing the impact of the finding: 20-50% adult patients with ALL undergo allogeneic hematopoietic stem cell transplantation as routine care. Using sibling donors with a potential carriership of CHEK2 variants endangers patients with a donor-derived hematological malignancy of any kind. Even though penetrance of the CHEK2 Ile200Thr seems incomplete, the risk for secondary HM is evident as the disease has already once been realized. Importantly, NGS panels used in diagnostics for ALL (or myeloid malignancies) do not routinely include CHEK2. We suggest this practice to be modified in the future. In Finland, we also face additional challenges as Ile200Thr is exceptionally frequent among Finns (2.8%), yet still clinically unappreciated.

We also scrutinized the exome-wide pathogenic variant burden which highlighted three novel genes, CUX2, RNPC3, and MFSD2A, not previously associated with hematological diseases. Although CUX2 variants have been reported in breast cancer (38), its potential role in hematological malignancies remains uncharacterized. CUX1, a paralog of CUX2, is typically impaired in myeloid malignancies and acts as a tumor suppressor (39). Further validation is needed in order to confirm the role of these candidate genes in HMs.

VUSes present an important challenge in clinical practice. In our study, 29% of variants remained VUS after PaVaDi analysis, each patient carrying an average of 116 VUSes. As a proof of concept to resolve such variants, we examined VUSes detected in MPO, which has been recently associated with HM predisposition (31). Multiple computational tools supported the pathogenicity of previously unreported variants Trp255Arg and Ile640Phe. However, there is still considerable discordance between different tools for assessing VUSes in general, posing a barrier to obtaining clinically useful information. The resolution of VUSes is paramount for clinical practice because it directly impacts patient care and clinical decision-making. Inability to find causative variants leaves patients and clinicians in a state of diagnostic uncertainty, limiting actionable interventions such as targeted therapies, surveillance strategies, or familial risk assessments. Furthermore, the accumulation of unresolved VUSes in genetic databases can complicate the interpretation of future variants for meaningful clinical insights.

Our findings highlight the advantages of conducting genetic studies within founder populations such as enhanced detection of pathogenic variants. However, the relevance and impact of these discoveries extend far beyond these populations, offering valuable insights into cancer predisposition on a global scale. By identifying pathogenic variants linked to hematological malignancies, we contribute to the growing body of evidence supporting the critical role of inherited genetic risk factors in the development of these diseases. These insights not only enhance our understanding of the genetic underpinnings of hematological malignancies but also have direct implications for clinical practice, including opportunities for early detection, targeted interventions, and improved patient care.

Conclusions

We have identified pathogenic germline variants across different hematological diseases identifying novel predisposition candidate genes and variants, and underscoring the importance of CHEK2 variants particularly in ALL.

Declarations

Ethics approval and consent to participate

The study is part of a larger project that has been approved by Helsinki University Hospital ethics review committee (#206/13/03/03/2016 and #303/13/03/01/2011). All samples and data have been derived after written informed consent.

Consent for publication

Not applicable.

Availability of data and materials

All unidentifiable data is available within the article and its supporting files. The individual-level sequencing data cannot be made publicly available because the research participant consent does not include authorization to share identifiable data.

Competing interests

The authors declare that they have no competing interests.

Funding

Research Council of Finland (#322675, #349760), Sigrid Jusélius Foundation, the Finnish Special Governmental Subsidy for Health Sciences, Research, and Training, the Helsinki University Hospital Comprehensive Cancer Research Funding, Cancer Foundation Finland, and iCAN Digital Precision Cancer Medicine Flagship.

Authors’ contributions

Study design: OK, EP, UWK; data acquisition: TR, MH, AKL, CH, UWK, OK; PaVaDi pipeline design and implementation: JRK; data analysis and interpretation: JRK, LL, UWK, EP, OK; writing the manuscript JRK, LL, UWK, EP, OK.

All authors have read and approved the final manuscript.

Acknowledgements

The authors would like to thank the FIMM Genomics unit supported by HiLIFE and Biocenter Finland for genome sequencing and bioinformatics services, and acknowledge CSC – IT Center for Science, Finland, for generous computational resources.

List of abbreviations

ACMG: American College of Medical Genetics and Genomics
AML: Acute myeloid leukemia
ALL: acute lymphoid leukemia
B: Benign
BM: Bone marrow
BMF: Bone marrow failure
CDA: Congenital dyserythropoietic anemia
CML: Chronic myeloid leukemia
CVI: common variable immunodeficiency
DDR: DNA damage response and repair
FDR: False-discovery rate
HIST: Histiocytosis syndrome
HM: Hematological malignancy
IMM: Autoimmune cytopenia and/or autoimmune lymphoproliferative syndrome
HSCT: Hematopoietic stem cell transplantation
LB: Likely benign
LOH: Loss-of-heterozygosity
LP: Likely pathogenic
LP: Chronic lymphoproliferative disease
Mb Waldenström: Waldenström’s macroglobulinemia
MDS: Myelodysplastic syndrome
MM: Multiple myeloma
MPA: Mixed phenotype acute leukemia
MPN: Myeloproliferative neoplasm
OR: Odds ratio
P: Pathogenic
VUS: variant of unknown significance

References

1.↵
Chin HL, Gazzaz N, Huynh S, Handra I, Warnock L, Moller-Hansen A, et al. The Clinical Variant Analysis Tool: Analyzing the evidence supporting reported genomic variation in clinical practice. Genet Med. 2022 Jul 1;24(7):1512–22.
OpenUrl PubMed Google Scholar
2.↵
Wilcox EH, Sarmady M, Wulf B, Wright MW, Rehm HL, Biesecker LG, et al. Evaluating the impact of in silico predictors on clinical variant classification. Genet Med Off J Am Coll Med Genet. 2022 Apr;24(4):924–30.
OpenUrl Google Scholar
3.↵
Richards S, Aziz N, Bale S, Bick D, Das S, Gastier-Foster J, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Med Off J Am Coll Med Genet. 2015 May;17(5):405–24.
OpenUrl Google Scholar
4.↵
Francesca Cristofoli, Daja M, Maltese PE, Guerri G, Tanzi B, Miotto R, et al. MAGI-ACMG: Algorithm for the Classification of Variants According to ACMG and ACGS Recommendations. Genes [Internet]. 2023 Aug [cited 2024 May 28];14(8). Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10454715/
Google Scholar
5.
Munté E, Feliubadaló L, Pineda M, Tornero E, Gonzalez M, Moreno-Cabrera JM, et al. vaRHC: an R package for semi-automation of variant classification in hereditary cancer genes according to ACMG/AMP and gene-specific ClinGen guidelines. Bioinformatics. 2023 Mar 1;39(3):btad128.
OpenUrl PubMed Google Scholar
6.
Nicora G, Zucca S, Limongelli I, Bellazzi R, Magni P. A machine learning approach based on ACMG/AMP guidelines for genomic variant classification and prioritization. Sci Rep. 2022 Feb 15;12(1):2517.
OpenUrl PubMed Google Scholar
7.↵
Preston CG, Wright MW, Madhavrao R, Harrison SM, Goldstein JL, Luo X, et al. ClinGen Variant Curation Interface: a variant classification platform for the application of evidence criteria from ACMG/AMP guidelines. Genome Med. 2022 Jan 18;14(1):6.
OpenUrl CrossRef PubMed Google Scholar
8.↵
Kim YE, Ki CS, Jang MA. Challenges and Considerations in Sequence Variant Interpretation for Mendelian Disorders. Ann Lab Med. 2019 Sep;39(5):421–9.
OpenUrl CrossRef PubMed Google Scholar
9.↵
Yang F, Long N, Anekpuritanang T, Bottomly D, Savage JC, Lee TL, et al. Identification and prioritization of myeloid malignancy germline variants in a large cohort of adult AML patients. Blood. 2021 Sep 5;
Google Scholar
10.↵
Lahtinen AK, Koski J, Ritari J, Hyvärinen K, Koskela S, Partanen J, et al. Clinically relevant germline variants in allogeneic hematopoietic stem cell transplant recipients. Bone Marrow Transplant. 2023 Jan;58(1):39–45.
OpenUrl PubMed Google Scholar
11.↵
Williams LS, Williams KM, Gillis N, Bolton K, Damm F, Deuitch NT, et al. Donor-Derived Malignancy and Transplantation Morbidity: Risks of Patient and Donor Genetics in Allogeneic Hematopoietic Stem Cell Transplantation. Transplant Cell Ther. 2024 Mar 1;30(3):255–67.
OpenUrl PubMed Google Scholar
12.↵
Kurki MI, Karjalainen J, Palta P, Sipilä TP, Kristiansson K, Donner KM, et al. FinnGen provides genetic insights from a well-phenotyped isolated population. Nature. 2023 Jan;613(7944):508–18.
OpenUrl CrossRef PubMed Google Scholar
13.↵
Wartiovaara-Kautto U, Hirvonen EAM, Pitkänen E, Heckman C, Saarela J, Kettunen K, et al. Germline alterations in a consecutive series of acute myeloid leukemia. Leukemia. 2018 Oct;32(10):2282–5.
OpenUrl PubMed Google Scholar
14.↵
Pemovska T, Kontro M, Yadav B, Edgren H, Eldfors S, Szwajda A, et al. Individualized Systems Medicine Strategy to Tailor Treatments for Patients with Chemorefractory Acute Myeloid Leukemia. Cancer Discov. 2013 Dec 10;3(12):1416–29.
OpenUrl Abstract/FREE Full Text Google Scholar
15.↵
Karczewski KJ, Francioli LC, Tiao G, Cummings BB, Alföldi J, Wang Q, et al. The mutational constraint spectrum quantified from variation in 141,456 humans. Nature. 2020 May;581(7809):434–43.
OpenUrl CrossRef PubMed Google Scholar
16.↵
Garrison E, Marth G. Haplotype-based variant detection from short-read sequencing [Internet]. arXiv; 2012 [cited 2023 Sep 5]. Available from: http://arxiv.org/abs/1207.3907
Google Scholar
17.↵
Danecek P, Bonfield JK, Liddle J, Marshall J, Ohan V, Pollard MO, et al. Twelve years of SAMtools and BCFtools. GigaScience. 2021 Feb 16;10(2):giab008.
OpenUrl CrossRef PubMed Google Scholar
18.↵
Tan A, Abecasis GR, Kang HM. Unified representation of genetic variants. Bioinforma Oxf Engl. 2015 Jul 1;31(13):2202–4.
OpenUrl Google Scholar
19.↵
McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A, et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res. 2010 Sep;20(9):1297–303.
OpenUrl Abstract/FREE Full Text Google Scholar
20.↵
Knijnenburg TA, Wang L, Zimmermann MT, Chambwe N, Gao GF, Cherniack AD, et al. Genomic and Molecular Landscape of DNA Damage Repair Deficiency across The Cancer Genome Atlas. Cell Rep. 2018 Apr 3;23(1):239.
OpenUrl CrossRef PubMed Google Scholar
21.↵
Li Q, Wang K. InterVar: Clinical Interpretation of Genetic Variants by the 2015 ACMG-AMP Guidelines. Am J Hum Genet. 2017 Feb 2;100(2):267–80.
OpenUrl PubMed Google Scholar
22.↵
Huang KL, Mashl RJ, Wu Y, Ritter DI, Wang J, Oh C, et al. Pathogenic Germline Variants in 10,389 Adult Cancers. Cell. 2018 Apr 5;173(2):355–370.e14.
OpenUrl CrossRef PubMed Google Scholar
23.↵
Katainen R, Donner I, Cajuso T, Kaasinen E, Palin K, Mäkinen V, et al. Discovery of potential causative mutations in human coding and noncoding genome with the interactive software BasePlayer. Nat Protoc. 2018 Nov;13(11):2580–600.
OpenUrl PubMed Google Scholar
24.↵
Rodrigues CH, Pires DE, Ascher DB. DynaMut: predicting the impact of mutations on protein conformation, flexibility and stability. Nucleic Acids Res. 2018 Jul 2;46(W1):W350–5.
OpenUrl CrossRef PubMed Google Scholar
25.↵
Cheng J, Novati G, Pan J, Bycroft C, Žemgulytė A, Applebaum T, et al. Accurate proteome-wide missense variant effect prediction with AlphaMissense. Science. 2023 Sep 19;381(6664):eadg7492.
OpenUrl PubMed Google Scholar
26.↵
Jaganathan K, Kyriazopoulou Panagiotopoulou S, McRae JF, Darbandi SF, Knowles D, Li YI, et al. Predicting Splicing from Primary Sequence with Deep Learning. Cell. 2019 Jan 24;176(3):535–548.e24.
OpenUrl CrossRef PubMed Google Scholar
27.↵
Krassowski M, Pellegrina D, Mee MW, Fradet-Turcotte A, Bhat M, Reimand J. ActiveDriverDB: Interpreting Genetic Variation in Human and Cancer Genomes Using Post-translational Modification Sites and Signaling Networks (2021 Update). Front Cell Dev Biol [Internet]. 2021 [cited 2023 Sep 5];9. Available from: https://www.frontiersin.org/articles/10.3389/fcell.2021.626821
Google Scholar
28.↵
Heyne HO, Karjalainen J, Karczewski KJ, Lemmelä SM, Zhou W, Havulinna AS, et al. Mono- and biallelic variant effects on disease at biobank scale. Nature. 2023;613(7944):519–25.
OpenUrl CrossRef PubMed Google Scholar
29.↵
Bao EL, Nandakumar SK, Liao X, Bick AG, Karjalainen J, Tabaka M, et al. Inherited myeloproliferative neoplasm risk affects haematopoietic stem cells. Nature. 2020 Oct 29;586(7831):769–75.
OpenUrl PubMed Google Scholar
30.↵
Hakkarainen M, Koski JR, Heckman CA, Anttila P, Silvennoinen R, Lievonen J, et al. A germline exome analysis reveals harmful POT1 variants in multiple myeloma patients and families. EJHaem. 2022 Nov;3(4):1352–7.
OpenUrl PubMed Google Scholar
31.↵
Kongkiatkamon S, Terkawi L, Guan Y, Adema V, Hasipek M, Dombrovski T, et al. Rare germline alterations of myeloperoxidase predispose to myeloid neoplasms. Leukemia. 2022 Aug;36(8):2086–96.
OpenUrl PubMed Google Scholar
32.↵
Marchetti C, Patriarca P, Solero GP, Baralle FE, Romano M. Genetic characterization of myeloperoxidase deficiency in Italy. Hum Mutat. 2004 May;23(5):496–505.
OpenUrl CrossRef PubMed Web of Science Google Scholar
33.↵
Li S, Wei R, Wang J, Adema V, Kuzmanovic T, Przychodzen BP, et al. MPO as a Novel Susceptibility Gene in Myeloid Malignancies. Blood. 2019 Nov 13;134(Supplement_1):5402.
OpenUrl Google Scholar
34.↵
Grishkovskaya I, Paumann-Page M, Tscheliessnig R, Stampler J, Hofbauer S, Soudi M, et al. Structure of human promyeloperoxidase (proMPO) and the role of the propeptide in processing and maturation. J Biol Chem. 2017 May 1;292(20):8244–61.
OpenUrl Abstract/FREE Full Text Google Scholar
35.↵
Freiman L, Larcher L, Tueur G, Vasquez N, Da Costa M, Raffoux E, et al. Characteristics of Germline CHEK2 Mutated Patients in a Large Cohort of 2322 Myeloid Malignancies. Blood. 2022 Nov 15;140(Supplement 1):4082–3.
OpenUrl Google Scholar
36.↵
Bychkovsky BL, Agaoglu NB, Horton C, Zhou J, Yussuf A, Hemyari P, et al. Differences in Cancer Phenotypes Among Frequent CHEK2 Variants and Implications for Clinical Care—Checking CHEK2. JAMA Oncol. 2022 Nov;8(11):1598–606.
OpenUrl PubMed Google Scholar
37.↵
Näslund-Koch C, Nordestgaard BG, Bojesen SE. Increased Risk for Other Cancers in Addition to Breast Cancer for CHEK2*1100delC Heterozygotes Estimated From the Copenhagen General Population Study. J Clin Oncol. 2016 Apr 10;34(11):1208–16.
OpenUrl Abstract/FREE Full Text Google Scholar
38.↵
Li L, Zhu G, Tan K, Jiang L, Li Y, Zhu X, et al. CUX2/KDM5B/SOX17 Axis Affects the Occurrence and Development of Breast Cancer. Endocrinology. 2022 Sep 1;163(9):bqac110.
OpenUrl PubMed Google Scholar
39.↵
Supper E, Rudat S, Iyer V, Droop A, Wong K, Spinella JF, et al. Cut-like homeobox 1 (CUX1) tumor suppressor gene haploinsufficiency induces apoptosis evasion to sustain myeloid leukemia. Nat Commun. 2021 Apr 30;12(1):2482.
OpenUrl CrossRef PubMed Google Scholar

Posted October 23, 2024.

Download PDF

Author Declarations

Supplementary Material

Data/Code

Citation Tools

Get QR code

Tweet Widget

Subject Area

Genetic and Genomic Medicine

Reviews and Context

Comment

TRIP Peer Reviews

Community Reviews

Automated Services

Blogs/Media

Author Videos

Subject Areas

All Articles

Addiction Medicine (413)
Allergy and Immunology (727)
Anesthesia (215)
Cardiovascular Medicine (3126)
Dentistry and Oral Medicine (351)
Dermatology (265)
Emergency Medicine (464)
Endocrinology (including Diabetes Mellitus and Metabolic Disease) (1107)
Epidemiology (13075)
Forensic Medicine (14)
Gastroenterology (868)
Genetic and Genomic Medicine (4893)
Geriatric Medicine (451)
Health Economics (755)
Health Informatics (3086)
Health Policy (1109)
Health Systems and Quality Improvement (1142)
Hematology (411)
HIV/AIDS (973)
Infectious Diseases (except HIV/AIDS) (14372)
Intensive Care and Critical Care Medicine (890)
Medical Education (457)
Medical Ethics (121)
Nephrology (502)
Neurology (4664)
Nursing (248)
Nutrition (691)
Obstetrics and Gynecology (849)
Occupational and Environmental Health (769)
Oncology (2405)
Ophthalmology (683)
Orthopedics (270)
Otolaryngology (333)
Pain Medicine (311)
Palliative Medicine (88)
Pathology (519)
Pediatrics (1249)
Pharmacology and Therapeutics (526)
Primary Care Research (526)
Psychiatry and Clinical Psychology (4001)
Public and Global Health (7231)
Radiology and Imaging (1613)
Rehabilitation Medicine and Physical Therapy (963)
Respiratory Medicine (945)
Rheumatology (462)
Sexual and Reproductive Health (480)
Sports Medicine (403)
Surgery (518)
Toxicology (65)
Transplantation (222)
Urology (192)

Comments

medRxiv aims to provide a venue for anyone to comment on a medRxiv preprint. Comments are moderated for offensive or irrelevant content (this can take ~24 h). Please avoid duplicate submissions and read our Comment Policy before commenting. The content of a comment is not endorsed by medRxiv.

medRxiv aims to inform readers about online discussion of this preprint occurring elsewhere. The content at the links below is not endorsed by either medRxiv or the preprint's authors.

Community reviews for this article:

There are no community reviews for this paper.

Automated Evaluations

Certain services provide automated analysis of preprints. Analyses invited by the authors are displayed at the top of this tab. Those done independently of authors are shown underneath . None of these analyses is endorsed by medRxiv.

Automated Evaluations:

There are no automated evaluations for this paper.

[1] 1.↵
Chin HL, Gazzaz N, Huynh S, Handra I, Warnock L, Moller-Hansen A, et al. The Clinical Variant Analysis Tool: Analyzing the evidence supporting reported genomic variation in clinical practice. Genet Med. 2022 Jul 1;24(7):1512–22.
OpenUrl PubMed Google Scholar

[2] 2.↵
Wilcox EH, Sarmady M, Wulf B, Wright MW, Rehm HL, Biesecker LG, et al. Evaluating the impact of in silico predictors on clinical variant classification. Genet Med Off J Am Coll Med Genet. 2022 Apr;24(4):924–30.
OpenUrl Google Scholar

[3] 3.↵
Richards S, Aziz N, Bale S, Bick D, Das S, Gastier-Foster J, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Med Off J Am Coll Med Genet. 2015 May;17(5):405–24.
OpenUrl Google Scholar

[4] 4.↵
Francesca Cristofoli, Daja M, Maltese PE, Guerri G, Tanzi B, Miotto R, et al. MAGI-ACMG: Algorithm for the Classification of Variants According to ACMG and ACGS Recommendations. Genes [Internet]. 2023 Aug [cited 2024 May 28];14(8). Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10454715/
Google Scholar

[5] 5.
Munté E, Feliubadaló L, Pineda M, Tornero E, Gonzalez M, Moreno-Cabrera JM, et al. vaRHC: an R package for semi-automation of variant classification in hereditary cancer genes according to ACMG/AMP and gene-specific ClinGen guidelines. Bioinformatics. 2023 Mar 1;39(3):btad128.
OpenUrl PubMed Google Scholar

[6] 6.
Nicora G, Zucca S, Limongelli I, Bellazzi R, Magni P. A machine learning approach based on ACMG/AMP guidelines for genomic variant classification and prioritization. Sci Rep. 2022 Feb 15;12(1):2517.
OpenUrl PubMed Google Scholar

[7] 7.↵
Preston CG, Wright MW, Madhavrao R, Harrison SM, Goldstein JL, Luo X, et al. ClinGen Variant Curation Interface: a variant classification platform for the application of evidence criteria from ACMG/AMP guidelines. Genome Med. 2022 Jan 18;14(1):6.
OpenUrl CrossRef PubMed Google Scholar

[8] 8.↵
Kim YE, Ki CS, Jang MA. Challenges and Considerations in Sequence Variant Interpretation for Mendelian Disorders. Ann Lab Med. 2019 Sep;39(5):421–9.
OpenUrl CrossRef PubMed Google Scholar

[9] 9.↵
Yang F, Long N, Anekpuritanang T, Bottomly D, Savage JC, Lee TL, et al. Identification and prioritization of myeloid malignancy germline variants in a large cohort of adult AML patients. Blood. 2021 Sep 5;
Google Scholar

[10] 10.↵
Lahtinen AK, Koski J, Ritari J, Hyvärinen K, Koskela S, Partanen J, et al. Clinically relevant germline variants in allogeneic hematopoietic stem cell transplant recipients. Bone Marrow Transplant. 2023 Jan;58(1):39–45.
OpenUrl PubMed Google Scholar

[11] 11.↵
Williams LS, Williams KM, Gillis N, Bolton K, Damm F, Deuitch NT, et al. Donor-Derived Malignancy and Transplantation Morbidity: Risks of Patient and Donor Genetics in Allogeneic Hematopoietic Stem Cell Transplantation. Transplant Cell Ther. 2024 Mar 1;30(3):255–67.
OpenUrl PubMed Google Scholar

[12] 12.↵
Kurki MI, Karjalainen J, Palta P, Sipilä TP, Kristiansson K, Donner KM, et al. FinnGen provides genetic insights from a well-phenotyped isolated population. Nature. 2023 Jan;613(7944):508–18.
OpenUrl CrossRef PubMed Google Scholar

[13] 13.↵
Wartiovaara-Kautto U, Hirvonen EAM, Pitkänen E, Heckman C, Saarela J, Kettunen K, et al. Germline alterations in a consecutive series of acute myeloid leukemia. Leukemia. 2018 Oct;32(10):2282–5.
OpenUrl PubMed Google Scholar

[14] 14.↵
Pemovska T, Kontro M, Yadav B, Edgren H, Eldfors S, Szwajda A, et al. Individualized Systems Medicine Strategy to Tailor Treatments for Patients with Chemorefractory Acute Myeloid Leukemia. Cancer Discov. 2013 Dec 10;3(12):1416–29.
OpenUrl Abstract/FREE Full Text Google Scholar

[15] 15.↵
Karczewski KJ, Francioli LC, Tiao G, Cummings BB, Alföldi J, Wang Q, et al. The mutational constraint spectrum quantified from variation in 141,456 humans. Nature. 2020 May;581(7809):434–43.
OpenUrl CrossRef PubMed Google Scholar

[16] 16.↵
Garrison E, Marth G. Haplotype-based variant detection from short-read sequencing [Internet]. arXiv; 2012 [cited 2023 Sep 5]. Available from: http://arxiv.org/abs/1207.3907
Google Scholar

[17] 17.↵
Danecek P, Bonfield JK, Liddle J, Marshall J, Ohan V, Pollard MO, et al. Twelve years of SAMtools and BCFtools. GigaScience. 2021 Feb 16;10(2):giab008.
OpenUrl CrossRef PubMed Google Scholar

[18] 18.↵
Tan A, Abecasis GR, Kang HM. Unified representation of genetic variants. Bioinforma Oxf Engl. 2015 Jul 1;31(13):2202–4.
OpenUrl Google Scholar

[19] 19.↵
McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A, et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res. 2010 Sep;20(9):1297–303.
OpenUrl Abstract/FREE Full Text Google Scholar

[20] 20.↵
Knijnenburg TA, Wang L, Zimmermann MT, Chambwe N, Gao GF, Cherniack AD, et al. Genomic and Molecular Landscape of DNA Damage Repair Deficiency across The Cancer Genome Atlas. Cell Rep. 2018 Apr 3;23(1):239.
OpenUrl CrossRef PubMed Google Scholar

[21] 21.↵
Li Q, Wang K. InterVar: Clinical Interpretation of Genetic Variants by the 2015 ACMG-AMP Guidelines. Am J Hum Genet. 2017 Feb 2;100(2):267–80.
OpenUrl PubMed Google Scholar

[22] 22.↵
Huang KL, Mashl RJ, Wu Y, Ritter DI, Wang J, Oh C, et al. Pathogenic Germline Variants in 10,389 Adult Cancers. Cell. 2018 Apr 5;173(2):355–370.e14.
OpenUrl CrossRef PubMed Google Scholar

[23] 23.↵
Katainen R, Donner I, Cajuso T, Kaasinen E, Palin K, Mäkinen V, et al. Discovery of potential causative mutations in human coding and noncoding genome with the interactive software BasePlayer. Nat Protoc. 2018 Nov;13(11):2580–600.
OpenUrl PubMed Google Scholar

[24] 24.↵
Rodrigues CH, Pires DE, Ascher DB. DynaMut: predicting the impact of mutations on protein conformation, flexibility and stability. Nucleic Acids Res. 2018 Jul 2;46(W1):W350–5.
OpenUrl CrossRef PubMed Google Scholar

[25] 25.↵
Cheng J, Novati G, Pan J, Bycroft C, Žemgulytė A, Applebaum T, et al. Accurate proteome-wide missense variant effect prediction with AlphaMissense. Science. 2023 Sep 19;381(6664):eadg7492.
OpenUrl PubMed Google Scholar

[26] 26.↵
Jaganathan K, Kyriazopoulou Panagiotopoulou S, McRae JF, Darbandi SF, Knowles D, Li YI, et al. Predicting Splicing from Primary Sequence with Deep Learning. Cell. 2019 Jan 24;176(3):535–548.e24.
OpenUrl CrossRef PubMed Google Scholar

[27] 27.↵
Krassowski M, Pellegrina D, Mee MW, Fradet-Turcotte A, Bhat M, Reimand J. ActiveDriverDB: Interpreting Genetic Variation in Human and Cancer Genomes Using Post-translational Modification Sites and Signaling Networks (2021 Update). Front Cell Dev Biol [Internet]. 2021 [cited 2023 Sep 5];9. Available from: https://www.frontiersin.org/articles/10.3389/fcell.2021.626821
Google Scholar

[28] 28.↵
Heyne HO, Karjalainen J, Karczewski KJ, Lemmelä SM, Zhou W, Havulinna AS, et al. Mono- and biallelic variant effects on disease at biobank scale. Nature. 2023;613(7944):519–25.
OpenUrl CrossRef PubMed Google Scholar

[29] 29.↵
Bao EL, Nandakumar SK, Liao X, Bick AG, Karjalainen J, Tabaka M, et al. Inherited myeloproliferative neoplasm risk affects haematopoietic stem cells. Nature. 2020 Oct 29;586(7831):769–75.
OpenUrl PubMed Google Scholar

[30] 30.↵
Hakkarainen M, Koski JR, Heckman CA, Anttila P, Silvennoinen R, Lievonen J, et al. A germline exome analysis reveals harmful POT1 variants in multiple myeloma patients and families. EJHaem. 2022 Nov;3(4):1352–7.
OpenUrl PubMed Google Scholar

[31] 31.↵
Kongkiatkamon S, Terkawi L, Guan Y, Adema V, Hasipek M, Dombrovski T, et al. Rare germline alterations of myeloperoxidase predispose to myeloid neoplasms. Leukemia. 2022 Aug;36(8):2086–96.
OpenUrl PubMed Google Scholar

[32] 32.↵
Marchetti C, Patriarca P, Solero GP, Baralle FE, Romano M. Genetic characterization of myeloperoxidase deficiency in Italy. Hum Mutat. 2004 May;23(5):496–505.
OpenUrl CrossRef PubMed Web of Science Google Scholar

[33] 33.↵
Li S, Wei R, Wang J, Adema V, Kuzmanovic T, Przychodzen BP, et al. MPO as a Novel Susceptibility Gene in Myeloid Malignancies. Blood. 2019 Nov 13;134(Supplement_1):5402.
OpenUrl Google Scholar

[34] 34.↵
Grishkovskaya I, Paumann-Page M, Tscheliessnig R, Stampler J, Hofbauer S, Soudi M, et al. Structure of human promyeloperoxidase (proMPO) and the role of the propeptide in processing and maturation. J Biol Chem. 2017 May 1;292(20):8244–61.
OpenUrl Abstract/FREE Full Text Google Scholar

[35] 35.↵
Freiman L, Larcher L, Tueur G, Vasquez N, Da Costa M, Raffoux E, et al. Characteristics of Germline CHEK2 Mutated Patients in a Large Cohort of 2322 Myeloid Malignancies. Blood. 2022 Nov 15;140(Supplement 1):4082–3.
OpenUrl Google Scholar

[36] 36.↵
Bychkovsky BL, Agaoglu NB, Horton C, Zhou J, Yussuf A, Hemyari P, et al. Differences in Cancer Phenotypes Among Frequent CHEK2 Variants and Implications for Clinical Care—Checking CHEK2. JAMA Oncol. 2022 Nov;8(11):1598–606.
OpenUrl PubMed Google Scholar

[37] 37.↵
Näslund-Koch C, Nordestgaard BG, Bojesen SE. Increased Risk for Other Cancers in Addition to Breast Cancer for CHEK2*1100delC Heterozygotes Estimated From the Copenhagen General Population Study. J Clin Oncol. 2016 Apr 10;34(11):1208–16.
OpenUrl Abstract/FREE Full Text Google Scholar

[38] 38.↵
Li L, Zhu G, Tan K, Jiang L, Li Y, Zhu X, et al. CUX2/KDM5B/SOX17 Axis Affects the Occurrence and Development of Breast Cancer. Endocrinology. 2022 Sep 1;163(9):bqac110.
OpenUrl PubMed Google Scholar

[39] 39.↵
Supper E, Rudat S, Iyer V, Droop A, Wong K, Spinella JF, et al. Cut-like homeobox 1 (CUX1) tumor suppressor gene haploinsufficiency induces apoptosis evasion to sustain myeloid leukemia. Nat Commun. 2021 Apr 30;12(1):2482.
OpenUrl CrossRef PubMed Google Scholar

Pathogenicity evaluation of coding germline variants identifies rare alleles enriched in hematological patients of a founder population

Abstract

Background