The T393C Polymorphism of the Gαs Gene (GNAS1) Is a Novel Prognostic Marker in Bladder Cancer

Transitional cell carcinoma (TCC) of the bladder is the fourth most frequent cancer in the United States with ∼57,400 new cases expected in the year 2003 (1). At first presentation, roughly 70% of cases are classified as superficial disease, whereas 30% of tumors have already invaded the bladder wall (2, 3). Ninety-five percent of urothelial neoplasms are of the transitional cell type (TCC). The overall 5-year survival rate for all patients with urothelial cancer is 81%; however, this rate varies greatly depending on multiple factors, including disease stage, and mutations in tumor suppressor genes (4, 5). Within the United States, TCC is twice as common among White people compared with Blacks, and incidence and mortality directly increases with age (1, 6). Still, the best established and routinely used clinical markers to predict prognosis are depth of tumor invasion, presence or absence of lymph node and distant metastases, and tumor differentiation (7). Tumor progression is thought to result from the accumulation of multiple genetic alterations, including the activation of oncogenes, loss of distinct chromosomal regions, and inactivation of tumor suppressor genes, such as p53 and Rb (8-10). Inactivation of these genes may, for example, result in impaired apoptosis of cancer cells, which contributes to tumor progression.

Recent investigations have shown that an increased concentration of the intracellular second messenger cyclic AMP (cAMP) promotes apoptosis in several cell types including leukemic cells (11), ovarian cancer cells (12), and lymphoma cells (13). Elevation of cAMP results from signal transduction from G protein–coupled receptors to the Gαs protein, which in turn activates its effector adenylyl cyclase to produce cAMP (14). A potential role of Gαs in promoting apoptosis has been proposed through experiments in which increased expression of Gαs has been shown to activate the adenylyl cyclase signal transduction cascade resulting in an accumulation of cAMP (15, 16). The Gαs subunit is encoded by the gene GNAS1, which is mapped to chromosome 20q13 and consists of 13 exons. Somatic activating mutations of GNAS1 have been implicated in the etiology of McCune Albright Syndrome (17) and sporadic, isolated endocrine tumors (18-20), which principally supports a role of GNAS1 in tumor initiation and/or progression. Moreover, a common synonymous single nucleotide polymorphism in exon 5 of the gene GNAS1 (T393C) was associated with hypertension although its functional role was not elucidated (21-23). The aim of the present study was to investigate a potential association between genotypes of the T393C polymorphism and disease progression in patients with TCC. Moreover, we investigated genotype-dependent expression of Gαs mRNA in different tissues to potentially uncover a functional mechanism that could explain the observed results.

Tumor tissue was stored from patients with bladder cancer (n = 389), who were treated by either transurethral resection and/or cystectomy for TCC of the bladder in the years between 1989 and 2001 at the Department of Urology at the University of Essen and who consented to tissue storage. This group comprised 290 men and 109 women. The mean age at first presentation was 64.2 ± 10.8 years and 153 patients were current smokers at primary diagnosis of TCC (Table 1). In collaboration with the affiliated primary urologists and general practitioners, we were able to work up the follow-up of 254 individuals, who then were finally included into the study. All patients were White Germans of Caucasian ethnicity. Another prerequisite for entry into the study was that histopathologic confirmation of TCC of the bladder as well as grading and staging were available. The ethics committee of the Essen University Hospital approved the study protocol and all patients provided written informed consent.

The tissue was placed into a plastic vial and snap-frozen in liquid nitrogen for storage. The tumors were staged and graded by standard histologic analysis according to the guidelines of the International Union Against Cancer (24). All patients were screened for the presence of metastasis by abdominal ultrasound and chest X-ray before surgical procedure. In the case of locally advanced disease (stage ≥ pT₂), patients received an abdominal CT scan.

The medical records of all patients were reviewed and follow-up data were collected in cooperation with the responsible urologist and/or family physician. The data were evaluated for the end points tumor progression, metastasis, tumor-related deaths, and overall survival. We assessed smoking habits at the time of hospitalization (yes versus no), but data on smoking history or pack-years were not routinely available. The cause of death was recorded using both hospital files and information from the responsible urologist/family physician. The last clinical assessment was used to define the follow-up period. The mean follow-up period was 45.2 months. Progression was defined as an increase in stage, occurrence of metastasis, or local recurrence after cystectomy or death due to bladder cancer. If a patient was initially lost to follow-up, we consulted the municipal registry to assess survival.

DNA was extracted from native tumor tissue using a commercially available kit (QIAAmp, Qiagen, Hilden, Germany). Genotypes for the T393C polymorphism were determined by PCR with the use of the following primers: forward primer 5′-CTCCTAACTGACATGGTGCAA-3′ and reverse primer 5′-TAAGGCCACACAAGTCGGGGT-3′. After denaturation at 94°C, 35 cycles of DNA amplification were done using Taq PCR Mastermix (Eppendorf, Hamburg, Germany) at 94°C for 45 seconds, 58°C for 40 seconds, and 72°C for 45 seconds. The 345 bp PCR products were digested using the restriction enzyme FokI and analyzed on a 2% agarose gel. The unrestricted products (345 bp) represent the TT genotype; the completely restricted products (259 and 86 bp) represent the CC genotype.

Tumor specimens from patients undergoing transurethral resection or cystectomy for bladder cancer was snap-frozen in liquid nitrogen directly after dissection. Adipose tissue was obtained from 45 patients undergoing mamma reduction or gastric banding operation. All adipose tissue samples were immediately snap frozen in liquid nitrogen. Right atrial specimens were obtained from 15 patients undergoing cardiac surgery. RNA was prepared from two independent tissue samples using the RNeasy kit (Qiagen) according to the manufacturer's instructions including DNase treatment of all samples.

First-strand cDNA was synthesized from 0.1 to 1.0 μg of total cellular RNA with oligo-dT primers (Roche, Mannheim, Germany) by using Superscript II reverse transcriptase as recommended by the supplier (Invitrogen, Karlsruhe, Germany). Quantification of Gαs was carried out by using primers 5′-GCACCATTGTGAAGCAGATG-3′ and 5′-TCAATCGCCTCTTTCAGGTT-3′ spanning introns 2 and 3. PCR and primers for the housekeeping gene ACTB encoding hβ-actin were used as described (25). The PCR reaction mix was prepared using the Quantitect SYBR Green kit (Qiagen) following the manufacturer's instructions. PCR was done in the ABI Prism 7700 Sequence Detection System (Applied Biosystems, Foster City, CA), which detects the signal from the SYBR green during PCR. A cDNA dilution series for Gαs confirmed a PCR efficiency >95% of which were comparable with the efficiency of hβ-actin (25). Results were confirmed using samples from independent reverse transcription and real-time PCR reactions. The relative expression of Gαs compared with β-Actin was calculated by using the following formula: Gαs / β-Actin = 2^{−[C_t(Gαs) − C_t(β-Actin)]}, where C_t is the number of cycles at which the fluorescence signal reaches a predefined threshold. Each C_t represents the mean C_t value of each sample duplicate.

The clinical outcomes analyzed in this study were survival, time to first progression, and time to first occurrence of metastases. As the mortality from bladder cancer directly increases with age, we used the receiver operating characteristic curve analysis for overall survival to define two age groups with ≤54 years to have a favorable and >54 years to have an unfavorable overall survival (26). Kaplan-Meier plots and the log-rank test for trend were used to evaluate the relationship between stage, age, T393C genotypes, and clinical outcome from the date of the primary diagnosis to a 5-year follow-up. Log-rank tests for T393C genotypes were adjusted for pathologic stage and comparison was done using all three genotypes unless stated otherwise. The impact of gender, tumor stage, age, and T393C genotypes as prognostic factors for the clinical outcome were analyzed by backward stepwise multivariate Cox regression analysis. Hazard ratios and 95% confidence intervals (95% CI) were calculated from the Cox regression model including all factors. Contingency tables and the Pearson's χ² test were used to calculate odds ratios (OR) and 95% CI for categorical variables using T393C allele frequency numbers or genotypes as indicated. Because the T393C polymorphism shows a gene-doses effect, linear ANOVA was used for comparison of continuous variables where appropriate. Differences were regarded significant at P <0.05. All statistical analyses were done using SPSS 11.0 (SPSS, Chicago, IL) or Graphpad Prism 4.0 (Graphpad Software, San Diego, CA).

Continuous variables are given as means ± SD or ±SE as indicated. ORs are given with 95% CIs.

Demographic characteristics of the whole group and the selected case group with complete follow-up are displayed in Table 1. There were no statistically significant differences regarding demographic characteristics between both groups, indicating that the case group is representative for the whole group. Mean age at diagnosis in our sample was 62.9 years, which is only 3.5 years younger than that described in two German epidemiologic studies surveying over 3,000 and 16,000 patients, respectively (27, 28). Grading and staging was higher in our study population compared with these latter studies in which 38.9% to 54.7% were pT_a, 14.9% to 25.0% pT₁, 15.6% to 20.0% pT₂, 8% to 10% pT₃, and 5.0% to 6.9% pT₄ tumors (27, 28).

To confirm that our sample displays the typical clinical course of patients with bladder cancer, we calculated Kaplan-Meier curves for overall survival depending on the established predictive parameter “pathologic stage.” As shown in Fig. 1A, overall survival was significantly dependent on pathologic stage (P < 0.0001) and survival rates were comparable with published data (3, 29). Moreover, as mortality increases with increasing age (29), we used the receiver operating characteristic curve analysis (see Materials and Methods) to define a cutoff value of 54 years to discriminate survival rates. Figure 1B clearly shows that the overall survival is increased in patients younger than 54 years compared with older patients.

We next analyzed possible associations between demographic data and genotypes of the T393C polymorphism (Table 2). Men were affected three times more often than women, which is compatible with data from the literature (1, 28, 30), but neither gender nor smoking habits were related to genotypes. Age at primary diagnosis was slightly higher in TT genotypes (65.1 years) compared with CC genotypes (61.4 years) with TC genotypes showing intermediate values (63.1 years, P = 0.059 ANOVA). Tumor grade was not significantly associated with T393C genotypes. On the other hand, the comparison of pathologic stages and genotype distribution showed a significant relationship between genotype and pathologic stage at the time of initial diagnosis. Only 22.9% with pT_a had CC genotypes, whereas 56.5% of pT₄ patients had CC genotypes (P = 0.021).

Progression, metastasis, and tumor-related mortality dependent on T393C genotypes were analyzed using Kaplan-Meier estimates (Table 3) and survival curves (Fig. 2). Because the 393C allele was significantly associated with advanced disease at the time of diagnosis (Table 2), log-rank tests for clinical outcome were adjusted for clinical stage and separate probabilities for the subgroups of superficial and muscle invasive carcinoma were calculated (Table 3).

Progression was significantly dependent on the T393C genotype with an apparent gene-dose effect (P = 0.011; Fig. 2A). GNAS1 393C homozygous patients displayed a higher risk for tumor progression than T393 homozygous patients, with heterozygous patients being at intermediate risk (hazard ratios: CC versus TT: 2.0, 95% CI 1.1-3.1, P = 0.015; CT versus TT: 1.5, 95% CI 0.9-2.4, P = 0.15).

Time to metastasis was also significantly associated with genotypes of the T393C polymorphism (P = 0.001; Fig. 2B). Whereas 393C homozygous patients were at higher risk for the development of metastases, T393 homozygosity is associated with a protective effect and heterozygous subjects were at intermediate risk (hazard ratios: CC versus TT: 3.7, 95% CI 1.5-5.6, P = 0.002; CT versus TT: 2.5, 95% CI 1.1-4.2, P = 0.03). The same observation emerged for the time to tumor-related death and the T393C polymorphism (P = 0.014; Fig. 2C). GNAS1 393C homozygous patients were at higher risk for death from bladder cancer compared with T393 homozygous individuals, with heterozygous patients being at intermediate risk (hazard ratio: CC versus TT: 2.7, 95% CI 1.2-4.6, P = 0.013; CT versus TT: 2.2, 95% CI 1.0-3.8, P = 0.039). The proportions of progression-free, metastases-free, and overall survival are given in Table 3, which shows that the probability of survival or being free of progression and metastases is reduced with increasing numbers of C alleles for both superficial and muscle-invasive stages.

Smoking is a known risk factor for the development of bladder cancer and an interaction between the T393C polymorphisms and smoking on hypertension has been reported for a Japanese population (23). We, therefore, calculated genotype-dependent Kaplan-Meier curves separately for current smokers and nonsmokers and found that genotype effects were not significantly different in smokers versus nonsmokers (data not shown).

Multivariate analysis, including age at diagnosis, gender, pathologic stage, and T393C genotypes, revealed that patients >54 years at time of diagnosis were at higher risk for tumor-related death than younger patients (hazard ratio, 2.37; 95% CI, 1.21-4.65; P = 0.012). Most interestingly, besides pathologic stage as a known established risk factor, the T393C polymorphism was an independent risk factor for progression, metastasis, and tumor-related death (Table 4). Patients with CC genotype had hazard ratios of 1.94 (95% CI, 1.11-3.38; P = 0.020) for progression, 3.49 (95% CI, 1.46-8.37; P = 0.005) for metastasis, and 2.49 (95% CI, 1.09-5.71; P = 0.031) for tumor-related death compared with the reference group consisting of T393 homozygous individuals. Heterozygous patients were at intermediate risk.

When we tested a potential interaction between the T393C polymorphism and smoking habits in this multivariate model, we found no significant interaction for progression (P = 0.993), metastasis (P = 0.207), or tumor-related death (P = 0.454), indicating that genotype-dependent clinical outcome is independent of smoking status.

Increased expression of Gαs has been shown to activate the adenylyl cyclase signal transduction cascade, resulting in an accumulation of cAMP (15). We, therefore, tested the hypothesis whether genotypes of the GNAS1 T393C polymorphism are associated with altered Gαs expression. To this end, we measured Gαs mRNA expression by means of quantitative real-time PCR using 38 tumor tissue specimens derived from different patients with bladder cancer. Gαs mRNA expression was significantly different between genotypes (P = 0.017; linear ANOVA) with highest expression associated with the TT genotype, followed by TC and CC genotypes (Fig. 3A). To investigate whether genotype-dependent Gαs mRNA expression could be a general phenomenon also in nontumor tissue, we measured its expression in adipose tissue and heart. Similar genotype-dependent Gαs mRNA expression was observed in these tissues with an apparent gene-dose effect in adipose tissue (P = 0.017; Fig 3B) and a comparable trend in human heart tissue that did not reach statistical significance possibly due to the small sample number (P = 0.150; Fig. 3B). These data, therefore, suggest a general mechanism on Gαs mRNA expression associated with GNAS1 genotypes.

The biology of initiation and progression of TCC is complex and its interindividual course is highly variable. Although clinical T stage (TNM) statistically correlates with outcome in a large percentage of patients, prediction of the individual course of disease in patients with identical tumor stages would require the availability of additional biomarkers. This could comprise presence or absence of somatic mutations in specific genes, gene expression profiles, but also genetic host factors (e.g., germ line single nucleotide polymorphisms). As an example, genetic instability and the accumulation of synergistically acting genetic lesions mainly involving p53, p21, retinoblastoma, and growth factors correlate with the progression of TCC (5, 10, 31, 32). In addition, a recent report examined combined p53, p21, and pRb expression in the progression of bladder cancer and detected additive effects of these alterations (33). Nevertheless, although these authors could predict the outcome for patients with tumors that show multiple alterations (p53, p21, or pRb), they also showed that it was still difficult to predict the outcome for patients with tumors that showed relatively few alterations. It was, therefore, suggested that progression of these tumors might be governed by alternative proteins or pathways (33).

We have investigated whether a genetic host factor, the common T393C polymorphism in the gene GNAS1 that encodes the G proteins subunit, may be predictive for clinical outcome in patients with TCC. The results presented here show that time to progression, metastasis, and tumor-related death was significantly accelerated in CC genotypes compared with TC or TT genotypes. Importantly, although 393C allele carriers presented at an advanced stage at first diagnosis, subanalysis of genotype-dependent progression at different stages (superficial and muscle-invasive stage), as well as log-rank statistics adjusted for pathologic stage, suggested a gene-dose effect for C-allele carriers with shorter time to progression, metastasis, and tumor-related death (Table 3). Moreover, the T393C polymorphism was an independent risk factor for clinical outcome when using staging as a covariate with the highest risk for an unfavorable clinical course in CC genotypes (Table 4). It should be emphasized here that the magnitude of the risk associated with C-allele carriage was not significantly different from that of stage, suggesting that genotyping of this polymorphism in patients with TCC could be of clinical relevance provided that our data are replicated by independent studies. Interestingly, older age (>54 years) was an independent risk factor for tumor-related death, whereas no correlation between gender and clinical outcome was observed. The question remains to be answered to what extent our sample is representative for patients with bladder cancer in Germany. According to available epidemiologic surveys, most cases (∼70%) are classified as superficial cancers at the time of first detection (28, 29). In contrast, our sample showed a significant enrichment of patients with higher staging and grading (see Table 1). This discrepancy arises from the following facts: (a) Our study population was recruited from patients treated at our university hospital, which specializes in the treatment of advanced tumors, whereas less advanced cancers are frequently treated by primary urologists. (b) Samples for genotyping were obtained predominantly from patients undergoing cystectomy due to invasive bladder cancer. In contrast, superficial bladder cancer is frequently treated by transurethral resection and obtaining sufficient tumor material from these patients is more difficult. Thus, there is a selection bias that explains why staging and grading of our patients differs from that found in population-based surveys. We have, nevertheless, no evidence to suggest that this selection bias has a major impact on the results described here as we found in our sample the known increased mortality of older patients and those with increased tumor stages (Fig. 1). Nevertheless, our findings will have to be replicated by independent studies until these results can be universally accepted.

The observed associations raise the question about the molecular mechanism(s) potentially explaining the differences in clinical outcome of patients with different GNAS1 genotypes. The T393C polymorphism located in exon 5 of the GNAS1 gene is synonymous (i.e., it neither affects the encoded amino acid sequence nor is it located in the promoter region). However, genotypes are significantly associated with Gαs mRNA expression, with TT genotypes displaying highest expression not only in urothelial tissue but also in adipose and heart tissue, suggesting a general mechanism of increased Gαs expression in TT genotypes (Fig. 3). This result would, therefore, be compatible to recent studies where an association of the T393C polymorphism to hypertension was described (21-23). Unfortunately, the small size of tissue specimens available from each patient prevented more direct assessment of Gαs expression through Western blot analysis. One potential explanation for the association between the T393C polymorphism and Gαs mRNA expression would be the existence of yet unidentified polymorphisms in regulatory regions of GNAS1 (e.g., in the promoter), which are in linkage disequilibrium with the T393C polymorphism. The detection of such mutations was beyond the scope of the present study. However, the availability of enhanced PCR techniques, like “slow-down PCR,” should now facilitate the sequencing of the GC-rich promoter regions of GNAS1, which was previously impeded by technical problems (34). On the other hand, the T393C polymorphism itself could influence the stability of the Gαs transcript, which then could result in altered mRNA expression. For example, synonymous polymorphisms in the human DRD2 gene alter the predicted mRNA folding, lead to a decrease in mRNA stability and translation, and dramatically change dopamine-induced up-regulation of DRD2 expression (35).

Data from in vitro experiments suggest that increased expression of Gαs possibly enhances apoptosis (15, 16, 36). The second messenger cAMP, which is generated by activated Gαs, seems to play a major role in this proapoptotic process. cAMP is able to inhibit the physiologic actions of growth factors (37) and to block the transformation phenotype in selected malignant cells (38). For example, in fibroblasts and smooth muscle cells, cAMP inhibits mitogen-activated protein kinase activation by growth factors (39, 40) and during anchorage-independent cell growth (41). These growth-inhibitory actions are thought to be mediated by G protein pathways that regulate cAMP (38) and the cAMP-dependent protein kinase A (37).

We show here a significantly increased Gαs mRNA expression in tumor tissue from patients homozygous for the 393T allele (Fig. 3). As a working hypothesis, these results are compatible with increased apoptosis in 393T allele carriers, which is especially important in malignant cells and could, therefore, contribute to a more favorable clinical course in these patients.

In conclusion, our data suggest that the GNAS1 T393C polymorphism is independently associated with disease progression in patients with bladder cancer.

It would now be important to replicate these dates in independent studies and to investigate a potential role of this polymorphisms in other cancer entities.

We thank Karl Worm for his excellent assistance with the real-time PCR experiments and Markus Neuhäuser for critical reading of the manuscript.