Cell‐free lncRNA expression signatures in urine serve as novel non‐invasive biomarkers for diagnosis and recurrence prediction of bladder cancer

Abstract Cell‐free long non‐coding RNAs (lncRNAs) are stably present in urine and can serve as non‐invasive biomarkers for cancer. We aimed to identify signatures of lncRNAs in urine for diagnosis and prognosis of bladder cancer (BC). Screening of lncRNAs by microarray analysis was performed using urine samples of 10 BC patients and 10 controls. Expressions of candidate lncRNAs were evaluated in the training and validation set including 230 BC patients and 230 controls by quantitative reverse transcription polymerase chain reaction (qRT‐PCR). A two‐lncRNA panel (uc004cox.4 and GAS5) was constructed and provided high diagnostic accuracy of BC with an area under the curve (AUC) of 0.885 (95% CI, 0.836‐0.924). The AUCs of the lncRNA panel for Ta, T1 and T2‐T4 were 0.843, 0.867 and 0.923, respectively, significantly higher than those of urine cytology (all P < .05). Kaplan‐Meier analysis revealed that higher level of uc004cox.4 was associated with poor recurrence‐free survival (RFS) of non‐muscle invasive BC (NMIBC) (P = .008). Additionally, Cox regression analysis indicated that uc004cox.4 was an independent prognostic factor for RFS of NMIBC (P = .018). Taken together, our findings indicated that urinary lncRNA signatures possessed potential clinical value for BC diagnosis. Moreover, uc004cox.4 could provide prognostic information for NMIBC.

widely used non-invasive approach, has high specificity for detection and monitoring of BC. However, this test has limited diagnostic sensitivity, which is necessary to rule out cancer. 2,3 The histological evaluation of cystoscopy-guided biopsy can provide high diagnostic accuracy, but it is invasive and often inconvenient, making it impractical for mass screening of BC in people without signs or symptoms of disease. These characteristics have prompted a search of novel biomarkers, particularly more reliable non-invasive markers for screening, initial evaluation and follow-up of BC.
As a new class of non-coding RNAs, long non-coding RNAs (lncRNAs) are more than 200 nucleotides in length with no or lower protein-coding potential. Moreover, lncRNAs have been continually reported to be involved in tumorigenesis and progress. [4][5][6] Dysregulated lncRNA expression has been found in many human malignancies. 7,8 Extracellular lncRNAs can circulate in body fluids and can be detected and strongly resist RNases. Several studies have investigated the potential of using lncRNAs as biomarkers for cancer diagnosis, and promising results have been obtained. In those studies, some lncRNAs were readily detected and remained stable in whole blood, plasma, gastric juice and saliva, which suggested that lncRNAs could be good candidates for tumour biomarkers. 9 Researchers have also found that lncRNAs in urine are detectable and may serve as new predictors of outcome in kidneytransplant patients with acute rejection. 10 However, detection of lncRNAs in urine of BC patients has not been investigated yet. Differentially expressed lncRNAs have already been identified between BC tissues and normal tissues. 11,12 The dysregulated expression of lncRNAs may imply their potential application as biomarkers for BC screening.
In this study, we firstly performed a microarray analysis to identify the genome-wide different expression profiles of lncRNAs between BC and control urine specimens. Then we selected 26 candidate lncRNAs and evaluated their expression levels by quantitative reverse transcription polymerase chain reaction (qRT-PCR) in the training set and validation set. Finally, we established a two-lncRNA panel with high diagnostic accuracy for BC. In addition, we further investigated the potential relationship between urinary lncRNA levels and recurrence of BC.

| Study design
A total of 240 BC patients and 240 controls were recruited from Qilu Hospital, Shandong University, between February 2008 and March 2011. All the participants were randomly divided into three phases. In the discovery set, pooled urine samples from 10 BC patients and 10 healthy controls were, respectively, subjected to microarray assay to initially identify dysregulated lncRNAs in BC. In the training set, the expression levels of candidate lncRNAs were firstly investigated in an independent cohort containing 36 BC patients and 36 controls. LncRNAs that were differentially expressed between BC group and control group were further examined in an additional test consisting of 168 participants including 84 BC patients and 84 controls. These above-mentioned 240 urine samples were used as the training set to construct the diagnostic panel based on the logistic regression model to distinguish BC patients from controls. In the validation set, the parameters of the logistic model from the training set were applied to an independent cohort consisting of 110 BC patients and 110 controls to confirm the diagnostic performance of the established lncRNA panel. Additionally, urine cytology was also performed on the same cohort to further assess the diagnostic value of the constructed lncRNA panel. Meanwhile, BC patients in the validation set were followed up at intervals of 3 months during the first 2 years and 6 months up to the fifth year.
The date of the latest retrieved record was April 2016.

| Urine sample preparation
Freshly voided urine samples were collected on the day before treatment. Cells and debris were removed from each urine sample by centrifugation at 15009 g for 10 minutes and 13 8009 g for 15 minutes at 4°C, and supernatant was stored at À80°C prior to further analysis. In addition, 15-mL mid-stream urine of BC patients was centrifuged at 13009 g for 10 minutes, and sediments were immediately processed for cytologic examination by two independent cytopathologists.   Specificity of qRT-PCR product was measured by melting curve analysis. GAPDH was selected as the housekeeping gene, and the relative expressions of lncRNAs were calculated using the 2 ÀMMct method. (sensitivity + specificity À 1) was used to set the optimal cut-off point. 13 Survival curves were established by Kaplan-Meier method, and differences were assessed using log-rank statistics. Univariate and multivariate Cox analyses were employed to identify independent prognostic factors of BC. A P-value < .05 was considered as statistically significant.       16 In our previous study, we have performed systematic analysis and identified seven differently expressed urinary miRNAs in BC patients. 17 However, little is known about the diagnostic utility of urinary lncRNAs in BC.

| Correlation between two lncRNAs and clinicopathological characteristics
Abnormal expressions of LncRNAs have been suggested as a major cause of oncogenesis, and various types of cancers can be distinguished because of their unique altered lncRNA signatures. [18][19][20] Previous studies have confirmed that the expressions of lncRNAs in human body fluids, including serum/plasma and urine, are abundant and quite stable. 9 Recently, the rapid development of microarray assays makes it possible to efficiently discover aberrantly expressed lncRNAs in cancer. Our laboratory has previously shown that lncRNAs can be detected in serum of BC patients, and serum lncRNAs (MEG3, SNHG16 and MALAT1) have potential values in BC diagnosis. 21 As urine samples are in direct contact with tumour cells in BC and can be collected easily and non-invasively, it is an ideal source for the detection of BC that needs constant monitoring. In the present study, the initial screening of urinary lncRNAs was performed by microarray in BC patients and controls, and potential differently expressed lncRNAs were firstly selected for further

CONFLI CTS OF INTEREST
The authors declare that no potential conflicts of interest were disclosed.