Contribution of MLH1, MSH2, and MSH6 large genomic rearrangements to Pakistani colorectal cancer patients.

Background
Colorectal cancer (CRC) is a major global health concern, with approximately 1.9 million new cases and over 0.9 million deaths reported annually [1]. In Pakistan, CRC ranks as the fourth most common cancer, with age-standardized (world) incidence and mortality rates of 5.4 and 3.1 per 100,000, respectively [1]. While the majority of CRC cases are sporadic, approximately 5–10% are hereditary, primarily due to germline pathogenic variants (PVs) in mismatch repair (MMR) genes (MLH1, MSH2, MSH6, and PMS2), as well as in other susceptibility genes (APC and EPCAM) [2].
Individuals harboring PVs in MMR genes are diagnosed with Lynch syndrome (LS), an autosomal-dominant cancer predisposition syndrome that accounts for 1–7% of all CRC cases [3]. Individuals with LS have a significantly increased lifetime risk of developing CRC, in addition to several extracolonic malignancies [4]. Among these individuals, large genomic rearrangements (LGRs) in MMR genes have been reported to contribute to 5–20% of all germline PVs [5, 6]. Furthermore, germline deletions in the 3′ end of EPCAM can inactivate the adjacent MSH2 gene, thereby also contributing to LS [7]. Molecular characterization of LGR breakpoints can offer crucial insights into the mechanisms driving these rearrangements. Identifying carriers of MMR gene LGRs is of clinical significance, as it directly impacts cancer surveillance and treatment strategies [8].
The prevalence and spectrum of MMR genes LGRs vary considerably across ethnic and geographic populations. In European cohorts, LGRs have been reported in 0% to 26.7% of CRC patients [9–21]. In Australia, prevalence rates range from 1.4% to 9.4% [22, 23], whereas studies from the United States report frequencies between from 1.4% to 12.7% [24–26]. Among Asian CRC patients, reported frequencies range from 0.7% to 27.3% [27–36], highlighting substantial variability across populations.
In Pakistan, however, the contribution of LGRs in MMR genes to CRC remains unknown. Previously, we reported that small-range PVs in MLH1 and MSH2 were detected in 34.5% of hereditary non-polyposis colorectal cancer (HNPCC) and suspected-HNPCC patients [37]. However, this frequency may be an underestimation, as prior studies utilized PCR-based screening methods, which are insufficient for detecting LGRs. Consequently, the true prevalence of MMR gene LGRs in Pakistani CRC patients remains unclear. The current study aims to address this gap by investigating the prevalence of LGRs in MLH1, MSH2, MSH6 and the 3′ end of EPCAM in a cohort of 199 Pakistani CRC patients who tested negative for small-range PVs in MMR genes. By elucidating the potential role of LGRs in CRC development within this population, our findings may provide valuable insights for the clinical management of hereditary CRC and contribute to the development of targeted diagnostic strategies in Pakistan.

Methods

Study subjects
This study included a cohort of unselected, consecutively enrolled CRC patients treated at Shaukat Khanum Memorial Cancer Hospital and Research Centre (SKMCH&RC) in Lahore, Pakistan, between November 2007 and March 2011. Patients were stratified into two groups: the HNPCC/suspected-HNPCC group (n = 29) and the non-HNPCC group (n = 183), as previously described [37].
The HNPCC group comprised patients fulfilling the revised International Collaborative Group on HNPCC (ICG-HNPCC) criteria, also known as the Amsterdam criteria II. These criteria included: (i) at least three relatives affected with histologically verified CRC or cancers of the endometrium, small bowel or urinary tract, with at least one being a first-degree relative of the other two; (ii) at least two of the above individuals were first-degree relatives from two different generations; (iii) at least one individual diagnosed before the age of 50 years; and (iv) exclusion of familial adenomatous polyposis (FAP) [38, 39]. The suspected-HNPCC subgroup included individuals meeting less stringent criteria: (i) at least one case of CRC, cancer of the endometrium, small bowel or urinary tract in a first-degree relative of a CRC patient (or in the patient him/herself); (ii) at least one of the above cancers diagnosed before age 50; (iii) exclusion of FAP [40]. The remaining 183 cases, who did not meet the criteria for either HNPCC or suspected-HNPCC, were classified as sporadic CRC patients and grouped under non-HNPCC.
Prior to the current study, all patients had undergone screening for small-range PVs in MLH1, MSH2, and MSH6 using denaturing high-performance liquid chromatography (DHPLC), followed by DNA sequencing of variant fragments [37]. Patients found to harbor small-range PVs affecting one or a few nucleotides, including frameshift, nonsense, or splice-site alterations, were excluded from further analysis.
The present study thus included 199 CRC patients who tested negative for small-range PVs in MMR genes. These individuals were re-stratified into the HNPCC/suspected-HNPCC group (n = 18) and the non-HNPCC group (n = 181). A detailed description of these patients is provided in Fig. 1. Clinical and histopathological data were collected from medical records and pathology reports. All participants provided written informed consent, and the study was approved by the Institutional Review Board (IRB) of SKMCH&RC (IRB approval number SKMCH-CRC-001).

Multiplex ligation-dependent probe amplification (MLPA) analysis
To comprehensively analyze LGRs in MLH1, MSH2, MSH6 (entire genes), and the 3′ end of EPCAM, we used SALSA MLPA Kits P003-C1 and P072-C1, following the manufacturer’s instructions (MRC Holland, Amsterdam, The Netherlands). DNA fragments were separated by capillary electrophoresis on an ABI 3500 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA), and the output data were analyzed using Coffalyser.Net software (MRC Holland, Amsterdam, The Netherlands). For quality control, visual peak pattern evaluation ensured that the standard deviation of reference sample probes was less than 0.10. The dosage quotients (DQ) of patient samples were interpreted as follows: DQ between 0.80 and 1.20 suggested normal copy number, DQ between 0.40 and 0.65 indicated deletion, while a DQ between 1.75 and 2.15 suggested duplication. Positive controls included DNA samples with known LGRs in MLH1 (exons 1–8, exon 6, and exon 10 deletions), MSH2 (exons 1–6, and exon 8 deletions), and EPCAM (exons 8–9 deletion). All samples showing deletions were confirmed using independent MLPA assays. Deletions were further analyzed using Sanger sequencing on an ABI 3500 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA) to rule out polymorphisms at probe ligation sites.

Identification and characterization of genomic breakpoints
DNA samples showing gross rearrangements in MSH2, specifically in the 5′ upstream region, exons 1–3, exons 1–6, exon 7, and exon 11, were subjected to further analysis to determine the precise breakpoints. Long-range PCRs were performed using AccuStart Long Range SuperMix (Quantabio, Beverly, USA), with multiple primer pairs designed to amplify regions of interest for Sanger sequencing. The locations of primer pairs (sequences available upon request) used to characterize the MSH2 5′ upstream deletion, exons 1–3 deletion, and exon 11 deletion are presented in Fig. 2.
Deletion-specific PCR products were successfully amplified with the following fragment sizes: 1405 bp (5′ MSH2 upstream deletion), 673 bp (exons 1–3 deletion), ~4 kb (exons 1–6 deletion), ~7 kb (exon 7 deletion), and 380 bp (exon 11 deletions). PCR products containing the breakpoint regions were purified using the QIAquick PCR purification kit (Qiagen, Hilden, Germany) and bidirectionally sequenced using the BigDye Terminator v3.1 Cycle Sequencing kit on the ABI 3500 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). Sequence alignments were performed using the February 2009 human genome assembly (GRCh37/hg19) (accessible at: https://genome.ucsc.edu/). All deletions were annotated based on the MSH2 reference sequence NM_000251.3 and chromosome 2 coordinates from the GRCh37/hg19 assembly. Additionally, sequences at the breakpoints were analyzed using the RepeatMasker program (http://www.repeatmasker.org) to investigate repetitive elements potentially involved in LGR mechanisms.

Results

Characteristics of the study participants
The clinical and demographic characteristics of 199 unrelated index patients from Pakistan, including both HNPCC/suspected-HNPCC and non-HNPCC groups, are summarized in Table 1. Of these, 198 patients were diagnosed with CRC (61 females and 137 males), while one female with ovarian cancer who met the suspected-HNPCC criteria. The median age at CRC diagnosis was 44 years for males and 40 years for females, with no statistically significant difference between the two groups (Mann–Whitney U test, P = 0.072). The majority of participants belonged to either the Punjabi (82/199; 41.2%) or Pathan (69/199; 34.7%) ethnic groups. In the HNPCC/suspected-HNPCC group, the median age of disease onset was 44 years for males (range 29–60 years; n = 14) and 31.5 years for females (range 20–38 years; n = 3). In the non-HNPCC group, the median age of disease onset was 42 years for both males (range 17–77 years; n = 123) and females (range 14–70 years; n = 58).

Frequencies of LGRs in HNPCC/suspected-HNPCC and non-HNPCC patients
Comprehensive screening for LGRs in MLH1, MSH2, MSH6, and the 3′ end of EPCAM using MLPA analyses identified five distinct MSH2 deletions (5′ upstream, exons 1–3, exons 1–6, exon 7, and exon 11). No LGRs were detected in MLH1, MSH6, or the 3′ end of EPCAM. Overall, LGRs were detected in 4.0% of CRC patients (8/199), including 11.1% (2/18) of HNPCC/suspected-HNPCC group and 3.3% (6/181) of non-HNPCC group (Table 2).

Genomic breakpoints were identified for all deletions except for the exons 1–6 and exon 7 deletions, which could not be characterized using long-range PCR and DNA sequencing (Fig. 2).

MSH2 5′ upstream deletion
A recurrent MSH2 5′ upstream deletion was identified in four unrelated CRC patients, all of Punjabi ethnicity. One patient belonged to the HNPCC/suspected-HNPCC group and was diagnosed with CRC at age 50 (Fig. 3a, II:1). The index patient’s mother (I:1) was diagnosed with liver cancer at age 60, while the father (I:2) was diagnosed with CRC at age 60 (Fig. 3a). The same deletion was also detected in three non-HNPCC patients, diagnosed with CRC at ages 60, 61, and 70 (Table 2; Fig. 3b, c, d).
Characterization of the genomic breakpoints revealed a 3,308 bp deletion (NC_000002.11:g.47624484_47627791del3308) (Fig. 2a, c, h). Sequence analysis at the junction points showed fusion between AluSp sequences in the MSH2 5′ upstream region at chr2: 47,624,455–47,624,483 and chr2: 47,627,763–47,627,791. The deletion involved a 29 bp core region with 100% sequence homology, suggesting unequal homologous recombination as the underlying mechanism.

MSH2 exons 1–3 deletion
The MSH2 exons 1–3 deletion was identified in a 45-year-old CRC patient from the non-HNPCC group (Fig. 3e, II:1), of Punjabi ethnicity. This deletion spanned 9,855 bp (NC_000002.11:g.47628637_47638491del9855) (Fig. 2a, d, i). Sequence analysis revealed a fusion between AluSx1 sequences in the MSH2 5′ upstream region (chr2:47,628,610–47,628,636) and AluSp sequences in MSH2 intron 3 (chr2:47,638,465–47,638,491). The deletion involved a 27 bp core region with 97% homology, suggesting unequal homologous recombination, as the underlying mechanism.

MSH2 exons 1–6 deletion
The MSH2 exons 1–6 deletion was identified in a 30-year-old CRC patient from the non-HNPCC group (Fig. 3f, III:1), of Punjabi ethnicity. The patient’s maternal aunt (II:1) had breast cancer at age 45, and the maternal grandmother (I:1) had bladder cancer at age 50 (Fig. 3f). This deletion spanned approximately 19.5 kb (Fig. 2f). Precise breakpoints for this deletion were not identified.

MSH2 exon 7 deletion
The MSH2 exon 7 deletion was identified in a 50-year-old CRC patient from the non-HNPCC group (Fig. 3g, III:4), of Pathan ethnicity. The index patient’s paternal aunt (II:5) was diagnosed with breast cancer before age 55 (Fig. 3g). This deletion spanned approximately 3 kb (Fig. 2g). The exact breakpoints for this deletion were not identified.

MSH2 exon 11 deletion
The MSH2 exon 11 deletion was identified in a 44-year-old CRC patient from the HNPCC/suspected-HNPCC group (Fig. 3h, III:3), of Punjabi ethnicity, with a strong family history of various cancers. Affected relatives included the patient’s mother (II:5) with uterine and tongue cancers, a maternal uncle (II:2) with CRC, a maternal aunt (II:3) with an abdominal malignancy, a niece (III:1) with a uterine tumor, and a nephew (III:2) with a brain tumor (Fig. 3h). Genomic breakpoints analysis identified a 3,723 bp deletion spanning MSH2 intron 10-intron 11 with a 6 bp insertion (NC_000002.11:g.47697567_47701289del3723insAGGTTG) (Fig. 2a, e, j).

Discussion
In this study, we investigated the prevalence of LGRs in key MMR genes (MLH1, MSH2, MSH6, and the 3′ end of EPCAM) among 199 CRC patients from Pakistan, who had previously tested negative for small-range PVs in MMR genes. We identified five distinct LGRs in the MSH2 gene, with a frequency of 11.1% (2/18) in HNPCC/suspected-HNPCC group and 3.31% (6/181) in non-HNPCC group. No LGRs were detected in the MLH1, MSH6, or the 3′ end of EPCAM genes.
The overall frequency of MSH2 LGRs in our cohort (4.0%) aligns with previous Asian reports, where frequencies range from 0.7% to 14.3% in CRC patients from China, Japan, Taiwan and Israel [27–36]. However, our findings show variability when compared to reported frequencies in Europe (0% to 56%) [9–21], Australia (0% to 6.2%) [22, 23], and the United States (0% to 7.1%) [24–26, 41]. This variation may reflect differences in study design, inclusion criteria, sample size, detection assays, or ethnic and geographic diversity.
A recurrent MSH2 5′ upstream deletion (3,308 bp) was identified in four unrelated CRC patients, all of Punjabi ethnicity. Similar deletions have been reported in three HNPCC patients of European descent, albeit with different breakpoints [42, 43]. The presence of Alu sequences at the breakpoints suggests unequal homologous recombination as the likely underlying mechanism for this recurrent deletion.
The spectrum of MSH2 LGRs detected in this study, including deletions of exons 1–3, exons 1–6, exon 7, and exon 11, mirrors findings from other populations, but also exhibits unique features, such as recurrent deletions specific to this cohort. These deletions affect exons critical for MSH2 function and are considered pathogenic, as they disrupt the interaction between MSH2 and its partners hMSH3/hMSH6, essential for mismatch repair.
The MSH2 exons 1–3 deletion (9,855 bp) was identified in a non-HNPCC patient of Punjabi ethnicity. This deletion removes the MSH2 promoter region and the first three coding exons, abolishing gene expression and function. Similar deletions have been reported in HNPCC patients from Italy [17], and in studies from Europe and China [15, 18–21, 27], although the breakpoints often differ or remain uncharacterized. The pathogenicity of this deletion arises from its impact on the promoter and coding exons, ultimately disrupting gene expression.
The MSH2 exons 1–6 deletion was identified in a non-HNPCC patient of Punjabi background. This deletion is among the most frequently reported MSH2 LGRs globally, as it affects the MSH2 promoter region and results in transcriptional silencing of the mutant allele. This deletion has been reported across Asian, European, and North American populations [17, 19, 21, 32, 35, 36, 41–46]. Although we could not identify the specific breakpoints, the affected region aligns with previously recognized hotspot regions.
A novel MSH2 exon 7 deletion was identified in a non-HNPCC patient of Pathan ethnicity. This deletion removes 68 amino acids from the hMSH3/hMSH6 interaction domain, which is critical for MSH2 function. Similar deletions have been reported in Asian, European, and South American CRC patients [19, 21, 27, 32, 36, 42, 45, 47, 48], with variable breakpoints. This deletion is considered pathogenic due to its disruption on the MSH2 protein’s interaction domain.
Additionally, we identified one MSH2 exon 11 deletion in a HNPCC/suspected-HNPCC patient of Punjabi ethnicity. This deletion removes 34 amino acids from the hMSH3/hMSH6 interaction domain. Similar deletions have been reported in CRC patients from Asia [32] and Europe [49], although breakpoints have largely remained uncharacterized. The pathogenicity of this deletion lies in its disruption of a crucial protein interaction domain, critical to mismatch repair function.
The absence of LGRs in MLH1, MSH6, and the 3′ end of EPCAM in our cohort is consistent with findings from other studies [50–52], suggesting that LGRs in these genes may be rare in the Pakistani population. In contrast, the spectrum of MSH2 LGRs identified in our study demonstrates both overlap with and divergence from those reported in other populations. Specifically, four MSH2 deletions (exons 1–3, exons 1–6, exon 7, and exon 11) have previously been reported in Asian [27, 32, 35, 36] and Caucasian [15, 17–21, 41–47] cohorts. While these deletions had different breakpoints except exons 1–3 deletion had the same breakpoints in our cohort and thereby supports its recurrent and global shared occurrence. We also identified a recurrent MSH2 5′ upstream deletion that has been previously reported in Caucasians [42, 43] but, to our knowledge, not yet in Asian cohorts. This deletion was observed only among Punjabi patients, suggests a founder effect or an independent recombination event within a similar genomic context. The identification of Alu-mediated breakpoints in our cases further suggests a common Alu-driven recombination mechanism, with potential ethnic specific variations in breakpoint. These findings suggest the need for population-tailored LS screening strategies to improve diagnostic yield in underrepresented populations.
In our study, 87.5% (7/8) of MSH2 deletions, including a recurrent MSH2 5′ upstream deletion, were identified among patients of Punjabi ethnicity. Punjabis represents ~ 50% of Pakistan’s total population and are further subdivided into various tribes, clans, and communities, each with unique backgrounds. This clustering pattern suggests a potential ethnic-specific founder deletion. However, the retrospective study design and the unequal representation of ethnic groups preclude definitive conclusions. Future multicenter studies with larger, and ethnically diverse cohorts are required to validate these observations.
Our findings have important clinical implications for LS management. According to the National Comprehensive Cancer Network (NCCN) guidelines (version 1.2025), MSH2 PV carriers should undergo high-quality colonoscopic surveillance every 1–2 years from age 20–25 years, or 2–5 years earlier than the youngest CRC diagnosis within the family. Preventive strategies, including aspirin chemoprevention, lifestyle modification, and weight control, may further reduce cancer risk. Genetic counseling and predictive genetic testing should be offered to first-degree relatives or if unavailable, more distant relatives for the known familial PV. These implications underscore the importance of gene-specific surveillance, personalized prevention, and therapy for MSH2 PV carriers [53].
This study has several limitations. First, universal LS screening using Bethesda guidelines (microsatellite instability testing, immunohistochemistry for MMR proteins, and BRAF V600E testing) was not performed due to the limited paired tumor/normal tissues and the retrospective nature of sample collection. Consequently, some LS cases may have been missed, although non-HNPCC cases helped mitigate bias, thereby minimizing potential selection bias. Second, our cohort’s male predominance (137 males vs. 61 females) consistent with national CRC epidemiology [1] but may reflect sociocultural disparities in healthcare access. Although no significant gender-based differences were observed within the overall cohort, the smaller HNPCC/suspected-HNPCC group showed a modest age difference between males and females, likely reflecting the small sample size (n = 3 females) rather than a true biological difference. Third, the predominance of Punjabi and Pathan patients may not fully represent the genetic diversity of CRC across Pakistan. These findings may reflect selection bias due to the single-center study participant’s enrollment. Future multicenter studies with balanced gender and ethnic representation are therefore needed to minimize bias and enhance generalizability. Fourth, segregation analysis and cascade testing were not performed due to the unavailability of their DNA, which limited confirmation of inheritance patterns and variant pathogenicity. Moreover, next generation sequencing (NGS)-based multigene panel testing, which is now the current standard approach for LS screening, was not performed. Nevertheless, MLPA remains a robust and complementary method for detecting LGRs that may be missed by NGS, ensuring reliable identification of clinically significant LGRs [54]. Despite these limitations, the study has several strengths. It represents one of the largest and most comprehensive South-Asian investigations of LGRs in MLH1, MSH2, MSH6, and the 3′ end of EPCAM. The inclusion of positive controls in MLPA assays and independent validation of deletions strengthen the reliability of the study results. Notably, the identification of a recurrent MSH2 5′ upstream deletion among Punjabi patients suggests a potential founder LGR, which could serve as the basis for population-specific, cost-effective targeted genetic testing in the future.

Conclusions
This study represents the first comprehensive analysis of LGRs in MMR genes among Pakistani HNPCC/suspected-HNPCC and non-HNPCC patients. Our findings highlight the significance of including MSH2 LGRs in genetic testing panels for CRC in Pakistan. The identification of a recurrent LGR in a specific ethnic group may inform the development of more efficient, cost-effective genetic testing strategies and facilitate early identification and management of at-risk individuals in this population.