Human oral keratinocytes (HOK; ScienCell Research Laboratories, Inc.) were incubated with Oral Keratinocyte Medium (ScienCell Research Laboratories, Inc.) in plates pre-coated with 2 µg/cm² poly-L-lysine. DOK oral precancerous cells were grown in DMEM supplemented with 10% FBS (both Thermo Fisher Scientific, Inc.), penicillin (100 U/ml), streptomycin (100 µg/ml), glutamine (2 mM) and hydrocortisone (5 µg/ml). Oral cancer SAS, OECM1, HSC-3 and Cal-27 cells were grown in Eagle's Minimum Essential Medium (Thermo Fisher Scientific, Inc.) with glutamine (2 mM) and FBS (10%), while human malignant glioma U87MG cells (American Type Culture Collection; glioblastoma of unknown origin) were grown in DMEM supplemented with 10% FBS. All cells were incubated in 5% CO₂ at 37°C.

Total cell lysates were extracted using RIPA buffer (Thermo Fisher Scientific, Inc.) and protein concentrations were detected using the BCA protein assay (Bio-Rad Laboratories, Inc.). Protein lysates (20 µg/lane) were separated via 10% SDS-PAGE and transferred to a PVDF membrane, which was then blocked for 1 h at room temperature with 5% non-fat milk in TBS-0.1% Tween-20 (TBST). Subsequently, the membrane was incubated overnight at 4°C with primary antibodies including XRCC1 monoclonal antibody (1:1,000; cat. no. GTX83411; GeneTex, Inc.), ERCC1 monoclonal antibody (1:1,000; cat. no. GTX22356; GeneTex, Inc.), ERCC2 polyclonal antibody (1:1,000; cat. no. GTX105357; GeneTex, Inc.) and β-actin antibody (1:10,000; cat. no. GTX629630; GeneTex, Inc.). After washing three times with TBST for 10 min, the membrane was incubated with HRP-conjugated secondary antibody (1:5,000; cat. no. GTX213110-01; GeneTex, Inc.) for 2 h at room temperature. The immunoblots were visualized using Chemiluminescence Reagent Plus (PerkinElmer, Inc.) and quantified using Image Lab™ software version 5.1 (Bio-Rad Laboratories, Inc.).

To study the mRNA expression levels of XRCC1, ERCC1 and ERCC2 in oral cancer and normal oral tissues, Oncomine™ [Estilo Head-Neck: XRCC1160033_s_at (31), Peng Head-Neck: XRCC13864445 (32), Cromer Head-Neck: ERCC11902_at (33), Ginos Head-Neck: ERCC1203720_s_at (34), Peng Head-Neck: ERCC13865378 (32), Peng Head-Neck: ERCC23865301 (32), Ginos Head-Neck: ERCC2213468_at (34), Cromer Head-Neck: ERCC241095_at (33)], an integrated cancer microarray database and web-based data-mining platform (35), was used.

Between September 2002 and December 2011, a total of 98 patients with OSCC (92 men and 6 women) were enrolled from the Department of Oral and Maxillofacial Surgery of Kaohsiung Medical University Hospital (Kaohsiung, Taiwan), with a median follow-up time of 40 months (range, 2.4–137.4 months). G*Power (version 3.1.9.4; http://ps-power-and-sample-size-calculation.software.informer.com/3.1/), a freely available windows application software, was used for sample size and power estimation, with α=0.05 and estimated effect size w=0.46. A total of 98 patients were recruited in the present study to achieve sufficient power of ≥90%. The current study was approved by the Institutional Review Board of Kaohsiung Medical University Hospital (approval no. 20140158) and patient informed consent was waived by the Institutional Review Board due to the retrospective nature of the study. OSCC pathology was determined by two pathologists independently, and the final diagnoses were made using clinical and histological data. Patients without previous history of any treatment for oral cancer were included. Patients who were <18 or >80 years old were excluded. Baseline characteristic data included patient age, sex, tumor location, grade, tumor size, lymph node metastases and tumor stage; additionally, substance use, such as alcohol consumption, betel nut chewing or cigarette smoking, and adjunct treatment details were recorded (Table I). The mean age of the study group was 51.4 years and the median age was 51 years (age range, 31–76 years). Clinical staging of the patients was determined using the TNM staging system according to the 1992 criteria of the American Joint Committee on Cancer/Union for International Cancer Control (36). The primary tumor locations were buccal mucosa (77.6%) and tongue (22.4%). All of the patients received surgery as primary treatment, and some patients received adjuvant treatment, such as radiotherapy and chemotherapy. A total of 62 patients received adjuvant chemotherapy, with the chemotherapy regimen consisting of cisplatin or carboplatin with or without the addition of 5-fluorouracil or paclitaxel. A total of 45 patients received adjuvant intensity-modulated radiotherapy, and the scheduled doses were given once per day, 5 days per week. Postoperative patients received the planned course of adjuvant radiotherapy of 60–66 Gy in 2-Gy fractions to the post-operative high-risk region.

Follow-up data was retrieved and updated from case records obtained from the medical records department until October 2014. Any patient not followed up within the last six months was contacted by phone to determine their current health status. Primary endpoints were disease-free survival (DFS) and overall survival (OS), with DFS being measured from the date of surgery to the date of locoregional recurrence, distant metastases or death from any cause, while OS was measured from the date of surgery to the date of death from any cause. Postoperative recurrence was defined as a lesion that exhibited postoperative regrowth at the same site after confirmation of healing of the surgical wounds.

To determine the expression levels of XRCC1, ERCC1 and ERCC2 in OSCC tissues by IHC staining, the tissues were fixed in 10% neutral buffered formalin for 48 h at room temperature to prepare paraffin-embedded tumor tissue blocks for IHC sections (4-mm-thick). For the normal control, the oral mucosa tissue from a delinked patient with fibroma was used after Institutional Review Board approval (approval no. KMUH-IRB-20140158). Sections were deparaffinized and rehydrated following standard methods. Briefly, the sections were deparaffinized with xylene for 5 min for three times and rehydrated in graded ethanol (80–100%) for 5 min. A microwave antigen retrieval procedure was performed for 20 min in citrate buffer (pH 6.0), and 3% hydrogen peroxide was used to block non-specific peroxidase reactions at room temperature for 10 min. After washing twice with TBS, non-specific blocking was performed with Protein Block (Novolink Polymer Detection System; Leica Microsystems, Inc.) for 5 min at room temperature and washed twice with TBS. Following washing with TBS, sections were incubated with the aforementioned XRCC1 monoclonal antibody (1:100), ERCC1 monoclonal antibody (1:200) and ERCC2 polyclonal antibody (1:100) at 4°C overnight. The sections were subsequently incubated with the Post Primary rabbit anti-mouse IgG antibody (Novolink Polymer Detection System; Leica Microsystems, Inc.) for 30 min at room temperature as secondary antibody application. The staining intensity of tumor tissues was determined as score 0 (negative), score 1 (weak), score 2 (moderate) and score 3 (strong) for antigens present in the cytoplasm and nucleus of cells using a light microscope (magnification, ×200), determined separately by two independent pathologists.

Statistical analyses were performed using SPSS 19.0 (IBM Corp.). Descriptive statistics were used and compared using independent t-test, χ² test or Fisher's exact test. Survival analyses were evaluated using the Kaplan-Meier method with the log-rank test, while the Cox proportional-hazards model was used for multivariate analysis. Furthermore, hazard ratios (HRs) and 95% CIs were calculated by multivariable Cox regression models and used to investigate the association between clinicopathological characteristics and survival. DFS was defined as the time after surgery during which the patient survived with no sign of recurrence. OS was defined as the time elapsed between surgery and death. All P-values were two-sided, with P<0.05 considered to indicate a statistically significant difference.

Since DNA repair proteins are essential for oral cancer cells in response to chemo-radiotherapy, the expression levels of XRCC1, ERCC1 and ERCC2 in HOK normal oral epithelial cells, DOK oral pre-cancer cells and oral cancer SAS, OECM1, HSC-3 and Cal-27 cell lines were first examined using western blotting (Fig. 1), with U87MG glioblastoma cells being included as a positive control for XRCC1, ERCC1 and ERCC2 proteins (37). The results revealed that the expression levels of XRCC1 and ERCC1 were weakly detected in all oral cancer cells, while ERCC2 expression was detected in all cell lines (Fig. 1). Notably, HOK cells had lower ERCC2 protein expression compared with DOK cells and the oral cancer cell lines.

In the present study, 98 patients with OSCC (92 men and 6 women) were included. The clinicopathological characteristics of these patients, including age, sex, tumor location, grade, tumor size, lymph node metastases, tumor stage, radiotherapy, chemotherapy, smoking habit and recurrence, are shown in Table I. XRCC1, ERCC1 and ERCC2 expression in oral cancer tissues, as well as in normal oral mucosa tissue of a patient with fibroma, was analyzed using an online database and IHC analysis. The Oncomine database revealed that the mRNA expression levels of XRCC1, ERCC1 and ERCC2 were significantly increased in oral cancer tissues compared with in normal epithelial tissues (Fig. S1), except in the Peng Head-Neck XRCC1-3864445 dataset, where there was no significant difference, but there was still a trend toward XRCC1 elevation in oral cancer tissues (P=0.061; Fig. S1A). Notably, IHC staining without addition of the primary antibody was used as a negative control for XRCC1, ERCC1 and ERCC2 staining (Fig. S2). According to the staining intensity in tumor tissues, the expression levels of the three proteins were categorized into four scores (0–3). XRCC1 and ERCC1 proteins were predominantly stained in the nuclei, while ERCC2 protein was predominantly stained in the cytoplasm of oral cancer tissues (Fig. 2). To analyze the association between the protein expression levels of the three proteins in OSCC tissues and clinicopathological characteristics, the expression levels of the three proteins were further divided into high expression (score 3) and low expression (score 0–2) groups. High expression groups of XRCC1 (P=0.020), ERCC1 (P=0.006) or ERCC2 (P<0.001) were significantly associated with OSCC recurrence (Tables II–IV). Furthermore, univariate and multivariate analyses were used to explore OSCC recurrence predictors in patients with OSCC. In univariate analysis, lymph node metastases (N1+N2), high XRCC1 expression, high ERCC1 expression and high ERCC2 expression were significantly associated with OSCC recurrence (P=0.006, P=0.020, P=0.006 and P<0.001, respectively; Table V). In multivariate analysis, lymph node metastases (N1+N2) (2.38-fold; 95% CI, 1.02–5.58; P=0.045) and high ERCC2 expression (4.84-fold; 95% CI, 2.56–9.16; P<0.001) were significantly associated with increased risk of OSCC recurrence (Table V).

Univariate and multivariate analyses were also used to explore OSCC survival predictors in patients with OSCC. In univariate analysis, tumor size, lymph node metastases (N1+N2), tumor stage, high XRCC1 expression, high ERCC1 expression and high ERCC2 expression were significantly associated with a worse survival in patients with OSCC (P=0.001, P=0.005, P=0.001, P=0.002, P=0.014 and P<0.001, respectively; Table VI). In multivariate analysis, tumor location (4.11-fold; 95% CI, 1.392–12.14; P=0.01) and high ERCC2 expression (15.55-fold; 95% CI, 4.34–55.67; P<0.001) were significantly associated with increased risk of death (Table VI).

The association between the expression levels of the three proteins and DFS or OS rates in patients with OSCC was analyzed using the Kaplan-Meier method. Significantly decreased DFS rates were observed in patients with high expression levels of XRCC1, ERCC1 and ERCC2 (P=0.020, P=0.040 and P<0.001, respectively), as well as significantly decreased OS rates in patients with high expression levels of XRCC1, ERCC1 and ERCC2 (P=0.002, P=0.014 and P<0.001, respectively), as determined using the log-rank test (Fig. 3). The findings of the survival analysis indicating that high ERCC2 expression indicated a poor prognosis were consistent with the results of the multivariate analysis.