RESEARCH ARTICLE
CD147 and Prostate Cancer: A Systematic
Review and Meta-Analysis
Fei Peng
1☯
, Hui Li
2☯
, Zhaoze Ning
1
, Zhenyu Yang
3
, Hongru Li
4
, Yonggang Wang
2
,
Fang Chen
1
, Yi Wu
1
*
1 People’s Hospital of Hunan Province, Chang Sha, Hu Nan Province, China, 2 Reproductive Department,
Xiangya Hospital, Central South University, Changsha, China, 3 Urology of Xiangya Hospital, Central South
University, Changsha, China, 4 Xiangya Medical School, Central South University, Changsha, China
☯ These authors contributed equally to this work.
*
wuyi19701210@outlook.com
Abstract
Background
Prostate cancer is one of the most common non-cutaneous malignancies in men. We
aimed to systemically evaluate the relationship between the expression of CD147 in tissues
and the clinicopathological features of prostate cancer.
Methods and Findings
PubMed (1966–2016), EMBASE (1980–2016), the Cochrane Library (1996–2016), Web of
Science (1945–2016), China National Knowledge Infrastructure (1982–2016), and the
WanFang databases (1988–2016) were searched. Literature quality assessment was per-
formed with the Newcastle-Ottawa Scale. Meta-analysis was performed by using Review
Manager 5.3 and Stata 13.0. A total of 12591 prostate cancer patients from 14 studies were
included. The results of the meta-analysis showed that there were significant differences in
the positive expression rate in the following comparisons: prostatic cancer tissues vs. nor-
mal prostate tissues (odds ratio [OR] = 26.93, 95% confidence interval [CI] 7.95–91.20,
P < 0.00001), prostatic cancer tissues vs. benign prostatic hyperplasia tissues (OR = 20.54,
95% CI 8.20–51.44, P < 0.00001), high Gleason score vs. low Gleason score (OR = 2.39,
95% CI 1.33–4.27, P = 0.03), TNM III to IV vs. TNM I to II (OR = 9.95, 95% CI 4.96–19.96,
P < 0.00001), low or moderate differentiation vs. high differentiation (OR = 8.12, 95%
CI 3.69–17.85, P < 0.00001), lymph node metastasis vs. non-lymph node metastasis
(OR = 4.31, 95% CI 1.11–16.71, P = 0.03), and distant metastasis vs. non-distant metasta-
sis (OR = 8.90, 95% CI 3.24–24.42, P < 0.00001).
Conclusion
The CD147 positive expression rate was closely related to the clinical characteristics of
prostate cancer, but more research is needed to confirm the findings owing to the results of
the subgroups.
PLOS ONE | DOI:10.1371/journal.pone.0163678 September 29, 2016 1 / 15
a11111
OPEN ACCESS
Citation: Peng F, Li H, Ning Z, Yang Z, Li H, Wang
Y, et al. (2016) CD147 and Prostate Cancer: A
Systematic Review and Meta-Analysis. PLoS ONE
11(9): e0163678. doi:10.1371/journal.
pone.0163678
Editor: Lucia R. Languino, Thomas Jefferson
University, UNITED STATES
Received: May 8, 2016
Accepted: September 11, 2016
Published: September 29, 2016
Copyright: © 2016 Peng et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: All relevant data are
within the paper and its Supporting Information
files.
Funding: The authors received no specific funding
for this work.
Competing Interests: The authors have declared
that no competing interests exist.
Introduction
The incidence of prostate cancer is the second highest among all malignant tumors in men,
and it is the most common cancer in men in developed countries [
1]. Although the incidence
of prostate cancer in developing countries is lower than in developed countries, it has shown a
continuous rapid increase in recent years[
2]. Prostate cancer prognosis varies significantly
among patients according to clinical stage and pathological grade. Early detection and early
treatment can help improve patient prognosis[
3].
Currently, we mainly use the detection of serum prostate-specific antigen (PSA) to carry out
an assessment for prostate cancer. PSA will increase in most clinically significant cases of pros-
tate cancer. However, we hoped to find a more effective index that has a close relationship with
the clinical features of prostate cancer[
4]. Hence, we performed this meta-analysis.
With the rapid development of molecular biology and immunology techniques in recent
years, tumor markers have played an important role in the differential diagnosis, prognosis
evaluation, and follow-up of malignant tumors[
5].CD147, also known as extracellular matrix
metalloproteinase inducer (EMMPRIN), is a member of the immunoglobulin family that is
expressed on the surface of many types of tumor cells[
6].CD147 has a high expression level in
many malignant tumors and there are significant differences both in the intensity and distribu-
tion of CD147 staining between malignant tumors and benign lesions; moreover, CD147
expression is reported to correlate with the clinical prognosis of patients with some malignant
tumors[7,8,9,10].
To date, there have been some case-control studies that investigated the expression of
CD147 in prostate cancer. They found prostate cancer tissues had a higher positivity rate that
is significantly different from prostatic hyperplasia and normal prostate tissues, and their
authors concluded that the expression of CD147 is related to TNM stage, aggressiveness, dis-
tant metastasis, and prognosis of prostate cancer; however, some studies had different conclu-
sions[11,12,13,14,15]. To provide better evidence of the clinical application of CD147 in
prostate cancer patients, this meta-analysis was conducted to assess the correlation between
CD147 and prostate cancer.
Methods and Materials
Criteria for including studies
1. Published case control study or randomized controlled trial that provides original data about
CD147 and prostate cancer with clinical pathological characteristics; 2. All cases had complete
clinical and pathological data, without radiotherapy or chemotherapy before sampling; 3. Path-
ological sections were all studied, and CD147 was detected by immunohistochemical staining;
4. When there was duplicate publication or similar information, the best quality study was
retained.
Criteria for excluding studies
1. Animal experiments; 2. The standard of pathological diagnosis was not clear; 3. CD147 was
not detected by immunohistochemical staining; 4. A duplicated report, a review, or a case
report.
Search Strategy
We searched PubMed (1966–2016), EMBASE (1980–2016), the Cochrane Library (1996–
2016), Web of Science (1945–2016), China National Knowledge Infrastructure (1982–2016),
and the WanFang databases (1988–2016). The trials were restricted to humans, but not by
CD147 and Prostate Cancer
PLOS ONE | DOI:10.1371/journal.pone.0163678 September 29, 2016 2 / 15
date, language, or publication status. The following combined search term was used: (prostate,
Prostatic, Prostat
, Prostatomegaly) AND (CD147, extracellular matrix metalloproteinase
inducer, EMMPRIN). 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 File.
Quality Evaluation
The Newcastle-Ottawa quality assessment scale of case control studies (NOS) [16] was adopted
to assess the quality of included studies. It has three categories (selection, comparability, and
exposure) and eight items. Two researchers performed the quality assessments separately. In
the selection category (adequate definition of the cases, representativeness of the cases, selec-
tion of controls, definition of controls) and exposure category (ascertainment of exposure,
same method of ascertainment for cases and controls, non-response rate), a quality research
item received one star, and a comparable category (comparability of cases and controls on the
basis of the design or analysis) could receive at most two stars. The quality assessment values
ranged from 0 to 9 stars. Each band indicates the percentage of the included studies that met
each of these quality criteria.
Statistical Analysis
Records retrieved from the initial search were independently scanned by two authors to
exclude clearly irrelevant studies. Then, the full text articles were independently reviewed by
two authors to see if they met the inclusion criteria, and differences of opinion were resolved
by a third author. All of the data were extracted independently by two authors. The corre-
sponding author of each study was contacted to provide information on missing or incomplete
data. The software Revman 5.3 and Stata 13.0 were used to analyze the data. Results were
expressed as odds ratios (OR) and 95% confidence intervals (95% CI). A fixed-effects model
was adopted in the case of no evidence of significant heterogeneity (P > 0.1 and I
2
< 50%); oth-
erwise, a random-effects model was used. If possible, heterogeneity was explored and subgroup
analyses were performed. Subgroup analyses were conducted based on patient age and study
area. If heterogeneity could not be explored, we conducted a sensitivity analysis to identify the
study with the most heterogeneity.
Sensitive analysis was also performed to evaluate the influences of individual studies on the
final effect size. When some studies were omitted or subgroup analyses were performed, if no
decreases in heterogeneity were observed, a qualitative systematic review method was used to
describe the results. All P values were 2-sided, and P < 0.05 was considered significant. 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.[17] Then, we replicated the funnel plot with their ‘‘miss-
ing” counterparts around the adjusted summary estimate.
Literature Search
A total of 259 studies were identified, and 95 studies were excluded because of duplication.
After reading the titles and abstracts, 114 studies were excluded. Fifty full text studies were
carefully reviewed (excluded for being animal studies [n = 19]; reviews and meta-analyses
[n = 1]; reported CD147 mRNA expression [n = 1]; no control group [n = 2]; and completely
irrelevant [n = 12]). Finally, 15 trials were included for qualitative analysis and 14 trials were
included for quantitative analysis (
Fig 1).
CD147 and Prostate Cancer
PLOS ONE | DOI:10.1371/journal.pone.0163678 September 29, 2016 3 / 15
Characteristics and Risk of Bias of Included Studies
A total of 15 studies [
9,10,11,12,13,14,15,18,19,20,21,22,23,24,25] were identified for qualitative
analysis. The study by Bauman et al. was not included in the quantitative analysis because it
provided the mean optical density of CD147 and not the positive expression rate of CD147.
Therefore, 14 studies [
9,10,11,12,13,14,18,19,20,21,22,23,24,25] were included for quantitative
analysis, including 12,591 prostate cancer tissues, 298 benign prostatic hyperplasia tissues, and
175 normal prostate tissues. Six articles provided a positivity rate of CD147 expression in pros-
tate cancer and normal prostate tissues, 8 articles reported on CD147 expression in prostate
cancer and benign prostatic hyperplasia, 12 articles involved Gleason score, 6 articles involved
Fig 1. Flow diagram of the literature filtering,14 studies are identified for analysis.
doi:10.1371/journal.pone.0163678.g001
CD147 and Prostate Cancer
PLOS ONE | DOI:10.1371/journal.pone.0163678 September 29, 2016 4 / 15
TNM stage, 5 articles involved differentiation, 6 articles involved lymph node metastasis, and 2
articles involved distant metastasis.
Eleven of the studies were performed in Asia and 3 elsewhere. All 14 studies adopted immu-
nohistochemistry (IHC) as the detection method, but the method for judging negative and pos-
itive staining was different among studies. For example, Wang used an IHC score. However,
Nelma Pertega-Gomes, Grupp K, and Weide Zhong used cell staining intensity (Nelma Per-
tega-Gomes and Grupp K’s cut-off value was 0%, while Weide Zhong’s was 5%). The source
of antibodies in all included studies was primarily three places (Fourth Military Medical Uni-
versity, Sigma, and ZYMED). The tissue slice thickness across the studies had some minor
differences.
We used the NOS scale to evaluate the literature, and all of the studies had a score greater
than 5, indicating that the quality of the literature is high. The statement “High CD147 expres-
sion” indicated that the positive rate of CD147 expression was high in the population and
“Low CD147 expression” indicated that the positive rate of CD147 expression was low in the
population.
We performed a subgroup analysis according to three aspects: country, antibody source,
and positivity scoring system because these three aspects may affect the results. The character-
istics of the studies are presented in
Table 1 and the NOS results can be seen in Table 2.
CD147 in prostate cancer and normal prostate tissues
Six studies [9,13,14,18,22,23] reported the positivity rate of CD147 in prostate cancer tissues
and normal prostate tissues, including 914 prostate cancer tissues and 163 normal prostate tis-
sues. There was significant heterogeneity (P = 0.01, I
2
= 67%), and a random-effects model
showed the CD147 positive expression rate in prostate cancer tissues was higher than that in
normal prostate tissues (OR = 26.93, 95% CI = 7.95–91.20, P < 0.00001) (
Fig 2A). To further
confirm the stability of the result, a sensitivity analysis was performed to evaluate the influences
of individual studies on the final effect. The sensitivity analysis showed that, irrespective of
which study was removed, the result was almost the same (
S1 Fig), indicating that the result
was stable. In addition, subgroup analysis found the CD147 positive expression rate was signifi-
cantly different between prostate cancer tissues and normal prostate tissues in all groups (
S1
Table
). Overall, the CD147 positive expression rate in prostate cancer tissues was higher than
that in normal prostate tissues.
CD147 with prostate cancer and benign prostatic hyperplasia tissues
Eight trials [
9,11,12,18,19,20,21,24] reported the difference of CD147 positivity rate between
prostate cancer and benign prostatic hyperplasia, including 575 prostate cancer tissues and 291
benign prostatic hyperplasia tissues. A random-effects model showed the CD147 positive
expression rate in prostate cancer tissues was higher than that in benign prostatic hyperplasia
tissues (OR = 20.54, 95% CI = 8.20–51.44, P < 0.00001) with significant heterogeneity
(P = 0.0009, I
2
= 71%) (Fig 2B). A sensitivity analysis was performed to evaluate the influences
of individual studies on the final effect, which showed that irrespective of which study was
ruled out, the result was almost the same (
S2 Fig), indicating that the result was stable. Sub-
group analysis showed that the CD147 positive expression rate was significantly different
between prostate cancer tissues and benign prostatic hyperplasia tissues except in the group of
other antibodies (
S1 Table). Therefore, more research is needed to confirm whether the CD147
positive expression rate is higher in prostate cancer tissues than in benign prostatic hyperplasia
tissues.
CD147 and Prostate Cancer
PLOS ONE | DOI:10.1371/journal.pone.0163678 September 29, 2016 5 / 15