TFAP2E and MLH1 Genes Methylation Pattern and Microsatellite Instability as Predictors of Rectal Cancer Response to Neoadjuvant Chemoradiotherapy

Background: Neoadjuvant chemoradiotherapy (nCRT) prior to surgery in rectal cancer has several adverse effects. Predictive biomarkers for response to nCRT are needed to save patients unnecessary toxicities and to take a timely tailored treatment decision. Epigenetic modifications like DNA methylation patterns have been suspected to be potential predictive biomarkers. Aim: To determine the role of TFAP2E and MLH1 genes’ methylation status and microsatellite instability (MSI) in predicting response to 5-fluorouracil – based nCRT in rectal cancer. Methods: DNA was extracted from 80 patients with newly diagnosed stage II / III rectal cancer. The methylation status of TFAP2E and MLH1 genes was determined by pyrosequencing and MSI was determined using 5 microsatellite loci by conventional polymerase chain reaction and capillary electrophoresis. Results: The cut-off values for TFAP2E & MLH1 genes ’methylation level were 40% and 15% by receiver operating characteristic curve analysis. Hypermethylated TFAP2E and MLH1 gene promotors and MSI were predominant among non-responders (p <0.001, <0.001 and =0.022; respectively). Other factors associated with significantly higher pathological response to nCRT were well/moderately differentiated adenocarcinoma, pretreatment carcinoembryonic antigen level ≤5 ng/ml and rectal tumor ≤5 cm from the anal verge. Conclusion: Hypermethylated TFAP2E and MLH1 gene promotors and MSI in rectal cancer tissue were associated with poor response to 5-fluorouracil – based nCRT. They might be of value in predicting the response of rectal cancer to nCRT and in tailoring its treatment.


Introduction
Colorectal cancer (CRC) is a disease with a major worldwide burden. It is the third most common cancer diagnosed in both males and females. It has heterogeneous clinical presentations, prognosis and individual treatment response 1 .
Rectal cancer has a striking complication, which is the high recurrence rate; 4-33% 2 . Aiming at reducing the recurrence rate and downstaging the tumor prior to surgery, neoadjuvant chemoradiotherapy (nCRT) is recommended for a group of rectal cancer patients by the National Comprehensive Cancer Network (NCCN) and others 3 ,4 .
The major benefits of pre-operative nCRT are to reduce tumor mass, decrease the thickness of invasion of tumor into the intestinal wall and achieve clinical down-staging of tumor 5 . However, its outcome varies from a favorable 'pathological complete response', which occurs in 10-20 % of patients, to 'no benefit from treatment' and delayed opportunity of surgery with tumor growth during this period and even the development of distant metastasis 6,7 .
Considering the severe secondary effects of nCRT observed, reliable predictive biomarkers are strongly needed to predict the response of patients to the treatment and to identify non-responders and thus avoiding costly, long, and ineffective therapy with its adverse effects and getting benefit from other personalized treatment 8 .
The recent introduction of epigenetic studies revealed characteristic DNA methylation pattern of cytosine guanine islands (CGIs) in rectal cancer. With these advances, a major potential lies in the use of DNA methylation patterns as a reliable and effective diagnostic, prognostic and predictive biomarkers to the response to 5-flurouracil (5-FU)based nCRT in rectal cancer 9 .
One of these predictive biomarkers is the transcription factor activating protein 2 epsilon (TFAP2E) gene which is located on chromosome 1p34 and has multiple CGIs with the potential for regulation of its expression by means of its CGIs' hypermethylation.
It was found that hypermethylation of this gene is associated with its decreased expression, that is associated with clinical non-responsiveness to 5FU-based nCRT. Moreover, induction of TFAP2E expression was triggered by using de-methylating drugs 10,11 .
Another one is the mut L homolog 1 (MLH1) gene, which is located on chromosome 3p22.2. It is a member of a group of genes known as DNA mismatch repair (MMR) genes that are responsible for recognition and repairing of DNA errors that occur during DNA replication 12 .
It was found that MLH1 gene hypermethylation with subsequent silencing of this gene correlates with loss of the MLH1 protein and DNA repairing mechanisms and provokes a characteristic molecular phenotype called microsatellite instability (MSI), displaying a form of genetic instability characterized by the accumulation of numerous mutations within repetitive sequences of DNA in non-encoding microsatellite regions, which has a predictive role in response to nCRT 13 ,14 .
It was reported that patients with MLH1 gene hypermethylation with MMR-deficient tumors and MSI did not respond to 5-FU-based nCRT, whereas those with MMR-proficient tumors and microsatellite stability (MSS) had a significant response to this treatment 15 . Moreover, it was reported that treatment of rectal cancer by demethylating agents (which de -methylate genes like MLH-1) results in enhanced chemo-radiation sensitivity 16 .
So, epigenetic methylation markers are suspected to be predictive markers to treatment response in rectal cancer. This information could be critical in the choice of chemotherapeutic treatment, especially considering recently developed new inhibitors of DNA methylation 17 .
Insufficient attention has been paid to this important clinical issue. Moreover, to the best of our knowledge, published studies did not separate rectal cancer and studied rectal and colon cancer together 18 .
In this study, we aimed at determining the predictive role of methylation markers (TFAP2E and MLH1 genes) and MSI status in response to nCRT among Egyptian patients with rectal cancer.

Patients
This study was conducted on 80 Egyptian patients with newly diagnosed rectal cancer stages II and III by magnetic resonance imaging (MRI) and biopsy, admitted to the Colorectal Surgery Department at Alexandria Main University Hospital, Alexandria, Egypt. All were scheduled to receive nCRT before surgery.
Full history was taken from all patients including; personal history (name, age, diet habits, smoking, alcohol intake and physical exercise), family history (CRC, gastro-intestinal, breast and genital cancers, age at diagnosis and outcome), past history (of exposure to previous chemotherapy, radiotherapy, X-ray, CT or MRI on colon and colorectal biopsy) and present history (regarding the main complaints as tenesmus, mucorrhae, anal bleeding, discharge, pain, abnormal bowel habits, weight loss and their duration). Digital rectal examination was done to assess the distance of the rectal mass from the anal verge.
They were dissolved in high purity water.

Statistical analyses
Data were fed to the computer and analyzed using IBM SPSS software package version 20.0. (Armonk, NY: IBM Corp). Qualitative data e.g. gender of patients was described using number and percent. The Kolmogorov-Smirnov test was used to verify the normality of distribution. Quantitative data e.g. age of patients was described using median and range (minimum and maximum) or mean and standard deviation. Significance of the obtained results was judged at the 5% level. The used tests were; Chi-square test to compare between different groups with categorical variables, Fisher's Exact or Monte Carlo correction for chi-square when more than 20% of the cells have expected count less than 5, Kappa test for agreement testing. Receiver operating characteristic (ROC) curve was generated by plotting sensitivity (true positive) on Y axis versus 1-specificity (false positive) on X axis at different cutoff values. The area under the ROC curve (AUC) denotes the diagnostic performance of the test. Area more than 50% gives acceptable performance and area about 100% is the best performance for the test. The ROC curve allows comparison of performance between two tests.

Ethical considerations
The study was approved by the Institutional Human Research Ethics Committee of the Faculty of Medicine -Alexandria University (Approval #: 020870) and a written informed consent was obtained from all participants before enrolment in the study. Patients were all coded and detailed information was stored confidentially.

Results
Among the 80 included patients, 34 (42.5 %) were males and 46 (57.5%) females and their age ranged from 41 to 70 years with a mean of 57.1 years.
Assessment of response to nCRT by pathological TRG and radiological RECIST criteria were in good agreement with each other by kappa test (0.751 with p<0.001). Pathological response is more sensitive to changes at cellular level. The relation between clinico-pathological data and pathological response to nCRT are summarized in Table 1. Factors associated with significantly higher pathological response to nCRT were well / moderately differentiated adenocarcinomas, pretreatment level of CEA ≤5 ng/ml, rectal tumors ≤5cm from the anal verge, microsatellite stability, hypomethylated TFAP2E gene and hypomethylated MLH1 gene.

Figure 3: ROC curve analysis of methylation % values for TFAP2E and MLH1 genes
There was no statistically significant association between methylation level of the TFAP2E gene promotor and age, gender, TNM stage and histological type (Table 2). Similarly, there was no statistically significant association between these variables and the methylation level of the MLH1 gene promotor (Table 2).
Hypomethylated MLH1 gene (≤15 %) was significantly associated with MSS of rectal tumors (p=0.011). There were significant associations between gender of patients, histological type and microsatellite status of rectal tumors with predominance of MSI among females and those having mucinous carcinoma ( Table 2).

Discussion
Hypermethylated TFAP2E and MLH1 gene promotors were found to be predominant among rectal cancer patients who did not respond to nCRT, while hypomethylated genes were predominant among responders. We did not find statistically significant associations between methylation levels of both TFAP2E and MLH1 genes and clinico-pathological characteristics of patients (age, gender, histological tumor type and TNM stage).
In agreement with our study, Ebert et al 10 stated that the probability of response to 5-FU-based nCRT was six times higher among rectal cancer with TFAP2E gene hypomethylation (≤30% methylation level) than among rectal cancer with TFAP2E gene hypermethylation (<30% methylation level). Also, they reported that there was insignificant association between methylation level of TFAP2E gene and gender, histological type and TNM stage of rectal cancer. Murcia et al 28 conducted their study on 783 patients with stages II-IV CRC who received 5-FU-based CRT; 532 patients from a populationbased multicenter cohort and 251 patients from a clinic-based trial. They aimed at evaluating the effectiveness of TFAP2E methylation and expression as a predictor of response of CRC patients to 5-FU-based CRT. The methylation status of TFAP2E gene was assessed by quantitative bisulfite pyrosequencing technique using PSQ HS 96A pyrosequencing system (QIAGEN). Optimal methylation cutoff was calculated by ROC curve analysis and was 40% with an AUC of 0.73. In contrast to our results, they reported that TFAP2E gene methylation level, either hypermethylated   (>40%) or hypomethylated (≤40%), did not significantly play a major role in predicting response to 5FU-based CRT in rectal cancer patients (p=0.2). This dissimilarity in results may be due to differences in ethnic groups, larger sample size (783 vs 80) and selection criteria of patients (stages II-IV vs stages II and III CRC). However, in agreement with our study, they reported that none of the clinico-pathological characteristics of patients enrolled in their study was significantly associated with TFAP2E gene methylation level. We found that hypermethylated MLH1 gene was significantly associated with MSI of rectal tumors. However, none of the clinico-pathological data such as age, gender, histological type and TNM stage of the tumor was significantly related to MLH1 gene methylation level.
Arnold et al 29 also reported that hyper methylation of MLH1 gene is significantly associated with the MSI phenotype in CRC. Also, Moreira et al 30 reported similar results. They found that MLH1 gene hypermethylation (<15%) was prevalent in MSI tumors and MLH1 gene hypomethylation (>15%) was present in MSS tumors. They also reported that some of the clinicopathological data, including age, gender and histological type of the tumor were not significantly related to MLH1 gene methylation status.
In our study, we found that MSS rectal tumors had a good response to 5-FU-based nCRT, while MSI tumors had lower response. MSI rectal tumors were predominant among female patients and patients with mucinous carcinoma with no statistically significant relation to either patients' age or rectal tumor stage.
Yang et al 15 conducted their study on patients with stage II rectal adenocarcinoma who received adjuvant therapy. In agreement with our study, they reported that 5-FU-based CRT had more benefits in MSS rectal cancer, but not with MSI. These results were supported by those of Sargent et al 14 and Ribic et al 31 . Ribic et al 31 reported that some of the clinico-pathological variables including age and tumor stage were not significantly related to microsatellite status of the tumor, in contrast with the poorly differentiated histological type of the tumor that was reported to be significantly related to its MSI.
In contrast to our study, Thomas et al 32 reported that the benefit from 5-FU-based CRT in patients with rectal cancer was not significantly related to microsatellite status of the tumor, whether stable or not, and that the beneficial effect of 5-FU-based CRT was found in both MSI and MSS tumors. This dissimilarity in results may be due to ethnicityrelated factors, difference in sample size (814 vs 80) and selection criteria of patients (stage C CRC, according to Australian Clinico-pathological Staging System, who received 5-FU-based chemotherapy vs stages II and III rectal cancer receiving nCRT). However, in agreement with our results, they reported that some of the clinico-pathological data enrolled in their study including female gender and poorly differentiated histological type of the tumor were significantly related to MSI of the tumor.
Among the strengths of the current study are the multiple clinical and laboratory factors studied to predict the response of rectal cancer patients to nCRT and the strict exclusion and inclusion criteria adopted for the selection of cases to avoid confounding factors that could affect the results as much as possible. Another strength is that the response to nCRT was assessed both pathologically and radiologically. The main limitation was the relatively small sample size due to limited budget.
To the best of our knowledge, this is the first study conducted on Egyptian patients with rectal cancer aiming at determining the role of some clinical and laboratory markers; TFAP2E and MLH1 genes and MSI in predicting their response to 5-FUbased nCRT.

Conclusion
Hypermethylation of TFAP2E and MLH1 gene promotors and MSI of rectal cancer are associated with a lower response rate to 5-FU-based CRT, indicating that they are potential predictors of treatment response in patients with rectal cancer who might benefit in the future from demethylating agents.