Sporadic deficient mismatch repair in colorectal cancer increases the risk for non‐colorectal malignancy: A European multicenter cohort study

Disparities between tumors arising via different sporadic carcinogenetic pathways have not been studied systematically. This retrospective multicenter cohort study evaluated the differences in the risk for non‐colorectal malignancy between sporadic colorectal cancer (CRC) patients from different DNA mismatch repair status.


Conclusion:
In this retrospective European multicenter cohort study, patients with sporadic dMMR CRC had a higher risk for non-colorectal malignancy than those with pMMR CRC.These findings indicate the need for further studies to establish the need for and design of surveillance strategies for patients with dMMR CRC.

| INTRODUCTION
2][3] Some dietary factors, obesity, inflammatory bowel disease, and smoking are known risk factors. 1 Hereditary diseases such as familial adenomatous polyposis and Lynch Syndrome are associated with genetic predisposition to cancer and constitute about 5%-10% of all CRC cases. 1 There are two main pathways leading to the transformation of normal colonic epithelium to histologically distinct precursor adenomatous lesions, and ultimately to CRC. 4 The first one, known as the chromosomal instability (CIN) pathway, is often associated with mutational inactivation of the APC-gene and found in approximately 85% of sporadic CRCs.The second pathway, known as the deficient mismatch repair (dMMR) pathway, involves defective/inactivation of DNA mismatch repair (MMR) genes that are responsible for correcting DNA replication errors and is found in approximately 15%-20% of all CRC cases.Inherited mutations in the MMR genes MLH1, MSH2, MSH6, and PMS2 are the main causes of Lynch Syndrome constituting 3%-5% of all dMMR cases.In contrast, sporadic dMMR tumors are mainly caused by epigenetic silencing of the MLH1 gene through hypermethylation of the MLH1 promoter and is found in approximately 15% of CRC cases. 5,6 recent decades, important advances have been made in our understanding of familial CRC and its clinical implications for diagnosis, prevention, and therapy.This applies specifically to Lynch-associated carcinogenesis, with its markedly increased lifetime risk for CRC and endometrial cancer, as well as cancers of the ovary, stomach, urogenital tract, small bowel, pancreas, and tumors from the biliary tract. 7,8However, there is little information on whether sporadic dMMR is associated with a higher risk for cancer in other organs before or after the CRC diagnosis.Previous reports have revealed dMMR in approximately 15%-30% of gastric cancer cases, with only a fraction having a hereditary background, as well as approximately 20% of endometrial cancer cases. 9,10Interestingly, endometrial cancer patients with dMMR tumors have better diseasefree and disease-specific survival than patients with proficient MMR (pMMR) tumors. 10Malignancies coming from the upper gastrointestinal tract reveal dMMR status only in a small fraction, and studies assessing the prognosis among these patients are lacking today. 11,12Despite these findings, there has been little change in the algorithms for cancer surveillance and prevention strategies in patients with sporadic dMMR CRC.This European multicenter cohort study aimed to retrospectively examine the association between sporadic dMMR status and the risk of non-colorectal malignancy among patients with known CRC.To adjust for variation in observation time, the incidence rate ratio (IRR) for non-colorectal malignancy over time, comparing patients with sporadic dMMR and pMMR CRC, was presented.
Our study was approved by the Regional Ethics Review Board,

| Study population
The present study included 1116 CRC patients treated between October 17, 1996, and March 31, 2009, at  were considered positive for Lynch Syndrome.Patients with loss of MLH1 were further analyzed for BRAF-mutation status.MLH1-loss cases with wild-type BRAF V600 were also considered positive for Lynch Syndrome.Patients with loss of MLH1 but with detected BRAF V600 mutations were further tested for hypermethylation of the MLH1 promoter.Cases with positive hypermethylation were considered to represent sporadic dMMR, and cases with no hypermethylation were classified as Lynch Syndrome (Figure 1).
Of the 2107 patients, 141 had inconclusive immunohistochemistry or missing data, and 67 had inconclusive BRAF status and were thus excluded from further analysis (Figure 2).The remaining patients were matched against the National Cancer Registries in Sweden and Finland, while medical file scrutiny was performed for the Czech population to identify non-colorectal malignancies within 20 years before or after CRC diagnosis.From this population, 62 patients with a second or recurrent CRC and 40 cases registered as non-melanoma skin cancer were excluded (Figure 2).A further 44 patients with BRAF V600 wild-type and 47 cases with BRAF V600E mutation and no hypermethylation of the MLH1 promoter were categorized as positive for Lynch Syndrome using the algorithm shown in Figure 1 and were thus excluded (Figure 2).

| Determination of MMR status
The proficiency or deficiency of the MMR system in tumor tissues was determined by IHC analysis of four protein products of genes involved in the MMR system.5][16] All specimens were assessed under the supervision of a pathologist with special interest in CRCl. 17

| BRAF tumor tissue analysis
Formalin-fixed, paraffin-embedded tumor tissues were collected at the departments involved.In the Swedish cohort, BRAF status was determined using a BRAF V600E mutation-specific TaqMan allelic discrimination assay. 18,19In the Czech Republic cohort, BRAF F I G U R E 1 Algorithm for selection of sporadic dMMR CRC cases.CRC, colorectal cancer.mutation testing was performed using Cobas ® BRAF V600 Mutation Test.In the Finnish cohort, BRAF mutation status was analyzed using a mutation-specific (V600E) monoclonal antibody in the same immunohistochemical setup used for MMR proteins.

| Hypermethylation analysis
DNA was extracted from FFPE tumor tissue (5 × 10 µm) using the QIAamp ® DNA FFPE Tissue Kit.The samples were treated with 40 µL Proteinase K and 4 µL RNAse.DNA concentration was measured using the Qubit TM dsDNA BR kit calculated with the Spectrophotometer/Fluorometer DS-11FX+ (DeNovix).Methylationspecific PCR was performed using a PyroMark PCR Kit.The PCR was run in a Veriti Thermal Cycler (ABI).Data was analyzed using PyroMark Q24 Advanced 3.0.0Software (Qiagen), and methylation in percentage (C m /(C m +C)) of each CpG site was calculated.

| Statistical analysis
Statistical analyses were conducted using IBM SPSS version 26.To adjust for variations in observation times, Poisson regression analysis was performed to analyze the relationships between the IRR for noncolorectal malignancy and age, sex, MMR status, site, and stage (I-IV).
The observation period was from 20 years before CRC diagnosis until death from any cause or end of follow-up of individuals still alive.For validation purposes, standardized incidence ratios (SIRs) were calculated for each of the Swedish cases reported (patients with pMMR and dMMR separately) and adjusted for sex, year (in 1-year classes), and age (in 5-year classes).Incidental cancer rates were calculated based on data from the Swedish Cancer Registry with Sweden as the reference population.Incidental cancer rates were calculated by multiplying the number of person-years for each sex, year, and age group by the corresponding cancer incidence rates in the reference population.SIR with 95% confidence interval (CI) was F I G U R E 2 Flow diagrams of the study population.calculated as the ratio of the number of observed and expected cases, assuming that the observed number of cases followed a Poisson distribution.

| Basic characteristics
In all, 1706 CRC, of whom 819 were female (48%), met the inclusion criteria.Among the cases included, 560 patients (33%) were diagnosed with a rectal tumor.The median age at diagnosis was 67 (interquartile range [IQR]: 60-75) years, and the patients were dichotomized at 70 years.Sporadic dMMR status was observed in 186 patients (11%).
A total of 324 patients (19%) were diagnosed with noncolorectal malignancies within 20 years before or after the index CRC diagnosis.A sporadic dMMR status was detected in 60 patients (18%) (p = 0.005) (Table 1).Non-colorectal malignancies are shown in Table 2.
IRR for the occurrence of non-colorectal malignancy within 20 years before and after CRC diagnosis in uni-and multivariable analysis.
IRR for the occurrence of non-colorectal malignancy within 20 years before CRC diagnosis in uni-and multivariable analysis.
IRR for the occurrence of non-colorectal malignancy during the observation period after CRC diagnosis in uni-and multivariable analysis.
SIRs, patients with dMMR CRC tumors, adjusted for year, sex, and age.
For the Swedish study population, calculation of SIR for validation purpose revealed that the expected and observed cancer cases among patients with sporadic dMMR tumors showed a statistically significant higher incidence of other non-colorectal malignancies among these patients.The incidence of other noncolorectal tumors among sporadic dMMR patients was increased (SIR = 1.23, 95% CI = 1.00-1.49,p = 0.042), especially among female patients (SIR = 1.10, 95% CI = 1.04-1.88,p = 0.020) (Table 6).
SIRs, patients with a pMMR CRC tumor, adjusted for year, sex, and age.The same calculation for the Swedish population revealed that the expected and observed cancer cases among patients with pMMR tumors showed a statistically significant lower incidence of other non-colorectal malignancies.The incidence of other non-colorectal tumors among pMMR patients was increased among these patients (SIR = 0.71, 95% CI = 0.71-0.95,p = 0.008), especially among male patients (SIR = 0.75, 95% CI = 0.61-0.92,p = 0.008) (Table 7).

| DISCUSSION
In this multicenter study, including a large cohort from three European medical centers, a significant association between sporadic dMMR CRC and the risk for non-colorectal malignancy before and after the diagnosis of CRC was observed.This suggests that the genetic background i.e., impaired ability to correct errors in DNA replication, may be associated with the development of tumors elsewhere, not only among the familial type but also among sporadic dMMR CRC cases.This relationship was significant in both uni-and multivariable analyses when estimating the IRR for non-colorectal malignancies.The higher IRR in sporadic dMMR cases was more obvious in the time period after CRC diagnosis.This higher risk for non-CRC in sporadic dMMR CRC cases was also seen in a validation analysis, where the number of observed cases was significantly higher than that expected among these patients.To the best of our knowledge, this is the first study to investigate the association between sporadic dMMR CRC and the risk for non-colorectal malignancy.Our results indicate that further research into the need for surveillance strategies, not only in hereditary cases, but also in sporadic cases, is warranted.Whether surveillance strategies for patients with sporadic dMMR CRC should be the same as those for hereditary dMMR cases or whether a tailored surveillance program would be more suitable is a topic of further research.
The association between sporadic dMMR status and a higher risk for non-colorectal malignancy may arise from mechanisms similar to those of familial Lynch Syndrome i.e., aberrancy in systems ensuring the fidelity of DNA replication. 20The theory that cancer develops from increased mutation rates due to uncorrected replication errors is based on previous findings revealing that in individuals with aberrant MMR activity, the accumulation of different mutations can result in the silencing of genes that normally inhibit tumorigenesis, such as tumor suppressor genes. 21In contrast, oncogenes that are normally turned off in healthy tissues may become activated or acquire a tumor-promoting function because of the higher mutation rate. 22The main idea of this study was that the same mechanism for carcinogenesis, i.e. impaired ability to correct errors in DNA replication, applies to both familial and sporadic dMMR cases.
In the study population, the defect mismatch repair process, resulting in the accumulation of insertions or deletions that secondarily predispose to malignancy, was associated with older age as an independent variable.This phenomenon is probably the result of a synergic interaction between age and dMMR rather than a strictly dependent relationship between age and carcinogenesis.In other words, increasing age also increases the risk of MLH1 promoter hypermethylation, resulting in dMMR carcinogenesis. 20It is wellknown that the incidence of cancer increases with age in both humans and animals.Different patterns of age-related distribution of tumors in different organs and tissues have been observed.Aging seems to increase the susceptibility of various tissues to the initiation of carcinogenesis and usually facilitates the promotion and progression of carcinogenesis. 23Aging may also predispose to cancer through mechanism such as tissue accumulation of cells in the late stages of carcinogenesis, alterations in homeostasis, alterations in the immune and endocrine systems, and telomere instability. 24For all these reasons, the effect of age on tumorigenesis cannot be restricted to dMMR cases, and the same should be the case in pMMR CRC, even though dMMR patients appear to develop malignancies at a slightly higher age than pMMR patients. 25In contrast, reports have revealed that there are no significant differences in molecular and clinicopathologic patterns between different CRC age groups, and that cases with sporadic dMMR CRC show similar molecular alterations in different age groups. 26This relationship between age and dMMR tumorigenesis seems to be much more complex since MMR plays a critical role in preserving the T A B L E 2 Non-colorectal malignancy in CRC patients with dMMR and those with pMMR cancer.T A B L E 3 IRR for the occurrence of a non-colorectal malignancy within 20 years before and after diagnosis between sporadic dMMR and pMMR cases in the uni-and multivariable models.integrity of the genome in virtually all organisms. 27Studies have suggested that the accumulation of DNA damage and the resulting genomic instability are not only risk factors for carcinogenesis, but also contribute to aging itself. 20,28This hand-in-hand relationship between dMMR and age is supported by studies on mice, suggesting that the capacity for MMR diminishes with age and that MMR can process certain forms of oxidative damage to DNA, which is thought to accumulate with age. 20,28broad spectrum of malignancies was observed in this study population.We excluded all non-melanoma skin neoplasms because these cancers are relatively common and have different epigenetic backgrounds due to UV radiation.Furthermore, since they are not usually fatal, the registration of non-melanoma skin cancer (ICD-10 C44) is probably not as complete as in other forms of cancer. 29This wide spectrum of non-colorectal malignancies is also seen in the majority of studies dealing with hereditary Lynch Syndrome, but in this study, some hematological malignancies were observed that have not been reported in studies on Lynch Syndrome. 30,31One possible explanation for this discrepancy is that hematologic malignancies are diseases that mainly affect older individuals. 32other difference is that endometrial cancer was less frequent in our study population than in studies on hereditary cases. 33This may be explained by the fact that the majority of Lynch endometrial cancers carry a deleterious germline mutation in MSH2 genes in contrast to sporadic dMMR cases where the majority carry a mutation in the MLH1 gene. 34As shown in previous reports, mutations in different genes lead to different penetrance and patterns of cancer expression. 35There are significant discrepancies in cancer risk estimations in Lynch Syndrome patients, with some studies stating that the "true" risk may never be established.The main reason seems to be that current reports are biased because T A B L E 5 Incidence rate ratio for other non-colorectal cancer after the diagnosis of CRC between sporadic dMMR and pMMR cases in the uni-and multivariable models.estimations are based on Lynch Syndrome registers that include Lynch families with highly penetrant alleles. 36The heterogeneity of allelic aberrations may lie behind the different etiologies of cancer syndromes encompassing a spectrum of similar clinical presentations and genetic profiles, such as Lynch-like Syndrome and familial CRC type X. 37,38 Heterogenicity could explain the differences observed between the results of previous studies on Lynch Syndrome and those in this study based exclusively on sporadic dMMR CRC, a subject that has seldom been investigated. 39nsiderable differences exist between colonic and rectal tumors, especially regarding tumor biology, relapse patterns, and treatment modalities. 40A limitation of this study is the inherent bias of the retrospective methodology.The strength of this study to determine the incidence of other non-colorectal malignancies in patients with sporadic dMMR CRC, is the validation results comparing expected and observed cancer cases separately for dMMR and pMMR cases, and that the study cohort consisted of populations from three different countries and different categories of hospitals, increasing the external validity and generalizability of the results. 41Furthermore, the unique possibility of using national cancer registry data for the Scandinavian population increasing the reliability and trustworthiness of our results. 42

| CONCLUSION
Patients with sporadic dMMR CRC have a higher IRR for other noncolorectal malignancies.Further studies are needed to determine the need for and design of surveillance screening programs for these patients to improve cancer diagnosis and clinical outcomes.
one university hospital and two regional hospitals in Västerbotten County, Sweden, and 577 CRC patients treated at Helsinki University Hospital, Finland, between September 1, 1998, and December 31, 2005.A further 414 colon cancer (CC) patients treated at the University Hospital of Pilsen, Czech Republic, between January 1, 2018, and December 31, 2019, were added to the Scandinavian cohorts.Inclusion criteria were: (i) histologically confirmed CRC; (ii) stage confirmed and transformed to the American Joint Committee on Cancer (AJCC) TNM classification; and (iii) available clinical and pathological data regarding the course of the disease and demographic data. 13Patients younger than 18 years were excluded.The following data were collected from the hospital records: age, sex, site and stage at diagnosis.All cases with conclusive immunohistochemistry (IHC) were evaluated for sporadic MMR-status as described below.Cases showing protein loss of MSH2 or MSH6 or isolated loss of PMS2

T A B L E 1
Non-colorectal cancer figures according to clinical characteristics and tumor pathology: entire study population.
IRR of non-colorectal malignancy within 20 years before CRC diagnosis between sporadic dMMR and pMMR cases in the Uni-and multivariable analyses.
aThe referance variable.T A B L E 4Abbreviations: CI, confidence Interval; CRC, colorectal cancer; IRR, incidence rate ratio.a The referance variable.
Standardized incidence ratios (SIR) for patients with pMMR tumor, adjusted for year, sex, and age.
Abbreviations: CI, confidence Interval; CRC, colorectal cancer; IRR, incidence rate ratio.aThereference variable.T A B L E 6 Standardized incidence ratios (SIR) for patients with dMMR tumor, adjusted for year, sex, and age.T A B L E 7