Association of caveolin-1 protein expression with hepatocellular carcinoma: a meta-analysis and literature review

Background

Hepatocellular carcinoma (HCC) is the sixth most common malignancy worldwide and the third leading cause cancer-related mortality globally.^1–3 Although many advances have been made in the diagnosis and management of HCC, the prognosis of patients with HCC remains poor due to late diagnosis, high recurrence, and metastasis.^4,5 Therefore, it is critical to identify biomarkers for diagnosis and predict the prognosis of patients with HCC.

Caveolins are a family of scaffolding proteins that coat 50 to 100 nm plasma membrane invaginations known as caveolae.⁶ There are three isoforms in the caveolin family: caveolin-1,⁷ caveolin-2,⁸ and caveolin-3.^8–13 Caveolin-1 is expressed in most cell types such as adipocytes, epithelial cells, and fibroblasts, where the protein is typically found co-expressed with caveolin-2.^8,14 Caveolin-2 is incapable of forming caveolae alone. Caveolin-3 has similar structure and function as caveolin-1, but caveolin-3 is expressed in skeletal, cardiac, and smooth muscle cells.¹⁵ Recent evidence indicates that CAV-1 plays important roles in carcinogenesis by regulating cell proliferation, and its roles differ among distinct histological tumor types.¹⁶ The aberrant expression of CAV-1 is involved in the pathogenesis of a variety of cancers. CAV-1 can act as a tumor suppressor or initiator, depending on the tissue of origin.¹⁷ Overexpression of CAV-1 in hepatocarcinogenesis has been shown to protect HCC cells from apoptosis and enhance HCC cells’ migration and invasion abilities.^18–21 However, the results were controversial due to the study’s small sample size. In the present study, we pooled nine studies and performed a meta-analysis to evaluate the relationship between the overexpression of CAV-1 and clinicopathological significance in HCC.

Methods

Search strategy and selection criteria

To identify relevant published literature, we searched PubMed (1966 ~ 2018), Web of Science (1945 ~ 2018), EMBASE (1980 ~ 2018), and Google scholar.

The following terms were used for searching: caveolin-1 or CAV-1 and HCC, liver cancer or HCC. We included studies that met the following criteria: 1) the association between CAV-1 overexpression and clinicopathological significance in HCC; 2) the association of CAV-1 overexpression and prognosis in HCC patients. Articles were excluded if: 1) the study utilized the same population or overlapping database, 2) the study is about the association between CAV-1 mRNA expression and clinicopathological significance, 3) the studies utilized cell lines, mice, or human grafts; 4) conference abstracts, case reports, reviews, letters, editorials, and expert opinion that contain insufficient data. The detailed information of nine relevant citations is summarized in Table 1.

Table 1 Main characteristics of included studies

Data extraction and study assessment

Two reviewers extracted data from selected studies independently. Any discontent was discussed, and a consensus was reached for all issues. The following items were collected from each study: first author’s name, year of publication, number of patients, stage of HCC, grade of HCC, overall survival (Kaplan Meier survival curve), CAV-1 expression, the quality of selection, comparability, exposure, and outcomes for study participants.

Statistics analysis

ORs with 95% CIs were calculated by using a fixed- or random-effect model depending on heterogeneity (a fixed-effect model for I²<50%, a random-effect model for I²>50%). Further analysis was performed to compare CAV-1 overexpression among HCC versus normal tissue, cirrhosis versus normal liver, high grade of HCC versus low grade of HCC, and advanced stage HCC versus early stage HCC. Raw data were obtained from Kaplan–Meier survival curve, replotting on the log–log scale. This estimates the Cox regression coefficient. The output is the log HR and its standard error. HR with a 95% confidence interval was calculated for the association between CAV-1 expression and prognosis. All p-values were two-sided. Funnel plots were used to detect publication bias. All analysis was performed with Review Manager 5.2.

Results

Five hundred and twenty-five articles from database PubMed, Web of Science, Embase, and Google Scholar were screened, and nine articles were included in the meta-analysis (Figure 1). There were 810 HCC and 172 cirrhosis patients involved in the present study. Table 1 summarizes the clinicopathological characteristics of the patients from all of the included studies.

Figure 1 Flowchart of study selection.

Newcastle Ottawa Quality Assessment Scale (NOQAS) was utilized to assess the quality of all included studies. The scales allocate a maximum of nine points for the quality of selection, comparability, exposure, and outcomes for study participants, and a score ≥7 indicates a good quality. Of the studies, three scored eight points, five scored seven points, and one scored six points. The studies were of a relatively high quality (Table S1) (Supplementary materials).

The rate of CAV-1 protein overexpression in HCC was not significantly higher than in normal liver tissue; OR was 5.36 with 95% CI 0.65–44.49, z=1.56, p=0.12 (Figure 2). However, CAV-1 protein overexpression was more frequently observed in cirrhosis than in normal liver tissues; OR was 3.25 with 95% CI 1.26–8.42, z=2.43, p=0.01 (Figure 3A). Furthermore, the frequency of CAV-1 protein overexpression in HCC with cirrhosis was significantly higher than in HCC without cirrhosis (Figure 3B); OR was 1.48 with 95% CI 1.01–2.17, z=2.00, p=0.05. The results indicate that the overexpression of CAV-1 protein can potentially contribute to the development of cirrhosis and subsequently lead to liver carcinogenesis. Whereas the frequency of CAV-1 protein overexpression in HCC was not significantly higher than in cirrhotic liver, OR was 1.53 with 95% CI 0.75–3.12, z=1.17, p=0.24 (Figure 3C).

Figure 2 The pooled OR from three studies including 272 HCC patients and 41 normal liver specimen. The studies included investigating the CAV-1 overexpression between HCC and normal liver tissues. 272 HCC patients and 41 normal liver specimen with the combined OR being 5.36 (95% CI=0.65–44.49, p =0.12) are shown.Abbrevitions: HCC, Hepatocellular carcinoma; NCT, Normal control tissue;OR, Odds ratio.

Figure 3 The pooled OR from studies in normal liver, cirrhosis, and HCC. (A) The studies included investigating the CAV-1 overexpression between 172 patients with cirrhosis and 41 normal liver specimens. The combined OR was 3.25, 95% CI=1.26–8.42, p=0.01. (B) The pooled OR from 5 studies included 594 HCC with cirrhosis patients and 221 HCC without cirrhosis patients, OR=1.48 (95% CI=1.01–2.17, p=0.03). (C) The studies included investigating CAV-1 overexpression between 272 patients with HCC and 172 patients with cirrhosis. The combined OR was 1.53, 95% CI=0.75–3.12, p=0.24.Abbreviations: HCC, Hepatocellular carcinoma; NCT, Normal control tissue;OR: Odds ratio.

Overexpression of CAV-1 protein was correlated with high grade of HCC; OR was 2.61 with 95% CI 1.09–6.26, z=2.16, p=0.03 (Figure 4). The rate of CAV-1 protein overexpression was higher in stage III–IV HCC than in stage I–II, but there was no significant difference; pooled OR was 1.43 with 95% CI 0.95–2.17, z=6.79, p=0.09 (Figure 5). Overexpression of CAV-1 protein was strongly associated with HCC invasion, and the rate of CAV-1 protein overexpression in HCC with metastasis was 2.71 times higher than in HCC without metastasis; 95% CI was 1.26–8.42, z=2.12, p=0.03 (Figure 6). No statistically significant correlation was found between the rate of CAV-1 protein overexpression and the size of tumor; OR was 0.97 with 95% CI 0.58–1.62, z=0.12, p=0.90 (Figure 7).

Figure 4 The pooled OR from studies in high and low grade of HCC. Pooled results of CAV-1 overexpression in different grades of HCC. The OR was 2.61, 95% CI=1.09–6.26, p=0.03.Abbreviations: HCC, Hepatocellular carcinoma; ORs, Odds ratio; CI, Confidence interval.

Figure 5 The pooled ORs from studies in different stages of HCC. Pooled results of CAV-1 overexpression in different stages of HCC. The pooled OR was 1.43, 95% CI=0.95–2.17, p=0.09.Abbreviations: HCC, Hepatocellular carcinoma; ORs, Odds ratio; CI, Confidence interval.

Figure 6 The pooled ORs from studies in different status of invasion. The pooled OR was 2.71, 95% CI=1.71–4.28, p<0.0001.Abbreviations: ORs, Odds ratio; CI, Confidence interval.

Figure 7 The pooled ORs from studies in different sizes of HCC. The pooled OR was 0.97, 95% CI=0.58–1.62, p=0.90.Abbreviations: HCC, Hepatocellular carcinoma; ORs, Odds ratio, CI: Confidence interval.

CAV-1 protein overexpression was also closely related to poor overall survival in patients with HCC; pooled HR was 0.40 with 95% CI 0.17–0.93, z=6.79, p=0.01 (Figure 8).

Figure 8 The pooled HR from studies for the association of CAV-1 expression with overall survival of HCC patients. Five studies are eligible for evaluation of OS and CAV-1 overexpression with 719 HCC patients. The pooled HR for OS showed that CAV-1 overexpression was associated with worse survival in HCC, HR=0.40, 95% CI=0.17–0.93, Z=2.12, p=0.03.Abbreviations: HCC, Hepatocellular carcinoma, HR: Hazard ratio; CI, Confidence interval.

A sensitivity analysis was performed by removing one study at a time, and the ORs were not significantly changed, indicating the stability of the present meta-analysis (S1, S2, S3, S4, S5, and S6). The funnel plots were largely symmetric (Figure 9), suggesting that publication biases do not exist in the meta-analysis of the relationship between CAV-1 overexpression and clinicopathological significance.

Figure 9 Funnel plots of publication bias in the meta-analysis of CAV-1 overexpression and clinicopathological features. (A) CAV-1 overexpression in HCC vs normal liver tissues, (B) Cirrhosis vs normal liver tissues, (C) HCC with cirrhosis vs HCC without cirrhosis, (D) HCC vs cirrhosis, (E) different grades of HCC, (F) different stages of HCC, (G) different status of invasion in HCC, (H) different sizes of HCC.Abbreviations: HCC, Hepatocellular carcinoma.

Discussion

Caveolins are a family of scaffolding proteins that coat 50 to 100 nm plasma membrane invaginations.^6,22 Lipid rafts play an important role in signaling crosstalk via bringing different proteins into proximity and promote interactions among receptors and intracellular signaling proteins.²³ Cirrhotic liver is related to macroregenerative and dysplastic nodules which are HCC precursor lesions. Our data indicate that the CAV-1 protein overexpression in cirrhosis is significantly higher than in normal liver, suggesting that CAV-1 protein overexpression plays an important role in cirrhosis which then acts as an initiator of carcinogenesis. However, the rate of CAV-1 protein overexpression in HCC is not significantly increased compared to the rate of CAV-1 protein overexpression in normal liver. This is in line with the report that CAV-1 was overexpressed in cirrhotic liver^24,25 and was underexpressed in HCC.²⁶ Additionally, we found that the frequency of CAV-1 protein overexpression in HCC with cirrhosis is higher than in HCC without cirrhosis. Our data indicate that CAV-1 overexpression likely plays an essential role in liver cirrhosis, subsequently contributing to the development of the HCC that derived from cirrhosis.

There are three isoforms in caveolin family: caveolin-1, caveolin-2, and caveolin-3. Engeman et al reported that the genes encoding caveolin-1 were co-localized at D7S522 locus on human chromosome 7q31.1, which was commonly deleted in cancers of breast,^27,28 colon,^29,30 kidney,³¹ prostate,^32,33 ovary,³⁴ lung,³⁵ and head and neck.³⁶ Therefore, they proposed that the caveolin-1 gene may be a tumor-suppressor gene candidate, its loss influences cancer cell survival and growth, and favor tumor progression,^37–42 indicating loss of CAV1 regulation is a critical step in the acquisition of transformed phenotype.⁴³ However, in later tumor stages, a re-expression of CAV1 seems to support invasion and metastasis.^44,45 CAV1 overexpression may act as an initiator during the early stage and promote invasion at late stages. Whether or not CAV1 functions as a suppressor during the development of HCC, further investigation is needed to be carried out.

Lipid rafts play a pivotal role in signaling crosstalk through bringing different proteins into proximity and facilitating interactions between receptors and intracellular signaling proteins.²³ CAV-1 function as scaffolding proteins that contribute to the assembly of a complex molecular platform, regulating endocytosis and transcytosis of molecules on the cell surface, and organizing signaling proteins that participate in cell proliferation, adhesion, and migration.^17,46 Our data indicate that the overexpression of CAV-1 in HCC is linked to an unfavorable clinicopathological status, including a low degree of differentiation and metastasis. Previous studies demonstrated that the disruption of CAV-1 expression in invasive HCC cells suppresses in vitro migration and invasion,^18,20,21 and also in vivo tumorigenicity and metastasis.^20,21,47 Our results are in agreement with the previous reports. Present data showed that the rate of CAV-1 protein overexpression was higher in stage III–IV HCC than in stage I-II, but there was no significant difference due to small number of sample.

Previous studies showed controversial results of the relationship between CAV-1 protein overexpression and overall survival due to a small samples size.^{19,20,41,48–50} Yang et al reported that CAV-1 overexpression was correlated with better prognosis. However, our pooled HR from six studies indicate that the CAV-1 overexpression was significantly correlated with poor prognosis in patients with HCC. The high heterogeneity is due to the result from Yang et al’s study. Taken together, CAV-1 protein overexpression can potentially be a prognostic predictor for patients with HCC.

There are a few limitations. First, the criteria for the definition of overexpression of CAV-1 protein varied across the included studies. Second, confounders may exist when looking at the association of CAV-1 overexpression with overall survival since some of the included studies did not adjust for important risk factors such as age, tumor size, and HCC stage.

Conclusions

Our results suggest that the overexpression of CAV-1 protein is associated with the risk of cirrhosis which is an early event of carcinogenesis in the liver and is associated with HCC that was derived from cirrhosis. CAV-1 is correlated with an unfavorable clinicopathological status, including a low degree of differentiation and metastasis. The overexpression of CAV-1 protein is associated with poor overall survival, suggesting that the overexpression of CAV-1 protein predicts a poor prognosis in patients with HCC.