A panel consisting of three novel circulating lncRNAs, is it a predictive tool for gastric cancer?

Abstract Early detection is vital for prolonging 5‐year survival for patients with gastric cancer (GC). Numerous studies indicate that circulating long non‐coding RNAs (lncRNAs) can be used to diagnose malignant tumours. This study aimed to investigate the capacity of novel lncRNAs for diagnosing GC. A lncRNA microarray assay was used to screen differentially expressed lncRNAs between plasma of patients with GC and healthy controls. Plasma samples from 100 patients with healthy controls were used to construct a multiple‐gene panel. An additional 50 pairs of GC patients with healthy controls were used to evaluate the diagnostic accuracy of the panel. Expression levels of lncRNAs were quantified through real‐time polymerase chain reaction. The receiver operating characteristic curve and area under curve (AUC) were used to estimate the diagnostic capacity. We identified three lncRNAs, CTC‐501O10.1, AC100830.4 and RP11‐210K20.5 that were up‐regulated in the plasma of GC patients with AUCs 0.724, 0.730 and 0.737, respectively (P < .01). Based on the logistic regression model, the combined AUC of the three lncRNAs was 0.764. The AUC of the panel was 0.700 in the validation cohort. These findings indicate that plasma lncRNAs can serve as potential biomarkers for detection of GC.

in multiple biological processes. 7 NcRNAs have been detected in many body fluids such as plasma, gastric juice, urine, and saliva. 8 Researchers have also demonstrated that extracellular ncRNAs, including microRNAs (miRNAs) and long non-coding RNAs (lncRNAs), can be used as diagnostic tools for conditions including heart failure, 9 infectious diseases, 10 type 2 diabetes 11 and malignant tumours such as GC, 12,13 lung cancer 14 and hepatocellular cancer. 15 Examples include lncRNA SPRY4-IT1 for diagnosis of hepatocellular carcinoma; 16 miR-22-3p, miR-642b-3p and miR-885-5p for diagnosis of pancreatic cancer; 17 and lncRNA MACC1 for diagnosis of non-smallcell lung cancer. 18 Although many publications have demonstrated high accuracies and efficiencies for ncRNAs-based diagnostic markers, the method is still controversial and far from clinical use due to the lack of largescale validation and consensus among researchers. 19 In this study, with the aim to find novel biomarkers for GC, we detected dysregulated lncRNAs in patients with GC vs healthy controls based on the results of lncRNA microarray and investigated the diagnostic potential of plasma-circulating lncRNAs.

| Study design, patients and plasma sample collections
A multiphase, case-control study was conducted, and the whole study was divided into four phases: the discovery phase, testing phase, training phase and validation phase. In the discovery phase, lncRNA profiling analyses were conducted between plasma from four GC patients with age-and gender-matched healthy controls. Clinicopathological data were obtained from the medical records.
Healthy controls consisted of patients with benign diseases such as hernias, haemorrhoids or varicose veins who had no evidence of any stomach disease or other malignancy. All clinical parameters were estimated according to the 8th AJCC/TNM staging system. Informed consent was obtained from all the participants, and this study was approved by the Research Ethics Committee of China Medical University.
Blood sampling was standardized. All blood samples were collected together with other blood-based tests to minimize extra injury to the patient, and all were obtained after admission, under fasting conditions and before any treatment including operation or chemotherapy. Two millilitres of whole peripheral venous blood was collected from each enrolled individual and held in a purple-top EDTA tube. Plasma samples were separated within 4 hours after collection following a two-step centrifugation protocol (3000 g for 10 minutes at 4°C, 12 000 g for 10 minutes at 4°C) to thoroughly remove cellular nucleic acids, transferred to RNase/DNase-free tubes and stored at À80°C until total RNA extraction. were cultured in Dulbecco's modified Eagle's medium (Invitrogen). All of the media contained 10% foetal bovine serum, and cells were kept in flasks at 37°C in a humidified atmosphere of 5% CO 2 . Every 2.0 9 10 6 cells were seeded in 10 cm dishes with 5 mL culture media, and 48 hours after initial seeding, the culture media were collected and processed exactly as that described for plasma collection.

| RNA extraction from plasma, cell lines and culture media
Total RNA was extracted from cultured cells using TRIzol reagent (Ambion) according to the manufacturer's protocol. Total RNA was extracted from plasma and culture media using the mirVana PARIS Kit (Ambion). 400 lL plasma or culture media were thoroughly mixed with an equal volume of 2 9 denaturing solution and incubated on ice for 5 minutes 800 lL acid-phenol:chloroform was added and thoroughly mixed via vortex, then centrifuged at 12 000 9 g for 10 minutes at room temperature. The upper aqueous phase was transferred to a fresh RNase-free tube. Washing steps were performed according to the manufacturer's protocol, and the RNA was finally eluted with 40 lL RNase-free water pre-heated to 95°C.
Plasma samples for evaluating the internal control were randomly selected and processed under identical conditions.

| Reverse transcription (RT) and real-time PCR
Any residual DNA in the RNA sample was eliminated, and complementary DNA was synthesized using a PrimeScriptâ RT reagent kit (TaKaRa) according to the manufacturer's protocol. The relative expression levels of lncRNAs in all samples were determined using SYBR Premix Ex Taq II (TaKaRa) on a Light Cycler 480 II Real-time PCR system (Roche). The reactions were incubated at 95°C for 30 seconds and then underwent 45 cycles of 95°C for 5 seconds and 60°C for 30 seconds. Samples were analysed in triplicate, and the products were confirmed by melting curve analysis following each reaction. Total cycles required for the SYBR signal to cross the threshold were identified as cycle threshold (Ct). Following the evaluation of the internal controls discussed in this study, the relative expression of each lncRNA in all samples was calculated using the Ct method normalized to b-actin. All the human gene-specific primers for real-time PCR are shown in Table S1. Samples with a Ct > 40 were considered negative, as previously described. 20

| General characteristics of the enrolled individuals
The median age of the patients with GC was 61 years (range, 29-83 years), and the median age of the healthy controls was 60 years (range, 37-89 years). A total of 108 patients with GC were male, and 42 were female. In the healthy controls group, 97 individuals were male, and 53 were female. There were no significant differences in age or gender between the two groups. Among the patients with GC, 26 of the 150 patients were diagnosed at an incurable stage and treated with chemotherapy or palliative operation.

| Evaluation of candidate endogenous controls
There is currently no consensus on the use of endogenous controls for the relative quantification of circulating RNAs. Thus, we tested  Table S2).

| Discovery phase: Screening of candidate lncRNAs by microarray
The Human lncRNA V6 Microarray (Agilent) was used to screen for differential expression of lncRNAs in plasma samples from 4 patients with GC and 4 healthy controls. Fold change (GC vs Healthy) and P value were calculated from the normalized expression. Clustering analysis and heatmaps were used to show differentially expressed lncRNAs between the plasma samples from patients with GC and healthy controls (P < .05). We found that 267 lncRNAs were differentially expressed between these two groups. Among them, 256 lncRNAs were up-regulated in the plasma of GC patients compared with healthy controls, and 11 lncRNAs were down-regulated (Figure 1). The detailed lncRNA microarray data are shown in Table S3.

| Correlation between the markers
To investigate the similarity in expression levels among the three lncRNAs, Pearson correlation coefficients were calculated for each marker. As presented in Table 1, the expression level of AC100830.4 showed a mutual relation with CTC-501O10.1 with a coefficient of 0.463 (P < .01) and also mutually related with RP11-210K20.5 with a coefficient of 0.268(P < .01).

| The correlation between expression level of lncRNAs and patient characteristics
We examined the correlations between the expression levels of circulating CTC-501O10.1, AC100830.4 and RP11-210K20.5 and patient clinicopathological parameters. No significant associations were observed between the lncRNAs and clinical parameters including age, gender, tumour location or tumour stage (Table S4).

| DISCUSSION
In recent years, numerous studies have demonstrated that ncRNAs (lncRNAs, miRNAs and circular RNAs) are differentially expressed in cancer tissues and bodily fluids, that they often play regulatory roles in biological procedures, and that they have highly specific expression among different tissues. 23,24 It has been demonstrated that miRNAs and lncRNAs are both stable in the plasma, even in extreme conditions and also resistant to RNase A. [25][26][27] These features make it plausible that ncRNAs could be used as diagnostic biomarkers for many types of cancer. 28 For example, PCA3 is a significantly upregulated lncRNA in prostate cancer tissue and metastatic sites, and, based on a real-time PCR assay, PCA3 is dysregulated in the urine of patients with prostate cancer. It has thus been reported to have clinical use as a diagnostic biomarker. 29,30 F I G U R E 3 The relative expression levels of lncRNAs in plasma from patients with gastric cancer and healthy controls. A, The expression level of lncRNA CTC-501O10.1 in plasma was higher in the gastric cancer group (P < .01). B, The expression level of lncRNA AC100830.4 in plasma was higher in the gastric cancer group (P < .01). C, The expression level of lncRNA RP11-210K20.5 in plasma was higher in the gastric cancer group (P < .01) The AUC for each lncRNA in the training phase. The AUC for lncRNA CTC-501O10.1, lncRNA AC100830.4 and lncRNA RP11-210K20.5 was 0.724, 0.730 and 0.737, respectively. The combined AUC was 0.764 (P < .01) F I G U R E 5 The AUC for the lncRNA panel in the validation phase. The AUC for the lncRNA panel was 0.700 (P < .01) in an independent cohort consisting of 50 patients with gastric cancer and 50 healthy controls Because there is still no consensus on an endogenous reference for use with circulating RNAs, we evaluated many common choices and chose b-actin for our endogenous control, consistent with other groups' works. 20,26 In the current study, we performed a microarray to screen for deregulated lncRNAs in plasma from patients with GC and healthy controls and identified three novel lncRNAs, CTC-501O10.1, AC100830.4 and RP11-210K20.5, which were differentially expressed between these two groups with AUCs of 0.724, 0.730 and 0.737, respectively. The diagnostic potential seems unsatisfactory, yet it is still higher than that of CEA and CA199, as previously reported. 31 The AUC of the single lncRNAs is similar to some previous studies. 32 Due to the limited accuracy of any single index, several studies use multiple-biomarker panels to elevate diagnostic accuracy, such as the miRNA panel established using miR-16-2, miR-195, miR-2861 and miR-497 for cervical cancer detection. 33 Likewise, a lncRNA panel consisting of lnc00152, CFLAR-AS1 and POU3F3 was reportedly used for diagnosing oesophageal squamous cell carcinoma. 34 The combination of three or more biomarkers can significantly elevate the diagnostic potential of blood-based panels. For GC, some panels have been established, but none are widely used in clinical practice. 20,35 With the aim to increase diagnostic capacity, we combined three lncRNAs using a logistic regression model, resulting in an AUC of 0.764 in the training phase. The panel could provide a moderate diagnostic accuracy, with an AUC of 0.700 in the validation phase. Using the combination of these three lncRNAs, the diagnostic sensitivity was 0.99, which was meaningful in the clinic. To investigate why the panel could not provide a more optimal diagnostic potential, we analysed the correlation between the expression levels of lncRNAs and found that they were mutually related with each other (P < .05). The limited accuracy of the panel could be explained by the correlation in expression between the markers. 36 The specific origins of circulating RNAs are largely unknown. It has been proposed that cancer tissues can secrete lncRNAs into the circulatory system and that they can be protected by nanoscale vesicles called exosomes. Thus, the correlation in expression may be due to the lncRNAs in bodily fluids arising from the same origin. 8,37 These circulating lncRNAs can be used as biomarkers and could even transfer drug-resistance signals to other sites. 38 CTC-501O10.1 is located on chromosome 17 with a length of 549nt, AC100830.4 is located on Chr21 (q22, 12) with a length of 554nt, and RP11-210K20.5 is located on Chr18 (q12, 1) with a length of 666nt. To the best of our knowledge, none of them have been investigated before. In this study, we found that CTC-501O10.1, AC100830.4, and RP11-210K20.5 could enter into the cell culture medium at a measurable level. And this provided evidence that lncRNAs were released into circulation and the dysregulation of the lncRNAs in circulation could be used as biomarkers for GC.
In our study, we found 3 novel dysregulated lncRNAs in plasma from patients with GC. Using these three lncRNAs, we could distinguish GC patients from healthy controls with better accuracy than some other biomarkers. It is not pessimistic when there was no significant association between the plasma lncRNAs and clinicopathological parameters. Many researchers reported that the lncRNAs in plasma showed no significant correlation with the clinic characteristics, such as some lncRNAs in clear cell renal cell carcinoma, 39 in hepatocellular carcinoma 40 and in gastric cancer. 20,41 The traditional biomarkers are the most common serum markers used for GC detection. Even in some studies, the traditional biomarkers such as CEA, CA19-9 and CA724 showed no significant correlation with the clinic parameters. 42 Although in the current study, the expression level of the lncRNAs is not correlated with any clinic parameters, but the diagnostic ability of the lncRNAs should not be neglected. They could provide better diagnostic ability than the traditional biomarkers, even better than the combination of CEA, CA19-9 and CA724. 43 Therefore, the circulation lncRNAs are potential diagnostic tools.
Interestingly, even the same lncRNA detected in different cohorts of patients may present different expression pattern. In our study, H19 and HOTAIR showed no significant differences between the plasma of patients with GC and healthy controls, even though it has been previously reported to be elevated in GC plasma. 20,21,44 The current study is not the only one argued these discrepancies. H19 was once reported to be up-regulated in plasma from patients with breast cancer; 45 however at almost the meantime, another research reported that H19 was not altered in another cohort of patients with breast cancer. 46 As to HOTAIR, it was also reported not altered in plasma from patients with GC in Otsuji's research, which was consistent with the current study. 41 The inconsistency was not only seen in H19 and HOTAIR, but also in lncRNA MALAT1 20

CONF LICT OF I NTEREST
The authors confirm that there are no conflict interests.