Loss of aquaporin 3 protein expression constitutes an independent prognostic factor for progression-free survival: an immunohistochemical study on stage pT1 urothelial bladder cancer

Being a matter of debate for more than 100 years, the molecular basis of water transport across epithelial surfaces was first described in red blood cells in the late 1980s by Peter Agre and associates [1]. Later on, water-transporting channels were also shown to be present in renal epithelial cells and subsequently termed aquaporins (AQP) [2‐4]. AQPs are a family of transmembrane proteins that selectively allow water or water plus other small, uncharged molecules such as urea and glycerol to pass along hydrostatic and osmotic gradients. Aquaporins are ubiquitously expressed in bacterial, animal and human cells. Hence, they are essential for cellular function [5]. To date, 13 different mammalian AQPs have been identified at the molecular level and localised to particular tissues [6]. Analysis of several human diseases has confirmed that AQPs are involved in various pathological conditions and provide promising drug targets [7, 8]. Moreover, there is strong presumptive evidence that AQPs play a role in carcinogenesis, specifically in tumour angiogenesis and cell migration [9]. The pro-tumourigenic effect of a lost AQP expression in neoplastic cells has been the subject of previous studies. Knockdown of AQP3 for instance has been shown to be associated with increased migration and proliferation of gastric cancer cells [10].

There is only very limited data on expression and biological significance of aquaporins in human urothelium. Rubenwolf et al. were the first to characterize human urothelium with regard to all 13 members of the human AQP family. While transcripts for AQP3, AQP4, AQP7, AQP9 and AQP11 were detected in freshly-isolated urothelium and cultured urothelial cells by reverse transcriptase-polymerase chain reaction (RT-PCR), AQP3 was unequivocally expressed also at the protein level, with intense immunohistochemical labelling of the cell membranes of basal and intermediate layers in normal bladder urothelium [11].

To date, investigations into the potential significance of aquaporins in urothelial bladder carcinoma (UBC) are lacking [12]. In a preliminary analysis of AQP3 expression in UBC of various stages, our group demonstrated loss of AQP3 in muscle-invasive disease whereas stage Ta specimens were shown to invariably express the marker. Interestingly, 60% of pT1 tumours were found to be AQP3 positive while the remaining specimens revealed complete absence of AQP3. This finding suggests that AQP3 could be of value as a prognostic marker, particularly in the highly heterogeneous subgroup of pT1 patients [13]. The objective of the present study was to analyze the prognostic value of AQP3 protein expression in stage pT1 UBC patients. Despite numerous previous attempts to identify clinical and histopathological prognostic parameters, Shahin’s rule of each 30% of patients developing either never recurrence, recurring or even dying of the disease is still valid [14].

Stage pT1 urothelial bladder carcinoma continues to be a challenging tumour entity for urologists due to unpredictable clinical courses. While about one third of patients never experience tumour recurrence, at least one third requires radical cystectomy for progression to muscle-invasive disease and the remaining 30% of patients ultimately die of UBC [14].

Numerous research groups worldwide have scrutinized a wide array of parameters to identify prognostic factors that may help to predict tumour progression in stage pT1 UBC [18‐24]. Among these parameters grading appears to constitute the most powerful prognostic histopathological feature. In their analysis of a large collective of 310 stage pT1 UBC patients, Otto et al. only recently were able to show that tumour grading, specifically the distinction between grade 2 and 3 tumours as classified by the WHO classification of 1973, was the most consistent and reliable predictor in respect to prognosis [16]. Moreover, a series of immunohistochemical investigations into the prognostic value of various proteins expressed by UBC was performed, most of which provided disappointing findings. In a cohort of 175 patients, 19% of whom diagnosed with pT1G3 UBC, Rodriguez Alonso and associates identified over-expression of tumour suppressor protein p53 by the tumour as prognostic factor. Overexpression of p53 and ki67 alongside CIS, multifocality and solid bladder tumours were associated with a worse progression-free survival in multivariate analysis [25]. Lopez-Beltran et al. were able to demonstrate a prognostic value of different biomarkers in relation to patient survival. In a total of 51 patients consisting of pT1G3 tumours, expression of p53 alongside other cell cycle regulators such as cyclin D1 and cyclin D3 showed prognostic significance for PFS in multivariate analysis [26]. By contrast, Park et al. were unable to find any prognostic value for seven markers including p53 and ki67 in 61 patients with pT1G3 UBC [27].

Our previous finding of AQP3 transcript and protein expression in normal human urothelium and cultured urothelial cells prompted us to investigate expression and potential significance in diseased urothelium including urothelial cancer [11, 13]. In a preliminary analysis of AQP3 expression in tumour specimens of various stages of UBC, we demonstrated loss of AQP3 protein expression in muscle-invasive disease whereas pTa specimens were shown to invariably express the marker. It was of note that 40% of stage pT1 UBC tumours exhibited loss of AQP3 protein expression [13]. Hence, we concluded that AQP3 might constitute a prognostic marker for progression to muscle-invasive disease.

Our present data indicate a statistically significantly worse PFS in patients in whom immunohistochemistry revealed loss of AQP3 expression. This finding was confirmed by multivariate Cox regression analysis. By contrast, recurrent disease was independent of the AQP3 expression status. Taking both our present and previous findings into account, we hypothesize that expression of AQP3 in stage pT1 UBC is pathogenetically associated with non-muscle-invasive disease while loss of AQP3 may be part of a molecular program associated with progression to muscle-invasive UBC [13]. Moreover, abnormalities of chromosome 9p, where the AQP3 gene is located, are frequently present in UBC, adding further presumptive evidence for a role of AQP3. However, such hypotheses have to be addressed in further studies aimed at elucidating the biological significance of AQP3 expression in UBC.

The pro-tumourigenic effect of a loss of AQP3 has been investigated in previous studies on other tumour entities. Knockdown of AQP3 expression for instance has been shown to be associated with increased migration and proliferation of gastric cancer cell lines [12]. In contrast, Kuseyama et al. found a high level of expression of AQP3 in oral and esophageal squamous cell carcinomas (SCC) [28]. Xu and associates investigated the effect of AQP3 on matrix metalloproteinase in human gastric carcinoma cells and concluded that AQP3 may be a promising drug target [29]. In the field of urological neoplasms, Ismail et al. provided evidence that prostate cancer cell lines in vitro become more sensitive to cryotherapy after inhibiting AQP3 by mercuric chloride and AQP3 siRNA [30].

Although the present study did not investigate the underlying molecular mechanisms of AQP expression in urothelial carcinoma, our findings suggest that AQP3 may be an independent predictor of tumour progression from stage pT1 towards stage pT2 tumours. This finding may be of clinical value in that patients exhibiting loss of expression of AQP3 in their TURB specimens could benefit from aggressive surgical treatment in the form of early cystectomy. However, we do not attempt to over-interprete our findings. The retrospective design of the study and the distinction between AQP3-positive versus AQP3-negative tumours may be oversimplistic. A detailed analysis of the various expression patterns of AQP3 in tumours by scoring both the labeling intensity and the proportion of positively-labeled tumour areas were not performed due to the low number of AQP3-positive tumours (n=51). The lack of such subanalyses is undoubtedly a limitation of the present study. Future work analyzing an appropriate number of AQP3-positive tumours is required to address this important aspect. Moreover, investigations into the molecular mechanisms of AQP3 expression in urothelial carcinoma are required to understand its role in the pathogenesis of UBC.

The present study is the first to investigate a larger patient cohort of pT1 urothelial bladder carcinomas for AQP3 protein expression. Lack of AQP3 protein expression in pT1 tumours was shown to be associated with progression towards muscle-invasive disease. By contrast, tumour recurrence was independent of AQP3 expression. Importantly, this association was independent of other clinicopathological parameters. We conclude that AQP3 expression could be used to help stratify the behavior of patients with pT1 UBC.

Our findings underline the importance of investigating expression and function of AQPs beyond water and solute transport in normal and neoplastic urothelium. Further studies into the molecular mechanisms of AQP3 expression as well as prospective multicentre studies are to be awaited before valid conclusions can be drawn from our present findings.