Hypermethylation of Multiple Genes in Tumor Tissues and Voided Urine in Urinary Bladder Cancer Patients

Silencing of genes by promoter hypermethylation is common in human cancers. The changes involving methylation of cytosine in CpG dinucleotide have been recognized as an alternative mechanism in Knudson’s two-hits hypothesis where tumor suppressor genes are inactivated. (1). Although the mechanisms of these epigenetic changes are not clearly understood, the list of aberrant methylation genes in cancer is rapidly growing. Nevertheless, a certain kind of tissue-specific methylated gene has been demonstrated. For example, a high frequency of methylation of GSTP1 has been found in prostate cancer (2) but is rare in other types of cancer. Other examples include frequent methylation of DAPK in lung cancer (3, 4) and retinoic acid receptor β2 (RARβ2) in both breast (5) and lung cancer (6). These findings may have potential diagnostic and therapeutic implications. Recently, these epigenetic changes have also been detected in DNA from plasma/serum (3, 7), urine (8, 9), and sputum (10), indicating that a noninvasive and early cancer detection method can be developed.

We are interested in bladder cancer, which is the sixth most common cancer in the world. The majority of bladder cancer is TCC.² One of the distinctive features of TCC is that multiple metachronous or synchronous cancers frequently develop. These arise from either polyclonal origin or metastasis from a single clone. Bladder cancer patients would then need to have a long-term follow-up with repeated urine cytology and cystoscopy for monitoring. Conventional urine cytology has been the standard noninvasive method for cancer detection and disease monitoring. However, the sensitivity of this method is known to be low, especially for low-grade TCC. Therefore, a more sensitive, noninvasive method for cancer detection is required. Methylation detection with appropriate markers may provide a more sensitive method for cancer detection.

We have analyzed the methylation patterns of 7 cancer-related genes including DAPK, RARβ, E-cadherin, p15, p16, MGMT, and GSTP1 in 98 bladder TCC samples. Frequent methylation was detected in RARβ, DAPK, E-cadherin, and p16, and they were chosen as markers to detect DNA methylation in 22 corresponding voided urine samples. Our results show that detection of DNA methylation in voided urine is feasible and appears to be more sensitive than conventional urine cytology.

Bladder tumor samples from transurethral resection specimens were obtained from 98 patients at the Prince of Wales Hospital. The samples consisted of 73 primary tumors and 25 recurrent tumors. In this study, four samples of carcinoma in situ and seven samples of normal urothelium from individuals without bladder cancer were also included. The clinical pathological data for all of the tissue samples are summarized in Table 1.

Paired voided urine samples were collected from 22 patients (Table 1). The urine samples were spun down, and the urine sediments were subjected to subsequent analysis. The corresponding urine samples were also subjected to conventional urine cytology examination by an experienced pathologist without knowledge of the methylation results. In addition, 17 normal voided urine sediments from age- and sex-matched controls were included.

DNA was extracted from formalin-fixed, paraffin-embedded sections or voided urine sediments using a high pure PCR template preparation kit (Boehringer Mannheim, Indianapolis, IN). H&E-stained sections from each tumor sample were examined by an experienced pathologist to confirm the histological diagnosis and assess the tumor content. If tumor content was <80%, tumor content was enriched by microdissection using a fine needle under a dissection microscope as described previously (11). Microdissection was performed for all of the carcinoma in situ and normal urothelium cases, and 20 five-μm-thick sections were used for DNA extraction.

Extracted DNA was bisulfite-modified by using the CpGenome DNA Modification kit (Intergen, Purchase, NY). The modified DNA was subject to MSP using specific primers. Primer sequences, annealing temperatures, and the expected product size are listed in Table 2. Two μl of bisulfate-modified DNA were amplified in a total volume of 25 μl containing 1× PCR buffer II (Perkin-Elmer Corp.), 2 mm MgCl₂, 0.25 mm deoxynucleotide triphosphate, 1 μm of each primer, and 1 unit of AmpliTaq Gold polymerase (Perkin-Elmer Corp.) at 95°C for 10 min; 40 cycles of 95°C for 30 s, the specific annealing temperature for 45 s, and 72°C for 45 s; followed by a final extension at 72°C for 10 min. IVD (Intergen) was used a positive control for methylation, and water was used as negative control. Ten μl of PCR products were loaded onto nondenaturing 10% polyacrylamide gels. The gels were then stained with ethidium bromide and visualized under UV illumination.

All of the statistical analysis was performed with the statistical package SPSS version 10.0. Association between parameters was assessed by χ² or Fisher’s exact test. The Mann-Whitney U test was used to compare parameters of different groups.

We have analyzed the frequency of methylation of RARβ, DAPK, E-cadherin, p16, p15, MGMT, and GSTP1 in 98 cases of bladder TCC, 4 cases of carcinoma in situ, and 7 samples of normal bladder epithelium by MSP (Figs. 1 and 2; Table 3). In the tumor samples, frequent methylation was detected in RARβ (87.8%), DAPK (58.2%), E-cadherin (63.3%), and p16 (26.5%). Methylation of p15 was detected in 13.3% of cases. However, methylation of GSTP1 (5.1%) and MGMT (5.1%) was rare. Methylation of one of four genes (DAPK, RARβ, E-cadherin, and p16) was found in 98 of 98 (100%) of TCC cases. We have compared the pattern of methylation of theses genes between primary and recurrent cases, and no significant difference was identified (Table 3). No statistically significant association was found between the methylation status of the genes and the clinical/pathological parameters. Methylation of E-cadherin has been reported to be associated with aging (12); however, no statistically significant difference in age was detected between unmethylated and methylated cases in our study (Table 4). We have also analyzed the number of genes that are methylated concurrently in the tumor (Table 5). Notably, more than three genes that are methylated concurrently in the tumor accounted for 20% of our cases. The frequency of concurrent gene methylation did not correlate with the grading of TCC or the presence or absence of muscle invasion.

Methylation of DAPK (75%), E-cadherin (50%), and RARβ (50%) was also detected in carcinoma in situ (Table 3). Normal urothelium controls did not show any aberrant hypermethylation except for RARβ, for which three of seven samples (42.9%) showed methylation.

To assess the feasibility of detecting methylated genes in urine, we investigated the methylation frequency of DAPK, RARβ, E-cadherin, and p16 in urine sediment from 22 patients (Fig. 1). Because these four genes had a higher frequency of methylation in TCC tumor samples, they were selected for urine analysis. Voided urine samples from 17 normal healthy individuals were included as control. MSP results are summarized in Tables 6 and 7. The results showed that gene promoter methylation could be detected in urine samples from the patients. The frequency of methylation was 45.5% for DAPK, 68.2% for RARβ, 59.1% for E-cadherin, and 13.6% for p16. Methylation of one of these four genes was found in 20 of 22 (90.9%) cases. Methylation could only be detected in those patients whose tumor tissue also showed gene methylation; in other words, no false positive was found. Besides, with the exception of RARβ [4 of 17 (23.5%) cases showed methylation], only unmethylated copies were detected in normal urine control. For comparison, urine cytology data were analyzed (Table 6). Only 10 cases (45.5%) were diagnosed as cancer or suspicious. The sensitivity was even lower in low-grade cases, in which only one of nine (11.1%) cases was positive (Table 7). Meanwhile, if we take any one of the four genes that showed methylation in the urine as a positive marker, the sensitivity of using a methylation marker to detect bladder TCC in urine was 90.9%, which was far greater than the sensitivity of urine cytology (Table 8). This difference was more striking when comparing low-grade cases (100% versus 11.1%). Methylation marker, on the other hand, has a lower specificity because of the presence of methylation copies of RARβ in normal urine. The sensitivity and specificity of individual gene methylation with respect to the grading of TCC is also tabulated in Table 8. For E-cadherin, DAPK, and p16, no methylated copies were detected in normal urine. Moreover, the methylation status of E-cadherin had a higher sensitivity in detection of bladder cancer, especially for grade 1 TCC, as compared with urine cytology. Similarly, methylation status of DAPK also demonstrated a higher sensitivity in detection of low-grade bladder TCC.

We have assessed the sensitivity of MSP for detection of methylated alleles of DAPK and RARβ. Different amounts of IVD were mixed with 1 μg of DNA from normal bladder urothelium with unmethylated alleles of all genes before bisulfite modification. The lower limit of MSP detection was 10 ng for these two genes (Fig. 3).

We have analyzed the methylation pattern of RARβ, DAPK, E-cadherin, p15, p16, MGMT, and GSTP1 in bladder TCC of different stages and grades. In our study, all of our samples have at least one gene methylated, and more than three genes that were methylated accounted for 20% of our cases. Thus, the epigenetic event of gene methylation was frequent in bladder cancer. However, this phenomenon did not appear to be correlated with disease grade or stage.

Reports on the methylation of various genes have been described in primary bladder cancer (12, 13, 14, 15, 16, 17, 18, 19). Among these reports, methylation of p16 was most commonly investigated. The frequency of p16 methylation in bladder TCC ranged from 9–67% (13, 14, 15, 16, 17, 19). Our study represented the largest series and demonstrated p16 methylation of 26.5%. Tumor suppressor gene p16 specifically inactivates cyclin-dependent kinase 4 and cyclin-dependent kinase 6, which interact with cyclin D1 and stimulate the progression of the cell cycle from G₁ to S phase. Thus, inactivation of the p16 gene was important for tumorigenesis in bladder cancer and other cancers (20). Another gene that is located on the same loci is p15^INK4b. Methylation of p15 has also been detected in several tumors (20, 21) but has not been reported in bladder cancer. In our study, we find that the frequency of methylation of the p15 gene is 13.3%, suggesting that alteration of the p15 gene occurs in at least a subset of bladder cancers. Other mechanisms of p15 inactivation such as deletion (22, 23) may also be involved.

In our results, we also found high methylation frequency for DAPK, RARβ, and E-cadherin in tumor tissue and carcinoma in situ. Methylation was demonstrated to be the mechanism of loss of expression of DAPK in bladder cancer cells and other cancer cells (24, 25). Recently, it has also been demonstrated that methylation of DAPK is associated with stage and poor prognosis in non-small cell lung cancer (26, 27). However, in our results, we do not find such a correlation. On the other hand, Esteller et al. (19) have demonstrated a low frequency of methylation of DAPK in bladder cancer using the same method. Apart from ethic or geographical factors, the differences in sample size may also account for the differences in frequency of DAPK methylation. Methylation of RARβ was first reported in breast cancer (5) and related to the development of retinoic acid resistance in cancer cells. A recent study has shown that methylation of RARβ is found in lung tumor tissue as well as in adjacent nonmalignant lung tissues (4). The presence of the methylated allele in adjacent nonmalignant tissue may represent premalignant changes. In our study, methylation of RARβ can also be detected in three of seven normal control urothelium samples. However, the significance of this finding needs to be further investigated. Methylation of the E-cadherin gene has been widely reported in different tumors as well as in bladder cancer (12, 28, 29). Recently, Bornman et al. (12) demonstrated a high frequency of methylation of the E-cadherin gene in bladder cancer tissues and in normal urothelial epithelium from elderly individuals. They suggested that methylation of E-cadherin in bladder epithelium was age-related. In our study, there is no difference in the median age of bladder TCC patients with methylated E-cadherin and those with unmethylated E-cadherin (Table 4). However, we cannot detect any methylation of E-cadherin in our normal urothelium controls. This is probably because the samples were taken from younger individuals (median age, 55 years). This result is consistent with that of Bornman et al. (12) because they only found methylation of E-cadherin in normal samples from individuals >70 years old. Methylation of RARβ, DAPK, and E-cadherin can also be detected in carcinoma in situ. The results suggest that inactivation of theses gene may be involved in both bladder TCC and carcinoma in situ.

We have also found a low frequency of gene methylation for MGMT and GSTP1. This finding is in keeping with those of previous studies (2, 19, 30).

Another aim of this study was to investigate whether cancer cells can be detected using methylation markers in corresponding voided urine samples of the patients. We choose DAPK, RARβ, E-cadherin, and p16 as methylation markers because these markers have a high frequency of methylation in our tumor tissue samples. Our results showed that unmethylated allele of these four genes could be detected in all urine samples. Recent findings in prostate cancer demonstrated that methylation of the GSTP1 gene can be detected in the urine of prostate cancer patients (8, 9). Our results confirmed that detection of gene methylation in urine was feasible. With regard to the sensitivity of the assay, methylation of any one of the four genes could be detected in 90.9% of the urine samples, whereas urine cytology could only detect cancer cells in 45.5% of the samples. This difference is more striking in low-grade cases, where conventional urine cytology was known to have a low sensitivity. The results suggest that methylation detection has a higher sensitivity than conventional urine cytology in cancer detection in urine. Combination of methylation makers, however, had a lower specificity, which was related to the detection of RARβ in normal urine control. However, using a specific marker, such as E-cadherin or DAPK, could result in a higher specificity and sensitivity as compared with urine cytology, especially for low-grade cases. Thus, the diagnostic assessment could be improved by using selected methylation markers. Furthermore, a combination of conventional urine cytology and selected methylation markers may improve diagnostic accuracy, especially with regard to low-grade cases.

In conclusion, we have demonstrated a distinct methylation pattern of multiple genes in urinary bladder cancer patients. Frequent methylation of RARβ, DAPK, E-cadherin, and p16 was detected. This was also the first time that methylation of RARβ and p15 was reported in bladder cancer. We have also demonstrated that detection of bladder cancer in urine using methylation markers appeared to be more sensitive then conventional urine cytology. Detection of methylated genes in routinely voided urine, as a potential noninvasive diagnostic and monitoring tool, deserves further investigation.