Colorectal Cancer Risk in Lynch Syndrome: Of Genes and More

Abstract

The International Mismatch Repair Consortium. Variation in the risk of colorectal cancer in families with Lynch syndrome: a retrospective cohort study. Lancet Oncol 2021;22:1014–1022.Lynch syndrome (LS) is the most common inherited colorectal cancer (CRC) predisposing syndrome. LS is due to germline mutations in the mismatch repair genes MLH1, MSH2, MSH6, and PMS2. Although a higher cancer risk for MLH1 and MSH2 mutation carriers is well-established, there is a wide range in reported CRC penetrance even within the same gene. In this study, the authors attempted to estimate the variation of CRC penetrance within the same gene accounting for sex and continent.The study participants were 5255 families with ≥1 confirmed carrier of a pathogenic variant in a DNA mismatch repair gene who resided in 15 countries in Europe, North America, and Australasia. First- and second-degree relatives were observed from birth to either first cancer diagnosis, first polypectomy or bowel resection, last known age alive, or death. The statistical method used was a segregation analysis fitted by maximum likelihood, which generated age-, sex-, continent-, and gene-specific hazard ratios comparing variant carriers with noncarriers, who were presumed to have the background population incidence of CRC for their age, sex, and continent. The model estimated the polygenic standard deviation (SD) for each continent, a measure of the variation in CRC risk between individual carriers with the same age, sex, and mutated gene. This analysis was repeated restricted to families carrying the c.942+3A>T MSH2 variant.Results confirmed a much higher penetrance of CRC in carriers of mutations in MLH1 and MSH2 than MSH6 and PMS2 across age, sex, and continent. The hazard ratio for 1 polygenic SD was 5.4 (5.4 × increased risk of CRC for each SD increment in polygenic factors) for carriers in Europe, 5.1 for carriers in North America, and 3.5 for carriers in Australasia (P < .0001 for all). Estimates of CRC penetrance varied greatly within a given mutated gene. For example, among male carriers of MLH1 variants in Europe, 23% (95% confidence interval [CI], 6–42) had a <20% penetrance, 14% (95% CI, 10–20) had a 40%–60% penetrance, and 33% (95% CI, 18–51) had a >80% penetrance. Wide variation in CRC penetrance persisted after restricting to families carrying the exact same variant c.942+3A>T in the MSH2 gene: 9%–15% of carriers had <20% penetrance and 33%–45% of carriers had >80% penetrance.The authors concluded that strong and common risk modifiers contribute to variation in CRC penetrance among individuals with LS with the same mutated gene, and even with the same variant in a given gene. Because the polygenic SD represents all factors correlated within families, the observed variation in CRC risk may be attributable to biological, social, and behavioral exposures.CommentIt is well-documented that LS has a highly variable penetrance for colorectal and other cancers depending on the specific mutated gene and the sex of the carrier (Cancer Epidemiol Biomarkers Prev 2017;26:404–412; Gut 2018;67:1306–1316; Hereditary Colorectal Cancer 2018;67–89; Genet Med 2020;22:15–25). Furthermore, studies have shown great variability in CRC penetrance, which is the mean age-specific cumulative cancer risk, even between individuals carrying a mutation in the same gene. This brings up the question of what factors determine these differences. Understanding these factors will help to assess cancer penetrance more accurately. This is key not only for patient counseling, but it also informs preventive and surveillance guidelines.The current study used an innovative approach to estimate the variation in penetrance of CRC between patients with LS attributable to factors other than the mutated mismatch repair gene, by looking at variation among families. This allowed them to assess unmeasured factors, what the authors call “polygenic factors,” the results of which could have meaningful real-world implications. The name “polygenic” suggests more proximate genetic ties than what this measure truly represents. As the authors explain, the polygenic SD captures all variation in risk that is correlated within families but is not attributable to the genetic mutation under study. This can include other heritable factors as well as nonheritable exposures more likely to be shared among family members. For example, a genetic predisposition to obesity would contribute, as would any dietary habits, health behaviors including physical activity, tobacco, and alcohol exposure, environmental exposures, and experiences with the health care system that tend to be more similar within families than between families. In fact, numerous behavioral and environmental risk factors have been identified in prior studies (Hereditary Colorectal Cancer 2018;67–89). Regarding genetic variants as potential risk modifiers of LS genetic defects, a study that evaluated rare high risk single nucleotide polymorphisms in cell cycle regulation and DNA repair genes as potential contributors to the variable penetrance of LS found that LS patients with ≥3 unfavorable genotypes had 4.37 times the risk of CRC as those with no unfavorable genotypes (Carcinogenesis 2013;34:299–306). Interestingly, a recent study found no association between a polygenic risk score associated with CRC in the general population and CRC among patients with LS (JNCI Cancer Spectrum 2021;5:2).Which are genetically driven modifiers and what is the magnitude of their effect and how this compares to the effect of environmental/life-style factors remains largely to be uncovered. In any case, although in this study the risk factors themselves are undefined, the magnitude and strength of association of the polygenic SDs are notable. As an example, MMR mutation carriers who have a polygenic risk level 1 SD above the mean have 5.4 times the CRC incidence of carriers with the mean polygenic risk level, and conversely carriers with a polygenic risk level one SD below the mean have 0.185 times or 18.5% of ([1/5.4] × 100%) the CRC incidence of carriers with the mean polygenic risk level. If we apply this to, for example, male MLH1 mutation carriers under age 40 in Europe, who have an average CRC incidence 37.5 (95% CI, 15.7–89.7) times the population incidence, those with a polygenic risk level 1 SD higher have 202.5 (5.4 × 37.5) times the population incidence, while those with a polygenic risk level one SD lower have 6.9 ([1/5.4] × 37.5) times the population incidence.The study did consider the variant-specific effect, meaning how cancer penetrance compares among individuals having exactly the same mutation in a given gene using the most common variant they had in the data set. They again saw a wide variability among individuals carrying the same mutation around the world, further supporting the notion that the factors mentioned play a bigger role than the actual mutation site.Taken together, these findings support the conclusion that factors other than the LS-causing genetic mutation drive the risk of CRC for individual patients with LS away from the average penetrance (up or down), and to do so these factors must both have a strong association with CRC risk and be common enough in the population to generate such an effect. Thus, the mean cumulative cancer risk as reported in many studies probably only applies to a small percentage of patients. Although this study highlights the magnitude of variation in CRC penetrance attributable to factors other than the mutated mismatch repair gene, it does not identify which candidate familial and environmental exposures are contributing to that variation or to what degree. Further research is needed to identify which risk factors, both heritable and modifiable, are responsible for modifying CRC risk in LS. Better understanding these factors should inform both risk stratification of carriers with LS for more targeted clinical management, as well as meaningful interventions to address modifiable risks at the individual and environmental levels.Meanwhile, because the important variation in risk is related to familial risk factors, paying close attention to family history of cancer could presumably be a good surrogate marker for this risk shared by family members. And probably this is the most important take home message from this study: until we have a proper understanding of the cancer risk modifiers, assessing risk in a specific patient should be made in the wider context of the patient’s family history of cancer. The International Mismatch Repair Consortium. Variation in the risk of colorectal cancer in families with Lynch syndrome: a retrospective cohort study. Lancet Oncol 2021;22:1014–1022. Lynch syndrome (LS) is the most common inherited colorectal cancer (CRC) predisposing syndrome. LS is due to germline mutations in the mismatch repair genes MLH1, MSH2, MSH6, and PMS2. Although a higher cancer risk for MLH1 and MSH2 mutation carriers is well-established, there is a wide range in reported CRC penetrance even within the same gene. In this study, the authors attempted to estimate the variation of CRC penetrance within the same gene accounting for sex and continent. The study participants were 5255 families with ≥1 confirmed carrier of a pathogenic variant in a DNA mismatch repair gene who resided in 15 countries in Europe, North America, and Australasia. First- and second-degree relatives were observed from birth to either first cancer diagnosis, first polypectomy or bowel resection, last known age alive, or death. The statistical method used was a segregation analysis fitted by maximum likelihood, which generated age-, sex-, continent-, and gene-specific hazard ratios comparing variant carriers with noncarriers, who were presumed to have the background population incidence of CRC for their age, sex, and continent. The model estimated the polygenic standard deviation (SD) for each continent, a measure of the variation in CRC risk between individual carriers with the same age, sex, and mutated gene. This analysis was repeated restricted to families carrying the c.942+3A>T MSH2 variant. Results confirmed a much higher penetrance of CRC in carriers of mutations in MLH1 and MSH2 than MSH6 and PMS2 across age, sex, and continent. The hazard ratio for 1 polygenic SD was 5.4 (5.4 × increased risk of CRC for each SD increment in polygenic factors) for carriers in Europe, 5.1 for carriers in North America, and 3.5 for carriers in Australasia (P < .0001 for all). Estimates of CRC penetrance varied greatly within a given mutated gene. For example, among male carriers of MLH1 variants in Europe, 23% (95% confidence interval [CI], 6–42) had a <20% penetrance, 14% (95% CI, 10–20) had a 40%–60% penetrance, and 33% (95% CI, 18–51) had a >80% penetrance. Wide variation in CRC penetrance persisted after restricting to families carrying the exact same variant c.942+3A>T in the MSH2 gene: 9%–15% of carriers had <20% penetrance and 33%–45% of carriers had >80% penetrance. The authors concluded that strong and common risk modifiers contribute to variation in CRC penetrance among individuals with LS with the same mutated gene, and even with the same variant in a given gene. Because the polygenic SD represents all factors correlated within families, the observed variation in CRC risk may be attributable to biological, social, and behavioral exposures. CommentIt is well-documented that LS has a highly variable penetrance for colorectal and other cancers depending on the specific mutated gene and the sex of the carrier (Cancer Epidemiol Biomarkers Prev 2017;26:404–412; Gut 2018;67:1306–1316; Hereditary Colorectal Cancer 2018;67–89; Genet Med 2020;22:15–25). Furthermore, studies have shown great variability in CRC penetrance, which is the mean age-specific cumulative cancer risk, even between individuals carrying a mutation in the same gene. This brings up the question of what factors determine these differences. Understanding these factors will help to assess cancer penetrance more accurately. This is key not only for patient counseling, but it also informs preventive and surveillance guidelines.The current study used an innovative approach to estimate the variation in penetrance of CRC between patients with LS attributable to factors other than the mutated mismatch repair gene, by looking at variation among families. This allowed them to assess unmeasured factors, what the authors call “polygenic factors,” the results of which could have meaningful real-world implications. The name “polygenic” suggests more proximate genetic ties than what this measure truly represents. As the authors explain, the polygenic SD captures all variation in risk that is correlated within families but is not attributable to the genetic mutation under study. This can include other heritable factors as well as nonheritable exposures more likely to be shared among family members. For example, a genetic predisposition to obesity would contribute, as would any dietary habits, health behaviors including physical activity, tobacco, and alcohol exposure, environmental exposures, and experiences with the health care system that tend to be more similar within families than between families. In fact, numerous behavioral and environmental risk factors have been identified in prior studies (Hereditary Colorectal Cancer 2018;67–89). Regarding genetic variants as potential risk modifiers of LS genetic defects, a study that evaluated rare high risk single nucleotide polymorphisms in cell cycle regulation and DNA repair genes as potential contributors to the variable penetrance of LS found that LS patients with ≥3 unfavorable genotypes had 4.37 times the risk of CRC as those with no unfavorable genotypes (Carcinogenesis 2013;34:299–306). Interestingly, a recent study found no association between a polygenic risk score associated with CRC in the general population and CRC among patients with LS (JNCI Cancer Spectrum 2021;5:2).Which are genetically driven modifiers and what is the magnitude of their effect and how this compares to the effect of environmental/life-style factors remains largely to be uncovered. In any case, although in this study the risk factors themselves are undefined, the magnitude and strength of association of the polygenic SDs are notable. As an example, MMR mutation carriers who have a polygenic risk level 1 SD above the mean have 5.4 times the CRC incidence of carriers with the mean polygenic risk level, and conversely carriers with a polygenic risk level one SD below the mean have 0.185 times or 18.5% of ([1/5.4] × 100%) the CRC incidence of carriers with the mean polygenic risk level. If we apply this to, for example, male MLH1 mutation carriers under age 40 in Europe, who have an average CRC incidence 37.5 (95% CI, 15.7–89.7) times the population incidence, those with a polygenic risk level 1 SD higher have 202.5 (5.4 × 37.5) times the population incidence, while those with a polygenic risk level one SD lower have 6.9 ([1/5.4] × 37.5) times the population incidence.The study did consider the variant-specific effect, meaning how cancer penetrance compares among individuals having exactly the same mutation in a given gene using the most common variant they had in the data set. They again saw a wide variability among individuals carrying the same mutation around the world, further supporting the notion that the factors mentioned play a bigger role than the actual mutation site.Taken together, these findings support the conclusion that factors other than the LS-causing genetic mutation drive the risk of CRC for individual patients with LS away from the average penetrance (up or down), and to do so these factors must both have a strong association with CRC risk and be common enough in the population to generate such an effect. Thus, the mean cumulative cancer risk as reported in many studies probably only applies to a small percentage of patients. Although this study highlights the magnitude of variation in CRC penetrance attributable to factors other than the mutated mismatch repair gene, it does not identify which candidate familial and environmental exposures are contributing to that variation or to what degree. Further research is needed to identify which risk factors, both heritable and modifiable, are responsible for modifying CRC risk in LS. Better understanding these factors should inform both risk stratification of carriers with LS for more targeted clinical management, as well as meaningful interventions to address modifiable risks at the individual and environmental levels.Meanwhile, because the important variation in risk is related to familial risk factors, paying close attention to family history of cancer could presumably be a good surrogate marker for this risk shared by family members. And probably this is the most important take home message from this study: until we have a proper understanding of the cancer risk modifiers, assessing risk in a specific patient should be made in the wider context of the patient’s family history of cancer. It is well-documented that LS has a highly variable penetrance for colorectal and other cancers depending on the specific mutated gene and the sex of the carrier (Cancer Epidemiol Biomarkers Prev 2017;26:404–412; Gut 2018;67:1306–1316; Hereditary Colorectal Cancer 2018;67–89; Genet Med 2020;22:15–25). Furthermore, studies have shown great variability in CRC penetrance, which is the mean age-specific cumulative cancer risk, even between individuals carrying a mutation in the same gene. This brings up the question of what factors determine these differences. Understanding these factors will help to assess cancer penetrance more accurately. This is key not only for patient counseling, but it also informs preventive and surveillance guidelines. The current study used an innovative approach to estimate the variation in penetrance of CRC between patients with LS attributable to factors other than the mutated mismatch repair gene, by looking at variation among families. This allowed them to assess unmeasured factors, what the authors call “polygenic factors,” the results of which could have meaningful real-world implications. The name “polygenic” suggests more proximate genetic ties than what this measure truly represents. As the authors explain, the polygenic SD captures all variation in risk that is correlated within families but is not attributable to the genetic mutation under study. This can include other heritable factors as well as nonheritable exposures more likely to be shared among family members. For example, a genetic predisposition to obesity would contribute, as would any dietary habits, health behaviors including physical activity, tobacco, and alcohol exposure, environmental exposures, and experiences with the health care system that tend to be more similar within families than between families. In fact, numerous behavioral and environmental risk factors have been identified in prior studies (Hereditary Colorectal Cancer 2018;67–89). Regarding genetic variants as potential risk modifiers of LS genetic defects, a study that evaluated rare high risk single nucleotide polymorphisms in cell cycle regulation and DNA repair genes as potential contributors to the variable penetrance of LS found that LS patients with ≥3 unfavorable genotypes had 4.37 times the risk of CRC as those with no unfavorable genotypes (Carcinogenesis 2013;34:299–306). Interestingly, a recent study found no association between a polygenic risk score associated with CRC in the general population and CRC among patients with LS (JNCI Cancer Spectrum 2021;5:2). Which are genetically driven modifiers and what is the magnitude of their effect and how this compares to the effect of environmental/life-style factors remains largely to be uncovered. In any case, although in this study the risk factors themselves are undefined, the magnitude and strength of association of the polygenic SDs are notable. As an example, MMR mutation carriers who have a polygenic risk level 1 SD above the mean have 5.4 times the CRC incidence of carriers with the mean polygenic risk level, and conversely carriers with a polygenic risk level one SD below the mean have 0.185 times or 18.5% of ([1/5.4] × 100%) the CRC incidence of carriers with the mean polygenic risk level. If we apply this to, for example, male MLH1 mutation carriers under age 40 in Europe, who have an average CRC incidence 37.5 (95% CI, 15.7–89.7) times the population incidence, those with a polygenic risk level 1 SD higher have 202.5 (5.4 × 37.5) times the population incidence, while those with a polygenic risk level one SD lower have 6.9 ([1/5.4] × 37.5) times the population incidence. The study did consider the variant-specific effect, meaning how cancer penetrance compares among individuals having exactly the same mutation in a given gene using the most common variant they had in the data set. They again saw a wide variability among individuals carrying the same mutation around the world, further supporting the notion that the factors mentioned play a bigger role than the actual mutation site. Taken together, these findings support the conclusion that factors other than the LS-causing genetic mutation drive the risk of CRC for individual patients with LS away from the average penetrance (up or down), and to do so these factors must both have a strong association with CRC risk and be common enough in the population to generate such an effect. Thus, the mean cumulative cancer risk as reported in many studies probably only applies to a small percentage of patients. Although this study highlights the magnitude of variation in CRC penetrance attributable to factors other than the mutated mismatch repair gene, it does not identify which candidate familial and environmental exposures are contributing to that variation or to what degree. Further research is needed to identify which risk factors, both heritable and modifiable, are responsible for modifying CRC risk in LS. Better understanding these factors should inform both risk stratification of carriers with LS for more targeted clinical management, as well as meaningful interventions to address modifiable risks at the individual and environmental levels. Meanwhile, because the important variation in risk is related to familial risk factors, paying close attention to family history of cancer could presumably be a good surrogate marker for this risk shared by family members. And probably this is the most important take home message from this study: until we have a proper understanding of the cancer risk modifiers, assessing risk in a specific patient should be made in the wider context of the patient’s family history of cancer.