NR1H3 Expression is a Prognostic Factor of Overall Survival for Patients with Muscle-Invasive Bladder Cancer

Abstract

Background: Nuclear receptors (NRs) are a class of transcription factors that regulate many cellular functions through manipulation of gene expression and also play important roles in tumorigenesis, proliferation, progression and prognosis in various kinds of cancers according to recent studies. This work aimed to determine the predictive ability of NRs in muscle-invasive bladder cancer (MIBC).

Patients and methods: A total of 308 MIBC patients with complete clinicopathological and RNASeq data from The Cancer Genome Atlas (TCGA) cohort were collected for filtration. Genes showed clear correlations with overall survival (OS) and recurrence free survival (RFS) were further validated in 123 MIBC patients recruited consecutively from 2008 to 2012 in Fudan University Shanghai Cancer Center (FUSCC) cohort. Cox proportional hazards regression model and Kaplan-Meier plot were used to assess the relative factors.

Results: In TCGA cohort, we found that high NR1H3 (HR=0.779, 95% CI: 0.634 - 0.957), NR2C1 (HR=0.673, 95% CI: 0.458 - 0.989) and NR2F6 (HR=0.750, 95% CI: 0.574 - 0.980) expressions were independent factors of favorable OS, while only low NR1H3 (log-rank test, P=0.0076) and NR2F6 (log-rank test, P=0.0395) expressions had the ability to predict poor prognosis for RFS. Further, in FUSCC validating cohort, we confirmed that low NR1H3 expression level was independent factor of poor OS (HR=1.295, 95% CI: 1.064 - 1.576) and it had the ability to predict poor RFS (log-rank test, P=0.0059).

Conclusions: Low NR1H3 expression level is an independent prognostic factor of poor OS, and can also predict worse RFS in MIBC patients. Our “TCGA filtrating and local database validating” model can help reveal more prognostic biomarkers and cast a new light in understanding certain gene function in MIBC.

Introduction

Bladder cancer is the seventh most common cancer in the world ¹ and is the most common urologic malignancy in China ². As newly estimated, there were 80,500 new cases in China in 2015, including 62,100 men and 18,400 women ³. Bladder cancer is staged via the tumor-node-metastasis system, which describes the extent of invasion (Tis-T4) ⁴. According to the 7^th edition of the TNM classification published by the UICC, bladder cancer can be divided into non-muscle invasive bladder cancer and muscle invasive bladder cancer (stage T2 or above) ⁵. Approximately 30% of patients suffer MIBC at the time of diagnosis and they have a less favorable prognosis with 5-year survival rate less than 50% ^{6, 7}. Despite improvements in surgical techniques and post-operative recovery, the complexity and high cost of procedure remains highly challenging ⁸, and the treatment has not been advanced for several decades ⁹. Therefore, to further understand the prognosis and to develop new biological therapies, knowledge of novel biomarkers for MIBC is in an urgent need.

Nuclear receptors are a class of proteins that are responsible for sensing steroid and thyroid hormones and certain other molecules. These proteins can work with other molecules to regulate expression of specific genes. In humans, there are 48 kinds of NRs ¹⁰, which can be classified according to mechanism ¹¹ or homolog ¹². Members of NR superfamily have the ability to regulate the development, homeostasis and metabolism, thus are correlated with a variety of diseases including cancer. Till today, many kinds of NRs are proved to be closely associated with the prognosis of cancer, including breast cancer ^{13, 14}, gastric cancer ¹⁵, colon cancer ¹⁶, prostate cancer ^{17, 18}, ovarian cancer ^{19, 20} and so many others. However, the potential roles of NRs in BCa, especially in MIBC, and their biological functions on tumorigenesis, recurrence and prognosis remain unknown. So in this study, we aim to find out if certain NR gene expression level is associated with patients' recurrence or prognosis with MIBC, and has the potential to serve as a novel biomarker.

Materials and methods

Patients and samples

Our study was carried out in accordance with the ethical standards of Helsinki Declaration II and approved by the Institution Review Board of Fudan University Shanghai Cancer Center. Written informed consent was obtained from each patient before any study-specific investigation was performed.

For TCGA cohort, patients' nuclear receptor gene expression and clinical information were downloaded from the website of Cancer Genomics Browser of University of California Santa Cruz (http://genome-cancer.ucsc.edu/). Forty-eight members of nuclear receptor gene family are included in the database as is presented in Supplementary Table 1. Originally, 436 lines' data was obtained. Then, we excluded patients with unclear event, age, gender, tumor subtype, tumor grade, stage and no RNASeq data, along with patients with a T stage of T0 or T1. After all, we included 308 patients of muscle-invasive bladder cancer from TCGA dataset.

For the FUSCC cohort, 123 muscle-invasive bladder cancer patients from 2008-2012 who underwent radical cystectomy were enrolled retrospectively and consecutively. All these tissue samples were collected during surgeries and stored at -70℃ in the tissue bank of FUSCC. Clinicopathological characteristics were obtained from our electronic records. Patients were regularly followed up by telephone or in the clinic once every 3 months. Events, including death and recurrence, were recorded.

RNA extraction, reverse transcription and qRT PCR analysis

In the FUSCC cohort, total RNA of tumor tissue was isolated from all 123 patients' samples using TRIzol reagent (15596-026, Invitrogen). A PrimeScript RT reagent kit (K1622, Thermo Scientific) was used to synthesize first-strand cDNA from total RNA. After that, we performed SYBR Green real-time PCR analysis using ABI 7900HT machine (Applied Biosystems, USA). We used β-actin mRNA expression level as a reference. NR1H3 and NR2F6 mRNA expression level was normalized to β-actin ²¹. The primers for qRT-PCR analysis were synthesized by Sangon (Shanghai, P.R. China), sequences of which are shown as following:

NR1H3: Forward primer: 5'-CCTTCAGAACCCACAGAGATCC-3'; Reverse primer: 5'-ACGCTGCATAGCTCGTTCC-3'

NR2F6: Forward primer: 5'-GAGCGGCAAGCATTACGGT-3'; Reverse primer: 5'-GGCAGGTGTAGCTGAGGTT-3'

β-actin: Forward primer: 5'-CATGTACGTTGCTATCCAGGC-3'; Reverse primer: 5'-CTCCTTAATGTCACGCACGAT-3'

Statistical analysis

Overall survival was calculated from the date of surgery to the date of death or last follow-up. Recurrence-free survival was defined as the time from the date of surgery to the date of first recurrence or last follow-up. Patients without events or death were recorded as censored at the time of last follow-up. Gender, tumor subtype (papillary vs non-papillary), histological grade (high vs low), family history, adjuvant chemotherapy and patient group (initial vs relapse) were considered binary variables and presented as proportions. Pathological stage, pT stage, pN stage, pM stage and smoking status were considered multiple categorical variables and shown as proportions. Age was deemed to be continuous variable and reported in the form of a median (range). Gene expression level from TCGA database, NR1H3 △Ct, NR2F6 △Ct were also deemed to be continuous variables. The higher the △Ct value, the lower the gene expression level. NR1H3 △Ct and NR2F6 △Ct were calculated using equations below ²²:

NR1H3 △Ct = Ct _(NR1H3)-Ct _(β-actin) (1)

NR2F6 △Ct = Ct _(NR2F6)-Ct _(β-actin) (2)

We used Kaplan-Meier method to construct survival curves and used log-rank tests to assess differences between groups. For continuous variables, we used X-tile 3.6.1 (Yale University, New Haven, CT, USA) to determine the best cutoff value for subgroups. Adjusted hazard ratio (HR) was calculated using Cox proportional hazards models. Univariate and multivariate Cox analysis for patients with muscle-invasive bladder cancer in TCGA database and FUSCC cohort were performed. Normally distributed data were compared using Student's t test. The Chi-square test was used to compare the distribution of categorical data between groups. All tests were two-tailed and P values less than 0.05 were considered to be statistically significant. SPSS software, version 22.0 (SPSS Inc., Chicago, IL, USA) were used for data analysis.

Results

Clinical characteristics of patients in TCGA and FUSCC database

In TCGA cohort, the median age of patients with muscle-invasive bladder cancer was 70.0 years old, ranging from 38 to 90 years old. In this database, 224 patients were male (72.7%) while 84 patients were female (27.3%). Information of TNM stage, tumor subtype, family history, smoking status and so on was recorded in Table 1. The median follow-up time was 14.5 months and 115 patients died during follow-up.

Table 1.

Patient demographic and clinical characteristics

Characteristics	TCGA cohort (N=308)		FUSCC cohort (N=123)		P value
	N	%	N	%
Age, median (range)	70.0	38 - 90	61.0	25 - 78	0.010a
Gender					0.016b
Male	224	72.7	103	83.7
Female	84	27.3	20	16.3
Tumor subtype
Papillary	93	30.2	-	-
Non-papillary	215	69.8	-	-
pT stage					<0.001b
T2	101	32.8	76	61.8
T3	159	51.6	34	27.6
T4	48	15.6	13	10.6
pN stage					<0.001c
N0	188	61.0	100	81.3
N1	35	11.4	10	8.1
N2	63	20.5	12	9.8
N3	6	1.9	1	0.8
Nx	16	5.2	0	0
pM stage
M0	146	47.4	123	100
M1	4	1.3	-	-
Mx	158	51.3	-	-
Pathological stage
Stage II	87	28.2	-	-
Stage III	115	37.3	-	-
Stage IV	106	34.4	-	-
Histological grade					0.020b
Low grade	14	4.5	13	10.6
High grade	294	95.5	110	89.4
Family history of cancer
Yes	107	34.7	-	-
No	201	65.3	-	-
Smoking status
Yes	212	68.8	-	-
No	86	27.9	-	-
Not mentioned	10	3.2	-	-
Adjuvant chemotherapy
Yes	-	-	46	37.4
No	-	-	77	62.6
Patient group
Initial case	-	-	90	73.2
Relapse case	-	-	33	26.8

a. Student's t test.

b. Chi-square test.

c. Chi-square test, N stage as a binary variable: No lymph node involvement versus otherwise.

In FUSCC database, the median age of 123 patients who underwent radical cystectomy was 61.0 years old, with a range from 25 to 78 years old. One hundred and three patients were men (83.7%) while twenty patients were women (16.3%). Records of TNM stage, adjuvant chemotherapy information, histological grade and patient catalog were also presented in Table 1, along with comparison of clinical characteristics between TCGA and FUSCC cohort. The median follow-up time in FUSCC cohort was 53.1 months and 42 people died.

NR1H3, NR2C1 and NR2F6 expression were independent prognostic factors of OS in TCGA cohort

To examine which nuclear receptor genes have prognostic value for predicting overall survival in muscle-invasive bladder cancer patients, we used TCGA cohort to do the filtration. Clinical pathological data and 48 nuclear receptor genes' expression profile were included in Cox proportion hazard ratio analysis. In univariate analysis, we found that age, pT stage, pN stage, pathological stage, ESRRA, NR1H3, NR1H4, NR2C1, NR2F6 and THRB expression level were significantly associated with overall survival in patients with bladder cancer in TCGA cohort (Table 2). Then, we performed reduced model of multivariate Cox analysis, only including variables with a statistical P value less than 0.1 in univariate analysis, to find independent prognostic factors for predicting OS. After adjustment, multivariate Cox regression for potential prognostic factors indicated that age (HR=1.043, 95% CI: 1.022 - 1.065), pT stage (HR=1.433, 95% CI: 1.061 - 1.936), pN stage (HR=2.007, 95% CI: 1.365 - 2.950), NR1H3 expression (HR=0.779, 95% CI: 0.634 - 0.957), NR2C1 expression (HR=0.673, 95% CI: 0.458 - 0.989) and NR2F6 expression (HR=0.750, 95% CI: 0.574 - 0.980) were independent prognostic factors for OS of patients with muscle-invasive bladder cancer in TCGA cohort (All P<0.05, shown in Table 2).

Table 2.

Univariate and multivariate Cox proportional hazards analysis of nuclear receptor gene expression and overall survival for patients with bladder cancer in TCGA cohort

Variable	Univariate analysis			Multivariate analysis
	HR	95% CI	P value	HR	95% CI	P value
Age	1.041	1.020 - 1.062	<0.001	1.043	1.022 - 1.065	<0.001
Gender	1.067	0.715 - 1.594	0.750
Tumor subtype	0.661	0.415 -1.054	0.082	0.933	0.571 - 1.527	0.784
pT stage	1.652	1.252 - 2.179	<0.001	1.433	1.061 - 1.936	0.019
pN stagea	2.437	1.679 - 3.537	<0.001	2.007	1.365 - 2.950	<0.001
pM stagea	1.443	0.995 - 2.092	0.053	1.290	0.882 - 1.887	0.189
Pathological stage	1.972	1.524 - 2.550	<0.001	1.221	0.748 - 1.991	0.424
Histological grade	21.658	0.146 - 3216.531	0.228
Family history	0.780	0.531 - 1.147	0.207
Smoking statusa	1.265	0.833 - 1.921	0.270
AR	0.984	0.912 - 1.062	0.679
ESR1	1.047	0.952 - 1.152	0.346
ESR2	0.853	0.717 - 1.015	0.074	0.860	0.715 - 1.035	0.110
ESRRA	0.709	0.519 - 0.968	0.030	0.906	0.627 - 1.310	0.600
ESRRB	0.875	0.735 - 1.040	0.131
ESRRG	0.945	0.867 - 1.029	0.191
HNF4A	0.989	0.884 - 1.105	0.840
HNF4G	0.939	0.873 - 1.009	0.088	0.988	0.912 - 1.070	0.760
NR0B1	1.057	0.967 - 1.156	0.220
NR0B2	0.966	0.778 - 1.200	0.758
NR1D1	1.123	0.933 - 1.352	0.218
NR1D2	0.883	0.677 - 1.152	0.360
NR1H2	0.723	0.472 - 1.109	0.137
NR1H3	0.784	0.641 - 0.958	0.018	0.779	0.634 - 0.957	0.018
NR1H4	0.926	0.860 - 0.997	0.041	0.962	0.879 - 1.053	0.405
NR1I2	0.956	0.825 - 1.108	0.549
NR1I3	0.928	0.790 - 1.090	0.362
NR2C1	0.522	0.365 - 0.745	<0.001	0.673	0.458 - 0.989	0.044
NR2C2	0.924	0.758 - 1.125	0.429
NR2E1	1.020	0.925 - 1.124	0.694
NR2E3	0.874	0.709 - 1.077	0.205
NR2F1	1.111	0.980 - 1.259	0.099	1.085	0.963 - 1.223	0.181
NR2F2	1.036	0.859 - 1.249	0.713
NR2F6	0.713	0.565 - 0.900	0.004	0.750	0.574 - 0.980	0.035
NR3C1	1.115	0.952 - 1.307	0.178
NR3C2	1.028	0.933 - 1.133	0.581
NR4A1	1.012	0.903 - 1.135	0.834
NR4A2	0.982	0.857 - 1.124	0.789
NR4A3	1.024	0.929 - 1.128	0.634
NR5A1	1.067	0.971 - 1.174	0.177
NR5A2	0.917	0.775 - 1.085	0.314
NR6A1	1.063	0.892 - 1.266	0.497
PGR	1.079	0.959 - 1.213	0.208
PPARA	1.013	0.811 - 1.264	0.911
PPARD	0.918	0.685 - 1.231	0.568
PPARG	0.928	0.861 - 1.001	0.053	1.013	0.891 - 1.151	0.847
RARA	0.994	0.741 - 1.335	0.970
RARB	1.022	0.916 - 1.141	0.694
RARG	0.855	0.714 - 1.024	0.089	0.897	0.739 - 1.089	0.273
RORA	1.033	0.898 - 1.190	0.647
RORB	1.014	0.892 - 1.153	0.826
RORC	0.946	0.881 - 1.016	0.126
RXRA	0.907	0.723 - 1.137	0.397
RXRB	0.879	0.564 - 1.371	0.571
RXRG	1.024	0.883 - 1.188	0.750
THRA	1.160	0.958 - 1.405	0.127
THRB	0.897	0.805 - 0.998	0.046	0.987	0.864 - 1.127	0.843
VDR	0.914	0.778 - 1.074	0.274

Binary variables: No lymph node involvement versus otherwise; Non-metastasis versus otherwise; Non-smoker versus otherwise.

For further study, we divided TCGA cohort into low expression and high expression groups by using X-tile to find out the best cutoff values for OS of NR1H3, NR2C1 and NR2F6 expression (Supplementary Figure 1). We also used the same cutoff value of OS as the cutoff of RFS. Thus, we found that low NR1H3 expression (log-rank test, P=0.0023), low NR2C1 expression (log-rank test, P=0.0005) and low NR2F6 expression (log-rank test, P<0.0001) were all significantly associated with poor prognosis for OS. However, by performing log-rank test for RFS, we found that only low NR1H3 expression (log-rank test, P=0.0076) and low NR2F6 expression (log-rank test, P=0.0395) were associated with poor prognosis for RFS. There's no obvious difference of RFS between high and low NR2C1 expression groups (log-rank test, P=0.0825) (Figure 1).

Figure 1.

Kaplan-Meier plots of survival in TCGA cohort are shown according to NR1H3, NR2C1 and NR2F6 expression. a, b, c Kaplan-Meier estimates of overall survival (OS) are shown according to NR1H3, NR2C1 and NR2F6 expression. d, e, f Kaplan-Meier estimates of recurrence free survival (RFS) are shown according to NR1H3, NR2C1 and NR2F6 expression.

NR1H3 expression was an independent prognostic factor of OS in FUSCC cohort

To validate the predictive value of NR1H3 and NR2F6 expression for OS and RFS in muscle-invasive bladder cancer patients, we tested tumor NR1H3 and NR2F6 expression levels in FUSCC cohort and performed statistical analysis. We used X-tile to find out the best cutoff values for OS of NR1H3 and NR2F6 expression level (△Ct) and divided patients into low expression and high expression groups (Supplementary Figure 2). Also we used the same cutoff value of OS as the cutoff of RFS. Firstly, we used Kaplan-Meier method to perform survival analysis and found that low NR1H3 expression was significantly associated with poor prognosis for both OS (log-rank test, P<0.0001) and RFS (log-rank test, P=0.0059). However, low NR2F6 expression level was relatively associated with poor prognosis for OS with a borderline significance (log-rank test, P=0.0623). There was no difference of RFS between high NR2F6 and low NR2F6 expression groups (Figure 2).

Figure 2.

Kaplan-Meier plots of survival in FUSCC cohort are shown according to NR1H3 and NR2F6 expression. a, b Kaplan-Meier estimates of overall survival (OS) are shown according to NR1H3 and NR2F6 expression. c, d Kaplan-Meier estimates of recurrence free survival (RFS) are shown according to NR1H3 and NR2F6 expression.

Furthermore, we performed univariate and multivariate Cox analysis to investigate if NR1H3 or NR2F6 gene expression was an independent prognostic factor of OS in FUSCC cohort. In univariate Cox model, patients who received adjuvant chemotherapy (HR=1.957, 95% CI: 1.067 - 3.587) and patients with low NR1H3 expression level (HR=1.312, 95% CI: 1.090 - 1.580) had significantly poor prognosis. However, in multivariate Cox analysis, only low NR1H3 expression level remained the significance (HR=1.295, 95% CI: 1.064 - 1.576) (Table 3). The same regression analysis was performed to determine whether gene expression was independent factor of RFS in muscle-invasive bladder cancer patients. It showed that only low NR1H3 expression level was significantly associated with poor prognosis of RFS in univariate Cox analysis (HR=1.259, 95% CI: 1.008 - 1.572). The statistical significance vanished in multivariate Cox regression model (Table 4).

Table 3.

Univariate and multivariate Cox proportional hazards analysis of overall survival for patients with bladder cancer in FUSCC cohort.

Variables	Univariate analysis			Multivariate analysis
	HR	95% CI	P value	HR	95% CI	P value
Age	1.028	0.990 - 1.067	0.146	1.029	0.990 - 1.070	0.147
Gender	0.474	0.169 - 1.331	0.156	0.499	0.177 - 1.407	0.189
pT stage	1.452	0.967 - 2.181	0.072	1.125	0.702 - 1.803	0.624
pN stagea	1.575	0.774 - 3.204	0.210	1.783	0.849 - 3.742	0.126
Histological grade	1.648	0.509 - 5.336	0.404	1.376	0.393 - 4.812	0.617
Adjuvant chemotherapy	1.957	1.067 - 3.587	0.030	1.754	0.955 - 3.222	0.070
Patient group	1.069	0.547 - 2.088	0.846	1.006	0.474 - 2.135	0.987
NR1H3 △Ct	1.312	1.090 - 1.580	0.004	1.295	1.064 - 1.576	0.010
NR2F6 △Ct	0.921	0.766 - 1.107	0.381	0.876	0.720 - 1.067	0.189

a. Binary variable: No lymph node involvement versus otherwise

Table 4.

Univariate and multivariate Cox proportional hazards analysis of recurrence-free survival for patients with bladder cancer in FUSCC cohort.

Variables	Univariate analysis			Multivariate analysis
	HR	95% CI	P value	HR	95% CI	P value
Age	1.002	0.962 - 1.044	0.917	0.996	0.951 - 1.044	0.877
Gender	0.157	0.021 - 1.160	0.070	0.144	0.020 - 1.068	0.058
pT stage	1.275	0.775- 2.098	0.339	0.966	0.527 - 1.771	0.910
pN stagea	1.423	0.576 - 3.511	0.445	1.912	0.756 - 4.833	0.171
Histological grade	1.720	0.408 - 7.253	0.460	1.133	0.250 - 5.141	0.871
Adjuvant chemotherapy	1.808	0.861 - 3.795	0.117	1.967	0.936 - 4.133	0.074
Patient group	0.631	0.256 - 1.557	0.318	0.643	0.242 - 1.706	0.375
NR1H3 △Ct	1.259	1.008 - 1.572	0.042	1.218	0.944 - 1.572	0.129
NR2F6 △Ct	0.924	0.775 - 1.101	0.375	0.909	0.750 - 1.103	0.333

a. Binary variable: No lymph node involvement versus otherwise

As low NR1H3 expression was found to be tightly correlated with overall survival in MIBC patients, we also evaluated if patients with low NR1H3 expression can benefit from adjuvant chemotherapy. However, it turned out that there was no significant benefit from adjuvant chemotherapy in patients with low NR1H3 expression (P=0.7627, Supplementary Figure 3).

Discussion

In our study, we were the first to demonstrate that NR superfamily gene expression was correlated with prognosis of patients with MIBC, in both OS and RFS. In TCGA filtrating cohort, we found that low NR1H3, NR2C1 and NR2F6 expressions were independent factors of poor OS, while only low NR1H3 and NR2F6 expressions had the ability to predict poor prognosis for RFS. Further, in our FUSCC validating cohort, we confirmed that low NR1H3 expression level was independent factor of poor OS and it had the ability to predict poor RFS. However, NR2F6 failed to maintain statistical significance in predicting prognosis of both OS and RFS.

Nuclear receptors are one of the largest classes of transcription factors and regulate many cellular functions through manipulation of gene expression ²³. They play important roles in many physiological and pathological processes ²⁴. Liver X receptor alpha (LXRα) is a nuclear receptor encoded by NR1H3 gene, and forms a subfamily of the NR superfamily along with liver X receptor β. Previously understood, LXRα was a key regulator of macrophage function ²⁵ and lipid homeostasis ^26-28. But recent researches revealed that LXRα was associated with various cancer types. Savic D et al found that activation of LXRα could inhibit colorectal cancer cell proliferation and decrease glutathione levels, consistent with increased cellular oxidative stress ^{29, 30}. In breast cancer, some researchers declared that NR1H3 was likely to be an onco-suppressor gene ^{31, 32}. Similar results were also achieved in prostate cancer ³³, gallbladder cancer ³⁴, ovarian carcinoma ³⁵ and so on. All the publications mentioned above illustrated that NR1H3 might be an onco-suppressor gene in various cancer types, which was in line with the unfavorable prognosis in MIBC patients with low NR1H3 expression level found in our investigation.

LXRs were closely related to cholesterol metabolism in cancer. LXR activation was absolutely a hot-spot in cancer research and was a potential bona fide therapeutic strategy. As reported, tumor-related proliferation status was associated with enhanced cholesterol requirement which is satisfied not only by up regulating de novo biosynthesis but also by increasing cholesterol uptake via LDLR. Furthermore, in cancer cells, despite the intracellular cholesterol abundance, LXR transcriptome was also down regulated ³⁶. In most studies, LXR activation promoted cholesterol removal from cells by increasing expression of membrane transporters ABCA1, ABCG5, ABCG8 and ABCG1 ³⁷, and inhibited uptake via LDLR by inducing IDOL-mediated LDLR degradation ³⁸, to inhibit cancer cell proliferation. Besides, some groups reported that activation of LXR can play an anti-tumor role through other mechanisms, like altering microenvironment by activating marcophages ³⁹, disrupting key growth pathways by suppressing β-catenin signaling ⁴⁰ or activating apoptotic processes through caspase pathway ⁴¹. Overall, LXRs play important role in tumor suppression through various mechanisms. Our study was the first to link LXRα to prognosis of MIBC and casted a new light in therapeutic potential of LXRα in MIBC.

Although adjuvant chemotherapy in MIBC remains controversial and many powerful guidelines, like EAU ⁴² and NCCN ⁴³ guidelines, are very restrictive with regard to its recommendation, adjuvant chemotherapy in a multimodal strategy will be probably used frequently in the future in bladder cancer treatment due to recent findings ⁴⁴. In our study, we found that MIBC patients with low NR1H3 expression not only had worse prognosis, but showed resistance to adjuvant chemotherapy as well. It indicated that maybe we should utilize other managing strategies instead of chemotherapy in those patients and a target therapy for those patients was in an urgent need. Indeed, subsequent well-designed large-cohort studies should be performed to confirm our finding. Response to neoadjuvant chemotherapy should also be evaluated in the future if possible.

A major strength of this study is that our data was got from two large populations with a relatively long follow-up, but there are certainly some limitations. Firstly, the limitations include the retrospective nature of this study and the selected bias of patients in FUSCC cohort, which from a single cancer center. Secondly, the prognosis of MIBC is affected by many factors, such as tumor stage, surgical performance, response to adjuvant chemotherapy and genetic background, so a single biological biomarker is not enough. Thirdly, qRT-PCR technique only measures the mRNA expression level of certain gene, not referring to protein level. So in the future, we will expand our cohort to multicenter cohorts for further validation, and perform immunohistochemistry or western blotting to measure protein expression level in tumor tissues. Furthermore, research in cell lines and deep molecular mechanism should be explored to make a step forward to potential application.

In conclusion, low NR1H3 expression level is an independent prognostic factor of poor OS, and can also predict worse RFS in MIBC patients. Our “TCGA filtrating and local database validating” model can help reveal more prognostic biomarkers of MIBC and cast a new light in understanding certain gene function in MIBC.

Supplementary Material

Supplementary figures and tables.

Abbreviations

NR

nuclear receptor

MIBC

muscle-invasive bladder cancer

TCGA

The Cancer Genome Atlas

OS

overall survival

RFS

recurrence free survival

FUSCC

Fudan University Shanghai Cancer Center

HR

hazard ratio

LXRα

Liver X receptor alpha.