The Prognostic Value of Deficient Mismatch Repair in Stage II–IVa Nasopharyngeal Carcinoma in the Era of IMRT

In the era of intensity-modulated radiotherapy (IMRT), it is important to analyse the prognostic value of deficient mismatch repair (dMMR) in nasopharyngeal carcinoma (NPC). In this study, in pretreatment biopsies of 69 patients with stage II–IVa NPC, the expression levels of MMR proteins, including MLH1, MSH2, MSH6 and PMS2, were assessed by immunohistochemistry (IHC). The median follow-up time was 37.5 months (3.1–87.4 months). 50.7% of cases (35/69) showed preserved expression of all 4 MMR proteins, which was interpreted as proficient mismatch repair (pMMR). Only 1.5% of cases (1/69) lost expression of all 4 MMR proteins, 26.1% of cases (18/69) have PMS2 loss alone and 21.7% of cases (15/69) lost expression of both PMS2 and MLH1. Thus, 49.3% of cases (34/69) lost expression of one or more MMR proteins, which was interpreted as dMMR. There was no significant difference (P > 0.05) in terms of sex, age, clinical stage, T category, N category or therapy regimens between the dMMR and pMMR groups. The multivariate Cox regression analysis revealed that dMMR was an independent significant prognostic factor for distant metastasis-free survival (DMFS) (dMMR vs pMMR: P = 0.01, HR = 0.25, 95% CI: 0.09~0.75). Therefore, NPC patients with dMMR had significantly superior DMFS compared with patients with pMMR. It can be expected that dMMR will become a new independent prognostic factor for NPC.


Treatment.
All patients were treated with the IMRT technique, and 52 (75.4%) patients were treated with additional chemotherapy. Cisplatin-based two-drug regimens were administered to 20 (29.0%) patients as asynchronous combination chemoradiotherapy (ACCRT, including induced chemotherapy or adjuvant chemotherapy). Cisplatin alone was administered to 32 (46.4%) patients as concurrent chemotherapy (CCRT).
Target volumes and organs at risk were delineated according to the consensus 14  Follow-up. Primary lesions, enlarged cervical lymph nodes and acute hematologic responses were closely observed during the treatment period. The follow-up strategy was the same as previously described 15 . All events were measured from the date of the histological diagnosis 16 . The last follow-up date was November 21, 2018. Local recurrence included primary and regional nodal recurrence. The median follow-up time was 37.5 months (3.1-87.4 months). Three patients were lost to follow up, and the follow-up rate was 95.7%. Immunohistochemistry (IHC) staining and evaluation. The tumour tissue MMR status was detected by IHC. The formalin-fixed, paraffin-embedded (FFPE) tumour tissue blocks were obtained for IHC examination using primary antibodies against MLH1 (ES05 clone, MXB, Fuzhou, China), MSH2 (RED2 clone, ZSGB-BIO, Beijing, China), MSH6 (EP49 clone, ZSGB-BIO), and PMS2 (EP51 clone, ZSGB-BIO) according to the immunohistochemical standard operating procedure (SOP). The normal IHC staining patterns for MLH1, MSH2, MSH6 and PMS2 MMR proteins were nuclear. Samples with >25% of tumour cells stained were interpreted as expression preserved of the MMR protein 17 . dMMR was interpreted as expression loss for at least one MMR protein 18 .
Statistical analysis. All data were analysed using SPSS statistical software (version 22.0; IBM, Armonk, NY). The chi-squared and Fisher's exact probability tests were used to analyse differences between qualitative data. Survival curves were plotted according to the Kaplan-Meier method and were compared using the log-rank test. A Cox proportional hazards model was used to identify significant prognostic factors. Forward LR method was use to selected variables. Statistical significance was set at P < 0.05 (two-sided).

Results
Expression of MMR proteins. The normal IHC staining patterns for all four MMR proteins were nuclear ( Fig. 1 Baseline characteristics and distribution of MMR. Of 69 patients with NPC, there was no significant difference (P > 0.05) in terms of sex, age, clinical stage, T category, N category or therapy regimens between the dMMR and pMMR groups. Histological pathology type was not analysed because only non-keratinizing squamous cell carcinoma was included ( Table 2). www.nature.com/scientificreports www.nature.com/scientificreports/ Univariate analysis of prognostic factors. Kaplan-Meier survival curves of NPC patients were analysed by log-rank test according to sex, age, clinical stage, T category, N-category, therapy regimens and MMR status ( Table 2). The analysis revealed that clinical stage (P = 0.01) or T category (P = 0.02) for overall survival (OS), age (P = 0.02) or T category (P = 0.03) for progression-free survival (PFS), age (P = 0.04) for local recurrence-free survival (LRFS), and age (P = 0.04) or T category (P = 0.04) or MMR status (P = 0.02, Fig. 2a) for distant metastasis-free survival (DMFS) were identified as significant prognostic factors (Table 2). In addition, the influence of N category or MMR status (Fig. 2b) on OS (P = 0.06), and clinical stage on PFS (P = 0.05) were the same as the cutoff values (Table 2).  Table 3) were independent significant prognostic factors. The prognostic value of T category for DMFS was marginally statistically significant (T1-2 vs. T3-4, P = 0.05, HR = 0.13, 95% CI: 0.02~1.02) (Fig. 3f, Table 3).

Discussion
In living cells, MMR proteins play an important role in not only safeguarding genetic stability by excising DNA mismatches introduced by DNA polymerase during replication, but also responding to and repairing cellular DNA damage [4][5][6][7][8] . For a start, in human cells, MMR proteins are recruited immediately to the sites of various types of DNA damage including single-strand breaks (SSBs), double-strand breaks (DSBs) and pyrimidine dimers 4-7 , for which recruitment is mediated by protein-protein interactions in nucleotide-excision-repair dependent or function domain of MMR protein dependent manners 4 . Then, the recruitment leads to degradation of licensing factor Cdt1 (a G1-specific cell-cycle regulatory protein) in the G1 phase and efficient repair of DNA damage 6 , and apoptosis is activated via the mitochondria and p53-independent mechanisms 7 . Furthermore, MMR proteins are involved in the responses to the genotoxicity of γ-radiation in human cells, including thymidine kinase gene mutation, micronucleus formation or apoptosis, through inducing the expression of p53 and delaying the cell cycle 8   www.nature.com/scientificreports www.nature.com/scientificreports/ because of the compensatory effects of other MMR proteins. MSH2 expression is preserved if MSH6 is lost, and MLH1 expression is preserved if PMS2 is lost 22,23 . This law was also observed in our studies. Furthermore, in our study, the expression of MSH2 and MSH6 was loss in only one patient (1.5%), and all the MLH1/PMS2-double preserved patients (50.7%) were the same as the pMMR patients, 26.1% of the patients showed the loss of PMS2 alone. Thus, loss of PMS2 expression may be the dominant factor for the dMMR in NPC. The multivariate Cox regression analysis confirmed this hypothesis (data was not shown).
MMR-defective cell lines are more resistant to cell death induced by several DNA-damaging agents, including methylation agents, cisplatin and UV radiation, whereas they are more sensitive to cell death caused by interstrand crosslinking agents 4 . Radiotherapy can cause many types of DNA and gene damage, including loss of heterozygosity, homozygosity deletion 1 , DSBs 2,3 . Misrepair of radiation -induced DNA damage underlies genomic instability and increased radiosensitivity 36 ; thus, MMR-defective cells may underlie increased radiosensitivity. www.nature.com/scientificreports www.nature.com/scientificreports/ Previous studies have reported that compared with patients with pMMR tumours, colorectal cancer patients with dMMR tumours had significantly superior OS and PFS 24,37 , and endometrial cancer patients with dMMR tumours had significantly superior PFS 38 . In this study, there was no significant difference (P > 0.05) between the dMMR and pMMR groups in sex, age, clinical stage, T category, N category or therapy regimens. Univariate and multivariate analyses of the prognostic factors showed that patients with dMMR tumours had significantly superior DMFS compared with patients with pMMR tumours (dMMR vs. pMMR, P = 0.01, HR = 0.25, 95% CI: 0.09~0.75) (Fig. 3e, Table 3). Therefore, dMMR promises to be a potential prognostic biomarker for NPC. The better prognosis of dMMR NPC patients might result from a stronger immunologic response driven by abundant tumour infiltrating lymphocytes (TILs) in the tumour microenvironment 39 . Immunotherapy has been approved for the treatment of patients with microsatellite instability-high (MSI-H) or dMMR solid tumours in 12 different tumour types, although NPC was not included in the 12 cancers 40 . In addition, several studies confirmed that dMMR cases had reduced levels of vascular endothelial growth factor (VEGF) compared to pMMR cases, which might partly explain why patients with dMMR tumours had more favourable prognosis 39 . www.nature.com/scientificreports www.nature.com/scientificreports/ In the IMRT era, distant metastasis predominates as the pattern of disease relapse in patients with stage II-IVa NPC, accounting for approximately 70% of cancer-specific mortality 41 . In this study, we found that dMMR was indicative of a favourable DMFS in patients with stage II-IVa NPC. It can be expected that dMMR will become a new independent prognostic factor of NPC. However, prospective clinical studies are needed to further investigate the prognostic value of dMMR in patients with advanced NPC and the role of dMMR in NPC immunotherapy. Thus, NPC should be screened for dMMR.

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.