Association of MMP-2, RB and PAI-1 with decreased recurrence-free survival and overall survival in bladder cancer patients

Background We previously reported an accurate urine-based bladder cancer (BCa)-associated diagnostic signature that can be used to non-invasively detect BCa. In this study, we investigated whether a component of this signature could risk stratify patients with BCa. Methods Utilizing immunohistochemistry, we investigated angiogenin, MMP-2, p53, RB and PAI-1 expression from 939 patients with BCa. The expression levels were scored by assigning a proportion score and an intensity score to yield a total staining score for each protein. The expressions of each protein individually and as an aggregate were then correlated with progression-free survival (PFS), cancer-specific survival (CSS) and overall survival (OS). Results Differential expressions of these markers were noted in BCa. With multivariate analysis in non-muscle invasive bladder cancer (NMIBC) age, tumor grade portended a worse PFS, while age, tumor grade, nodal status, MMP2, RB and PAI-1 expression portended a worse OS. As for multivariate analysis in muscle invasive bladder cancer (MIBC), age MMP-2 and RB were associated with a worse PFS, while age, nodal status, MMP-2, RB and PAI-1 were associated with a worse OS. Using Kaplan-Meier survival analysis, we noted a significant reduction in OS as more of the five biomarkers were expressed in a tumor. Thus, overall, high expressions of MMP-2, RB and/or PAI-1 in bladder tumors were markers of poor prognosis. Conclusion Individually, MMP-2, RB and PAI-1, as well as in aggregate correlated with poor survival in patients with BCa. Thus, patients whose bladder tumors express these biomarkers may benefit from early radical treatment and/or neoadjuvant or adjuvant therapies.


INTRODUCTION
Cancer of the urinary bladder is the fourth most common malignancy in men and the ninth most common malignancy in women in the United States [1]. Urothelial carcinomas constitute approximately 90% of all bladder cancer (BCa) cases [2]. At presentation, more than 80% of bladder tumors are non-muscle invasive bladder cancer (NMIBC, i.e., Tis, Ta or T1) and the remaining 20% of bladder tumors are muscle-invasive bladder cancers (MIBC) or metastatic. NMIBC harbors a 5-year survival rate of approximately 94% [3,4]; however, approximately 70% of patients with these lesions develop tumor recurrence within two years of initial diagnosis. The recurrence phenomenon of NMIBC makes it one of the most prevalent cancers worldwide; in America, it is second only to colorectal cancer. Therefore, NMIBC is a significant burden to healthcare systems [1,5]. As for MIBC, 5-year survival rate is approximately 50%, which drops off to < 20% in the case of metastatic disease [6,7].
Methods to identify NMIBC and MIBC patients, who are at increased risk of disease recurrence and succumbing to disease, have previously relied on conventional histopathologic evaluation. These histopathologic features have failed to properly risk stratify these patients. Thus, prognostic markers are urgently needed to identify high-risk patients who would benefit from early radical treatment and/or neoadjuvant or adjuvant therapies.
In the present study, we set out to validate the tissue expression of ANG, MMP-2, p53, PAI-1 and RB in human BCa with immunohistochemistry conducted on tissue microarrays (TMA). Demographic, clinical, disease and treatment characteristics are presented in Table 1. The aims of the validation were to a) analyze the tissue immunoreactivity of these BCa candidate biomarkers and b) investigate if the expression of these BCa candidate biomarkers possesses prognostic value in determining recurrence-free survival and overall survival in BCa patients.

RESULTS
P53, RB and PAI-1 are preferentially associated with aggressive bladder cancer Figure 1 shows representative immunohistochemical stainings for each of the five targets in a high-grade non-muscle invasive tumor. The relationship between immunophenotype for each target and tumor grade is summarized in Table 2. We found the highest expression levels of p53, RB and PAI-1 (total staining score ≥3) associated with high-grade disease compared to lowgrade disease (58% vs. 27%, 85% vs. 65%, 65% vs. 49%, respectively). Furthermore, the highest expression of MMP-2 was noted with low-grade disease compared to high-grade disease (64% vs. 52%). These differences were statistically significant. The relationships between immunophenotype for each target and tumor stage are summarized in Table 3. We found the highest expression levels of p53 and RB in T2 stage tumors (compared to Ta/Tis stage tumors (62% vs. 29% and 87% vs. 68%, respectively). Similar to its expression in low-grade tumors, MMP-2 expression was higher in Ta/Tis stage tumors compared to T2 stage tumors (62% vs. 47%). These differences were statistically significant. However, we observed no difference in p53, RB and MMP-2 expression between T2 and T3/T4 stage tumors.
MMP-2, p53, RB, PAI-1 are associated with a reduction in recurrence-free survival and overall survival in bladder cancer In (Table 4b).
Using the Kaplan-Meier survival analysis with the logrank test, we found significantly reduced progression-free survival in NMIBC but not in MIBC when more of the five biomarkers were expressed in a tumor (p = 0.01 and p = 0.16, respectively), Figure 2a and 2b. As noted in Table  4, three of the five biomarkers (MMP-2, RB and PAI-1) were critical biomarkers influencing survival analysis of NMIBC and MIBC. We noted no difference in cancerspecific survival in NMIBC or MIBC associated with the expression of these five biomarkers (Supplementary  OS. Using the Kaplan-Meier survival analysis with the log-rank test, we noted a significant reduction in OS in both NMIBC and MIBC as more of the five biomarkers were expressed in a tumor (p = 0.001 and p = 0.03, respectively), Figure 3a and 3b. Thus, the overall, high expression of MMP-2, RB and/or PAI-1 in bladder tumors was a marker of poor prognosis.

DISCUSSION
In this retrospective analysis, we demonstrated that the expression of MMP-2, RB and PAI-1 correlated with a reduction in recurrence-free survival and overall survival in patients with BCa. Our results also demonstrated the prognostic value of combining ANG, MMP-2, p53, RB and PAI-1 expression for all patients with BCa. Thus, patients whose bladder tumors express these biomarkers may benefit from early radical treatment and/or neoadjuvant or adjuvant therapies. To the best of our knowledge, this represents the largest and most comprehensive study to date demonstrating the prognostic significance of MMP-2, RB and PAI-1 in patients with BCa.  The biomarkers that comprised this prognostic signature (i.e., ANG, MMP-2, p53, RB and PAI-1) have a varied range of ascribed functions including angiogenesis, breakdown of extracellular matrix, serine protein inhibitor, DNA binding and transcription factor. ANG and PAI-1 [22][23][24][25][26] have also been associated with angiogenesis. Angiogenesis, the development of new blood vessels from existing blood vessels, is essential for normal growth and the development of tissues and organs. Furthermore, in addition to the degradation extracellular matrix by MMP-2, recent studies have suggested that ANG and PAI-1 can breakdown the extracellular matrix [27,28]. Degradation of the extracellular matrix allows cells to become more motile. Thus, these extracellular matrix-degrading proteins may work in conjunction with an increase in vasculature, thereby, increasing the probability that motile-invasive tumor cells may enter the circulation to disseminate to distant organs [29]. P53 has both transactivation and transrepression activity and, thus, controls the transcription of numerous genes [30][31][32]. RB also controls the expression of numerous genes although it does so primarily by recruiting transcription factors and chromatin remodelling proteins [33][34][35]. Furthermore, we have identified that ANG regulates MMP-2 [17], as well as PAI-1 and p53 expression, [36] while PAI-1 activity is controlled by p53 and RB [unpublished data]. Previous in vitro and in vivo analyses of ANG and PAI-1 began to unravel their potential contributions to BCa pathobiology. Briefly, there is clear evidence that ANG expression positively regulates MMP-2 expression via ERK1/2 signaling [17]. In addition, ANG regulates MMP-2 expression through two major pathways, a) p53 regulation via ERK pathway activation and b) induction of hypomethylation by DNMT3b down-regulation through the ERK pathway [36]. MMPs are known to affect extracellular matrix (ECM) remodeling, angiogenesis, apoptosis, epithelial-to-mesenchymal transition and cell proliferation and, thus, induce aggressive invasion and metastasis of different cancer types, including BCa [37]. Assessing the GEO database, we found that MIBC specimens had increased levels of ANG and MMP-2 compared to NMIBC. These findings suggest ANG may   influence MMP-2 gene, which may be a mechanism to promote BCa growth and angiogenesis. Interestingly in the current study, ANG expression was not associated with tumor grade or tumor stage. Perhaps the inability to probe for ANG was due to the condition of the paraffin embedded tissue (i.e., older sections may have some degradation of proteins). Thus, more research is needed to confirm ANG's presence in human BCa. On the other hand, MMP-2 expression was greater in low-grade and low-stage BCa and was associated with a reduction in recurrence-free survival as well as reduction in overall survival. PAI-1 is the primary inhibitor of tissue-type plasminogen activator (tPA) and urokinase-type plasminogen activator (uPA), and it acts to suppress tissue and plasma fibrinolysis via plasmin conversion [38].   We found that the down-regulation of PAI-1 led to a) an inhibition in cell proliferation and b) a potent arrest in the G 1 to S phase of the cell cycle, a phenomena associated with reduction in cyclin D3/cdk4/6 and cyclin E/cdk2 and an increase in cell cycle inhibitors, p53, p21 Cip1/Waf1 , and p27 kip1 [18]. Little attention has been given to PAI-1 in the human urinary bladder. For example, only two groups have reported PAI-1 levels for BCa patients. Urquidi et al. noted a significant increase in urothelial cell PAI-1 levels in patients bearing bladder tumors compared to non-tumor bearing patients [39]. Becker et al. reported significantly higher PAI-1 levels in tissue and plasma samples, but not in urine, from patients with BCa compared with controls [40]. Although the function of PAI-1 is complex, further investigation is warranted in the hope that it can provide insight into specific aspects of tumor biology and the identification of tumor cell vulnerabilities for therapeutic exploitation. Interestingly in the current study, higher PAI-1 expression was associated with high-grade compared to low-grade, while higher expression of p53 and RB were associated with highgrade and high stage disease. High PAI-1, p53 and RB were associated with a reduction in survival. Perhaps, the ability to demonstrate the difference in PAI-1 expression in this cohort compared to our previous cohort is related to the increase in the sample sizes of the current study. The high expression levels of p53 and RB have been well documented in BCa [41,42]. In fact several studies have reported on the prognostic significance of p53 expression in bladder tumors. For example, Cordon-Cardo and others, noted an association between the p53-positive phenotype and disease progression (p < 0.001), as well as reduced survival (p < 0.001) in a small cohort comprised of 29 NMIBC [43]. Similarly in MIBC, Shariat and others, reported that alteration in p53 (or p21, pRB, p16) was independently associated with disease progression (p ≤ 0.038) and disease-specific survival (p < 0.039) in a small cohort of 127 MIBC treated with cystectomy [44]. Differences in outcomes associated with these studies and the current study may be related to our large, diverse cohort of 939 subjects. In the only phase III randomized study assessing the prognostic significance of p53 in BCa, Stadler and others, could not validate the prognostic significance of p53 in 521 patients undergoing adjuvant chemotherapy [45]. However, the current report begins to elucidate a specific pathway involving PAI-1, p53 and RB in human BCa. On the other hand, MMP-2 expression was greater in low-grade and low-stage BCa and was associated with a reduction in recurrence-free survival, as well as a reduction in overall survival. Our findings in this report also have therapeutic implications regarding the potential utility of targeting these proteins directly. We and others have previously reported that blocking ANG, MMP-2, p53, RB or PAI-1 in bladder cell lines can significantly reduce cell proliferation/survival [17,18], migration and invasion [18,45,46] and angiogenesis [17,[47][48][49]. Our current results, when combined with these previous observations, further strengthen the rationale of targeting MMP-2, RB and PAI-1 directly for treatment of BCa.
In summary, we have demonstrated in this study that MMP-2, RB and PAI-1 expression is correlated with poor prognosis of patients with BCa. Our findings support the hypothesis that targeting these three molecules may improve the survival of patients with BCa.

Patients and clinicopathologic information
The study was performed after approval by the Western Institutional Review Board under a request of waiver of consent on archived pathologic specimens. Tissue microarrays associated with two independent patient cohorts were used in the current study. Cohort #1 was comprised of 587 patients diagnosed with BCa from 1979-2007 at Aarhus University Hospital (Aarhus, Denmark) [21], including 2 sub-cohorts; one with 223 primary Ta/T1 tumors from patients treated by transurethral resection of the bladder and one with 364 tumors from radical cystectomy patients. Cohort #2 was comprised of 352 patients diagnosed with BCa from 1984-2005 at Uppsala University Hospital (Uppsala, Sweden) [50]. Thus, samples from 939 patients were available for analysis. Demographic, clinical, disease and treatment characteristics are presented in Table 1. The median follow-up for the entire cohort was 5.37 years.
Follow-up data also included progression-free survival (PFS), cancer-specific survival (CSS) and overall survival (OS). The endpoints of recurrence-free survival and OS were calculated from the date of surgery to the date of the recurrence or last follow-up.
The expression levels of ANG, MMP-2, p53, RB and PAI-1 were scored by assigning a proportion score and an intensity score [17,18,51,52]. The estimated proportion (0 = 0% of cells; 1 = 1% to 40%; 2 = 41% to 75% and 3 = 76% to 100%) and the average intensity (0 = none; 1 = weak; 2 = intermediate and 3 = strong) of immunoreactive tumor cells were assessed. The proportion and intensity scores were combined to obtain a total staining score for each protein, which ranged from 0 to 6. Thus, the protein expression levels were determined based on the total ANG, MMP-2, p53, RB and PAI-1 staining score as follows: none = 0, low = 1 or 2, moderate = 3 or 4, high = 5 or 6. Two investigators (OTMC, YS), who were blinded to the clinicopathologic data and clinical outcomes, scored each core. When there was a discrepancy in the scoring, a third investigator reviewed consensus scoring was obtained.
Human lung (SERPINE1, MMP-2), liver (ANG) and tonsil (p53, Rb) were used as positive controls, and omitting the primary antibody served as the negative controls. SAS 9.4 (Cary, NC) was used to perform the statistical analyses. The relationship between immunoexpression of the five targets and clinicopathological features were assessed with the Fisher's exact test (crosstabs), log-rank tests (progressionfree survival, cancer-specific survival and overall survival), and multivariate Cox proportional hazards regression (used to adjust for demographic and clinical parameters). All tests were 2-tailed.

Author contributions
Owen T.M. Chan: performed IHC, assisted in interpreting IHC and drafted manuscript.
Hideki Furuya: participated in study design and helped to draft the manuscript.
Ian Pagano: organized and analyzed data. Yoshiko Shimizu: assisted with IHC, organized data. Kanani Hokutan: assisted with IHC, organized data. Lars Dyrskjøt: participated in study design, provided TMA and helped to draft the manuscript.
Jørgen Bjerggaard Jensen: participated in study design and helped to draft the manuscript.
Per-Uno Malmstrom: participated in study design, provided TMA and helped to draft the manuscript.
Ulrika Segersten: participated in study design and helped to draft the manuscript.
Filip Janku: participated in study design and helped to draft the manuscript.
Charles J. Rosser: conceived the study, and participated in its design and coordination and helped to draft the manuscript.