Formalin-fixed and paraffin-embedded tissues from 37 cases of salivary CXPA with complete clinical and pathological data were retrieved from the Department of Oral Pathology at Shanghai Ninth People's Hospital, Shanghai Jiao Tong University in Shanghai, China. Tissue sections (4 µm) were stained with hematoxylin and eosin (H&E) and were reviewed by two investigators. The tumors were histologically examined and classified as high or low grade (19). High-grade tumors exhibited ≥2 of the following features: i, anaplasia with nuclear pleomorphism and prominent nucleolus; ii, frequent mitoses: ≥5 per 10 high-power fields; iii, atypical mitosis; and iv, extensive coagulative tumor necrosis. The clinical stage of each patient's disease was determined according to criteria of the tumor-lymph node-metastasis (TNM) classification system (2002) International Union Against Cancer (20). CXPAs could be classified into 2 main subtypes according to their morphological and immunohistochemical features. The classification of CXPA-L and CXPA-NL in our study followed the methods detailed by Kim et al (19). This study was approved by the ethics committee of Shanghai Jiao Tong University.

For our in vitro experiments, 2 CXPA cell lines (SM-AP1 and SM-AP4) (21) were cultured in DMEM supplemented with 100 U/ml penicillin, 100 mg/ml streptomycin, 2 mM glutamine and 10% fetal bovine serum (FBS) and were incubated at 37°C in a humidified atmosphere containing 5% CO₂. Induction of CDH1 promoter hypermethylation in SM-AP1 cancer cells was initiated by the addition of 10 ng/ml TGF-β1 (Peprotech, NJ, USA) to the medium for up to 72 h. Induction of CDH1 promoter demethylation in SM-AP4 cancer cells was initiated by the addition of a demethylation agent, 5-Aza-2′-deoxycytidine (5-Aza-dC; Selleck, TX, USA).

Genomic DNA was extracted from paraffin-embedded tumor tissues and cultured cells using the QIAamp DNA formalin-fixed paraffin-embedded (FFPE) tissue kit (Qiagen, Duesseldorf, Germany) and the QIAamp DNA Mini kit (Qiagen), respectively, according to the manufacturer's instructions. We selected regions of malignant tumor as much as possible when extracting DNA from tissue sections. Extracted DNA was treated with sodium bisulfite using the EpiTect Bisulfite kit (Qiagen) according to the manufacturer's protocol. Nested PCR primer sequences were as follows: first round, 5′-GGTAAAAGAAAAAAAAATTAGTTTG-3′ and 5′-AATACCTACAACAACAACAACAA-3′; second round, 5′-TAGAGAGGTTGGGGTTAGAG-3′ and 5′-AACCCCTCCCCAAAACRAAACTAA-3′. Methylation status was assessed at the CDH1 promoter region by sequencing the PCR-amplified bisulfite-treated DNA using the automated ABI PRISM 3730XL DNA sequencer (Applied Biosystems, USA), as previously described (22). The CDH1 promoter region studied here contains 10 CpG dinucleotides located in the -571- to -230-bp fragment upstream of the transcription start site. Ten random clones were selected from each sample for sequencing. Methylation status was defined as low [methylation rate (MR)≤20%], medium (20%<MR≤40%), or high (MR≥40%).

Immunohistochemistry (IHC) was performed on 4-µm paraffin-embedded sections according to the protocol. An anti-E-cadherin receptor antibody (monoclonal mouse anti-human, dilution 1:200; Life Technologies, USA) was applied as the primary antibody for IHC detection. The IHC procedure was performed by the Envision™ method (Dako, Glostrup, Denmark) according to the manufacturer's protocol. In the negative control samples, primary antibodies were replaced by PBS. Normal salivary gland tissue slices served as a positive control. In the CXPA samples, E-cadherin was located on the cell membrane and in the cytoplasm. Five random high-power fields were chosen from every slice to assess the E-cadherin score. The score of each slice was based on the percentage and intensity of positively stained cells. The percentage scoring system was as follows: no positive cells (0), <50% positive tumor cells (1), 50–75% positive tumor cells (2), and >75% positive tumor cells (3). The intensity scoring system was as follows: no staining (0), light yellow (1), yellow brown (2), and dark brown (3). The percentage score was multiplied by the intensity score and sections were divided into 2 groups based on the resulting product, as follows: low expression (score ≤6) and high expression (score >6). IHC slides were scored by two pathologists without knowledge of the clinical data in order to eliminate bias. Discrepancies were eliminated by consensus.

Western blotting and qRT-PCR were carried out as previously described (23,24). Protein lysates were separated by 10% SDS-PAGE and electrophoretically transferred to a PVDF membrane (Millipore, MA, USA). Subsequently, the membrane was incubated with a primary monoclonal antibody followed by a fluorescent secondary antibody. β-tubulin was used as a protein loading control. Primary antibodies used for western blotting included those against E-cadherin (Abcam), vimentin (Abcam), and β-tubulin (Santa Cruz, CA, USA). Western blot bands were visualized using Imaging system (LI-COR Biosciences, Lincoln, NE, USA), and protein density was quantified using Odyssey version 1.2 software (LI-COR Biosciences). qRT-PCR was performed using SYBR-Green PCR master mix (Applied Biosystems) on an ABI 7300 system. PCR primers were as follows: CDH1 (5′-AGAACAGCACGTACACAGCCCTAA-3′ and 3′-ATCAGCAGAAGTGTCCCTGTTCCA-5′) and β-actin (5′-CTCCATCCTGGCCTCGCTGT-3′ and 3′-GCTGTCACCTTCACCGTTCC-5′).

The data were analyzed using version 13 of the Statistical Package for Social Sciences (SPSS Inc., Chicago, IL, USA). Quantitative data were summarized using the means and standard deviations and were compared using the Student's t-test. Qualitative and ranked data were compared using the χ² test. Associations between clinicopathological variables and CDH1 promoter methylation status were evaluated using Pearson's χ² test. Patient survival analysis was performed by the Kaplan-Meier method, and differences were evaluated with the log-rank test. Hazard ratios (HR) and their 95% confidence intervals (CI) were estimated using univariate and multivariate Cox proportional hazard models. All statistical analyses were considered significant when the P-value was ≤0.05.

A total of 37 CXPAs from 26 males (70.27%) and 11 females (29.73%) were investigated in this study. The male to female ratio was 2.36. The age range of the patients was 26–83 years, and the mean age was 61.62 years. Twenty-nine tumors (78.38%) originated from the major salivary glands, and 8 tumors (21.62%) originated from the minor salivary glands. Histologically, 16 (43.24%) tumors were classified as low grade, and 21 (56.76%) were classified as high grade. Perineural invasion was observed in 11 of 37 patients (29.73%). Sixteen patients (43.24%) developed lymph node metastases. The mean follow-up time was 28.86 months. There were 17 deaths, 16 patients died of CXPA and 1 of another disease. Of those that survived, 17 patients survived without tumors and 3 survived with tumors.

The methylation status of CDH1 was analyzed in 37 salivary CXPA tissues using BSP. As shown in Table I, low, medium and high methylation of the CDH1 promoter (Fig. 1) was found in 16 (43.24%), 9 (24.32%) and 12 (32.43%) CXPA samples, respectively. IHC analysis was performed to investigate E-cadherin expression. The results showed that 13 (35.14%) of 37 cases had low E-cadherin expression, while 24 (64.86%) cases showed high E-cadherin expression (Table I and Fig. 2). As shown in Table I, we found that CDH1 promoter methylation was significantly lower in the high E-cadherin expression group as compared with the low E-cadherin expression group (P<0.01). This result indicates that the methylation status of CDH1 strongly correlates with E-cadherin expression.

CDH1 promoter methylation was detected in both SM-AP1 and SM-AP4 cell lines (Fig. 3A). Consistent with the notion that methylation contributes to gene inactivation, CDH1 hypermethylation (MR=48%) status resulted in lower CDH1 mRNA and protein expression in SM-AP4 cell lines than (MR=23%) in SM-AP1 cell lines (Fig. 3B and C). When treated with 5-Aza-dC, a demethylating agent, SM-AP4 cells showed increased CDH1 mRNA and protein expression and decreased CDH1 promoter methylation (Fig. 4A–C). However, SM-AP1 cells showed decreased CDH1 mRNA and E-cadherin expression but an elevated methylation status after TGF-β1 treatment (Fig. 4D–F). Taken together, we suggest that promoter methylation is a predominant factor regulating CDH1 expression in CXPA cell lines.

To evaluate the clinical significance of CDH1 promoter methylation, we investigated the association between methylation status and clinicopathological features in CXPA patients. As presented in Table II, CDH1 methylation status was differentially detected according to sex, histological subtype, histological grade, and tumor N-stage and TNM-stage. CXPA cases with high histological grade (42.9% versus 7.1%, P=0.005), lymph node metastasis (56.2% versus 14.3%, P=0.024), or advanced TNM-stage (41.7% versus 7.7%, P=0.038) were more likely to display high CDH1 promoter methylation, which indicates that promoter methylation may be a prognostic factor in CXPA. Interestingly, compared with males, females tended to present with higher CDH1 methylation rates (P=0.028). CDH1 methylation status was not significantly correlated with other clinicopathological parameters, such as age, tumor site or neural invasion.

Survival curves were generated for all 37 salivary CXPA cases. Methylation of the CDH1 promoter was significantly associated with overall survival (log-rank test, P=0.026) (Fig. 5). In univariate analyses, lymph node metastasis (P= 0.004) and CDH1 promoter hypermethylation (P= 0.030) were significantly associated with poor overall survival (Table III). To determine whether the association between CDH1 promoter methylation and survival was independent of other parameters, a multivariate analysis was performed including N-stage and CDH1 promoter methylation as co-factors. The multivariate analysis showed that lymph node metastasis (P=0.010) is independently associated with overall survival (Table III) and is an independent prognostic factor in CXPA.