Effects of FGFR4 G388R, V10I polymorphisms on the likelihood of cancer

The correlation between G388R or V10I polymorphisms of fibroblast growth factor receptor (FGFR) 4 gene and the risk of carcinoma has been investigated previously, but the results are contradictory. Odds ratios (ORs) with 95% confidence intervals (95%CIs), in silico tools, and immunohistochemical staining (IHS) were adopted to assess the association. In total, 13,793 cancer patients and 16,179 controls were evaluated in our pooled analysis. Summarization of all the studies showed that G388R polymorphism is associated with elevated susceptibility to cancer under homozygous comparison (OR = 1.21, 95%CI = 1.03–1.43, P = 0.020) and a recessive genetic model (OR = 1.21, 95%CI = 1.04–1.41, P = 0.012). In the stratification analysis by cancer type and ethnicity, similar findings were indicated for prostate cancer, breast cancer, and individuals of Asian descendant. Polyphen2 bioinformatics analysis showed that the G388R mutation is predicted to damage the protein function of FGFR4. IHS analysis indicated that FGFR4 expression is increased in advanced prostate cancer. These findings may guide personalized treatment of certain types of cancers. Up-regulation of FGFR4 may be related to a poor prognosis in prostate cancer.

Inclusion criteria and exclusion criteria. The inclusion criteria were as follows: (a) investigations of the relationship between FGFR4 G388R or V10I polymorphisms and risk of cancer; (b) cohort or case-control studies; (c) sufficient genotype information to calculate Odds ratios (ORs) and 95% confidence intervals (95%CIs); and (d) P-values greater than 0.05 for Hardy-Weinberg equilibrium (HWE) of controls. Articles that departed from HWE were removed. We also excluded studies with no control population. When repeated studies appeared, only the latest or largest articles were included. Data extraction. Two authors independently searched the articles and extracted data from individual studies according to the inclusion criteria. Information collected from all eligible studies included the name of the first author, publication date, the ethnicity of subjects in the study, source of control, number of genotyped cases and controls, P-values for HWE of controls, and genotyping method. If a type of cancer appeared in only one study, then this cancer was classified in to 'other cancer' group. A total of 37 eligible studies were included. Statistical analysis. We adopted ORs with 95% CI to explore the correlation between FGFR4 G388R or V10I polymorphisms and the risk of cancer. For the G388R variant, five genetic models were used (allelic contrast, R vs. G, heterozygous model, RG vs. GG, homozygous model, RR vs. GG; dominant model, RR + RG vs. GG, recessive model RR vs. RG + GG). For V10I polymorphism, the five models were as follows: I versus V; IV versus VV; II versus VV; II + IV versus VV; II versus IV + VV. The homogeneity of the study was calculated by a chi-square-based Q-test. P-value ≥ 0.05 indicated a lack of heterogeneity; the summary OR was evaluated by the fixed-effects model (Mantel-Haenszel method). Otherwise, the random-effects model was employed. Begg's funnel plot and sensitivity analysis were performed to assess publication bias. Stratification analysis was applied to evaluate the impact of ethnicity and cancer type. All statistical analyses were performed using Stata software (Stata Corporation 2009. Stata Statistical Software: Release 11. College Station, TX: StataCorp LP.).
In silico and IHS analysis of FGFR4 expression. We employed an online database to assess the expression of FGFR4 in prostate cancer and normal tissues (https ://www.cance r.gov/about -nci/organ izati on/ccg/resea rch/struc tural -genom ics/tcga). Moreover, gene-gene interaction of FGFR4 was also evaluated by an online database (http://ualca n.path.uab.edu/analy sis.html). The Cancer Genome Atlas (TCGA) samples were also applied to investigate the effect of FGFR4 expression on overall survival (OS) time (http://genom ics.jeffe rson.edu/progg ene/resul ts.php). The relationship between G388R or V10I polymorphisms and FGFR4 protein damage was analyzed by Polyphen2 tools (http://genet ics.bwh.harva rd.edu/pph2/). Furthermore, we used IHS to test the tissue expression of FGFR4 in prostate cancer subjects recruited by our centers. From February 2013 to July 2018, a total of 220 patients diagnosed with prostate cancer by puncture biopsy were enrolled in our study. These patients underwent radical prostatectomy in our hospitals. Before the IHS analysis, each participant signed an informed consent. In addition to the routine pathological examination, the remaining part of the tissue removed during the operation was used for immunohistochemical examination. Paraffin-embedded samples were stained with hematoxylin and eosin to confirm cancer. Tissue sections were dewaxed in xylene, dehydrated in alcohol and washed in phosphate buffer (PBS). Each slice was incubated overnight with rabbit anti-FGFR4 monoclonal antibody at 4 °C. After being rinsed with PBS for three times, the slices were incubated with secondary antibody at 20 °C for 30 min. PBS was used instead of a primary antibody as a negative control. Two authors evaluated the prostate cancer sections separately. We investigated the intensities of FGFR4 reactivities in different samples utilizing the image J software (Version 1.45, a java-based image analysis program designed by National Institutes of Health, Bethesda, Maryland, USA, Available from: URL: https ://rsb.info.nih.gov/ij/) (range from score 1 to 8) 34

Results
Characteristics of included studies. A total of 269 articles were initially involved based on the inclusion criteria (Supplement Fig. 1 Main results. The overall results showed that the FGFR4 G388R variant was associated with elevated susceptibility to cancer under homozygous comparison (OR = 1.21, 95%CI = 1.03-1.43, P heterogeneity < 0.001, P = 0.020) and recessive genetic modeling (OR = 1.21, 95%CI = 1.04-1.41, P value for heterogeneity < 0.001, P = 0.012, Table 2). The stratification analysis by cancer type revealed that individuals with the RR + RG allele had a 1.20-fold higher susceptibility to prostate cancer than those with the GG allele (95%CI = 1.06-1. 35, P heterogeneity = 0.892, P = 0.004, Fig. 2A). Individuals with the RR + RG allele had a 1.26-fold higher risk of breast cancer than those with the wild type (95%CI = 1.14-1.54, P heterogeneity = 0.197, P < 0.001). In subgroup analysis by ethnicity, we observed that Asian descendants carrying the RR allele had a 1.28-fold increased risk of cancer compared with those carrying the RG + GG allele (95%CI = 1.02-1.60, P heterogeneity < 0.001, P = 0.034, Fig. 3A). However, we did not In silico and IHS analyses of FGFR4 expression. We used in silico tools to investigate whether G388R and V10I mutations affect the protein function of FGFR4. Polyphen2 bioinformatics analysis showed that FGFR4 G388R was predicted to damage protein function, with a score of 0.700 (Fig. 4A). However, the V10I variation was predicted to be benign, with a score less than 0.001 (Fig. 4B). We also utilized an online database to assess the expression of FGFR4 in prostate cancer participants and normal controls. As described in Fig. 5A, FGFR4 expression is elevated in prostate cancer compared with that in the control. TCGA samples were also analyzed to investigate the effect of FGFR4 expression on OS time. No significant difference in OS time was observed between the high FGFR4 expression group and the low expression group (P > 0.05, Fig. 5B).
In order to demonstrate the expression of FGFR4 in prostate cancer tissues, we applied IHS to evaluate its expression among prostate cancer patients at our centers. A total of 220 prostate cancer participants were enrolled in our centers. The feature distribution from prostate cancer volunteers has been provided in our previous       www.nature.com/scientificreports/ article 36 . Immunohistochemistry of FGFR4 in prostate cancer specimens is described in Fig. 6. The intensity of immunoreactivity was mainly concentrated in the cytoplasm of prostate cancer epithelial cells (Fig. 6C,D). The expression of FGFR4 was up-regulated in more advanced cases (Fig. 6D) compared with early stage cases (Fig. 6A,B, P < 0.05). Moreover, the gene-gene correlation of FGFR4 was also assessed. At least 24 genes were shown to participate in interactions with FGFR4 (Fig. 7A). The most related genes to FGFR4 include: CORIN (corin, serine peptidase, Fig. 7B), NKD1 (Naked1, NKD inhibitor 1, Fig. 7C), and CALML3 (calmodulin like 3, Fig. 7D).

Publication bias and sensitivity analysis.
A Begg's funnel plot was employed to investigate publication bias. No evidence of asymmetry was observed for FGFR4 G388R (t = − 1.52, P = 0.140, Fig. 8A) or V10I variants (t = 0.07, P = 0.945, Fig. 8B). Sensitivity analysis of FGFR4 G388R or V10I polymorphisms and the risk of cancer was performed by removing individual studies in turn. No single study influenced the overall OR, indicating that the results of the above analysis for FGFR4 G388R (Fig. 8C) and V10I (Fig. 8D) polymorphisms are reliable.

Discussion
The etiology of cancer has not been totally elucidated. Clinically, SNPs have various influences on the development of diseases, including cancer [37][38][39] . The association between FGFR4 G388R or V10I variants and the susceptibility of cancer has been evaluated previously, but the results are conflicting. For example, Ma et al. performed a case-control study and found that the R-allele of the G388R variant had a significant impact on the development and progression of prostate cancer in Japanese patients 27 . However, this conclusion could not be confirmed by the research of FitzGerald et al., who observed no positive correlation between FGFR4 G388R or V10I polymorphisms and prostate cancer susceptibility in Caucasians or African Americans 23 . Xiong et al. performed a meta-analysis using articles published before October 2016 to assess the effect of the FGFR4 G388R variant 8 . They showed that the G388R variant was correlated with increased susceptibility to prostate and breast cancer and reduced risk of lung cancer. In our analysis, all eligible studies based on inclusion criteria were included to extensively evaluate the association between FGFR4 G388R or V10I variants and the susceptibility of cancer. We further adopted in silico and IHS analysis to confirm the above conclusion. We performed a pooled analysis of studies that included 9416 cancer participants and 11,187 control subjects to investigate the relationship between the FGFR4 G388R variant and susceptibility to cancer. In the current analysis, we found that G388R polymorphism is associated with an elevated risk of cancer. Furthermore, stratifying by type of cancer, we observed that this variant is correlated with prostate and breast cancer, but not with lung cancer. Our results are consistent with those of Wei et al. and Xu et al. 37,40 . In subgroup analysis by ethnicity, we also found that the FGFR4 R-allele is correlated with an increased risk of cancer in individuals with Asian descent. For FGFR4 V10I polymorphism, no significant relationship was indicated in either overall or stratifying analysis. The conclusions derived from our analysis were consistent with a previous meta-analysis published in 2010 41 . In 2017, another meta-analysis found that the FGFR4 388R variation was a reduced risk factor for lung cancer 8 . However, we did not come to this conclusion in the current analysis. The reason may be that there were few studies in our analysis that were focused on lung cancer 15,33 . Therefore, further well-designed studies with large sample sizes are needed to confirm the role of FGFR4 G388R polymorphisms in lung cancer in future. Furthermore, we employed an in silico tool to investigate whether the G388R and V10I mutations could affect the protein function of FGFR4. It showed that the G388R mutation, but not the V10I mutation, could damage the protein function of FGFR4. We further utilized TCGA samples to assess the expression of FGFR4 in prostate cancer participants. The FGFR4 expression was elevated in prostate cancer compared with that in the control group. Nevertheless, no significant difference in the OS time could be identified between the high FGFR4 expression group and the low expression group. In addition, we applied IHS to evaluate its expression among prostate cancer subjects in our centers. The expression of FGFR4 is increased in more advanced cases, which indicates that up-regulation of FGFR4 is related to a poor prognosis of prostate cancer. There are some potential limitations in the present analysis. First, the P-value of heterogeneity was less than 0.05 in five genetic models when all studies were pooled to assess FGFR4 G388R polymorphism. Although the Der Simonian and Laird method (random-effect model) was used 42 , the analysis may have been influenced by potential bias. Second, the number of eligible studies on the FGFR4 V10I variant in the present analysis remains insufficient for a comprehensive analysis. In our subgroup analysis by cancer type, only two studies concentrated  41 . For prostate cancer, further in vitro experiments are required to confirm the results from our pooled analysis. More functional research is warranted to determine whether the G388R mutation is responsible for increased expression of FGFR4. Moreover, genotyping the same patients will provide more persuasive evidence of a correlation between genotype or alleles and tissue expression of FGFR4 observed by IHS analysis. Finally, at least 24 genes are involved in the interaction with FGFR4. Since few studies on these specific relationships can be retrieved from an online database, future studies are needed to ascertain these correlations in more detail. In summary, our study showed that FGFR4 G388R polymorphism is associated with an elevated risk of cancer, especially for prostate and breast cancer. The R-allele of the FGFR4 G388R variant is correlated with an increased risk of cancer in individuals with Asian descent. The G388R mutation, but not the V10I mutation, is predicted to damage the protein function of FGFR4. Up-regulation of FGFR4 may be related to a poor prognosis in prostate cancer. These findings may guide personalized treatment of certain types of cancers.

Data availability
All the data generated in the above research can be acquired from the corresponding authors upon reasonable request. All methods were conducted in accordance with relevant guidelines and regulations.