Reduced Glucocorticoid Receptor Expression Predicts Bladder Tumor Recurrence and Progression

Abstract

Objectives

To assess the levels of glucocorticoid receptor (GR) expression in bladder tumors because the status and its prognostic value remain largely unknown.

Methods

We immunohistochemically stained for GR in bladder tumor and matched non-neoplastic bladder tissue specimens.

Results

Overall, GR was positive in 129 (87%) of 149 urothelial tumors, which was significantly (P = .026) lower than in non-neoplastic urothelium (90 [96%] of 94). Forty-two (79%) of 53 low-grade tumors vs 45 (47%) of 96 high-grade carcinomas (P < .001) and 61 (73%) of 84 non–muscle-invasive (NMI) tumors vs 26 (40%) of 65 muscle-invasive (MI) carcinomas (P < .001) were moderately to strongly immunoreactive for GR. Kaplan-Meier and log-rank tests revealed that loss or weak positivity of GR significantly or marginally correlated with recurrence of NMI tumors (P = .025), progression of MI tumors (P = .082), and cancer-specific survival of MI tumors (P = .067). Multivariate analysis identified low GR expression as a strong predictor for recurrence of NMI tumors (P = .034).

Conclusions

GR expression was downregulated in bladder tumors compared with nonneoplastic bladder tumors and in high-grade/MI tumors compared with low-grade/NMI tumors. Decreased expression of GR, as an independent prognosticator, predicted recurrence of NMI tumors. These results support experimental evidence suggesting an inhibitory role of GR signals in bladder cancer outgrowth.

Bladder cancer is the second most common genitourinary malignancy, leading to significant morbidity and mortality.^1–3 Two-thirds to three-fourths of patients with bladder tumor initially present with non–muscle-invasive (NMI; pTa or pT1) tumors that can often be treated with conservative approaches, but they experience recurrences, occasionally with grade and/or stage progression. Patients with muscle-invasive (MI; ≥pT2) bladder cancer have high risks of disease progression and metastasis in spite of available aggressive treatment modalities. However, current molecular markers remain insufficient to precisely predict the potential for tumor recurrence and progression.

Glucocorticoid receptor (GR) belongs to the steroid hormone receptor superfamily. As is the case with other steroid receptors, glucocorticoid (GC)-bound GR translocates into the nucleus and binds to GC response elements, which in turn transactivates various genes.^4,5 The activated GC-GR complex also interacts with transcription factors, such as nuclear factor (NF)-κB, that are involved in cell proliferation/survival and cell cycle regulation.^5,6 Thus, GC-regulated genes via the GR pathway play a vital role in maintaining various cellular, molecular, and physiologic networks.

GCs are frequently used, for instance, in the treatment of inflammatory and autoimmune disorders. In cancer patients, GCs are also used for alleviating the adverse effects of chemotherapy or radiotherapy.⁷ In addition, antitumor activities of GCs have been observed, and several GCs are indeed being used clinically as cytotoxic drugs, with or without other chemotherapeutic agents, mainly for hematologic malignancies and castration-resistant prostate cancer.^8–10 In bladder cancer, using cell line models, GC treatment was shown to induce resistance to cytotoxic effects of cis-platin.^11,12 We found that GR activation by a synthetic GC dexamethasone strongly inhibited cell invasion and metastasis of bladder cancer in vitro and in vivo.¹³ Nonetheless, the functional role of GR signaling in bladder cancer progression needs to be further investigated.

It appears that GR expression status had never been examined in human bladder cancer. In our previous study with 24 cystectomy cases,¹³ GR expression tended to be weaker in tumor cells than in nonneoplastic urothelial cells, and strong GR positivity tended to correlate with a better prognosis. However, the findings were unlikely conclusive, presumably because of the relatively small number of cases studied, with no lower grade tumors. In the current study, we aim to validate our earlier results in a larger patient cohort with a wider range of histopathologic features and further elucidate the potential role of GR expression as a biomarker in bladder cancer.

Materials and Methods

We retrieved 152 bladder tissue specimens obtained through transurethral resection or cystectomy performed at the Johns Hopkins Hospital (Baltimore, MD) or the University of Rochester Medical Center (Rochester, NY). All the sections were reviewed for confirmation of original diagnoses according to the 2004 World Health Organization/ International Society of Urological Pathology classification system for urothelial neoplasms¹ (149 urothelial tumors and three squamous cell carcinomas). Appropriate approval was obtained from the institutional review board at each institution before construction and use of the tissue micro-array (TMA). Bladder TMAs, including 152 tumor tissues and 94 benign-appearing tissues from bladders of patients with tumors, were constructed from formalin-fixed paraffin-embedded specimens, as described previously.¹⁴ The tumors included 10 papillary urothelial neoplasms of low malignant potential (PUNLMPs), 43 noninvasive (pTa) low-grade urothelial carcinomas, 31 NMI high-grade urothelial carcinomas, and 65 MI high-grade urothelial carcinomas, in addition to three pT2N0 squamous cell carcinomas. All 65 patients with MI urothelial carcinoma underwent cystectomy. None of the patients had received radiotherapy or systemic chemotherapy preoperatively, whereas 17 cases had intravesical bacillus Calmette-Guérin treatment before radical cystectomy. All 149 cases with urothelial tumor were included in our prior study analyzing 188 cases for the expression of androgen receptor (AR), estrogen receptor (ER)-α, and ERβ.¹⁴

Immunohistochemical staining was performed using the primary antibody to GR (H-300, 1:200 dilution; Santa Cruz Biotechnology, Santa Cruz, CA), which recognizes both main human isoforms, GRα and GRβ, as described previously.^13,14 All the stains were manually scored by one pathologist (H.M.) blinded to patient identity. The German immunoreactive score was calculated by multiplying the percentage of immunoreactive cells (0% = 0; 1%–10% = 1; 11%–50% = 2; 51%–80% = 3; 81%–100% = 4) by staining intensity (negative = 0; weak = 1; moderate = 2; strong = 3). The immunohistochemical scores (ranging from 0–12) were considered negative (0; 0 or 1), weakly positive (1+; 2, 3 or 4), moderately positive (2+; 6 or 8), and strongly positive (3+; 9 or 12) for GR expression.

Statistical analyses were performed using PASW statistics 18 software (IBM, Chicago, IL). The Fisher exact test was used to evaluate the association between categorized variables. Survival rates in 149 patients with urothelial tumor were calculated with the Kaplan-Meier method and comparison was made with the log-rank test. Tumor recurrence was evaluated in patients with NMI tumor. Tumor progression was separately evaluated in patients with NMI tumor (development of high-grade carcinoma [in those with initial PUNLMP or low-grade carcinoma], pT1-pT4 tumor [in those with initial pTa tumor], or pT2-pT4 tumor [in those with initial pT1 tumor]) and in patients with MI tumor (development of local recurrence or metastatic tumor after cystectomy). Disease-specific survival was evaluated in patients with MI tumor. In addition, the Cox proportional hazards regression analysis was used to determine statistical significance of predictors in a multivariate setting. P values less than .05 were considered to be statistically significant.

Results

Using immunohistochemistry, we investigated the expression of GR in 149 bladder urothelial neoplasm specimens and corresponding 94 non-neoplastic bladder tissues. Positive signals were detected predominantly in the nuclei of benign and malignant epithelial cells Image 1A, Image 1B, Image 1C, and Image 1D. Correlations of the expression status with different non-neoplastic and neoplastic bladder tissues are summarized in Table 1.

Image 1.

Immunohistochemistry showing glucocorticoid receptor expression (A–D). A, strong staining in non-neoplastic urothelium; B, strong staining in non–muscle-invasive tumor; C, strong staining in muscle-invasive (MI) tumor; and D, weak staining in MI tumor.

Table 1.

Correlation of Glucocorticoid Receptor Expression with Clinicopathologic Profile of the Patients

	No.	Negative Expression, No. (%)	Positive Expression, No. (%)			P Value
		0	1+	2+	3+	0 vs 1+/2+/3+	0/1+ vs 2+/3+	0/1+/2+ vs 3+
Tissue						.026	.687	.761
Non-neoplastic urothelium	94	4 (4.3)	32 (34.0)	33 (35.1)	25 (26.6)
Urothelial neoplasm	149	20 (13.4)	42 (28.2)	51 (34.2)	36 (24.2)
Sex						.785	.847	1.000
Male	114	16 (14.0)	32 (28.1)	38 (33.3)	28 (24.6)
Female	35	4 (11.4)	10 (28.6)	13 (37.1)	8 (22.9)
Tumor grade						.011	<.001	.046
PUNLMP + LG	53	2 (3.8)	9 (17.0)	24 (45.3)	18 (34.0)
HG	96	18 (18.8)	33 (34.4)	27 (28.1)	18 (18.8)
Tumor invasiveness						<.001	<.001	.084
NMI	84	3 (3.6)	20 (23.8)	36 (42.9)	25 (29.8)
MI	65	17 (26.2)	22 (33.8)	15 (23.1)	11 (16.9)
Lymph node involvement						.759	.199	.197
pN0	47	12 (25.5)	11 (23.4)	12 (25.5)	12 (25.5)
pN+	21	4 (19.0)	10 (47.6)	5 (23.8)	2 (9.5)

HG, high-grade urothelial carcinoma; LG, low-grade urothelial carcinoma; MI, muscle-invasive tumor; NMI, non–muscle invasive tumor; PUNLMP, papillary urothelial neoplasm of low malignant potential.

GR was positive in 90 (96%; 32 were 1+, 33 were 2+, and 25 were 3+) of 94 non-neoplastic urothelial tissues and 129 (87%; 42 were 1+, 51 were 2+, and 36 were 3+) of 149 urothelial tumors. Of the 94 cases in which both benign and tumor tissues were stained, only 20 (21%) cases showed higher scores of GR expression in tumor, compared with benign urothelium (P = .046 with the sign test). Overall, the rate of GR positivity was significantly lower in tumors than in non-neoplastic tissues (P = .026). On the other hand, GR was negative in one of three pure squamous cell carcinomas, whereas the remaining two cases showed moderate positivity (2+).

Next we evaluated the correlation of GR expression levels in urothelial tumors with the clinicopathologic profile available for our patient cohort. There was no significant difference in expression pattern between male and female cases. Fifty-one (96%) of 53 lower grade tumors (PUN-LMPs + low-grade carcinomas) were GR-positive, whereas 78 (81%) of 96 high-grade carcinomas were GR-positive (P = .011). Similarly, 81 (96%) of 84 NMI tumors expressed GR compared with 48 (74%) of 65 MI tumors (P < .001). Differences in GR expression (0/1+ vs 2+/3+) remained significant by grade (lower: 79% vs high: 47%, P < .001) and stage (NMI: 73% vs MI: 40%, P < .001). Even higher rates of increased GR expression (2+/3+) were seen in lower grade (P = .042) or NMI (P = .151) tumors than in non-neoplastic bladders (62%). In contrast, there was no significant difference in GR expression between pN0 tumors and those with lymph node involvement.

Concurrent urothelial carcinoma in situ (CIS) lesions seen in high-grade MI urothelial carcinoma cases were also stained for GR. Similar to its expression status in high-grade or MI tumors, GR was 0 in five cases (22%), 1+ in eight cases (35%), 2+ in seven cases (30%), and 3+ in three cases (13%). Thus, differences in GR levels between lower grade vs CIS (0 vs 1+/2+/3+: P = .024; 0/1+ vs 2+/3+: P = .003; 0/1+/2+ vs 3+: P = .093) and between NMI vs CIS (0 vs 1+/2+/3+: P = .011; 0/1+ vs 2+/3+: P = .013; 0/1+/2+ vs 3+: P = .119) were statistically significant.

We then performed Kaplan-Meier analysis coupled with the log-rank test to assess possible associations between GR staining and patient outcomes. There were no statistically significant differences between GR positivity (0 vs 1+/2+/3+) and tumor recurrence or progression. However, lower expression levels (0/1+) of GR in NMI tumors were strongly associated with tumor recurrence (P = .025) Figure 1A, but not with grade/stage progression (P = .165) Figure 1C. In addition, there was a statistically significant difference in progression-free survival in patients with lower grade (PUNLMP + low-grade carcinoma) NMI tumor (GR 0/1+ vs 2+/3+, P = .003). Patients with strongly GR-positive MI tumor also had lower risks of disease progression (0/1+ vs 2+/3+, P = .082, Figure 1D; 0/1+/2+ vs 3+, P = .030, Figure 1E) and disease-specific death (0/1+ vs 2+/3+, P = .067, Figure 1F; 0/1+/2+ vs 3+, P = .101, figure not shown). Moreover, there was a significant difference in progression-free survival in patients with pN0 MI tumor (GR 0/1+ vs 2+/3+, P = .043). Interestingly, when analyzed separately in men and women, GR expression showed more significant differences in the survival rates of, for instance, recurrence of NMI tumors (P = .019) Figure 1B, progression of MI tumors (P = .029 or P = .030) Figure 1G and Figure 1H, and disease-specific mortality of MI cases (P = .029) Figure 1I in male patients. By contrast, analysis in female patients failed to find a prognostic significance of GR expression.

Figure 1.

Results of Kaplan-Meier analysis according to the levels of glucocorticoid receptor expression in non–muscle-invasive (NMI) tumors (A–C) and muscle-invasive (MI) tumors (D–I). Recurrence-free survival in all cases (A; P = .025) or male patients (B; P = .019) with NMI tumor. Progression-free survival in all cases with NMI tumor (C; P = .165). Progression-free survival in all cases (D, E; P = .082 and P = .030, respectively) or male patients (G, H; P = .029 and P = .030, respectively) with MI tumor. Disease-specific survival in all cases (F; P = .067) or male patients (I; P = .029) with MI tumor. Comparisons were made using the log-rank test.

To see whether GR expression was an independent prognosticator, multivariate analysis was performed with Cox model Table 2. We excluded lymph node involvement in NMI tumors and tumor grade in MI tumors from the analyses, because lymph node dissection was performed in only a few NMI cases and all MI tumors showed high-grade carcinoma. In the NMI subgroup, low GR (0/1+) was found to correlate with tumor recurrence (hazard ratio [HR] = 0.444; 95% confidence interval [CI] = 0.210–0.939; P = .034), but not with tumor progression (HR = 0.704; 95% CI = 0.184–2.703; P = .610). However, GR status (0/1+ vs 2+/3+ or 0/1+/2+ vs 3+), as well as sex or lymph node involvement, was not an independent predictor of progression or survival in our cohort of MI tumors.

Table 2.

Multivariate Cox Regression Analysis

	Recurrence			Progression			Disease-Specific Death
	HR	95% CI	P Value	HR	95% CI	P Value	HR	95% CI	P value
NMI
GR expression (0/1+ vs 2+/3+)	0.444	0.210 – 0.939	.034	0.704	0.184 – 2.703	.610
Sex	0.509	0.242 – 1.068	.074	0.782	0.148 – 4.147	.773
Tumor grade	1.959	0.948 – 4.048	.069	8.511	1.658 – 43.993	.010
MI
GR expression (0/1+ vs 2+/3+)				0.597	0.263 – 1.356	.218	0.478	0.160 – 1.425	.185
Sex				1.235	0.492 – 3.103	.653	1.213	0.405 – 3.629	.730
Lymph node involvement				1.949	0.905 – 4.197	.088	2.014	0.796 – 5.097	.140
MI
GR expression (0/1+/2+ vs 3+)				0.271	0.064 – 1.151	.077	0.387	0.088 – 1.707	.210
Sex				1.152	0.463 – 2.866	.761	1.092	0.368 – 3.243	.874
Lymph node involvement				1.758	0.814 – 3.796	.151	1.861	0.732 – 4.730	.192

CI, confidence interval; GR, glucocorticoid receptor; HR, hazard ratio; MI, muscle-invasive tumors; NMI, non–muscle-invasive tumors.

Finally, we analyzed the correlations between expressions of GR vs AR, ERα, and ERβ in our cohort of 149 bladder urothelial tumors in which the expression of all four proteins was examined. Overall, AR, ERα, and ERβ were positive in 65 (44%), 38 (26%), and 76 (51%) tumors, respectively. Of AR-, ERα-, and ERβ-positive tumors, 59 (91%), 38 (100%), and 60 (79%), respectively, were also GR-positive, whereas 70 (83%) of 84 AR-negative (P = .230), 91 (82%) of 111 ERα-negative (P = .004), and 69 (95%) of 73 ERβ-negative (P = .007) tumors, respectively, were GR-positive. Thus, GR and ERα were positively correlated, and GR and ERβ were negatively correlated.

Discussion

An increasing amount of evidence suggests the involvement of steroid hormones and their receptor signals in bladder carcinogenesis and cancer progression.¹⁵ Among them, the roles of androgens/AR and estrogens/ERs in bladder cancer have been relatively well studied.^14–18 In contrast, little is known about GC-mediated GR functions in bladder cancer growth. Although prolonged systemic use of GCs has been correlated with an increased risk of bladder cancer, this is presumably because of immunosuppression.¹⁹ Thus, it remains unclear whether GCs directly affect the development of bladder cancer via the GR pathway. Recently, using preclinical models, we demonstrated that dexamethasone-enhanced GR signals suppressed bladder cancer cell invasion and metastasis potentially via NF-κB inactivation and interleukin-6 downregulation as well as by reversing epithelial-to-mesenchymal transition, while stimulating cell proliferation by preventing apoptosis.¹³ Nonetheless, the status of GR expression in bladder cancer specimens has not been assessed, except in our pilot study of 24 cases of high-grade urothelial carcinoma¹³; however, it has been reported in other types of malignancy, including non–small cell lung carcinoma²⁰ and renal cell carcinoma.²¹

In the present study, we confirmed our recent observation in immunohistochemistry¹³ and further showed that nuclear GR expression was significantly reduced in bladder tumors, compared with non-neoplastic bladders, and in high-grade MI tumors, compared with lower grade NMI tumors. Higher expression (2+/3+) was even more frequently seen in lower grade or NMI tumors compared with non-neoplastic bladders. However, “normal” urothelium was derived from non-neoplastic tissues in patients with bladder tumor, which was a limitation of the current study. The levels of GR expression in CIS lesions were also found to be similar to those in high-grade or MI urothelial carcinomas. It is still important to determine if a low-grade component potentially coexisting in a high-grade tumor shows the same degree of staining intensity. Furthermore, although we included three cases of squamous cell carcinoma in this study, other histologic types of bladder tumor, such as adenocarcinoma and small cell carcinoma, may also need to be stained for GR.

Consistent with our preclinical data,¹³ reduced expression of GR was found to correlate with poorer outcomes. In particular, GR levels independently predicted tumor recurrence in patients with initial NMI tumor. These findings suggest that GR, as a transcription factor, is functionally active in most non-neoplastic or neoplastic bladder cells; the findings also support the idea that GR activation in urothelial cells prevents the development of recurrent tumor and possibly invasion or metastasis. Accordingly, low-dose GC treatment that does not induce immunosuppression has the potential to be a chemopreventive and therapeutic approach in patients with bladder cancer. Of note, our results also indicate that GR status can act as a prognosticator of bladder tumor.

Recent observations have suggested a sex-specific action of GCs.^22–24 Moreover, GC treatment has been shown to result in inhibition of androgen biosynthesis and subsequent modulation of AR activity^25,26 as well as lowering the circulating level of estrogens.²⁷ In breast cancer cells, estrogen was able to inhibit GC-induced kinase genes and reduce GC-induced GR phosphorylation, a transcriptionally active form of GR.²⁸ More recently, GCs were suggested to mediate growth inhibition of breast cancer cells via a direct interaction between GR and ERα.²⁹ Higher GR expression was also shown to correlate with better and worse outcomes in patients with ERα-positive and ERα-negative breast tumors, respectively.³⁰ In addition, in prostate cancer tissues, coexpression of GR and AR³¹ as well as increased GR expression after androgen deprivation therapy³² was demonstrated. As mentioned earlier, both androgen-mediated AR signals and estrogen-mediated ER signals are known to play an important role in bladder tumorigenesis and tumor progression.^15–18 This may indeed explain some of the sex-specific differences in the incidence and progression of bladder cancer, though no studies have identified a significant difference in the expression status of AR or ERs between bladder tumors in men and women.^14–16 Our current results showing that the prognostic value of GR expression is especially evident in male patients may therefore provide evidence suggesting potential cross-talk between GC-GR and androgen-AR/estrogen-ER pathways in bladder cancer. There were no statistically significant correlations between GR status and the prognosis of female patients, presumably as a result of the small number of cases. We also showed that GR was likely to coexpress with ERα in bladder tumors, whereas the expression of GR and ERβ was inversely correlated. Further investigation of GR in preclinical models that determine biological functions of GCs and interactions between GR and AR/ER signals in bladder cancer cells may provide new insights into not only tumor recurrence and progression but also novel preventive and/or therapeutic approaches.

In conclusion, compared with non-neoplastic urothelium, the expression of GR in urothelial neoplasms, especially high-grade MI tumors, was observed to be significantly decreased. Lower GR expression was associated with disease progression, especially in male patients with MI tumor, and was also an independent predictor of tumor recurrence in patients with NMI tumor. These results support experimental evidence suggesting an inhibitory role of GR signals in bladder cancer progression. Further analysis of GR expression in larger cohorts, particularly with female patients, is necessary to definitively determine the prognostic significance of GR status in bladder cancer.