Prognostic value of melanoma cell adhesion molecule expression in cancers: a meta-analysis

Abstract

Melanoma cell adhesion molecule (MACM) has been reported in many studies as a novel bio-marker for its prognosis value in cancers. But the prognosis significance of MACM expression in cancer remains inconclusive. Therefore, we conducted a system review and meta-analysis to assess its prognosis value in cancers. A systematic search through Pubmed, EMBASE and Cochran Library database was conducted. Hazard Ratios (HRs) and 95% confidence intervals (CIs) were used to evaluate the prognosis value of MACM expression. Eleven studies with 2657 cases were included after sorting out 462 articles for this meta-analysis. The results of the fixed-model depending on the heterogeneity in studies demonstrated that MACM expression was significantly associated with overall survival (OS) in cancer (HR=2.84, 95% CI: 1.10-7.31, P<0.00001). Furthermore, subgroup analysis indicated that high expressed MACM predicted a poor OS in both Asian (HR=2.52, 95% CI: 1.80-3.52, P<0.00001) and Caucasian (HR=2.40, 95% CI: 2.01-2.88, P<0.00001). In conclusion, high expression of MACM was significantly associated with a poor prognostic outcome in cancer. MACM can be regarded as a novel bio-marker in different types of cancers and can be used to evaluate the prognosis of therapeutic effect during clinical practices.

Introduction

Carcinoma is an uncontrolled cell growth disease with complex mechanism and severe consequence, invading and destroying nearby parts of the body [1]. In general, about 50% of people accepting treatment for infiltrating cancer die from cancer or its treatment. Outcome of cancer tends to be poor in developing countries because of the inefficient diagnosis and limited treatment [2]. Thus, the accurate diagnostic methods with high sensitivity and specificity are indispensable to the detection of cancer and remain a key obstacle for efficient treatment and prognosis [3]. Latest research results indicated that tumor markers could be helpful along the continuum of care for patients [4].

MACM is originally expressed at the intercellular junction of endothelial cells [5] and considered as a tightly connection between endothelial cells. MACM contains a characteristic V-V-C2-C2-C2 immunoglobulin-like domain structure belonging to immunoglobulin supergene family [6]. Most often, this adhesion molecule exists in many normal tissues, such as vascular endothelial cells, smooth muscle cells and pericytes [7]. The expression of MACM was reported to increase during pathological conditions after endothelial injury, such as cardiovascular diseases, organ transplant rejection, inflammation, nephritic syndromes and surgical trauma [8]. To date, an increasing number of studies have demonstrated that MACM plays an important role in cancers [9], including melanoma [10-12], breast cancer [13,14], nasopharyngeal carcinoma [15], epithelial ovarian cancer [16], and non-small cell lung cancer [17]. However, the underlying mechanisms of MACM associated with cancer progression have not been elaborately studied [18]. Recently, some research data indicated that expression of MACM was associated with prognosis of cancers. These implied MACM could be a bio-marker for the prediction of prognosis in various cancers.

Therefore, we conducted a comprehensive study search and performed a meta-analysis to investigate the correlation between MACM and prognosis of cancers.

Materials and methods

Search strategy

We conducted a comprehensive search of PubMed, EMBASE and Cochrane Library database though March of 2015. We used the following search terms: “MACM,” “MCAM,” and “MUC18” in combination with “cancer,” “carcinoma,” “neoplasm,” and “tumor”. Potential references in the retrieved articles were also reviewed to consummate our search. Authors of the primary studies were also contacted for more specific information when required. G. Zhu and X. Zhang performed the literature search independently.

Selection criteria

The studies were included when they conformed to the following criteria: 1) Positive expression of MACM needs to be identified. 2) They focused on prognosis of the carcinoma and analyzed the relationship with MACM. 3) They offered sufficient data to calculate the hazard ratio (HR) and the corresponding 95% CI (either direct data or Kaplane-Meier survival curves). We also excluded studies based on the following criteria: 1) Review articles or the whole text couldn’t be provided. 2) If the studies overlapping with the same population, we preferred the study with the longest average follow-up period. 3) Case-control studies or the exposure not associated with our analysis were also excluded. G. Zhu and H. Xiong excluded the articles independently according to the selection criteria.

Data abstraction

All data were extracted using a standard form to get the recorded information, including: first author, publication year, country, ,number of cases, type of carcinoma, detection method, duration of follow-up, identification of MACM positive expression and the HRs and the corresponding 95% CI, overall survival or progression-free survival. If the studies provided univariate and multivariate analysis, we selected the multivariate analysis. If the studies only provided a Kaplan-Meier curve instead of HRs and CIs, we used software Engauge Digitizer 4.1 to extract data.

Quality evaluation

A Newcastle-Ottawa Quality Assessment Scale (NOS), composed of three items (selection, comparability and exposure), was used to evaluate the quality of every single study included in our analysis. Each item could be awarded with stars based on NOS evaluation criterion. All stars were added up to get a total points and the identify threshold score for high quality selection is six. Studies with 6 points or more were regarded as high quality choices.

Statistical analysis

Hazard ratios (HRs) and corresponding 95% confidence intervals (CIs) were recorded by Review Manager 5.1. They were used as the common measure of prognosis value of MACM expression. Q test and I² test were used to measure the heterogeneity among studies. We selected a Fixed or Random model to calculate the pooled HR depending on the heterogeneity test. A Begg rank correlation test and Egger linear regression test were used to evaluate the publication bias. Subgroup analysis were conducted for further estimation based on duration of follow-up, MACM expression detection methods, races and approaches to extract data. Sensitivity analyses were executed to investigate the influence of every single study on the overall risk. The Begg’s test, Egger’s test and sensitivity analyses were conducted with STATA.

Results

Literature search

We initially searched 462 potentially eligible studies from the Pubmed, Cochrane Library, EMBASE databases. 445 articles were excluded after screening the titles and abstracts, mainly because they were case-control studies or the exposure were not associated with our analysis. Therefore, 17 articles were left for further evaluation. Three studies were excluded because they can’t offer sufficient data to calculate the HR and 95% CI. Two studies were excluded because they used a retrospective cohort design, and one was excluded because the length of follow-up was too short. Finally, 11 studies were included in our meta-analysis [10,14,16-24]. The process of selecting the eligible studies was presented in Figure 1.

Figure 1.

Flow diagram of selection of studies included in our meta-analysis.

Study characteristics

The characteristics of the 11 eligible studies were listed in Table S1. Studies included were published between 2007 and 2014. Of these, six studies were conducted in China; two were in France; three were in UK, Italy, and Japan respectively. Furthermore, two studies were concerned with NSCLC; one was concerned with each of colorectal cancer, ovarian cancer, clear cell renal cell carcinoma, breast cancer, melanoma, lung adenocarcinoma, gastric cancer, gallbladder adenocarcinoma and ESCC. The average cases were 242, ranging from 63 to 1080 with a total number of 2659. The testing methods of MACM were multiple. 8 studies used immunohistochemistry (IHC) to evaluate the level of MACM; one study used RNA expression as a measurement of MACM expression; one study applied flow cytometry and one applied ELISA to test MACM expression. Besides, 7 studies detected MACM expression on both membrane and cytoplasm, and the left 4 studies detected only on membrane. OS was used as survival condition in the selected 11 studies and PFS/TTP was used in only two studies. The median duration of the follow-up was 5.3 years, ranging from 2 to 12 years.

Quality assessment

The Newcastle-Ottawa Scale (NOS) was conducted to evaluate the quality of every study included in our meta-analysis. We assessed three items of each study: selection, comparability and outcome [25]. After all stars added up, the 11 studies were scored ranging from 7 to 9, which can be regarded as high quality choices. The evaluation was presented in Table S2.

Main analysis

Hazard ratio (HR) and the corresponding 95% confidence interval (CI) were used to evaluate the correlation between MACM expression and cancers. The results, presented in Figure 2, implicated that MACM expression was significantly correlated with OS in cancers (HR=2.84, 95% CI: 1.10-7.31, P<0.00001). Heterogeneity test were conducted among the studies and results showed no evidence of heterogeneity (I²=34%, P=0.13). A fixed-effect model was used to evaluate the correlation in this analysis.

Figure 2.

Main analytical results of investigating the correlation between CD146 expression and overall survival of patients with cancers. A. The forest plot exhibiting the pooled HR and the corresponding 95% CI. B. Funnel plot analysis investigating the publication bias.

Subgroup and sensitivity analysis

The subgroup results were presented in Table S3. We conducted the subgroup analysis based on the ethnic groups of studies (Figure 3). The investigation suggested that the high MACM expression predicted a poor prognosis both in Caucasian (HR=2.52, 95% CI: 1.80-3.52, P<0.00001) and Asian (HR=2.40, 95% CI: 2.01-2.88, P<0.00001). No significant heterogeneity was observed (Caucasian: I²=43%, P=0.10; Asian: I²=33%, P=0.21). We further stratified the extracted data according to the length of follow-up (Figure S1), detection methods (Figure S2) and approaches to data extraction (Figure S3). For duration of less than 60 months, the high MACM expression predicted poor OS of patients (HR=1.97, 95% CI: 1.54-2.52, P<0.00001); and duration of more than 60 months predicted even worse outcome (HR=2.83, 95% CI: 2.30-3.50, P<0.00001). As for the subgroup of detection methods, the results showed similar outcome. 8 Studies used IHC to detect MACM expression also suggested a poor outcome of patients (HR=2.40, 95% CI: 2.03-2.84, P<0.00001). Three studies applied flow cytometry, RNA expression and ELISA respectively, making the results rely on the individual study entirely. Six studies reported HRs and 95% CI directly implying the significant value of MACM expression in cancers (HR=2.50, 95% CI: 1.81-3.45, P<0.00001). The left five studies which we used software to estimate the HR also showed a parallel conclusion (HR=2.41, 95% CI: 2.01-2.89, P<0.00001).

Figure 3.

Subgroup analysis results stratified on races.

We performed sensitivity analysis to investigate the potential sources of heterogeneity by omitting one study in each time. The results (Figure 4) implicated overall risk estimate was not obviously influenced by any single study. Combined HRs and the corresponding 95% CI ranged from 2.11 (95% CI: 1.73-2.58) to 2.58 (95% CI: 2.17-3.08) with one study omitted each time.

Figure 4.

Sensitivity test among studies included.

Publication bias

No substantial asymmetry was observed in funnel plot. But the Begg’s and Egger’s test showed different results (Figure S4). The P value of Begg’s test was 0.043, and the Egger’s regression test showed no evidence of publication bias (P=0.468). To a further evaluation of the publication bias, a Trim and Filling method was conducted to explain the paradox between the Begg’s test and Egger’s test. The results of Trim and Filling method (Figure S5) showed that two more studies should be added to eliminate the publication bias (variance=0.061, P=0.075).

Discussion

MACM is expressed in a limited quantity of normal tissues, such as hair follicle, smooth muscle, cerebellum, mammary, lung, and nasopharynx [26]. As a receptor for laminin alpha 4 [27], this matrix molecule is prominent associated with a variety of biological and pathological processes. MACM has been demonstrated to play a significant role in tumor progression, metastasis and invasion [28]. Evidence shows that MCAM expression is associated with the progression of human breast cancer [14,29]. It can also promote the metastasis of other tumor types, including melanoma [30], endometrial cancer [31], nasopharyngeal carcinoma [15] and others. However, Wu et al [26] suggest that MCAM may have the opposite effect on modulating the progression of different cancers, stimulating the metastasis of some certain cancers and also inhibiting the malignant progression in other tumors. Thus, we conducted this meta-analysis to further investigate the association between MACM and the prognosis of cancers.

It is the first meta-analysis to evaluate the prognosis value of MACM expression in cancers, and the results illustrated the significant correlation between MACM expression and poor outcome of cancers (HR=2.84, 95% CI: 1.10-7.31, P<0.00001). Accumulating studies showed that inflammation can promote melanoma progression and increased MACM expression was found in inflammatory lesions [12], also leading to VEGF-induced vascular permeability in MACM KO mice [30,32]. Wu et al [33] elucidated that suppressing MCAM may increase the cell apoptosis via mediating the dysregulation of small RhoGTPase (RhoA and Cdc42) in ovarian epithelial tissues. Another study showed that high expression of MACM may lead to a loss of cell-cell contacts and induce phenotypic by transiting epithelial to mesenchymal [34] (EMT) in human breast cancer cell lines. And the mechanism was demonstrated to be related with the estrogen receptor (EP) [13]. However, MCAM was reported to restrain the progression of colon carcinoma [35] while other studies demonstrated a facility effect on the metastasis of colon carcinoma [36]. Increasing studies have found that MACM was significantly associated with metastasis of cancers and predicted a poor outcome of patients, which is consistent with our results of this meta-analysis. In our subgroup analysis of ethnic groups, the prognostic value of high expression of MACM seems more pronounced in Asian (HR=2.40, 95% CI: 2.01-2.88, P<0.00001) than in Caucasian (HR=2.52, 95% CI: 1.80-3.52, P<0.00001). Genetic inheritance may account for this slight difference. Hence, more investigation about the mechanisms of MACM’s function in cancers needs to be performed to confirm the hypothesis.

However, several limitations also impose restrictions on our meta-analysis. First, we included 11 eligible studies involving 10 types of cancers; each type has insufficient studies to summarize the main effect. Second, the Begg’s test indicated the existence of publication bias. After performing a Trim and Filling method, additional two studies need to be added to eliminate the bias. Third, some studies included only provided Kaplan-Meier curve, we used software Engauge Digitizer 4.1 to estimate the data. Thus the calculation error was unavoidable. Fourth, the detection method of MACM expression varied among studies in our meta-analysis and may account for the heterogeneity. Finally, the cut-off value to determine positive expression of MACM was quite different among studies. There is no consolidated standard of MACM positive expression. Therefore, more studies are needed to consummate our analysis for further investigation to assess the prognostic value of MACM in cancers.

Conclusion

This meta-analysis demonstrated that MACM was significantly associated with overall survival of patients and predicted a poor prognosis outcome in cancers. MACM may become a novel prognostic bio-maker and could be used as an efficient therapeutic target with further research to elucidate the mechanism.

Disclosure of conflict of interest

None.