A Previously Unrecognized Molecular Landscape of Lynch Syndrome in the Mexican Population

Lynch syndrome (LS) is the main hereditary colorectal cancer syndrome. There have been few reports regarding the clinical and molecular characteristics of LS patients in Latin America; this is particularly true in the Mexican population, where no information is available. The present study aims to describe the clinical and molecular spectrum of variants in a cohort of patients diagnosed with LS in Mexico. We present a retrospective analysis of 412 patients with suspected LS, whose main site of cancer diagnosis was the colon (58.25%), followed by the endometrium (18.93%). Next-generation sequencing analysis, with an extensive multigene panel, showed that 27.1% (112/414) had a variant in one of the genes of the mismatch repair pathway (MMR); 30.4% (126/414) had a variant in non-MMR genes such as CHEK2, APC, MUTYH, BRCA1, and BRCA2; and 42.5% (176/414) had no genetic variants. Most of the variants were found in MLH1. Pathogenic variants (PVs) in MMR genes were identified in 65.7% (96/146) of the total PVs, and 34.24% (45/146) were in non-MMR genes. Molecular and clinical characterization of patients with LS in specific populations allowed personalized follow-up, with the option for targeted treatment with immune checkpoint inhibitors and the development of public health policies. Moreover, such characterization allows for family cascade testing and consequent prevention strategies.


Introduction
Lynch syndrome (LS) is the most common hereditary syndrome of colorectal cancer. It is also associated with predisposition to several extracolonic neoplasms, with endometrial carcinoma being the most common, and an increased risk of cancer of the ovary, small intestine, stomach, urinary tract, pancreas, and brain [1]. It is an autosomal dominant syndrome caused by germline mutations in one of the genes of the mismatch repair (MMR) pathway: MLH1, MSH2, MSH6, and PMS2 [2,3]. The germline pathogenic variant (PV), combined with a second somatic acquired pathogenic variant in the wildtype allele, results in complete loss of the MMR pathway function in affected cells [4], resulting in an inability to repair mismatch errors and microsatellite instability (MSI), which is present in 95% of all tumors associated with LS [4,5]. Part of the process of identifying individuals at risk of LS relies on the review of the individual and family history of cancer, where the Amsterdam and Bethesda criteria are evaluated [3]. Two types of tests are used to screen for LS in

Results
A total of 412 patients with suspected LS were recruited from 2016 to 2021, through the Hereditary Cancer Clinic of the National Cancer Institute in Mexico City, Mexico. Patients underwent a cancer risk assessment by medical geneticists, with the corresponding preand post-test genetic counseling. LS patients were selected using Amsterdam and Bethesda criteria, as well as the NCCN guidelines. For the purposes of describing our population, we grouped patients under the term "suspected Lynch syndrome" for those who met the clinical criteria and underwent genetic testing. Based on molecular analysis results, we further divided them into "MMR", comprising those patients with pathogenic variants in MMR genes, and therefore with LS; "no variant identified", for those patients with a negative result; and, "non-MMR", comprising those patients with a variant in a gene other than an MMR gene.
Of the 412 patients, 71.84% (296/412) were female and 28.16% (116/412) were male. The mean age at diagnosis was 40.82 years, with a minimum age of 16 and maximum age of 78 years (SD 12.09). The mean current age was 44.27 years (SD 12.87). A positive cancer family history was present in 47.33% (195/412) of patients.
The distribution according to the first cancer diagnosis and sex in the patients studied is shown in Figure 1. In women, the most common site corresponded to colorectal cancer, in 45.94% of cases (136/296). Of these, 36 (Table 3).
By analyzing the results of the multigene panel performed on the patients with LS, 42.5% (176/414) had no variant identified ( Figure 2). To complement the molecular approach, because of important clinical LS suspicion, MLH1 and MSH2 multiplex ligationdependent probe amplification (MLPA) was performed on 36 of the 412 patients (8.73%). The result was negative for all of them, since no deletions or duplications of exons of the explored genes were identified.
Of the patients with no variant identified, 67.43% (118/175) were women and 32.57% (57/175) were men; the mean age at diagnosis was 40.56 years, with a minimum age of 17 years and a maximum of 78 years. The mean current age was 40.656 years. A total of 57.06% (101/175) were alive without disease at the end of this study. Clinical stage IV disease was present in 19.77% (35/175) of patients. The remaining socio-demographic characteristics of this group are shown in Table 1. By analyzing the results of the multigene panel performed on the patients with LS, 42.5% (176/414) had no variant identified ( Figure 2). To complement the molecular approach, because of important clinical LS suspicion, MLH1 and MSH2 multiplex ligationdependent probe amplification (MLPA) was performed on 36 of the 412 patients (8.73%). The result was negative for all of them, since no deletions or duplications of exons of the explored genes were identified.   When a variant was identified in one of the MMR genes, the majority (58.56%, 65/111) corresponded to colorectal cancer, followed by endometrial cancer (18.02%, 20/111), ovarian cancer (10.81%, 12/111), breast cancer (4.50%, 5/111), and gastric cancer (1.80%, 2/111). In addition, a variant was identified for cervical, renal, skin, thyroid, and pancreatic cancers ( Table 2).
According to the ClinVar classification of the variants identified in the MMR genes, a total of 59 variants were obtained in MLH1, most of which were pathogenic (46 variants) followed by variants that were not reported (9), likely pathogenic variants (2), variants with a conflicting interpretation of pathogenicity, and one variant of uncertain significance (VUS). In MSH2, 24 variants were reported, where the majority were pathogenic (15), followed by those that were not reported (4), VUS (3), those with conflicting interpretation (1) and pathogenic/likely pathogenic variants (1). In MSH6, 21 variants were identified, where the majority were VUS (9), followed by variants that were not reported (6), were pathogenic (5), had conflicting interpretation (2), were pathogenic/likely pathogenic (1), and were likely pathogenic variants (1). In PMS2, eight variants were identified: three pathogenic, two likely pathogenic, two VUS, and one that was not reported (Figure 3).
With regard to the PV identified, considering those reported in ClinVar and those not previously reported, a total of 96 (96/146) patients had a PV in one of the MMR genes; of these, 69.79% (67/96) had a first-degree cancer family history; 70.83% (69/96) were women, and 29.17% (29/96) were men. The mean age at diagnosis was 41.21 years with a minimum age of 19 and a maximum of 66 years (SD 10.39). In women, the mean age at diagnosis was 41.76 years, with a minimum age of 20 years and a maximum age of 66 years (SD 10.05). In men, the mean age was 39.89 years (SD 9.48), with a minimum age of 19 years and a maximum of 66 years. Regarding clinical stage, the majority of carriers (24.02%, 23/96) presented at stage IIA, followed by stage IIIB in 11.45% (11/96) of the cases (Table 4).
pretation (1) and pathogenic/likely pathogenic variants (1). In MSH6, 24 variants were identified, where the majority were VUS (9), followed by variants that were not reported (6), were pathogenic (5), had conflicting interpretation (2), were pathogenic/likely pathogenic (1), and were likely pathogenic variants (1). In PMS2, eight variants were identified: three pathogenic, two likely pathogenic, two VUS, and one that was not reported ( Figure  3). With regard to the PV identified, considering those reported in ClinVar and those not previously reported, a total of 96 (96/146) patients had a PV in one of the MMR genes; of these, 69.79% (67/96) had a first-degree cancer family history; 70.83% (69/96) were women, and 29.17% (29/96) were men. The mean age at diagnosis was 41.21 years with a minimum age of 19 and a maximum of 66 years (SD 10.39). In women, the mean age at diagnosis was 41.76 years, with a minimum age of 20 years and a maximum age of 66  The site of the first cancer diagnosis in PV carriers was, for women and men, colorectal tumors (55.89% and 89.29%, respectively), with a predominance of the adenocarcinoma subtype ( Figure 4 and Table 5). In the case of women, the second most common cancer incidence was endometrial cancer, in 23.53% of the cases, all of which were adenocarcinomas. For ovarian cancer (11.76% of the cases), most of them had endometrioid adenocarcinoma histology. Breast cancer represented 5.88% of the PV cases, and they all had invasive ductal carcinoma histology (Figure 4).
Most of the PVs identified were in MLH1 (58 PV), and the majority were identified in women (42 women vs. 15 men) ( Figure 5). Forty-two of the MLH1 variants were identified in patients diagnosed with colorectal cancer, eight with endometrial cancer, three with ovarian cancer, two with breast cancer, one with cervical cancer, and one with skin cancer (Table 4). Following this, 20 PVs were identified in MSH2, 13 identified in women, and 7 in men ( Figure 6). The majority were identified in patients diagnosed with colorectal cancer, followed by endometrial cancer (Table 5). In MSH6, 14 PV were identified; 9 identified in women, and 4 in men. Of these, tumor diagnosis corresponded to four PVs in colorectal cancer, four PVs in endometrial cancer, three PVs in ovarian, one PV in breast cancer, and one PV in thyroid cancer. Finally, for PMS2, six PVs were identified; four in women and two in men ( Figure 5). Three of these were in colorectal cancer, one in breast cancer, and one each in gastric and renal cancer (Table 5). We report four patients with more than one variant in the MMR genes, and seven patients with an MMR gene variant plus a variant in a non-MMR gene, either pathogenic or VUS. It is worth mentioning that there were two patients with a VUS in an MMR gene alongside a pathogenic variant in another gene (Table A1, Appendix A).
The most represented type of PV in the cohort was nonsense mutations, in 31.63% of cases, followed by frameshift deletions (17.35%), frameshift duplications (14.29%), whole exon deletions (12.24%) missense (11.22%), splicing alterations (4.08%), in-frame deletions (1.02%), and indels (1.02%).     in patients diagnosed with colorectal cancer, eight with endometrial cancer, three with ovarian cancer, two with breast cancer, one with cervical cancer, and one with skin cancer (Table 4). Following this, 20 PVs were identified in MSH2, 13 identified in women, and 7 in men ( Figure 6). The majority were identified in patients diagnosed with colorectal cancer, followed by endometrial cancer (Table 5). In MSH6, 14 PV were identified; 9 identified in women, and 4 in men. Of these, tumor diagnosis corresponded to four PVs in colorectal cancer, four PVs in endometrial cancer, three PVs in ovarian, one PV in breast cancer, and one PV in thyroid cancer. Finally, for PMS2, six PVs were identified; four in women and two in men ( Figure 5). Three of these were in colorectal cancer, one in breast cancer, and one each in gastric and renal cancer (Table 5). We report four patients with more than one variant in the MMR genes, and seven patients with an MMR gene variant plus a variant in a non-MMR gene, either pathogenic or VUS. It is worth mentioning that there were two patients with a VUS in an MMR gene alongside a pathogenic variant in another gene (Table A1, Appendix A).    The most represented type of PV in the cohort was nonsense mutations, in 31.63% o cases, followed by frameshift deletions (17.35%), frameshift duplications (14.29%), whole  (Table 1). When a variant in non-MMR genes was identified, 61.11% (77/126) corresponded to colorectal cancer, followed by endometrial cancer in 17.46% (22/126), ovarian cancer in 10.32% (13/126), breast cancer in 7.14% (9/126), and gastric and vaginal cancer in 1.59% (2/126 each); in addition, a variant was identified in cervical cancer ( Table 2).
Regarding the variants identified in non-MMR genes, 69.23% (108 variants) were VUS. The genes that presented more than 10 variants were CHEK2, APC, FANCA, MUTYH, ATM, BRCA1 and POLE (Figure 7). We report 23 patients who presented more than one germline variant in different genes. Of these, 12 patients had two VUS, one with two PVs, one with one pathogenic variant with two VUS, and one with one pathogenic variant with three VUS (Table A1, Appendix A).
Regarding the variants identified in non-MMR genes, 69.23% (108 variants) were VUS. The genes that presented more than 10 variants were CHEK2, APC, FANCA, MUTYH, ATM, BRCA1 and POLE (Figure 7). We report 23 patients who presented more than one germline variant in different genes. Of these, 12 patients had two VUS, one with two PVs, one with one pathogenic variant with two VUS, and one with one pathogenic variant with three VUS (Table A1, Appendix A).

Discussion
Through comprehensive hereditary cancer risk assessment, we identified a PV frequency of 28.15% (116/412) in cancer susceptibility genes, in a population of 412 Mexican patients with LS, through NGS. In terms of PVs, 71 were identified in MMR genes, which represents a mutational frequency of 17.23% (71/412).
Regarding the tumor site diagnosis for PV carriers, most of them presented a LSspectrum cancer, such as colorectal, endometrial, and ovarian cancer. For women, breast cancer was found in fourth place in terms of tumor frequency. The assessment of breast cancer risk for LS is controversial [17,18]. Recent studies have suggested that breast cancer may be included in the cancer risk associated with LS, where a significant increase in breast cancer incidence rates have been found with an earlier age of diagnosis compared to the general population. Nevertheless, other studies show no association and, therefore, do not recommend increased breast cancer surveillance for patients with LS [17][18][19]. This study adds to the evidence to indicate a broader spectrum of associated tumors, where breast cancer represented 5.88% of the MMR pathogenic variant cases, contributing to the comparison and reflection in other populations.
The variants in our studied population were found to be dominated by those identified in MLH1 (52.67%), followed by MSH2 (21.42%), MSH6 (18.75%), and PMS2 (7.14%). Comparing these results with previous reports in Latin America, our results concur with those of Rossi et al. (2017) and DellaValle et al. (2019), where they report a higher prevalence of MLH1 at 53.9% and 43%, respectively, followed by MSH2 at 32.4% and 37%, respectively. Our results differ with those previously reported regarding the percentage of variants found in MSH6, where the prevalence cited in the literature ranges from 7 to 9% [13,15,20]; we found a higher percentage of variants in our report. DellaValle identified variants in MLH1 in 39% of women and 50% of men; MSH2 in 37% of women and men; MSH6 in 13% of women and 3% of men; and PMS2 in 11% of women and 8% of men. In our cohort, a higher percentage of variants was observed in women in all of the genes of the MMR pathway; however, this may be due to the fact that 70% of the cohort were women.
The most frequent type of PV identified was the frameshift type, considering insertions, duplications, and deletions; this was followed by nonsense variants, large deletions, missense variants, and splice-site alterations. These data are comparable with those reported by Vaccaro et al. (2016), who reported that, in their cohort of 98 PVs, frameshift alterations were the most common, followed by nonsense mutations and large deletions; our data differ in the frequency of missense variants, where they report a lower frequency than we identified in our patients.
A recurrent pathogenic variant in MLH1 was identified in 13 unrelated patients of the cohort. The variant MLH1 c.676C > T creates a premature termination codon. It was first reported in three siblings from a Turkish family [21]. This variant has been identified in multiple families with LS [22][23][24][25][26], Lynch-like syndrome [27], early onset colorectal cancer [28][29][30], as well as epithelial ovarian cancer [31]. In addition, this variant has been found in the Latin American population in families meeting the Amsterdam and Bethesda criteria in Argentina [13,32], and in one study of the Hispanic population in the United States [33] and Mexico [13].
A variant of uncertain significance (VUS) is defined as a genetic sequence variant, whose association with disease risk is unknown. Since the risk is unknown, it is not clear whether the sequence change is a typical variant, a polymorphism, or represents a disease-causing variant [34].  2017) report a total of 220 variants, of which, 37 were VUS and 1 variant had a conflict of interpretation. In our cohort, we found a lower number of VUS, reporting 15 VUS and 4 variants with conflicting interpretations. However, we have also found a number of VUS in MMR genes, in tumors outside of the LS spectrum. The reporting of a VUS represents a dilemma since it is not known where on the spectrum, from pathogenic to benign, a given VUS falls; carrier status does not stratify family members into those at higher or lower risk. Therefore, it does not provide any benefit in terms of medical management for the carrier or their family [35]. Given this uncertainty, it is worth emphasizing the importance of elucidating the pathogenicity of VUS, in order to provide management and follow-up for both patients and their families.
For the cancer susceptibility genes identified in the cohort that are not involved in LS, we found several genes implicated in the hereditary colorectal cancer landscape: APC, MUTYH, POLE, POLD1, and BLM, as well as genes resulting in an increased risk of colon cancer as CHEK2. The highest number of variants in the cohort was in the MUTYH gene, mostly PVs and pathogenic/likely pathogenic variants. Of the 21 MUTYH variants identified, 7 were found in patients diagnosed with colorectal cancer. MUTYH is a gene whose protein is involved in the base excision repair pathway that detects and protects DNA from oxidative stress [36,37]. The presence of monoallelic variants and medical management of carriers is controversial. Several studies have reported an association with colorectal cancer [37,38], endometrial cancer, and breast cancer [38], sites which were also reported in the present study. In addition, 13 variants, mostly pathogenic, were found in the APC gene. Familial adenomatous polyposis (FAP), characterized by the presence of dozens to hundreds of colorectal adenomas, is a hereditary disease caused by germline variants in APC, a key tumor suppressor gene in the regulation of the WNT signaling pathway [35,39]. FAP patients have a high risk of developing colorectal cancer, and an increased risk of gastric, small bowel, pancreatic, and thyroid carcinoma, in addition to bone and ophthalmological alterations [39]. In this cohort, 11 of the variants were identified in patients diagnosed with colorectal cancer, but without polyps or other clinical manifestation of FAP. Finally, related to hereditary colorectal cancer, 11 variants in POLE and 6 variants in POLD1 were found. Pathogenic variants in POLE and POLD1 cause PPAP syndrome (polymerase proofreading-associated polyposis), where there is an increased risk of developing colorectal cancer [40]. In the cohort, all variants identified were VUS in colorectal cancer patients. Different NGS studies have reported that up to 18% of patients under 50 years old, diagnosed with colorectal cancer, have a PV in genes that are not traditionally associated with this neoplasm such as ATM, CHEK2, and BRCA1/2 [16]. In the cohort, 13 variants in CHEK2 were identified, of which, 9 were found in patients diagnosed with colorectal cancer. Most of these variants were of uncertain significance, but five pathogenic variants were identified. CHEK2 encodes a serine/threonine kinase, activated in response to DNA damage, regulating downstream effector proteins such as p53, BRCA1, and BRCA2. CHEK2 is a susceptibility gene for several types of cancer such as breast, ovarian, and colorectal cancer, among others [41,42]. In this cohort, 11 variants in ATM were identified, most of them diagnosed with colorectal cancer. ATM is a master regulatory kinase which is active in response to DNA damage, carriers of heterozygous variants are at risk of breast cancer and, potentially, pancreatic cancer [43]. The most recent version of the NCCN Guidelines for Genetic and Familial CRC Syndromes (version 1.2021), indicates that no evidence has been established; therefore, the risk of colorectal cancer in carriers of ATM variants is uncertain.
Likewise, our report showed 11 BRCA1 variants, 6 PVs and 5 VUS, as well as 9 BRCA2 variants, 5 PV and 4 VUS. Different studies have found the presence of variants in BRCA1 and BRCA2: in a total of 450 patients with early onset colorectal cancer, four variants were identified in BRCA1 and one variant in BRCA2 [44]; in 1260 individuals with a family history of LS, 8% had variants in these genes [45]; and in a cohort of 961 endometrial cancer patients, 1.04% presented with a PV [46]. PVs in BRCA1 and BRCA2 are classically associated with hereditary breast and ovarian cancer syndromes [47]. Colorectal cancer is not included in the BRCA1/2 tumor spectrum; published guidelines do not recommend increased detection of this type of neoplasm in carriers of mutations in these genes. Some studies have suggested a modest link between the risk of developing colorectal cancer and BRCA genes, although the data are often contradictory and cannot establish a causal link between variants and the diagnosis of colorectal cancer in probands [48].
The limitations of this study include: The National Cancer Institute is a countrywide reference oncology hospital, which reflects the population from the central region of Mexico, and the results obtained cannot be generalized to the rest of the country. However, the population of this study is relevant due to the large number of patients analyzed.
Hereditary Cancer Clinic patients must be referred by the attending physician (medical and/or surgical oncologist). If the physician does not suspect a hereditary cancer risk, these patients are not referred and therefore, they are not included. Nowadays, in our Institution there is greater awareness of the importance of cancer risk assessment in the Gynecology and Breast Cancer Units; as a result, our cohort had a higher representation of female patients. The identification of patients with a PV, hence with a new diagnosis of a hereditary cancer syndrome such as LS, benefit from management recommendations for enhanced surveillance, prophylactic surgeries, and cascade testing for cancer prevention [49].
Regarding therapeutic implications, the NCCN guidelines recommend searching for protein loss by immunohistochemistry (IHC) and/or microsatellite instability in all colorectal and endometrial cancer tumors [12]. In our Institute, due to hospital criteria and budget constraints, the pathologist only performs IHC for some cases. Immunotherapy of immune checkpoint inhibition targeting PD1 and CTLA-4 proteins has been shown to be effective in dMMR-MSI-H tumors and has become the standard of care for patients with metastatic colorectal cancer. The FDA has approved the use of the anti-PD1 antibodies pebrolizumab and nivolumab, anti-CTLA4 termelimunab, and the combination of nivolumab and ipilimumab for the treatment of dMMR-MSI-H. It is noteworthy that pMMR-MSI-L tumors demonstrate a lack of response to treatment by immunotherapy, which is attributed to the lower mutational burden of this type of tumor [7][8][9][10]. It is important to know the MSI status of tumors in order to provide appropriate and timely therapeutic options. In addition, we depend on multidisciplinary management so that patients have therapeutic adherence to follow-up and the availability of specific therapies.
To the best of our knowledge, there are no studies of LS that have dealt exclusively with Mexico. There are previous reports on the Mexican population in the Latin American context, with limited LS data [15,16]. This is the first study to report the spectrum of genetic variants in a cohort of patients with LS in the Mexican population, which provides a comprehensive molecular and clinical overview. It is imperative to highlight the importance of identifying carriers of PV in any of the MMR genes or other clinically actionable highrisk genes, who would benefit from either surveillance, targeted therapy, or risk-reducing surgery. The recognition of a population with specific health care needs will make it possible to outline public health strategies and policies.

Patients
This study is a retrospective analysis of a cohort of patients with suspected LS from the Hereditary Cancer Clinic, of the National Cancer Institute, from 2016 to 2021 (Ramirez-Otero et al., 2022, under review). Hereditary Cancer Clinic patients must have been referred by the attending physician (medical and/or surgical oncologist).
Suspected LS patients were selected using the Amsterdam and Bethesda criteria, as well as the NCCN guidelines. For the purposes of describing our population, we grouped patients under the term "suspected Lynch syndrome" for those who met the clinical criteria and underwent genetic testing. Based on molecular analysis results, we further divided them into "MMR", comprising those patients with pathogenic variants in MMR genes, and therefore with LS; "no variant identified", for those patients with a negative result; and, "non-MMR", comprising those patients with a variant in a gene other than an MMR gene.
Data such as gender, age at diagnosis, first-degree family history, molecular test result, site of incidence, and subtype of disease were considered.

Molecular Testing
Blood samples were used for molecular testing by next-generation sequencing by an extensive multigene panel, as previously reported (Ramirez-Otero et al., 2022, under review). In brief, 4 milliliters of peripheral blood were obtained from all patients through venipuncture. Genomic DNA was extracted from peripheral blood leukocytes using the commercial Wizard Genomic DNA Purification kit (Promega, Madison, WI, USA). Next generation sequencing (NGS) was performed on the Illumina commercial platform (Illumina, San Diego, CA, USA), using two panels of 263 and 322 genes associated with cancer. The genes analyzed were selected according to their function and association with high, moderate, or low susceptibility to cancer. The patients were assigned to the version panel, according to the date they were enrolled in the study.
When clinical suspicion was high, multiplex ligation-dependent probe amplification (MLPA; MLPA reagent kit and SALSA MLPA Reagents (MRC-Holland P003, P072) in MLH1 and MSH2, genes was used to identify genomic rearrangements in this MMR genes.
All patients had pre-and post-testing genetic counseling. Informed consent was obtained from each patient. The research protocol was current and was approved by the local research committees.
The variants were classified according to the five-tier class system, as follows: pathogenic, likely pathogenic, uncertain significance, likely benign, and benign variants [51]. Likely benign and benign variants were considered where no variant was identified.
Data were collected in Microsoft Excel, and descriptive statistics such as frequencies and percentages were calculated to describe the characteristics of the population. The mean was calculated with the statistical program R (version 3.6.3). All graphs were made using GraphPad Prism (version 9.0.1). involved for participating in this project.

Conflicts of Interest:
The authors declare no conflict of interest.