Expression parameters of the inhibitors of apoptosis cIAP1 and cIAP2 in renal cell carcinomas and their prognostic relevance

Abstract

The expression of the inhibitor of apoptosis (IAP) family members cIAP1 and cIAP2 have been shown to be altered in various cancer entities. This study was done to characterize the tumor-related expression profile of cIAP1 and cIAP2 in patients with renal cell carcinomas (RCC) and to evaluate its potential predictive value after curative resection. Expression of cIAP1 and cIAP2 was analyzed by real-time RT-PCR in RCC and corresponding normal tissue samples obtained from a cohort of 127 RCC patients (median follow-up: 48 months) undergoing surgical treatment. Expression data was correlated to histopathological variables and outcome. Overexpression of cIAP1 and cIAP2 occurred in most RCC specimens (p < 0.001), but 20% of the patients had lower cIAP levels in malignant than in normal tissue. The cIAP1 expression correlated with the tumor stage, levels being higher in pT1 tumors than in advanced pathological stages (p = 0.002). Decreased cIAP1 expression in RCC relative to paired normal samples predicted an abbreviated time to recurrence (hazard rate 2.96; 95% CI: 1.23–7.09) and tumor-specific survival (hazard rate 2.78; 95% CI: 1.22–6.38) irrespective of the tumor stage and grade. The prognostic effect of cIAP1 was most pronounced in patients with pT3 disease (log rank test p = 0.001). The results of uni- and multivariate analysis suggest a prognostic value of cIAP1 expression for RCC patients, downregulation indicating an aggressive, potentially lethal phenotype. © 2006 Wiley-Liss, Inc.

Renal cell carcinoma (RCC) is an entity characterized by extraordinary resistance to conventional chemotherapy and ionizing radiation, which results in a poor outcome in cases with metastatic disease.1 The ability of cells to evade apoptosis plays a crucial role during carcinogenesis as well as tumor progression and contributes to treatment resistance.2, 3 The inhibitor of apoptosis protein family (IAP) protects against various endogenous and exogenous apoptotic triggers.4 Eight human IAPs have thus far been identified and grouped into three classes based on the presence or absence of a RING domain and the homology of their BIR (baculovirus IAP repeat) domains: c-IAP1, c-IAP2, XIAP, livin, ILP2 (class 1); NAIP (class 2); survivin and BRUCE (class 3).5 One important apoptotic mechanism of IAPs is the inhibition of caspases as key effector molecules of the apoptotic process.6, 7 However, IAPs have a functional complexity that goes far beyond merely inhibiting caspases: they have also been reported to affect cell division, cell cycle progression, signal transduction pathways and protein degradation.5, 8, 9, 10

The IAP-mediated inhibition of apoptosis can be countered by IAP antagonistic proteins like the critical death inducer Smac/Diablo.11, 12 Downregulation of Smac/Diablo was recently shown to predict a worse prognosis in RCC.13 Other studies found that increased XIAP levels upset the delicate balance of XIAP and Smac/Diablo and were associated with advanced tumor stages and tumor dedifferentiation in RCC.14, 15 cIAP1 and cIAP2 are very similar to XIAP, containing three BIR domains and a RING finger, and were identified by their interaction with the TNF receptor-associated factor II (TRAF2).16, 17 Overexpression of cIAP1 and cIAP2 has been reported to suppress chemotherapy-induced apoptosis in vitro (reviewed in Refs.4 and10). These IAPs also interact with Smac/Diablo but are less potent caspase inhibitors than XIAP.11, 12, 18, 19, 20, 21 However, only cIAP1 and cIAP2, but not XIAP, have been found to promote the proteasome-dependent degradation of Smac/Diablo through E3 ubiquitin ligase activity of their RING domains.19 Various cell line experiments have revealed that cIAP1 and its cleavage products, generated during the apoptotic cascade, also have apoptosis-enhancing potency, but it is not yet known whether these proapoptotic effects are relevant in vivo.22, 23

The expression of cIAP1 and cIAP2 in RCC has not been examined up to now. Their structural relationship to XIAP and their ability to interact with Smac/Diablo led us to consider whether cIAP1 and cIAP2 expression might relate to tumor features of RCC and to the surgical treatment outcome. In this study, we determined the quantitative mRNA expression profile of cIAP1 and cIAP2 in RCC and corresponding normal renal cortex tissue, and correlated the gene expression data with the clinicopathological tumor features and outcome.

Material and methods

Patients and specimens

Surgical specimens were prospectively collected from 127 consecutive patients undergoing surgical treatment for primary RCC at the Department of Urology, Charité–Universitätsmedizin Berlin, Campus Benjamin Franklin, between November 1995 and November 2005. Their use was approved by the Ethics Committee of the Free University of Berlin, and all patients gave their informed consent prior to surgery. Tumor tissue samples and paired specimens of normal renal cortex tissue were obtained during surgery and snap-frozen immediately in liquid nitrogen. Only patients with histologically confirmed RCC were included. Histopathologic classification of renal cell carcinoma was based on the World Health Organization and 1997 TNM classification guidelines (International Union Against Cancer, 1997). Samples from 15 study participants were not eligible for RT-PCR analysis. Follow-up information was not available for 8 of the remaining 112 patients, leaving a final data set of 104 patients (Table I).

Table I. Clinicopathological Characteristics of the Study Population

	Patients
Eligible patients n (%)	104 (100)
Age (years)
Median (range)	62.5 (36–87)
Sex n (%)
Female	37 (35.6)
Male	67 (64.4)
Histology n (%)
Clear cell	94 (90.4)
Chromophile	4 (3.8)
Chomophobe	3 (2.9)
Sarcomatoid	3 (2.9)
Stage n (%)
T1	45 (43.2)
T2	27 (26.0)
T3	32 (30.8)
Grade n (%)
G1	2 (1.9)
G2	62 (59.6)
G3	40 (38.5)
Follow-up (months)
Median (range)	48 (2–122)
Recurrence n (%)	20 (19.2)
Tumor-related death n (%)	22 (21.2)

RNA extraction and real-time PCR

Total RNA was extracted from tissue specimens using the RNeasy® Mini Kit with DNAse treatment (Qiagen GmbH, Hilden, Germany) according to the manufacturer's instructions. Its concentration and integrity were determined using the NanoDrop ND-1000 (PEQLAB, Biotechnology GmbH, Erlangen, Germany) and the Agilent Bioanalyzer 2100 (Agilent Technologies, Waldbronn, Germany).

Quantitative real-time RT-PCR for cIAP1 (Genbank accession number: NM_001166) and cIAP2 (Genbank accession number: NM_001165) was performed on the LightCycler® system using the LightCycler RNA Amplification Kit Hybridization Probe (Roche Diagnostics GmbH, Penzberg, Germany). RT-PCR conditions and primer sequences are given in Table II. Analysis of the reference gene porphobilinogen deaminase (PBDG) was concomitantly performed for each sample using the LightCyler h-PBDG Housekeeping Gene Set (Roche Diagnostics, Mannheim, Germany). Standard curves were established for each of the target genes. RT-PCR efficiencies of target and housekeeping genes were calculated from the standard curves as follows: cIAP1: 1.872; cIAP2: 2.041; PBGD: 1.955. The relative gene expression, i.e. the ratio of the respective target gene copy number to the PBDG copy number, was calculated for each sample, thus normalizing for sample-to-sample differences in RNA input, quality and RT efficiency.24 The detection limit determined from serial dilutions of target and housekeeping gene templates was 10 copy numbers in 5 μl of PCR template. This limit provided the basis for calculating the cIAP tumor-to-normal (T/N) expression ratio in cases where the copy number of the respective target gene was below the detection limit despite sufficient housekeeping gene expression. All amplifications were repeated twice.

Table II. RT-PCR Characteristics for CIAP1 and CIAP2

Gene	Sequences of primers and probes	PCR conditions	Product size
cIAP1	F 5′-GCTCAGTAACTGGGAACCAAA-3′	0.5 s at 95°C, 15 s at 55°C, 20 s at 72°C	229 bp
	R 5′-ATCATTGCGACCCACATAATA-3′
	FL: 5′-GCTCGAATGAGAACATTTATGTACTGGCC-FL
	LC: 5′-TCTAGTGTTCCAGTTCAGCCTGAGCAGC-PH
cIAP2	F 5′-GATGTTTCAGATCTACCAGTG-3′	0.5 s at 95°C, 15 s at 55°C, 20 s at 72°C	196 bp
	R 5′-GAAATGTACGAACTGTACCCT-3′
	FL: 5′-TCCTTGTGGTCATCTAGTAGTATGCAA-FL
	LC: 5′-GATTGTGCTCCTTCCTTTAAGAAAGTGTC-PH

Statistical analysis

The SPSS 13.0 software package (SPSS, Chicago, IL) was used to perform statistical analysis and create diagrams. The Wilcoxon test was applied to compare gene expression data between the RCC samples and the paired normal tissue. Mann–Whitney tests were used to correlate gene expression data with various histopathological features. Associations between the cIAP gene expression parameters and the outcome were analyzed by multivariate Cox proportional hazards models. Kaplan–Meier estimates were calculated to detail differences in recurrence-free and disease-specific survival by cIAP expression and were tested for statistical significance using the log-rank test. All p-values are two-tailed, and those <0.05 were considered statistically significant.

Results

cIAP1 and cIAP2 are overexpressed in renal cell carcinoma

cIAP1 was expressed in all RCC specimens and in 103 of 104 paired normal renal cortex samples. cIAP2 expression was seen in 99 (95.2%) of 104 RCC specimens but only in 72 (69%) of 104 normal renal tissue samples. Quantitative expression analysis revealed a median cIAP1 expression of 0.68 (range of 0.0–5.6) in normal renal cortex tissue. The median expression of cIAP1 in RCC samples was 1.66 (range of 0.02–28.3), indicating a median 2.4-fold overall increase compared to normal tissue. cIAP2 was equally overexpressed in RCC. Its median expression was 0.75 (range of 0.0–110) in normal samples but 1.94 (0.0–45.4) in RCC tissue, suggesting an overall 2.6-fold elevation in RCC samples (Fig. 1a).

Figure 1. Descriptive statistics of cIAP expression parameters. Boxplots show median, 25th and 75th percentiles and range, excluding extreme values. (a) Relative cIAP1 and cIAP2 mRNA expression in tissue form normal renal cortex and renal cell carcinoma (RCC). Expression of cIAP1 and cIAP2 was significantly higher in RCC. (b) cIAP1- and cIAP2-ratio (Tumor/Normal) of paired samples. y-axis is graduated logarithmically as data was skewed toward the right. cIAP1 and cIAP2 expression was significantly increased in RCC as compared to matching normal renal tissue samples. About 80% of the paired samples analyzed showed an elevated cIAP expression in tumor specimens as compared to normal reference tissue (log cIAP-ratio >1), while the remaining 20% were characterized by decreased cIAP-ratio in tumor tissue (log cIAP-ratio <1).

Pairwise comparison of corresponding normal and malignant samples confirmed that cIAP1 and cIAP2 expression was significantly higher in RCC tissue than in normal renal tissue (p < 0.001 and p < 0.001). cIAP1 expression was increased in RCC samples compared to normal tissue in 79.8% of the patients, while a decrease was observed in the remaining 20.2%. cIAP2 expression in RCC tissue was elevated relative to normal tissue in 83.0% of the patients and decreased in 17.0%. The tumor-to-normal (T/N) ratio of cIAP expression was calculated to better reflect the expression levels in RCC specimens relative to normal reference tissue. The median ratio (T/N) for paired samples was 1.96 (range of 0.19–6389) for cIAP1, which is in line with the 2.4-fold increase seen in unpaired analysis of tumor versus normal samples, and 4.0 (0.001–2964) for cIAP2. Figure 1b depicts the descriptive statistics of cIAP1 and cIAP2 ratios.

Correlation of cIAP1 and cIAP2 expression parameters with histopathological variables and patients' characteristics

When considering the relative gene expression in RCC tissue, both cIAP1 and cIAP2 levels were significantly higher in Stage 1 tumors (median: 2.1 and 2.6) than in the pooled subgroup of pT2 and pT3 tumors (median: 0.6 and 1.3) (p = 0.002 and p = 0.002). Expression of cIAP1 and cIAP2 is not selective for RCC but also occurs in most normal renal cortex tissue samples. Thus, rather than being tumor-specific, the different cIAP expression levels in the RCC samples may merely reflect individual variances found in both normal and malignant tissue. To address this problem, we tested whether the cIAP ratio (T/N) also correlated with the tumor stage. The cIAP1 ratio was significantly higher in pT1 tumors than in locally advanced tumor stages (median ratio: 2.7 vs. 1.6; p = 0.02). Patients with pT1 tumors were more likely to have a cIAP1 ratio >1, i.e. higher cIAP1 level in RCC versus normal renal tissue, than those with pT2 and pT3 tumors (89.1 vs. 71.6%; p = 0.03). No such correlations were seen for the cIAP2 ratio (T/N). The cIAP1 and cIAP2 expression parameters were not associated with the histological tumor subtype or tumor grade, but tumor types other than clear cell RCC or Grade 1 tumors were rare in this study (Table I). The check for associations between cIAP expression parameters and patients' characteristics (sex and age) revealed that the cIAP1 ratio (T/N) correlated with the patients' sex. A cIAP1 ratio (T/N) >1 was observed in 34 of 37 (91.8%) female and 49 of 67 (73.1%) male RCC patients (p = 0.02). However, male participants were more likely to present with pT2 and pT3 tumors than female subjects (63.2 vs. 44.7%), which might at least in part explain the sex-related differences in the cIAP1 ratio.

Correlation of cIAP1 and cIAP2 expression parameters with outcome

Associations between the cIAP expression parameters and the risk of recurrence or tumor-related death were analyzed by Cox proportional hazards models. For these analyses, patients were dichotomized into subgroups with increased or decreased cIAP tumor versus normal tissue levels using a cut point of one for the cIAP ratio (T/N).

In multivariate analysis adjusted for sex and age, both the tumor stage and the cIAP1 ratio (T/N) were significantly associated with the outcome, but the tumor grade was not (Table III). The hazard rate (95% confidence interval) for patients with a cIAP1 ratio <1 was 2.96 (CI 95%: 1.23–7.09) for tumor recurrence and 2.78 (CI 95%: 1.22–6.38) for tumor-related death (p = 0.015/p = 0.016). Results remained essentially the same when analyzing the cIAP1 ratio and the tumor stage and grade in univariate models. Taken together, these data suggest that the cIAP1-ratio has a prognostic effect independent of the tumor stage and grade, a decreased cIAP1 expression (T/N ratio < 1) predicting an increased risk of tumor recurrence and tumor-related death.

Table III. Cox Proportional Hazards Model for Tumor Recurrence and Tumor-Related Death in RCC Patients Following Surgical Treatment

	Relative risk (CI 95%); p value
	Tumor recurrence	Tumor-related death
Univariate model
cIAP1 ratio (T/N) < 1	3.64 (1.54–8.60); p = 0.003	3.35 (1.47–7.63); p = 0.004
Tumor stage
pT1	1.00	1.00
pT2	7.57 (1.64–34.95); p = 0.01	5.68 (1.20–26.80); p = 0.028
pT3	7.95 (1.70–37.30); p = 0.009	10.46 (2.33–46.95); p = 0.002
Tumor grade
G1/2	1.00	1.00
G3	1.37 (0.58–3.27); p = 0.474	1.47 (0.65–3.34); p = 0.357
Multivariate model
cIAP1 ratio (T/N) < 1	2.96 (1.23–7.09); p = 0.015	2.78 (1.22–6.38); p = 0.016
Tumor stage
pT1	1.00	1.00
pT2	5.97 (1.27–28.02); p = 0.024	4.76 (1.00–22.57); p = 0.05
pT3	7.21 (1.53–34.10); p = 0.013	9.20 (2.03–41.68); p = 0.004

Kaplan–Meier estimates were calculated for the subgroups of patients with high versus low cIAP1 ratios (T/N) to illustrate the prognostic effect of this parameter (Figs. 2 and 3). The mean time to recurrence was 102.8 months (CI 95%: 92.0–113.7) in patients with a cIAP1 ratio (T/N) above the selected cutoff but only 61.7 months (CI 95%: 41.1–81.3) in those with a lower one (p = 0.002). The mean time to tumor-related death was 103.1 months (CI 95%: 93.1–113.1) in patients with a cIAP1 ratio (T/N) >1 and only 65.1 months (CI 95%: 47.3–82.9) in those with a cIAP1 ratio (T/N) ≤1 (p = 0.002). The 5-year recurrence-free and disease-specific survival was 84 and 82%, respectively, for patients with a cIAP1 ratio >1, but only 64 and 59% for those with a cIAP1 ratio (T/N) ≤1 (Fig. 2).

Figure 2. Kaplan–Meier estimates of cumulative (a) recurrence-free and (b) disease-specific survival in RCC patients. Significant differences were observed among the subgroups of high (solid line) and low cIAP1-ratio (dashed line), indicating the predictive value of the cIAP1-ratio.

Figure 3. Kaplan–Meier estimates of cumulative (a) recurrence-free and (b) disease-specific survival in RCC patients with locally advanced (pT3) RCC. Strong differences were observed among the subgroups of high (solid line) and low cIAP1-ratio (dashed line), indicating favorable prognosis of patients with a cIAP1-ratio (T/N) >1 within the stratum of pT3 tumors.

To control for the prognostic effect of the tumor stage, we stratified accordingly and performed stratum-specific Kaplan–Meier analyses. Recurrence-free and disease-specific survival rates were not significantly related to the cIAP1 ratio in patients with pT1 and pT2 tumors. In the subgroup of patients with pT3 tumors, however, the cIAP1 ratio (T/N) was predictive of prognosis. In this subgroup, the mean time to recurrence was 82.8 months (CI 95%: 76.6–92.1) in patients with a cIAP1 ratio (T/N) >1 but only 23.9 months (CI 95%: 0.0–49.2) in those with a ratio ≤1 (p < 0.001). The mean time to tumor-related death was 99.0 months (CI 95%: 78.8–119.1) in patients with a cIAP1 ratio (T/N) >1 but only 28.7 months (CI 95%: 8.6–48.7) in those with a ratio ≤1 (p= 0.001). The estimated 5-year recurrence-free and disease-specific survival rates were 92 and 77% for patients with a cIAP1 ratio (T/N) >1 but only 17 and 13% for those with a ratio ≤1 (Fig. 3). Similar correlations were observed when patients were dichotomized by the median cIAP1 expression in RCC samples. Our stratum-specific analyses indicated a prognostic effect of cIAP1 expression, especially among patients with pT3 tumors. On the other hand, analyses within the strata of pT1 and pT2 tumors were hampered by the low number of events, i.e. recurrences and deaths. In contrast to the findings obtained for cIAP1, parameters of cIAP2 expression were not predictive of outcome.

Discussion

Renal cell carcinoma (RCC) is characterized by its resistance to chemo- and radiotherapy, but XIAP and survivin are the only apoptosis regulators of the IAP group that have been investigated in this tumor entity.1, 3, 14, 25, 26, 27 The present study demonstrates for the first time that the inhibitors of apoptosis cIAP1 and cIAP2 are overexpressed in the majority of RCC compared to normal renal tissue. Surprisingly, however, downregulation of the cIAP1 gene at the mRNA level in tumor versus normal reference tissue was associated with locally advanced disease and predicted a worse prognosis in RCC patients after surgical treatment.

The pathologic tumor stage is known to be the most important prognostic factor in RCC.1, 28 Cox regression analyses of our data suggest that cIAP1 expression has a prognostic effect independent of tumor stage and grade. However, it must be kept in mind that the findings from the multivariable analyses in our study are based on a rather limited number of cases and events and have to be confirmed in larger cohorts or after longer follow-up periods. Remarkably, the prognostic effect of cIAP1 seemed to be most pronounced in patients with locally advanced tumors, the 5-year disease-specific survival being 77% in cases with elevated cIAP1 levels but only 13% in those with decreased levels. However, the fact that the cIAP1 ratio was not predictive of recurrence or tumor-related death in patients with pT1 and pT2 tumors may be attributable to the limited number of events in these strata.

At first glance, our findings demonstrating the favorable prognostic effect of cIAP1 overexpression are at odds with data on other IAPs like survivin, whose overexpression proved to be a reliable marker of unfavorable histopathological parameters and outcome in several malignancies (reviewed in Ref.29). However, there have been observations similar to ours in that an association between increased IAP levels and a favorable prognosis was also reported for another class 1 IAP, namely XIAP. Increased XIAP levels correlated with longer recurrence-free survival in prostate cancer and with longer disease-specific survival in nonsmall cell lung cancer.30, 31

A number of studies have shown that cIAP1 and cIAP2 mRNA as well as proteins are broadly expressed at different levels in normal human tissues, including the kidney.17, 32, 33 Thus cIAP1 is expressed in nearly all normal renal cortex samples, while cIAP2 expression in the normal kidney appears to be more restricted than previously thought. Expression analyses in RCC as well as pairwise tumor-to-normal comparisons have demonstrated for the first time that both IAPs are overexpressed at the transcriptional level in most RCC. The reason for this is not known. However, downregulation of cIAP1 and cIAP2 expression at the transcriptional level has been observed in conjunction with the expression of functionally intact von Hippel–Lindau tumor suppressor protein (pVHL) and attenuation of NF-κB activity in RCC cell lines. Accordingly, RCC cell lines lacking functional pVHL showed increased expression of NF-κB-driven antiapoptotic genes, including cIAP1 and cIAP2. This in turn was associated with resistance to TNF-α-induced cytotoxicity.34 Although the incidence of VHL disease is rare, biallelic inactivation of the VHL gene has been documented in a majority of RCC.35 Overexpression of cIAP1 and cIAP2 may play a role in tumor biology, since it also occurs in other malignancies, including prostate cancer and malignant pleural mesothelioma.31, 36 On the other hand, downregulation of cIAP1 expression has been demonstrated in cervical carcinoma, and decreased levels of both cIAP1 and cIAP2 have been reported in colon cancer.37, 38 All these observations indicate that cIAP expression levels are closely related to the tumor type and also suggest that the multifaceted functions of various IAPs may be dependent on cell-type-specific factors. Indeed, besides exhibiting antiapoptotic properties, cIAP1 may also potentiate TNF-α-induced apoptosis, and cIAP1-overexpressing cells have recently been shown to have difficulty in completing mitosis, which results in slower cell growth.23, 39 These proapoptotic and cell-growth-inhibiting effects are basically in line with our findings. However, our data cannot be used as a basis for drawing conclusions regarding the functional implications of cIAP expression. In this context, it remains to be elucidated whether differences in cIAP mRNA levels translate to the protein level, as shown for XIAP in RCC.15 Furthermore, studies focusing on functional aspects of cIAPs in RCC would also require the investigation of IAP antagonistic proteins like Smac/Diablo and Omi/HtrA2.21, 40

However, our data raise the question of whether downregulation of cIAP1 may play a role in the progression of RCC to locally advanced and metastatic disease. Interestingly, the closest cIAP paralog XIAP has recently been shown to be associated with RCC progression. Its expression levels increased from early to advanced tumor stages and also with tumor dedifferentiation.14, 15 Unexpectedly, we found cIAP1 and cIAP2 mRNA expression to be inversely correlated with the tumor stage. However, when the tumor-to-normal ratio was considered, correlations with the tumor stage were confirmed for cIAP1 but not cIAP2. Thus individuals with a high cIAP2 expression in both normal and paired tumor specimens were more likely to present with early-stage tumors. It remains to be elucidated, however, whether low individual expression levels of cIAP2 predispose for the development of more advanced RCC stages.

Interestingly, the significantly higher cIAP1 ratio (T/N) in the group of patients with early-stage tumors, i.e. pT1 RCC, shows a striking similarity to the high cIAP1 expression observed in low-stage adenocarcinomas of the lung.41 This seems to suggest that, irrespective of a probable implication of cIAPs in tumorigenesis and tumor progression, increased cIAP1 expression parameters are associated with less aggressive tumor phenotypes. This is strongly supported by the associations we found between the cIAP1 expression parameters and the outcome.

In summary, the results of our study demonstrate a high variability of cIAP1 and cIAP2 expression levels in normal and malignant renal tissue, but overexpression of both IAPs commonly occurs in RCC. However, downregulation of cIAP1, which occurs in 20% of RCC cases, independently predicts an unfavorable outcome in RCC patients submitted to tumor surgery, particularly those presenting with locally advanced disease. Thus, decreased cIAP1 expression may relate to a more aggressive, potentially lethal RCC phenotype but further studies are needed to clarify, whether a decrease in cIAP1 mRNA expression is functionally relevant or whether this is just an associated phenomenon of RCC progression. However, since decreased cIAP1 mRNA levels predicted a worse prognosis, especially in patients who might be considered for adjuvant approaches, cIAP1 may constitute a helpful molecular parameter for optimizing the selection of patients who might profit from adjuvant therapy.

Acknowledgements

The authors wish to thank Ms. Petra von Kwiatkowski, Ms. Waltraud Jekabsons and Ms. Antonia Maass from the Department of Urology for their excellent support in assessing the samples.