To determine whether the DNA-methylation status of CDH1 or CDH13 in serum samples has a diagnostic value for cervical cancer, we used MethyLight analysis to investigate the frequency of DNA-methylation of CDH1 and CDH13 in serum samples from 40 patients with non-malignant diseases and 49 cervical cancer patients. In non-malignant serum samples, aberrant DNA-methylation for CDH1 was present in 10 of 40 serum samples (75% specificity). Among the patients with invasive cervical cancer, the DNA-methylation frequency for CDH1 was 55% (27 of 49; 55% sensitivity). For CDH13, we identified in non-malignant serum samples aberrant DNA-methylation in two of 40 serum samples (95% specificity), but in only five of 49 serum samples from patients with invasive cervical cancer (sensitivity 10%).

A statistically significant higher frequency of DNA-methylation for CDH1, but not for CDH13, was observed in cervical cancer patients in comparison to patients with benign diseases (p = 0.004).

To determine whether any prognostic significance was connected to differences in the detection of CDH1 or CDH13 DNA-methylation in serum samples of cervical cancer patients, we compared the clinical outcomes of cervical cancer patients with and without CDH1 or CDH13 DNA-methylation. Neither CDH1 nor CDH13 DNA-methylation correlated with clinicopathological features (Table 1).

Furthermore, we compared the CDH1 results of the MethyLight assay with the results of a DHPLC PCR in all 89 serum samples. We found only a modest correlation between these data (r = 0.496; p < 0.001). Only 15 of all 37 (41%) MethyLight positive samples could also be detected by means of DHPLC PCR, whereas 15 of the 16 (94%) DHPLC positive samples were detected by MethyLight PCR.

Moreover, the DHPLC method showed a statistically significant higher frequency of DNA-methylation for CDH1 in cervical cancer patients than in patients with benign diseases (p = 0.020). Among the patients with invasive cervical cancer, the DNA-methylation frequency for CDH1 was only about 27% (13 of 49). In non-malignant serum samples, we detected aberrant DNA-methylation in three of 40 serum samples (7.5%). Consequently, we identified a specificity of 92.5% and a sensitivity of only 27% for the CDH1 DHPLC PCR analysis. With this method we were not able to identify a statistically significant difference between cervical cancer patients and patients with benign diseases.

Cervical cancer is a leading cause of cancer-related death in women. Several clinical and histopathological characteristics, i.e., tumor stage, lymph node status, and vascular invasion, have been shown to be prognostic factors for recurrent disease [16,17]. New objective diagnostic, prognostic and predictive biomarkers for cervical cancer are needed.

Recently we identified CDH1 and CDH13 DNA-methylation in serum samples taken at the time of diagnosis as an independent prognostic marker in cervical cancer patients with no concurrent chemo- or radiation therapy [15].

In the present study, the methylation status of CDH1 and CDH13 genes in serum samples from 49 cervical cancer patients who were treated according to the guidelines and 40 patients with diseases other than cancer was reevaluated using MethyLight assay technology. The DNA-methylation frequency for CDH1 was 55% in the serum samples from patients with invasive cervical cancer. In our previous study, we identified a frequency of 42% in serum [15]. Other groups describe frequencies of about 51.1–80.5% in cervical cancer tissue samples but not in serum samples. [18,19]. For CDH13 methylation, a frequency of only 10% was observed in serum samples from cervical cancer patients. Distribution of CDH1 but not CDH13 methylation within the cervical cancer and non-cancer group showed statistically significant differences. For the detection of cervical cancer based on the CDH1 methylation analysis in serum samples, however, the specificity and sensitivity values of 75% and 55% are not convincing. Therefore the serological detection of CDH1 and CDH13 DNA-methylation in serum does not appear to be a suitable method for cervical cancer detection. There is currently no methylation marker that can be readily translated for use in cervical cancer screening or triage settings [20].

To determine whether the methylation status of CDH1 and CDH13 in serum samples has also a prognostic value for cervical cancer, we compared serum DNA-methylation of these genes with the overall and relapse-free survival of patients. We found that there was a significant difference in the overall and the relapse-free survival for patients with methylated CDH1 DNA in their pre-treatment serum samples. These findings are in accordance with our previously published study where only patients without concurrent chemo- or radiation therapies were analyzed [15]. In that study, we observed a higher CDH1 DNA-methylation frequency in serum samples of patients with FIGO Stage III and no association with tumor grade. In the present study, we found only a non-statistically significant trend of a higher CDH1 DNA-methylation frequency in serum samples of patients with FIGO IV and again no association with tumor grade. The low number of patients in the present study may explain the slight difference compared to the previous study. In our present study the combination of CDH1 and CDH13 DNA-methylation data did not improve the survival results. However, further studies with larger numbers of patients are required to confirm our findings.

Finally, we compared MethyLight technology and DHPLC-PCR for CDH1 DNA-methylation analysis. Using DHPLC-PCR, we detected only 41% of all MethyLight positive samples, but the majority (94%) of the DHPLC positive samples was also detected by MethyLight PCR.

The comparison of methods showed a higher detection rate by MethyLight PCR. Using DHPLC-PCR for CDH1, it was not possible to distinguish statistically significantly between patients with cervical cancer and patients with benign diseases. The high sensitivity of MethyLight PCR technology could be a reason for the discrepancy in the methylation frequencies detected by MethyLight and DPHLC. MethyLight PCR can detect a single methylated allele in 10⁵ unmethylated alleles [21]. Previously, Eads et al. showed that MethyLight PCR is at least 10 times more sensitive than reported for the highly sensitive MSP technology [21].

The findings described in this manuscript suggest that the more labor-intensive DHPLC-PCR is not an ideal method for CDH1 methylation analysis in serum samples.

For this retrospective study we analyzed serum samples from 49 cervical cancer patients (ages 31.4–93.4 years; median age, 60.5 years) and 40 patients with non-malignant gynecological diseases (ages 30.7–85.8 years; median age 56.7). All patients were treated at the Department of Obstetrics and Gynecology, Innsbruck Medical University, Austria. Treatment was conducted according to international standards. All patients were monitored within the department’s outpatient follow-up program. The median observation period of the cervical-cancer patients was 2.2 years (1 month to 8.0 years). Overall, 24% of the patients had surgery (n = 12), 53% received chemotherapy (n = 26) and 78% received radiotherapy (n = 38). Radiation therapy was applied in combination with chemotherapy in 49% (n = 24) of patients and 18% of patients received no adjuvant therapy (n = 9). The study was approved by the ethics committee of the Innsbruck Medical University (reference number: AN3568) and conducted in accordance with the Declaration of Helsinki.

All serum samples were taken on the date of diagnosis and before initial treatment. Serum samples were stored at −50 °C until analysis was performed. The median storage age of the serum samples from cervical cancer patients was 4.4 years (5.5 months to 5.8 years), and 4.9 years (1.2–5.9 years) for patients with non-malignant gynecological diseases.

DNA from 1 mL serum samples was isolated using the ChargeSwitch gDNA kit from Invitrogen (Life technologies, Paisley, UK) according to the manufacturer’s protocol. Sodium bisulfite conversion of genomic DNA was performed using the EZ DNA Methylation-Gold KitTM (Zymo Research, Orange, CA, USA) according to the manufacturer’s recommendations. Sodium bisulfite-treated genomic DNA was analyzed by means of the MethyLight method, a fluorescence-based, real-time PCR assay, as previously described [15,21]. Two sets of primers and probes, designed specifically for bisulfite-converted DNA, were used: a methylated set for the gene of interest and a reference set, collagen (COL2A1), to normalize for input DNA. Specificity of the reactions for methylated DNA was confirmed separately using SssI-treated (New England Biolabs, Beverly, MA, USA) human white blood cell DNA, which is heavily methylated. The percentage of fully methylated molecules at a specific locus was calculated by dividing the GENE:COL2A1 ratio of a sample by the GENE:COL2A1 ratio of SssI-treated white blood cell DNA and multiplying by 100. The abbreviation PMR (percentage of fully methylated reference) indicates this measurement. For each MethyLight reaction, 10 μL of bisulfite-treated genomic DNA were used. A gene was deemed unmethylated if the PMR value was 0. To verify the reproducibility of each assay, the normalized value (GENE:COL2A1) of the standard sample was checked for all PCR runs. The primers and probes used for MethyLight reactions have been described recently [15].

The primers used in PCR for DHPLC were determined with the assistance of the software Methyl Primer Express v1.0 (Applied Biosystems, Foster City, CA, USA). The primers are designed to amplify 403 bp (CDH1) and 348 bp (CDH13) of bisulfate-modified DNA, irrespective of methylation status. CDH1 Forward primer: 5′-TTTTAGTTTGGGTGAAAGAGT-3′; CDH1 Reverse Primer: 5′-AACTCACAAATACTTTACAATTCC-3′; CDH13 Forward Primer: 5′-TTTGTTTTAGGTAGGGAAGAGG-3′; CDH13 Reverse Primer: 5′-AAAACCAAAATTACCCCACTTA- 3′.

PCR reactions were performed in a final volume of 25 μL containing 1 U of HotStarTaq DNA Polymerase (Qiagen, Hilden, Germany), 0.2 μM dNTP mix (Qiagen, Hilden, Germany), 250 nM of each primer, 1× buffer and 10 μL of bisulfite modified DNA. All reactions were performed under the following thermal cycling conditions: 95 °C for 15 min, followed by 35 cycles of 94 °C for 1 min, 50 °C for 30 sec, 72 °C for 1 min, and a final extension step at 72 °C for 10 min. After amplification, PCR products were visualized by gel electrophoresis.

Afterwards, PCR products were analyzed directly by the Transgenomic WAVE DHPLC instrument (San Jose, CA, USA) and a DNASep ™ 50 × 4.6 mm i.d. cartridge was used as a column (Transgenomic, Omaha, NE, USA). Separation of products was conducted at 54.3 °C and 57.6 °C for CDH1 and CDH13, respectively, by continuously mixing buffer B (0.1 M triethylammonium acetate, 25% acetonitrile) with buffer A (0.1 M triethylammonium acetate), either over 5 min: 57–66% (CDH1) or 50–75% (CDH13) at 0.9 mL per minute flow rate. The partially denaturing temperature at which samples were run was determined for each assay using Wavemaker software version 4.1.40 (Transgenomic, Omaha, NE, USA). The average partially denaturing temperature of the methylated amplicon was used to separate the target fragments with various degrees of methylation.

Associations between categorical variables were tested with Pearson’s Chi square test. The Kaplan-Meier method was used for univariate survival analysis, and the log rank test was used to assess the difference between groups. A Cox proportional hazards model applying a backward variable selection procedure based on the likelihood-ratio test was used for multivariate survival analysis. The correlation between the MethyLight PCR data and the DHPLC-PCR data was analyzed using the Pearson’s correlation coefficient. A p value of <0.05 was considered statistically significant. These statistical calculations were performed using SPSS, version 13.0.