Renal cell carcinoma (RCC), which is the fifth most common cancer worldwide, accounts for 2-3% of all malignant diseases in adults [
CXC chemokine receptor 4 (CXCR4) is one of 19 acknowledged chemokine receptors and belongs to the family of G-protein coupled chemokine receptor [
Two independent authors comprehensively searched the PubMed and EMBASE/Cochrane Library for relevant articles published up to June 1, 2015. The key terms included renal cell carcinoma (“renal cancer” OR “renal cell carcinoma” OR “renal carcinoma” OR “renal tumor”), CXCR4 (“CXCR4” OR “chemokine receptor 4”), prognosis, and survival. The language of articles was limited to English. In addition, we also checked reference lists of identified studies for the other potential eligible trials. This progress was stopped when there were not additional articles.
The eligible studies included in this meta-analysis must met the following criteria to reduce the heterogeneity of articles: (1) patients with distinctive renal cell carcinoma diagnosis by pathology without the limitation of age and gender; (2) using immunohistochemistry method to detect the expression of CXCR4; (3) articles focused on the association of high CXCR4 expression and poor prognosis of patients with RCC; and (4) articles having the hazard ratios (HRs) of overall survival (OS) or progression-free survival (PFS) about CXCR4 expression and survival. The case reports, letters, and expert opinions were excluded. The exclusion criteria of studies also included (1) articles about cell lines or animals; (2) no definition of expression of CXCR4; (3) articles’ lack of original data and control groups; (4) no relevant outcome data of OS or PFS; and (5) repetitive articles.
Relevant characteristics and outcome data were collected by two independent reviewers. The main characteristics of articles were listed as follows: (1) first author’s name; (2) publication year; (3) country; (4) study period; and (5) median follow-up. The relevant clinical data of studies included (1) patients’ number; (2) gender (male/female); (3) age (years); (4) pathological pattern; (5) Fuhrman grade; (6) histologic origin; (7) antibody source; (8) dilution; (9) evaluation method of CXCR4 expression level; and (10) low versus high CXCR4. The HR with 95% confidence interval (CI) was the outcome data. If an article provided the results of univariate and multivariate analyses, we chose the latter. Any disagreement was resolved by discussion. The quality of articles included in this meta-analysis was assessed by Newcastle-Ottawa Scale (NOS).
A pooled analysis of HRs and 95% CIs was used to evaluate the effect of CXCR4 expression on the survival of renal cell carcinoma in this meta-analysis. Subgroup analyses were conducted according to the ethnicity, the proportion of Fuhrman grade III-IV patients, and duration of follow-up. If the number of studies is not insufficient in subgroup analysis, then it will be listed simply. We further conducted sensitivity analyses to detect the possible reasons for heterogeneity and to evaluate the effect of each study on the overall pooled estimate. Chi square (
There were 7 relevant articles finally included in our meta-analysis [
Main characteristics and NOS score of each study included in meta-analysis.
First author | Year | Country | Study period | Median follow-up (year) | NOS score |
---|---|---|---|---|---|
An [ |
2014 | China | 1999–2006 | TS: 5.2 (0.6–9.7) |
9 |
Li [ |
2013 | France | 1999–2005 | 6.6 (1.0–15.3) | 8 |
D’Alterio [ |
2012 | Italy | 2005–2009 | 2.4 | 8 |
Wang [ |
2012 | China | 2002-2003 | NA | 7 |
Li [ |
2011 | China | 2001–2005 | 4.3 (0.2–8.3) | 8 |
D’Alterio [ |
2010 | Italy | 1999–2007 | 5.8 | 7 |
D’Alterio [ |
2010 | Italy | NA | 5.3 | 8 |
TS: training set, VS: validation set, and NA: not available.
Clinical data of studies included in meta-analysis.
First author/ |
Samples | Gender (M/F) | Age (years) | Pathological pattern | Fuhrman stage | Immunohistochemistry | Evaluation of expression level (CXCR4) | Low versus high (CXCR4) | |||
---|---|---|---|---|---|---|---|---|---|---|---|
Histologic origin | Antibody source | Dilution | |||||||||
An, 2014 [ |
TS |
125 |
84/41 |
57.6 ± 12.7 |
ccRCC |
I-II/III-IV; 75/50 |
Tumor sections | Mouse; R&D Systems, USA | 1 : 400 | Five-staged score (0, 1, 2, 3, and 4) | High (scores 0, 1, and 2) versus low (scores 3 and 4) |
|
|||||||||||
Li, 2013 [ |
104 | 69/35 | 64.5 (34–86) | ccRCC | I-II/III-IV; 63/41 | Tissue sections | Rabbit; Abcam, UK | NA | Range of positive cells: high; moderate 25–85%; low <25%; and absence = 0% | ≥85% versus <85% | |
|
|||||||||||
D’Alterio, 2012 [ |
62 | 45/17 | 55 (31–82) | mRCC | Fuhrman grading: |
Histologic sections | Mouse; R&D Systems, USA | 1 : 1000 | The rate of stained (positive) tumor cells: 0–5% low, >5–20% intermediate, and >20% high | Negative/low versus | |
|
|||||||||||
Wang, 2012 [ |
97 | 60/37 | ≤60/>60; 63/34 | ccRCC and others | Fuhrman grading: |
Tissue sections | Mouse; R&D Systems, USA | NA | The 25th percentile value of the average percentage of positive tumor cells | High (≥30%) versus low (<30%) | |
|
|||||||||||
Li, 2011 [ |
117 | 78/39 | ≥60/<60; 54/63 | ccRCC and LARCC | Fuhrman grading: |
Tissue sections | Mouse; R&D Systems, USA | 1 : 100 | Conventional four-tiered semiquantitative scoring system: scores 0–3 for negative, weak, moderate, and strong staining | Positive (+) versus negative (−) | |
|
|||||||||||
D’Alterio, 2010 [ |
223 | 121/102 | ≥70/<70; 99/124 | RCC (chromophobe, conventional, papillary, bellini and other) | Fuhrman grading: |
Histologic sections | Mouse; R&D Systems, USA | 1 : 2000 | The rate of positive tumor cells in 10 high power field (400x)/slide: 0–5% low, >5–20% intermediate, and >20% high | 20%> versus >20% | |
|
|||||||||||
D’Alterio, 2010 [ |
240 | 139/101 | 61 (26–84) | RCC (chromophobe, papillary, clear cell and other) | Fuhrman grading: |
Tissue sections | Mouse; R&D Systems, USA | 1 : 2000 | The rate of positive tumor cells in 10 high power field (400x)/slide: 0–5% low, >5–20% intermediate, and >20% high | 20%> versus >20% |
TS: training set, VS: validation set, M/F: male/female, ccRCC: clear cell RCC, mRCC: metastasis RCC, and LARCC: locally advanced RCC.
Flow diagram for articles included in this meta-analysis.
There were 5 articles involved with 605 patients providing the HR of OS. The result of meta-analysis showed that high CXCR4 expression predicts a poor OS (random effect model (REM) HR = 2.77, 95% CI = 1.80–4.27) with obvious heterogeneity (
The forest plot of HRs for OS with 5 studies included in this meta-analysis.
Forest plot of HRs for PFS with 6 studies included in this meta-analysis.
The plot of result of sensitivity analysis for OS.
The plot of result of sensitivity analysis for PFS.
The association between high expression and poor OS was similar in Asian patients and non-Asian patients when grouped by ethnicity and median follow-up period. When grouped by the proportion of Fuhrman grade III-IV patients with RCC, it seemed that there was no association between high CXCR4 expression and poor OS (HR = 1.48, 95% CI = 0.93–2.37). However, only one article was involved. The results of subgroup analyses were shown in Table
Subgroup analysis based on characteristics of various studies.
Variables | T/P | Overall survival | |||||
---|---|---|---|---|---|---|---|
HR (95% CI) |
|
Model |
|
|
|
||
Overall | 5/605 | 2.77 (1.80, 4.27) | 51.7 | Random effect | 0.066 | 4.61 | <0.001 |
Ethnicity | |||||||
Asian | 3/439 | 3.78 (2.47, 5.78) | 0.0 | Fixed effect | 0.60 | 6.13 | <0.001 |
Non-Asian | 2/166 | 1.68 (1.14, 2.48) | 0.0 | Fixed effect | 0.35 | 2.63 | 0.009 |
Fuhrman grades III-IV (%) | |||||||
>70 | 1/62 | 1.48 (0.93, 2.37) | NA | NA | NA | 1.64 | 0.102 |
≤70 | 4/543 | 3.26 (2.27, 4.67) | 0.0 | Fixed effect | 0.471 | 6.41 | <0.001 |
Median follow-up (months) | |||||||
>60 | 2/329 | 2.70 (1.73, 4.21) | 0.0 | Fixed effect | 0.727 | 4.39 | <0.001 |
≤60 | 3/276 | 3.10 (1.24, 7.80) | 78.6 | Random effect | 0.009 | 2.41 | 0.016 |
T/P: number of trials/number of patients,
The interpretability of publication bias assessed by Begg and Egger tests was limited when only 7 studies were included in this meta-analysis.
Meta-analytical technique is qualitative and quantitative tool to evaluate those subjects which are still controversial. The results of meta-analysis always were regarded as the highest level of evidence. Nowadays, people do not fully understand which factors affect the prognosis of RCC patients. Many studies had been identifying the suitable molecular biological prognostic markers for RCC. Recently, a series of studies focused on the relationship of CXCR4 expression levels and the prognosis of patients with RCC. However, these studies did not achieve consensus. This is the first meta-analysis performed to elucidate the prognosis value of high CXCR4 expression in OS and PFS of RCC patients. The results of our meta-analysis suggested that high CXCR4 expression predicted poor OS and PFS. The CXCR4 may serve as a useful prognosis marker and a therapeutic target for the RCC.
Chemokines are peptide mediators involved in normal development, immune and hematopoietic regulation, inflammation, and wound healing [
A series of meta-analyses had investigated the prognosis value of CXCR4 for patients with other system tumors. Han et al. considered that CXCR4 could help predict prognosis of gastric cancer patients [
However, the results of this meta-analysis need to be interpreted cautiously due to some limitations. First, only 7 relevant articles were only involved with 1068 patients and designed in Asia and Europe. Moreover, studies in other languages were excluded except for English, so language bias may exist in our meta-analysis. Second, there was a significant heterogeneity between these studies. The clinical characteristics of patients in each study such as age, gender, and performance status would lead to bias obviously. Third, the variable histologic type and immunohistochemistry method might affect the accuracy of this meta-analysis. Further researches should be conducted to investigate whether these factors would affect the results of meta-analysis.
The authors declare that there are no other competing financial interests.
This work was supported by the National Science Foundation of China (Grant no. 81372736/H1619).