You are here

Article Text

Article menu
Cancer genetics
Original research
Ability of a polygenic risk score to refine colorectal cancer risk in Lynch syndrome
  1. Nuria Dueñas1,2,3,4,
  2. Hannah Klinkhammer5,6,
  3. Nuria Bonifaci2,3,
  4. Isabel Spier4,7,8,
  5. Andreas Mayr5,
  6. Emadeldin Hassanin6,9,
  7. Anna Diez-Villanueva10,11,12,
  8. Victor Moreno10,11,12,13,
  9. Marta Pineda1,2,3,4,
  10. Carlo Maj6,
  11. Gabriel Capellà1,2,3,4,
  12. Stefan Aretz4,7,8,
  13. Joan Brunet1,2,3,4,14
  1. 1 Hereditary Cancer Program, Catalan Institute of Oncology - ICO, L’Hospitalet de Llobregat, Spain
  2. 2 Hereditary Cancer Group, Molecular Mechanisms and Experimental Therapy in Oncology Program, Institut d’Investigació Biomèdica de Bellvitge - IDIBELL, L'Hospitalet de Llobregat, Spain
  3. 3 Biomedical Research Centre Network for Oncology (CIBERONC), Instituto Salud Carlos III, Madrid, Spain
  4. 4 European Reference Network on Genetic Tumour Risk Syndromes (ERN GENTURIS), Nijmegen, Netherlands
  5. 5 Institute for Medical Biometry, Informatics and Epidemiology, Medical Faculty, University of Bonn, Bonn, Germany
  6. 6 Institute for Genomic Statistics and Bioinformatics, Medical Faculty, University of Bonn, Bonn, Germany
  7. 7 Institute of Human Genetics, Medical Faculty, University of Bonn, Bonn, Germany
  8. 8 National Center for Hereditary Tumor Syndromes, University of Bonn, Bonn, Germany
  9. 9 Luxembourg Centre for Systems Biomedicine, University of Luxembourg, Esch-sur-Alzette, Luxembourg
  10. 10 Oncology Data Analytics Program (ODAP), Catalan Institute of Oncology - ICO, L'Hospitalet de Llobregat, Spain
  11. 11 Colorectal Cancer Group (ONCOBELL), Institut d’Investigació Biomèdica de Bellvitge - IDIBELL, L’Hospitalet de Llobregat, Spain
  12. 12 Biomedical Research Centre Network for Epidemiology and Public Health (CIBERESP), Instituto Salud Carlos III, Madrid, Spain
  13. 13 Department of Clinical Sciences, Faculty of Medicine and Health Sciences and Universitat de Barcelona Institute of Complex Systems (UBICS), Universitat de Barcelona, Barcelona, Spain
  14. 14 Hereditary Cancer Program, Catalan Institute of Oncology - ICO, Girona, Spain
  1. Correspondence to Dr Joan Brunet, Hereditary Cancer Program, Catalan Institute of Oncology, Girona 17190, Spain; jbrunet{at}iconcologia.net; Professor Stefan Aretz; stefan.aretz{at}uni-bonn.de

Abstract

Background Polygenic risk scores (PRSs) have been used to stratify colorectal cancer (CRC) risk in the general population, whereas its role in Lynch syndrome (LS), the most common type of hereditary CRC, is still conflicting. We aimed to assess the ability of PRS to refine CRC risk prediction in European-descendant individuals with LS.

Methods 1465 individuals with LS (557 MLH1, 517 MSH2/EPCAM, 299 MSH6 and 92 PMS2) and 5656 CRC-free population-based controls from two independent cohorts were included. A 91-SNP PRS was applied. A Cox proportional hazard regression model with ‘family’ as a random effect and a logistic regression analysis, followed by a meta-analysis combining both cohorts were conducted.

Results Overall, we did not observe a statistically significant association between PRS and CRC risk in the entire cohort. Nevertheless, PRS was significantly associated with a slightly increased risk of CRC or advanced adenoma (AA), in those with CRC diagnosed <50 years and in individuals with multiple CRCs or AAs diagnosed <60 years.

Conclusion The PRS may slightly influence CRC risk in individuals with LS in particular in more extreme phenotypes such as early-onset disease. However, the study design and recruitment strategy strongly influence the results of PRS studies. A separate analysis by genes and its combination with other genetic and non-genetic risk factors will help refine its role as a risk modifier in LS.

  • Digestive System Neoplasms
  • Congenital, Hereditary, and Neonatal Diseases and Abnormalities
  • Early Diagnosis
  • Genetic Association Studies
  • Genetic Counseling

Data availability statement

Data are available upon reasonable request. Data supporting the results were stored in local databases at both centres.

https://doi.org/10.1136/jmg-2023-109344

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    WHAT IS ALREADY KNOWN ON THIS TOPIC

    • Great variability in the incidence of colorectal cancer (CRC) has been described in individuals with LS, even within the same family.

    • Polygenic risk scores (PRSs) can help stratify colorectal cancer risk and, thus, adjust surveillance or treatment procedures.

    WHAT THIS STUDY ADDS

    • PRS performed on family-based registries slightly influences CRC risk in subgroups of individuals with LS, even though with weak effects.

    • Our study showed a weak association between PRS and multiple and young CRC cases, pointing to a possible risk-modifying role in extreme phenotypes.

    HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

    • Gene-based PRS analysis and its combination with other genetic and non-genetic factors may contribute to refining cancer risk in patients with LS.

    Introduction

    Colorectal cancer (CRC) is the third most incident cancer overall and the second leading cause of cancer-related death worldwide. Incidence rates are four times higher in the Global North, associated with lifestyle and dietary risk factors.1

    About 5% of CRC is considered hereditary due to highly penetrant pathogenic germline variants in cancer-predisposing genes.2 3 The main cause of hereditary CRC is Lynch syndrome (LS), with an estimated carrier frequency in the general population of around 1:279.4 It is characterised as an autosomal dominant inherited defect in any of the mismatch repair (MMR) genes (MLH1, MSH2, MSH6 and PMS2) or EPCAM gene deletions, resulting in silencing of the MSH2 gene in epithelial tissues.5 Median CRC cumulative incidences at 75 years show an important variability according to mutated gene and gender: 48/57%, 47/51% and 18/20% for male and female carriers of mutations in MLH1, MSH2 and MSH6, respectively, and 10% for both genders in carriers of mutations in PMS2.6 Differences in CRC risk have also been identified based on the ethnic or geographical origin of carriers, with lower risks reported for European versus American and Australasian individuals.7 Moreover, LS carriers have an increased risk of developing multiple CRCs, CRC at a younger age and other LS-associated cancers such as endometrial cancer (EC) or ovarian cancer.6

    In LS, as in other hereditary cancer predisposition syndromes characterised by incomplete penetrance, one of the main challenges is to identify which risk-modifying factors may modulate the expression of the cancer syndrome.7 8 In recent years, multiple, common, low-penetrance CRC risk variants have been identified through genome-wide association studies (GWASs).9–11 Each risk allele individually confers a small risk, but their combined effect as a polygenic risk score (PRS) exhibits significant risks of developing CRC in the general population. Being in the highest PRS percentiles was shown to increase the risk of CRC twofold to sevenfold.10 12–16 Moreover, PRS might be particularly relevant in patients with a more extreme, that is, severe, phenotype: a study performed in individuals diagnosed with CRC before 50 years of age (early-onset disease) demonstrated the existence of an interaction between PRS and CRC risk, with an OR of 3.73 (95% CI 3.28 to 4.24) in the highest PRS quartile.17 Another study on familial CRC (individuals who fulfil Amsterdam or Bethesda criteria without a pathogenic germline MMR variant) identified an increased CRC risk in individuals in the highest 5% of the PRS distribution, with an OR of 4.89 (95% CI 2.37 to 10.07).18

    To date, the modulating effect of PRS on CRC risk in individuals with LS is still controversial. Two studies on a population-based repository from the UK Biobank (UKBB), including 76 and 388 LS carriers, respectively, reported that PRS may strongly influence CRC risk16 19; however, another analysis of the clinic-based registry of the Colon Cancer Family Registry (CCFR), including 826 European-descendant individuals with LS, found no evidence of association, irrespective of sex or mutated gene.20

    Our objective was to evaluate whether differences in CRC penetrance in European-descendant individuals with LS can, in part, be explained by the accumulation of low-risk CRC alleles using a validated set of 91 SNPs for PRS analysis.

    Methods

    Study participants

    Individuals with LS

    A total of 1465 European-descendant individuals with genetically confirmed LS (557 MLH1, 517 MSH2/EPCAM, 299 MSH6 and 92 PMS2) from two independent cohorts were included: 918 individuals with LS (353 families) identified at the Catalan Institute of Oncology (ICO, Spain) and 547 individuals with LS (392 families) from the University Hospital of Bonn (UKB, Germany). Patients were recruited based on the fulfilment of Bethesda or Amsterdam criteria or via an EC and CRC-based LS screening programme (since 2016 at the ICO).21 Patients included were affected index patients and affected or unaffected carriers among the relatives identified through cascade testing. In the ICO LS cohort, there was a lower percentage of pathogenic MSH2 variant carriers (mainly due to the existence of MLH1 founder mutations in the ICO series) and a higher percentage of pathogenic MSH6 and PMS2 variant carriers (mainly identified through an EC/CRC-based LS screening) when compared with the UKB LS cohort. In addition, the ICO cohort included a higher proportion of non-index individuals. There were no significant differences in the distribution of affected genes between early-onset cases and the entire cohort (table 1).

    View this table:
    Table 1

    Main characteristics of the population studied

    Individuals without LS

    A total of 5656 unselected CRC-free individuals from the same population were included in the analysis (CRC-free population controls): 1642 individuals from Spain and 4014 from Germany. The controls from Spain included individuals from the Colorectal Cancer Genetics & Genomics (CRCGEN) study and individuals participating in a population-based CRC screening programme, most of whom had a positive faecal immunochemical test result and a colonoscopy without cancer or advanced adenoma (AA), as described elsewhere.22 The German controls were drawn from the population-based Heinz Nixdorf (HNR) Risk Factors, Evaluation of Coronary Calcification and Lifestyle (RECALL) study as described recently.23 (table 1).

    Data collection

    Clinical data included demographic, personal and oncological history, and follow-up carried out from birth to June 2021. In individuals with LS, histories of colorectal polyps or other LS-related cancers were also collected. Data supporting the results were stored in local databases at both centres.

    SNP selection

    The selected SNPs (n=95) and associated risks were obtained from the meta-analysis of CRC risk alleles performed by Huyghe et al 10(online supplemental table S1) and were commonly used to study sporadic CRC risk at the initiation of the study.16 19 Individual CRC risk-associated SNPs reached independent genome-wide significance (p<5×10−8) in a large-scale GWAS.

    Supplemental material

    Genotyping

    ICO blood DNA samples were genotyped with the Illumina Global Screening Array-24 (GSA) V.2.0 and V.3.0 (https://emea.illumina.com/science/consortia/human-consortia/global-screening-consortium.html) and UKB samples with GSA V.3.0. Of note, 48% of the ICO population of CRC-free individuals were previously included in the meta-analysis by Huyghe et al 10; however, they corresponded to ~1% of the total number of cases and controls in the analysis. Details regarding quality control procedures and correlation between arrays have been described previously.18 22

    Non-European-descendant individuals were excluded from the analysis. To assess ethnicity, Spanish samples were compared with 1397 HapMap samples, while German samples were compared with 1000 Genomes Project samples. Classification into different ethnicity groups was performed by selecting ancestry-informative marker SNPs and using a principal component analysis approach.

    Imputation

    Of the 95 variants of interest, 13 and 18 were included in the Illumina GSA-24 V.2.0 and V.3.0, respectively. Variants not directly genotyped by the corresponding arrays were imputed in the ICO with the Michigan Imputation Server (HRC V.r1.1.2016 panel)24 and in the UKB with a comparable pipeline based on the bioinformatic tools bcftools, minimac and vcftools, using GRCh37 as the reference genome (1000 Genomes Project, phase III, V.5).25 Missing variants and variants with an imputation quality (r2) of <0.3 (considering all genotyped samples) were not included in the final PRS analysis, which resulted in the exclusion of rs6058093, rs35470271, rs145364999 and rs755229494 (online supplemental table S1).

    PRS calculation

    For each participant, PRS was computed using the PLINK score function26 based on the 91 quality-controlled CRC risk alleles (coded as 0, 1, or 2) and effect sizes as reported by Huyghe et al (PRS) and averaged over the number of observed variants per individual10 (weighted PRS - wPRS). To ease interpretation, wPRS values were rescaled (rescaled weighted PRS - rwPRS) to indicate risk per allele (using the ratio of non-averaged PRS and wPRS values in controls as a scaling factor) as previously reported.18

    Study events

    Two events were considered: (i) CRC and (ii) AA (adenoma with significant villous features (>25%), size≥1.0 cm, high-grade dysplasia, or early invasive cancer).

    Two subgroups were defined for the primary analysis: affected individuals (individuals with LS with CRC and CRC or AA) and unaffected individuals (CRC-free or CRC-free and AA-free individuals with LS). For the subanalysis of multiple CRCs, three subgroups were defined: multiple events (individuals with LS with multiple CRC and multiple CRC or AA), single event (individuals with LS with single CRC and single CRC or AA) and no-event (CRC-free and CRC-free and AA-free individuals with LS). CRC-free population controls were only compared with individuals with LS with CRC or multiple CRC when considering CRC as a study event as no reliable information was available regarding AA in this population (online supplemental tables S2 and S3 and figure S1).

    View this table:
    Table 2

    Meta-analysis results of the association between polygenic risk score (rwPRS) and CRC risk or CRC and AA risk

    Statistical methods

    Statistical analyses and graphical representations were conducted with R V.4.0.5. For the primary analysis, the association between rwPRS and CRC and CRC or AA risk was tested by considering time to CRC (years since birth to event of study) using a Cox proportional hazard regression model with family as a random effect (frailty model). Observations in the control cohort were right censored at the age of last contact, and CRC diagnosis (yes/no) was used as an event variable. The date of the first polypectomy for adenoma was used as a time-dependent variable. Additionally, sex, birth cohort (<1940, 1940–1949, 1950–1959, 1960–1969, 1970–1979 and >1980) and other LS-related cancers were included as covariates.

    For the subanalysis of multiple CRCs, the association between rwPRS and multiple CRCs or AAs was tested using a mixed effects logistic regression, including age, sex, birth cohort, polypectomy before the second CRC, the occurrence of other cancers and family (random effect) as covariates.

    Results from both cohorts (ICO and UKB) were combined and analysed via a fixed-effect meta-analysis and the inverse-variance method. The combined rwPRS effect was estimated as the weighted average of the estimates of the individual studies, and weights were derived as the inverse of the variance of the individual effect estimate. The population was stratified according to rwPRS tertiles using the medium category as a reference. Additionally, to test for heterogeneity, Cochran’s Q was computed on the derived estimates and a χ2 test with 1 df was performed. Results with p values of <0.05 in the test for heterogeneity were not considered. The meta-analysis was conducted via R package meta.27 To correct for multiple testing, analyses were grouped by study event and control group, and p values inside these groups were corrected via false discovery rate (FDR) correction.28 Only results with p values of <0.05 after FDR correction (p-FDR) were considered statistically significant.

    Results

    No differences in PRS distribution were observed when comparing CRC-free individuals with LS and CRC-free population controls in any of the cohorts studied (online supplemental figure S2).

    Primary analysis

    CRC as the study event

    A statistically significant association between rwPRS and CRC risk was found in LS carriers under 50 years of age compared with CRC-free individuals with LS (HR=1.022 [1.007–1.038], p-FDR=0.01). We found a tendency for an association between rwPRS and CRC risk in the entire cohort and MSH6 variant carriers. We found no statistically significant association between rwPRS and CRC risk when comparing CRC LS to CRC-free population individuals (table 2 and online supplemental table S4).

    Additionally, rwPRS tended to be associated with higher CRC risk in MSH2/EPCAM (tertile low: HR=0.716, 95% CI 0.505 to 1.016, p-FDR=0.53, vs tertile high: HR=1.058, 95% CI 0.769 to 1.455, p-FDR=0.96) and MSH6 variant carriers (tertile low: HR=0.617, 95% CI 0.299 to 1.271, p-FDR=0.53, vs tertile high: HR=1.594, 95% CI 0.929 to 2.735, p-FDR=0.53); however, results were not statistically significant (figure 1).

    Figure 1
    Figure 1

    Interplay of each mutated germline gene and PRS (rwPRS) for CRC risk. CRC risk for each mutated gene, stratified according to rwPRS tertiles using the intermediate rwPRS tertile as the reference group. Ninety-five per cent CIs are indicated by vertical lines. PMS2 carriers are not included due to the low sample size. CRC, colorectal cancer; PRS, polygenic risk score.

    CRC or AA as study events

    A statistically significant association between rwPRS and CRC or AA risk was observed in the entire cohort (HR=1.019, 95% CI 1.005 to 1.032, p-FDR=0.03) and in LS carriers under 50 years of age (HR=1.022, 95% CI 1.006 to 1.038, p-FDR=0.006). We observed a tendency for an association between rwPRS and CRC or AA risk in MSH2/EPCAM and MSH6 carriers (table 2 and online supplemental table S5).

    Even though no statistically significant associations were observed (figure 2), rwPRS tended to be associated with a higher risk of CRC and AA in MSH6 variant carriers (tertile low: HR=0.669, 95% CI 0.322 to 1.393, p-FDR=0.57, vs tertile high: HR=2.015, 95% CI 1.169 to 3.471, p-FDR=0.39).

    Figure 2
    Figure 2

    Interplay of each mutated germline gene and PRS (rwPRS) for CRC or AA risk. Risk of CRC or AA for each mutated gene, stratified according to rwPRS tertiles using the intermediate rwPRS tertile as the reference group. Ninety-five per cent CIs are indicated by vertical lines. PMS2 carriers are not included due to the low sample size. AA, advanced adenoma; CRC, colorectal cancer; PRS, polygenic risk score.

    Subanalysis: multiple CRCs

    Multiple CRCs as the study event

    No statistically significant association between rwPRS and multiple CRC risk was observed (irrespective of the gene involved) when comparing single-CRC LS cases, CRC-free individuals with LS or CRC-free population controls (table 3 and online supplemental table S6). These analyses could not be performed in MSH6 or PMS2 carriers due to the low sample sizes.

    View this table:
    Table 3

    Meta-analysis results of the association between polygenic risk score (rwPRS) and multiple CRC risk or multiple CRC or AA risk

    Multiple CRCs or AAs as study events

    A significant association between rwPRS and multiple CRC or AA risk was observed in individuals with LS under 60 years when comparing with single-CRC or AA LS (HR=1.057, 95% CI 1.010 to 1.100, p-FDR=0.04) and CRC-free and AA-free LS (HR=1.043, 95% CI 1.008 to 1.079, p-FDR=0.03). A tendency was observed for an association between rwPRS and multiple CRC or AA risk in the entire cohort and MLH1 carriers (table 3 and o nline supplemental table S7). These analyses could not be performed in MSH6 and PMS2 carriers due to the low sample size.

    Discussion

    PRS is regarded as an important addition to the assessment of an individual’s genetic risk in patients with sporadic and hereditary cancers; it can be used to identify individuals with a CRC risk several times lower or higher than that of the average population. In this way, its implementation seems to be a promising approach for a more individualised risk stratification. Several studies described the impact of PRS on modelling CRC risk in the general population.10 12–16 In line with this, the risk alleles of those SNPs were found to accumulate in unexplained familial and early-onset CRC cases.17 18 However, the interplay between a PRS based on sporadic CRC-associated SNPs and LS CRC risk remains controversial.

    It is well known that among patients with hereditary CRC, in particular LS, the age of onset and cumulative CRC incidence is very heterogeneous, even within the same family transmitting the same pathogenic germline variant.6 The estimated gene-specific, individual lifetime CRC risks of patients with LS with MLH1 or MSH2 variants can be lower than 10% or as high as 90%–100% in a considerable fraction, highlighting relevant genetic and non-genetic modifiers of CRC risk.7 8 Initially, a small subset of common CRC-associated SNPs was analysed in selected LS cohorts.29–32 More recently, some studies used a more comprehensive set of around 100 CRC-associated SNPs in large population-based or familial CRC cohorts with conflicting results.16 19 20

    Herein, we used a large, combined cohort of 1465 affected and unaffected patients with LS with pathogenic MMR germline variants, recruited at two European centres based on the fulfilment of clinical criteria (revised Bethesda or Amsterdam criteria) or as a result of an EC-based or CRC-based LS screening programme to evaluate to what extent the polygenic background modulates CRC risk. When we compared LS carriers with CRC against population-based CRC-free controls (mean age 71 years), we did not observe any significant effect of PRS on CRC risk, neither in the entire cohort nor in subgroups (gene-specific groups and early-onset group). Nevertheless, the PRS was associated with a modestly increased risk of CRC or AA in the entire LS cohort. These results are in line with the work by Jenkins et al, which is based on a similar study design, recruitment strategy and a set of 107 SNPs used for PRS calculation.20 In that work, 826 European-descent LS carriers from the CCFR were included, and the authors found no statistical evidence of an association between PRS and CRC risk, irrespective of sex or mutated gene.

    Regarding the analysis between CRC and CRC-free LS probands, we did not find a statistically significant association between CRC and PRS in the entire cohort or the different subgroups except for early-onset LS CRC cases (<50 years) and LS with multiple CRCs or AAs (<60 years), where a slightly increased CRC risk was evidenced.

    In contrast, two recent studies using UKBB data and the same 95 SNPs for PRS calculation demonstrated that the polygenic background substantially influences CRC risk in LS with ORs ranging from 8 to 118 (estimated effect of PRS) or from 4 to 16 (calculated effect of PRS) compared with the median tertile of the CRC-free population.16 19 According to these results, PRS would account for parts of the interindividual variation in CRC risk among LS carriers and might contribute towards a clinically relevant individualised risk stratification.

    The most obvious explanation for the apparently discrepant results between family-based (Jenkins et al 20 and the present study) and population-based (Fahed et al 16 and Hassanin et al 19) studies is differences in study design and recruitment strategies. The LS probands from the two familial CRC registry studies were mainly recruited based on established clinical criteria, in particular early-onset and familial clustering of CRC and other LS-related tumours. Consistent with this ascertainment approach, the vast majority of participants carry pathogenic variants in the highly penetrant MLH1 and MSH2 genes (table 4), which are likely to be less influenced by the genetic background.

    View this table:
    Table 4

    Characteristics of inclusion criteria and distribution of affected MMR genes in the different studies analysing PRS in individuals with LS

    In contrast, studies using individuals from a population-based repository (UKBB) show a different distribution of affected MMR genes, with the vast majority of individuals with LS. carrying pathogenic variants in the moderate and low-penetrance genes MSH6 or PMS2 (table 4). In a gene-specific analysis, Hassanin et al found that the modifying effect of the PRS is inversely related to the penetrance of the MMR gene, with the strongest effect in MSH6 and PMS2 carriers,19 which are clearly under-represented in the studies of Jenkins et al 20 and the present.

    This is in line with hereditary breast cancer, where PRS has proven most relevant as a cancer risk modifier in carriers of pathogenic variants in moderate penetrance genes such as CHEK2, ATM or PALB2 compared with BRCA1/2.33 34 While it can be expected that PRS may have a major influence in less penetrant CRC risk genes, we have not been able to show a significant effect, likely due to the small numbers of MSH6 and PMS2 variant carriers present in our family-based cohorts due to the aforementioned selection bias.

    Another plausible explanation for the differences observed may be the sample size. In this study, we included twice as many individuals with LS as Jenkins et al 20 and 4 and 20 times more than the UKBB analyses.16 19 Moreover, the distribution and composition of cases and controls differ between family registry-based and population-based studies (table 4). In Jenkins et al and our study, the percentages of affected (LS carriers with CRC) and unaffected (CRC-free LS carriers) individuals are similar and some are members of the same family. Hence, the CRC-free LS controls are relatives of the cases and, thus, they likely share parts of the polygenic background and other risk factors with their affected relatives (cases) to a certain extent, which may explain the observed missing effect of PRS. In contrast, the UKBB studies include 10 times more controls, supposedly healthy LS carriers apparently unrelated to the CRC LS cases. In this regard, it was shown that in both sporadic CRC35 and LS CRC,19 family history and PRS are largely independent and provide complementary information about CRC risk.

    On the other hand, there are differences in the results obtained between Jenkins et al 20 and the present study. These differences can be explained by the sample size, as discussed previously, and methodological differences. Individuals with LS in Jenkins et al were censored after a polypectomy, while we considered the first polypectomy as a time-dependent variable of CRC risk, as per studies showing a reduction in CRC incidence in individuals with LS undergoing regular colonoscopies.36 37 However, since alternative pathways of colorectal carcinogenesis seem to exist in LS carriers, which originate doubts regarding the risk-reducing impact of colonoscopies, especially in MLH1 carriers, future evidence will determine whether it is useful to apply this time-dependent variable correction in individuals with LS.38 39 Taken together, this and previous PRS studies on LS demonstrate that the study design and recruitment strategy strongly influence the results and conclusions of the PRS.

    Finally, when we analysed extreme phenotypes, such as early-onset CRC (<50 years) and young (<60 years) LS cases with multiple CRCs or AAs, a significant, although low, association between PRS and risk was observed, pointing to a possible contribution of PRS to these higher-risk situations.

    Some authors questioned whether the same CRC-associated SNPs identified in the general population and their specific effect sizes can be applied to stratify CRC risk in individuals with LS and whether both specific SNPs and their risk-modifying power may differ for each mutated gene.8 20 29 The potential identification of LS risk-modifier SNPs in large GWASs might contribute to the description of more specific risk-modulating factors in the future.

    Considering the high incidence of EC in LS,6 it would be helpful to eventually analyse the relevance of an EC-associated PRS in this context. However, both the limited sample size and the lack of a currently validated PRS for EC40–42 make it non-suitable in our work.

    PRS studies in much larger, international, multicentric LS cohorts are needed to more precisely estimate the PRS effect size in individuals with LS, especially in those with extreme phenotypes, to evaluate the relevance of the polygenic background and interplay with other genetic and non-genetic risk factors. This will enable its eventual application in routine clinical practice.

    In summary, this work shows, for the first time in a family-registry LS cohort, that the PRS can influence the CRC risk in specific subgroups of individuals with LS, although with very weak effect sizes, which contrasts with the clearer modulating effect of the PRS in LS carriers identified in population-based cohorts.

    Data availability statement

    Data are available upon reasonable request. Data supporting the results were stored in local databases at both centres.

    Ethics statements

    Patient consent for publication

    Ethics approval

    This study involves human participants and was approved by the ethics committee of the Bellvitge University Hospital (code PR225/11). The participants gave informed consent to participate in the study before taking part.

    Acknowledgments

    We thank the participating patients and families and all members of the Units of Genetic Counseling and Genetic Diagnostic of the Hereditary Cancer Program of the Catalan Institute of Oncology and the Institute of Human Genetics of the University Hospital Bonn as well as the BufaLynch Association for their support and funding of ICO’s Lynch Syndrome Database. We thank Gemma Aiza for technical support. The authors also acknowledge the Department of Medicine at the Universitat Autònoma de Barcelona and the CERCA Program/Generalitat de Catalunya for institutional support. This research is supported (not financially) by the European Reference Network on Genetic Tumour Risk Syndromes (ERN GENTURIS). ERN GENTURIS is funded by the European Union.

    References

    1. ↵
      2. Sung H ,
      3. Ferlay J ,
      4. Siegel RL , et al
      . Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2021;71:209–49. doi:10.3322/caac.21660
      OpenUrlCrossRefPubMed
    2. ↵
      2. Czene K ,
      3. Lichtenstein P ,
      4. Hemminki K
      . Environmental and heritable causes of cancer among 9.6 million individuals in the Swedish family-cancer database. Int J Cancer 2002;99:260–6. doi:10.1002/ijc.10332
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      2. Moreira L ,
      3. Balaguer F ,
      4. Lindor N , et al
      . Identification of Lynch syndrome among patients with colorectal cancer. JAMA 2012;308:1555. doi:10.1001/jama.2012.13088
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      2. Win AK ,
      3. Jenkins MA ,
      4. Dowty JG , et al
      . Prevalence and penetrance of major genes and Polygenes for colorectal cancer. Cancer Epidemiol Biomarkers Prev 2017;26:404–12. doi:10.1158/1055-9965.EPI-16-0693
      OpenUrlAbstract/FREE Full Text
    5. ↵
      2. Hampel H ,
      3. de la Chapelle A
      . How do we approach the goal of identifying everybody with Lynch syndrome. Fam Cancer 2013;12:313–7. doi:10.1007/s10689-013-9611-5
      OpenUrlCrossRefPubMed
    6. ↵
      2. Dominguez-Valentin M ,
      3. Sampson JR ,
      4. Seppälä TT , et al
      . Cancer risks by gene, age, and gender in 6350 carriers of pathogenic mismatch repair variants: findings from the prospective Lynch syndrome database. Genet Med 2020;22:15–25. doi:10.1038/s41436-019-0596-9
      OpenUrlCrossRefPubMed
    7. ↵
      1. International Mismatch Repair Consortium
      . Variation in the risk of colorectal cancer in families with Lynch syndrome: a retrospective cohort study. Lancet Oncol 2021;22:1014–22. doi:10.1016/S1470-2045(21)00189-3
      OpenUrlCrossRefPubMed
    8. ↵
      2. Scott RJ
      . Modifier genes and Lynch syndrome: some considerations. Hered Cancer Clin Pract 2022;20:35. doi:10.1186/s13053-022-00240-2
    9. ↵
      2. Law PJ ,
      3. Timofeeva M ,
      4. Fernandez-Rozadilla C , et al
      . Association analyses identify 31 new risk Loci for colorectal cancer susceptibility. Nat Commun 2019;10:2154. doi:10.1038/s41467-019-09775-w
    10. ↵
      2. Huyghe JR ,
      3. Bien SA ,
      4. Harrison TA , et al
      . Discovery of common and rare genetic risk variants for colorectal cancer. Nat Genet 2019;51:76–87. doi:10.1038/s41588-018-0286-6
      OpenUrlCrossRefPubMed
    11. ↵
      2. Fernandez-Rozadilla C ,
      3. Timofeeva M ,
      4. Chen Z , et al
      . Deciphering colorectal cancer genetics through multi-Omic analysis of 100,204 cases and 154,587 controls of European and East Asian Ancestries. Nat Genet 2023;55:519–20. doi:10.1038/s41588-023-01334-w
      OpenUrl
    12. ↵
      2. Hsu L ,
      3. Jeon J ,
      4. Brenner H , et al
      . A model to determine colorectal cancer risk using common genetic susceptibility Loci. Gastroenterology 2015;148:1330–1339. doi:10.1053/j.gastro.2015.02.010
      OpenUrlCrossRefPubMed
    13. ↵
      2. Jenkins MA ,
      3. Makalic E ,
      4. Dowty JG , et al
      . Quantifying the utility of single nucleotide Polymorphisms to guide colorectal cancer screening. Future Oncology 2016;12:503–13. doi:10.2217/fon.15.303
      OpenUrl
    14. ↵
      2. Frampton MJE ,
      3. Law P ,
      4. Litchfield K , et al
      . Implications of Polygenic risk for Personalised colorectal cancer screening. Ann Oncol 2016;27:429–34. doi:10.1093/annonc/mdv540
      OpenUrlCrossRefPubMed
    15. ↵
      2. Thomas M ,
      3. Sakoda LC ,
      4. Hoffmeister M , et al
      . Genome-wide modeling of Polygenic risk score in colorectal cancer risk. Am J Hum Genet 2020;107:432–44. doi:10.1016/j.ajhg.2020.07.006
      OpenUrlCrossRefPubMed
    16. ↵
      2. Fahed AC ,
      3. Wang M ,
      4. Homburger JR , et al
      . Polygenic background modifies penetrance of Monogenic variants for tier 1 Genomic conditions. Nat Commun 2020;11:3635. doi:10.1038/s41467-020-17374-3
    17. ↵
      2. Archambault AN ,
      3. Su Y-R ,
      4. Jeon J , et al
      . Cumulative burden of colorectal cancer-associated genetic variants is more strongly associated with early-onset vs late-onset cancer. Gastroenterology 2020;158:1274–86. doi:10.1053/j.gastro.2019.12.012
      OpenUrlCrossRefPubMed
    18. ↵
      2. Mur P ,
      3. Bonifaci N ,
      4. Díez-Villanueva A , et al
      . Non-Lynch familial and early-onset colorectal cancer explained by accumulation of low-risk genetic variants. Cancers (Basel) 2021;13:3857. doi:10.3390/cancers13153857
    19. ↵
      2. Hassanin E ,
      3. Spier I ,
      4. Bobbili DR , et al
      . Clinically relevant combined effect of Polygenic background, rare pathogenic Germline variants, and family history on colorectal cancer incidence. BMC Med Genomics 2023;16:42. doi:10.1186/s12920-023-01469-z
    20. ↵
      2. Jenkins MA ,
      3. Buchanan DD ,
      4. Lai J , et al
      . Assessment of a polygenic risk score for colorectal cancer to predict risk of Lynch syndrome colorectal cancer. JNCI Cancer Spectr 2021;5:pkab022. doi:10.1093/jncics/pkab022
    21. ↵
      2. Dueñas N ,
      3. Navarro M ,
      4. Sanjuán X , et al
      . Lessons learnt from the implementation of a colorectal cancer screening programme for Lynch syndrome in a tertiary public hospital. Cancer Epidemiol 2023;82:102291. doi:10.1016/j.canep.2022.102291
      OpenUrl
    22. ↵
      2. Obón-Santacana M ,
      3. Díez-Villanueva A ,
      4. Alonso MH , et al
      . Polygenic risk score across distinct colorectal cancer screening outcomes: from Premalignant polyps to colorectal cancer. BMC Med 2021;19:261. doi:10.1186/s12916-021-02134-x
    23. ↵
      2. Schmermund A ,
      3. Möhlenkamp S ,
      4. Stang A , et al
      . Assessment of clinically silent Atherosclerotic disease and established and novel risk factors for predicting myocardial infarction and cardiac death in healthy middle-aged subjects: rationale and design of the Heinz Nixdorf RECALL study. Am Heart J 2002;144:212–8. doi:10.1067/mhj.2002.123579
      OpenUrlCrossRefPubMedWeb of Science
    24. ↵
      2. Das S ,
      3. Forer L ,
      4. Schönherr S , et al
      . Next-generation genotype imputation service and methods. Nat Genet 2016;48:1284–7. doi:10.1038/ng.3656
      OpenUrlCrossRefPubMed
    25. ↵
      2. Auton A ,
      3. Brooks LD ,
      4. Durbin RM , et al
      . A global reference for human genetic variation. Nature 2015;526:68–74. doi:10.1038/nature15393
      OpenUrlCrossRefPubMed
    26. ↵
      2. Purcell S ,
      3. Neale B ,
      4. Todd-Brown K , et al
      . PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 2007;81:559–75. doi:10.1086/519795
      OpenUrlCrossRefPubMed
    27. ↵
      2. Balduzzi S ,
      3. Rücker G ,
      4. Schwarzer G
      . How to perform a meta-analysis with R: a practical Tutorial. Evid Based Ment Health 2019;22:153–60. doi:10.1136/ebmental-2019-300117
      OpenUrlPubMed
    28. ↵
      2. Benjamini Y ,
      3. Hochberg Y
      . Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Series B Stat Methodol 1995;57:289–300. doi:10.1111/j.2517-6161.1995.tb02031.x
      OpenUrl
    29. ↵
      2. Win AK ,
      3. Hopper JL ,
      4. Buchanan DD , et al
      . Are the common genetic variants associated with colorectal cancer risk for DNA mismatch repair gene mutation carriers. Eur J Cancer 2013;49:1578–87. doi:10.1016/j.ejca.2013.01.029
      OpenUrlCrossRefPubMed
    30. ↵
      2. Ten Broeke SW ,
      3. Elsayed FA ,
      4. Pagan L , et al
      . SNP Association study in PMS2-associated Lynch syndrome. Fam Cancer 2018;17:507–15. doi:10.1007/s10689-017-0061-3
      OpenUrlCrossRef
    31. ↵
      2. Talseth-Palmer BA ,
      3. Wijnen JT ,
      4. Brenne IS , et al
      . Combined analysis of three Lynch syndrome cohorts confirms the modifying effects of 8Q23.3 and 11Q23.1 in MLH1 Mutation carriers. Int J Cancer 2013;132:1556–64. doi:10.1002/ijc.27843
      OpenUrlCrossRefPubMed
    32. ↵
      2. Houlle S ,
      3. Charbonnier F ,
      4. Houivet E , et al
      . Evaluation of Lynch syndrome modifier genes in 748 MMR Mutation carriers. Eur J Hum Genet 2011;19:887–92. doi:10.1038/ejhg.2011.44
      OpenUrlCrossRefPubMed
    33. ↵
      2. Gallagher S ,
      3. Hughes E ,
      4. Wagner S , et al
      . Association of a Polygenic risk score with breast cancer among women carriers of high- and moderate-risk breast cancer genes. JAMA Netw Open 2020;3:e208501. doi:10.1001/jamanetworkopen.2020.8501
    34. ↵
      2. Hassanin E ,
      3. May P ,
      4. Aldisi R , et al
      . Breast and prostate cancer risk: the interplay of polygenic risk, rare pathogenic Germline variants, and family history. Genet Med 2022;24:576–85. doi:10.1016/j.gim.2021.11.009
      OpenUrlCrossRef
    35. ↵
      2. Mars N ,
      3. Lindbohm JV ,
      4. Della Briotta Parolo P , et al
      . Systematic comparison of family history and Polygenic risk across 24 common diseases. Am J Hum Genet 2022;109:2152–62. doi:10.1016/j.ajhg.2022.10.009
      OpenUrl
    36. ↵
      2. Järvinen HJ ,
      3. Aarnio M ,
      4. Mustonen H , et al
      . Controlled 15-year trial on screening for colorectal cancer in families with hereditary Nonpolyposis colorectal cancer. Gastroenterology 2000;118:829–34. doi:10.1016/s0016-5085(00)70168-5
      OpenUrlCrossRefPubMedWeb of Science
    37. ↵
      2. Kalady MF ,
      3. Lipman J ,
      4. McGannon E , et al
      . Risk of colonic neoplasia after proctectomy for rectal cancer in hereditary Nonpolyposis colorectal cancer. Ann Surg 2012;255:1121–5. doi:10.1097/SLA.0b013e3182565c0b
      OpenUrlCrossRefPubMed
    38. ↵
      2. Ahadova A ,
      3. Gallon R ,
      4. Gebert J , et al
      . Three molecular pathways model colorectal carcinogenesis in Lynch syndrome. Int J Cancer 2018;143:139–50. doi:10.1002/ijc.31300
      OpenUrlCrossRefPubMed
    39. ↵
      2. Seppälä T ,
      3. Pylvänäinen K ,
      4. Evans DG , et al
      . Colorectal cancer incidence in Path_Mlh1 carriers subjected to different follow-up protocols: a prospective Lynch syndrome database report. Hered Cancer Clin Pract 2017;15:18. doi:10.1186/s13053-017-0078-5
    40. ↵
      2. Bafligil C ,
      3. Thompson DJ ,
      4. Lophatananon A , et al
      . Association between genetic Polymorphisms and endometrial cancer risk: a systematic review. J Med Genet 2020;57:591–600. doi:10.1136/jmedgenet-2019-106529
      OpenUrlAbstract/FREE Full Text
    41. ↵
      2. Choi J ,
      3. Jia G ,
      4. Wen W , et al
      . Evaluating Polygenic risk scores in assessing risk of nine solid and hematologic cancers in European descendants. Int J Cancer 2020;147:3416–23. doi:10.1002/ijc.33176
      OpenUrl
    42. ↵
      2. Bafligil C ,
      3. Thompson DJ ,
      4. Lophatananon A , et al
      . Development and evaluation of Polygenic risk scores for prediction of endometrial cancer risk in European women. Genet Med 2022;24:1847–56. doi:10.1016/j.gim.2022.05.014
      OpenUrl

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • ND, HK and NB are joint first authors.

    • SA and JB are joint senior authors.

    • Twitter @NDuenas5

    • Contributors NDC: study concept and design, analysis and interpretation of data, drafting of the manuscript and critical revision of the manuscript for important intellectual content. HK, NB and AD-V: study concept and design, statistical analysis, analysis and interpretation of data, drafting of the manuscript and critical revision of the manuscript for important intellectual content. AM, EH and CM: study concept and design, statistical analysis, analysis and interpretation of data and critical revision of the manuscript for important intellectual content. AD-V and VM: study concept and design and critical revision of the manuscript for important intellectual content. IS, MP, GC, SA and JB: study concept and design, Analysis and interpretation of data, critical revision of the manuscript for important intellectual content and study supervision. SA and JB act as guarantor of the study.

    • Funding This research was partially funded by the Spanish Ministry of Economy and Competitiveness and the Spanish Ministry of Science and Innovation, cofunded by FEDER Funds: a Way to Build Europe (grants SAF2015-68016-R and PID2019-111254RB-I00), CIBERONC (CB16/12/00234), the Government of Catalonia (SGR_01112), the Spanish Association Against Cancer Scientific Foundation (grant GCTRA18022MORE) and Spanish Ministry for Economy and Competitivity, Instituto de Salud Carlos III, cofunded by FEDER funds: a Way to Build Europe (FIS PI14-00613). ND was funded by the Instituto de Salud Carlos III and cofunded by the European Social Fund investing in your future (grant CM19/00099), the Catalan-Balearic Society of Oncology (2018 grant of the Catalan-Balearic Society of Oncology), the European Union’s Horizon 2020 research and innovation programme under the EJP RD COFUND-EJP number 825575. AD-V was supported by PERIS contract SLT017/20/000042. The GSA genotyping was performed at the Spanish National Cancer Research Centre, in the Human Genotyping lab, a member of CeGen, PRB3, and is supported by grant PT17/0019, of the PE I+D+i 2013-2016, funded by ISCIII and ERDF.

    • Competing interests None declared.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.