Background
Gastric cancer was the third leading cause of cancer-related death, and was responsible for 723,000 deaths all over the world [
1]. At present, proliferation markers such as the histopathological TNM (tumor-node-metastasis) classification and Lauren classification have been extensively adopted for predicting survival in gastric cancer [
2]. Nonetheless, there is a large variation in survival of gastric cancer patients with similar TNM stage or Lauren classification. Therefore, it is required to explore more precise markers at a molecular level to increase the precision of prognosis prediction for individual patients [
3]. For decades there has been a growing interest in chromosomal instability, which was defined by DNA aneuploidy, to classify gastric carcinoma into molecular subtype groups [
1].
DNA aneuploidy was found in the majority (70–90%) of cancer cells, and was defined as the set of an abnormal number of chromosomes in cells [
4,
5]. DNA aneuploidy reflected a high genotypic instability, leading to malignant behavior of cells. Numerous articles which investigated DNA aneuploidy in various cancers have been published. Its role in non-small cell lung cancer, colorectal cancer and breast cancer seems clear. It was observed that patients with DNA aneuploid tumor had worse clinical outcomes in non-small cell lung cancer, colorectal cancer and breast cancer [
6‐
9].
Previous studies [
10‐
34] which focused on the correlation between DNA aneuploidy and survival in gastric cancer reported conflicting results. Consequently, we conducted this meta-analysis to explore the prognostic value of DNA aneuploidy status in gastric cancer patients.
Methods
Search strategy and study selection
Our meta-analysis was conducted following the Preferred Reported Items for Systematic Reviews and Meta- Analysis (PRISMA) statement [
35]. We searched relevant articles from two databases (PubMed, and Web of Science March 20th 2018). Articles were identified using the following keywords: ‘DNA ploidy/aneuploidy’, ‘gastric cancer/carcinoma’ and ‘survival’. There was no restriction of English language during our search.
Articles were picked out on the basis of inclusion criteria as follows: 1. sufficient information of baseline clinicopathological characteristics of gastric cancer patients; 2. the research focused on the association between overall survival (OS) and DNA aneuploidy; 3. proven histological diagnosis of gastric cancer prior to anticancer therapies; 4. DNA aneuploidy status were presented specificly; 5. value of Hazard ratio (HR) with a 95% confidence interval (CI). Alternatively, the Kaplan-Meier (K-M) curve or a Cox regression model was provided..
Literature retrieval and selection was cautiously performed by two authors (JX and RZ). To avoid overlap of patient cohort, we included the latest studies with largest samples from the same organization. All authors reached consensus and made final decision through thorough discussion.
Two authors (JX, RZ) independently extracted the useful data from all eligible articles. The collected information was listed as follows: the last name of the first author, publication date, sample size, mean or median age of population, DNA ploidy methods, specimens, disease stage, follow-up period, treatments and OS. The risk ratios (RRs) of DNA aneuploidy to tumor stage, depth of invasion (T), lymph node metastasis (N), distant metastasis (M), differentiation (G), tumor types (intestinal and diffuse, based on Lauren classification), and HRs of OS with 95% CIs were checked. If the value of HRs were not provided straightforwardly, survival data were estimated using K-M curves by Engauge Digitizer 4.1.
Quality assessment of primary studies
The Newcastle-Ottawa Scale (NOS) scale was occupied to assess the quality of included papers.
Statistical analysis
We conducted statistical analyses on the basis of the Cochrane Collaboration Guidelines. The pooled RRs were estimated by the Mantel-Haenszel’s method, whereas the pooled HRs were calculated by the inverse variance method. The homogeneity assumption was investigated by Cochran’s Q statistic. The Higgins’
I2 index was calculated to evaluate inconsistency among the included studies. If the heterogeneity was absent, the pooled RR and HRs were calculated by the fixed-effect model (
χ2P < 0.100 and
I2 < 50%). Otherwise the meta-analysis was performed by employing the DerSimonian random-effects model [
36,
37]. Significance was identified if the value of
P was lower than 0.05. Egger’s tests were generated to detect potential publication bias (
P > 0.05 was considered representative of no statistically significant publication bias) [
38]. All the statistical calculations of meta-analyses were performed by RevMan 5.3. And Egger’s tests were carried out through Stata11.0. All the
P values were two-sided.
Discussion
The analysis of DNA ploidy is based on the evaluation of cell viability, which exhibits the percent of DNA in S phase cells. Nevertheless, various factors, e.g., the overlap between diploid cells and aneuploid cells, background fragments, and insufficient number of cells, might disturb the evaluation of S phase. Therefore, the assessment of DNA aneuploidy is usually applied for the analysis of DNA content. Although DNA aneuploidy was supposed to play a crucial role in varies cancers, its prognostic value in gastric cancer was still under debate. To the best of our knowledge, our study is the first meta-analysis exploring the prognostic value of aneuploidy in gastric cancer patients.
Recently, antitumor strategy mainly relies on the TNM system and histologic classification which are common prognostic factors for the disease-free survival and overall survival in cancer patients. However, patients sharing the same clinicopathologic characteristics may present various clinical outcomes. It has been suggested that the measurement of DNA nuclear content via flow cytometry may help distinguish gastric cancer patients who have different disease relapse risk and cancer-related death risk [
39‐
41]. Nevertheless, the results of previous studies were often influenced by the heterogeneity in the composition and size of the samples. From our study, the prognostic value of DNA aneuploidy in GC is clearer.
DNA aneuploid tumors were usually regarded as high metastatic risk tumors with aggressive behaviors [
42,
43]. We did not, however, find any associations between depth of tumor invasion and distant metastases. Likewise, there is a negative association between aneuploidy and tumor differentiation. It may suggest that the malignant potential of a cell clone does not directly rely on the DNA amount. Moreover, only one sample was extracted from each tumor patient, which may lead to a relatively low incidence of DNA aneuploidy.
Our study revealed that DNA aneuploidy was remarkably correlated with advanced TNM staging and lymph-node metastases. This further indicated that DNA aneuploidy played a critical role in tumor progression and aggressiveness. This also suggested the combination of these three factors were crucial to predict the clinical course of gastric cancer patients.
Diffuse tumor was supposed to have a poorer prognosis compared with intestinal tumor [
44]. Interestingly our results show a remarkable correlation between DNA aneuploidy and intestinal histotype rather than the diffuse type. Such heterogeneity may suggest that in these two different tumor types, DNA aneuploidy have various biological and clinical prognostic implications. According to Wyatt et al.’s study, intestinal tumor patients with DNA aneuploidy suffered from shorter survival [
12]. Further studies of this patient population are required to validate this association.
There were some advantages of our meta-analyses. For example, our data processing was based on sufficient literature retrieval and eligible statistical methods. Moreover, we adopted NOS scale to evaluate all the included articles, and found that the quality of their data was relatively high. Nevertheless, some limitations in our study should be mentioned. Firstly, our analyses were based on published studies instead of individual data. Secondly, the included studies were not randomized work. Some of them were prospective studies, while the rest were retrospective studies. Thirdly, the enrolled patients were diagnosed from 1968 to 2006. In this period, the treatment for gastric cancer improved greatly. Fourthly, the methodological change in aneuploidy measurements across these years is a confounding factor. Fifthly, the methodological differences in each study might affect the final conclusion. Subgroup analyses on DNA aneuploidy measurements, cell numbers and specimen preparations were performed. We found that these factors did not influence the association between DNA aneuploidy and clinical outcomes of gastric cancer. Sixthly, stomach contains acid and enzymes which might affect the sample preparations for DNA aneuploidy detection. This factor could affect the obtained data in each study. To avoid this, some of the studies specified the time limit to fix the tissues. And others used deep site or multiple sites of tumor for preparation. However, a standard is required in the future experiments. Our meta-analysis provide an indication of the prognostic value of DNA aneuploidy, we hope our study could encourage future studies to draw a standard guideline in this aspect.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.