The value of MMPs and TIMPs expression in bladder tissue
Despite the clinical significance on the pathogenetic impact of MMPs in human bladder cancer only a limited number of studies are available in the literature. MMP2, MMP9, and TIMP2 are the most frequently investigated metalloproteinases and inhibitors in human bladder tissue [4, 32–34]. It has been described in the literature that elevated expressions of MMP2 and MMP9 in bladder cancer tissue at the mRNA and protein level are associated with advanced tumour stage, grade, and a decreased survival rate [32–34]. Kanayama et al.  found an elevated mRNA expression of MMP2 and MMP9 in tissue of muscular invasive bladder tumours compared to non-invasive tumours. The authors concluded that MMP2 and TIMP2 contribute to bladder tumour invasiveness and therefore might be useful prognostic markers in the future. Overexpressions of MMPs are associated with various pathological events . Gakiopoulou et al.  found that TIMP2 is involved in regulation of apoptosis and is associated with an adverse prognosis in patients with TCC of the bladder. One group investigated MMP1 and MMP3 protein expression by immunohistochemistry . They showed a correlation for MMP1 expression and tumour aggressiveness, but no detectable expression for MMP3. Vasala et al.  demonstrated an overexpression of MMP2 immunohistochemically. MMP2 overexpression correlated with bladder cancer stage, but not with grade. The 5-year survival rate for patients with bladder cancer was significantly lower in the MMP2 positive group.
In the most recent study on MMPs and their inhibitors urothelial carcinoma specimens were profiled for 24 MMPs and the four endogenous tissue inhibitors and their receptors using quantitative real time RT-PCR . This study showed that MMP2, MT1-MMP, and MMP28 were very highly expressed in bladder tumour samples and MMP1, 7, 9, 11, 15, 19, and 23 were also highly expressed. TIMP1 and TIMP3 correlated with an increasing tumour grade. By laser capture microdissection of RNA it was possible to locate the MMP expression within the tissue. The study revealed that the most highly expressed MMPs are located in the stroma, except MMP13, which was located in the epithelium. Wallard et al.  confirmed in their study the impact of MMP1, MMP2, MMP9, and TIMP1 as potential diagnostic and therapeutic targets and supported the impact of those MMPs and TIMP1 as clinical markers for bladder cancer.
In neoplastic diseases an imbalance of MMPs and TIMPs, leading to an excess of degradative activity, is supposed to be linked to the invasive character of tumour cells [12, 13]. Their proteolytic activity is activated by a complex cascade, which is not yet completely understood. Activated MMP1, MMP3 and latent forms of MMP2 and MMP9 are regulated and inhibited by endogenous proteins known as tissue inhibitors of metalloproteinase TIMP1 and TIMP2 . Studies have shown that TIMP1 binds preferably to MMP9 and TIMP2 preferably to MMP2 . Kugler et al.  stated for renal cell carcinoma (RCC) that the balance of MMP2- and MMP9- to TIMP1- and TIMP2-mRNA expression is a prognostic factor of tumour aggressiveness. Increased expression of TIMP1 in RCC correlates with poor prognostic variable including shortened patient survival. The paradoxical poor prognostic implication of TIMP overexpression documented in the literature complements the dual function of TIMPs and warrants further investigation .
Diagnostic values of MMPs, TIMPs and their combination tested in blood plasma
There are only limited data available on circulating MMPs and TIMPs in patients with bladder cancer. Gohji et al.  analyzed MMP2 and MMP3 concentrations in blood serum of patients with bladder cancer. They found elevated concentrations of MMP2 and MMP3 in patients with advanced TCC of the bladder (pT2-T4, N+, M+) in comparison to the serum of patients with superficial tumors (Ta-T1, N0, M0). Their studies revealed further elevated MMP2, MMP3, and TIMP2 concentrations from patients after tumour resection, which might function as a predictive value for early detection of tumour recurrence [20, 40]. Further, Guan et al.  correlated MMP9 expression in the serum to tumour grade and stage as a clinical prognostic factor. However, it has been shown that all MMP measurements in serum give equivocal values .
To our knowledge, there is no publication so far available on detection of MMPs and TIMPs in blood plasma for bladder cancer detection. On the basis of the concept that MMPs are synthesized in tissue and released into the blood stream we evaluated the levels of MMPs, their inhibitors, and MTC1 in blood plasma from patients with TCC of the bladder at different grades, stages, and with and without metastasis. Although citrate was recently suggested to be the most suitable anticoagulant to prepare blood samples for MMP measurements , comparable MMP activities in heparin and citrate plasma samples were found . The plasma concentration of MMP2 was significantly higher in comparison to the control group whereas the concentrations of MMP1, TIMP1, TIMP2, and MTC1 were significantly lower from patients with non-metastasized TCC of the bladder in comparison to the healthy control group (Table I). In comparison to the expression data in tissue described above, these results confirm that circulating MMPs and TIMPs do not always reflect the direct tissue situation. That phenomenon, well known in clinical enzymology as enzyme distortion, has also to be considered for MMPs and TIMPs . Since the mechanisms of release into extracellular space, the distribution in the intravascular compartment, and the elimination from that compartment can differently affect these analytes. Both changes of quantitative and qualitative relations between them are possible. For the complex interconnections between the various MMP components and pathways in tissue, Overall and Kleifeld  recently introduced the term "protease web" and underlined that the interactions between MMPs, TIMPs, and related components are more important than the single components. We believe that this complexity may essentially determine the distortion between tissue expression and blood pattern of MMPs and TIMPs. That aspect makes it also understandable that the levels of MMPs and TIMPs in plasma have a limited value to attribute these values to their expression in tissue and vice versa. Another important aspect is the fact that we only measured non-active MMPs. However, it is well known that, for example, the ratios of active to total MMP2 and MMP9, respectively are changed in tumour tissue and also in plasma [45, 46].
Plasma concentrations from patients with metastasized tumours showed statistically significant higher values for TIMP1 compared to samples from non-metastasized tumours. That phenomenon could be a result of the dual function of the TIMP1 as already discussed in the previous chapter. Due to the possible pre-analytical interferences mentioned above it is not possible to compare our data measured in plasma to the data from investigations in serum. In this study the diagnostic performance criteria sensitivity, specificity, and ROC data revealed the best values for MMP2 as a separate tumour marker for TCC. Comparing our data on different MMPs MMP2 was proved to be the best marker alone. It was statistical significantly elevated in superficially, invasive bladder tumours and in metastasized bladder tumours in comparison to the healthy control group (p < 0.001) (Table 1). MMP2 was also shown to stand alone as clinical tissue marker in other studies as described above.
Using the statistical mROC program we have calculated the different sensitivities and specificities for all MMPs and TIMPs at their different cutoff points (Figures 1, 2 and Table 4) and additionally determined the best combination of two and three markers (Table 4, Figure 3). The best single indicators with the highest sensitivity and specificity (MMP2, TIMP1) did not necessarily reached the highest sensitivities and specificities in combination with the next best indicators. However, MMP2 does not reach a satisfying sensitivity when tested alone. At the cutoff point it reaches only 75% sensitivity and specificity than when tested in combination with MMP9 and TIMP1 (97% sensitivity, 94% specificity). Our results show that different indicators in specific combinations could result in an improved sensitivity and specificity in comparison to the best single indicator test by itself, which has never been shown so far in the literature (Table 3, 4, Figure 2). We presented in this study that single markers have only a limited diagnostic value. Separate analyzes of MMP1, MMP2, MMP3, MMP9, TIMP1, TIMP2, and MTC1 in plasma do not allow a prognostic prediction for tumour grading, staging, or metastasis, but certain combinations could be very helpful. Although the limitations of our study should be considered that the results are based on a small number of patients and patients of benign urological diseases (inflammations) were not included. We suggest specific MMPs and their combinations as potential helpful diagnostic indicators in early and advanced bladder cancer. To some degree there might exist an overfitting in the resulting data of the two virtual mROC classifiers because MMP9 and TIMP1 in the two best combinations are the markers with the highest correlation to each other.
There are controversial opinions in the literature. However, we used a similar approach as Piironen et al. , who included covariates of a correlation coefficient <0.7 into a common multivariate model. Further prospective studies should prove the usefulness of these marker combinations. However, so far it is not economic efficient and not practical to combine those three markers for routine tumour screening. Further investigations are still necessary for easier analyzes of MMPs and TIMPs and a more economical application to the clinical routine.