Expression and clinical significance of CD147 in renal cell carcinoma: a meta-analysis

Objective To assess clinical significance of CD147 in renal cell carcinoma. Methods Collect case-control studies which focus on CD147's expression in renal cell carcinoma. Trails were retrieved from CBM, CNKI, Wan-fang database, PubMed, Cochrane Library and Embase. According to the inclusion and exclusion criteria, data extraction and quality assessment were done by two researchers independently, and outcomes were pooled with Revman5.3 and STATA14.0. Results A total of 11 studies were confirmed, among which renal cell carcinoma 887 cases, non-cancer 505cases. As for the positive rate of CD147, there are statistical differences among survival, renal cell carcinoma tissue vs. non-cancer tissues [OR= 8.19, P= 0.0002], with vs. without lymph node metastases [OR= 6.52, P= 0.001], clinical stage III~IV vs. II~I [OR= 4.07, P< 0.00001], histopathological stage III~IV vs. II [OR= 3.01, P= 0.002], histopathological stage III~IV vs. I [OR= 7.50, P< 0.00001], tumor size [OR= 5.01, P= 0.0007]. No significant difference was tested among different age, gender, histological types and Position of cancer. Conclusion As shown in our results, CD 147 may participate the whole course of carcinogenesis of renal cell carcinoma, which might be valuable for the diagnosis, treatment and prognosis.


INTRODUCTION
Renal cell carcinoma (RCC) is the most common malignancy of kidneys found in adults [1]. It is the seventh most common cancer in man and the tenth in women, which account for 2%~3% of all adult malignancies [2]. The incidence of RCC is increasing rapidly on average 1.1% over the last ten years [3]. In the United States, there are more than 62,000 new cases of RCC and at least 14,000 deaths expected in 2016 [2]. Tumor metastasis is the main cause of mortality and treatment failure in RCC patients [4]. Unfortunately, because of the lack of obvious symptoms during the early stage, up to 20-30% RCC patients present with metastatic disease [5]. Therefore, it is of great significance to search for sensitive and specific markers that can provide valuable information for the early diagnosis and prognosis of RCC. A number of biomarkers such as survivin, MCT1, MCT4, Cullin 1 and CD147 have been found to be involved in its development and progression according to some reports [6][7][8]. Among these markers, extracellular matrix metalloproteinase inducer (EMMPRIN, also known as CD147 or basigin) is highly expressed in a variety of tumors, facilitating tumor invasion and metastasis [9]. Studies show that CD147 is up-regulated in a lot of malignancies, including breast, lung, oral, esophageal, laryngeal and renal cancers [6,[10][11][12][13][14]. So there is great possible that CD147 can serve as a prognosis biomarker for RCC. But the expression profiles of CD147 in RCC are controversial. For example, some studies show a significant difference among the subtypes of RCC while some studies argue about this [15][16][17][18]. The same conflict is also seen in the studies of metastasis status and TNM stage [16,[18][19][20].

Research Paper
These controversies could be a result of differences in sample sizes and other factors, such as the criterial of the positive expression of CD147, and unfortunately evidence-based confirmation by large-scale clinical trials is still lacking. Therefore, we conducted this meta-analysis to quantitatively inspect the relationship between CD147 and clinicopathological features and survival of renal cancer patients.

CD147 expression and survival
We investigated the relationship between CD147 expression 1-,3-,5-,10-and 15-survival. Fixed-effects model was used for all them without heterogeneity. As expected, the positive expression of CD147 indicated worse long-term survival with significant difference (Table 2, and Supplementary Figure 1). In addition, five data sets [16,18,19,23,24] with 5-year overall survival HR (hazard ratio) are avalialble, and all the studies reported the univariate result. We combined the data with fixed model effect (I2=0, P=0.728), and found that CD147 positive expression bring a worse 5-year survival comparing with negative patients (HR=1.61, 95%CI= 1.04-2.49) ( Figure 2).

Bioinformatics analysis
First, by using the GOpubmed, we explored that the discovery of CD147 was in 1990 and since then it is becoming more and more popular expecially in resecnt years ( Figure 5A). In addition, most of research related to CD147 was investigated in China (142 publications) and the USA (141publications), followed by Japan (65 pablications) ( Figure 5B and 5C). Besides, we can see the author networks in Figure 5D. And the top 20 author was show in Figure 5E. Second, string version10 analysis result showed the interaction between CD147 and other protiens ( Figure 6A-6D). Figure 6A, active interaction sources restricted only on experiment data and B-D without sources restriction. A-B with high confidence (0.700) while C-D with medium confidence (0.400). A-C with max number of interactors 50 for first shell and 5 for the second shell, while D with 50 for the first shell and the second shell. For more details of information can be seen in supplymentary file. Third, IPA (Ingenuity PathwayAnalysis) result showed in Figure 7A-7D without species and sources restriction of the data seting. Figure 7A showed total interaction network of CD147 with other moleculars (252 moleculars) including protein and micro-RNA and their cellular location (extracellular space, plasma membrane, cytoplasma, nuleus, and other organelle) and red labeled mean the molecular is related to caner. We can know almost of all the interaction molecular is related to cancer (233/252, 92.5%). Figure  7B, red labeled in the molecular related to tumorgenesis (221/252, 87.7%). Figure 7C labeled with Blue linked line molecular is related to molecular mechanisms of cancer. And Figure 7D show the molecular mechanism signaling pathway and the red labeled molecular is the same molecular linked with bule line in Figure 7C. besides, we explored the interaction between CD147 and MAPK, AKT, ERK, VEGF which is the most core molecular in their signaling pathway ( Figure 8A-8D).

DISCUSSION
CD 147, also known as extracellular matrix metalloproteinase inducer (EMMPRIN), is a heavily glycosylated immunoglobulin [27]. Also there are numerous clinical study in renal carcinoma to test whether CD 147 can be a biomarker. However, the clinical relevance of CD147 remains controversial. In the current meta-analysis, we pooled the data from 13 studies together [15-20, 22-26, 35, 36], and demonstrated a remarkable association between CD147 expression of patients with renal cancer.
As shown, CD 147 is a good predictor for the survival of renal cancer patients. No matter for 1-, 3-year short term or 5-, 10-, 15-year long term survival, patients with positive expression of CD 147 did much worse than negative expression patient.
When we come to investigate the different CD 147 expression among cancer tissue and non-cancer tissue, results consist with the conclusion in other cancer [30,[32][33][34], which show cancer tissue is positive for CD 147 with significant statistic difference on both mRNA (OR= 7.93, 95%CI=(2.44, 25.77), P= 0.0006) and protein level (OR= 8.59, 95%CI= (2.20, 33.52), P= 0.002). So this is a good proof for CD 147 to be a potential biomarker to renal cancer.
We compared the expression of CD 147 among different lymph metastasis, clinical stage, histopathological stage, age, gender, histological type and the position of tumor to test the relationship between CD 147 and clinicopathologic characteristics. In a word, CD 147 is strongly associated with poorclinical features.    The molecular mechanism signaling pathway and the red labeled molecular is the same molecular linked with bule line in Figure 7C. www.impactjournals.com/oncotarget CD147 has been shown to be involved in the regulation of tumor cell invasion, metastasis, angiogenesis, energy metabolism and anti-apoptosis. Firstly, by combininginteracting with integrin alpha6beta1, CD 147 take part in FAK P13K-Ca (2+) and MARK signal pathway, interstitial collagenase (MMP-1) is producted and CD147-MMP-1 complex is formed at the tumor cell surface, thus modifying the tumor cell pericellular matrix to promote invasion [37][38][39]. Secondly, CD147 is a receptor for platelet GPVI and mediates platelet rolling via GPVI-EMMPRIN combination, incerasing the potential metastasis [40]. Thirdly, it can influence the VEGF/ VEGF receptor system of endothelial cells and increase the blood supply for tumor [41]. For glycolytic energy metabolism, CD147 promote the lactate transportion to feed the tumor with the help of AMPK and MCT1/2 [42, 43]. Last but not least, CD147 can down-regulate Beclin 1 and inhibit starvation-induced autophagy through the PI3K/Akt/mTOR pathway, modulating the apoptosis of tumor [44,45]. And our results demonstrated that poor clinical performance for renal cancer patients, such as lymph metastasis, higher TMN and histopathologic stage, bigger tumor size, is usually accompanied by positive CD 147. So the survival time for them would be much shorter than CD 147 negetive patients.
Efforts were made to conduct a comprehensive analysis, but some limitations need to be acknowledged. First, although we have tried but no unpublished data was found, so all included studies were published data. Publication bias existed in 10-year survival and the tumor size. But trim and fill analysis (S3 Appendix) showed the same results which means the result is reliable. Second, survival analysis was not performed by multivariate analyses in most reported studies, this could bring some bias. Third, external virtuality was limited as most included patients were Chinese. Last, the definition of CD147 positive expressin criteria in all studies isn't unified, which could bring some potential bias, fortunately, the sensitivity analyses showed a stable result.
To our knowledge, this meta-analysis is the first study which systematically estimates the association between CD147 expression and the survival analysis and its clinicopathological parameters. Nowadays, early diagnosis and earlytreatment are the fundamental approaches to improve prognosis [46]. Our results found that CD147 positive expression was significant associated with renal caner tissues, which supported that CD147 could be applied as a potential clinical marker for the early diagnosis of renal cancer. In our study, we demonstrated that CD147 positive expression strongly predicted poorer TNM stage, histopathologic stage, lymph node metastasis and worse survival in the patients with renal cancer. Further studies using additional putative renal cancer surface markers in combination with CD147 are required to evaluate their potential use in predicting patients' outcome.  (1988-2016). The studies were restricted to humans, but not restricted by date, language, or publication status. The following combined search term was used: (kidney neoplasms, renal neoplasms, kidney tumor, renal tumor, kidney cancer, renal cancer, kidney carcinoma, renal carcinoma, clear cell carcinoma) AND (CD147, extracellular matrix metalloproteinase inducer, EMMPRIN, BSG). We combined the term appropriately with MeSH Terms and used an appropriate adjustment for different databases. Details of the search strategies can be found in S1 Appendix.

Criteria for including studies
1. Published or unpublished case control study or cohort study in English or Chinese with the full text available; 2. Cases have survival data or clinical pathological characteristic data, without radiotherapy or chemotherapy or biological therapy before sampling; 3. Pathological methods for confirming renal cancer. 4. CD147 expression based on primary renal cancer tissue, rather than serum or any other kinds of indirect specimen were included. 5. Choosing best quality study for duplication.
Criteria for excluding studies 1. Cell or animal studies, case reports, letters, reviews.
2. The standard of pathological diagnosis was not clear;

Statistical analysis
Bibliographies were scanned by two authors independently to exclude unrelated studies. Then, full text were reviewed, data were extracted independently and controversy were solved by disscussion. The software Revman 5.3 and Stata 14.0 were applied to analyze the data. Results were showed with odds ratios (OR) or standard mean difference (SMD) and 95% confidence intervals (95% CI). Fixed-effects model was adopted for non-heterogeneity data (P > 0.1 and I 2 < 50%); otherwise, random-effects model. If possible, heterogeneity subgroup analyses were performed. All P values were 2-sided, and P < 0.05 was considered significant. Sensitive analysis was also performed to evaluate the influences of individual studies on the final effect. Egger's test was used to assess publication bias (P < 0.05 was considered statistically significant). If publication bias was confirmed, a trim-and-fill method developed by Duval and Tweedie was implemented to adjust for this bias [21]. Then, we replicated the funnel plot with their ''missing'' counterparts around the adjusted summary estimate.

Bioinformatics analysis
Firstly, we used GOpubmed (http://www.gopubmed. com/web/gopubmed/) with the Search term""Antigens, CD147"[mesh]"to get information about the current related published studies with a global perspective. Secondly, we applied String vesrion10 (http://string-db. org/) to explore the interaction between CD147 and other moleculars. We choosed the organism with Homo sapiens, and mathed the information with different minimum required interaction score, active interaction sources and max number of interactors. Thirdly, IPA (Ingenuity PathwayAnalysis) (http://www.ingenuity.com/) was employed to uncover the interaction network between CD147 and other moleculars with different threshold without species restriction. And IPA pathyway explorer was used to explore some important signaling pathways which CD147 has take parted in.

Author contributions
Hui Li conceived the study and drafted the paper. Dongwen Wu participated in drafting the paper and carried out the Statistical analysis. Shupeng Shi, Yadong Xu and Lin Wei searched the lectures. Jing Liu and Yanting Liu did data extraction and quality assessment.

CONFLICTS OF INTEREST
The authors declare that they have no conflict of interests.