Purpose: Metastatic prostate cancer is a major cause of death of men in the United States. Expression of met, a receptor tyrosine kinase, has been associated with progression of prostate cancer.

Experimental Design: To investigate met as a biomarker of disease progression, urinary met was evaluated via ELISA in men with localized (n = 75) and metastatic (n = 81) prostate cancer. Boxplot analysis was used to compare the distribution of met values between each group. We estimated a receiver operating characteristic curve and the associated area under the curve to summarize the diagnostic accuracy of met for distinguishing between localized and metastatic disease. Protein-protein interaction networking via yeast two-hybrid technology supplemented by Ingenuity Pathway Analysis and Human Interactome was used to elucidate proteins and pathways related to met that may contribute to progression of disease.

Results: Met distribution was significantly different between the metastatic group and the group with localized prostate cancer and people with no evidence of cancer (P < 0.0001). The area under the curve for localized and metastatic disease was 0.90, with a 95% confidence interval of 0.84 to 0.95. Yeast two-hybrid technology, Ingenuity Pathway Analysis, and Human Interactome identified 89 proteins that interact with met, of which 40 have previously been associated with metastatic prostate cancer.

Conclusion: Urinary met may provide a noninvasive biomarker indicative of metastatic prostate cancer and may be a central regulator of multiple pathways involved in prostate cancer progression.

Translational Relevance

Prostate cancer is the most common cancer in men. Although prostate-specific antigen has been used to monitor disease progression and for screening purposes, it is not a good marker for tumor progression as the disease becomes hormone insensitive. In this study, we examined the urinary biomarker met as a possible marker of metastatic PCa. Our data shows that urinary met is a significant marker for patients with metastatic disease. Using yeast hybrid protein-protein interaction studies in combination with a bioinformatics approach, we show additional proteins that are related to met and may be part of the conversion of prostate cancer form a local disease to a systemic one. Our study sets the foundation for additional studies of biomarkers of metastatic prostate cancer.

Prostate cancer (PCa) is the most common non–skin cancer in men and the second most common cause of cancer deaths of men in the United States, with nearly all deaths secondary to metastatic disease (1). In the last two decades, serum prostate-specific antigen (PSA) levels have been used for screening for PCa, as a marker of recurrence after therapy, and as a marker of metastatic disease; however, the utility of this test for screening and prediction of treatment outcome remains controversial (2). One limitation of PSA measurement is the finding that patients with metastatic disease in some cases have undetectable PSA levels (3). Conversely, patients with elevated PSA levels posttreatment may never develop metastatic disease (4). Furthermore, it has been reported that PSA values may not directly correlate and, in some cases, may be inversely correlated, with disease progression. This phenomenon may be related to high levels of PSA resulting from architectural distortion and low levels due to the loss of the ability to secrete PSA by tumor cells (5, 6). Despite the widespread use of PSA, concerns about the utility of PSA as a stand-alone test for monitoring disease progression have prompted investigators to seek alternative markers of disease progression.

Met, a receptor tyrosine kinase, has received much attention for its overexpression and/or mutation in a number of malignancies (7). Multiple lines of evidence point to the importance of met in PCa initiation and progression. In vitro studies have shown that met expression is inversely correlated with androgen receptor expression (8). In murine models, met ectodomain shedding has been shown to correlate with tumor burden across many histologies including PCa, and as tumor burden increases, there is an increase in urinary and plasma met levels (9). In prostatectomy specimens, met expression was shown to correlate with progressive disease (10). A subsequent study showed similar results reporting uniform met positivity in prostate bone metastasis (11). Naughton et al. (12) extended these results by showing hepatocyte growth factor (HGF; the ligand of met) to be significantly elevated in the serum of metastatic PCa patients when compared with patients with localized cancer. Thus, both experimental as well as clinical data have associated met activity with PCa progression.

Met has not been evaluated as an independent predictor of disease progression in PCa patients. To investigate the potential of met as a minimally invasive biomarker of disease progression, we have evaluated urinary met in patients with both localized and metastatic PCa. Furthermore, we used met as a bait in a protein-protein interaction (PPi) study combined with an informatics approach to elucidate the proteins and pathways that may be involved in the progression of PCa from a local to a metastatic disease. The data presented indicate that PSA and met levels are inversely correlated in prostate tumor cell lines and that higher urinary met levels are found in patients with metastatic PCa. Moreover, the PPi-informatics approach identified a number of proteins previously recognized as being involved in PCa, but more importantly revealed a number of interactions between molecules that had not been previously reported in PCa metastases.

Cell lines and culture. The LNCaP [lymph node metastasis, (+) PSA], PC-3 [bone metastasis, (−) PSA], DU145 [brain metastasis, (−) PSA], and 22Rv1 [prostate, (+) PSA] cell lines (American Type Culture Collection), were maintained in DMEM supplemented with 10% fetal bovine serum in 5% CO2/95% air at 37°C.

In vitro met and PSA quantitation. Cells were grown to 60% confluence, media changed, and 12 h later, cells were collected, protein quantified (Bradford assay), and ELISAs (total met and PSA; R&D Systems) were carried out in duplicate (40 μg) according to manufacturer's protocol.

Urine collection and processing. Urine was collected prospectively on clinical study (02C0064), approved by the Institutional Review Board of the National Cancer Institute, NIH, from 2002 to present. Specimens were collected from 75 patients with localized PCa (18, 32, and 25 with low, intermediate and high risk, respectively) before receiving radiation therapy and 20 males without PCa. Low-risk PCa is defined as clinical stage of T1 to T2a, Gleason score of 2 to 6, and PSA of <10 ng/mL; intermediate risk is defined as clinical stage of T2b to T2c, Gleason score of 7, and PSA of 10 to 20 ng/mL; and high risk is defined as clinical stage of T3a, Gleason score of 8 to 10, and PSA of >20 ng/mL. Three patients from the intermediate-risk group and one patient from the low-risk group had specimens with nondetectable levels of met and were excluded from analysis (poor specimen). Thus, we analyzed met from 71 patients with localized disease. Excluding the data from these four patients will attenuate any differences when comparing localized and metastatic disease groups; thus, a test of our primary comparison will be conservative. The clinical and pathologic characteristics of patients with localized PCa are shown in Table 1. We subsequently analyzed 81 patient's samples that had been collected on Protocols 00-C-0083 and 04-C-0257, clinical trials using chemotherapy in patients with radiographic evidence of metastases from PCa. Urine was stored at −20°C, then thawed, centrifuged at 3,000 rpm for 10 min at 4°C, and supernatant was used for assay. Urinary creatinine levels were obtained on the Bayer DCA 2000+ Analyzer (Bayer HealthCare) according to manufacturer's protocol.

Table 1.

Clinical and pathologic characteristics of patients with localized PCa

Overall (n = 71)Low (n = 17)Intermediate (n = 29)High (n = 25)
PSA* 7.9 (0, 149) 3.2 (0, 8.7) 4.0 (0, 12) 16.4 (0,149) 
Gleason     
5-6 31% (22) 100% (17) 10% (3) 8% (2) 
41% (29) 0% (0) 90% (26) 12% (3) 
8-10 28% (20) 0% (0) 0% (0) 80% (20) 
Clinical stage     
T1 51% (36) 53% (9) 55% (16) 44% (11) 
T2a 27% (19) 41% (7) 31% (9) 12% (3) 
T2b 10% (7) 6% (1) 14% (4) 8% (2) 
T2c 1% (1) 0% (0) 0% (0) 4% (1) 
T3a 8% (6) 0% (0) 0% (0) 24% (6) 
TX 3% (2) 0% (0) 0% (0) 8% (2) 
Hormones     
At enrollment 20% (14) 0% (0) 21% (6) 32% (8) 
Previously 7% (5) 12% (2) 7% (2) 4% (1) 
Overall (n = 71)Low (n = 17)Intermediate (n = 29)High (n = 25)
PSA* 7.9 (0, 149) 3.2 (0, 8.7) 4.0 (0, 12) 16.4 (0,149) 
Gleason     
5-6 31% (22) 100% (17) 10% (3) 8% (2) 
41% (29) 0% (0) 90% (26) 12% (3) 
8-10 28% (20) 0% (0) 0% (0) 80% (20) 
Clinical stage     
T1 51% (36) 53% (9) 55% (16) 44% (11) 
T2a 27% (19) 41% (7) 31% (9) 12% (3) 
T2b 10% (7) 6% (1) 14% (4) 8% (2) 
T2c 1% (1) 0% (0) 0% (0) 4% (1) 
T3a 8% (6) 0% (0) 0% (0) 24% (6) 
TX 3% (2) 0% (0) 0% (0) 8% (2) 
Hormones     
At enrollment 20% (14) 0% (0) 21% (6) 32% (8) 
Previously 7% (5) 12% (2) 7% (2) 4% (1) 
*

PSA is mean of each group, and () are ranges. () for other parameters are frequencies.

Electrochemiluminescence immunoassays. Met in urine specimens was measured, in triplicate, using electrochemiluminescent immunoassay (Meso Scale Discovery), after adjusting to pH 6.5 to 7.5 according to the protocol described by Athauda et al. (9). Assays were done blinded to study end point.

Statistics. All analyses were done using SPLUS version 7.0 (Insightful corporation). Boxplots were used to descriptively compare the distribution of met values across low-risk, intermediate-risk, high-risk, normal, and metastatic patients. A Kruskal-Wallace nonparametric ANOVA test was used to compare the distributions of the localized disease group. A Wilcoxon rank-sum test was used to compare values of normal individuals to the localized disease group. We estimated the receiver operating characteristic curve (ROC) for distinguishing between the localized and metastatic groups using an empirical ROC curve. Area under the ROC curve (AUC) was estimated using a trapezoidal rule. A 95% confidence interval for the estimated AUC was obtained with a bootstrap using the percentile method with 5,000 bootstrap samples.

PPis. Myriad National Cancer Institute ProNet7

Yeast Two-hybrid technology was used to identify PPis with met. Interactions with met were tested using nine baits, which covered the length of the met protein, and probed into three activation domain libraries (see Supplementary Table S1). Additional PPis were identified through Unified Human Interactome (UniHI),8 an integrated resource from 10 major interaction maps derived by computational and experimental methods. Ingenuity Pathways Analysis (IPA) was used as a third database of interactions. Once met interacting proteins were identified, a systematic PubMed search was completed to distinguish which proteins have been implicated in metastatic PCa.

REMARK criteria. We used the REMARK (13) criteria as a guideline in conducting and reporting this study. This includes marker stated (met), study objectives, prespecified hypotheses, patient characteristics (Table 1), biological material used (urine), methods of preservation and storage, assay method used, quality control, blinded to study end point, stated retrospective study with prospective collection, time period of collection, specified statistical methods, and flow of patient study. Sample size for this hypothesis was based on using all available samples. The relation to standard prognostic variables, results interpreted in the context of prespecified hypotheses, limitations of study and implications for future research are described in the “Discussion” section. This study does not specifically discuss all patient treatments received, as all patient specimens from those with localized disease were obtained pretreatment. This study does not report univariate or multivariate analyses in relation to outcome as the study was not designed to assess patient outcome, and only 3 of 71 patients with localized disease failed within the study period.

Recent in vitro studies have shown an inverse correlation between met protein levels and androgen receptor expression (8). To expand these initial findings and to better understand the relationship between PSA and met in the progression of PCa, we measured the level of met and PSA in a panel of PCa cell lines from both local and metastatic sites (14). The cell lines derived from a lymph node (LNCaP) or a primary prostate specimen (22Rv1) had the highest PSA levels (68 and 9.8 ng/mL, respectively). PC3 and DU145, derived from metastatic bone and brain deposits, respectively, showed nondetectable values for PSA. Conversely, PC3 and DU145 exhibited the greatest level of total met (3.4 and 3.76 ng/mL, respectively), whereas 22Rv1 and LNCaP produced less total met (2.1 and 1.17 ng/mL, respectively). Thus, as shown in Fig. 1, the level of total met and PSA are inversely correlated. Cell lines derived from distant metastatic sites expressed the highest levels of met.

Fig. 1.

PSA and total met protein levels are inversely correlated in PCa cell lines. LNCaP, 22Rv1, PC-3, and DU145 PCa cell lines were grown to 60% confluency. Cells were given fresh medium and were harvested 12 h later. Protein (40 μg) was quantified by ELISA. Protein concentration (ng/mL) for each cell line is represented at % expression compared with the cell line with the highest concentration (LNCaP for PSA, DU145 for met).

Fig. 1.

PSA and total met protein levels are inversely correlated in PCa cell lines. LNCaP, 22Rv1, PC-3, and DU145 PCa cell lines were grown to 60% confluency. Cells were given fresh medium and were harvested 12 h later. Protein (40 μg) was quantified by ELISA. Protein concentration (ng/mL) for each cell line is represented at % expression compared with the cell line with the highest concentration (LNCaP for PSA, DU145 for met).

Close modal

We have previously shown that met could be detected in the urine of mice bearing human xenograft tumors (9). To determine if soluble met could be measured in the urine of patients with PCa, urine was analyzed for met from patients with localized PCa, divided into their clinical risk categories; low, intermediate, or high, metastatic PCa, and a group of normal volunteers. As shown in Fig. 2, met could be detected in the urine of patients with both localized and metastatic PCa as well as the urine from men without cancer. There was no significant difference in met levels among any subgroups of patients with localized disease (P = 0.44, Kruskal-Wallace nonparametric ANOVA test comparing the three groups of localized disease). Furthermore, there was no significant difference between individuals without PCa and those with localized disease (P = 0.17, Wilcoxon rank-sum test). However, there was a highly significant difference between the normal group and the metastatic group (P < 0.0001) and between the localized disease group and the metastatic group (P < 0.0001). Further correlation of met and PSA in patients with early metastatic disease may be important in determining if met is an accurate marker of PCa progression.

Fig. 2.

Urinary met levels in normal volunteers and patients with PCa. Urinary met values were measured by electrochemiluminescent immunoassay. Samples were run in triplicate and normalized with urinary creatinine levels. Low (T1-T2a, Gleason of 2-6, or PSA of <10 ng/mL), intermediate (T2b-T2c, Gleason of 7, or PSA of 10-20 ng/mL), and high (T3a, Gleason of 8-10, or PSA of >20 ng/mL) are risk classification groups as defined by National Comprehensive Cancer Network guidelines. For each patient group, the line in the middle of the box represents the median. The lower and the upper edges of the box are the 1st and 3rd quartile, respectively. The fences are drawn to the nearest value not exceeding 1.5 (interquartile range), and observations beyond the fences are denoted as lines and are considered outliers. P value of <0.0001 between metastatic and localized groups and metastatic and normal groups.

Fig. 2.

Urinary met levels in normal volunteers and patients with PCa. Urinary met values were measured by electrochemiluminescent immunoassay. Samples were run in triplicate and normalized with urinary creatinine levels. Low (T1-T2a, Gleason of 2-6, or PSA of <10 ng/mL), intermediate (T2b-T2c, Gleason of 7, or PSA of 10-20 ng/mL), and high (T3a, Gleason of 8-10, or PSA of >20 ng/mL) are risk classification groups as defined by National Comprehensive Cancer Network guidelines. For each patient group, the line in the middle of the box represents the median. The lower and the upper edges of the box are the 1st and 3rd quartile, respectively. The fences are drawn to the nearest value not exceeding 1.5 (interquartile range), and observations beyond the fences are denoted as lines and are considered outliers. P value of <0.0001 between metastatic and localized groups and metastatic and normal groups.

Close modal

To further assess the performance of urinary met to distinguish between men with localized disease and men with metastatic disease, an empirical ROC curve was computed (Fig. 3). The AUC was 0.90 with a 95% confidence interval ranging from 0.84 to 0.95. This interval was evaluated using a bootstrap using the percentile method with 5,000 bootstrap samples.

Fig. 3.

ROC analysis for localized versus metastatic PCa patients. AUC = 0.90; 95% confidence interval (CI), 0.84-0.95.

Fig. 3.

ROC analysis for localized versus metastatic PCa patients. AUC = 0.90; 95% confidence interval (CI), 0.84-0.95.

Close modal

The data presented above suggest that met may provide a biomarker indicative of metastatic PCa. As an initial step toward understanding the biology underlying this putative relationship between met and the metastatic progression of PCa, PPi involving met were identified using a laboratory-based approach and two bioinformatics tools. First, an experimental method using a yeast two-hybrid system was used to probe for proteins that directly bind to met. Three human tumor libraries were probed using overlapping met fragments as baits. Six direct met PPi, as well as nine secondarily derived interactions were identified (Table 2). In addition, the UniHI database, which contains data from multiple PPi studies as well as literature-derived, computer-curated interactions, was queried. The UniHI database contained 43 met interacting proteins, of which 8 overlapped with the Myriad data. Finally, IPA was used to define proteins interacting with met. IPA contains interaction data that is literature-derived but human-curated; thus, it can differ from UniHI data. The IPA study revealed 58 proteins that directly interact with met. Of the 58 IPA-derived proteins, 8 overlapped with those identified through the yeast two-hybrid system, 19 overlapped with those from the UniHI database, and 8 were identified in all three methods. Thus, application of these three methods led to the identification of a total of 89 unique proteins that can interact with met.

Table 2.

met interacting proteins as identified through 3 independent bioinformatic databases

Myriad ProNet
UniHI
IPA
ExperimentalExperimental/computationalLiterature
GRAP BAG1 BAG1 
GRAP2 CASP3 CBL 
GRB2 CBL CBLB 
PIK3R3 CNR1 CD44 
PLCG1 CTNNB1 CD82 
ZYX CTTN CRKL 
Literature DAPK3 CTNNB1 
CBL DNAJA3 CTTN 
GLMN EGFR DSG1 
GAB1 FAS EGFR 
HGF FGF7 EHF 
mABL1 GAB1 ELF3 
mGRB10 GLMN F2 
mPIK3p55 GRB2 FAS 
mPIK3R1 HGF FOS 
RANBP9 HGFAC GAB1 
 HGS GAB2 
 IL32 GLMN 
 INPP5D GRB2 
 INPPL1 Heparan sulfate 
 ITGB4 HGF 
 MST1 JUN 
 MUC20 JUP 
 PCBD2 MBP 
 PIK3R1 MITF 
 PLCG1 MST1R 
 PLXNB1 MYH4 
 PTPN11 NRP1 
 PTPRB PIK3R1 
 PTPRJ PIK3C2B 
 RANBP10 PIK3R2 
 RANBP9 PLAU 
 SH3KBP1 PLAUR 
 SHC1 PLC GAMMA 
 SMC1L1 PLCG1 
 SNAPAP PLG 
 SNX2 PLXNB1 
 SPSB1 PLXNB2 
 SRC PLXNB3 
 STAT3 PP2A 
 USP4 PTPN11 
 VAV1 PTPRF 
 YWHAZ RANBP9 
  RAS 
  RASA1 
  SHC1 
  SP1 
  SP3 
  SPSB1 
  SPSB2 
  SPSB3 
  SPSB4 
  SRC 
  STAT3 
  TGFα 
  TP53 
  VIL2 
  WT1 
Myriad ProNet
UniHI
IPA
ExperimentalExperimental/computationalLiterature
GRAP BAG1 BAG1 
GRAP2 CASP3 CBL 
GRB2 CBL CBLB 
PIK3R3 CNR1 CD44 
PLCG1 CTNNB1 CD82 
ZYX CTTN CRKL 
Literature DAPK3 CTNNB1 
CBL DNAJA3 CTTN 
GLMN EGFR DSG1 
GAB1 FAS EGFR 
HGF FGF7 EHF 
mABL1 GAB1 ELF3 
mGRB10 GLMN F2 
mPIK3p55 GRB2 FAS 
mPIK3R1 HGF FOS 
RANBP9 HGFAC GAB1 
 HGS GAB2 
 IL32 GLMN 
 INPP5D GRB2 
 INPPL1 Heparan sulfate 
 ITGB4 HGF 
 MST1 JUN 
 MUC20 JUP 
 PCBD2 MBP 
 PIK3R1 MITF 
 PLCG1 MST1R 
 PLXNB1 MYH4 
 PTPN11 NRP1 
 PTPRB PIK3R1 
 PTPRJ PIK3C2B 
 RANBP10 PIK3R2 
 RANBP9 PLAU 
 SH3KBP1 PLAUR 
 SHC1 PLC GAMMA 
 SMC1L1 PLCG1 
 SNAPAP PLG 
 SNX2 PLXNB1 
 SPSB1 PLXNB2 
 SRC PLXNB3 
 STAT3 PP2A 
 USP4 PTPN11 
 VAV1 PTPRF 
 YWHAZ RANBP9 
  RAS 
  RASA1 
  SHC1 
  SP1 
  SP3 
  SPSB1 
  SPSB2 
  SPSB3 
  SPSB4 
  SRC 
  STAT3 
  TGFα 
  TP53 
  VIL2 
  WT1 

NOTE: Bolded proteins are those found in all three databases (Myriad, UniHI, and IPA). Italicized proteins are those found in both UniHI and IPA.

As the three methods/databases are constructed with different types of data and curated using various methods, we used the entire set of 89 proteins in our literature search through the PubMed database. We entered, “protein name or synonym and prostate, PCa and/or metastatic PCa” in the search field. This search indicated that 40 of the 89 proteins had been associated with metastatic PCa as listed in Supplementary Table S1. Of these 40 proteins, 30 have been shown to be up-regulated and 10 proteins have been shown to be down-regulated in metastatic PCa. Four proteins (CBL, GRB2, HGF, and PLCG1) were common to all three databases and 10 proteins were common to two databases [BAG1, CTNNB1, CTTN, epidermal growth factor receptor, FAS, PLXNB1, PTPN11, SHC1, SRC, and signal transducers and activators of transcription 3 (STAT3)]. Of the 40 proteins, 19 were previously derived from the laboratory studies and 21 of the proteins were reported to interact with met strictly from a literature curation (Fig. 4). These data suggest that met may be a central regulator of multiple pathways involved in the progression of PCa from a local to a metastatic disease.

Fig. 4.

PPis with met were identified using Myriad National Cancer Institute ProNet, UniHI, and IPA. Of the 89 interacting proteins, the 40 illustrated here were found to be related to metastatic PCa. Inner circle, those experimentally/computationally found to be involved in metastatic PCa; outer circle, literature derived. Red and tan, up-regulated; green and blue, down-regulated. Bold text, those identified experimentally using Myriad.

Fig. 4.

PPis with met were identified using Myriad National Cancer Institute ProNet, UniHI, and IPA. Of the 89 interacting proteins, the 40 illustrated here were found to be related to metastatic PCa. Inner circle, those experimentally/computationally found to be involved in metastatic PCa; outer circle, literature derived. Red and tan, up-regulated; green and blue, down-regulated. Bold text, those identified experimentally using Myriad.

Close modal

PSA testing in patients with PCa has been used clinically for almost 20 years. Although originally approved for use in monitoring disease progression after therapy, PSA monitoring is now widely used in screening for PCa. Interestingly, PSA is neither a very sensitive nor specific marker of disease progression (15). Thus, much work is currently under way to evaluate alternative markers (16). One such protein is the met receptor, the expression of which has been correlated with metastatic PCa. In contrast to other oncogenic proteins associated with prostate tumor progression, met ectodomain is shed from tumor cells and ultimately appears in the urine (9), thus providing the potential for a noninvasive assessment. The studies described here evaluate the urinary met levels as a biomarker for metastatic PCa.

As shown in Fig. 2, met is measurable in the urine of patients with PCa. Importantly, patients with metastatic PCa had higher levels of met than those with localized disease. However, met was also detectable in the urine of normal volunteers at levels not significantly different from patients with localized PCa. The small sample size for each subgroup may have obscured a small elevation in patients with localized PCa, a hypothesis which could be tested in a larger prospective clinical trial. Another possible explanation is biologically based. In the normal prostate, HGF is produced by stromal cells and met is present at high levels in basal and intermediate cells, together acting in a paracrine fashion (17). Normal luminal cells express met only 2.1% of the time; however, expression increases to 44.1% with either proliferation or inflammation, commonly seen in patients with benign prostatic hypertrophy. Thus, a number of physiologic conditions can potentially influence urinary met levels. However, in advanced PCa, the paracrine HGF/met activity is thought to switch to an autocrine pathway; thus, as HGF becomes overexpressed in androgen-independent cancer cells, it may lead to hyper-stimulating met leading to met overexpression (18). One possible consequence of such met overexpression would be increased shedding and elevated levels selectively in the urine of patients with metastatic PCa.

The mechanism accounting for the increased met levels in the urine of patients with advanced PCa remains speculation. To gain a better understanding of the biology underlying the elevated met expression in metastatic PCa, we used both experimental and bioinformatic approaches to develop a met interactome (Fig. 4). Met interacts with proteins that mediate numerous processes characteristic of a metastatic phenotype including inhibition of apoptosis, loss of cell adhesion, and enhancement of cell motility and migration (19). HGF, the high-affinity ligand of met, binds to and activates met triggering a broad spectrum of biological responses. Several proteins we identified are involved in the stabilization/binding of HGF, with subsequent activation of met, such as CD44 variant 3, PLAU and heparin sulfate (20). Other molecules that we identified are involved in the downstream signaling after HGF binding including, GRB2, PI3K, PLCγ, RAS, SHC1, SRC, and Stat3 (21, 22). SHC1 binds directly to the phosphorylated receptor tail of met and, in combination with GRB2, links met activation and the Ras–mitogen-activated protein kinase pathway, previously shown to be important for cellular proliferation and transformation (23). PI3K has been shown to be important for cell motility, PLCγ for branching morphogenesis, and STAT3 for epithelial tubulogenesis (24). In an Oncomine analysis of microarray data from patients with localized PCa versus those with metastatic cancer, both GRB2 and STAT3 were over expressed in the metastatic cancers (P = 3.8 × 10−19 and 1.9 × 10−5, respectively).9

Furthermore, HGF (also known as Scatter factor) is known to induce migration of cells, while inhibiting intercellular adhesion (25). As shown in Supplementary Table S2, CD82, DSG1, INPPL1, JUP, PTPRF, VIL2, and zyxin (ZYX) are related to cell adhesion and are all down-regulated. Activation of met is known to lead to a loss of cell-to-cell contact. For example, intact CD82 has been shown to suppress integrin-induced invasion of PCa cells in vitro and CD82 loss may promote metastasis (26). Likewise, ZYX is known to bind myopodin leading to slower migration of PCa cells and a reduced invasiveness; thus, a loss of ZYX might enhance migration and invasion (27). Thus, it follows that either an increase in total HGF, an increase in HGF binding, or an overexpression of met can lead to an increased signaling and in turn to a more invasive and metastatic phenotype.

A number of studies have used immunohistochemical analyses of tissue specimens to associate the overexpression of met with metastatic PCa. The data reported here are the first to show a significant increase in met levels in urine in patients with metastatic PCa. The use of urine as a source for biomarker analysis has the benefits of its ease of acquisition, noninvasiveness, and simplicity of storage and processing. As noted above, there was no apparent difference between cancer-free males and those with localized PCa with respect to urinary met levels, indicating that met is unlikely to be applicable in initial diagnosis. However, the results presented here suggest that urinary met may serve as a marker for metastatic PCa and may be particularly useful as a marker of recurrence after therapy, in situations where PSA fails to identify such patients. Future investigation should compare ROC curves for both met and PSA, and based on this preliminary data, we would expect the AUC to be significantly smaller for PSA. Furthermore, studies should also include comparison of met with bone scans and other imaging modalities to determine if met can accurately identify patients with metastatic disease before having clinical or radiological evidence of such disease.

No potential conflicts of interest were disclosed.

Grant support: Intramural Research Program of the NIH, National Cancer Institute and by the Howard Hughes Medical Institute through the Howard Hughes Medical Institute-NIH Research Scholars Program.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1
Jemal A, Siegel R, Ward E, Murray T, Xu J, Thun MJ. Cancer statistics, 2007.
CA Cancer J Clin
2007
;
57
:
43
–66.
2
Bensalah K, Lotan Y, Karam JA, Shariat SF. New circulating biomarkers for prostate cancer. Prostate Cancer Prostatic Dis 2007.
3
Leibovici D, Spiess PE, Agarwal PK, et al. Prostate cancer progression in the presence of undetectable or low serum prostate-specific antigen level.
Cancer
2007
;
109
:
198
–204.
4
Fall K, Garmo H, Andren O, et al. Prostate-specific antigen levels as a predictor of lethal prostate cancer.
J Natl Cancer Inst
2007
;
99
:
526
–32.
5
Sandblom G, Ladjevardi S, Garmo H, Varenhorst E. The impact of prostate-specific antigen level at diagnosis on the relative survival of 28,531 men with localized carcinoma of the prostate.
Cancer
2008
;
112
:
813
–9.
6
Aihara M, Lebovitz RM, Wheeler TM, Kinner BM, Ohori M, Scardino PT. Prostate specific antigen and gleason grade: an immunohistochemical study of prostate cancer.
J Urol
1994
;
151
:
1558
–64.
7
Ma PC, Maulik G, Christensen J, Salgia R. c-Met: structure, functions and potential for therapeutic inhibition.
Cancer Metastasis Rev
2003
;
22
:
309
–25.
8
Verras M, Lee J, Xue H, Li TH, Wang Y, Sun Z. The androgen receptor negatively regulates the expression of c-Met: implications for a novel mechanism of prostate cancer progression.
Cancer Res
2007
;
67
:
967
–75.
9
Athauda G, Giubellino A, Coleman JA, et al. c-Met ectodomain shedding rate correlates with malignant potential.
Clin Cancer Res
2006
;
12
:
4154
–62.
10
Pisters LL, Troncoso P, Zhau HE, Li W, von Eschenbach AC, Chung LW. c-met proto-oncogene expression in benign and malignant human prostate tissues.
J Urol
1995
;
154
:
293
–8.
11
Knudsen BS, Gmyrek GA, Inra J, et al. High expression of the Met receptor in prostate cancer metastasis to bone.
Urology
2002
;
60
:
1113
–7.
12
Naughton M, Picus J, Zhu X, Catalona WJ, Vollmer RT, Humphrey PA. Scatter factor-hepatocyte growth factor elevation in the serum of patients with prostate cancer.
J Urol
2001
;
165
:
1325
–8.
13
McShane LM, Altman DG, Sauerbrei W, Taube SE, Gion M, Clark GM. Reporting recommendations for tumor marker prognostic studies (REMARK).
J Natl Cancer Inst
2005
;
97
:
1180
–4.
14
Sobel RE, Sadar MD. Cell lines used in prostate cancer research: a compendium of old and new lines-part 1.
J Urol
2005
;
173
:
342
–59.
15
Catalona WJ, Richie JP, Ahmann FR, et al. Comparison of digital rectal examination and serum prostate specific antigen in the early detection of prostate cancer: results of a multicenter clinical trial of 6,630 men.
J Urol
1994
;
151
:
1283
–90.
16
Laxman B, Morris DS, Yu J, et al. A first-generation multiplex biomarker analysis of urine for the early detection of prostate cancer.
Cancer Res
2008
;
68
:
645
–9.
17
Nakashiro K, Hara S, Shinohara Y, et al. Phenotypic switch from paracrine to autocrine role of hepatocyte growth factor in an androgen-independent human prostatic carcinoma cell line, CWR22R.
Am J Pathol
2004
;
165
:
533
–40.
18
Nakashiro K, Hayashi Y, Oyasu R. Immunohistochemical expression of hepatocyte growth factor and c-Met/HGF receptor in benign and malignant human prostate tissue.
Oncol Rep
2003
;
10
:
1149
–53.
19
Birchmeier C, Birchmeier W, Gherardi E, Vande Woude GF. Met, metastasis, motility and more.
Nat Rev
2003
;
4
:
915
–25.
20
van der Voort R, Taher TE, Wielenga VJ, et al. Heparan sulfate-modified CD44 promotes hepatocyte growth factor/scatter factor-induced signal transduction through the receptor tyrosine kinase c-Met.
J Biol Chem
1999
;
274
:
6499
–506.
21
Liu X, Yao W, Newton RC, Scherle PA. Targeting the c-MET signaling pathway for cancer therapy.
Expert Opin Invest Drugs
2008
;
17
:
997
–1011.
22
Dietrich S, Abou-Rebyeh F, Brohmann H, et al. The role of SF/HGF and c-Met in the development of skeletal muscle.
Development
1999
;
126
:
1621
–9.
23
Ponzetto C, Zhen Z, Audero E, et al. Specific uncoupling of GRB2 from the Met receptor. Differential effects on transformation and motility.
J Biol Chem
1996
;
271
:
14119
–23.
24
Boccaccio C, Ando M, Tamagnone L, et al. Induction of epithelial tubules by growth factor HGF depends on the STAT pathway.
Nature
1998
;
391
:
285
–8.
25
Sato N, Sadar MD, Bruchovsky N, et al. Androgenic induction of prostate-specific antigen gene is repressed by protein-protein interaction between the androgen receptor and AP-1/c-Jun in the human prostate cancer cell line LNCaP.
J Biol Chem
1997
;
272
:
17485
–94.
26
Sridhar SC, Miranti CK. Tetraspanin KAI1/CD82 suppresses invasion by inhibiting integrin-dependent crosstalk with c-Met receptor and Src kinases.
Oncogene
2006
;
25
:
2367
–78.
27
Yu YP, Luo JH. Myopodin-mediated suppression of prostate cancer cell migration involves interaction with zyxin.
Cancer Res
2006
;
66
:
7414
–9.