Abstract
Background: Early diagnosis of prostate cancer and identification of new prognostic factors remain main issues in prostate cancer research. In this study, we sought to test a panel of cancer-specific markers in urine samples as an aid for early cancer diagnosis. Materials and Methods: Sedimented urine samples of 66 candidates for needle biopsy were tested. Real time-polymerase chain reaction (RT-PCR) was applied to detect the expression of transmembrane protease serine-2 and Ets-related gene fusion (TMPRSS2–ERG), Ets-related gene (ERG), prostate cancer antigen-3 (PCA3), and serine peptidase inhibitor kazal type-1 (SPINK1) transcripts. For testing of the methylation status of Glutahione S-tranferase P (GSTP1) and Ras association domain family member-1(RASSF1A) promoter region, methylation-specific PCR (MSP-PCR) was applied. Results: Among the tested parameters, the presence of TMPRSS2–ERG (OR=9.044, 95% CI=2.207-37.066, p=0.002), as well as a positive test result for PCA3 (OR=7.549, 95% CI=1,858-30,672, p=0.005) were associated with the subsequent diagnosis of prostate cancer. A multivariable logistic regression including all the significantly associated variables [prostate-specific antigen (PSA), digital rectal examination (DRE), TMPRSS2-ERG and PCA3], yielded a model with area under the receiver-operating characteristic curve (AUC) =0.894 (95% CI=0.772-1.00). Conclusion: A multiplexed quantitative PCR analysis on sedimented urine, in conjunction with the results of serum PSA levels and DRE, has the potential to accurately foresee subsequent needle biopsy outcomes. On the basis of the above, algorithms may be designed to guide decisions for needle biopsy.
Prostate cancer is one of the most frequently diagnosed types of non-skin cancer and a leading cause of cancer-related deaths for men worldwide (1). Diagnosis and management are complicated by the lack of cancer-specific markers to assist for diagnosis during the early stages of the disease, and to predict and monitor response to therapy.
The use of prostate-specific antigen (PSA) as a screening and monitoring marker for prostate cancer is widespread (2). Although PSA monitoring has led to higher prostate cancer detection rates, it has also substantial drawbacks. PSA is specific for tissues of prostatic origin but is not cancer-specific. Serum PSA levels are often elevated in benign prostatic hyperplasia and prostatitis. Thus, the PSA testing is associated with a significant false-positive rate, with a high proportion (>50%) of the resulting biopsies proving negative for cancer. Furthermore, studies have also indicated that low levels of PSA do not preclude prostate cancer, and that 15% of men with PSA 0-4 mg/ml developed prostate cancer (3). Moreover large randomized trials assigned modest effects upon mortality rates of PSA screening during the first decade of follow-up (4, 5).
Thus the need for non-invasive methods that can accurately assist in the early detection of prostate cancer, still exists. To address this issue, additional markers have been investigated including genetic and epigenetic alterations. Among those the most promising are: Genes that are specifically overexpressed in prostate cancer cells, such as prostate cancer antigen-3 (PCA3), alpha methyl-CoA racemase (AMACR), serine peptidase inhibitor kazal type1 (SPINK1) (6-8); prostate cancer-specific gene alterations, mainly the fusion genes involving transmembrane protease serine-2 (TMPRSS2) and E-twenty six (ETS) family members (9); and prostate cancer-specific methylation alterations of gene promoter regions Glutahione S-tranferase P (GSTP1), Ras association domain family member 1(RASSF1A) (10, 11).
Taking advantage of the fact that prostate cells can be detected in blood and urine, prostate cancer-specific markers can be tested through a urine or blood diagnostic test (12).
The goal of the present study was to assess the diagnostic efficacy of a number of markers associated with malignant transformation when used individually or combined, in order to achieve a pre-biopsy prediction of prostate cancer and contribute to early diagnosis in a non-invasive manner.
Materials and Methods
Urine collection and DNA and RNA isolation. This study was approved by the St. Savvas Anticancer Hospital Scientific Committee and informed consent was obtained from all participants. Urine samples were collected from 66 men who were admitted for transrectal ultrasound (TRUS)-guided prostate biopsy at the Department of Urology in St. Savvas Anticancer Hospital on the basis of PSA level and/or abnormal DRE. At least 30 ml of urine sample were collected from each patient immediately following a DRE. All patients included in this study were referred for prostate biopsy for the first time.
Urine was voided into sterile collection cups containing 5 ml of 0.5 M EDTA. A minimum of 30 ml of the collected urine was centrifuged at 5000 ×g for 20 min at 4°C and the resulting cell pellet was processed for DNA and RNA extraction.
Total RNA and DNA was isolated using an RNA, DNA extraction kit (NucleoSpin RNAXS, NucleoSpin RNA/DNA buffer set; Macherey-Nagel, Duren, Germany), according to the manufacturer's instructions.
Quantitative reverse transcription-polymerase chain reaction (qRT-PCR). Quantitative RT-PCR was used to assess the expression of four biomarkers [transmembrane protease serine2 and Ets related gene fusion (TMPRSS2–ERG), Ets related gene (ERG), PCA3, SPINK1] and the PSA transcript as control for prostate-derived cells. Total RNA (8 μl) was reverse-transcribed in 20 μl reactions using the SuperScript III First-strand Synthesis Super Mix (Invitrogen, Carlsbad, California, USA) and random hexamers as primers.
The cDNA was subjected to TaqMan PCR amplification for both the TMPRSS2–ERG fusion gene and PSA transcripts using primers and PCR conditions, as described in Table I.
All PCR amplifications were performed in 20 μl reaction mixture, containing 1× Platinum Quantitative PCR Super Mix (Invitrogen), 5 pmol of each primer and 5 pmol of TaqMan probes. Syber Green qRT-PCR was applied for PCA3, ERG, SPINK1 and PSA control expression using primers and PCR conditions described in Table I.
All PCR amplifications were performed in 20μl reactions, containing 1× Platinum Syber Green Quantitative PCR Super mix (Invitrogen), and 5 pmol of each primer.
The qPCR results for all the genes were calculated using RelQuant software (Roche Molecular Biochemicals, Mannheim, Germany) and are expressed as the ratio of the target gene/PSA ×1000. The results were considered valid only if the PSA Ct was <35 cycles.
Methylation-specific PCR (MSP). The extracted DNA was subjected to sodium bisulphide modification using the EpiTec Bisulfite kit (Qiagen, GmbH, Germany), following the manufacturer's instructions. For the detection of methylated and unmethylated GSTP1 and RASSF1A alleles MSP was applied using primers and PCR conditions, as previously described (13, 14) (Table I). PCR products were electrophoresed on 2.5% agarose gels and visualised with ethidium bromide staining.
Sequence of primers, probes and annealing temperatures (TM).
Statistical analysis. Pre-biopsy clinical parameters were compared between men diagnosed with prostate cancer vs. men without prostate cancer using Mann-Whitney tests for continuous variables and Fisher's exact tests for categorical variables. Exact univariate logistic regression was used to examine the association between PSA levels, DRE outcome, PCA3 expression levels and TMPRSS2–ERG fusion with the presence or absence of prostate cancer upon prostate needle biopsy (PNB).
The performance of each biomarker as a screening test was evaluated, and sensitivity, specificity, and the area under the receiver-operating characteristic curve (AUC), with 95% confidence intervals (CI), were calculated (15). A multivariable logistic regression model predicting the diagnosis of prostate cancer on biopsy was then developed using backward selection. The initial model contained PSA, DRE, PCA3 and TMPRSS2–ERG fusion as potential predictor variables. The inclusion of PSA and PCA3 in the multivariable model was evaluated both as a continuous variable and as a categorical variable at a cut-off of 10 ng/ml and a ratio value of 30, respectively. Calibration was evaluated by calculating the AUC. All analyses were carried out using SPSS ver. 17.0 (SPSS Inc., Chicago, Illinois, USA) at the 0.05 level of significance and two-sided p-values are reported.
Pre-biopsy clinical parameters and biomarker status according to diagnosis on prostate needle biopsy.
Performance of prostate-specific antigen (PSA), digital rectal examination (DRE), prostate cancer antigen (PCA3) and transmembrane protease serine-2 and Ets-related gene (TMPRSS2–ERG) fusion gene in predicting prostate cancer diagnosis following prostate needle biopsy (PNB): Univariate analyses.
Results
The mean age of men included in this study was 66.3 (range 45-83) years. The pre-biopsy clinical parameters and biomarker status of 66 men are presented in Table II. Fourteen (21%) men received diagnosis of prostate cancer upon biopsy and 52 (79%) men were found to be free of prostate cancer. For 10 of the cancer-positive patients, the Gleason score was 6 (3+3), it was7 (3+4) in 3, and one had a Gleason score of 10 (5+5).
The univariate analysis of the pre-biopsy clinical parameters revealed that men diagnosed with prostate cancer had significantly higher PCA3 levels compared to men free of prostate cancer (p<0.01). In addition, significant associations were found between prostate cancer and the TMPRSS2–ERG fusion gene, positivity and abnormal DRE (both p<0.01). The other pre-biopsy clinical variables did not differ significantly between men found upon biopsy to have prostate cancer and those free of cancer.
The predicted classification of the presence or absence of prostate cancer on biopsy using PSA as a binary variable and the experimental urine biomarkers is as shown in Table II. Twenty-six (39%) men had positive TMPRSS2–ERG fusion status, while 28 (42%) men tested with a high level of urine PCA3, and 31 (47%) men had an abnormal DRE. In univariate logistic analysis of these biomarkers, men diagnosed with prostate cancer were more likely to have a positive test for TMPRSS2–ERG fusion (OR=9.044, 95% CI=2.207-37.066; p=0.002), a positive test result for PCA3 (OR=7.549, 95% CI=1.858-30.672; p=0.005) and an abnormal test result for DRE (OR=5.867, 95% CI 1.456-23.636; p=0.013).
Clinical application of the multivariate model in selecting patients for biopsy.
Nineteen (29%) men had elevated serum PSA (cut-off of 10 ng /ml) but were not more likely to have prostate cancer on biopsy, as compared to men with PSA <10 ng /ml (OR=1.508, 95% CI=0.431-5.280; p=0.521).
Although PCA3, TMPRSS2–ERG fusion and DRE had the highest sensitivity of 79% in predicting the presence of prostate cancer, PSA was more specific (specificity=73%). TMPRSS2–ERG fusion had the greatest discriminatory value, with an AUC of 0.75 (95% CI=0.604-0.893), whereas the performance of serum binary PSA at a cut-off of 10 ng/ml (AUC=0.544, 95% CI=0.370-0.718) was not as effective.
A predictive model including only those biomarkers that showed significant association with malignancy on univariate analysis and PSA was constructed. A multivariable logistic regression, where PSA and PCA3 are considered as binary variables (at cut-offs of 10 ng /ml and ratio of 30, respectively), yielded a model wherein DRE and TMPRSS2–ERG fusion were both significant (p-values=0.028 and 0.005, AUC=0.818, 95% CI=0.690-0.946). Furthermore, a multivariable logistic regression including all the variables, where PSA and PCA3 were considered as continuous variables, yielded a model with AUC=0.894, 95% CI=0.772-1.000). When the parameters PSA, DRE, TMPRSS2–ERG fusion and PCA3 were removed from the model one at a time the p-values were=0.039, 0.046, 0.001 and 0.079l, respectively. Finally, the discriminatory ability of a multivariate model combining all the variables as binary variables was slightly weaker than the model where the PSA and PCA3 were considered as continuous variables (AUC=0.852, 95% CI=0.723-0.980).
Table III shows the efficacy of the latter model to support the decision to advise a patient to undergo prostate biopsy. In a less restricted biopsy selection scenario at high sensitivity (92.9%), individuals with either a TMPRSS2–ERG fusion or an abnormal DRE and either PSA or PCA3 at a high level will be recommended for prostate biopsy. However, with the more restricted biopsy selection scenario at high sensitivity and specificity (85.75 and 78.8%, respectively), a positive TMPRSS2–ERG fusion and either abnormal DRE or a high level of PCA3 are required simultaneously in order to make a recommendation for prostate biopsy.
Discussion
The performance of serum PSA as a biomarker for prostate cancer diagnosis is undermined by several limitations, including a relatively high proportion of false-positive and false-negative test results because PSA is expressed in normal as well as in cancerous cells. Hence, there is a need to identify biomarkers that are more specific for prostate cancer, which are easily tested and that may thereby improve prospects for prostate cancer screening.
Research during the past five years has provided a large inventory of candidate biomarkers that might prove to be more selective and potentially more useful than individual biomarkers when combined in a multiplex model (16, 17).
The parameters tested in this study were chosen either because they are being used in routine screening (DRE, serum PSA), or have been recently approved for diagnostic use (PCA3), or were found – in most previous studies – to be significantly associated with prostate cancer.
Among them are the recurrent fusion genes involving the androgen-regulated gene TMPRSS2 and members of the ETS family transcription factors (ERG, ETV1, and ETV4). TMPRSS2–ERG is by far the most common subtype of ETS fusion accounting for approximately 85% of all ETS fusion-positive samples (18, 19). One consequence of the presence of fusion gene is the overexpression of ERG, assigning ERG expression a putative role as a prostate cancer biomarker.
PCA3 is a non-coding RNA and its expression is restricted solely to prostate (it not being expressed in any other normal human tissue nor any other tumour type). PCA3 RNA is highly overexpressed in 95% of tumours compared to normal or benign hyperplastic prostate tissue (6, 20).
The expression of SPINK1 has been found to be increased in prostate cancer. It was also shown that SPINK1 is exclusively expressed in tumours without the TMPRSS2–ERG fusion (7).
Methylation of regulatory sequences at the GSTP1 and RASSF1A gene loci is a common molecular alteration in prostate cancer and it has also been demonstrated in high-grade PIN, but not in normal prostate tissue (21).
Some genes that are overexpressed, hypermethylated, or mutated in prostate cancer tissues may be detectable in urine. Prostate cancer cells that are ex-foliated or which enter the circulation even during early stages of tumour growth might display characteristics of cancer that is either likely to metastasize or remain indolent.
Recent studies have shown that all these markers can be reliably detected in the urine of patients collected after DRE, which has the advantage of being a non-invasive procedure (22, 23).
Although urine-based testing for PCA3 expression has already been documented in large screening programs, the feasibility of testing based on other markers has not been rigorously evaluated. More importantly, single marker tests, such as those based on PSA or PCA3, ignore the heterogeneity of cancer development and the innate heterogeneity within tumour cells and may only detect a proportion of cancer cases (24). To overcome this limitation, multiplexing, or combining biomarkers for cancer detection can improve the performance of a test in predicting the biopsy result (25-27).
In this study we used the simplest procedure so far described for simultaneous extraction of RNA and DNA and real-time PCR without any kind of target enrichment.
All the markers were first evaluated by univariate analysis with DRE (p=0.008), PCA3 (p=0.008) and TMPRSS2–ERG (p=0.001) showing significant association with biopsy-confirmed prostate cancer. Serum PSA levels before biopsy were not associated with subsequent prostate cancer detection in this cohort (p=0.301)
SPINK1 overexpression in urine sediments was previously shown to be associated with prostate cancer detection but was not found to be significantly associated in this study (28). This may be due to sample variability or differences in the methods used, as there was no whole-transcriptome amplification step in our study.
Whereas TMPRSS2–ERG fusion was significantly associated with the detection of prostate cancer on biopsy, ERG overexpression was not, suggesting that cells from other tissues may also contribute to ERG transcripts in urine, thus masking any differences.
Neither GSTP1 nor RASSF1A promoter hypermethylation, found in previous studies to be specific markers for prostate cancer in tissues and urine sediments, were found to be statistically significant predictors for prostate cancer detection in this study (11, 29). The sample heterogeneity and differences in method sensitivity may account for the observed discrepancies.
However, the main issue addressed in this study is the feasibility of using urine sediment-based tests for prostate cancer screening.
In this study, we detected TMPRSS2–ERG fusion in post DRE urine sediments in 11 out of 14 prostate cancer-positive cases on biopsy, whereas 37 out of 52 biopsy-negative patients were TMPRSS2–ERG negative, resulting in a specificity of 71% and a sensitivity of 79%. This findings are in agreement with all the previous studies that assessed the use of urine-detected TMPRSS2–ERG fusion gene as a prostate cancer-specific marker (17-19).
At this point, the TMPRSS2–ERG fusion gene outperformed all the other markers tested in this study namely PCA3 (specificity 67%, sensitivity 79%), PSA (specificity 73%, sensitivity 36%), ERG (specificity 23%, sensitivity 88.6%), RASSF1A methylation (specificity 69.8%, sensitivity 38.4%), GSTP1 methylation (specificity 75.4%, sensitivity 38.4%).
The performance of the individual tests in prostate cancer prediction at biopsy was found to be improved when all the significant cancer-associated markers (DRE, PCA3, TMPRSS2–ERG) were combined with serum PSA. In a multivariate regression analysis where PSA and PCA3 were considered as continuous variables, DRE, PSA, PCA3 and TMPRSS2–ERG fusion were significantly associated with prostate cancer detection on biopsy. The discriminatory ability of a multivariate model combining all the markers as binary variables (AUC=0.852, 95% CI=0.723-0.980) (PSA at cut of 10 ng/ml and PCA3 at 30) was found to exceed that of PSA-alone (AUC=0.544, 95% CI=0.370-0.718) DRE-alone (AUC=0.701, 95% CI=0.551-0.850) and both combined (AUC=0.724, 95% CI=0.557-0.891).
Our findings demonstrate the value of and support the efficacy of combining the detection of TMPRSS2–ERG fusion gene and of PCA3 score with the routinely used DRE and serum PSA findings to formulate a clinical algorithm guiding the decision for prostate biopsy.
The multistep model is illustrated in Table IV, suggesting two possible selection scenarios: a less restricted one with a sensitivity and specificity at 92.9% and 60%, respectively, and a more restricted one with a lower sensitivity of 85.7% but a much higher specificity of 78.8%. Selecting the less restricted scenario 32 patients out of 66 would have avoided an unnecessary biopsy, whereas all but one of the cancer patients would have been identified. On the basis of the alternative scenario, 43 patients out of 66 would have avoided an unnecessary biopsy, whereas all but two of the cancer patients would have been identified.
To sum up, this pilot study demonstrates that a multiplexed quantitative PCR analysis of two specific genes, on sedimented urine from patients presenting for prostate biopsy, in conjunction with the results of routine practices, such as serum PSA level monitoring and DRE outcome, has the potential to accurately foresee needle biopsy outcome.
These results support the examination of larger cohorts for further validation of this relatively simple and feasible procedure for assessing the necessity of patients to undergo biopsy.
- Received September 25, 2012.
- Revision received November 9, 2012.
- Accepted November 9, 2012.
- Copyright© 2013 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
