Polymorphisms of Mismatch Repair Pathway Genes Predict Clinical Outcomes in Oral Squamous Cell Carcinoma Patients Receiving Adjuvant Concurrent Chemoradiotherapy

Background: We aimed to investigate the association between single-nucleotide polymorphisms (SNP) in mismatch repair (MMR) pathway genes and survival in patients with oral squamous cell carcinoma (OSCC) who received adjuvant concurrent chemoradiotherapy (CCRT). Methods: Using the Sequenom iPLEX MassARRAY system, five SNPs in four major MMR genes were genotyped in 319 patients with OSCC who received CCRT treatment. Kaplan–Meier survival curves and Cox proportional hazard regression models were used to assess overall survival (OS) and disease-free survival (DFS) among MMR genotypes. Results: The results of Kaplan–Meier survival analysis revealed that the MutS homolog 2 (MSH2) rs3732183 polymorphism showed a borderline significant association with DFS (log-rank p = 0.089). Participants with the MSH2 rs3732183 GG genotype exhibited a relatively low risk of recurrence (hazard ratio (HR) = 0.45; 95% confidence interval (CI) = 0.22–0.96; p = 0.039). In addition, the MutL homolog 1 (MLH1) rs1800734 GG genotype carriers exhibited higher OS (HR = 0.52, 95% CI = 0.27–1.01; p = 0.054) and DFS (HR = 0.49, 95% CI = 0.26–0.92; p = 0.028) rates. Conclusions: Our results indicated that the GG genotypes of MSH2 rs3732183 and MLH1 rs1800734 are associated with relatively high survival in OSCC patients treated using adjuvant CCRT. These polymorphisms may serve as prognosis predictors in OSCC patients.


Introduction
Approximately 3% of all cancers occur in the oral cavity; approximately 90% of all oral cancers (OCs) are oral squamous cell carcinoma (OSCC), which is one of the most prevalent cancers and the fourth commonest causes of cancer mortality among men in Taiwan [1].Despite advancements in the diagnosis and treatment of OSCC, the 5-year overall survival (OS) rate, which is currently 56.6%, has not improved significantly in the past decades [1].This is mainly because approximately 51.5% of OSCC cases are diagnosed at advanced stages (III and IV) [1].For the early stages, namely stages I and II, surgery constitutes the standard treatment of choice and often results in a permanent cure.Adjuvant treatments of chemotherapy and radiotherapy (RT) with surgery are used to treat advanced stages of OSCC, depending on the overall health status of the patients and invasive pathological parameters [2,3].Despite aggressive therapy, the 5-year survival rate among patients with advanced stages of OSCC is low, and most patients die in the first 30 months of the disease [4,5].The clinical outcomes of OSCC often vary with factors related to the tumor and treatment [6].However, inherent patient genetic characteristics may affect patient prognosis [7].
The response of tumor cells to DNA damage involves numerous complicated molecular mechanisms [8,9].For example, DNA repair mechanisms can affect the survival of tumor cells.Both chemotherapy and RT induce DNA lesions, including replication errors, thereby activating the DNA damage and repair response [10].The DNA mismatch repair (MMR) pathway proteins, which include MutL homolog 1 (MLH1), MutS homolog 2 (MSH2), MutS homolog 3 (MSH3), and exonuclease 1 (EXO1), play significant roles in the repair process by recognizing DNA damage caused by endogenous and exogenous agents [11,12].MSH2, which dimerizes with other proteins (MSH6 or MSH3), and are responsible for recognizing and initiating the mismatch repair.The MLH1 protein joins another protein (MLH4) to form a protein complex that coordinates downstream repair events, which involves the EXO1 protein [12].Genetic variations in MMR genes alter the capacity of individuals to repair DNA damage induced by radiotherapeutic and chemotherapeutic agents, leading to variations in outcomes ranging from tumor cell apoptosis to resistance [13,14].For example, the G-alleles of MLH1 rs1800734 [15], MSH3 rs26279 [14,16], and EXO1 rs1047840 [14] have been reported to be associated with higher levels of protein expression and/or higher survival rates than the variant alleles.Thus, the G-alleles represent higher DNA repair capacity than the other alleles, indicating that these single-nucleotide polymorphisms (SNPs) may have the potential to become prognostic biomarkers for individualized therapy.
Studies on the effects of genetic variations in MMR genes on the prognosis of cancer have generally included patients receiving different treatments and/or patients with different ethnic backgrounds [14,[16][17][18].Given that different treatments may cause varying DNA damage and exhibit different repair efficacy, we investigated the association between polymorphisms and outcomes in OSCC patients who received identical treatment with adjuvant concurrent chemoradiotherapy (CCRT) and exhibited similar disease stages.Five SNPs of MSH2, MSH3, EXO1, and MLH1 were selected according to their effects on the risk and/or survival of tumors [14,16,19,20].

Discussion
The MMR is a complicated network with numerous functions.One prominent function is genomic stability, which is achieved by eliminating mismatched or distorted DNA strands.SNPs in this pathway are shown to alter the anticancer effects of therapeutic agents.Therefore, identifying such SNPs may determine prognostic markers in patients with OSCC.In this retrospective cohort study, we investigated the association between genetic variants of MMR genes and clinical outcomes in the patients with OSCC who received adjuvant CCRT.The results indicated that the MSH2 rs3732183 GG genotype was a favorable prognostic indicator of relapse, and the MLH1 rs1800734 GG genotype was a favorable prognostic indicator of both relapse and death.In addition, patients with ENE exhibited a high risk of death, while those of Mainland Chinese ethnicity, with primary tumor size in the T3-T4 range and nodal involvement in the N2-N3 range exhibited high risks of relapse.
MSH2 is a prominent member of the MMR pathway, and its inactivation can have far-reaching pathological effects on DNA.Studies have reported that polymorphisms in this gene affect the DNA damage and repair mechanism [14,21].Similarly, we found that variations in MSH2 affected clinical outcomes in patients with OSCC.Patients with the rs3732183 GG genotype exhibited a lower risk of relapse than the patients with the AA genotype.The SNP rs3732183 is an intronic SNP that may affect MSH2 expression through cis-acting regulatory elements (such as enhancers, silencers, insulators, and transcription factors) that positively control gene expression [22].The A to G substitution may result in high MSH2 expression, which is favorable to the anticancer effects of therapeutic agents, thus reducing the risk of relapse.Low expression levels of MSH2 were previously found to have unfavorable prognostic value for different cancers, including head and neck squamous cell carcinoma (HNSCC) [18,23,24].Huang et al. reported MSH2 overexpression to correlate with better survival in Taiwanese colon cancer patients [25].Alternatively, the SNP may be in high linkage disequilibrium with other nearby polymorphisms that may affect MSH2 expression.Previously, Kang et al., in their study on Korean colorectal cancer patients, also reported the GG genotype of rs3732183 to show a favorable prognostic factor for DFS [26].
Consistent with the results of previous studies [7,27], we also observed that the carriers of the MLH1 rs1800734 GG genotype exhibited lower risks of relapse and death than did the carriers of the wild-type AA genotype.The SNP rs1800734 is located at an MLH1 promoter CpG island transcription factor-binding site that can cause differences in individual susceptibility by regulating the activity of MLH1 and other downstream proteins [28].Several studies have reported that rs1800734 is associated with high levels of methylation and low levels of protein expression in different types of tumors [29][30][31], which suggests that this polymorphism may contribute to gene dysfunction by altering transcription activity.The GG genotype carriers may be at a relatively low risk because the G-allele provides a favorable binding site for transcription factors, such as AP-4, which recruits RNA polymerase II and c-Myc, thereby leading to MHL1 transcription [28].By contrast, a repressor protein may bind to the A-allele to recruit epigenetic modifying factors leading to promoter methylation and low level of MHL1 transcription [29].Extensive promoter methylation was associated with MLH1 inactivation [32].The AA genotype of MLH1 rs1800734 was previously found to be correlated with poor prognosis in Taiwanese patients with lung cancer [27].Our findings support the proposition that reduced DNA damage efficacy offers a survival advantage in patients with OSCC.
A few studies have reported relatively poor survival outcomes in the EXO1 rs1047840 and MSH3 rs26279 GG genotypes among patients with HNSCC [14,16].Nogueira et al. reported that the GG genotypes of EXO1 rs1047840 and MSH3 rs26279 were associated with DFS and OS, respectively, in a Brazilian population with HNSCC.A similar Brazilian study on patients with HNSCC showed that the MSH3 rs26279 GA or AA genotypes are associated with an approximate nine-fold higher risk of partial response to cisplatin chemoradiation or achieving stable disease than are patients with other genotypes [14].The EXO1 rs1047840 GA genotype was associated with complete recovery in patients with laryngeal cancer treated using cisplatin and RT [33].However, we did not observe any association between the SNPs and clinical outcomes in our sample.The difference between the results of the previous study and those of the present study may be attributable to differences in ethnicity, tumor characteristics, treatment, and median follow-up times.Our sample was highly homogeneous and included patients with advanced OSCC of identical ethnicity who received identical treatment, whereas the study samples in the aforementioned studies consisted of patients with diverse tumor characteristics who received different treatment modalities.In addition, despite the long follow-up time of 199 months in our study, the median follow-up time in this study was shorter than that in other studies [14,16] and could be attributed to the poor tumor characteristics of our patients.The difference in findings may also be attributed to the low statistical power of our study.Additional studies with larger samples than that in the present study may be required to confirm the results of the present study.
As previously reported [14,16,34], we observed that the participants with ENE exhibited an increased risk of death and those with primary tumor size in the T3-T4 range and nodal involvement exhibited poorer DFS than their counterpart with milder characteristics.Epidemiological studies from Taiwan showed ethnic difference in oral cancer survival [35,36].Similarly, in our study mainland Chinese participants exhibited increased risk for recurrence than does participants of Taiwanese ethnicity.
The strength of our study is the relative homogeneity of our patient population with respect to ethnicity, treatment modality, and tumor stage, which eliminated the effects of differences in DNA damage and repair mechanisms and their effects on clinical outcomes.The relatively small sample owing to the strict inclusion criteria and the recruitment of patients from only one hospital may have limited the generalizability of our findings.Despite the aforementioned limitation, this is the first study to investigate the association between genetic polymorphisms in MMR genes and clinical outcomes in patients of Chinese ethnicity with advanced OSCC receiving adjuvant CCRT.

Ethical Statement
The study was conducted in accordance with the Declaration of Helsinki.The ethics review committees of the Chang Gung Memorial Hospital (IRB No. 201800213B0) and the Taipei Medical University (IRB No. N201802083) granted final approval for the study.Written informed consent to participate in the study was obtained from all the participants after a detailed explanation of the study objectives.

Participants and Data Collection
For this study, participants were recruited from the Chang Gung Memorial Hospital, Linkou, Taiwan, and constituted a part of the Head and Neck Surgery Department's Cancer Registry from 1999 to 2016.The registry included 2528 cases of OSCC.After excluding 51 cases who had incomplete questionnaires, there were 973 cases who received surgery only, 513 cases who had surgery plus radiotherapy, 503 cases who had post-operative CCRT, and 488 cases who received other treatment or had missing information on treatment.We selected those patients who had received CCRT and restricted the sample to male subjects, which left a sample of 473 for the present study.Patients were further excluded if they had no blood specimen (n = 105), were of aboriginal ethnicity (n = 13), had early-stage OSCC (n = 12), had missing clinicopathological information (n = 12), or had failed the genotyping experiment (n = 7).Finally, 319 participants were included in the analysis.The sociodemographic data and lifestyle habits of the participants were obtained using an interviewer-administered questionnaire.The collected information included age, education level, ethnicity, lifestyle habits (history of alcohol drinking, betel quid chewing, and cigarette smoking), and family history of cancers.Lifestyle habits were dichotomized into never (never engaged in the habit for over 1 year continuously) and ever (had ever engaged in the habit for over 1 year continuously).Detailed clinical information was also collected by taking medical history, evaluating complete blood count, and conducting complete physical examination, routine blood chemistry tests, whole-body bone scan or positron emission tomography scan, abdominal ultrasound and magnetic resonance imaging (MRI), or computed tomography (CT) of the head and neck.The clinicopathological characteristics of the patients included tumor stage, depth, and differentiation, nodal status, and ENE, as well as perineural, skin, and bone invasions.The participants' weight and height were also recorded, and BMI was computed using the following formula: weight/height 2 (kg/m 2 ).

DNA Extraction and Genotyping
Pairs of samples from the tumor and normal adjacent nontumor tissue that were previously obtained from each participant and stored in liquid nitrogen (at −80 • C) were used for pathological examination and staging by a pathologist.The Seventh Edition of the American Joint Committee on Cancer (AJCC)-TNM staging system [37] was used for staging.Only participants diagnosed with histological squamous cell carcinoma and TNM stages III and IV were considered for the study.Prior to treatment, the participants' venous blood samples were collected and centrifuged, and buffer-coated cells were collected for DNA extraction.Genomic DNA extraction was performed using the standard phenol-chloroform method.
Extracted DNA was used for genotyping the MSH2 SNP rs3732183, MSH3 SNPs rs12515548 and rs26279, EXO1 SNP rs1047840, and MLH1 SNP rs1800734, which were selected based on their previously reported association with the risk or prognosis of cancers [14,16,19,20].The Sequenom iPLEX MassARRAY system (Sequenom, Inc., San Diego, CA, USA) was used for genotyping.Using the Sequenom MassARRAY platform and iPLEX GOLD chemistry, matrix-assisted laser desorption ionization time of flight (MALDI-TOF) spectroscopy was conducted.Then, 10 ng of genomic DNA was used as a template, and a polymerase chain reaction (PCR) mixture containing Qiagen HotStarTaq (Qiagen, Valencia, CA, USA) was prepared.Primer extension and cleanup using shrimp alkaline phosphatase were performed according to the Sequenom guidelines.The PCR primers were obtained from Integrated DNA Technology (Coralville, IA, USA).The MassARRAY Assay Design software Version 3.1 (Sequenom) was used for assay design.For quality control, 10% of samples were randomly selected, and genotypes showed 100% concordance for all SNPs.

Adjuvant CCRT and Follow-Up
Prior to CCRT treatment, all the patients had undergone surgery that included radical tumor excision with neck dissection based on clinical stage after tumor survey.The primary tumor resection was performed at 1 cm above safety margins (both peripheral and deep margins), and neck dissection was performed according to examination status.All the patients received CCRT with radiation doses between 6000 and 6600 cGy, bi-weekly or tri-weekly intravenous cisplatin 40-60 m 2 /kg, and oral 5-fluorouracil for 4-8 weeks after the surgical procedure.The patients were evaluated during and after treatment through regular clinical and radiological examinations.The examinations involved check-ups every month for the first 6 months, every 2 months for the next 6 months, every 3 months within the second year, and every 6 months thereafter.Monitoring included analysis of medical history and physical examination (including complete oral examination), X-rays, CT or MRI, and laboratory examination.History of biopsy or imaging studies were used to confirm relapse, and deaths resulting from OSCC were recorded based on death certificates.

Statistical Analysis
The major clinical outcome was relapse (DFS) of OSCC or death (OS) caused by OSCC.The DFS time (in months) was calculated from the day of treatment commencement to the time of relapse, metastasis, or death from any cause.The OS time was defined as the duration (in months) between the time of treatment commencement and date of death.The patients who did not experience any event as of the date of the last follow-up visit were censored or subject to administrative censoring by the end of the study period (10 April 2019).The demographic and clinicopathological parameters of the participants were summarized using descriptive statistics.Cumulative survival in different genotypes was analyzed using Kaplan-Meier curves, and the differences between the genotypes were tested using the log-rank test.The association among the demographic and clinicopathological parameters, candidate SNPs, and OS or DFS were tested using univariate and multivariate Cox proportional hazard regression models.Multivariate analysis was conducted to examine the associations between individual SNPs and OS or DFS after adjusting for age and clinicopathological variables that were significant at p < 0.10 in the univariate model.Different genetic models were used to test the effect of individual SNPs on OS and DFS (including codominant, additive, dominant, recessive, and allelic models).The most significant genetic model was finally used in the multivariate analysis.The relative risk of relapse or death was estimated using HR and their corresponding 95% CIs.For the two SNPs located in MSH3, LD analysis was performed using Haploview (version 4.2, Broad Institute, Cambridge, MA, USA).All analyses were two-sided, and p < 0.05 was considered statistically significant.All analyses were conducted using the SAS software (version 9.4 for Windows; SAS Institute, Cary, NC, USA).

Figure 1 .
Figure 1.Kaplan-Meier analysis and log-rank test results of OS and DFS curves for the MSH2 rs3732183 and MLH1 rs1800734 polymorphisms in patients with oral squamous cell carcinoma treated with concurrent chemoradiotherapy.(A-1) No significant difference in OS among the MSH2 rs3732183 genotypes (log-rank test p = 0.5185).(A-2) The GG genotype showed a borderline significant better DFS time than the AA genotype (log-rank test p = 0.0887).(B-1,2) No significant difference in death and recurrence among genotypes of MLH1 rs1800734 (log-rank test p = 0.1911 and p = 0.1907, respectively).

Table 1 .
Demographic and clinical characteristics of the patients with oral squamous cell carcinoma (OSCC) receiving adjuvant concurrent chemoradiotherapy.

Table 2 .
Univariate association of demographic and clinical factors with survival in patients with OSCC receiving adjuvant concurrent chemoradiotherapy.

Table 3 .
Univariate association between candidate single-nucleotide polymorphisms (SNPs) in mismatch repair (MMR) genes and survival in patients with OSCC receiving concurrent chemoradiotherapy.
HR, hazard ratio; CI, confidence interval; OS, overall survival; DFS, disease-free survival.a Values is less than the total due to missing variables.* Significance at p < 0.05.
CI, confidence interval; OS, overall survival; DFS, disease-free survival.a Values is less than the total due to missing variables.* Significance at p < 0.05.

Table 4 .
Multivariate association between polymorphisms of MMR pathway genes and survival in patients with OSCC receiving adjuvant concurrent chemoradiotherapy.