Association between GPER gene polymorphisms and GPER expression levels with cancer predisposition and progression.

A meta-analysis was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [ 21 ]. A literature search was conducted in MEDLINE and EMBASE using keywords such as “GPER/GPR30”, “polymorphisms”, “immunohistochemistry”, “expression/level”, and “cancer”, singularly and in combination. The literature search was updated until July, 2020. The inclusion criteria of studies were as follows: (1) evaluating the association between GPER rs3808350, rs11544331, and rs3808351 polymorphisms and cancer predisposition, (2) conducted with a case-control design, and (3) evaluating GPER expression level (immunohistochemistry) and cancer progression. Data were extracted as follows: (1) name of the first author, (2) year of publication, (3) type of cancer, (4) the number of cases and controls, (5) number of genotypes in cases and controls, (6) number of haplotypes of rs3808350/rs3808351/rs11544331 in cases and controls, and (7) number of patients with GPER+/– or high/low. Meta-analysis for each gene polymorphism was performed for two or more studies, as previously described [ 3 , 4 , 5 , 6 , 22 , 23 , 24 , 25 ]. Genotypic frequency of GPER gene polymorphism was tested for deviation from the Hardy–Weinberg equilibrium (HWE) in the control subjects if HWE was not reported. The genetic association was examined using different genetic models, including allelic (a vs. A), recessive (aa vs. Aa + AA), dominant (aa + Aa vs. AA), over dominant (Aa vs. aa + AA), homozygous (aa vs. AA), and heterozygous (Aa vs. AA) models [ 5 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 ]. The associations between GPER gene polymorphisms or GPER expression levels with cancer predisposition and progression were calculated by the pooled odds ratio (OR) and 95% confidence interval (CI). Heterogeneity among studies was evaluated using Q test and I 2 statistic. A significant Q-statistic ( p < 0.10) indicated heterogeneity across studies. The I 2 values indicated no (0–24.9%), low (25–49.9%), moderate (50–74.9%), or high (75–100%) heterogeneity. The random-effect model (REM) was used if heterogeneity existed; otherwise, the fixed-effect model (FEM) was used [ 31 , 32 , 33 , 34 , 35 , 36 , 37 ]. Subgroup analysis was conducted by stratifying based on ethnicity, type of cancer, and localization of GPER expression. In addition, we also evaluated the association between rs3808351 and tumor size, as well as the involvement of haplotypes rs3808350/rs3808351/rs11544331 with cancer predisposition. Potential publication bias was assessed by Begg's funnel plots and Egger's regression test. Begg's funnel plot was applied if the pooled effect size consisted of 10 or more studies. The Newcastle Ottawa Scale (NOS) was adopted to assess the quality of the case-control study, with a score of 8–9 for all included studies, indicating a low risk of bias ( Supplementary Table 1 ). A sensitivity analysis was performed by sequentially omitting each study one at a time, and the results remained unchanged (data not shown), implying the robustness and stability of the findings. A quantified result of p < 0.05 was indicative of statistical significance.

For GPER gene polymorphisms, a total of 142 articles were screened, among which 11 were reviewed. Six studies were excluded due to not relating to cancer or GPER rs3808350, rs11544331, and rs3808351 polymorphisms. Five studies were then included in this meta-analysis [ 10 , 13 , 16 , 17 , 38 ]. From 5 studies, Chevalier et al. [ 10 ] and Giess et al. [ 17 ] recruited testicular and breast cancer patients, respectively, while Kasap et al. [ 16 ] and Hong et al. [ 13 ] enrolled patients with uterine leiomyoma and adenomyosis/uterine leiomyoma/another precancerous lesion of uterine-cervix, respectively. The last included study recruited gynecomastia patients [ 38 ], and although it should be noted that some reports have classified the condition as a non-malignant male breast disorder [ 39 ], gynecomastia has shown strong association with GPER [ 38 ], exhibiting a nearly 10-fold increased risk of breast cancer in men [ 40 ]. A total of 1,288 (case: 601, control 687), 5,565 (case: 729, control: 4,836), and 1,294 (case: 610; control: 684) subjects for GPER rs3808350, rs11544331, and rs3808351 polymorphisms, respectively, were further analyzed. All studies complied with the HWE except for the study from Chevalier et al. (for rs11544331 and rs3808351) [ 10 ]. Details of the retrieved studies are shown in Table 1 . Table 1 Characteristics of individual studies for the association between GPER polymorphisms and cancer. Table 1 No Author (year) Disease Country/Ethnicity Sample size SNP Definition of allele ∗ p HWE Genotype distribution Case Control Ref. Alt. Case Control AA Aa aa AA Aa aa 1 Korkmaz et al (2014) Gynecomastia Turkey/Asian 109 104 rs3808350 A G 0.5295 37 41 31 44 45 15 2 Chevalier et al (2014) Testicular cancer France/Caucasian 89 224 0.6146 45 41 3 82 110 32 3 Hong et al (2019) Adenomyosis Korea/Asian 35 34 0.2115 8 12 15 4 20 10 4 Giess et al (2010) Breast cancer Germany/Caucasian 257 247 0.4092 100 121 36 96 111 40 5 Kasap et al (2016) Uterine leiomyoma Turkey/Asian 111 78 0.8309 41 33 37 27 37 14 1 Korkmaz et al (2014) Gynecomastia Turkey/Asian 109 104 rs11544331 C T 0.0873 61 43 5 66 37 1 2 Chevalier et al (2014) Testicular cancer France/Caucasian 223 4,374 0.0005 78 43 2 2843 1321 210 3 Hong et al (2019) Adenomyosis Korea/Asian 35 34 1 33 2 0 34 0 0 4 Giess et al (2010) Breast cancer Germany/Caucasian 251 246 0.5145 146 88 18 128 96 22 5 Kasap et al (2016) Uterine leiomyoma Turkey/Asian 111 78 0.5106 44 45 22 23 36 19 1 Korkmaz et al (2014) Gynecomastia Turkey/Asian 109 104 rs3808351 G A 0.4980 36 46 27 52 41 11 2 Chevalier et al (2014) Testicular cancer France/Caucasian 100 222 0.0031 16 69 15 96 114 12 3 Hong et al (2019) Adenomyosis Korea/Asian 35 34 0.0989 27 7 1 19 15 0 4 Giess et al (2010) Breast cancer Germany/Caucasian 255 246 0.0545 133 99 23 130 89 27 5 Kasap et al (2016) Uterine leiomyoma Turkey/Asian 111 78 0.1432 57 45 9 28 32 18 Alt., alternative allele; Ref., reference allele; SNP, single nucleotide polymorphism; ∗ p for Hardy–Weinberg equilibrium test in controls; A, Wild type; a, mutant type. Bold values indicate statistically significant p < 0.05. Characteristics of individual studies for the association between GPER polymorphisms and cancer. Alt., alternative allele; Ref., reference allele; SNP, single nucleotide polymorphism; ∗ p for Hardy–Weinberg equilibrium test in controls; A, Wild type; a, mutant type. Bold values indicate statistically significant p < 0.05. The pooled result of the analyses is shown in Table 2 . Overall, there was no significant association between GPER rs3808350, rs11544331, and rs3808351 polymorphisms with cancer predisposition in all inheritance models, even when the studies evaluating gynecomastia or/and study deviated from HWE were excluded ( Table 2 ). However, subgroup analyses stratified by ethnicity revealed a significant association between rs3808350 (G vs. A, OR = 1.38, 95%CI = 1.06–1.79, p = 0.015; GG vs. AG + AA, OR = 2.20, 95%CI = 1.42–3.43, p = 0.000 or OR = 2.11, 95%CI = 1.19–3.74, p = 0.010; GG vs. AA, OR = 1.83, 95%CI = 1.10–3.04, p = 0.019; AG vs. AA, OR = 0.51, 95% = CI 0.28–0.95, p = 0.033; Table 2 ) and rs3808351 (A vs. G, OR = 0.51, 95%CI = 0.34–0.75, p = 0.000; AA vs. GA + GG, OR = 0.34, 95%CI = 0.14–0.78, p = 0.011; AA + GA vs. GG, OR = 0.48, 95%CI = 0.29–0.81, p = 0.006; AA vs. GG, OR = 0.28, 95%CI = 0.11–0.69, p = 0.005; GA vs. GG, OR = 0.56, 95% = CI 0.32–0.98, p = 0.043; Table 2 ) with cancer predisposition. Ethnicity did not associate with predisposition of cancer for rs11544331 (data not shown). In addition, no association was also observed in any haplotypes of rs3808350/rs3808351/rs11544331 with cancer predisposition ( Table 3 ). Table 2 Meta-analysis for the association between GPER polymorphisms and cancer. Table 2 SNP Genetic model Group No. of studies Test of association Stat. Model Test of heterogeneity Publication bias p -value (Egger's test) OR 95% CI p -value p -value I 2 (%) rs3808350 G vs. A Overall 5 1.02 [0.72; 1.45] 0.888 Random 0.003 74.50 0.815 Overall∗ 4 0.91 [0.64; 1.29] 0.604 Random 0.027 67.14 0.943 Asian 3 1.38 [1.06; 1.79] 0.015 Fixed 0.548 0 0.357 Asian∗ 2 1.22 [0.86; 1.74] 0.252 Fixed 0.595 0 NA Caucasian 2 0.74 [0.44; 1.24] 0.268 Random 0.025 79.89 NA GG vs. AG + AA Overall 5 1.20 [0.59; 2.45] 0.602 Random 0.001 77.16 0.840 Overall∗ 4 0.99 [0.43; 2.31] 0.996 Random 0.003 77.71 0.837 Asian 3 2.20 [1.42; 3.43] 0.000 Fixed 0.902 0 0.057 Asian∗ 2 2.11 [1.19; 3.74] 0.010 Fixed 0.700 0 NA Caucasian 2 0.47 [0.12; 1.81] 0.273 Random 0.036 77.15 NA GG + AG vs. AA Overall 5 0.90 [0.71; 1.13] 0.379 Fixed 0.114 46.18 0.584 Overall∗ 4 0.81 [0.63; 1.05] 0.121 Fixed 0.237 29.12 0.403 Asian 3 1.06 [0.72; 1.57] 0.753 Fixed 0.221 33.66 0.359 Asian∗ 2 0.79 [0.46; 1.38] 0.424 Fixed 0.342 0 NA Caucasian 2 0.77 [0.44; 1.34] 0.359 Random 0.068 69.93 NA GG vs. AA Overall 5 0.95 [0.44; 2.04] 0.913 Random 0.003 74.06 0.583 Overall∗ 4 0.73 [0.32; 1.68] 0.468 Random 0.021 69.04 0.558 Asian 3 1.83 [1.10; 3.04] 0.019 Fixed 0.355 3.36 0.246 Asian∗ 2 1.43 [0.72; 2.86] 0.302 Fixed 0.314 1 NA Caucasian 2 0.42 [0.08; 2.05] 0.287 Random 0.018 82.03 NA AG vs. AA Overall 5 0.84 [0.65; 1.08] 0.182 Fixed 0.216 30.81 0.132 Overall∗ 4 0.80 [0.61; 1.05] 0.114 Fixed 0.171 40.05 0.080 Asian 3 0.74 [0.48; 1.15] 0.186 Fixed 0.168 43.81 0.403 Asian∗ 2 0.51 [0.28; 0.95] 0.033 Fixed 0.396 0 NA Caucasian 2 0.89 [0.66; 1.21] 0.484 Fixed 0.183 43.50 NA rs11544331 T vs. C Overall 5 0.91 [0.76; 1.08] 0,299 Fixed 0.204 32.48 0.283 Overall∗∗ 3 0.80 [0.63; 1.01] 0,064 Fixed 0.462 0 0.389 TT vs. CT + CC Overall 4 0.76 [0.49; 1.19] 0,244 Fixed 0.234 29.66 0.662 Overall∗∗ 2 0.77 [0.48; 1.24] 0,295 Fixed 0.966 0 0.265 TT + CT vs. CC Overall 5 0.93 [0.75; 1.16] 0,555 Fixed 0.205 32.41 0.502 Overall∗∗ 3 0.76 [0.56; 1.03] 0,080 Fixed 0.398 0 0.543 TT vs. CC Overall 4 0.68 [0.42; 1.09] 0,114 Fixed 0.214 32.91 0.522 Overall∗∗ 2 0.66 [0.40; 1.11] 0,122 Fixed 0.748 0 0.723 CT vs. CC Overall 5 0.97 [0.78; 1.22] 0,854 Fixed 0.252 25.31 0.606 Overall∗∗ 3 0.78 [0.56; 1.08] 0,136 Fixed 0.418 0 0.532 rs3808351 A vs. G Overall 5 1.07 [0.61; 1.87] 0.809 Random 0.000 89.43 0.657 Overall∗∗ 3 0.68 [0.41; 1.12] 0.135 Random 0.032 70.73 0.481 Asian 3 0.83 [0.30; 2.27] 0.716 Random 0.000 90.93 0.732 Asian∗ 2 0.51 [0.34; 0.75] 0.000 Fixed 0.974 0 NA Caucasian 2 1.44 [0.65; 3.17] 0.364 Random 0.000 92.33 NA AA vs. GA + GG Overall 5 1.27 [0.49; 3.29] 0.618 Random 0.000 82.14 0.817 Overall∗∗ 3 0.60 [0.37; 0.97] 0.040 Fixed 0.103 55.95 0.903 Asian 3 1.14 [0.17; 7.38] 0.886 Random 0.000 86.74 0.970 Asian∗ 2 0.34 [0.14; 0.78] 0.011 Fixed 0.174 45.90 NA Caucasian 2 1.53 [0.41; 5.71] 0.526 Random 0.007 85.87 NA AA + GA vs. GG Overall 5 1.16 [0.56; 2.38] 0.680 Random 0.000 87.44 0.882 Overall∗∗ 3 0.66 [0.36; 1.21] 0.182 Random 0.054 65.68 0.219 Asian 3 0.77 [0.26; 2.24] 0.640 Random 0.000 85.67 0.593 Asian∗ 2 0.48 [0.29; 0.81] 0.006 Fixed 0.571 0 NA Caucasian 2 1.98 [0.52; 7.50] 0.313 Random 0.000 93.24 NA AA vs. GG Overall 5 1.56 [0.44; 5.51] 0.484 Random 0.000 88.31 0.798 Overall∗∗ 3 0.54 [0.19; 1.57] 0.263 Random 0.069 62.47 0.995 Asian 3 1.12 [0.12; 9.73] 0.916 Random 0.000 89.02 0.988 Asian∗ 2 0.28 [0.11; 0.69] 0.005 Fixed 0.210 36.18 NA Caucasian 2 2.42 [0.28; 20.91] 0.419 Random 0.000 93.40 NA GA vs. GG Overall 5 1.15 [0.61; 2.19] 0.653 Random 0.000 82.10 0.728 Overall∗∗ 3 0.73 [0.40; 1.32] 0.307 Random 0.081 60.15 0.001 Asian 3 0.78 [0.33; 1.81] 0.564 Random 0.020 74.22 0.402 Asian∗ 2 0.56 [0.32; 0.98] 0.043 Fixed 0.243 26.59 NA Caucasian 2 1.93 [0.59; 6.31] 0.271 Random 0.000 90.87 NA ∗analysis by excluding Korkmaz et al (2014); ∗∗analysis by excluding Korkmaz et al (2014) and Chevalier et al (2014); CI. confidence interval; OR. odds ratio; Stat. model, statistical model. Bold values indicate statistically significant differences between cases and control, p < 0.05. Table 3 Characteristics of individual studies and meta-analysis for the association between rs3808350/rs3808351/rs11544331 haplotypes and cancer risk. Table 3 No Author (year) Haplotypes Case Control OR (95% CI) [Random] p -value Events Total Events Total 1 Korkmaz et al (2014) AGC 70 109 87 104 0.55 (0.23–1.34) 0.193 2 Kasap et al (2016) 60 111 45 78 1 Korkmaz et al (2014) AGT 9 109 12 104 1.00 (0.58–1.73) 0.990 2 Kasap et al (2016) 30 111 18 78 1 Korkmaz et al (2014) AAC 25 109 28 104 0.59 (0.29–1.19) 0.143 2 Kasap et al (2016) 11 111 17 78 1 Korkmaz et al (2014) GGC 31 109 34 104 1.87 (0.35–0.89) 0.458 2 Kasap et al (2016) 44 111 10 78 1 Korkmaz et al (2014) GGT 8 109 12 104 0.86 (0.49–1.50) 0.598 2 Kasap et al (2016) 25 111 17 78 1 Korkmaz et al (2014) AAT 11 109 6 104 1.13 (0.50–2.56) 0.755 2 Kasap et al (2016) 14 111 12 78 1 Korkmaz et al (2014) GAC 39 109 20 104 1.64 (0.76–3.54) 0.205 2 Kasap et al (2016) 18 111 12 78 1 Korkmaz et al (2014) GAT 25 109 9 104 1.09 (0.14–8.40) 0.930 2 Kasap et al (2016) 20 111 28 78 CI, confidence interval; OR, odds ratio. Meta-analysis for the association between GPER polymorphisms and cancer. ∗analysis by excluding Korkmaz et al (2014); ∗∗analysis by excluding Korkmaz et al (2014) and Chevalier et al (2014); CI. confidence interval; OR. odds ratio; Stat. model, statistical model. Bold values indicate statistically significant differences between cases and control, p < 0.05. Characteristics of individual studies and meta-analysis for the association between rs3808350/rs3808351/rs11544331 haplotypes and cancer risk. CI, confidence interval; OR, odds ratio. In addition to the association of GPER polymorphism with cancer predisposition, we also evaluated the association between rs3808351 and tumor size ( Table 4 ). The analysis showed that rs3808351 (AA + GA vs. GG, OR = 0.46, 95%CI = 0.28–0.76, p = 0.002; GA vs. GG, OR = 0.46, 95% = CI 0.27–0.79, p = 0.004; Table 5 ) was associated with smaller tumor size. Table 4 Characteristics of individual studies for the association between rs3808351 and tumor size. Table 4 No Author (year) Sample size SNP Definition of allele ∗ p HWE Genotype distribution ≥ T2 < T2 Ref. Alt. ≥ T2 < T2 GG GA AA GG GA AA 1 Chevalier et al (2014) 56 56 rs3808351 G A 0.086 2 9 6 8 32 12 2 Giess et al (2010) 104 246 0.9729 67 30 7 61 64 17 Alt., alternative allele; Ref., reference allele; SNP, Single nucleotide polymorphism. ∗ p for Hardy–Weinberg equilibrium test in controls. Table 5 Meta-analysis for the association between rs3808351 and tumor size. Table 5 SNP Genetic model No. of studies Test of association Stat. Model Test of heterogeneity Publication bias p -value (Egger's test) OR 95% CI p -value p -value I 2 (%) rs3808351 A vs. G 2 0.79 [0.29; 2.09] 0.637 Random 0.028 79.25 NA AA vs. GA + GG 2 0.84 [0.40; 1.74] 0.643 Fixed 0.107 61.37 NA AA + GA vs. GG 2 0.46 [0.28; 0.76] 0.002 Fixed 0.180 44.33 NA AA vs. GG 2 0.53 [0.23; 1.23] 0.142 Random 0.111 60.47 NA GA vs. GG 2 0.46 [0.27; 0.79] 0.004 Fixed 0.292 9.78 NA CI, confidence interval; OR, odds ratio; SNP, single nucleotide polymorphism; Stat. model, statistical model. Bold values indicate statistically significant differences between ≥ T2 and < T2. Characteristics of individual studies for the association between rs3808351 and tumor size. Alt., alternative allele; Ref., reference allele; SNP, Single nucleotide polymorphism. ∗ p for Hardy–Weinberg equilibrium test in controls. Meta-analysis for the association between rs3808351 and tumor size. CI, confidence interval; OR, odds ratio; SNP, single nucleotide polymorphism; Stat. model, statistical model. Bold values indicate statistically significant differences between ≥ T2 and < T2. A total of 204 articles were first screened to evaluate the association between GPER expression levels with cancer progression. After reviewing the title, abstract, and removing duplications, 151 articles were excluded, and 53 articles were then further evaluated. Among them, 33 articles were subsequently removed either because the data cannot be extracted, or the studies did not provide immunohistochemistry results. Finally, 20 articles were included in this meta-analysis [ 41 , 42 , 43 , 44 , 45 , 46 , 47 , 48 , 49 , 50 , 51 , 52 , 53 , 54 , 55 , 56 , 57 , 58 , 59 , 60 ]. The characteristics of the included studies are shown in Table 6 . Table 6 Characteristics of individual studies for the association between GPER expression levels and cancer progression. Table 6 No Author (year) Cancer type Tumor size ≥2 cm Tumor stage ≥2 Tumor grade ≥2 GPER+/High Total GPER–/Low Total GPER+/High Total GPER–/Low Total GPER+/High Total GPER–/Low Total 1 Aiad et al (2014) Breast cancer – – – – – – – – 11 33 6 18 2 Aquino et al (2018)a Salivary Gland Tumors – – – – – – – – 16 26 1 5 Aquino et al (2018)b Salivary Gland Tumors – – – – – – – – 3 4 14 27 3 Friese et al (2017)a Cervical cancer – – – – 96 114 35 42 102 111 38 41 Friese et al (2017)c Cervical cancer – – – – 109 129 22 27 118 125 22 27 4 Heublein (2011)-1 Ovarian Granulosa Cell Tumors – – – – 1 8 2 7 – – – – Heublein (2011)-2 Ovarian Granulosa Cell Tumors – – – – 1 3 2 12 – – – – 5 Ignatov et al (2011) Breast cancer 84 183 72 140 – – – – 164 182 126 140 6 Ignatov et al (2013) Ovarian cancer – – – – – – – – 85 103 21 21 7 Ignatov et al (2013)∗∗ Breast cancer 34 65 39 99 84 99 60 65 – – – – 8 Ignatov et al (2018) Breast cancer 149 352 40 83 – – – – 280 352 74 83 9 Ino et al (2019)∗ Uterine cervical adenocarcinoma – – – – 9 19 1 34 – – – – 10 Kolkova et al (2012) Ovarian cancer – – – – 36 50 77 100 48 50 95 100 11 Krakstad et al (2012)∗ Endometrial cancer – – – – – – – – 68 333 79 141 12 Liu et al (2019)a NSCLC – – – – 63 120 8 30 – – – – Liu et al (2019)b NSCLC – – – – 39 78 32 72 – – – – 13 Luo et al (2011) Breast cancer – – – – – – – – 138 198 10 40 14 Martin et al (2018)a∗ Breast cancer 124 327 372 910 132 327 351 910 263 327 775 910 Martin et al (2018)b∗ Breast cancer 111 370 384 864 124 370 348 864 287 370 748 864 15 Samartzis et al (2014)a Breast cancer – – – – 99 189 486 789 136 185 680 781 Samartzis et al (2014)b Breast cancer – – – – 313 528 272 450 443 520 373 446 16 Smith et al (2009)∗ Ovarian cancer – – – – – – – – 39 52 37 82 17 Steiman et al (2013) Breast cancer – – – – 21 27 14 21 – – – – 18 Tian et al (2018)∗ Gastric cancer 8 26 18 58 17 26 40 58 4 26 33 58 19 Ye et al (2019)∗ Breast cancer – – – – 46 74 127 175 62 73 149 176 20 Yu et al (2014) Breast cancer – – – – 48 66 13 30 53 66 23 30 a. cytoplasmic GPER; b. nuclear GPER; c. membrane GPER; ∗Expression level classified as high/low; ∗∗Expression level classified as increase/decrease; 1-Iimunoreactive score; 2-Intensity. Characteristics of individual studies for the association between GPER expression levels and cancer progression. a. cytoplasmic GPER; b. nuclear GPER; c. membrane GPER; ∗Expression level classified as high/low; ∗∗Expression level classified as increase/decrease; 1-Iimunoreactive score; 2-Intensity. The meta-analysis results regarding pooled GPER expression levels and cancer progression are shown in Table 7 . In brief, no associations were found between GPER expression levels with tumor size, stage, nor grade. Subgroup analysis by ethnicity and cancer type were also performed, yielding similar findings, with the exception of a significant association between GPER expression with higher tumor stage in the Asian population (OR = 2.22, 95% CI = 1.12–4.41, p = 0.022, Table 7 ). No association was also observed when the analysis was performed based on the localization of GPER (data now shown). Table 7 Meta-analysis for the association between GPER expression levels and cancer progression. Table 7 Group No. of studies OR (95% CI) [Random] p -value Tumor size ≥ 2 cm  Overall (GPER+/–) 2 0.80 (0.58–1.10) 0.168  Overall (GPER high/low) 4 0.88 (0.55–1.39) 0.575  Breast cancer (GPER+/–) 3 0.87 (0.51–1.46) 0.590  Breast cancer (GPER high/low) 2 0.80 (0.58–1.10) 0.168 Tumor stage ≥ 2  Overall (GPER+/–) 11 1.18 (0.85–1.64) 0.326  Overall (GPER high/low) 6 0.87 (0.58–1.31 0.497  Asian (GPER+/–) 3 2.22 (1.12–4.41) 0.022  Asian (GPER high/low) 3 1.65 (0.38–7.21) 0.505  Caucasian (GPER+/–) 8 0.86 (0.72–1.04) 0.120  Caucasian (GPER high/low) 3 0.85 (0.58–1.20) 0.345  Breast cancer (GPER+/–) 4 1.15 (0.69–1.92) 0.595  Breast cancer (GPER high/low) 4 0.80 (0.59–1.09) 0.153  Ovarian cancer (GPER+/–) 3 0.78 (0.38–1.61) 0.505 Tumor grade ≥ 2  Overall (GPER+/–) 12 1.22 (0.68–2.20) 0.507  Overall (GPER high/low) 5 0.54 (0.25–1.17) 0.117  Caucasian (GPER+/–) 10 0.94 (0.56–1.60) 0.829  Caucasian (GPER high/low) 4 0.69 (0.31–1.55) 0.368  Breast cancer (GPER+/–) 5 1.06 (0.44–2.54) 0.894  Ovarian cancer (GPER+/–) 2 0.49 (0.05–5.12) 0.549 Bold values indicate statistically significant p < 0.05. Meta-analysis for the association between GPER expression levels and cancer progression. Bold values indicate statistically significant p < 0.05. Publication biases were examined by Begg's funnel plots and Egger's regression tests. Overall, funnel plots were symmetrical (data not shown) and p -values of Egger's regression test greater than 0.05, suggesting that publication biases did not likely influence the results.

To date, this study is the first to summarize the association between GPER gene polymorphisms and GPER expression levels with cancer. The pooled meta-analyses results demonstrated that GPER rs3808350 and rs3808351, but not rs11544331, were significantly associated with cancer predisposition, specifically in the Asian population. Patients harbouring the A allele of rs3808351 exhibited a lower risk of developing cancer and displayed smaller tumor size. Moreover, GPER expression levels in cancerous tissues were correlated with higher tumor stage in the Asian population. Our finding reinforces previous reports that A allele carriers of rs3808350 and rs3808351 exhibit protective effects against uterine leiomyoma and gynecomastia risks in the Turkish population [ 17 , 38 ]. Similar to our findings, Giess et al. [ 17 ] observed that AA and AG genotypes of rs3808351 were correlated with lower tumor stage and grade. Although we did not observe a significant association between rs11544331 and cancer risk, it has been suggested that rs11544331 (P16L) can alter the conformational structure and localization of GPER, resulting in defective GPER function and the aggravated migration of carcinoma cells [ 61 ]. We also found no significant relationship between haplotypes of rs3808350/rs3808351/rs11544331 with cancer predisposition, possibly because our analysis was pooled from two studies reporting different cancer type/disease. Considering the potential functional significance of rs3808350 and rs3808351, further studies should try to estimate the relationship between rs3808350 and rs3808351 with cancer in a larger population and other ethnicities to test whether our findings are statistically robust. Because rs3808350 and rs3808351 are located in the 5′ region of the GPER gene (rs3808350 (–642) is located in the 5′-regulatory region, while rs3808351 (+124) is located in the 5′-untranslated region and containing the gene promoter) [ 10 ], these polymorphisms may influence the transcription level of GPER. However, no related studies are currently available. Since our results showed that GPER expression in cancerous tissues correlate with the aggressiveness of malignancies and that the A allele of rs3808351 exhibits protective effects against tumor progression in the Asian population, it is reasonable to speculate that the G allele of rs3808351 may be associated with the upregulation of GPER transcription. However, only one study has reported the functional role of GPER polymorphisms in relation to post-transcriptional expression. The study reported that only rs10235056 was significantly correlated with GPER mRNA expression [ 20 ]. Therefore, further studies are still required to reveal the exact molecular mechanism underlying our significant findings. Although in general we did not find any relationship between expression level and localization of GPER with cancer progression, other studies have reported that GPER overexpression is strongly associated with lower survival rates in several cancer types [ 43 , 54 , 55 , 59 , 60 , 62 , 63 ]. Contrastingly, some studies demonstrated that loss of GPER protein corresponds with low GPER mRNA and poorer prognosis of endometrial and breast cancer patient [ 50 , 64 ], possibly due to GPER promoter hypermethylation [ 64 ]. Moreover, it seems that the localization of GPER in the plasma membrane is responsible for cancer aggressiveness [ 63 ]. Thus, in order to evaluate the prognostic value of GPER in cancer patients, GPER protein level, localization, and promoter hypermethylation must be examined simultaneously. Despite being the first meta-analysis in the field, several limitations of this study should be noted. First, only a limited number of studies were included for meta-analysis of GPER gene polymorphisms and cancer. Consequently, further studies are still warranted to test our findings with a larger sample size. Second, because the etiologies of cancer are complex, other genetic and environmental factors need to be addressed and may influence the relationship between GPER gene polymorphism, GPER level, and its localization in different cancer types. Hence, publication bias might affect the accuracy of our pooled studies. Notwithstanding, detailed functional analyses are still needed to uncover the exact molecular mechanisms of the observed significant association between GPER and cancer. It is notable that rs3808350 and rs3808351 have the potential to be used as a prospective biomarker for cancer, with potential use of rs3808351 in particular as a prognostic marker for cancer progression, particularly in Asians. Thus, future studies should address the possibility of GPER polymorphisms can be used as an early detection marker for malignancies in clinical settings. Altogether, our findings indicate that GPER plays a crucial role in cancer pathogenesis and progression.

Zulvikar Syambani Ulhaq: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data; Wrote the paper. Gita Vita Soraya, William Ka Fai Tse: Analyzed and interpreted the data; Wrote the paper. Alvi Milliana: Contributed reagents, materials, analysis tools or data. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Data included in article/supplementary material/referenced in article. The authors declare no conflict of interest. No additional information is available for this paper.

Estradiol (E 2 ) is a major form of estrogen and displays pleiotropic steroid function that play regulatory roles in many physiological processes [ 1 , 2 ]. Biosynthesis of E 2 is determined by the conversion of testosterone by a rate-limiting enzyme, aromatase (CYP19A1) [ 1 , 2 , 3 , 4 , 5 , 6 ]. E 2 -mediated effects are modulated through both genomic and non-genomic pathways by the nuclear and membrane estrogen receptor (ER), respectively [ 2 ]. Recent reports have suggested a pivotal role of E 2 in both the development and malignant progression of multiple cancers [ 7 ]. Several meta-analysis have demonstrated that cancer risk is associated with the polymorphism of ER-alpha ( ER α) [ 8 ], but not ER-beta ( ER β) [ 9 ]. However, the role of membrane ERs, such as the G protein-coupled estrogen receptor (GPER), with cancer pathogenesis remains elusive. GPER has been identified as a novel ER, and is a seven-transmembrane domain protein that is structurally distinguished from the classical ERα and ERβ [ 10 ]. GPER mediates rapid E 2 -induced nongenomic signaling events, resulting in long-term transcriptional changes and a broad range of response among a large variety of cell types [ 10 , 11 ]. Such evidence was supported by the expression of GPER in various human tissues, including lung, heart, brain, liver, skeletal muscle, and lymphoid tissues [ 12 ]. Additionally, E 2 exerts ten times higher binding capacity to GPER than ERα [ 13 ], implying a critical role of GPER in regulating normal physiological functions. GPER overexpression has been reported in several hormone-dependent malignancies, including cancers of the breast, ovaries, and endometrium [ 10 ]. The upregulation of GPER is also evident in seminoma and lung cancer [ 10 , 14 ]. Additionally, GPER overexpression has also been associated with poor treatment outcomes such as lowered efficacy of primary endocrine treatment in breast cancer patients [ 15 ] and poor-prognosis of endometrial cancers, uterine carcinosarcoma, and endometriosis [ 16 ]. The finding indicates that GPER expressed in ERα/β-negative breast cancer could induce the expression of connective tissue growth factor (CTGF) [ 17 ], and thus binding of E 2 to GPER for cell proliferation and migration. Hence, several studies have been proposed to identify novel GPER ligands with specific antiproliferative effects against estrogen-based malignancies [ 18 , 19 ]. Single nucleotide polymorphisms (SNPs) are variations in the genomic sequence that could potentially result in modifications of gene expression level as well as protein structure, level, and function [ 17 ]. The expression level of GPER mRNA is possibly affected by its polymorphism [ 20 ]. Although several SNPs have been identified in the GPER gene, only three were reported to have higher biological relevance with human neoplasms, which are rs3808350, rs3808351, and rs11544331 [ 10 ]. However, the role of GPER polymorphism in cancer remains inconclusive as shown by different results in various studies [ 10 , 13 , 16 , 17 ]. Therefore, this meta-analysis was conducted in order to understand the role of GPER with cancer predisposition and progression.