Gene and pathway based burden analyses in familial lymphoid cancer cases: Rare variants in immune pathway genes

Abstract

Genome-wide association studies have revealed common genetic variants with small effect sizes associated with diverse lymphoid cancers. Family studies have uncovered rare variants with high effect sizes. However, these variants explain only a portion of the heritability of these cancers. Some of the missing heritability may be attributable to rare variants with small effect sizes. We aim to identify rare germline variants associated with familial lymphoid cancers using exome sequencing. One case per family was selected from 39 lymphoid cancer families based on early onset of disease or rarity of subtype. Control data was from Non-Finnish Europeans in gnomAD exomes (N = 56,885) or ExAC (N = 33,370). Gene and pathway-based burden tests for rare variants were performed using TRAPD. Five putatively pathogenic germline variants were found in four genes: INTU, PEX7, EHHADH, and ASXL1. Pathway-based association tests identified the innate and adaptive immune systems, peroxisomal pathway and olfactory receptor pathway as associated with lymphoid cancers in familial cases. Our results suggest that rare inherited defects in the genes involved in immune system and peroxisomal pathway may predispose individuals to lymphoid cancers.

Introduction

Lymphoid cancers arise at various stages of B- and T-cell differentiation. There are more than 80 subtypes of lymphoid cancer [1] that can be grouped into three types: Hodgkin lymphoma, non-Hodgkin lymphoma, and other lymphoproliferative disorders [2]. Non Hodgkin lymphoid neoplasms are the 7th most common cancer in North America [3]. Lymphoid cancer types and subtypes represent heterogeneous group of diseases with different genetic mechanisms, clinical presentation and histological appearance and ages of onset [4, 5]. Lymphoid cancers are complex disorders caused by various genetic [6, 7] and environmental factors [8, 9].

Genetic factors for lymphoid cancers include common and rare variants [10]. Familial studies mostly detect rare variants with high effect sizes. Family-based sequencing studies have identified rare lymphoid cancer predisposition variants in several genes, including KDR [11], KLHDC8B [12], POT1 [13], ACAN [14], NPAT [15], and others [7]. In addition, significant linkage has been found for chronic lymphocytic leukemia (CLL), at 2q21.2, which contains the chemokine receptor (CXCR4) gene [16].

In contrast, genome wide association studies (GWAS) of cases and controls often identifies common variants with smaller effect sizes. Susceptibility genes that are involved in B-cell development, immune response [17], DNA repair [18] and immune function [19] have been associated with various subtypes of lymphoid cancers. The NHGRI-EBI GWAS catalog [20] contained 654 associations for lymphoid cancers as of October 3, 2022 with the most significant association being for B-cell acute lymphoblastic leukemia with a variant rs10821936-C (p-value: 1X10^-106) in ARID5B [21]. The heritability estimate for Hodgkin lymphoma was 28.4% in a Swedish population study [22]. The heritability of non-Hodgkin lymphoma varies by subtype; for instance, for CLL it is 16–34% [17, 23] and for DLBCL it is 16% [24]. Although many susceptibility loci have been identified for lymphoid cancers, a substantial fraction of the disease heritability remains unexplained. Some of this ‘missing heritability’ could be due to rare variants with small effect sizes [25–27].

Rare variant association studies (RVAS) are used to identify low allele frequency variants with small effect sizes. In RVAS, the number of cases with rare damaging variants in each gene or group of genes (e.g., pathway) is compared to that of controls to generate one statistical test result per gene or group of genes, instead of one per variant. Various statistical tests have been developed to detect rare variant associations such as the burden test, the adaptive burden test, the variance-component test, and the omnibus test, all of which have been reviewed previously [28, 29]. Study designs and considerations for performing RVAS have also been described [29–31]. RVAS has been applied in lymphoid cancer cases with a mature B-cell non-Hodgkin lymphoma subtype, and identified a burden of rare germline variants in CHMP6 and GSTA4 [32]. Unlike GWAS, where the genome-wide significance threshold is defined at p = 5×10⁸, there is no standard threshold for RVAS, as the number of tests performed depends on the sequencing platform used, the depth of coverage, and how variants are aggregated in the test [33].

To improve our understanding of the etiology of lymphoid cancers, we performed gene- and pathway-based RVAS to test for an aggregated burden of rare, deleterious variants in gene(s) and pathway(s), in familial lymphoid cancer cases compared to controls from public databases. Test Rare vAriants with Public Data (TRAPD) was used [34], as it allows for the use of controls from public databases such as the large exome sequence databases Genome Aggregation Database (gnomAD) [35] and ExAC [36].

Methods

Participants

The study was approved by the Research Ethics Board of BC Cancer and the University of British Columbia with Simon Fraser University as the harmonized partner board. All participants provided written informed consent.

Cases were derived from the Lymphoid Cancer Families Study (LCFS) [37]. Eligible families had two or more members, living or deceased, with a lymphoid cancer. The type and subtype of the cancers were classified according to the hierarchical classification provided by the International Lymphoma Epidemiology Consortium (InterLymph) Pathology Working Group [2]. Families (N = 39) were prioritized for exome sequencing based on the number of family members affected by lymphoid cancer and the number of germline samples available for sequencing.

Whole exome sequencing

Samples from 39 lymphoid cancer families, including 39 early onset cases and relatives, were part of a batch of 92 samples subject to whole exome sequencing, performed using an Illumina HiSeq2500 instrument in High Output mode with HCS (v2.2.68) at the Genome Sciences Centre (Vancouver, British Columbia, Canada). DNA extraction for 37 early onset cases was from blood; for two cases, saliva samples were used. Briefly, library preparation involved shearing of DNA using Covaris E-Series (Covaris, E210) to an average fragment size of 250 bp and using NEB Paired-End Sample Prep Kit (NEB, E6000B-25B; protocol as per kit version 1.1) for phosphorylation, d-A tailing, adaptor ligation, and PCR enrichment of sheared samples. Capture was performed with Agilent SureSelect v5 + UTR kit capture probes using four libraries per capture, for a total of 23 captures. The protocol was carried out according to the Agilent SureSelect Target Enrichment System for Illumina Multiplexed Sequencing. Amplification was performed using the HiSeq PE Cluster Kit v4 (Illumina, PE-401-4001) and cBot Cluster Station. Sequencing was performed using the HiSeq SBS Kit v4 (Illumina, FC-401-4003) and the Illumina HiSeq2500 System, generating 125 bp reads.

Variant calling

Reads were mapped and variants called at the Bioinformatics core facility at the Genome Sciences Centre. Paired-end reads were mapped against the human reference genome (GrCh37) using the Burrows-Wheeler Aligner (version 0.5.7) [38]. Aligned reads were filtered and sorted using Sambamba (version 0.5.52) [39] and the BAM file was generated using GATK’s (version 4.0.2.1) HaplotypeCaller [40] with the–L parameter to restrict the analysis to exome specific intervals. Variants were jointly called to produce genotypes for all variant positions in all samples, according to GATK’s best practice recommendations [41] to generate a single raw VCF file with 290,549 variants.

Variant quality filtering

The raw VCF file containing single nucleotide variants (SNV) and insertions and deletions (indels) was filtered for variant quality using Variant Quality Score Recalibration (VQSR) with truth sensitivity at 99.0% for both SNVs and indels. Variants that did not pass VQSR were removed. VQSR filtering was performed on a batch of 92 exomes for multiple projects including this study. A subset of VCF file containing only the genotype data for the early onset 39 cases was generated using bcftools (version 1.13) [42]. Workflow for the rare variant association study is summarized in Fig 1. Multiallelic sites were converted into biallelic sites with left-alignment, and normalization of variants were performed by using Bcftools [42]. Sequencing, variant calling, and filtering methods for control set GnomAD and ExAC were previously described [35, 36]. Variants that had a PASS filter status in both cases and controls were retained for RVAS.

[Figure omitted. See PDF.]

Estimation of ancestry

Input.bed files for ancestry estimation were generated using PLINK (version 1.9) [43, 44]. A principal component analysis (PCA) was conducted using KING (version 2.2.7) [45] with the 1000 Genomes Project dataset as reference. PCA plots were produced using the command—rplot with the e1071 package in R version 3.6.0 (S1 Fig). For gnomAD and ExAC controls, the ancestries were as described [35, 36].

Read depth filtering

Different exon capture methods were used for cases and controls. To reduce potential bias due to differences in coverage, the proportion of individuals covered at a read depth of > 10x was calculated separately for cases and controls. Bases with at least 90% of the samples covered at >10x in both cases and controls were retained for analysis. The coverage files for gnomAD exome and ExAC controls were obtained from the gnomAD consortium [46].

Annotation

Annotation was performed using Snpsift (version 5.0) [47] and Variant Effect Predictor (VEP) (version 103) [48] using the Genome Aggregation Database (gnomAD 2.2.1) [35], ExAC [36], and Combined Annotation Dependent Depletion (CADD) (version 1.6) [49]. European allele frequencies were annotated from gnomAD and ExAC databases. Variants not annotated for allele frequency in these databases were annotated manually with the dbSNPs Allele Frequency Aggregator (ALFA) European allele frequency.

Pre-burden test filtration

Variants in cases and controls were filtered to retain only those with CADD Phred score ≥ 20 and high impact variants such as splice acceptor, splice donor, stop-gain, frameshift, stop lost, start lost from VEP (version 103). Analyses were limited to variants with allele frequency ≤ 1% (rare variants) or ≤ 0.1% (very rare variants). Variant data from cases was filtered to retain only those genotypes where at least one case had an alternate allele.

Burden tests

Burden testing was performed to identify genes and pathways with a higher frequency of rare deleterious variants in cases compared to publicly available European controls from gnomAD (N = 56,885) and ExAC (N = 33,370) using the Test Rare vAriants with Public Data (TRAPD) method. TRAPD utilizes summary statistics for the variants from publicly available control data and allows for the application of both dominant and recessive models. TRAPD generates three files: a SNP file with qualifying rare variants mapped to genes, a count file with tabulated count of rare variants in the genes, and a burden file with p-value for each gene and pathway using a two-sided Fisher’s exact test.

Gene-based analysis included only genes that, after filtering, had one or more variants in at least one case. Therefore, for gene-based burden analysis, the same genes with rare damaging variants present in cases were selected from the controls using the TRAPD count file before carrying out the burden tests. In addition, a subset of these genes was chosen to include genes with two or more variant counts in cases before carrying out burden tests.

Pathway-based burden analysis was performed by aggregating genes into pathways based on three widely-used databases: Reactome [50], KEGG [51], and Biocarta. The molecular signatures database from the Broad institute [52] (http://www.broadinstitute.org/gsea/downloads.jsp) was used to download KEGG and Biocarta pathways. Reactome was downloaded from https://reactome.org/download-data. The Reactome database has low-level and high-level pathway hierarchies; here we used the low-level pathways. Genes with rare variants present in cases were selected from the controls using the TRAPD count file and grouped into pathways. The variant counts were summed to obtain a per-pathway rare variant count for pathway-based burden analysis.

Burden analysis was performed for the dominant model. The p-values generated by TRAPD were corrected for the inflation factor, a measure of population stratification. To measure the inflation factor, lambda 0.5 to 0.95 was calculated which fit a line to the points within 0.5 and 0.95 quantiles, as described previously [53]. Quantile-Quantile (QQ) plots were constructed using R (v3.6.0). Multiple testing corrections were performed based on the number of genes analysed at allele frequencies ≤1% and ≤0.1%. Genes or pathways were considered to be associated with lymphoid cancer if the p-value after correction for the inflation factor and multiple testing was ≤ 0.05. For the genes and pathways with significant associations, variants were visually inspected using the Integrative Genomics Viewer (version 2.4) [54] and excluded if deemed an artifact. Variants are deemed as artifacts if they show strand bias or allele frequency inconsistent with the heterozygous variant, have low base-quality of the alternate allele in the reads or low mapping quality.

Results

Exome sequencing included an early onset case from each lymphoid cancer family (N = 39). The types/subtypes of the 39 cases are shown in Table 1. Estimation of ethnicity resulted in 38 cases of European ancestry that were included in rare variant burden analyses, and one case of South Asian ancestry (S1 Fig) that was excluded from further analysis.

[Figure omitted. See PDF.]

ExAC (N = 33,370) and GnomAD exomes (N = 56,885) with Non-Finnish European ethnicity were used as controls. Read depth filtering to exclude any bases that did not have read depth 10 or more, in at least 90% of cases and 90% of a control set, resulted in slightly different sets of variants being included in the analyses with GnomAD controls vs. those with ExAC controls. After read depth filtering using gnomAD controls, 102,385 variants remained, whereas 103,427 variants remained when controls were from ExAC. After filtering to retain high impact rare deleterious variants using an allele frequency cut-off of ≤1%, there were 467 rare variants with GnomAD as controls, and 400 rare variants with ExAC as controls, whereas at an allele frequency cut-off of ≤0.1%, there were 294 rare variants with GnomAD as controls and 276 rare variants with ExAC as controls.

Gene-based burden analysis

Burden tests were performed at the gene and pathway levels to identify if there was a higher burden of rare variants in lymphoid cancer cases than in controls from GnomAD exomes and ExAC. First, we analyzed genes with one or more variant counts in cases. There were 404 and 258 genes containing deleterious variants with allele frequencies ≤1% and ≤0.1% analysed when the controls were GnomAD exomes, and 355 and 210 genes at allele frequencies ≤1% and ≤0.1% when the control set was ExAC. There were no significant associations between lymphoid cancer and any individual gene, with either control set or at either allele frequency, after correcting for inflation and for multiple testing. The inflation factors for gene-based analysis for both the controls and allele frequency are summarised in S1 Table.

We then limited analysis to genes with two or more variant counts present in any of the 39 cases. At allele frequency ≤1%, 43 and 39 genes were analysed when the controls were GnomAD exomes and ExAC, respectively. Statistical power for the cases and controls when two or more variants were analysed using genpwr [55] are summarized in S2 Table. Lymphoid cancer associated genes and variants are summarized in Table 2. INTU was the only gene associated with lymphoid cancer after multiple testing correction for 43 genes with GnomAD exomes and 39 genes with ExAC controls. For allele frequency ≤0.1%, 7 and 5 genes were analysed when the controls were GnomAD exomes and ExAC, respectively. Four genes, INTU (p-value 0.005), ASXL1 (p-value 0.009), EHHADH (p-value 0.03) and PEX7 (p-value 0.02), were associated with lymphoid cancer after multiple testing correction for 7 genes when GnomAD exomes were used as controls. Three genes, INTU (p-value 0.004), PEX7 (p-value 0.02), and EHHADH (p-value 0.02) were associated with lymphoid cancers after multiple testing correction for 6 genes with ExAC controls. Data underlying the results of burden testing are summarized in S3 Table.

[Figure omitted. See PDF.]

Examination of variants of interest in family members.

For cases with rare variants in genes associated with lymphoid cancer, we looked at whether any affected relatives also exome sequenced carried the variant or not. INTU has two stop-gain variants of interest: NM_015693.4:c.466C>T (rs1347837845) and NM_015693.4:c.581G>A (rs1450337930), these novel variants were not reported in Clinvar. NM_015693.4:c.466C>T was found in a case from family 129 with Burkitt lymphoma at the age of 9. This participant had one family member (a sister) diagnosed with nodular sclerosis, not otherwise specified, at age 34; the sister does not carry the NM_015693.4:c.466C>T variant. Given that the sibling does not share the NM_015693.4:c.466C>T variant in INTU, it is likely that additional genetic factors might contribute to familial lymphoid cancer in this family. NM_015693.4:c.581G>A was observed in an individual with CLL at the age of 62 from family 100; his/her mother, who was diagnosed with CLL at the age of 71, also carried this variant. As both the parent and offspring with the same lymphoma subtype carry the NM_015693.4:c.581G>A variant, it is plausible that this variant could be involved in susceptibility to lymphoid cancer in this family.

A stop-gain variant NM_000288.4:c.875T>A (rs1805137) in PEX7, which is indicated in Clinvar as pathogenic, was observed in two cases from different families; one case had nodular sclerosis at age 30, while the other had Burkitt lymphoma at age 9. The case with nodular sclerosis had two additional family members affected with lymphoid cancer; a father and a maternal cousin, as well as a mother without lymphoid cancer but with basal cell carcinoma (BCC), who were exome sequenced. The father developed lymphoplasmacytic lymphoma at 61 years, and the maternal cousin developed nodal marginal zone lymphoid cancer at 28 years; they did not carry the NM_000288.4:c.875T>A variant. The mother, who developed BCC at 45, carried the variant. Given that the mother who is not affected with lymphoid cancer had the variant in PEX7, but the affected father and the cousin did not, the NM_000288.4:c.875T>A variant seems unlikely to contribute to lymphoid cancer in this family. The case with Burkitt lymphoma had one additional exome sequenced family member, a sister diagnosed with nodular sclerosis at the age of 34. She carries the NM_000288.4:c.875T>A variant. The case who developed Burkitt lymphoma had variation in both the PEX7 and INTU and developed cancer at the age of 9. The shared NM_000288.4:c.875T>A variant amongst the siblings implies that the gene might contribute to the susceptibility of lymphoid cancer in this family.

A frameshift variant, NM_015338.6:c.1934dup (rs750318549) in ASXL1, was observed in two cases, one with lymphoplasmacytic lymphoma at the age of 73 and one with CLL at the age of 50. This variant is also reported in Clinvar as pathogenic/likely pathogenic. The case with lymphoplasmacytic lymphoma had a sister sequenced who developed CLL at the age of 78 but did not carry the NM_015338.6:c.1934dup variant. For the case with CLL, no additional family members were sequenced.

A splice acceptor deletion NM_001966.4:c.594_595del (rs781090244) in EHHADH was observed in two unrelated familial lymphoid cancer cases, one diagnosed with CLL at the age of 34 and one with DLBCL at the age of 34. This variant has unknown significance in Clinvar. For the case with DLBCL, a family member (mother) was exome sequenced who developed CLL at 72 and did not carry the NM_001966.4:c.594_595del variant. The other case with CLL also had a family member (sister) who was exome sequenced. She developed Follicular Lymphoma at 44 and did not carry the NM_001966.4:c.594_595del variant. Since affected family members of both cases did not have the same variant in ASXL1 or EHHADH, it is plausible that other factors contribute to familial lymphoid cancer in these families.

Pathway-based burden analysis

We obtained 1761, 186, and 292 pathways from the Reactome, KEGG, and Biocarta databases, respectively. Gene and variant counts generated in the gene-based burden analysis were collapsed into pathways. The variant allele frequency thresholds of ≤1% and ≤0.1% were applied while aggregating the variant counts and genes to pathways. P-values were corrected for genomic inflation when necessary. The inflation factors for pathway-based analysis for both the controls and allele frequency are summarised in S1 Table. For variants with an allele frequency of ≤1% with GnomAD exomes as controls, a total of 483 pathways were identified: 333 Reactome, 106 KEGG, and 44 Biocarta pathways. Twelve pathways (10 Reactome and 2 KEGG) showed significant association after corrections (Table 3). With ExAC controls, 308 Reactome, 94 KEGG, and 34 Biocarta pathways were detected, for a total of 436 pathways. Two KEGG pathways were significantly associated with lymphoid cancer after corrections (Table 3).

[Figure omitted. See PDF.]

To test for associations between lymphoid cancers and rare variants with allele frequency ≤0.1%, 346 pathways were analyzed with GnomAD exomes as controls. These included 231 Reactome, 90 KEGG, and 25 Biocarta pathways. Eighteen pathways (12 Reactome and 6 KEGG) showed a significant association with lymphoid cancer (Table 3). There were 284 pathways, with 197 from the Reactome database, 71 from KEGG, and 16 from Biocarta, when the controls were from ExAC. Five pathways showed significant association with lymphoid cancer, of which four were from Reactome and three were from KEGG (Table 3).

To identify which genes in the lymphoid cancer associated pathways had rare variants, we summarized the count of rare variants aggregated by genes in cases and controls (Table 4). The 18 pathways (6 from KEGG and 12 from Reactome) associated with lymphoid cancers included 34 genes that had rare variant counts. One of the pathways associated with lymphoid cancer is the antigen activated B cell receptor, which was significant with a corrected p-value for multiple testing and inflation factor of 0.0008 and 0.0007 for allele frequency ≤0.1% and ≤1% respectively, with gnomAD controls. This pathway was then collapsed into high-level Reactome pathway hierarchies for the adaptive immune system. The remaining 11 pathways from Reactome were grouped into three high-level pathways. One of these high-level pathways is for the innate immune systems comprising seven low-level hierarchy pathways. Five genes with rare variants, SYK, IGLV3-25, IGLV7-43, BLK, and GRAP2, are included in these adaptive and innate immune system pathways. The five variants, one in each gene, are present in 5 different lymphoid cancer cases. The pathways of olfactory transduction and peroxisome from KEGG were also associated with the lymphoid cancer across both the allele frequencies and control sets. Data underlying these findings are summarized in S4 Table.

[Figure omitted. See PDF.]

Discussion

We performed gene- and pathway-based RVAS using lymphoid cancer cases with a strong family history and early cancer onset. Choosing cases with a strong family history and early onset is expected to enrich for cases with a genetic component to their disease. These families often have members with different lymphoid cancer types and subtypes. We suspect that some genes could predispose to more than one type of lymphoid cancer in such families. Previous rare variant association studies have analysed only non-Hodgkin lymphoid cancers collectively [32]. The genes EHHADH, INTU, PEX7, and ASXL1 were found to have a significantly higher burden of deleterious variants in cases than controls; variants in these genes may contribute to risk of lymphoid cancers in specific individuals. However, the variants of interest segregated with lymphoid cancer in other affected members of the same family, in only two families, a variant in INTU in family 100 and a variant in PEX7 in Family 129.

PEX7 encodes the peroxin 7 protein required for the peroxisomal importation of proteins. To our knowledge, the gene has not been previously implicated in lymphoid neoplasms, but peroxisomal proteins may have a role in the regulation of innate and adaptive immune cells by controlling the synthesis and breakdown of immune bioactive metabolites such as reactive oxygen species and reactive nitrogen species [56, 57]. Drosophila Schneider 2 (S2) macrophage-like cells with PEX7 and PEX5 knocked down by RNA interference fail to perform phagocytosis, suggesting that compromised peroxisome function within macrophages may possibly reduce the immune system’s ability to eliminate cancer cells [58].

INTU is involved in Hedgehog signalling, a pathway activated in various cancers, including basal cell carcinoma, medulloblastoma and hematological malignancies [59, 60].

Variants in EHHADH and AXSL1 did not segregate with lymphoid cancer in the families studied, which implies that the variants do not play a role in all affected members of these families, it does not rule out their making a contribution to cancer predisposition in the original case. EHHADH encodes a peroxisome enzyme that oxidises medium-chain and large-chain fatty dicarboxylic acids and has delta 3, delta 2-enoyl-CoA isomerase activity [61]. It has been suggested that EHHADH affects T-cell function in tumor infiltration, affecting immune surveillance in cancer [62]. The EHHADH variant NM_001966.4:c.594_595del is reported in Clinvar as a variant of unknown significance. ASXL1 plays a role in chromatin remodelling and is a tumor suppressor gene [63]. Several ASXL1 somatic mutations have been described in CLL [64, 65]. The NM_015338.6:c.1934dup (rs750318549) variant observed in ASXL1 is a known mutation found in >50% of myeloid malignancies [66]. Germline variants in ASXL1 must be interpreted with caution, particularly in older individuals, as variants in this gene observed in blood DNA could be of somatic in nature [67, 68]. A candidate gene-based study by Hamadou et al. (2016) identified a c.1205 G>A germline missense substitution in ASXL1 in two sisters, one with large B-cell lymphoma and the other with mucosa-associated lymphoid tissue lymphoma [69]. The c.1205 G>A missense variant is different than the one identified here, but the affected family members in Hamadou et al. (2016) and our study have different lymphoid cancer subtypes; this may imply a role for different ASXL1 variants in cases with different lymphoid cancer subtypes.

Pathway-based burden analysis uses pathways from KEGG, Reactome, and Biocarta. These databases are generated mainly as an independent effort by experts, with each database having specific criteria for annotation and speed of curation; there are redundant pathways amongst the databases which would make the multiple testing correction overly conservative. The Reactome pathways of innate and adaptive immune systems showed association with lymphoid cancers. Previous studies have demonstrated the involvement of variants in immune system related genes in different subtypes of lymphoid cancer [70–72]. These studies, combined with our results, support that genetic variation in immune-related genes plays a role in the development of lymphoid neoplasms. SYK, IGLV3-25, and IGLV7-43 are involved in the innate and adaptive immune pathways defined in Reactome. SYK is a spleen tyrosine kinase activated by binding to the phosphorylated Immunoreceptor tyrosine- based activation motif, triggering a downstream signalling cascade regulated by its tyrosine phosphorylation [73]. This gene plays a role in cell adhesion, innate pathogen recognition, platelet activation, and vascular development [74]. SKY amplification has been identified as a genetic alteration in primary BCR-type DLBCL tumours [75]. IGLV3-25 and IGLV7-43 are immunoglobin light chain variable regions involved in antigen recognition. Cells that express a point mutation that changes amino acid position 110 from glycine to arginine in IGLV3-21 trigger autonomous B-cell receptor (BCR) signalling by BCR-BCR interaction and confer aggressive behaviour in CLL [76]. The pattern of somatic mutation in sporadic CLL for IGLV3-21*04 and R110 is similar to that of familial cases of CLL [77]. BLK, a B-lymphoid tyrosine kinase involved in the adaptive pathway, is a proto-oncogene and a member of the SRC family kinases involved in B-cell development [78]. This gene has been identified in a family of two members affected with Hodgkin lymphoma [79]. The case with the variant in the proto-oncogene BLK also had a variation in AXSL1, a tumour suppressor gene. GRAP2 is a tumour suppressor gene involved in T cell signalling, development and regulate NF-AT activation [80]. GRAP2 has been implicated in extranodal NK/T cell lymphoma, a rare and aggressive subtype of T cell or NK lineage extranodal lymphoma [81]. GARP2 is a specific marker of T lineage differentiation; it was interesting to observe a variant in this gene in a case affected with CLL, suggesting a role of GRAP2 in B cell lymphoma as well. These genes would not have been detected if pathway-based burden analysis had not been performed.

The KEGG pathway that was associated, for both ≤1% and ≤0.1% allele frequencies and both control groups, was olfactory transduction; this pathway had the highest number of genes, 12, with rare variants. Although there is no previously known involvement of olfactory genes in lymphoid cancers, these genes have been studied in other neoplasms such as prostate cancer; for instance the activation of OR51E2 facilitates the cellular transformation of the disease into a more aggressive form [82]. Olfactory gene OR2B6 is expressed in over 80% of breast carcinoma tissues and its expression has been suggested as a potential biomarker in breast carcinoma [83]. It is also not implausible that this pathway could influence the sense of smell of individuals and contribute to avoidance of exposure to some carcinogens, thus affecting cancer risk. Olfactory receptor genes are a large gene family with over 900 genes and pseudogenes [84], which can complicate the mapping of sequenced reads and lead to erroneous genotypes and potentially spurious association. Though all variants of interest in this study were individually verified by examining them in IGV, this may not be sufficient to eliminate this type of error.

This study has three strengths: first, it uses cases with early onset and a strong family history of lymphoid cancers; it thus enriches for cases that are likely to have a stronger genetic component to their disease. Second, it enables investigation of variants of interest in other affected family members. Third, we assess rare variant associations at both the gene- and pathway-level, improving the possibility of making meaningful inferences regarding the etiology of lymphoid cancers, which might not have been possible by analysing individual genes.

This work also has several limitations. The use of public control datasets can give rise to issues due to differences in population structure and sequencing methods, including differences in depth of sequencing and in data processing. Some of these potential issues were limited; however, by restricting analyses to variants with ≥10x coverage in both controls and cases, as suggested by Guo et al. [34].

A second limitation is calculation of the inflation factor lambda. The inflation factor lambda reflects the inflation in the test statistic due to ancestry; not correcting for this factor can lead to false associations. TRAPD calculates the observed and expected -log₁₀(p values) of lambda at the 95^th percentile with the highest expected -log₁₀ (p value) for genes with p-value 1.0; however, in our study we only included genes with rare variants present in the cases, making it unfeasible to use the TRAPD inflation factor calculation. Instead, the inflation factor lambda_0.5–0.95 was calculated as proposed by Chen et al., 2021 for burden analysis using public datasets as controls; this method excludes inflated value at the extremes of the plot [53]. The inflation factor in some instances was >1, indicating potential confounding by genetic background; this was offset by correcting for the inflation factor by multiplying it with the p-value. Yang et al. (2011) showed substantial inflation in the context of polygenic inheritance, even when data was controlled for population structure and technical artifacts [85]. The inflation >1 observed in lymphoid neoplasms might also be due to polygenic inheritance.

The success of rare variant association analyses depends on various factors, two of which are statistical power and a suitable method for the association study. RVAS suffers from low statistical power due to the rarity of the variants being tested. Attempts to improve statistical power in RVAS are made by increasing the sample sizes or collapsing rare variants into gene annotation categories and biological functions. In this RVAS of lymphoid cancers, we considered genes with two or more variants, as done by others [86, 87], and also carried out pathway-based analyses. These analyses reduce the number of tests, thereby preserving power. For disease-causing genes in lymphoid cancers, we expect that a majority of rare causal variants will increase disease risk. Given that burden analysis assumes that all the variants influence the phenotype in the same direction [29], the selection of this test seems appropriate for our analyses.

Conclusion

RVAS was performed using lymphoid cancer cases with a strong family history and early disease onset. Cases with early–onset and family history are more likely to have a genetic basis than later onset cases. Publicly available data sets with large sample sizes were used as controls to maximize statistical power of burden analysis. Nevertheless, the case sample size was small and therefore statistical power was limited; large studies are necessary to substantiate these findings. We identified putative pathogenic germline variants in four genes, and biological pathways related to immune systems, as possible contributors to lymphoid cancer risk. Stop-gain variant NM_015693.4:c.581G>A in INTU in the Hedgehog signalling pathway and stop-gain variant NM_000288.4:c.875T>A in PEX7 in peroxisomal pathway each segregated with lymphoid cancers in families, suggesting that these variants in these genes may be related to risk of lymphoid cancer.

Supporting information

S1 Fig. Principal component analysis of 39 lymphoid cancer cases performed using KING (version 2.2.7) with the 1000 Genome dataset as reference.

Thirty-eight cases with European ethnicity were selected for the rare variant association study while one case of non-European ancestry was removed.

https://doi.org/10.1371/journal.pone.0287602.s001

(TIFF)

S1 Table. The inflation factor (lambda_0.5–0.95) for the association tests of lymphoid cancer cases vs. GnomAD exome and ExAC controls, at allele frequency ≤0.1% and ≤1%.

https://doi.org/10.1371/journal.pone.0287602.s002

(XLSX)

S2 Table. Statistical power calculation for more than 2 variants analysis with lymphoid cancer cases and GnomAD or ExAC controls, for allele frequencies ≤1% and ≤0.1%.

https://doi.org/10.1371/journal.pone.0287602.s003

(XLSX)

S3 Table. Genes with two or more variants and count of variants in lymphoid cancer cases and GnomAD or ExAC controls, for allele frequencies ≤1% and ≤0.1%.

https://doi.org/10.1371/journal.pone.0287602.s004

(XLSX)

S4 Table. Pathways and count of variants in lymphoid cancer cases and GnomAD or ExAC controls, for allele frequencies ≤1% and ≤0.1%.

https://doi.org/10.1371/journal.pone.0287602.s005

(XLSX)

Acknowledgments

We thank the families and their family members for participating in the Lymphoid Cancer Families study. We thank the Canada’s Michael Smith Genome Sciences Centre (GSC) at BC cancer Sequencing and Bioinformatics platforms for the expert generation of high-quality data. We would also like to thank Martin Krzywinski, Genome Sciences Centre, for his help with the workflow figure design.

Citation: Ralli S, Jones SJ, Leach S, Lynch HT, Brooks-Wilson AR (2023) Gene and pathway based burden analyses in familial lymphoid cancer cases: Rare variants in immune pathway genes. PLoS ONE 18(6): e0287602. https://doi.org/10.1371/journal.pone.0287602

About the Authors:

Sneha Ralli

Roles: Formal analysis, Investigation, Methodology, Visualization, Writing – original draft

Affiliations: Department of Biomedical Physiology and Kinesiology, Simon Fraser University, Burnaby, British Columbia, Canada, Canada’s Michael Smith Genome Sciences Centre, BC Cancer, Vancouver, British Columbia, Canada

Samantha J. Jones

Roles: Data curation, Resources, Writing – review & editing

Affiliation: Department of Biomedical Physiology and Kinesiology, Simon Fraser University, Burnaby, British Columbia, Canada

Stephen Leach

Roles: Data curation, Resources, Writing – review & editing

Affiliation: Department of Biomedical Physiology and Kinesiology, Simon Fraser University, Burnaby, British Columbia, Canada

Henry T. Lynch

† Deceased.

Roles: Resources

Affiliation: Hereditary Cancer Center, Creighton University, Omaha, Nebraska, United States of America

Angela R. Brooks-Wilson

Roles: Conceptualization, Methodology, Supervision, Writing – review & editing

E-mail: [email protected]

Affiliations: Department of Biomedical Physiology and Kinesiology, Simon Fraser University, Burnaby, British Columbia, Canada, Canada’s Michael Smith Genome Sciences Centre, BC Cancer, Vancouver, British Columbia, Canada

ORICD: https://orcid.org/0000-0003-1009-6408

References

1. De Leval L, Jaffe ES. Lymphoma classification. Cancer J (United States). 2020;26(3):176–85. pmid:32496451

2. Turner JJ, Morton LM, Linet MS, Clarke CA, Kadin ME, Vajdic CM, et al. InterLymph hierarchical classification of lymphoid neoplasms for epidemiologic research based on the WHO classification (2008): Update and future directions. Blood. 2010;116(20):90–8. pmid:20699439

3. Ferlay J, Ervik M LFCM, Mery L, Piñeros M, Znaor A, Soerjomataram I, et al. Global cancer observatory: cancer today. [Internet]. Lyon, France: international agency for research on cancer. 2018 [cited 2021 Dec 22]. p. 1–6. Available from: https://gco.iarc.fr/today

4. Weniger MA, Küppers R. Molecular biology of Hodgkin lymphoma. Leukemia [Internet]. 2021;35(4):968–81. Available from: pmid:33686198

5. Pratap S, Scordino TS. Molecular and cellular genetics of non-Hodgkin lymphoma: Diagnostic and prognostic implications. Exp Mol Pathol [Internet]. 2019;106(November 2018):44–51. Available from: pmid:30465756

6. Skibola CF, Curry JD, Nieters A. Genetic susceptibility to lymphoma. Haematologica. 2007;92(7):960–9. pmid:17606447

7. Szmyd B, Mlynarski W, Pastorczak A. Genetic predisposition to lymphomas: Overview of rare syndromes and inherited familial variants. Mutat Res—Rev Mutat Res [Internet]. 2021;788(November 2020):108386. Available from: pmid:34893151

8. Royer RH, Koshiol J, Giambarresi TR, Vasquez LG, Pfeiffer RM, McMaster ML. Differential characteristics of Waldenström macroglobulinemia according to patterns of familial aggregation. Blood. 2010;115(22):4464–71.

9. Jones SJ, Stroshein S, Williams AM, Liu D, Spinelli JJ, Connors JM, et al. Birth order, sibship size, childhood environment and immune-related disorders, and risk of lymphoma in lymphoid cancer families. Cancer Epidemiol Biomarkers Prev. 2020;29(6):1168–78. pmid:32169998

10. Cerhan JR, Slager SL. Familial predisposition and genetic risk factors for lymphoma. Blood. 2015;126(20):2265–73. pmid:26405224

11. Rotunno M, McMaster ML, Boland J, Bass S, Zhang X, Burdett L, et al. Whole exome sequencing in families at high risk for hodgkin lymphoma: Identification of a predisposing mutation in the KDR gene. Haematologica. 2016;101(7):853–60. pmid:27365461

12. Salipante SJ, Mealiffe ME, Wechsler J, Krem MM, Liu Y, Namkoong S, et al. Mutations in a gene encoding a midbody kelch protein in familial and sporadic classical Hodgkin lymphoma lead to binucleated cells. Proc Natl Acad Sci USA. 2009;106(35):14920–5. pmid:19706467

13. McMaster ML, Sun C, Landi MT, Savage SA, Rotunno M, Yang XR, et al. Germline mutations in Protection of Telomeres 1 in two families with Hodgkin lymphoma. Br J Haematol. 2018;181(3):372–7. pmid:29693246

14. Ristolainen H, Kilpivaara O, Kamper P, Taskinen M, Saarinen S, Leppä S, et al. Identification of homozygous deletion in ACAN and other candidate variants in familial classical Hodgkin lymphoma by exome sequencing. Br J Haematol. 2015;170(3):428–31. pmid:25715982

15. Saarinen S, Aavikko M, Aittomäki K, Launonen V, Lehtonen R, Franssila K, et al. Exome sequencing reveals germline NPAT mutation as a candidate risk factor for Hodgkin lymphoma. Blood. 2011;118(3):493–8. pmid:21562039

16. Sellick GS, Goldin LR, Wild RW, Slager SL, Ressenti L, Strom SS, et al. A high-density SNP genome-wide linkage search of 206 families identifies susceptibility loci for chronic lymphocytic leukemia. Blood. 2007;110(9):3326–33. pmid:17687107

17. Law PJ, Berndt SI, Speedy HE, Camp NJ, Sava GP, Skibola CF, et al. Genome-wide association analysis implicates dysregulation of immunity genes in chronic lymphocytic leukaemia. Nat Commun. 2017;8(14175). pmid:28165464

18. Rendleman J, Antipin Y, Reva B, Adaniel C, Przybylo JA, Dutra-Clarke A, et al. Genetic variation in DNA repair pathways and risk of non-Hodgkin’s lymphoma. PLoS One. 2014;9(7):1–14. pmid:25010664

19. Cozen W, Li D, Best T, Van Den Berg DJ, Gourraud PA, Cortessis VK, et al. A genome-wide meta-analysis of nodular sclerosing Hodgkin lymphoma identifies risk loci at 6p21.32. Blood. 2012;119(2):469–75. pmid:22086417

20. Buniello A, Macarthur JAL, Cerezo M, Harris LW, Hayhurst J, Malangone C, et al. The NHGRI-EBI GWAS Catalog of published genome-wide association studies, targeted arrays and summary statistics 2019. Nucleic Acids Res. 2019;47(D1):D1005–12. pmid:30445434

21. Vijayakrishnan J, Qian M, Studd JB, Yang W, Kinnersley B, Law PJ, et al. Identification of four novel associations for B-cell acute lymphoblastic leukaemia risk. Nat Commun [Internet]. 2019;10(5384). Available from: http://dx.doi.org/10.1038/s41467-019-13069-6

22. Shugart YY, Hemminki K, Vaittinen P, Kingman A, Dong C. A genetic study of Hodgkin’s lymphoma: An estimate of heritability and anticipation based on the familial cancer database in Sweden. Hum Genet. 2000;106(5):553–6. pmid:10914686

23. Berndt SI, Skibola CF, Joseph V, Camp NJ, Nieters A, Wang Z, et al. Genome-wide association study identifies multiple risk loci for chronic lymphocytic leukemia. Nat Genet. 2013;45(8):868–76. pmid:23770605

24. Cerhan JR, Berndt SI, Vijai J, Ghesquières H, McKay J, Wang SS, et al. Genome-wide association study identifies multiple susceptibility loci for diffuse large B cell lymphoma. Nat Genet. 2014;46(11):1233–8. pmid:25261932

25. McCarthy MI, Abecasis GR, Cardon LR, Goldstein DB, Little J, Ioannidis JPA, et al. Genome-wide association studies for complex traits: Consensus, uncertainty and challenges. Nat Rev Genet. 2008;9(5):356–69. pmid:18398418

26. Manolio TA, Collins FS, Cox NJ, Goldstein DB, Hindorff LA, Hunter DJ, et al. Finding the missing heritability of complex diseases. Nature [Internet]. 2009;461(7265):747–53. Available from: pmid:19812666

27. Zuk O, Hechter E, Sunyaev SR, Lander ES. The mystery of missing heritability: Genetic interactions create phantom heritability. Proc Natl Acad Sci USA. 2012;109(4):1193–8. pmid:22223662

28. Povysil G, Petrovski S, Hostyk J, Aggarwal V, Allen AS, Goldstein DB. Rare-variant collapsing analyses for complex traits: guidelines and applications. Nat Rev Genet [Internet]. 2019;20(12):747–59. Available from: pmid:31605095

29. Lee S, Abecasis GR, Boehnke M, Lin X. Rare-variant association analysis: Study designs and statistical tests. Am J Hum Genet. 2014;95(1):5–23. pmid:24995866

30. Zuk O, Schaffner SF, Samocha K, Do R, Hechter E, Kathiresan S, et al. Searching for missing heritability: Designing rare variant association studies. Proc Natl Acad Sci U S A. 2014;111(4). pmid:24443550

31. Auer PL, Lettre G. Rare variant association studies: Considerations, challenges and opportunities. Genome Med. 2015;7(1):1–11.

32. Orgueira AM, López MC, Raíndo AP, Arias JÁD, Rodríguez BA, Pérez LB, et al. Detection of rare germline variants in the genomes of patients with b-cell neoplasms. Cancers (Basel). 2021;13(6):1–18.

33. Jiang D, Wang M. Recent developments in statistical methods for GWAS and high-throughput sequencing association studies of complex traits high-throughput sequencing association studies of complex. Biostat Epidemiol [Internet]. 2018;9360(2:1):132–59. Available from: https://doi.org/10.1080/24709360.2018.1529346

34. Guo MH, Plummer L, Chan YM, Hirschhorn JN, Lippincott MF. Burden Testing of Rare Variants Identified through Exome Sequencing via Publicly Available Control Data. Am J Hum Genet. 2018;103(4):522–34. pmid:30269813

35. Karczewski KJ, Francioli LC, Tiao G, Cummings BB, Alföldi J, Wang Q, et al. Variation across 141,456 human exomes and genomes reveals the spectrum of loss-of-function intolerance across human protein-coding genes. bioRxiv [Internet]. 2019;531210. Available from: https://www.biorxiv.org/content/10.1101/531210v3

36. Lek M, Karczewski KJ, Minikel E V., Samocha KE, Banks E, Fennell T, et al. Analysis of protein-coding genetic variation in 60,706 humans. Nature. 2016;536(7616). pmid:27535533

37. Jones SJ, Voong J, Thomas R, English A, Schuetz J, Slack GW, et al. Nonrandom occurrence of lymphoid cancer types in 140 families. Leuk Lymphoma. 2017;58(9):2134–43. pmid:28278712

38. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics. 2009;25(14):1754–60. pmid:19451168

39. Tarasov A, Vilella AJ, Cuppen E, Nijman IJ, Prins P. Sambamba: fast processing of NGS alignment formats. Bioinformatics. 2015;31(12):2032–2034. pmid:25697820

40. Poplin R, Ruano-Rubio V, DePristo MA, Fennell TJ, Carneiro MO, Van der Auwera GA, et al. Scaling accurate genetic variant discovery to tens of thousands of samples. bioRxiv. 2017;201178:1–22.

41. Germline short variant discovery (SNPs + Indels)–GATK [Internet]. [cited 2021 Dec 23]. Available from: https://gatk.broadinstitute.org/hc/en-us/articles/360035535932-Germline-short-variant-discovery-SNPs-Indels-

42. Danecek P, Bonfield JK, Liddle J, Marshall J, Ohan V, Pollard MO, et al. Twelve years of SAMtools and BCFtools. Gigascience. 2021;10(2):1–4. pmid:33590861

43. Chang CC, Chow CC, Tellier LCAM, Vattikuti S, Purcell SM, Lee JJ. Second-generation PLINK: Rising to the challenge of larger and richer datasets. Gigascience. 2015;4(1):1–16.

44. Chang Christopher, Purcell Shaun. PLINK 1.90 [Internet]. [cited 2021 Dec 22]. Available from: https://www.cog-genomics.org/plink/1.9/

45. Manichaikul A, Mychaleckyj JC, Rich SS, Daly K, Sale M, Chen W min. Robust relationship inference in genome-wide association studies. Bioinformatics. 2010;26(22):2867–73.

46. 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;581(7809):434–43. pmid:32461654

47. Cingolani P, Patel VM, Coon M, Nguyen T, Land SJ, Ruden DM. Using Drosophila melanogaster as a model for genotoxic chemical mutational studies with a new program, SnpSift. Front Genet. 2012;3(March):1–9.

48. McLaren W, Gil L, Hunt SE, Riat HS, Ritchie GRS, Thormann A, et al. The Ensembl Variant Effect Predictor. Genome Biol [Internet]. 2016;17(122):1–14. Available from: pmid:27268795

49. Rentzsch P, Witten D, Cooper GM, Shendure J, Kircher M. CADD: Predicting the deleteriousness of variants throughout the human genome. Nucleic Acids Res. 2019;47(D1):D886–94. pmid:30371827

50. Croft D, Mundo AF, Haw R, Milacic M, Weiser J, Wu G, et al. The Reactome pathway knowledgebase. Nucleic Acids Res. 2014;42:D472–D477. pmid:24243840

51. Kanehisa M, Goto S. KEGG: Kyoto Encyclopedia of Genes and Genomes. Nucleic Acids Res. 2000;28(1):27–30. pmid:10592173

52. Liberzon A, Subramanian A, Pinchback R, Thorvaldsdóttir H, Tamayo P, Mesirov JP. Molecular signatures database (MSigDB) 3.0. Bioinformatics. 2011;27(12):1739–40. pmid:21546393

53. Chen W, Wang S, Tithi SS, Ellison D, Wu G. CoCoRV: a rare variant analysis framework using publicly available genotype summary counts to prioritize germline disease-predisposition genes. bioRxiv [Internet]. 2021; Available from: https://doi.org/10.1101/2021.09.29.462472

54. Thorvaldsdóttir H, Robinson JT, Mesirov JP. Integrative Genomics Viewer (IGV): High-performance genomics data visualization and exploration. Brief Bioinform. 2013;14(2):178–92. pmid:22517427

55. Moore CM, Jacobson SA, Fingerlin TE. Power and Sample Size Calculations for Genetic Association Studies in the Presence of Genetic Model Misspecification. Hum Hered. 2019;84(6):256–71. pmid:32721961

56. Di Cara F, Andreoletti P, Trompier D, Vejux A, Bülow MH, Sellin J, et al. Peroxisomes in immune response and inflammation. Int J Mol Sci. 2019;20(16):1–29. pmid:31398943

57. Vijayan V, Srinu T, Karnati S, Garikapati V, Linke M, Kamalyan L, et al. A new immunomodulatory role for peroxisomes in macrophages activated by the TLR4 ligand lipopolysaccharide. J Immunol. 2017;198(6):2414–25. pmid:28179495

58. Di Cara F, Sheshachalam A, Braverman NE, Rachubinski RA, Simmonds AJ. Peroxisome-mediated metabolism is required for immune Response to microbial infection. Immunity [Internet]. 2017;47(1):93–106.e7. Available from: pmid:28723556

59. Yang N, Leung E, Lui C, Eguether T, Jun Yao X, Jones E, et al. INTU is essential for oncogenic Hh signaling through regulating primary cilia formation in basal cell carcinoma. Oncogene. 2017;36(35):4997–5005. pmid:28459465

60. Jiang J, Hui C chung. Hedgehog signaling in development and cancer. Dev Cell [Internet]. 2008;15(6):801–12. Available from: http://dx.doi.org/10.1016/j.devcel.2008.11.010

61. Houten SM, Denis S, Argmann CA, Jia Y, Ferdinandusse S, Reddy JK, et al. Peroxisomal L-bifunctional enzyme (Ehhadh) is essential for the production of medium-chain dicarboxylic acids. J Lipid Res. 2012;53(7):1296–303. pmid:22534643

62. Rogers LM, Wang Z, Mott SL, Dupuy AJ, Weiner GJ. A genetic screen to identify gain- And loss-of-function modifications that enhance T-cell infiltration into tumors. Cancer Immunol Res. 2020;8(9):1206–14. pmid:32611665

63. Abdel-Wahab O, Adli M, Saunders L, Gao J, Shih AH, Pandey S, et al. ASXL1 Mutations Promote Myeloid Transformation Through Inhibition of PRC2-Mediated Gene Repression. Cancer Cell. 2012;22(2):180–93.

64. Katoh M. Functional and cancer genomics of ASXL family members. Br J Cancer [Internet]. 2013;109(2):299–306. Available from: pmid:23736028

65. Quesada V, Conde L, Villamor N, Ordóñez GR, Jares P, Bassaganyas L, et al. Exome sequencing identifies recurrent mutations of the splicing factor SF3B1 gene in chronic lymphocytic leukemia. Nat Genet. 2012;44(1):47–52.

66. Abdel-Wahab O, Kilpivaara O, Patel J, Busque L, Levine RL. The most commonly reported variant in ASXL1 (c.1934dupG;p.Gly646TrpfsX12) is not a somatic alteration. Leukemia. 2010;24(9):1656–64. pmid:20596031

67. Roberts NJ, Norris AL, Petersen GM, Bondy ML, Brand R, Gallinger S, et al. Whole genome sequencing defines the genetic heterogeneity of familial pancreatic cancer. Cancer Discov. 2016;6(2):166–75. pmid:26658419

68. Xie M, Lu C, Wang J, McLellan MD, Johnson KJ, Wendl MC, et al. Age-related cancer mutations associated with clonal hematopoietic expansion. Nat Med. 2014;20(12):1472–8.

69. Hamadou WS, Abed RE, Besbes S, Bourdon V, Fabre A, Youssef YB, et al. Familial hematological malignancies: ASXL1 gene investigation. Clin Transl Oncol. 2016;18(4):385–90. pmid:26286068

70. Purdue MP, Lan Q, Menashe I, Zheng T, Zhang Y, Yeager M, et al. Variation in innate immunity genes and risk of multiple myeloma. Hematol Oncol. 2011;29(1):42–6. pmid:20658475

71. Hosgood HD, Purdue MP, Wang SS, Zheng T, Morton LM, Lan Q, et al. A pooled analysis of three studies evaluating genetic variation in innate immunity genes and non-Hodgkin lymphoma risk. Br J Haematol. 2011;152(6):721–6. pmid:21250972

72. Cerhan JR, Ansell SM, Fredericksen ZS, Kay NE, Liebow M, Call TG, et al. Genetic variation in 1253 immune and inflammation genes and risk of non-Hodgkin lymphoma. Blood. 2007;110(13):4455–63. pmid:17827388

73. Geahlen RL. Syk and pTyr’d: Signaling through the B cell antigen receptor. Biochim Biophys Acta. 2009;1793(7):1115–27. pmid:19306898

74. Mócsai A, Ruland J, Tybulewicz VLJ. The SYK tyrosine kinase: A crucial player in diverse biological functions. Nat Rev Immunol. 2010;10(6):387–402. pmid:20467426

75. Chen L, Monti S, Juszczynski P, Ouyang J, Chapuy B, Neuberg D, et al. SYK Inhibition Modulates Distinct PI3K/AKT- Dependent Survival Pathways and Cholesterol Biosynthesis in Diffuse Large B Cell Lymphomas. Cancer Cell [Internet]. 2013;23(6):826–38. Available from: pmid:23764004

76. Nadeu F, Royo R, Clot G, Duran-Ferrer M, Navarro A, Martín S, et al. IGLV3-21R110 identifies an aggressive biological subtype of chronic lymphocytic leukemia with intermediate epigenetics. Blood. 2021;137(21):2935–46. pmid:33211804

77. Armand M, Verrier A, Theves F, Bravetti C, Garff-Tavernier M Le, Choquet S, et al. Prevalence of IGLV3-21R110 among familial CLL: a retrospective study of 45 cases. Blood Adv. 2022;6(12):3632–3635. pmid:35294518

78. Petersen DL, Berthelsen J, Willerslev-Olsen A, Fredholm S, Dabelsteen S, Bonefeld CM, et al. A novel BLK-induced tumor model. Tumor Biol. 2017;39(7):1–8. pmid:28670978

79. Flerlage JE, Myers JR, Maciaszek JL, Oak N, Rashkin SR, Hui Y, et al. Discovery of novel predisposing coding and noncoding variants in familial Hodgkin lymphoma. Blood. 2022;

80. Liu SKW, Berry DM, McGlade CJ. The role of Gads in hematopoietic cell signalling. Oncogene. 2001;20(44 REV. ISS. 5):6284–90. pmid:11607830

81. Miyata-Takata T, Chuang SS, Takata K, Toji T, Maeda Y, Sato Y, et al. Expression of T-cell receptor signalling pathway components in extranodal NK/T-cell lymphoma. Histopathology. 2018;73(6):1030–8. pmid:30102799

82. Abaffy T, Bain JR, Muehlbauer MJ, Spasojevic I, Lodha S, Bruguera E, et al. A testosterone metabolite 19-hydroxyandrostenedione induces neuroendocrine trans-differentiation of prostate cancer cells via an ectopic olfactory receptor. Front Oncol. 2018;8(162):1–15. pmid:29892571

83. Weber L, Maßberg D, Becker C, Altmüller J, Ubrig B, Bonatz G, et al. Olfactory receptors as biomarkers in human breast carcinoma tissues. Front Oncol. 2018;8(33):1–12. pmid:29497600

84. Glusman G, Yanai I, Rubin I, Lancet D. The complete human olfactory subgenome. Genome Res. 2001;11(5):685–702. pmid:11337468

85. Yang J, Weedon MN, Purcell S, Lettre G, Estrada K, Willer CJ, et al. Genomic inflation factors under polygenic inheritance. Eur J Hum Genet. 2011;19(7):807–12. pmid:21407268

86. Zheng Q, Zhang Y, Jiang J, Jia J, Fan F, Gong Y, et al. Exome-Wide Association Study Reveals Several Susceptibility Genes and Pathways Associated With Acute Coronary Syndromes in Han Chinese. Front Genet. 2020;11(April):1–10.

87. Liu Y, Lusk CM, Cho MH, Silverman EK, Qiao D, Zhang R, et al. Rare Variants in Known Susceptibility Loci and Their Contribution to Risk of Lung Cancer. J Thorac Oncol. 2018;13(10):1483–95. pmid:29981437