Prognostic impact of urokinase-type plasminogen activator system components in clear cell renal cell carcinoma patients without distant metastasis

Renal cell carcinoma (RCC), comprising 3-4% of all malignant neoplasms, is the third most common urological tumour entity and the 13th most common malignancy worldwide [1, 2]. Clear cell RCC (ccRCC) is the most common and the most aggressive subtype of this disease [3]. About one third of the patients already present distant metastases at the time of diagnosis, whereas nearly another third develops metastases during the course of the disease [1, 4]. Although the application of target-directed therapeutics improved the outcome of patients with metastatic ccRCC their prognosis remains unfavourable [5].

Since the disease progression of RCC is still not sufficiently predictable by clinical and histopathological parameters, there is an urgent need for additional powerful biological prognostic factors that may help to refine individual risk stratification of RCC patients. A reliable prediction of outcome after nephrectomy by such markers would be valuable to tailor individualised follow-up schedules and to provide adjuvant therapies to patients at high risk of relapse [6]. Several clinicopathological parameters and molecular biomarkers or their combination have been described to be prognostically useful for RCC [7‐9]. Nevertheless, further research and validation of these molecular markers for diagnostic and prognostic purposes in RCC patients are still necessary.

The urokinase-type plasminogen activator (uPA) system consisting mainly of uPA, the uPA receptor (uPAR) and the plasminogen activator inhibitor-1 (PAI-1) has been shown to play a key role in physiological and pathological pathways including carcinogenesis [10‐12]. In addition to extracellular proteolysis, uPA in concert with uPAR and/or PAI-1 induces cell signaling pathways that affect multiple steps of cancer progression such as angiogenesis, cell growth, cell adhesion and migration, chemotaxis and cell survival [10, 11, 13, 14].

High expression levels of uPA and/or PAI-1 in tumour tissue extracts have been found to be strong predictors of poor prognosis in patients afflicted with different types of solid malignant tumours [15, 16]. High uPAR levels are also associated with poor prognosis in a variety of cancer types, however, the prognostic impact of uPAR expression is not as pronounced as that of uPA and PAI-1 [17, 18].

To date, only few studies exist that characterise the role of uPA system members in RCC and analyse their association with clinicopathological parameters and their impact on the outcome of RCC patients. High uPA, uPAR and/or PAI-1 protein expression in tumour tissue of RCC patients, detected by immunohistochemistry, was significantly associated with higher tumour stage, metastasis and poor disease-specific survival in a cohort of 106 RCC patients [19]. Furthermore, significant associations were observed between high immunoexpression of PAI-1 in tumour tissue of ccRCC patients and older age, advanced tumour stage, high nuclear grade, and disease progression [20, 21]. In survival analyses, strong PAI-1 immunostaining was associated with a shorter disease-free survival of ccRCC patients [20]. Furthermore, high PAI-1 immunoexpression was identified as an independent predictor of cancer-specific survival in a cohort of 172 ccRCC patients [21].

Increased uPA, uPAR and/or PAI-1 antigen levels in tumour tissue extracts, as determined by ELISA, were found to be related to high tumour grade of RCC [22, 23]. Hofmann et al. [24, 25], who also measured antigen levels of uPA, uPAR and PAI-1 in renal tissue extracts from RCC patients by ELISA, revealed all three factors as strong and independent predictors for early relapse.

Thus, the above-mentioned studies point to a rather strong clinical relevance of the members of the uPA-system in RCC, however, with some important limitations. Most of these data were obtained using tissue samples originating from different RCC subtypes, inconsistent cohort sizes, optimised cut-off points, the use of non-detergent tissue extracts or the three factors were not analysed in parallel. Therefore, in the present study, uPA, uPAR and PAI-1 antigen levels were determined in detergent tissue extracts derived from corresponding malignant and non-malignant renal tissues of 112 RCC patients of the clear-cell subtype without clinically detectable distant metastases at the time of diagnosis by ELISA. The antigen concentrations of uPA system components in tissue extracts were analysed for correlations between these factors themselves and for potential associations with clinicopathological parameters. In addition, we assessed the prognostic power of these markers for the prediction of disease-specific survival (DSS) and overall survival (OS) to validate their usefulness for outcome prediction of ccRCC patients without distant metastasis.

In the present study, we aimed at analysing the protein levels of uPA system components in tissue extracts from matched malignant and corresponding non-malignant tissue kidney specimens in a cohort of 112 ccRCC patients without distant metastasis and at assessing their potential associations with clinicopathological parameters and the prognostic relevance in ccRCC patients.

The antigen levels of uPA, uPAR and PAI-1 (ng analyte/mg of total protein) determined in detergent extracts of corresponding malignant and non-malignant kidney tissues in our study were comparable to those reported in previous studies applying the same commercially available ELISA kits (IMUBIND from Sekisui/American Diagnostica) or using in-house ELISA formats (see Table 6). Conversely, in cytosolic tumour tissue extracts, considerably lower uPA concentrations were measured, whereas the PAI-1 antigen levels were comparable [22].

The concentration of uPA system components has been reported to differ between malignant and non-malignant renal tissues (see Table 6). All mentioned studies including our present analyses demonstrated significantly higher protein levels of PAI-1 in RCC tissues compared to normal renal tissues (see Table 6). With regard to uPA and uPAR, the observations in the different studies vary. While Swiercz et al. [23] observed significantly elevated uPA levels in the tumour specimens, others found no significant differences between malignant and non-malignant renal tissues [22, 24, 25, 32]. In contrast, in our study uPA protein levels were significantly decreased in ccRCC tissues compared to the corresponding non-malignant specimens. The reasons for these discrepancies between the different reports remain unclear so far. Hofmann et al. [24, 25], who used the same strategy and analysed a similar number of patients as we did, measured comparable uPA levels in the corresponding renal tissue specimens, whereas Span et al. [32] obtained a similar trend of uPA levels in malignant and non-malignant renal tissues as in our study, which did not reach statistical significance (P = 0.081).

For uPAR, we found no differences between corresponding malignant and non-malignant tissue of ccRCC patients. In contrast, significantly higher uPAR antigen concentrations have been observed in RCC tissues compared to benign renal tissue in previous reports [23, 24]. These discrepancies might originate from differences in the numbers of analysed tissue samples. While 112 pairs of corresponding malignant and non-malignant renal tissues were used in our study, Hofmann et al. [24] assessed uPAR levels only in 49 matched tissue pairs and Swiercz et al. [23] compared 52 RCC tissues with 28 normal tissues of different origin (Table 6). Furthermore, levels of uPA system components might differ between the RCC subtypes. We analysed only patients with ccRCC, but no information on the subtype is given in the mentioned reports [23, 24]. Interestingly, Chautard et al. [22] observed clear differences in uPA and PAI-1 levels between tissue specimens from ccRCC (n = 85), papillary RCC (n = 11) and sarcomatoid RCC (n = 4) indicating that the RCC subtype is of importance and the use of mixed patient cohorts hampers the drawing of definite conclusions (Table 6).

In the tumour tissues, we found significant, moderately high correlations between the antigen levels of all three components of the uPA system (r_s values between 0.51 and 0.65). Similar positive correlations have in fact been previously reported between uPA, uPAR, and PAI-1 in tumour tissue extracts of RCC patients [24, 25]. Strikingly, significant correlations between antigen levels of these factors in tumour tissues have been also observed in various other types of cancer [33‐35]. The regulated expression of uPA system components in a concerted manner in tumour growth and metastasis may be attributed to the various interactions within the uPA system [11, 16]. In contrast, we found only low correlations between uPA system components in non-malignant renal tissues, which is in line with observations on tumour-free samples of other tumour entities [32, 36]. This may be caused by various mechanisms of expression regulation and functional interplay between these factors under physiological conditions [11].

Only few studies have analysed the protein levels of uPA system components in RCC specimens with regard to their dependence on clinicopathological parameters and to their impact on prognosis of RCC patients. In the present study, we observed significantly elevated antigen levels of all three uPA system components in tissue specimens from non-organ confined tumours compared to organ-confined tumours. Furthermore, significantly increased levels of uPAR and PAI-1 could be detected in high grade tumours in comparison to low grade tumours. In accordance to our results, Swiercz et al. [23] reported elevated protein levels of uPA, uPAR and PAI-1 in tissue extracts from high grade RCC. Chautard et al. [22] also detected a gradual increase of uPA and PAI-1 protein levels in RCC extracts with higher tumour grade and stage, whereas only the relation between PAI-1 and grade reached statistical significance. A positive relation between the immunoreactive scores of uPA, uPAR or PAI-1 and tumour grade, tumour stage or metastasis was described by Ohba et al. [19] who performed immunohistochemical analyses on 106 RCC patients. Two other immunohistochemical studies on ccRCC patients also described significant positive associations of PAI-1 staining with tumour grade as well as with tumour stage [20, 21]. On the contrary, the studies by Hofmann et al. [24, 25] did not reveal any association between protein levels of uPA, uPAR or PAI-1 in tissue extracts and tumour grade or stage.

Moreover, data on the influence of the protein expression of uPA system members on survival of RCC patients are partially contradictory depending on the detection method used [19, 22, 24]. In the present study, all three uPA system components were significant predictors for DSS in univariate Cox’s regression analyses and high levels of uPA and uPAR remained independent prognostic factors for DSS in multivariate analyses. The combination of both factors further improved the prediction of DSS in multivariate analysis considerably (Table 4; HR=14.5, 95% CI 1.88-111.1, P=0.010). These results are in line with the data of the immunohistochemical analyses reported by Ohba et al. [19] who revealed that increased protein levels of all three factors were associated with poor DSS. Nevertheless, only PAI-1 emerged as independent predictive factor in their multivariate Cox’s regression analyses [19].

In the report of Hofmann et al. [24] who considered – in contrast to our study – the time from surgery to tumour progression, all three factors appeared also as strong predictors for a higher risk of relapse. In contrast, uPA and PAI-1 protein levels in cytosolic tissue extracts were not independently predictive for metastasis as reported by Chautard et al. [22] but elevated uPA levels were significantly associated with a shorter disease-free survival (DFS) in univariate Cox’s regression analyses. Similarly, Choi et al. [20] described a significant relation between high PAI-1 immunoexpression and a shorter DFS, albeit this could not be confirmed in multivariate Cox’s regression analyses. In comparison to this, the immunohistochemical study by Zubac et al. [21] revealed PAI-1 as a significant prognostic factor for disease-specific survival both in univariate and multivariate Cox’s regression analyses. Taken together, previous studies demonstrated a prognostic potential of uPA and uPAR, and most obviously, for PAI-1 protein levels in RCC tissues. In contrast, high levels of uPA and uPAR in ccRCC extracts were shown to be associated with a significantly shorter DSS of ccRCC patients in our study. The differing study results may originate from variances in the applied methods and used cut-off points as well as in the patient cohorts with regard to the inclusion of patients with diverse RCC subtypes, metastatic status and follow-up endpoints.

Finally, we evaluated the predictive value of the three uPA system components for patients with an organ-confined tumour who are usually expected to have a favourable outcome but still can progress after tumour nephrectomy given the highly aggressive nature of ccRCC. Surprisingly, the DSS of those patients could be further stratified on the basis of combined tissue protein levels of uPA and uPAR (Additional file 1: Table S2, Table 5). The HR of 9.83 (P = 0.032) obtained for patients with high levels of uPA and/or uPAR in multivariate analysis in the pT1+2 subgroup further emphasises the potential prognostic utility of uPA and uPAR protein expression for patients with ccRCC. Based on these findings, it might be also of interest to specifically determine the concentration of the uPA/uPAR complex with hybrid sandwich-type ELISAs and its relation to prognosis of the patients [37].

Moreover, Zubac et al. [21] could show that PAI-1 immunoexpression turned out to be a highly significant predictor for metastatic relapse in patients with organ-confined RCC. These results suggests that the determination of protein levels of uPA system components in patients with pT1 and pT2 tumours may provide additional prognostic information that may allow for individual, risk-adapted therapy decisions. Patients at higher risk of relapse could be treated by targeted agents in an adjuvant setting as currently evaluated in a number of ongoing prospective clinical trials [6]. For primary breast cancer, the determination of uPA and PAI-1 antigen levels in tumour tissue extracts has already entered clinical practice for risk stratification and individual therapy decisions in patients with lymph node-negative disease [16, 38]. This approach might be also imaginable for ccRCC patients without distant metastasis, particularly for those with organ-confined tumours, but needs further elucidation in prospective clinical trials.

The present study, to our knowledge, is the first report on the parallel assessment of protein levels of uPA, uPAR and PAI-1 in detergent tissue extracts of renal tissue specimens in a consistent cohort of 112 patients with ccRCC and without clinically detectable distant metastasis at the time of diagnosis. Comparable previous studies were performed using tissue samples originating from different RCC subtypes and follow-up endpoints, inconsistent cohort sizes, optimised cut-off points or non-detergent tissue extracts or analysed not all three factors in parallel (see Table 6). Our study tried to overcome these shortcomings, and revealed a considerable, independent prognostic potential of uPA and uPAR protein levels in tumour tissue extracts for DSS. Best prediction of DSS might be achieved by the combination of elevated protein levels of uPA and/or uPAR compared to low levels of both factors. Particularly, this holds true for the subgroup of ccRCC patients with organ-confined tumours, whose outcome in general is assumed to be favourable but can be further stratified on the basis of the protein levels of uPA and uPAR. Despite the strengths of our retrospective study on 112 ccRCC patients, future prospective validation studies have to elucidate independently the prognostic value of the uPA system components for ccRCC patients.