Quantitative Methylation-Specific Polymerase Chain Reaction Gene Patterns in Urine Sediment Distinguish Prostate Cancer Patients From Contro
http://www.100md.com
《临床肿瘤学》
the Department of Otolaryngology-Head and Neck Surgery, The Johns Hopkins School of Medicine
Department of Pathology, Johns Hopkins Medical Institutions, Baltimore, MD
Bioinformatics and Computational Genomics, Department of Molecular Biotechnology, Faculty of Agricultural and Applied Biological Sciences, Gent University, Gent, Belgium
ABSTRACT
PURPOSE: Aberrant promoter hypermethylation of several known or putative tumor suppressor genes occurs frequently during the pathogenesis of prostate cancers and is a promising marker for cancer detection. We sought to develop a test for prostate cancer based on a quantitative methylation-specific polymerase chain reaction (QMSP) of multiple genes in urine sediment DNA.
PATIENTS AND METHODS: We tested urine sediment DNA for aberrant methylation of nine gene promoters (p16INK4a, p14ARF, MGMT, GSTP1, RAR?2, CDH1 [E-cadherin], TIMP3, Rassf1A, and APC) from 52 patients with prostate cancer and 21 matched primary tumors by quantitative fluorogenic real-time polymerase chain reaction. We also analyzed urine sediments from 91 age-matched individuals without any history of genitourinary malignancy as controls.
RESULTS: Promoter hypermethylation of at least one of the genes studied was detected in urine samples from all 52 prostate cancer patients. Urine samples from the 91 controls without evidence of genitourinary cancer revealed no methylation of the p16, ARF, MGMT, and GSTP1 gene promoters, whereas methylation of RAR?2, TIMP3, CDH1, Rassf1A, and APC was detected at low levels.
CONCLUSION: Overall, methylation found in urine samples matched the methylation status in the primary tumor. A combination of only four genes (p16, ARF, MGMT, and GSTP1) would theoretically allow us to detect 87% of prostate cancers with 100% specificity. Our data support further development of the noninvasive QMSP assay in urine DNA for early detection and surveillance of prostate cancer.
INTRODUCTION
In recent years, prostate cancer has shown approximately a 3% annual increase worldwide. In American men, this cancer is clinically diagnosed in one of every 11 men; one third of men diagnosed will develop significant life-threatening disease, making prostate cancer the second most lethal neoplasia.1 Diagnosis and management are confounded by the lack of symptoms and the lack of cancer-specific diagnostic techniques to be used during early stages of disease. Prostate cancer is curable if detected early, while still localized within the capsule.2 To date, curative therapeutic options for the majority of patients depend on early detection, including digital rectal examination and measurement of prostate-specific antigen (PSA) in the serum. PSA is regarded as one of the best conventional serum tumor markers; however, determination of PSA levels alone is neither sensitive nor specific enough for a definite diagnosis of prostate cancer.3 Most men with either an abnormal finding on digital rectal examination or elevated PSA require transrectal biopsies with ultrasound guidance, and up to one third of these men are found to be free of disease. Moreover, by this approach, false-negative biopsies still occur in approximately 25% of patients. Novel approaches for the definitive detection and control of this cancer are urgently needed.
DNA hypermethylation in cytosine phospho guanine island within gene promoter regions is associated with binding of methylcytosine-binding proteins and the recruitment of histone deacetylases and corepressors.4 The methylated DNA–protein complex formation is critical for constructing transcriptionally repressive chromatin structure. Among some human cancers, some genes are epigenetically altered as a group in a tumor type–specific manner.5 In other cases, specific methylation patterns are shared among different tumor types.6 Some cytosine phospho guanine islands may be differentially susceptible to hypermethylation under certain unknown growth selection pressures, which may drive characteristic pathways leading to the development of certain tumor types.
Bodily fluids from several types of cancer have been successfully used for the molecular detection of neoplasia, including stool in colon and pancreatic cancer, urine in bladder cancer, and sputum and bronchoalveolar lavage in lung cancer.7,8 Recently, promoter hypermethylation has been used successfully to detect neoplastic DNA in sputum9; bronchial lavage fluid10 and serum11 from lung cancer patients; and serum from liver,12 head and neck,13 and breast cancer patients.14 Because prostate tumors occur in the peripheral zone that contains three quarters of the gland and secretary ducts empty their contents into the urethra, urine detection is conceivable. In a previous study testing conventional methylation-specific polymerase chain reaction (MSP) in urine sediment DNA, we identified GSTP1 methylation in 27% of urine samples and no false-positive results when compared with methylation in the corresponding tumor.15 To further develop this approach, we examined nine genes, the expression of which is frequently silenced by aberrant methylation, in urine from patients and controls using high-throughput quantitative MSP (QMSP).10 We then correlated our findings with disease states and clinicopathologic features in prostate cancer patients.
PATIENTS AND METHODS
Sample Collection and DNA Preparation
We evaluated urine samples of 52 patients with prostate cancer who underwent curative surgery at the Johns Hopkins University School of Medicine. Detailed data on these patients are listed in Table 1. Urine samples from 91 age-matched individuals (median age, 56.5 years; range, 28 to 84 years) without a history of genitourinary malignancy were used as controls. Of these 91 individuals, nine were diagnosed with benign prostate hyperplasia, 10 harbored atypical cells by urine cytology examination, five had primary cancers in other sites (non–small-cell carcinoma of lung, n = 1; basal cell carcinoma of skin, n = 1; malignant melanoma of leg, n = 1; Kaposi's sarcoma of the leg, n = 1; and infiltrating ductal carcinoma of the breast, n = 1), one had fibroepithelial polyp of the bladder, three had tubular adenomas of the colon, one had organizing thrombus in the vagina, 25 visited the hospital for routine physical examination, 20 had either macroscopic or microscopic hematuria, and 17 were seen for vague urologic symptoms without malignancy. Among the 91 controls, 66 were male, and 25 were female. Fifty milliliters of voided urine were collected from all controls and patients before definite surgery. Urine samples were spun at 3,000 x g for 10 minutes and washed twice with phosphate-buffered saline. All samples were stored at –80°C. Frozen urine cell pellets were digested with 1% sodium dodecyl sulfate and 50 μg/mL of proteinase K (Boehringer, Mannheim, Germany) at 48°C overnight, followed by phenol/chloroform extraction and ethanol precipitation of DNA, as previously described.16
Bisulfite Treatment
DNA from urine sediment or from primary tumors was subjected to bisulfite treatment, as described previously.17 Briefly, 2 μg of genomic DNA was denatured in 0.2 M of NaOH for 20 minutes at 50°C. The denatured DNA was diluted in 500 μL of freshly prepared solution of 10 mmol/L hydroquinone and 3 M of sodium bisulfite and incubated for 3 hours at 70°C. After incubation, the DNA sample was desalted through a column (Wizard DNA Clean-Up System; Promega, Madison, WI), treated with 0.3 M of NaOH for 10 minutes at room temperature, and precipitated with ethanol. The bisulfite-modified genomic DNA was resuspended in 120 μL of LoTE (EDTA 2.5 mmol/L and Tris-HCl 10 mmol/L) and stored at –80°C.
Methylation Analysis
The bisulfite-modified DNA was used as a template for fluorescence-based real-time polymerase chain reaction (PCR), as previously described.18 In brief, primers and probes were designed to specifically amplify the bisulfite-converted promoter of the gene of interest. The ratios between the values of the gene of interest and the internal reference gene, ?-actin, which was obtained by Taqman analysis, were used as a measure for representing the relative level of methylation in the particular sample (gene of interest/reference gene x 1,000). Fluorogenic PCRs were carried out in a reaction volume of 20 μL consisting of 600 nmol/L of each primer; 200 nM of probe; 0.75 U of platinum Taq polymerase (Invitrogen, Carlsbad, CA); 200 μmol/L each of 2’-Deoxyadenosine 5’-triphosphate, 2’-Deoxycytidine 5’-triphosphate, 2’-Deoxyguanosine 5’-triphosphate, and 2’-Deoxythymidine 5’-triphosphate; 16.6 mmol/L of ammonium sulfate; 67 mmol/L of Trizma (Sigma, St Louis, MO); 6.7 mmol/L of MgCl2 (2.5 mmol/L for p16); 10 mmol/L of mercaptoethanol; and 0.1% dimethylsulfoxide. Three microliters of treated DNA solution were used in each real-time MSP reaction. Amplications were carried out in 384-well plates in a 7900 Sequence Detector System (Perkin-Elmer Applied Biosystems, Norwalk, CT). Each plate consisted of patient samples and multiple water blanks, as well as positive and negative controls. Leukocytes from a healthy individual were methylated in vitro with excess SssI methyltransferase (New England Biolabs Inc, Beverly, MA) to generate completely methylated DNA, and serial dilutions of this DNA were used for constructing the calibration curves on each plate. Identical lab procedures and intermixing were performed in the same laboratory for each batch tested.
Statistical Analysis
First, for all the markers, individual receiver operating characteristic curves were generated. This was performed by sorting the different percent methylation scores and checking for sensitivity and specificity in each unique score of the end point to be tested (cancer v normal samples). The positive likelihood ratio was calculated at each cut point. Then maximal likelihood ratio–positive values for all the different markers were combined, and learning sets were created. In this way, the original continuous and rather complex information in the QMSP data was transformed to a discrete binary read-out. Then, all the learning sets were tested for all possible combinations of markers.
In the cross-validation procedure, the samples were randomly divided into nine tenths of the original data as a training set, the remaining one tenth was used as a test set to calculate performance. The sampling procedure ensured equal class representation in the training set (stratification constraints). This procedure was repeated 10 times by maximizing the chance that each instance was used in the test set. Over the 10 experiments, a general sensitivity and specificity score was computed. Because the procedure was a stochastic, we repeated the procedure multiple times, and as can be expected from a 10-fold cross validation, the computed results were robust.
In the final step, machine learning was applied. This was done by applying orthologous data analysis techniques using the WEKA System's Bayes Network approach.19 Orthologous data analysis looks at data from different perspectives and is capable of detecting completely different patterns in datasets (technically independent). It also adds to the interpretability of the results and gives additional information to the learning set and its saturation.
All P values were calculated using the 2 test. When observed frequencies were below 5, a Fisher correlation test was performed. All statistical tests were two sided. All differences were considered statistically significant if P .05. The associations between methylation of an individual gene and clinical and pathologic variables were assessed using logistic regression.
RESULTS AND DISCUSSION
We tested the methylation status of nine gene promoters in the urine sediment DNA of 52 cancer patients and 91 controls without genitourinary malignancy. The demographic and clinical characteristics of cancer patients included in this study are listed in Table 1. Methylation levels of selected genes in urine sediment of prostate cancer patients and control urine sediments are shown in Figure 1. Aberrant promoter hypermethylation of at least one of the genes investigated was detected in the urine sediment of all the 52 prostate cancer patients (100%), and 42 of these urine DNA samples (80%) were positive for at least three genes simultaneously. Moreover, 87% of the samples from patients with prostate cancer demonstrated methylation in at least one of the four genes (p16, ARF, MGMT, and GSTP1) with 100% specificity (ie, all of the 91 control samples were negative for methylation in these four genes). Based on a Bayes Network statistical approach, additional tests are available in Supplementary Table 1 (available online only). The frequency and median methylation values (gene/?-actin x 1,000) for each gene in urine DNAs are listed in Table 2. Methylation-positive urine samples from prostate cancer patients ranged from 19% in MGMT to 77% in CDH1 (Table 2). On the basis of 66 male controls, sensitivity and specificity were calculated and are detailed in Supplementary Table 1. Interestingly, most of the methylation-positive controls came from patients with benign prostate hyperplasia (Supplementary Table 2, available online only).
To confirm whether the epigenetic alterations in urine sediments were identical to the matched tumors, we analyzed five genes (p16, ARF, MGMT, GSTP1, and RAR?2) in 21 paired primary tumor samples. The methylation patterns of these five genes in primary tumor and matched urine DNA are shown in Figure 2. The five genes were selected because of absence or near absence of methylation in normal prostate tissue.20 We did 21 matched available primary prostate cancer samples. We extracted the DNA from a high Gleason score area. Overall, identical methylation patterns were found in the urine and corresponding tumor DNA. Aberrant methylation was detected in only one urine DNA sample of a prostate cancer patient without methylation in the corresponding tumor (sample No. 14, Fig 2). In this patient, we found methylation of ARF and RAR?2 only in the urine sample. This urine sample may contain tumor cells from an area separate from where the tissue DNA was extracted. The analytic sensitivity of these five genes is shown in Figure 2.
The development of real-time PCR has simplified the study of genes inactivated by promoter hypermethylation in human cancer. It is a highly sensitive assay that is capable of detecting methylated alleles in the presence of a more than 1,000-fold excess of unmethylated alleles. Yet, it is more stringent and more specific because, in addition to the two PCR primers, the fluorescent-labeled hybridization probe has to anneal correctly between the two primers. QMSP is often more sensitive than conventional MSP, but this varies based on the promoter, primers, and condition. Others have found a higher frequency of APC methylation by QMSP compared with conventional MSP in cell lines.21 In general, the methylation frequency in primary tumors of each tested gene was higher than previous reports because of the use of QMSP or our selective dissection of a higher Gleason score area for DNA extraction.
Aberrant methylation in the urine sediment of primary prostate carcinoma had no significant level of correlation with patient demographic data, including age, histologic subtype, and staging of the tumor (data not shown). However, others have found a significant correlation between methylation and Gleason score, preoperative serum PSA, and tumor stage.22 The reason behind these discrepancies may be the indirect measurement of methylation in urine instead of primary tumor DNA and the different clinical subgroups represented in various studies (Table 3).
We investigated two reported methylated DNA repair genes (GSTP1 and MGMT) commonly found in various tumor types including prostate cancer. Using conventional MSP, we detected methylated GSTP1 alleles in the urine sediment from 27% of the patients with a methylated primary tumor.15 In the present study, GSTP1 was methylated in 48% of urine sediment samples. The reason for this discrepancy may be the primer design, but it should be noted that the sample size was different in both studies and that tumor stages and grade also differed. We and others also demonstrated that, for prostate cancer, there was no case in which a urine sediment DNA sample gave a positive GSTP1 methylation result in the absence of methylation in the corresponding tumor.15,23 We found MGMT methylation in 19% of urine sediment samples compared with less than 25% of primary prostate tumors by conventional MSP.22-24
Three cell cycle regulators (p16, ARF, and possibly Rassf1A) were included in our study. Previous reports of methylation in primary tumor tissues and prostate cancer cell lines ranged from 3% to 69% for p16 methylation,22,24-28 6% for ARF methylation,24 and 53% to 100% for Rassf1A methylation.22,29,30 These discrepancies may be a result of differences in the methylation assays used and the inclusion of tumors with different stages and grades.
Two metastatic suppressor genes, CDH1 and TIMP3, were frequently methylated in the urine sediment of prostate cancer patients (77% and 37%, respectively). Our finding of CDH1 methylation is similar with other studies based on conventional MSP in primary prostate tumors.31,32 Li et al31 reported that the severity of CDH1 methylation correlated with tumor progression. However, we found no correlation between CDH1 methylation and tumor grade and stage. Despite establishing a cutoff value (Table 2) in our controls, we found low levels of CDH1 methylation in five (6%) of 91 samples from individuals without any known genitourinary malignancy. TIMP3 is the third member of the TIMP family of proteins and is believed to play a significant role in controlling extracellular matrix remodeling. TIMP3 was found to be methylated in 24% to 28% of various human cancers.33-35 We found TIMP3 methylation in 37% of urine sediments from prostate cancer patients. As a diagnostic marker in urine DNA, TIMP3 may be limited by a persistent low frequency of methylation in normal controls. The use of retinoids to suppress tumor development has been evaluated in several animal models of carcinogenesis, including models of skin, breast, oral cavity, lung, hepatic, GI, prostatic, and bladder cancer.36 Retinoids act primarily via nuclear receptors encoded by the RAR? gene. Because the isoforms RAR?2 and RAR?4 are frequently methylated in other cancers,37-39 we investigated methylation of the RAR?2 promoter in urine sediment DNA. We and others have also reported methylation of RAR?2 in 53% to 95% of primary prostate tumor tissues.22,40
The APC protein is an integral part of the wnt-signaling mechanism, but it also plays a role in cell-cell adhesion, stability of the microtubular cytoskeleton, cell cycle regulation, and possibly apoptosis. We and others have demonstrated that the promoter regions of APC gene are aberrantly methylated in many types of cancer.18,20,41-44 In other studies, APC was found to be hypermethylated in 27% to 95% of primary prostate tumors20,22 compared with 54% methylation in urine sediment DNA reported in the present study.
To our knowledge, there have been few studies15,23,45-50 using an extended panel of methylation markers for the detection of prostate cancer in urine sediment. Thus, our methylation assay using nine different genes in the urine DNA confirms and extends previous observations. The high sensitivity (87%) using just four genes (p16, ARF, MGMT, and GSTP1) with undetectable methylation levels in all control samples (Table 2) is promising. The detection of tumor molecular signatures in body fluids has implications for the identification of high-risk patients and patients with preinvasive or early-stage lesions and for monitoring residual disease. Molecular approaches characterized by high specificity have variable sensitivity, perhaps because of the presence of low tumor DNA quantities in urine or the high level of contamination with normal DNA. Several approaches to improve assay sensitivity have been applied to clinical samples. Sensitivity has been improved over conventional MSP by performing a semi-nested MSP after a DNA preamplification step50 or a nested two-stage PCR,51 usually with decreased specificity for clinically definable disease. The sensitivity of QMSP in urine sediment could be further increased by isolating the aberrant cells from urine before DNA extraction or increasing the number of prostate cancer–specific markers. However, more sensitive assays may result in imperfect specificity and lack of quantitation, and tumors must be further validated in clinical samples.
Exfoliative material (present in urine, stool, sputum, bronchoalveolar lavage, bronchial brushings, and so on) offers diagnostic possibilities, but the sensitivity of current cytologic tests is low and virtually not used for prostate cancer detection. Diagnostic tools based on DNA alterations able to provide high specificity and sensitivity would clearly be of enormous benefit to patients, particularly if the specimens could be obtained by noninvasive means. To this end, the detection of aberrant methylation in urine DNA may offer a promising approach for the noninvasive diagnosis of prostate cancer. Apart from prostate cancer detection, it would be interesting to see whether the detection of aberrant methylation in the urine can be used in disease monitoring after curative surgery. If methylated DNA disappears shortly in urine after curative surgery, the reappearance of these markers may suggest recurrence of disease that may require more intensive screening and aggressive treatment. Thus, this simple and noninvasive method for detecting prostate cancer is readily automated and has many potential clinical applications, including primary diagnosis, monitoring for relapse, and measurement of therapeutic response. This study was performed on patients referred after PSA screening or other clinical suspicion. Thus, additional studies are necessary to elucidate the role of detecting aberrant methylation in urine as a tool for early detection and surveillance of prostate cancer either alone or in combination with serum PSA or digital rectal examinations. Moreover, other cancers, including bladder and kidney cancer, contribute cellular DNA to urine sediment. Thus, a panel of carefully selected methylation markers in urine sediment could be envisioned that both detects and then discriminates among a variety of urologic tumors.
Supplemental Tables
The supplemental tables are included in the full-text version of this article, available on-line at www.jco.org. They are not included in the PDF (via Adober Acrobat Readerr) version.
Authors' Disclosures of Potential Conflicts of Interest
Although all authors completed the disclosure declaration, the following author or immediate family members indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.
NOTES
Supported by National Cancer Institute grant No. U01-CA84986 and Oncomethylome Sciences, SA.
M.O.H. and O.T. contributed equally to this study.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Konishi N, Cho M, Yamamoto K, et al: Genetic changes in prostate cancer. Pathol Int 47:735-747, 1997
Alers JC, Krijtenburg PJ, Vis AN, et al: Molecular cytogenetic analysis of prostatic adenocarcinomas from screening studies: Early cancers may contain aggressive genetic features. Am J Pathol 158:399-406, 2001
Labrie F, Candas B, Dupont A, et al: Screening decreases prostate cancer death: First analysis of the 1988 Quebec Prospective Randomized Controlled trial. Prostate 38:83-91, 1999
Jones PA, Laird PW: Cancer epigenetics comes of age. Nat Genet 21:163-167, 1999
Toyota M, Ahuja N, OheToyota M, et al: CpG island methylator phenotype in colorectal cancer. Proc Natl Acad Sci U S A 96:8681-8686, 1999
Costello JF, Fruhwald MC, Smiraglia DJ, et al: Aberrant CpG-island methylation has non-random and tumour-type-specific patterns. Nat Genet 24:132-138, 2000
Sidransky D: Nucleic acid-based methods for the detection of cancer. Science 278:1054-1059, 1997
Hoque MO, Lee J, Begum S, et al: High-throughput molecular analysis of urine sediment for the detection of bladder cancer by high-density single-nucleotide polymorphism array. Cancer Res 63:5723-5726, 2003
Belinsky SA, Nikula KJ, Palmisano WA, et al: Aberrant methylation of p16(INK4a) is an early event in lung cancer and a potential biomarker for early diagnosis. Proc Natl Acad Sci U S A 95:11891-11896, 1998
Topaloglu O, Hoque MO, Tokumaru Y, et al: Detection of promoter hypermethylation of multiple genes in the tumor and bronchoalveolar lavage of patients with lung cancer. Clin Cancer Res 10:2284-2288, 2004
Esteller M, Sanchez-Cespedes M, Rosell R, et al: Detection of aberrant promoter hypermethylation of tumor suppressor genes in serum DNA from non-small cell lung cancer patients. Cancer Res 59:67-70, 1999
Wong IH, Lo YM, Zhang J, et al: Detection of aberrant p16 methylation in the plasma and serum of liver cancer patients. Cancer Res 59:71-73, 1999
Sanchez-Cespedes M, Esteller M, Wu L, et al: Gene promoter hypermethylation in tumors and serum of head and neck cancer patients. Cancer Res 60:892-895, 2000
Silva JM, Dominguez G, Garcia JM, et al: Presence of tumor DNA in plasma of breast cancer patients: Clinicopathological correlations. Cancer Res 59:3251-3256, 1999
Cairns P, Esteller M, Herman JG, et al: Molecular detection of prostate cancer in urine by GSTP1 hypermethylation. Clin Cancer Res 7:2727-2730, 2001
Sidransky D, Von Eschenbach A, Tsai YC, et al: Identification of p53 gene mutations in bladder cancers and urine samples. Science 252:706-709, 1991
Herman JG, Graff JR, Myohanen S, et al: Methylation-specific PCR: A novel PCR assay for methylation status of CpG islands. Proc Natl Acad Sci U S A 93:9821-9826, 1996
Harden SV, Tokumaru Y, Westra WH, et al: Gene promoter hypermethylation in tumors and lymph nodes of stage I lung cancer patients. Clin Cancer Res 9:1370-1375, 2003
Witten IH, Frank E, Kaufmann M: Data Mining: Practical Machine Learning Tools With Java Implementations. Morgan Kaufmann Publishers, San Francisco, CA, 2000
Jeronimo C, Henrique R, Hoque MO, et al: A quantitative promoter methylation profile of prostate cancer. Clin Cancer Res 10:8472-8478, 2004
Virmani AK, Tsou JA, Siegmund KD, et al: Hierarchical clustering of lung cancer cell lines using DNA methylation markers. Cancer Epidemiol Biomarkers Prev 11:291-297, 2002
Maruyama R, Toyooka S, Toyooka KO, et al: Aberrant promoter methylation profile of prostate cancers and its relationship to clinicopathological features. Clin Cancer Res 8:514-519, 2002
Goessl C, Krause H, Muller M, et al: Fluorescent methylation-specific polymerase chain reaction for DNA-based detection of prostate cancer in bodily fluids. Cancer Res 60:5941-5945, 2000
Konishi N, Nakamura M, Kishi M, et al: DNA hypermethylation status of multiple genes in prostate adenocarcinomas. Jpn J Cancer Res 93:767-773, 2002
Yamanaka M, Watanabe M, Yamada Y, et al: Altered methylation of multiple genes in carcinogenesis of the prostate. Int J Cancer 106:382-387, 2003
Konishi N, Nakamura M, Kishi M, et al: Heterogeneous methylation and deletion patterns of the INK4a/ARF locus within prostate carcinomas. Am J Pathol 160:1207-1214, 2002
Herman JG, Merlo A, Mao L, et al: Inactivation of the CDKN2/p16/MTS1 gene is frequently associated with aberrant DNA methylation in all common human cancers. Cancer Res 55:4525-4530, 1995
Jarrard DF, Bova GS, Ewing CM, et al: Deletional, mutational, and methylation analyses of CDKN2 (p16/MTS1) in primary and metastatic prostate cancer. Genes Chromosomes Cancer 19:90-96, 1997
Liu L, Yoon JH, Dammann R, et al: Frequent hypermethylation of the RASSF1A gene in prostate cancer. Oncogene 21:6835-6840, 2002
Kuzmin I, Gillespie JW, Protopopov A, et al: The RASSF1A tumor suppressor gene is inactivated in prostate tumors and suppresses growth of prostate carcinoma cells. Cancer Res 62:3498-3502, 2002
Li LC, Zhao H, Nakajima K, et al: Methylation of the E-cadherin gene promoter correlates with progression of prostate cancer. J Urol 166:705-709, 2001
Kallakury BV, Sheehan CE, Winn-Deen E, et al: Decreased expression of catenins (alpha and beta), p120 CTN, and E-cadherin cell adhesion proteins and E-cadherin gene promoter methylation in prostatic adenocarcinomas. Cancer 92:2786-2795, 2001
Kang GH, Lee S, Kim JS, et al: Profile of aberrant CpG island methylation along the multistep pathway of gastric carcinogenesis. Lab Invest 83:635-641, 2003
Alonso ME, Bello MJ, Gonzalez-Gomez P, et al: Aberrant promoter methylation of multiple genes in oligodendrogliomas and ependymomas. Cancer Genet Cytogenet 144:134-142, 2003
Schagdarsurengin U, Wilkens L, Steinemann D, et al: Frequent epigenetic inactivation of the RASSF1A gene in hepatocellular carcinoma. Oncogene 22:1866-1871, 2003
Evans TR, Kaye SB: Retinoids: Present role and future potential. Br J Cancer 80:1-8, 1999
Widschwendter M, Berger J, Hermann M, et al: Methylation and silencing of the retinoic acid receptor-beta2 gene in breast cancer. J Natl Cancer Inst 92:826-832, 2000
Yang Q, Mori I, Shan L, et al: Biallelic inactivation of retinoic acid receptor beta2 gene by epigenetic change in breast cancer. Am J Pathol 158:299-303, 2001
Ivanova T, Petrenko A, Gritsko T, et al: Methylation and silencing of the retinoic acid receptor-beta 2 gene in cervical cancer. BMC Cancer 2:4, 2002
Carmen J, Henrique R, Hoque MO, et al: Quantitative RARbeta2 hypermethylation: A promising prostate cancer marker. Clin Cancer Res 10:4010-4014, 2004
Esteller M, Sparks A, Toyota M, et al: Analysis of adenomatous polyposis coli promoter hypermethylation in human cancer. Cancer Res 60:4366-4371, 2000
Kawakami K, Brabender J, Lord RV, et al: Hypermethylated APC DNA in plasma and prognosis of patients with esophageal adenocarcinoma. J Natl Cancer Inst 92:1805-1811, 2000
Usadel H, Brabender J, Danenberg KD, et al: Quantitative adenomatous polyposis coli promoter methylation analysis in tumor tissue, serum, and plasma DNA of patients with lung cancer. Cancer Res 62:371-375, 2002
Dong SM, Kim HS, Rha SH, et al: Promoter hypermethylation of multiple genes in carcinoma of the uterine cervix. Clin Cancer Res 7:1982-1986, 2001
Gonzalgo ML, Pavlovich CP, Lee SM, et al: Prostate cancer detection by GSTP1 methylation analysis of postbiopsy urine specimens. Clin Cancer Res 9:2673-2677, 2003
Jeronimo C, Usadel H, Henrique R, et al: Quantitative GSTP1 hypermethylation in bodily fluids of patients with prostate cancer. Urology 60:1131-1135, 2002
Goessl C, Muller M, Straub B, et al: DNA alterations in body fluids as molecular tumor markers for urological malignancies. Eur Urol 41:668-676, 2002
Goessl C, Muller M, Heicappell R, et al: DNA-based detection of prostate cancer in blood, urine, and ejaculates. Ann N Y Acad Sci 945:51-58, 2001
Goessl C, Muller M, Heicappell R, et al: DNA-based detection of prostate cancer in urine after prostatic massage. Urology 58:335-338, 2001
Kersting M, Friedl C, Kraus A, et al: Differential frequencies of p16 (INK4a) promoter hypermethylation, p53 mutation, and K-ras mutation in exfoliative material mark the development of lung cancer in symptomatic chronic smokers. J Clin Oncol 18:3221-3229, 2000
Palmisano WA, Divine KK, Saccomanno G, et al: Predicting lung cancer by detecting aberrant promoter methylation in sputum. Cancer Res 60:5954-5958, 2000(Mohammad Obaidul Hoque, O)
Department of Pathology, Johns Hopkins Medical Institutions, Baltimore, MD
Bioinformatics and Computational Genomics, Department of Molecular Biotechnology, Faculty of Agricultural and Applied Biological Sciences, Gent University, Gent, Belgium
ABSTRACT
PURPOSE: Aberrant promoter hypermethylation of several known or putative tumor suppressor genes occurs frequently during the pathogenesis of prostate cancers and is a promising marker for cancer detection. We sought to develop a test for prostate cancer based on a quantitative methylation-specific polymerase chain reaction (QMSP) of multiple genes in urine sediment DNA.
PATIENTS AND METHODS: We tested urine sediment DNA for aberrant methylation of nine gene promoters (p16INK4a, p14ARF, MGMT, GSTP1, RAR?2, CDH1 [E-cadherin], TIMP3, Rassf1A, and APC) from 52 patients with prostate cancer and 21 matched primary tumors by quantitative fluorogenic real-time polymerase chain reaction. We also analyzed urine sediments from 91 age-matched individuals without any history of genitourinary malignancy as controls.
RESULTS: Promoter hypermethylation of at least one of the genes studied was detected in urine samples from all 52 prostate cancer patients. Urine samples from the 91 controls without evidence of genitourinary cancer revealed no methylation of the p16, ARF, MGMT, and GSTP1 gene promoters, whereas methylation of RAR?2, TIMP3, CDH1, Rassf1A, and APC was detected at low levels.
CONCLUSION: Overall, methylation found in urine samples matched the methylation status in the primary tumor. A combination of only four genes (p16, ARF, MGMT, and GSTP1) would theoretically allow us to detect 87% of prostate cancers with 100% specificity. Our data support further development of the noninvasive QMSP assay in urine DNA for early detection and surveillance of prostate cancer.
INTRODUCTION
In recent years, prostate cancer has shown approximately a 3% annual increase worldwide. In American men, this cancer is clinically diagnosed in one of every 11 men; one third of men diagnosed will develop significant life-threatening disease, making prostate cancer the second most lethal neoplasia.1 Diagnosis and management are confounded by the lack of symptoms and the lack of cancer-specific diagnostic techniques to be used during early stages of disease. Prostate cancer is curable if detected early, while still localized within the capsule.2 To date, curative therapeutic options for the majority of patients depend on early detection, including digital rectal examination and measurement of prostate-specific antigen (PSA) in the serum. PSA is regarded as one of the best conventional serum tumor markers; however, determination of PSA levels alone is neither sensitive nor specific enough for a definite diagnosis of prostate cancer.3 Most men with either an abnormal finding on digital rectal examination or elevated PSA require transrectal biopsies with ultrasound guidance, and up to one third of these men are found to be free of disease. Moreover, by this approach, false-negative biopsies still occur in approximately 25% of patients. Novel approaches for the definitive detection and control of this cancer are urgently needed.
DNA hypermethylation in cytosine phospho guanine island within gene promoter regions is associated with binding of methylcytosine-binding proteins and the recruitment of histone deacetylases and corepressors.4 The methylated DNA–protein complex formation is critical for constructing transcriptionally repressive chromatin structure. Among some human cancers, some genes are epigenetically altered as a group in a tumor type–specific manner.5 In other cases, specific methylation patterns are shared among different tumor types.6 Some cytosine phospho guanine islands may be differentially susceptible to hypermethylation under certain unknown growth selection pressures, which may drive characteristic pathways leading to the development of certain tumor types.
Bodily fluids from several types of cancer have been successfully used for the molecular detection of neoplasia, including stool in colon and pancreatic cancer, urine in bladder cancer, and sputum and bronchoalveolar lavage in lung cancer.7,8 Recently, promoter hypermethylation has been used successfully to detect neoplastic DNA in sputum9; bronchial lavage fluid10 and serum11 from lung cancer patients; and serum from liver,12 head and neck,13 and breast cancer patients.14 Because prostate tumors occur in the peripheral zone that contains three quarters of the gland and secretary ducts empty their contents into the urethra, urine detection is conceivable. In a previous study testing conventional methylation-specific polymerase chain reaction (MSP) in urine sediment DNA, we identified GSTP1 methylation in 27% of urine samples and no false-positive results when compared with methylation in the corresponding tumor.15 To further develop this approach, we examined nine genes, the expression of which is frequently silenced by aberrant methylation, in urine from patients and controls using high-throughput quantitative MSP (QMSP).10 We then correlated our findings with disease states and clinicopathologic features in prostate cancer patients.
PATIENTS AND METHODS
Sample Collection and DNA Preparation
We evaluated urine samples of 52 patients with prostate cancer who underwent curative surgery at the Johns Hopkins University School of Medicine. Detailed data on these patients are listed in Table 1. Urine samples from 91 age-matched individuals (median age, 56.5 years; range, 28 to 84 years) without a history of genitourinary malignancy were used as controls. Of these 91 individuals, nine were diagnosed with benign prostate hyperplasia, 10 harbored atypical cells by urine cytology examination, five had primary cancers in other sites (non–small-cell carcinoma of lung, n = 1; basal cell carcinoma of skin, n = 1; malignant melanoma of leg, n = 1; Kaposi's sarcoma of the leg, n = 1; and infiltrating ductal carcinoma of the breast, n = 1), one had fibroepithelial polyp of the bladder, three had tubular adenomas of the colon, one had organizing thrombus in the vagina, 25 visited the hospital for routine physical examination, 20 had either macroscopic or microscopic hematuria, and 17 were seen for vague urologic symptoms without malignancy. Among the 91 controls, 66 were male, and 25 were female. Fifty milliliters of voided urine were collected from all controls and patients before definite surgery. Urine samples were spun at 3,000 x g for 10 minutes and washed twice with phosphate-buffered saline. All samples were stored at –80°C. Frozen urine cell pellets were digested with 1% sodium dodecyl sulfate and 50 μg/mL of proteinase K (Boehringer, Mannheim, Germany) at 48°C overnight, followed by phenol/chloroform extraction and ethanol precipitation of DNA, as previously described.16
Bisulfite Treatment
DNA from urine sediment or from primary tumors was subjected to bisulfite treatment, as described previously.17 Briefly, 2 μg of genomic DNA was denatured in 0.2 M of NaOH for 20 minutes at 50°C. The denatured DNA was diluted in 500 μL of freshly prepared solution of 10 mmol/L hydroquinone and 3 M of sodium bisulfite and incubated for 3 hours at 70°C. After incubation, the DNA sample was desalted through a column (Wizard DNA Clean-Up System; Promega, Madison, WI), treated with 0.3 M of NaOH for 10 minutes at room temperature, and precipitated with ethanol. The bisulfite-modified genomic DNA was resuspended in 120 μL of LoTE (EDTA 2.5 mmol/L and Tris-HCl 10 mmol/L) and stored at –80°C.
Methylation Analysis
The bisulfite-modified DNA was used as a template for fluorescence-based real-time polymerase chain reaction (PCR), as previously described.18 In brief, primers and probes were designed to specifically amplify the bisulfite-converted promoter of the gene of interest. The ratios between the values of the gene of interest and the internal reference gene, ?-actin, which was obtained by Taqman analysis, were used as a measure for representing the relative level of methylation in the particular sample (gene of interest/reference gene x 1,000). Fluorogenic PCRs were carried out in a reaction volume of 20 μL consisting of 600 nmol/L of each primer; 200 nM of probe; 0.75 U of platinum Taq polymerase (Invitrogen, Carlsbad, CA); 200 μmol/L each of 2’-Deoxyadenosine 5’-triphosphate, 2’-Deoxycytidine 5’-triphosphate, 2’-Deoxyguanosine 5’-triphosphate, and 2’-Deoxythymidine 5’-triphosphate; 16.6 mmol/L of ammonium sulfate; 67 mmol/L of Trizma (Sigma, St Louis, MO); 6.7 mmol/L of MgCl2 (2.5 mmol/L for p16); 10 mmol/L of mercaptoethanol; and 0.1% dimethylsulfoxide. Three microliters of treated DNA solution were used in each real-time MSP reaction. Amplications were carried out in 384-well plates in a 7900 Sequence Detector System (Perkin-Elmer Applied Biosystems, Norwalk, CT). Each plate consisted of patient samples and multiple water blanks, as well as positive and negative controls. Leukocytes from a healthy individual were methylated in vitro with excess SssI methyltransferase (New England Biolabs Inc, Beverly, MA) to generate completely methylated DNA, and serial dilutions of this DNA were used for constructing the calibration curves on each plate. Identical lab procedures and intermixing were performed in the same laboratory for each batch tested.
Statistical Analysis
First, for all the markers, individual receiver operating characteristic curves were generated. This was performed by sorting the different percent methylation scores and checking for sensitivity and specificity in each unique score of the end point to be tested (cancer v normal samples). The positive likelihood ratio was calculated at each cut point. Then maximal likelihood ratio–positive values for all the different markers were combined, and learning sets were created. In this way, the original continuous and rather complex information in the QMSP data was transformed to a discrete binary read-out. Then, all the learning sets were tested for all possible combinations of markers.
In the cross-validation procedure, the samples were randomly divided into nine tenths of the original data as a training set, the remaining one tenth was used as a test set to calculate performance. The sampling procedure ensured equal class representation in the training set (stratification constraints). This procedure was repeated 10 times by maximizing the chance that each instance was used in the test set. Over the 10 experiments, a general sensitivity and specificity score was computed. Because the procedure was a stochastic, we repeated the procedure multiple times, and as can be expected from a 10-fold cross validation, the computed results were robust.
In the final step, machine learning was applied. This was done by applying orthologous data analysis techniques using the WEKA System's Bayes Network approach.19 Orthologous data analysis looks at data from different perspectives and is capable of detecting completely different patterns in datasets (technically independent). It also adds to the interpretability of the results and gives additional information to the learning set and its saturation.
All P values were calculated using the 2 test. When observed frequencies were below 5, a Fisher correlation test was performed. All statistical tests were two sided. All differences were considered statistically significant if P .05. The associations between methylation of an individual gene and clinical and pathologic variables were assessed using logistic regression.
RESULTS AND DISCUSSION
We tested the methylation status of nine gene promoters in the urine sediment DNA of 52 cancer patients and 91 controls without genitourinary malignancy. The demographic and clinical characteristics of cancer patients included in this study are listed in Table 1. Methylation levels of selected genes in urine sediment of prostate cancer patients and control urine sediments are shown in Figure 1. Aberrant promoter hypermethylation of at least one of the genes investigated was detected in the urine sediment of all the 52 prostate cancer patients (100%), and 42 of these urine DNA samples (80%) were positive for at least three genes simultaneously. Moreover, 87% of the samples from patients with prostate cancer demonstrated methylation in at least one of the four genes (p16, ARF, MGMT, and GSTP1) with 100% specificity (ie, all of the 91 control samples were negative for methylation in these four genes). Based on a Bayes Network statistical approach, additional tests are available in Supplementary Table 1 (available online only). The frequency and median methylation values (gene/?-actin x 1,000) for each gene in urine DNAs are listed in Table 2. Methylation-positive urine samples from prostate cancer patients ranged from 19% in MGMT to 77% in CDH1 (Table 2). On the basis of 66 male controls, sensitivity and specificity were calculated and are detailed in Supplementary Table 1. Interestingly, most of the methylation-positive controls came from patients with benign prostate hyperplasia (Supplementary Table 2, available online only).
To confirm whether the epigenetic alterations in urine sediments were identical to the matched tumors, we analyzed five genes (p16, ARF, MGMT, GSTP1, and RAR?2) in 21 paired primary tumor samples. The methylation patterns of these five genes in primary tumor and matched urine DNA are shown in Figure 2. The five genes were selected because of absence or near absence of methylation in normal prostate tissue.20 We did 21 matched available primary prostate cancer samples. We extracted the DNA from a high Gleason score area. Overall, identical methylation patterns were found in the urine and corresponding tumor DNA. Aberrant methylation was detected in only one urine DNA sample of a prostate cancer patient without methylation in the corresponding tumor (sample No. 14, Fig 2). In this patient, we found methylation of ARF and RAR?2 only in the urine sample. This urine sample may contain tumor cells from an area separate from where the tissue DNA was extracted. The analytic sensitivity of these five genes is shown in Figure 2.
The development of real-time PCR has simplified the study of genes inactivated by promoter hypermethylation in human cancer. It is a highly sensitive assay that is capable of detecting methylated alleles in the presence of a more than 1,000-fold excess of unmethylated alleles. Yet, it is more stringent and more specific because, in addition to the two PCR primers, the fluorescent-labeled hybridization probe has to anneal correctly between the two primers. QMSP is often more sensitive than conventional MSP, but this varies based on the promoter, primers, and condition. Others have found a higher frequency of APC methylation by QMSP compared with conventional MSP in cell lines.21 In general, the methylation frequency in primary tumors of each tested gene was higher than previous reports because of the use of QMSP or our selective dissection of a higher Gleason score area for DNA extraction.
Aberrant methylation in the urine sediment of primary prostate carcinoma had no significant level of correlation with patient demographic data, including age, histologic subtype, and staging of the tumor (data not shown). However, others have found a significant correlation between methylation and Gleason score, preoperative serum PSA, and tumor stage.22 The reason behind these discrepancies may be the indirect measurement of methylation in urine instead of primary tumor DNA and the different clinical subgroups represented in various studies (Table 3).
We investigated two reported methylated DNA repair genes (GSTP1 and MGMT) commonly found in various tumor types including prostate cancer. Using conventional MSP, we detected methylated GSTP1 alleles in the urine sediment from 27% of the patients with a methylated primary tumor.15 In the present study, GSTP1 was methylated in 48% of urine sediment samples. The reason for this discrepancy may be the primer design, but it should be noted that the sample size was different in both studies and that tumor stages and grade also differed. We and others also demonstrated that, for prostate cancer, there was no case in which a urine sediment DNA sample gave a positive GSTP1 methylation result in the absence of methylation in the corresponding tumor.15,23 We found MGMT methylation in 19% of urine sediment samples compared with less than 25% of primary prostate tumors by conventional MSP.22-24
Three cell cycle regulators (p16, ARF, and possibly Rassf1A) were included in our study. Previous reports of methylation in primary tumor tissues and prostate cancer cell lines ranged from 3% to 69% for p16 methylation,22,24-28 6% for ARF methylation,24 and 53% to 100% for Rassf1A methylation.22,29,30 These discrepancies may be a result of differences in the methylation assays used and the inclusion of tumors with different stages and grades.
Two metastatic suppressor genes, CDH1 and TIMP3, were frequently methylated in the urine sediment of prostate cancer patients (77% and 37%, respectively). Our finding of CDH1 methylation is similar with other studies based on conventional MSP in primary prostate tumors.31,32 Li et al31 reported that the severity of CDH1 methylation correlated with tumor progression. However, we found no correlation between CDH1 methylation and tumor grade and stage. Despite establishing a cutoff value (Table 2) in our controls, we found low levels of CDH1 methylation in five (6%) of 91 samples from individuals without any known genitourinary malignancy. TIMP3 is the third member of the TIMP family of proteins and is believed to play a significant role in controlling extracellular matrix remodeling. TIMP3 was found to be methylated in 24% to 28% of various human cancers.33-35 We found TIMP3 methylation in 37% of urine sediments from prostate cancer patients. As a diagnostic marker in urine DNA, TIMP3 may be limited by a persistent low frequency of methylation in normal controls. The use of retinoids to suppress tumor development has been evaluated in several animal models of carcinogenesis, including models of skin, breast, oral cavity, lung, hepatic, GI, prostatic, and bladder cancer.36 Retinoids act primarily via nuclear receptors encoded by the RAR? gene. Because the isoforms RAR?2 and RAR?4 are frequently methylated in other cancers,37-39 we investigated methylation of the RAR?2 promoter in urine sediment DNA. We and others have also reported methylation of RAR?2 in 53% to 95% of primary prostate tumor tissues.22,40
The APC protein is an integral part of the wnt-signaling mechanism, but it also plays a role in cell-cell adhesion, stability of the microtubular cytoskeleton, cell cycle regulation, and possibly apoptosis. We and others have demonstrated that the promoter regions of APC gene are aberrantly methylated in many types of cancer.18,20,41-44 In other studies, APC was found to be hypermethylated in 27% to 95% of primary prostate tumors20,22 compared with 54% methylation in urine sediment DNA reported in the present study.
To our knowledge, there have been few studies15,23,45-50 using an extended panel of methylation markers for the detection of prostate cancer in urine sediment. Thus, our methylation assay using nine different genes in the urine DNA confirms and extends previous observations. The high sensitivity (87%) using just four genes (p16, ARF, MGMT, and GSTP1) with undetectable methylation levels in all control samples (Table 2) is promising. The detection of tumor molecular signatures in body fluids has implications for the identification of high-risk patients and patients with preinvasive or early-stage lesions and for monitoring residual disease. Molecular approaches characterized by high specificity have variable sensitivity, perhaps because of the presence of low tumor DNA quantities in urine or the high level of contamination with normal DNA. Several approaches to improve assay sensitivity have been applied to clinical samples. Sensitivity has been improved over conventional MSP by performing a semi-nested MSP after a DNA preamplification step50 or a nested two-stage PCR,51 usually with decreased specificity for clinically definable disease. The sensitivity of QMSP in urine sediment could be further increased by isolating the aberrant cells from urine before DNA extraction or increasing the number of prostate cancer–specific markers. However, more sensitive assays may result in imperfect specificity and lack of quantitation, and tumors must be further validated in clinical samples.
Exfoliative material (present in urine, stool, sputum, bronchoalveolar lavage, bronchial brushings, and so on) offers diagnostic possibilities, but the sensitivity of current cytologic tests is low and virtually not used for prostate cancer detection. Diagnostic tools based on DNA alterations able to provide high specificity and sensitivity would clearly be of enormous benefit to patients, particularly if the specimens could be obtained by noninvasive means. To this end, the detection of aberrant methylation in urine DNA may offer a promising approach for the noninvasive diagnosis of prostate cancer. Apart from prostate cancer detection, it would be interesting to see whether the detection of aberrant methylation in the urine can be used in disease monitoring after curative surgery. If methylated DNA disappears shortly in urine after curative surgery, the reappearance of these markers may suggest recurrence of disease that may require more intensive screening and aggressive treatment. Thus, this simple and noninvasive method for detecting prostate cancer is readily automated and has many potential clinical applications, including primary diagnosis, monitoring for relapse, and measurement of therapeutic response. This study was performed on patients referred after PSA screening or other clinical suspicion. Thus, additional studies are necessary to elucidate the role of detecting aberrant methylation in urine as a tool for early detection and surveillance of prostate cancer either alone or in combination with serum PSA or digital rectal examinations. Moreover, other cancers, including bladder and kidney cancer, contribute cellular DNA to urine sediment. Thus, a panel of carefully selected methylation markers in urine sediment could be envisioned that both detects and then discriminates among a variety of urologic tumors.
Supplemental Tables
The supplemental tables are included in the full-text version of this article, available on-line at www.jco.org. They are not included in the PDF (via Adober Acrobat Readerr) version.
Authors' Disclosures of Potential Conflicts of Interest
Although all authors completed the disclosure declaration, the following author or immediate family members indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.
NOTES
Supported by National Cancer Institute grant No. U01-CA84986 and Oncomethylome Sciences, SA.
M.O.H. and O.T. contributed equally to this study.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Konishi N, Cho M, Yamamoto K, et al: Genetic changes in prostate cancer. Pathol Int 47:735-747, 1997
Alers JC, Krijtenburg PJ, Vis AN, et al: Molecular cytogenetic analysis of prostatic adenocarcinomas from screening studies: Early cancers may contain aggressive genetic features. Am J Pathol 158:399-406, 2001
Labrie F, Candas B, Dupont A, et al: Screening decreases prostate cancer death: First analysis of the 1988 Quebec Prospective Randomized Controlled trial. Prostate 38:83-91, 1999
Jones PA, Laird PW: Cancer epigenetics comes of age. Nat Genet 21:163-167, 1999
Toyota M, Ahuja N, OheToyota M, et al: CpG island methylator phenotype in colorectal cancer. Proc Natl Acad Sci U S A 96:8681-8686, 1999
Costello JF, Fruhwald MC, Smiraglia DJ, et al: Aberrant CpG-island methylation has non-random and tumour-type-specific patterns. Nat Genet 24:132-138, 2000
Sidransky D: Nucleic acid-based methods for the detection of cancer. Science 278:1054-1059, 1997
Hoque MO, Lee J, Begum S, et al: High-throughput molecular analysis of urine sediment for the detection of bladder cancer by high-density single-nucleotide polymorphism array. Cancer Res 63:5723-5726, 2003
Belinsky SA, Nikula KJ, Palmisano WA, et al: Aberrant methylation of p16(INK4a) is an early event in lung cancer and a potential biomarker for early diagnosis. Proc Natl Acad Sci U S A 95:11891-11896, 1998
Topaloglu O, Hoque MO, Tokumaru Y, et al: Detection of promoter hypermethylation of multiple genes in the tumor and bronchoalveolar lavage of patients with lung cancer. Clin Cancer Res 10:2284-2288, 2004
Esteller M, Sanchez-Cespedes M, Rosell R, et al: Detection of aberrant promoter hypermethylation of tumor suppressor genes in serum DNA from non-small cell lung cancer patients. Cancer Res 59:67-70, 1999
Wong IH, Lo YM, Zhang J, et al: Detection of aberrant p16 methylation in the plasma and serum of liver cancer patients. Cancer Res 59:71-73, 1999
Sanchez-Cespedes M, Esteller M, Wu L, et al: Gene promoter hypermethylation in tumors and serum of head and neck cancer patients. Cancer Res 60:892-895, 2000
Silva JM, Dominguez G, Garcia JM, et al: Presence of tumor DNA in plasma of breast cancer patients: Clinicopathological correlations. Cancer Res 59:3251-3256, 1999
Cairns P, Esteller M, Herman JG, et al: Molecular detection of prostate cancer in urine by GSTP1 hypermethylation. Clin Cancer Res 7:2727-2730, 2001
Sidransky D, Von Eschenbach A, Tsai YC, et al: Identification of p53 gene mutations in bladder cancers and urine samples. Science 252:706-709, 1991
Herman JG, Graff JR, Myohanen S, et al: Methylation-specific PCR: A novel PCR assay for methylation status of CpG islands. Proc Natl Acad Sci U S A 93:9821-9826, 1996
Harden SV, Tokumaru Y, Westra WH, et al: Gene promoter hypermethylation in tumors and lymph nodes of stage I lung cancer patients. Clin Cancer Res 9:1370-1375, 2003
Witten IH, Frank E, Kaufmann M: Data Mining: Practical Machine Learning Tools With Java Implementations. Morgan Kaufmann Publishers, San Francisco, CA, 2000
Jeronimo C, Henrique R, Hoque MO, et al: A quantitative promoter methylation profile of prostate cancer. Clin Cancer Res 10:8472-8478, 2004
Virmani AK, Tsou JA, Siegmund KD, et al: Hierarchical clustering of lung cancer cell lines using DNA methylation markers. Cancer Epidemiol Biomarkers Prev 11:291-297, 2002
Maruyama R, Toyooka S, Toyooka KO, et al: Aberrant promoter methylation profile of prostate cancers and its relationship to clinicopathological features. Clin Cancer Res 8:514-519, 2002
Goessl C, Krause H, Muller M, et al: Fluorescent methylation-specific polymerase chain reaction for DNA-based detection of prostate cancer in bodily fluids. Cancer Res 60:5941-5945, 2000
Konishi N, Nakamura M, Kishi M, et al: DNA hypermethylation status of multiple genes in prostate adenocarcinomas. Jpn J Cancer Res 93:767-773, 2002
Yamanaka M, Watanabe M, Yamada Y, et al: Altered methylation of multiple genes in carcinogenesis of the prostate. Int J Cancer 106:382-387, 2003
Konishi N, Nakamura M, Kishi M, et al: Heterogeneous methylation and deletion patterns of the INK4a/ARF locus within prostate carcinomas. Am J Pathol 160:1207-1214, 2002
Herman JG, Merlo A, Mao L, et al: Inactivation of the CDKN2/p16/MTS1 gene is frequently associated with aberrant DNA methylation in all common human cancers. Cancer Res 55:4525-4530, 1995
Jarrard DF, Bova GS, Ewing CM, et al: Deletional, mutational, and methylation analyses of CDKN2 (p16/MTS1) in primary and metastatic prostate cancer. Genes Chromosomes Cancer 19:90-96, 1997
Liu L, Yoon JH, Dammann R, et al: Frequent hypermethylation of the RASSF1A gene in prostate cancer. Oncogene 21:6835-6840, 2002
Kuzmin I, Gillespie JW, Protopopov A, et al: The RASSF1A tumor suppressor gene is inactivated in prostate tumors and suppresses growth of prostate carcinoma cells. Cancer Res 62:3498-3502, 2002
Li LC, Zhao H, Nakajima K, et al: Methylation of the E-cadherin gene promoter correlates with progression of prostate cancer. J Urol 166:705-709, 2001
Kallakury BV, Sheehan CE, Winn-Deen E, et al: Decreased expression of catenins (alpha and beta), p120 CTN, and E-cadherin cell adhesion proteins and E-cadherin gene promoter methylation in prostatic adenocarcinomas. Cancer 92:2786-2795, 2001
Kang GH, Lee S, Kim JS, et al: Profile of aberrant CpG island methylation along the multistep pathway of gastric carcinogenesis. Lab Invest 83:635-641, 2003
Alonso ME, Bello MJ, Gonzalez-Gomez P, et al: Aberrant promoter methylation of multiple genes in oligodendrogliomas and ependymomas. Cancer Genet Cytogenet 144:134-142, 2003
Schagdarsurengin U, Wilkens L, Steinemann D, et al: Frequent epigenetic inactivation of the RASSF1A gene in hepatocellular carcinoma. Oncogene 22:1866-1871, 2003
Evans TR, Kaye SB: Retinoids: Present role and future potential. Br J Cancer 80:1-8, 1999
Widschwendter M, Berger J, Hermann M, et al: Methylation and silencing of the retinoic acid receptor-beta2 gene in breast cancer. J Natl Cancer Inst 92:826-832, 2000
Yang Q, Mori I, Shan L, et al: Biallelic inactivation of retinoic acid receptor beta2 gene by epigenetic change in breast cancer. Am J Pathol 158:299-303, 2001
Ivanova T, Petrenko A, Gritsko T, et al: Methylation and silencing of the retinoic acid receptor-beta 2 gene in cervical cancer. BMC Cancer 2:4, 2002
Carmen J, Henrique R, Hoque MO, et al: Quantitative RARbeta2 hypermethylation: A promising prostate cancer marker. Clin Cancer Res 10:4010-4014, 2004
Esteller M, Sparks A, Toyota M, et al: Analysis of adenomatous polyposis coli promoter hypermethylation in human cancer. Cancer Res 60:4366-4371, 2000
Kawakami K, Brabender J, Lord RV, et al: Hypermethylated APC DNA in plasma and prognosis of patients with esophageal adenocarcinoma. J Natl Cancer Inst 92:1805-1811, 2000
Usadel H, Brabender J, Danenberg KD, et al: Quantitative adenomatous polyposis coli promoter methylation analysis in tumor tissue, serum, and plasma DNA of patients with lung cancer. Cancer Res 62:371-375, 2002
Dong SM, Kim HS, Rha SH, et al: Promoter hypermethylation of multiple genes in carcinoma of the uterine cervix. Clin Cancer Res 7:1982-1986, 2001
Gonzalgo ML, Pavlovich CP, Lee SM, et al: Prostate cancer detection by GSTP1 methylation analysis of postbiopsy urine specimens. Clin Cancer Res 9:2673-2677, 2003
Jeronimo C, Usadel H, Henrique R, et al: Quantitative GSTP1 hypermethylation in bodily fluids of patients with prostate cancer. Urology 60:1131-1135, 2002
Goessl C, Muller M, Straub B, et al: DNA alterations in body fluids as molecular tumor markers for urological malignancies. Eur Urol 41:668-676, 2002
Goessl C, Muller M, Heicappell R, et al: DNA-based detection of prostate cancer in blood, urine, and ejaculates. Ann N Y Acad Sci 945:51-58, 2001
Goessl C, Muller M, Heicappell R, et al: DNA-based detection of prostate cancer in urine after prostatic massage. Urology 58:335-338, 2001
Kersting M, Friedl C, Kraus A, et al: Differential frequencies of p16 (INK4a) promoter hypermethylation, p53 mutation, and K-ras mutation in exfoliative material mark the development of lung cancer in symptomatic chronic smokers. J Clin Oncol 18:3221-3229, 2000
Palmisano WA, Divine KK, Saccomanno G, et al: Predicting lung cancer by detecting aberrant promoter methylation in sputum. Cancer Res 60:5954-5958, 2000(Mohammad Obaidul Hoque, O)