Elucidating the Risk of Colorectal Cancer for Variants in Hereditary Colorectal Cancer Genes

An important subset of colorectal cancer (CRC) is caused by rare pathogenic variants in more than 20 high-risk genes,1Huyghe J.R. Bien S.A. Harrison T.A. et al.Nat Genet. 2019; 51: 76-87Crossref PubMed Scopus (265) Google Scholar, 2Seifert B.A. McGlaughon J.L. Jackson S.A. et al.Genet Med. 2019; 21: 1507-1516Abstract Full Text Full Text PDF PubMed Scopus (14) Google Scholar, 3Belhadj S. Terradas M. Munoz-Torres P.M. et al.Hum Mutat. 2020; 41: 1563-1576Crossref PubMed Scopus (23) Google Scholar The National Comprehensive Cancer Network clinical practice guidelines (2022) recommend that physicians consider multigene panel testing for these high-risk genes in all newly diagnosed CRC patients2Seifert B.A. McGlaughon J.L. Jackson S.A. et al.Genet Med. 2019; 21: 1507-1516Abstract Full Text Full Text PDF PubMed Scopus (14) Google Scholar,3Belhadj S. Terradas M. Munoz-Torres P.M. et al.Hum Mutat. 2020; 41: 1563-1576Crossref PubMed Scopus (23) Google Scholar to identify carriers of pathogenic variants and promote testing of family members who may also be carriers and would benefit from increased screening for CRC prevention. However, clinical challenges remain, such as (1) understanding the risk of CRC associated with individual variants within high-risk genes and (2) for recessively inherited CRC genes, where both alleles of the gene are defective (biallelic) because of the same pathogenic variant (homozygous carriers) or 2 different pathogenic variants (compound heterozygote carriers), understanding if CRC risk is increased if only 1 pathogenic variant is present (monoallelic carriers). Research addressing these challenges will improve clinical actionability regarding the intensity of screening and surveillance for carriers. We combined genetic data from 58,998 CRC-affected individuals and 71,171 control individuals of European ancestry from 3 major CRC consortia, namely, the Genetics and Epidemiology of Colorectal Cancer Consortium (GECCO), the Colorectal Cancer Transdisciplinary Study (CORECT), and the Colon Cancer Family Registry (CCFR)1Huyghe J.R. Bien S.A. Harrison T.A. et al.Nat Genet. 2019; 51: 76-87Crossref PubMed Scopus (265) Google Scholar (Supplementary Table 1). To enable analysis of rare genetic variants in this dataset, we used the largest available imputation panel based on whole-genome sequencing data from 97,256 samples in the National Heart, Lung, and Blood Institute Trans-Omics for Precision Medicine (TOPMed) study4Consortium G.P. Abecasis G.R. Altshuler D. et al.Nature. 2010; 467: 1061-1073Crossref PubMed Scopus (6069) Google Scholar to impute variants into genome-wide array data for CRC-affected case and control individuals. We examined the association of variants with a minor allele frequency (MAF) of <0.001 in 22 moderate- to high-penetrance CRC genes.2Seifert B.A. McGlaughon J.L. Jackson S.A. et al.Genet Med. 2019; 21: 1507-1516Abstract Full Text Full Text PDF PubMed Scopus (14) Google Scholar For the recessive CRC genes MUTYH, NTHL1, MSH3, and MBD4, we assessed the risk of CRC associated with biallelic or monoallelic carriers. If different variants within a gene increase CRC risk, testing all variants simultaneously can be more powerful; therefore, we conducted gene-based tests using the set-based Mixed-Effects Score Test (MiST).5Sun J. Zheng Y. Hsu L. Genet Epidemiol. 2013; 37: 334-344Crossref PubMed Scopus (91) Google Scholar Detailed methods are provided in the Supplementary Material. We investigated the association of single variants with CRC by modeling the variants as a log-additive effect, which is a more general model. Two significant variants were identified, the APC c.3920T>A:p.Ile1307Lys variant (odds ration [OR], 1.82; P =1.62 × 10–14), which is more common in the Ashkenazi Jewish population,6Stern H.S. Viertelhausen S. Hunter A.G. et al.Gastroenterology. 2001; 120: 392-400Abstract Full Text Full Text PDF PubMed Scopus (39) Google Scholar and the c.1187G>A:p.Gly396Asp pathogenic variant in MUTYH (OR, 1.28; P = 2.17 × 10–5) (Table 1). Overall, 24 variants in 11 of 22 high-risk genes had a P value of <.01, but aside from the 2 variants in APC and MUTYH, none surpassed multiple comparison correction (Supplementary Table 2). Applying a recessive model for genes known to act recessive demonstrated that the recessive model was a better fit for MUTYH because the ORs were larger and P values were lower than for the log-additive model. Biallelic carriers of MUTYH c.1187G>A:p.Gly396Asp (OR, 32.1; P = 1.41 × 10–6) or c.536A>G:p.Tyr179Cys (OR, 16.1; P = 0.02) and compound heterozygote carriers of these 2 variants had a substantially increased CRC risk (OR, 58.03; P = 4.2 × 10–4) (Table 1). Monoallelic carriers of either one of these 2 pathogenic variants in MUTYH did not demonstrate an increased risk of CRC. Furthermore, we investigated the association between monoallelic MUTYH carriers and CRC risk stratified by the presence or absence of 1 or more first-degree relatives with CRC. There was no evidence that monoallelic MUTYH carriers had an increased risk of CRC regardless of a family history of CRC (Supplementary Table 3). See the Supplementary Material for an analysis of monoallelic carriers (Supplementary Table 5) in additional recessive genes and candidate pathway genes (Supplementary Table 6).Table 1Imputed Germline Variants in High-Risk CRC Genes and the Association With CRC Risk Based on Individual Variant Level Analysis, Recessive Model Analysis of the 2 Common Pathogenic Variants in the MUTYH Gene, and the Set-Based MiST Gene Burden AnalysisIndividual variant–level analysisGeneVariantgnomAD NFE frequencyImputation accuracyCase/control individualsOR (95% CI)P valueAPCc.3920T>A, p.Ile1307Lys (rs1801155)6.4 × 10–4 (AJ, 0.036)0.948519/3061.82 (1.56–2.12)1.62 × 10–14aStatistically significant P value based on Bonferroni correction to account for multiple comparisons.MUTYHbThe number of carriers among case or control individuals for each of these 2 pathogenic variants includes counts of all carriers regardless of whether they occur as heterozygous/monoallelic, compound heterozygous, or homozygous/biallelic carriers.c.1187G>A, p.Gly396Asp (rs36053993)5.4 × 10–30.995747/7691.28 (1.15–1.42)2.17 × 10–5aStatistically significant P value based on Bonferroni correction to account for multiple comparisons.MUTYHbThe number of carriers among case or control individuals for each of these 2 pathogenic variants includes counts of all carriers regardless of whether they occur as heterozygous/monoallelic, compound heterozygous, or homozygous/biallelic carriers.c.536A>G, Tyr179Cys (rs34612342)2.3 × 10–30.872232/2391.35 (1.11–1.67)3.5 × 10–3NTHL1c.268C>T p.Gln90Ter (rs150766139)1.8 × 10–30.758195/2241.11 (0.89–1.39)0.54NTHL1c.859C>T p.Gln287Ter (rs146347092)3.6 × 10–40.55815/290.43 (0.18–0.90)0.6Recessive analysisMUTYHrs34612342 (c.536A>G, Tyr179Cys or Y179C)TTTCCCCase/control individualsOR (95% CI)Case/control individualsOR (95% CI)Case/control individualsOR (95% CI)rs36053993 (c.1187G>A, Gly396Asp or G396D)CC58,049/70,4751.00 (Reference)194/2371.13 (0.91–1.40)8/116.1 (1.42–181.72)P = .26P = .02CT665/7661.09 (0.98–1.22)30/158.03 (6.1–552.8)0/0—P = .12P = 4.2 × 10–4TT52/232.1 (7.8–131.5)0/0—0/0—P = 1.4 × 10–6Gene set-based MiST analysisGeneNumber of variantsLead variantOR (95% CI)MiSTP valueMiSTP valuecIndicates MiST P value excluding the lead variant. (minus lead variant)APC15c.3920T>A, p.Ile1307Lys (rs1801155)1.82 (1.56–2.12)0.0007.85MLH17c.1321G>A, p.Ala441Thr (rs63750365)1.55 (1.02–2.37)0.0362.16MSH3dGenes inherited in an autosomal recessive pattern.12c.2262A>G, p.Ile754Met (rs200819607)0.406 (0.184–0.897)0.0216.21NOTE. The lead variant is the most associated variant at the locus. The reference single-nucleotide polymorphism cluster ID is based on the National Center for Biotechnology Information dbSNP Build 150. Alleles are on the positive strand. P values are based on fixed-effects inverse variance–weighted meta-analysis. For recessive analysis, OR is the OR estimate for the risk allele. P values reported in this section are based on pooled data analysis. Estimates were adjusted for age, sex, and genome-wide association study genotyping platform. For set-based MiST analysis, genetic variants in each gene were restricted to missense, stop-gained, frameshift, and splice site variants with a minor allele count of >10, an MAF of <5%, and an imputation R2 of >0.3. CADD and REVEL prediction scores were included as continuous functional weights: 1 if CADD > 20 or REVEL > 0.5.AJ, Ashkenazi Jewish gnomAD frequency; CI, confidence interval; NFE, non-Finnish European gnomAD frequency.a Statistically significant P value based on Bonferroni correction to account for multiple comparisons.b The number of carriers among case or control individuals for each of these 2 pathogenic variants includes counts of all carriers regardless of whether they occur as heterozygous/monoallelic, compound heterozygous, or homozygous/biallelic carriers.c Indicates MiST P value excluding the lead variant.d Genes inherited in an autosomal recessive pattern. Open table in a new tab NOTE. The lead variant is the most associated variant at the locus. The reference single-nucleotide polymorphism cluster ID is based on the National Center for Biotechnology Information dbSNP Build 150. Alleles are on the positive strand. P values are based on fixed-effects inverse variance–weighted meta-analysis. For recessive analysis, OR is the OR estimate for the risk allele. P values reported in this section are based on pooled data analysis. Estimates were adjusted for age, sex, and genome-wide association study genotyping platform. For set-based MiST analysis, genetic variants in each gene were restricted to missense, stop-gained, frameshift, and splice site variants with a minor allele count of >10, an MAF of <5%, and an imputation R2 of >0.3. CADD and REVEL prediction scores were included as continuous functional weights: 1 if CADD > 20 or REVEL > 0.5. AJ, Ashkenazi Jewish gnomAD frequency; CI, confidence interval; NFE, non-Finnish European gnomAD frequency. Previous studies based on substantially smaller sample sizes have reported an increased risk of CRC for monoallelic carriers of MUTYH pathogenic variants, especially in the presence of a first-degree relative with CRC,7Win A.K. Dowty J.G. Cleary S.P. et al.Gastroenterology. 2014; 146: 1208-1211.e1-5Abstract Full Text Full Text PDF PubMed Scopus (148) Google Scholar although the effect size was small. Accordingly, our study provides strong evidence that biallelic, but not monoallelic, inactivation of the MUTYH gene predisposes to CRC. This is further supported by observations that COSMIC tumor mutational signatures (SBS18/SBS36), indicative of defective base excision repair, are present only in biallelic MUTYH pathogenic variants carriers but not in CRC tumors of monoallelic MUTYH pathogenic variant carriers.8Georgeson P. Pope B.J. Rosty C. et al.Gut. 2021; 70: 2138-2149Crossref PubMed Scopus (13) Google Scholar Similarly, monoallelic carriers of either of the 2 common pathogenic variants in NTHL1, c.268C>T:p.Gln90Ter or c.859C>T:p.Gln287Ter, were not associated with an increased risk of CRC. Our finding confirms a recent analysis of 5942 individuals with CRC or unexplained polyposis that showed no evidence of an increased risk of CRC in monoallelic NTHL1 loss of function variant carriers and the absence of SBS30 mutational signature unique to biallelic inactivation of NTHL1.9Elsayed F.A. Grolleman J.E. Ragunathan A. et al.Gastroenterology. 2020; 159: 2241-2243.e6Abstract Full Text Full Text PDF PubMed Scopus (14) Google Scholar When we tested the association with CRC risk for all predicted pathogenic variants within a gene using gene-based MiST testing, we found that the combined burden of rare predicted pathogenic variants in the APC (P = .0007), MSH3 (P = .0216), and MLH1 (P = .0362) genes increased CRC risk, but these associations were driven by a single rare variant per gene. Exclusion of each of these lead variants resulted in a diminished and nonsignificant signal (Table 1). Imputation to TOPMed allowed us to test a substantially larger number of genetic variants at higher imputation quality than our previous efforts using the Haplotype Reference Consortium1Huyghe J.R. Bien S.A. Harrison T.A. et al.Nat Genet. 2019; 51: 76-87Crossref PubMed Scopus (265) Google Scholar (26.4 million vs 14.8 million single-nucleotide polymorphisms with MAFs between 0.1% and 1%) (Supplementary Table 4). We note that although sequencing provides the best quality, it has been shown that imputation of rare variants is more reliable than direct genotyping using arrays.10Hanks S.C. Forer L. Schonherr S. et al.Am J Hum Genet. 2022; 109: 1653-1666Abstract Full Text Full Text PDF PubMed Scopus (5) Google Scholar Nevertheless, the TOPMed imputation remains limited for variants with MAFs as low as ∼0.01%, and large-scale whole-genome sequencing studies are required to identify those ultrarare genetic variants. Following the lead where this has been successful in cardiometabolic and lung diseases via the TOPMed program, dedicated funding is required to enable a well-powered discovery effort for cancer. The study population is limited to individuals of European descent and needs to be expanded to include other ancestry populations. This well-powered study provides strong evidence that monoallelic pathogenic variant carriers in MUTYH, regardless of family history of CRC, as well as other recessive-acting genes, such as NTHL1, MSH3, and MBD4 do not have an increased CRC risk. Accordingly, intensified surveillance of monoallelic carriers of these recessive genes, which have a relatively high population frequency close to 1%, is not warranted. The GECCO-CCFRC Consortium includes Xiaoliang Wang,1 Jeroen R. Huyghe,1 Jihoon E. Joo,2,3 Peter Georgeson,2,3 Volker Arndt,4 Sonja I. Berndt,5 Stéphane Bézieau,6 Stephanie A. Bien,1 D. Timothy Bishop,7 Hermann Brenner,4,8,9 Stefanie Brezina,10 Peter C. M. J. A. I. A. C. R. I. A. D. C. Stephanie L. E. L. S. M. Y. Stephanie J. M. and A. from the Cancer of The of of Center for Cancer Comprehensive Cancer of Epidemiology and Cancer Research of Cancer Epidemiology and National Cancer National of Institute of Research at of of Cancer Research Center and National Center for Cancer Cancer Research of Cancer of Medicine for of Epidemiology and of for of of of and of Network and and and Epidemiology and Institute of and Institute of of of of and of of Cancer Cancer Research Center Cancer Center of and of of of Research and in Medicine and of of of of of of Comprehensive Cancer Institute for of of Cancer Research of Epidemiology Cancer for Epidemiology and of and The of of at and for Research on of Medicine of of The Comprehensive Cancer of of for and of of of of Cancer of of of Family of of of Medicine and Research Institute of of of Research Institute of of of Cancer in of and Cancer of Medicine and of Epidemiology of and of of of of of Center for and of Medicine and and of Institute of National Cancer of of Medicine and Cancer Comprehensive Cancer of Medicine and of of of of Cancer of of Medicine and Cancer Institute and of of of and the of Center for of of Institute of Medicine of the of of and of of Medicine and Center in Cancer Research of of of of The The of Medicine and Family Cancer The of of and of and of National We are to D. this would not have We also all those who to in this the and the control as well as all the and Colon Cancer Family Registry The the of study of study and the from the National Cancer which this important would not The would to the study and of the Colon Cancer Family Registry and the and Colon Cancer study We the of and the efforts of the at the Center for Research and in the of the and the would to the and at and Research and the in the We all individuals who to in the Furthermore, we all physicians and and the of the of Cancer The the and Study for to this The would also to the to this study from cancer supported the for and National of Cancer and cancer supported by the National Cancer Institute Epidemiology and We are to all in major in the the study would not be We are also to all in this We all and as well as who provided We all to the of the at of of of and of and of are as of the for Research on the are for the in this and they do not the or of the for Research on We are to all in this study who were as of the We the National of and for the of the The was at the and The study was by the of the and and of and those of as We would to the and of the and for as well as the cancer for and The for and of these We would to the at the Cancer Colorectal Cancer We the of and in this study and the of and We would to the study and the Colorectal Cancer Study The the the the Research at and at for data and samples and the from supported by the Research The the Cancer screening and the from Information and we the study for that this study Cancer data have been provided by the of Cancer Cancer Cancer Cancer Cancer Cancer Cancer Cancer Cancer and Cancer are supported in by from the for and National for or the National Cancer and The reported and the are the of the We the We the and clinical at the that in the the would not have been We are also to the dedicated who in The the and for and the study for the of can be found at D. We combined studies from 3 the Genetics and Epidemiology of Colorectal Cancer Consortium (GECCO), the Colorectal Cancer Transdisciplinary Study (CORECT), and the Colon Cancer Family Registry studies have been in J.R. Bien S.A. Harrison T.A. et al.Nat Genet. 2019; 51: 76-87Crossref PubMed Scopus Google Scholar, S. et al.Gastroenterology. 2013; Full Text Full Text PDF PubMed Scopus Google Scholar, et Cancer 2019; PubMed Scopus Google Scholar, S. et al.Nat PubMed Scopus Google Scholar and additional studies not included are in quality we included CRC or case individuals and 71,171 control individuals of European We genetic ancestry by et al.Genet Epidemiol. 2010; PubMed Scopus Google et al.Am J Hum Genet. Full Text Full Text PDF PubMed Scopus Google Scholar on a of autosomal single-nucleotide variants and comparison with individuals from the G.R. Altshuler D. et 2010; 467: 1061-1073Crossref PubMed Scopus Google Scholar Supplementary Table 1 provides on the sample and of the study provided and each study was by the or of genotyping and quality control for studies included are J.R. Bien S.A. Harrison T.A. et al.Nat Genet. 2019; 51: 76-87Crossref PubMed Scopus Google Scholar, S. et al.Gastroenterology. 2013; Full Text Full Text PDF PubMed Scopus Google Scholar, et Cancer 2019; PubMed Scopus Google Scholar, S. et al.Nat PubMed Scopus Google Scholar and samples not included quality control analysis using methods as in et et al.Genet Epidemiol. 2010; PubMed Scopus Google Scholar The genotyping data were to the National Heart, Lung, and Blood Institute TOPMed Consortium panel using the of Imputation D. et al.Nature. 2021; PubMed Scopus Google Scholar The TOPMed panel whole-genome sequencing data from 97,256 samples and genetic variants. To improve imputation and imputation was studies or genotyping that used the same or genotyping (Supplementary Table 1). We a genome-wide association analysis of individual variants with CRC risk using the test the log-additive for sex, age, and 20 to account for population by genotyping we the and for each variant. We combined the by and and the P value based on the test et al.Nat Genet. 2014; PubMed Scopus Google Scholar to genome-wide analysis, we examined the association of variants 1%) in 22 moderate- to high-penetrance genes that were reported to be or with evidence for association with CRC B.A. McGlaughon J.L. Jackson S.A. et al.Genet Med. 2019; 21: 1507-1516Abstract Full Text Full Text PDF PubMed Scopus Google Scholar This of genes includes CRC and polyposis risk genes MUTYH, NTHL1, and and genes with evidence for association with CRC polyposis MSH3, MBD4, and B.A. McGlaughon J.L. Jackson S.A. et al.Genet Med. 2019; 21: 1507-1516Abstract Full Text Full Text PDF PubMed Scopus Google Scholar, S. Terradas M. Munoz-Torres P.M. et al.Hum Mutat. 2020; 41: 1563-1576Crossref PubMed Scopus Google Scholar, J.E. F.A. et 2019; Full Text Full Text PDF PubMed Scopus Google Scholar, C. E. et al.Am J Hum Genet. 2022; 109: Full Text Full Text PDF PubMed Scopus Google Scholar, et al.Nat Genet. PubMed Scopus Google Scholar to the log-additive association we assessed the association analysis a recessive model for MUTYH because this gene is known to have a recessive of S. Terradas M. Munoz-Torres P.M. et al.Hum Mutat. 2020; 41: 1563-1576Crossref PubMed Scopus Google Scholar we examined the association of biallelic carriers (compound and carriers) for variants within MUTYH, NTHL1, MSH3, and MBD4 the frequency of biallelic carriers is we pooled data from all samples for the analysis and the model for age, sex, and genotyping than a meta-analysis. we tests at the gene using our set-based J. Zheng Y. Hsu L. Genet Epidemiol. 2013; 37: 334-344Crossref PubMed Scopus Google Scholar MiST tests the association of the effect of the variants in a by the burden test that for functional of variants and the test for the associations of individual variants that have not been by the burden using the We in functional effect prediction scores from M. P. et al.Nat Genet. 2014; PubMed Scopus Google P. D. et 2019; PubMed Scopus Google Scholar and et al.Am J Hum Genet. Full Text Full Text PDF PubMed Scopus Google Scholar and population frequency from gnomAD et al.Nature. 2020; PubMed Scopus Google Scholar as to the burden MiST among the best in of a of L. C. et Genet. PubMed Scopus Google Scholar and has been to using R. et Genet. 2020; Scholar We used the as the and genotyping data = from to estimate Variants with a minor allele count of or imputation R2 of were For genome-wide and set-based we used Bonferroni correction to account for multiple and a P value of × and genes = as For all other we a P value of as evidence for Variants were and were using Variant L. et PubMed Scopus Google Scholar on gene function was predicted for variants using in S. L. et al.Nat 2010; PubMed Scopus Google Scholar P. S. PubMed Scopus Google S. 2001; PubMed Scopus Google Scholar M. P. et al.Nat Genet. 2014; PubMed Scopus Google P. D. et 2019; PubMed Scopus Google Scholar and et al.Am J Hum Genet. Full Text Full Text PDF PubMed Scopus Google Scholar Variants with CADD of or REVEL of were to be The variants were to the M. et PubMed Scopus Google Scholar to variant allele were from the reference dataset, gnomAD where the variants are from a of whole-genome et al.Nature. 2020; PubMed Scopus Google Scholar Recessive genes MUTYH, NTHL1, MSH3, and MBD4 are associated with CRC and We did not that carriers of a single pathogenic variant in these genes are associated with an increased risk of therefore, increased screening for CRC may not be in these individuals. the frequency of these MUTYH variants in Jewish excluding samples from the study conducted in showed associations as shown in Table 1 not Similarly, monoallelic carriers of a pathogenic variant in the recessively inherited CRC genes NTHL1, MSH3, and MBD4 showed no evidence of an increased risk of CRC. We not test for CRC for biallelic (homozygous or compound carriers for the NTHL1 variants p.Gln90Ter P = and p.Gln287Ter P = because only a single biallelic was in the biallelic MSH3 or MBD4 carriers were of high-risk variants in other cancer genes found no evidence for an association with CRC risk for the cancer risk allele in P = or the risk allele in P = We were not to impute the and variants. The A. E. A. et PubMed Scopus Google Scholar variant was no significant association was An additional genes (Supplementary Table were from the base excision repair, repair, and of Genes and to our into recessive genes or compound For we all variants with an MAF of and minor allele count within a gene and tested either for or compound carriers of all variant the we did not identify additional recessive gene for multiple (Supplementary Table with Supplementary

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.