Introduction

Prostate cancer remains as one of the most common cause of cancer related death among men in the US, with an estimated 29,720 deaths projected in 2013 [1,2]. Prostate cancer is curable if detected early by digital rectal examination and prostate specific antigen (PSA) in the serum [3]. With a sensitivity of 80%, PSA screening significantly increases the early diagnosis of prostate cancer, however the PSA screening has only 20% specificity, which may lead to unnecessary biopsies and overtreatment [4]. Hence diagnosis and management are confounded by the lack of cancer-specific diagnostic techniques to be used during early stages of disease. Novel approaches for the definitive detection and control of this cancer are urgently needed.

Molecular studies have revealed important events in prostate cancer development and progression and the emergence of new biomarkers may improve the diagnosis of prostate cancer [5]. Aberrant DNA methylation of gene promoters is characteristic of cancer cells, and several genes are epigenetically altered in a cancer specific manner. GSTP1 gene promoter methylation is widely characterized by several independent groups and is found to be have diagnostic value in prostate cancer [6,7]. To date, more than one hundred genes are reported as methylation targets in prostate cancer, which may represent a promising method for monitoring the occurrence and progression of cancer [7-9]. Recently, next generation sequencing based studies have also identified additional candidate genes [10]. The chromosomal region 3p21 is subject to frequent loss (heterozygous and homozygous) in multiple tumor types, the RAS association domain family protein 1 A (RASSF1A) gene located in this region, is a putative tumor suppressor which shares high sequence homology with a known mouse protein (Nore1) [11,12]. Studies suggests the role for RASSF1 as the effector that propagates the apoptotic effects of RAS by binding RAS in a guanosine triphosphate dependent manner [13]. Hypermethylation of the CpG islands within the RASSF1A promoter region are the major cause of loss of expression [14]. In many solid tumors, methylation of RASSF1A has been identified and its frequency varies between 30% to 50% [12]. Therefore, RASSF1A methylation affects its tumor suppressor role and is associated with unfavorable prognosis [15,16].

Despite a number of individual studies performed in prostate cancer patients, the diagnostic value of RASSF1A methylation status in prostate cancer and the relationship between RASSF1A methylation and pathological features or tumor stage remains controversial. Therefore, the purpose of this study was to conduct a meta-analysis on the prognostic value of RASSF1A methylation status in prostate cancer, and the relationship between RASSF1A methylation and pathological stage, Gleason score, and PSA level. We also assessed the sensitivity and specificity of RASSF1A methylation in the body fluids and tissues on prostate cancer detection.

Materials and Methods

Results

Association between RASSF1A promoter methylation and pathological stage, Gleason score, and PSA levels in PCa cases

Under the random-effects model, the pooled OR of RASSF1A methylation in prostate cancer cases, compared to non-cancer controls, was 14.73 with 95%CI = 7.58–28.61(Table 2). In the stratified analysis by sample, significantly increased risk was associated with RASSF1A methylation in both fluid samples (OR = 26.27, 95%CI = 7.79–88.61) and in tissues (OR = 12.28, 95%CI = 5.86–25.76). As stratified analysis by method, significantly increased risk was also found in MSP (OR 6.75, 95% CI = 3.42–13.22) and QMSP (OR = 31.50, 95%CI = 10.95–90.62) (Figure 1A). We also conducted an analysis of the relationship between the pathological stage, Gleason score, and PSA levels among PCa cases and RASSF1A methylation. Seven studies [17,19,21-23,25,26] had sufficient information to perform this analysis (Table S2), we found no significant association between groups with the appropriate models except for the Gleason score (OR=2.35, 95% CI: 1.56–3.53, I²=32.3%). Details are shown in Table 3.

Variables	pa	OR (95% CIs)	Heterogeneity Test(I2, p-value)
RASSF1A
Total	16	14.73 (7.58-28.61)	75.5%,0.000
Sample Type
Fluid	4	26.27 (7.79-88.61)	56.6%,0.075
Tissue	12	12.28 (5.86-25.76)	75.5%,0.000
Method
QMSP	7	31.50 (10.95-90.62)	61.8%,0.015
MSP	8	6.75 (3.42-13.22)	67.9%,0.003
Pyrosequencing	1	148.50 (25.45-866.36)	NA

Table 2. Stratification analyses of RASSF1A methylation and prostate cancer risk.

MSP, Methylation-Specific PCR; QMSP, Quantitative real time Methylation Specific PCR; OR,Odds Ratio;

CI, Confidence Interval; ^a Number of studies; NA, Not Applicable.

CSV

Download CSV

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Forest plot showing the association between RASSF1A methylation and prostate cancer.

(A) Forest plot of 16 studies showed that RASSF1A methylation was associated with PCa risk as stratified analysis by methods. (B) Forest plot after sensitivity analysis removing 6 studies showed the same risk without heterogeneity (p= 0.089). The diamond symbol represents the overall estimate from the meta-analysis and its CI (confidence interval), while its center is positioned on the value for the overall effect estimate. Diamond’s width depicts the width of the overall CI.

https://doi.org/10.1371/journal.pone.0075283.g001

Clinicopathology	Category	OR (95% CIs)	Heterogeneity test (I2, P-value)	Publication bias test (P-value)
Gleason score	GS≥7	2.35 (1.56-3.53)	32.3%,0.182	0.40
	GS<7
PSA levels	PSA>4	2.19 (0.27-17.76)	67.4%,0.046	0.04
	PSA≤4
Pathological stage	Ⅲ&Ⅳ	2.01 (0.77-5.23)	68.4%,0.007	0.73
	Ⅰ&Ⅱ

Table 3. Association between RASSF1A promoter methylation and Pathological stage, Gleason score, and PSA levels in PCa cases.

OR, Odds Ratio; CI, Confidence Interval; PSA, Prostate Specific Antigen; GS Gleason Score.

CSV

Download CSV

The results of meta-regression indicated that the trend in ORs correlated with methylation detection methods, which accounted for part of the heterogeneity (coefficient =−1.69, p=0.024, adjusted R² = 47.41%), However, other factors such as sample types (p=0.525), year of publication (p=0.608), and the origin of the patients (p=0.847) could not explain the heterogeneity. We also performed sensitivity analyses to determine the effects of omitting a single study on the overall effect where six independent studies were found to introduce some source of heterogeneity. Hence when we excluded the six studies that would likely introduce high heterogeneity due to different types of cases and controls, we observed a decrease in heterogeneity (p = 0.089) with stable conclusion (OR = 14.45, 95%CI = 8.35–25.01) (Figure 1B).

We did not observe any publication bias (Egger’s test; p=0.066), and further confirmed this observation by implementing a trim-and-fill method. Meta-analysis with or without the trim-and-fill method did not draw different conclusions, indicating that our results were statistically robust. The Egger’s test suggested the absence of publication bias (P>0. 05) in studies of association between RASSF1A methylation and pathological stage or Gleason score (Table 3).

Specificity and sensitivity of RASSF1A promoter methylation using different types of samples

The pooled specificity for all included studies was 0.87 (95% CI: 0.72–0.94), and the pooled sensitivity was 0.76 (95% CI: 0.55–0.89). For the traditional biomarker, the sensitivity of PSA varied, but the specificity was generally low at about 20%, which suggested that the RASSF1A methylation test has a much higher specificity than the PSA test. The p-value for heterogeneity was <0.001, indicating a significant heterogeneity. There was no evidence of publication bias (p=0.13). In addition, we classified all samples into two groups according to specimen type (fluid/tissue). Among the fluid studies (urine/blood/plasma), the pooled sensitivity and specificity was 0.60 (95% CI: 0.08–0.96) and 0.93 (95% CI: 0.75–0.98), respectively. For the tissue studies, the pooled sensitivity and specificity was 0.79 (95% CI: 0.64–0.89) and 0.84 (95% CI: 0.63–0.94), respectively and the S–ROC curve is presented in Figure 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Meta-analysis with the S-ROC curve.

SENS: Sensitivity, SPEC: Specificity, AUC: Area under the curve.

https://doi.org/10.1371/journal.pone.0075283.g002

Discussion

The results of our meta-analysis showed that RASSF1A methylation in prostate cancer was significantly associated with cancer risk when monitored in urine, blood or tissue samples. It was also associated with increased risk for high Gleason score. The pooled specificity (0.87) of RASSF1A was much higher than the specificity of PSA (20%). In addition the specificity in each subgroup stratified by sample type remained above 0.84 and the sensitivity of RASSF1A also remained high above 0.60. These results suggested that RASSF1A promoter methylation would be a valuable biomarker in diagnosing PCa.

DNA methylation is a common mechanism for inactivating tumor-suppressor genes in cancers [10]. Monitoring aberrant methylation patterns have aided in the detection of tumor cells in clinical specimens such as tissue biopsies or body fluids. RASSF1A is a putative tumor suppressor gene and plays an important role in the regulation of different types of human tumors. Previous reports demonstrated that genetic variation of RASSF1A affect prostate cancer susceptibility and the frequency of RASSF1A methylation was found to be significantly higher in patients’ group compared with control [13]. To further confirm RASSF1A promoter methylation status in PCa patient’s diagnosis, we carried out a meta-analysis of 16 studies involving 1431 cases and 565 controls to derive a more precise estimation of the association. Our results suggested that RASSF1A methylation is a potential risk factor for PCa as detected both in body fluid and tissues. The frequency of RASSF1A methylation in PCa cases was 14.73 times higher than that in control individuals. In addition, due to differences between cases and controls between studies, we performed sensitivity analyses and removed select studies to make the data more homogeneous and observed no major changes to our conclusion. Importantly, frequent methylation of RASSF1A gene showed significant associations with Gleason score, although it did not correlate with pathological stage or PSA levels in this study.

As shown previously, the PSA test had varying sensitivity and poor specificity (about 20%) [1]. Tests that provide higher specificity rather than sensitivity and complement PSA testing is the current requirement. Promisingly, our results show that, RASSF1A methylation testing has a potential to complement PSA screening due to its high specificity. Here we believe a sequential testing model will be more helpful where PSA test will be initially used to identify patients with elevated PSA levels, followed by the RASSF1A methylation test to increase true positivity. Patients with negative RASSF1A test can be placed on watchful waiting, while individuals with positive results can be recommended for biopsies. Thus sequential testing will greatly reduce the unnecessary biopsy recommendations based solely on PSA testing.

The present study has several limitations. First, the heterogeneity in the methods employed such as MSP, Q-MSP and pyrosequencing to monitor gene promoter methylation. Based on studies conducted, it will be helpful if a consensus could be arrived for a standard testing. Sequencing based techniques are generally considered to have superior performance as compared to methodologies that employ PCR alone. The rapidly gaining momentum in using next generation sequencing (NGS) assays to monitor genomic and epigenomic changes will certainly usher in advancements in this area [31]. One can foresee development of comprehensive diagnostic tests that will simultaneously measure transcript expression and DNA methylation changes of a panel of biomarkers in addition to identifying genetic aberrations such as gene fusions and copy number variations for each patient [32]. The current study along with others helps in shortlisting biomarkers of great potential that will be given special attention while analyzing NGS results which usually produces a long list of methylated regions. Next, the diverse sample types employed in these studies which include plasma, serum, urine, whole blood and tissues is a potential source of heterogeneity. Here again development of high performance robust assays with high sensitivity and specificity may help overcome this in the near future.

Conclusion

This meta-analysis has demonstrated that RASSF1A methylation in prostate cancer was associated with cancer risk across different study populations. RASSF1A is a promising biomarker for screening and identifying PCa especially when combined with the PSA test to bring down the rate of unnecessary biopsies. Future studies with larger sample cohort and novel assays based on next generation sequencing approaches will help in further strengthening our observations.

Supporting Information

Acknowledgments

We would like to thank Dr. Srinivasan Yegnasubramanian (Johns Hopkins University) for providing related raw data and Dr. Saravana Mohan Dhanasekaran (University of Michigan) for discussion and manuscript preparation.