Prostate Cancer Risk in Relation to a Single Nucleotide Polymorphism in the Insulin-like Growth Factor-binding Protein-3 (IGFBP3) Gene: a Meta-analysis

As one of the most prominent public health problems in the western, prostate cancer (PCa) was estimated to have claimed 33,720 deaths in the United States in 2011, ranking the second leading cause of cancer death (Brawley, 2012). In China, PCa is more and more concerned ,for its growing incidence rate leaping from 1.6/105 PY (person per year) in 2002 to 4.3/105 PY in 2008 (Zhang et al., 2011). Although increasing risk factors such as age, family history of the disease, and race/ethnicity have been identified , the etiology of prostate cancer is still complex and elusive. One of the factors involved is insulin-like growth factor binding protein-3 ( IGFBP3), which regulates IGFs bioavailability to facilitate or inhibit IGF—IGF receptor interaction via binding to circulating IGFs (CollettSolberg et al., 1996; Kelley et al., 1996). Some studies have demonstrated that decreased circulating IGFBP3 concentration portends higher cancer risk including breast, colorectal, lung and gastric cancer (Hankinson et al., 1998;


Introduction
As one of the most prominent public health problems in the western, prostate cancer (PCa) was estimated to have claimed 33,720 deaths in the United States in 2011, ranking the second leading cause of cancer death (Brawley, 2012). In China, PCa is more and more concerned ,for its growing incidence rate leaping from 1.6/105 PY (person per year) in 2002 to 4.3/105 PY in 2008 (Zhang et al., 2011). Although increasing risk factors such as age, family history of the disease, and race/ethnicity have been identified , the etiology of prostate cancer is still complex and elusive.
One of the factors involved is insulin-like growth factor binding protein-3 ( IGFBP3), which regulates IGFs bioavailability to facilitate or inhibit IGF-IGF receptor Ye-Qing Mao, Xin Xu, Yi-Wei Lin, Hong Chen, Zheng-Hui Hu, Xiang-Lai Xu, Yi Zhu, Jian Wu, Xiang-Yi Zheng, Jie Qin, Li-Ping Xie* Ma et al., 1999;Yu et al., 1999;Pham et al., 2007), and the individual variation of gene expression level may largely be attributed to genetic factors. IGFBP3 A-202C polymorphisms, an A-C transversion which is located 202 bp upstream of the transcription start site of IGFBP3, has been confirmed to be associated with basal promoter activity both in vitro and in vivo (Rohrbacher et al., 2005;Wagner et al., 2005), The [A] possessing stronger promoter activity yields higher IGFBP3 gene expression, while the [C] allele or A-202C leads to be a lower one (Deal et al., 2001;D'Aloisio et al., 2003;Costalonga et al., 2009).
Recently, more and more studies have focused on A-202C polymorphism and cancer susceptibility. As for PCa, the results are conflicting. The inconsistency might come from various study design, sample size , recruitment criteria or insignificant effect of polymorphisms. Therefore, it is necessary to perform a meta-analysis reviewing all the published case-control studies to reach a more reliable conclusion on the relation between IGFBP3 A-202C polymorphisms and PCa susceptibility.

Literature search
All the publications until Sep.20 2012 in PubMed, Scopus, Web of Science and Chinese National Knowledge Infrastructure (CNKI) were identified with the search terms'IGFBP3' or 'insulin-like growth factor-binding protein-3', 'polymorphism','variants', 'variation' and 'prostate' with restriction of 'Human'. The potentially associated articles as well as their bibliographies were read in full text or abstract to assess the appropriateness.

Inclusion criteria
The eligible studies should be case-control ones pertaining to IGFBP3 A-202C polymorphisms and PCa, with sufficient data for odds ratio (OR) or relative risk (RR) and 95% confidence interval (CI) calculation. All the eligible studies with full text articles were retrieved.

Data extraction
The following data were carefully extracted from every identified article independently by two authors including: first author's name, publication year, ethnicity, subject source, number of cases and controls, IGFBP3 A-202C genotypes distribution frequency. Necessary data for calculation in two articles were retrieved by email, if omitted by the authors . Ethnicity covered in this paper was classified as 'Caucasian', 'Asian', and 'Mixed' which could be further divided in subgroup analysis. Populationbased and hospital-based studies were two kinds of subject source.

Other results
To examine the publication bias, Begg's and Egger's tests for the alleles comparison were performed with a Begg's funnel plot (Figure 3) provided for visual judgment. Both tests revealed no publication bias in this analysis (Begg's z=0.44 p=0.66, Egger's t=0.84 p=0.42), and no significant asymmetry was found in the funnel plot. To explore whether the ORs were sufficiently robust under various genetic model and contrasts, the sensitivity tests were performed where the remaining studies were pooled after every single one was deleted; The results showed that none of the studies could considerably affect the overall risk estimates in our meta-analysis (data were not shown).

Discussion
This meta-analysis including 9,238 cases and 8,741 controls represents the largest study to date investigating the association between IGFBP3 A-202C polymorphisms and PCa susceptibility as far as we know. Our results revealed that C allele/[CC] genotype were slightly more frequent than A allele/[AA] genotype at IGFBP3 A-202C SNP site and A-202C is a potential risk factor for PCa, which was especially more prominent within 'populationbased control' and 'large' subgroups with negligible heterogeneity. This finding is generally in line with some former reports, but in discrepancy with Li's result derived from a smaller sample size and a sole stratification (Li et al., 2010). An increased cancer risk of C allele carriers among the PBC groups but rather among HBC ones could be attributed to suboptimal representativeness of hospital controls with potential disease conditions involving the SNP polymorphisms under investigation and potential biases producing significant heterogeneity. One available population-based study including 2,626 cases and 2,876 controls screened from seven well-established cohort studies as the largest weight in our analysis swayed the overall calculation to some extent (Schumacher et al., 2010). Hence, a large population-based control is more reliable in meta-analysis. For race stratification, we didn't find any regular genotype distribution or association between different races especially in Africans Despite a comprehensive study with substantial data and insignificant publication bias, there were still some limitations in our study: First, heterogeneity of various levels existed among most subgroups and genetic models, which meant some heterogeneity factors were yet to be analyzed. One of the reasons might come from inconformity of raw data that should be adjusted by age, smoking status, drinking status, obesity, and environmental/ lifestyle factors. Second, unavailable details of race subdistribution in two studies prevented themselves from inclusion for subgroup analysis, which lead to insufficient samples in Africans and Asians subgroups compared with Caucasians (Nam et al., 2003;Li et al., 2004). Besides, it should be noteworthy that our conclusion actually owed much to Safarinejad's report (Safarinejad et al., 2011) with the most prominently positive result of all. While most other included studies yielded insignificant results, which meant our conclusion was seemingly less robust. Recently, it has been reported that IGFBP3 as a multifunctional anti-proliferative protein gets involved in benign prostatic hyperplasia (BPH) development in a similar way with PCa (Neuhouser et al., 2008;Safarinejad et al., 2011), suggesting that it was possible for some BPH cases to be improperly grouped as controls, not to mention the asymptomatic or underdiagnosed PCa cases. On the other hand, the widely accepted hypothesis that circulating concentration of IGFBP3 runs inversely with PCa risk has been more and more challenged by case-control studies (Severi et al., 1999;Li et al., 2004;Hong et al., 2008). Recently, one study has focused on intracellular level of instead of circulating level of IGFBP3 and identified a high expression of IGFBP3 in nucleus as a poor prognostic biomarker (Seligson et al., 2012). If it is further strengthened, the significance of IGFBP3 A-202C should be re-defined.
In conclusion, our meta-analysis indicated that there is a marginal association between IGFBP3 A-202C polymorphisms and PCa risk. However, further studies using well-defined large-scale controls, adjusted data should be carried out critically with more detailed stratifications. Only in this way, a more comprehensive and insightful understanding of the IGFBP3 A-202C polymorphism could be obtained.