Background
Gastric cancer is one of the leading causes of cancer mortality in the world, and it has a particularly high incidence in Asian countries including China. Despite the declining incidence of gastric cancer, there are still more than 1 million cases newly diagnosed and 850,000 deaths globally each year [
1]. The mortality rate remains high due to late presentation, since early stage of gastric cancer is either asymptomatic or presents with non-specific symptoms. For advanced diseases, the overall 5-year survival following surgical resection is 30–40 %, as compared to 70–90 % in early stage [
2]. To date, endoscopic and pathological examinations are the common techniques for cancer diagnosis. Despite their sensitivity and specificity to visualize and locate the site of malignancy, these approaches are invasive in nature which impede patients from routine screening for gastric cancer. Conventional serum tumour markers, namely, carcinoembryonic antigen (CEA) and carbohydrate antigen 19.9, however, are not tissue specific and expressed in most of the gastrointestinal cancers. Our group has shown that serum migration-inhibitory factor (MIF) had a better diagnostic value than CEA, however, combined serum MIF and CEA would have a better 5-year survival prognosis than individual marker [
3]. Serum E-cadherin level was found to be positively correlated with disease recurrence, and this could be a better marker than CEA in predicting disease recurrence [
4,
5]. Therefore, there is an urge to look for a non-invasive diagnostic biomarker that could be easily detected in serum, plasma or urine for early diagnosis for gastric cancer to greatly improve the mortality rate.
Many recent studies revealed that microRNAs (miRNAs) were actively involved in development, differentiation, inflammation, and pathogenesis of various malignancies. They belong to a class of small non-coding RNAs about 19–25 nucleotides in length, and are able to bind complementary sequences in 3′-untranslated regions (3′-UTR) of various target mRNAs to promote degradation or translational repression [
6]. Studies have shown that over 30 % of human genes are regulated by miRNAs, in which a single miRNA controls over hundreds of RNA [
7,
8]. MiRNAs also function as tumour suppressors or oncogenes in various types of cancer. Because miRNAs are very specific for different types of tissues and even for types of cells within those tissues, they are potentially useful for diagnosis, predicting clinical outcome or therapeutic targets in cancer patients. By comparing the miRNA expression profiles in tumour versus adjacent non-tumour tissues, distinct patterns of up- or down-regulation of miRNAs could be found in different types of cancers [
9‐
12]. These cancer specific miRNAs expression patterns could be used for diagnosis or monitor the efficiency of follow-up treatment.
With the development of microarray platforms, researchers could easily differentiate the oncogenic or tumour suppressive miRNAs in various human malignancies. We have previously developed a robust protocol to profile miRNA expression in the circulation of colorectal cancer patients [
12]. With the use of PCR-based miRNA array, we could profile the miRNA expression in plasma, as well as in paired tumour and adjacent non-tumour tissues. There are studies reporting that miRNA is more stable than mRNA in the circulation and yet specific miRNA is originate from tumour site [
13,
14]. In this study, we compare the miRNA expression profile in plasma of gastric cancer patients with healthy controls. We further validated the levels of miRNAs in three independent sets of gastric cancer patients and associated with tumour progression. We proposed that circulating miRNA signature could act as a potential molecular marker for diagnosis and therapeutic targets for gastric cancer.
Discussion
Gastroscopy remains the gold standard for detection of gastric cancer, which usually has a high false negative rate, due to the lack of noticeable symptoms of gastric cancer. Therefore, patients usually diagnosed with advanced cancer when underwent these examinations. Hence, the development of gastric cancer screening tool would effectively reduce the overall mortality. To date, there is no reliable non-invasive blood based classifier for the detection of gastric cancer.
Several studies reported that miRNA pattern is distinct in different cancer types. MiR-21 is one of the miRNAs that is known to play a role in various types of carcinomas [
15,
16], including gastric carcinoma [
17]. Studies have shown that miR-21 is known to increase in gastric cancer, and associated with tumor cell growth and invasion [
18,
19]. However, it is not suitable to be a tissue-specific diagnostic marker of gastric cancer, due to its high expression in most of the cancerous tissues. This prompted us to discover the feasibility of other miRNAs in plasma for diagnosis.
In this study, we used a miRNA microarray platform to screen the differentially expressed miRNAs in plasma of gastric cancer patients, and several upregulated miRNAs (miR-18a, miR-140-5p, 199a-3p, miR-627, miR-629 and mir-652) were selected for validation. We then selected the three miRNAs (miR-627, miR-629 and miR-652) with highest ROC curve and performed a large validation study on the expression levels. In both training (
n = 50) and validation sets (
n = 58), these 3 miRNAs were highly expressed in gastric cancer cases when compared with healthy controls. The algorithm we used to optimize the gastric cancer classifier (by multiplication of two highest expression levels among 3 miRNAs), which can be used as a biomarker to discriminate gastric cancer patients with healthy controls. The classifier was validated again in TS and VS, and AUC were determined to be 0.902 and 0.969 respectively with improved sensitivity and specificity when compared to individual miRNA, implicating that this classifier is robust biomarker for gastric cancer. We further validated the classifier in a RS (mixture of gastric cancer and healthy controls samples), and could discriminate gastric cancer patients from healthy controls with AUC calculated to be 0.942. There is no association between the expression levels of the classifier (or individual miRNAs) and TNM stage in plasma of gastric cancer patients. It is widely accepted that miRNA is being released into the circulation from the primary gastric cancer site, therefore it is expected to detect a similar trend of elevation of miRNA expression in gastric cancer tissues. A consistent higher expression levels of miR-627, miR-629 and miR-652 were detected in tumor tissues than adjacent non-tumor counterparts (
n = 36) (Fig.
3b), which echoed with the findings in plasma of gastric cancer patients.
To the best of our knowledge, this is the first study reported that expression of plasma miR-627 is more than 10-fold higher in gastric cancer patients than in healthy controls. It is worthwhile to investigate the specificity of this miRNA by examining its expression in other cancerous tissues. High level of miR-629* was detected in the plasma of xenograft mice which is originated from human prostate cancer xenograft [
14]. This study provide evidence that tumor-derived miRNA could be detected in the circulation. Several studies reported that mir-652 is associated with tumor progression of osteosarcoma [
20] and hepatocellular carcinoma [
21]. Another group reported that deregulation of miR-652 was identified as a biomarker for schizophrenia [
22]. However, in primary squamous cell lung carcinoma tissues, the levels of miR-652 were downregulated as compared with normal counterparts [
23]. These studies revealed that the function of miRNA is different in various carcinomas and yet further study mechanistic on the oncogenic pathways may help to understand the progression of the disease.
Conclusion
In conclusion, our data demonstrated that circulating miRNAs could be a sensitive biomarker for diagnosis of gastric cancer. This study revealed a new algorithm to discriminate gastric cancer cases and non-gastric cancer cases, the classifier (by combining 3 miRNAs) illustrated a promising discrimination of gastric cancer cases in different validation sets, which is far more sensitive than conventional tumor marker (eg CEA and CA 19.9). Further studies are warranted to examine the expression levels of selected miRNAs after surgery to verify the usefulness of the classifier to predict recurrence or therapeutic strategy during follow-up. This study may open up new opportunity to develop an economical non-invasive diagnostic tool for early detection of gastric cancer to reduce mortality.
Methods
Clinical samples
Pre-resection plasma and respective paired primary tumor/adjacent non-tumor tissue samples were obtained at Queen Mary Hospital between 2006 to 2010, while healthy volunteers were recruited at the Hong Kong Sanatorium and Hospital from 2009–2010. To discover and validate a specific biomarker for gastric cancer, this study consists of 3 independent sets of cases and controls. The training set (TS) contained 50 gastric cancer cases from 2006–2007, validation set (VS) contain 58 gastric cancer cases from 2008–2010 and random set contain 15 gastric cancer cases from 2008–2010, each independent set includes stage I to IV patients.
Plasma and tissue samples are stored at the frozen tissue bank of the Department. The collection and storage of such tissue samples have been approved by the Institutional Review Board. Informed consent has been obtained from each patient. Characteristics of patients such as gender, age, co-morbidity, presenting symptoms and signs, operative findings and staging will be prospectively collected into our standard electronic database. Clinical characteristics of gastric cancer patients were summarized in Table
3. These gastric plasma/tissues will be subjected to miRNA profiling and correlated with the clinicopathological factors in gastric cancer patients.
Table 3
Clinical characteristics of gastric cancer patient
Age [years; mean (SD)] | 62.8 (18.3) | 67.2 (16.4) | 67.1 (13.4) | 0.366 | 64.40 (16.88) |
Sex | | | | 0.499 | |
Men | 31 | 33 | 11 | | 75/123 (61 %) |
Women | 19 | 25 | 4 | | 48/123 (39 %) |
Depth of invasion (T) | | | | 0.932 | |
T1 | 7 | 4 | 2 | | 13/123 (11 %) |
T2 | 12 | 13 | 3 | | 28/123 (23 %) |
T3 | 18 | 25 | 6 | | 49/123 (40 %) |
T4 | 13 | 16 | 4 | | 33/123 (27 %) |
Lymph-node metastasis | | | | 0.361 | |
(N) | 13 | 10 | 4 | | 27/123 (22 %) |
N0 | 10 | 8 | 2 | | 20/123 (16 %) |
N1 | 26 | 40 | 8 | | 74/123 (60 %) |
N2 | 1 | 0 | 1 | | 2/123 (2 %) |
N3 | | | | | |
Distant metastasis | | | | 0.554 | |
No | 37 | 46 | 13 | | 96/123 (78 %) |
Yes | 13 | 12 | 2 | | 27/123 (22 %) |
Stage | | | | 0.088 | |
I | 8 | 8 | 4 | | 20/123 (16 %) |
II | 11 | 2 | 2 | | 15/123 (12 %) |
III | 15 | 26 | 5 | | 46/123 (38 %) |
IV | 16 | 22 | 4 | | 42/123 (34 %) |
Total RNA containing small RNA was extracted from 500 μl of plasma using Trizol LS reagent (Invitrogen, Carlsbad, California, USA) and miRNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol. After phase separation by chloroform addition and centrifugation, 1.5 volumes of 100 % ethanol was added to the aqueous phase and the mixture was loaded into miRNeasy column (Qiagen). DNase treatment (Qiagen) was carried out to remove any contaminating DNA. The final elution volume was 30 μl. The concentrations of all RNA samples were quantified by NanoDrop 1000 (Nanodrop, Wilmington, Delaware, USA).
MiRNA microarray
In the screening phase, we profiled 10 age- and sex- matched individuals (5 gastric cancer patients vs 5 normal controls) using a miRCURY LNA Array (Exiqon) which contained 730 human miRNAs. This system is a real-time PCR-based array containing a panel of 384 well-established mature miRNA assays. The kit contains all reagents and primers, reverse transcription and qPCR. In brief, a poly-A tail is added to the mature miRNA template and then synthesized to cDNA by a poly-T primer with a 3′ degenerate anchor and 5′ universal tag. The cDNA is amplified by miRNA-specific and LNA™ -enhanced forward and reverse primers. SYBR Green PCR will be performed in LC480 Real-time PCR system (Roche).
MiRNA validation by real-time quantitative RT-PCR
Plasma/tissues RNA containing miRNA is reverse transcribed to cDNA using miScript Reverse Transcription kit (Qiagen) according to the manufacturer’s instructions. qPCR is performed using SYBR real-time PCR using miScript SYBR Green PCR kit (Qiagen) with the manufacturer provided miScript Universal primer and the miRNA-specific forward primers in ABI 7900 Real-time PCR system (Applied Biosystems). The miRNA-specific primer sequences are designed by us based on the miRNA sequences obtained from the miRBase database (release 19). Each sample is run in duplicates and the expression levels of miRNAs are normalized to an endogenous control RNU6B (U6). Fold change in expression of each gene is calculated by a comparative threshold cycle (Ct) method using the formula: 2-[ΔCt(tumor)- ΔCt(control)].
Statistical analysis
The significance of plasma miRNA levels was determined by the Mann–Whitney test, Wilcoxon test, t-test or Kruskal–Wallis test where appropriate. The Spearman rank order correlation test was used to examine correlation relationships between the levels of the miRNA markers. Multivariate logistic regression model will be established and leave-one-out cross validation will be performed to find the logistic model. Receiver operating characteristic (ROC) curves were established for discriminating patients with or without gastric cancer. All P-values are two-sided and a value less than 0.05 were considered statistically significant. All statistical calculations were performed by the SPSS software (version 17.0).
Acknowledgements
This study was supported by Seed Funding Programme for Basic Research from Committee on Research and Conference Grants, The University of Hong Kong (201109159010). We thank Miss Shan Shan Lu for sample collection and processing, and Mr Jack Chau for statistical data analysis. We also thank Tissue Bank of Department of Surgery (The University of Hong Kong) for providing gastric cancer plasma and tissues samples, and Hong Kong Sanatorium and Hospital for provided healthy control plasma samples.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (
http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (
http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Competing interests
All authors have declared no competing interest.
Authors’ contributions
VY Shin and KM Chu responsible for experimental design. VY Shin and VW Chan performed the experiments. EK Ng and A Kwong analyzed the data. All the authors approved the final draft of the manuscript.