Eligible studies
As shown in Fig. 1, a total of 584 articles were initially obtained by searching the databases. After duplicate check by Endnote 20, 435 articles remained. Excluded 37articles based on browsing the titles and abstracts. According to the inclusion criteria, 5 of the remaining 23 records were further excluded based on full-text review. In total, 21 studies (18 articles) were eligible including 2692 virus-related hepatocellular carcinoma cases and 5835 controls were identified and included in this meta-analysis.
Characteristics Of Study
The characteristics of the 21 included studies are shown in Table 1. Of all eligible studies, 11 were conducted in Asian populations, 2 in European populations, 5 in African populations, and 3 in mixed populations. In all studies, the cases were histologically confirmed (17 studies) or diagnosed by elevated α-fetoprotein and different iconography changes (abdominal ultrasound and triphasic computed tomography). All controls were free of cancer. 4 studies used healthy populations, 5 studies used patients with chronic liver diseases (HBV infection, HCV infection, cirrhosis), and 12 studies included healthy subjects and patients with chronic liver diseases as controls. The sample size of the total participants ranged from 75 to 1774, with a mean of 406. Quality scores for individual studies ranged from 11.5 to 21, with 10 of the 21 studies classified as high quality. 20 studies used peripheral blood, and one study used FFPE to extract genome DNA. 14 studies used the polymerase chain reaction-restriction fragment length polymorphism assay (PCR-RFLP), 6 studies used the Taqman method, and 1 study used Matrix-Assisted Laser Desorption/ Ionization Time of flight mass spectrometry (MALDI-TOF-MS) to genotype the EGF + 61 A/G polymorphism. The genotype distribution in the controls of all studies was consistent with HWE.
Table 1
Main characteristics of eligible studies included in the meta-analysis
first author | year | country(Ethnicity) | source of controls | type of controls | sample origin | genotyping methods | sample size (case/control) | genotype frequency (case/control) | G allele frequency | HWE | Quality score |
GG | GA | AA |
Tanabe-FRA | 2008 | France(European) | HB | Cirrhosis | Peripheral bolld | PCR-RFLP | 44/77 | 15/12 | 17/37 | 12/28 | 39.60% | Y | 13.5 |
Tanabe-USA | 2008 | USA(mixed) | HB | HBV/HCV/Cirrhosis | Peripheral bolld | PCR-RFLP | 59/148 | 23/32 | 27/65 | 9/51 | 43.60% | y | 14.5 |
Qi | 2009 | China(Asian) | HB and PB | Healthy/HBV | Peripheral bolld | PCR-RFLP | 215/380 | 102/182 | 98/160 | 15/38 | 68.90% | Y | 21 |
Wang-GX | 2009 | China(Asian) | HB | Healthy/HBV | Peripheral bolld | PCR-RFLP | 376/477 | 190/208 | 154/221 | 32/48 | 66.80% | Y | 17.5 |
Wang-JS | 2009 | China(Asian) | HB | Healthy/HBV | Peripheral bolld | PCR-RFLP | 186/198 | 107/93 | 65/88 | 14/17 | 69.20% | Y | 18 |
Li | 2010 | China(Asian) | HB and PB | Healthy/Cirrhosis | Peripheral bolld | PCR-RFLP | 186/338 | 96/161 | 82/145 | 8/32 | 69.10% | Y | 19.5 |
Abu Dayyeh | 2011 | USA(mixed) | HB | HCV | Peripheral bolld | PCR-RFLP | 66/750 | 26/178 | 25/350 | 15/222 | 47.10% | Y | 16.5 |
Chen | 2011 | China(Asian) | HB | Healthy/HBV/Cirrhosis | Peripheral bolld | PCR-RFLP | 120/240 | 62/106 | 51/110 | 7/24 | 67.10% | Y | 19 |
Abbas | 2012 | Egypt(African) | HB | Healthy/HCV/Cirrhosis | Peripheral bolld | PCR-RFLP | 20/60 | 7/9 | 9/28 | 4/23 | 38.30% | Y | 12 |
Cmet | 2012 | Italy(European) | HB and PB | Healthy/HBV | Peripheral bolld | PCR-RFLP | 18/361 | 4/66 | 10/172 | 4/123 | 42.10% | Y | 16 |
Shi | 2012 | China(Asian) | HB | Healthy | Peripheral bolld | PCR-RFLP | 73/117 | 18/13 | 31/52 | 24/52 | 33.30% | Y | 13.5 |
El-Bendary | 2013 | Egypt(African) | HB | HCV/Cirrhosis | Peripheral bolld | PCR-RFLP | 133/105 | 57/9 | 43/36 | 33/60 | 25.70% | Y | 12 |
Suenaga | 2013 | Japan(Asian) | HB | Healthy/HBV/HCV | Peripheral bolld | PCR-RFLP | 208/290 | 108/161 | 89/104 | 11/25 | 73.40% | Y | 11.5 |
Wu | 2013 | China(Asian) | HB and PB | Healthy/HBV | Peripheral bolld | TaqMan | 404/1370 | 206/647 | 153/576 | 45/147 | 68.20% | Y | 17.5 |
Yuan-USA | 2013 | USA(mixed) | HB | Healthy | Peripheral bolld | TaqMan | 117/225 | 28/63 | 61/102 | 28/60 | 50.70% | Y | 19 |
Yuan-CHN | 2013 | China(Asian) | HB | Healthy/HBV/HCV | Peripheral bolld | TaqMan | 250/245 | 25/20 | 99/107 | 126/118 | 30.00% | Y | 15 |
Wei | 2016 | China(Asian) | HB | HCV | Peripheral bolld | MALDI-TOF-MS | 47/213 | 30/101 | 15/98 | 2/14 | 72.1% | Y | 12.5 |
El Sergany | 2017 | Egypt(African) | HB | Healthy | Peripheral bolld | TaqMan | 50/50 | 42/2 | 5/6 | 3/42 | 49.50% | Y | 15.5 |
Gholizadeh | 2017 | Iranian(Asian) | HB | Healthy | FFPE/Healthy | PCR-RFLP | 40/106 | 4/34 | 25/48 | 11/24 | 51% | Y | 13 |
Asar | 2020 | Egypt(African) | HB | Healthy/HCV | Peripheral bolld | TaqMan | 30/60 | 11/11 | 10/34 | 9/15 | 48.89% | Y | 13 |
Baghdadi | 2020 | Egypt(African) | HB | Healthy/Cirrhosis | Peripheral bolld | TaqMan | 50/25 | 20/4 | 23/13 | 7/8 | 56% | Y | 21 |
Abbreviations: HB, Hospital-based; PB, Population-based; HBV, control subjects were hepatitis B virus carriers; HCV, control subjects were hepatitis C virus carriers; PCR-RFLP, polymerase chain reaction-restriction fragment length polymorphism; MALDI-TOF-MS, Matrix-Assisted Laser Desorption/ Ionization Time of Flight Mass Spectrometry; HWE: Hardy-Weinberg equilibrium in control population; Y, yes; N, no. |
Meta-analysis Results
The results of pooling all studies showed that the EGF + 61A/G polymorphism was significantly associated with an increased virus-related hepatocellular carcinoma risk under all genetic models (G vs. A:OR = 1.56, P < 0.001, 95% CI: 1.26–1.94, I2 = 86.8%; GG vs. GA + AA: OR = 1.67, P < 0.001, 95% CI 1.29–2.15, I2 = 79%; GG + GA vs. AA: OR = 1.67, P < 0.001, 95% CI: 1.26–2.20, I2 = 70.2%; GG vs. AA: OR = 2.18, P < 0.001, 95% CI 1.50–3.16, I2 = 76.6%; GA vs. AA: OR = 1.20, P < 0.001, 95% CI 1.03–1.39, I2 = 23.7%)(Table 2).
Table 2
Main results of meta-analysis for EGF + 61A/G polymorphism and HCC risk
Subgroup | No. comparisons | sample size ( case/control) | GG vs. GA + AA | GG + GA vs. AA | GG vs. AA | GA vs. AA | G vs. A |
OR (95% CI) | I2(%) | OR (95% CI) | I2(%) | OR (95% CI) | I2(%) | OR (95% CI) | I2(%) | OR (95% CI) | I2(%) |
overall | 21 | 2680/5835 | 1.67(1.29–2.15)* | 79# | 1.67(1.26–2.20)* | 70.2# | 2.18(1.50–3.16)* | 76.6# | 1.20(1.03–1.39)* | 23.7 | 1.56(1.26–1.94)* | 86.8# |
Racial descent | | | | | | | | | | | | |
Asian | 11 | 2105/3974 | 1.21(1.01–1.45)* | 52.5# | 1.18(0.99–1.42) | 6.5 | 1.39(1.07–1.82)* | 34 | 1.11(0.92–1.32) | 4.5 | 1.15(1.02–1.29)* | 40.1 |
European | 2 | 62/438 | 2.07(0.98–4.38) | 12.6 | 1.61(0.84–3.12) | 0 | 2.51(1.10–5.72)* | 0 | 1.30(0.64–2.63) | 0 | 1.59(1.05–2.41)* | 0 |
African | 5 | 271/300 | 6.97(2.34–20.71)* | 79.8# | 4.12(1.23–13.81)* | 86.5# | 9.74(2.43–39.12)* | 82.9# | 1.55(0.99–2.41) | 60.8# | 4.46(1.53–13.02)* | 93# |
Mixed | 3 | 242/1123 | 1.55(0.79–3.06) | 77.3# | 1.57(0.96–2.58) | 46.9 | 1.94(0.87–4.33) | 73.1# | 1.37(0.94-2.00) | 11 | 1.45(0.93–2.26) | 77.1# |
Study quality | | | | | | | | | | | | |
High quality | 9 | 1720/4003 | 1.26(1.07–1.48)* | 38.4 | 1.26(1.04–1.52)* | 0 | 1.46(1.13–1.89)* | 29.2 | 1.16(0.95–1.42) | 0 | 1.20(1.08–1.34)* | 24.2 |
Low quality | 12 | 960/1832 | 2.34(1.30–4.20)* | 85.6# | 2.03(1.22–3.38)* | 80.7# | 3.00(1.48–6.08)* | 82.3# | 1.24(1.00-1.54)* | 40 | 1.98(1.25–3.14)* | 91.7# |
Source of controls | | | | | | | | | | | | |
Population-based | 4 | 823/2449 | 1.12(0.95–1.32) | 0 | 1.37(0.90–2.10) | 41.9 | 1.35(0.93–1.95) | 21.6 | 1.18(0.88–1.58) | 50.9# | 1.12(0.99–1.27) | 0 |
Hospital-based | 17 | 1857/3386 | 1.94(1.38–2.73)* | 81.6# | 1.75(1.24–2.45)* | 73.6# | 2.44(1.53–3.89)* | 78.8# | 1.20(1.01–1.43)* | 20.3 | 1.72(1.29–2.29)* | 88.8# |
Type of controls | | | | | | | | | | | | |
Healthy controls | 15 | 2124/4919 | 1.47(1.10–1.95)* | 78.5# | 1.40(1.03–1.90)* | 69.1# | 1.68(1.14–2.47)* | 72.9# | 1.08(0.92–1.27) | 20.4 | 1.40(1.10–1.79)* | 86.4# |
Patients with chronic liver diseases | 6 | 556/916 | 2.39(1.33–4.28)* | 79.4# | 2.77(1.99–3.86)* | 0 | 4.13(2.29–7.45)* | 52# | 1.89(1.32–2.70)* | 0 | 2.02(1.32–3.09)* | 82.7# |
Number of cases | | | | | | | | | | | | |
> 100 | 17 | 2530/5640 | 1.42(1.14–1.77)* | 72.1# | 1.44(1.17–1.78)* | 45.9 | 1.78(1.29–2.45)* | 67.7# | 1.18(1.01–1.37)* | 0 | 1.33(1.12–1.56)* | 76.7# |
≤ 100 | 4 | 150/195 | 6.99(1.51–32.40)* | 84.3# | 4.44(0.70-28.16) | 89.9# | 9.73(1.31–71.99)* | 86.7# | 1.53(0.83–2.82) | 70.6# | 4.58(0.95–22.09) | 94.7# |
genotyping methods | | | | | | | | | | | | |
PCR-RFLP | 14 | 1732/3647 | 1.56(1.17–2.07)* | 75.6# | 1.68(1.35–2.09)* | 22.6 | 2.14(1.46–3.13)* | 65.5# | 1.39(1.14–1.70)* | 0 | 1.44(1.17–1.77)* | 78.1# |
TaqMan | 6 | 901/1975 | 2.41(1.14–5.11)* | 87.7# | 1.87(0.91–3.84) | 88.3# | 2.74(1.08–6.95)* | 88# | 0.98(0.78–1.23) | 61.9# | 2.15(1.15–4.01)* | 94.4# |
MALDI-TOF-MS | 1 | 47/213 | 1.96(1.02–3.76)* | / | 1.58(0.35–7.21) | / | 2.08(0.45–9.67 | / | 1.07(0.22–5.19) | / | 1.66(0.96–2.86) | / |
OR, odds ratio; 95% CI, 95% confidence interval. |
*Significant results, P-value < 0.05. |
#Random effect estimate. |
In subgroup analyzes based upon ethnicity, a significantly elevated association was observed between EGF + 61A/G polymorphism and the risk of virus-related hepatocellular carcinoma in Asian populations (G vs. A: OR = 1.15, P < 0.001, 95% CI: 1.02–1.29, I2 = 40.1%), European populations (G vs. A: OR = 1.59, P < 0.001, 95% CI: 1.05–2.41, I2 = 0.0%) and African populations (G vs. A: OR = 4.46, P < 0.001, 95% CI 1.53–13.02, I2 = 93%), respectively (Fig. 2). When stratifying by study quality, the results showed that EGF + 61A/G polymorphism was associated with an increased virus-related hepatocellular carcinoma risk both in high-quality studies (G vs. A: OR = 1.20, P < 0.001, 95% CI: 1.08–1.34, I2 = 24.2%) and in low-quality studies (G vs. A: OR = 1.98, P < 0.001, 95% CI: 1.25–3.14, I2 = 91.7%). In subgroup analyzes by source of controls, the results showed that the EGF + 61A/G polymorphism was significantly associated with the risk of virus-related hepatocellular carcinoma in hospital-based studies (G vs. A: OR = 1.72, P < 0.001, 95% CI 1.29–2.29, I2 = 88.8%), but not in population-based studies (G vs. A: OR = 1.12, P = 0.202, 95% CI 0.99–1.27, I2 = 0.0%). Furthermore, according to the chronic liver disease status in Asian controls, a significant association was obtained between EGF + 61A/G polymorphism and virus-related hepatocellular carcinoma in patients with chronic liver diseases (G vs. A: OR = 2.02, P < 0.001, 95% CI 1.32–3.09, I2 = 82.7%), and in healthy controls (G vs. A: OR = 1.40, P < 0.001, 95% CI 1.10–1.79, I2 = 86.4%) (Table 2).
Heterogeneity Analysis
The Q-statistic indicated statistically significant heterogeneity between all studies under all genetic models except for the comparison of heterozygote comparison (Table 2). However, in the subgroup analyses by ethnicity, the between-study heterogeneity was not observed in Asian populations, European populations or African populations. Moreover, meta-regression indicated that both ethnicity and study quality significantly contributed to the heterogeneity for EGF + 61A/G polymorphism (Table 3).
Table 3
Main results of meta-regression for EGF + 61A/G polymorphism and HCC risk
Factor | GG vs. GA + AA | GG + GA vs. AA | GG vs. AA | GA vs. AA | G vs. A |
t | p | t | p | t | p | t | p | t | p |
Racial descent | -2.43 | 0.025 | -1.76 | 0.095 | -2.29 | 0.034 | -1.29 | 0.212 | -2.17 | 0.043 |
Source of controls | 1.07 | 0.298 | 0.36 | 0.722 | 0.73 | 0.473 | -0.08 | 0.939 | 0.84 | 0.41 |
Type of controls | 0.91 | 0.373 | 2.12 | 0.048 | 1.7 | 0.106 | 2.77 | 0.012 | 0.71 | 0.486 |
Genotyping methods | 0.68 | 0.507 | -0.03 | 0.975 | 0.25 | 0.803 | -2.05 | 0.054 | 0.73 | 0.477 |
Sample size | -1.22 | 0.236 | -0.89 | 0.386 | -1.15 | 0.265 | -0.50 | 0.626 | -1.11 | 0.28 |
Quality score | 2.6 | 0.018 | 1.9 | 0.073 | 2.34 | 0.031 | 0.75 | 0.461 | 2.54 | 0.02 |
Sensitivity Analysis And Publication Bias
Sensitivity analysis was performed by sequential omission of individual studies. Pooled ORs were consistently significant in general populations or Asian populations by omitting one study at a time under the allelic contrast, recessive model, and homozygote comparison, suggesting robustness of our results. And Begg’s funnel plots were prepared and Egger’s test was performed on the final set of 21 studies to assess publication bias for reported comparisons of EGF + 61A/G polymorphism and virus-related hepatocellular carcinoma risk. The results showed that risk of publication bias may exist in overall population, but a low risk of publication bias in Asian populations(Fig. 3, 4, Table 4).
Table 4
The results of the Begg and Egger’s test
Population | genetic model | Begg (z|p) | Egger (t|p) |
All | G vs. A | 3.11|0.002 | 2.87|0.01 |
Asian | G vs. A | 1.09|0.276 | 0.43|0.68 |
All | GG vs. GA + AA | 2.39|0.017 | 2.78|0.012 |
Asian | GG vs. GA + AA | 0.93|0.35 | 0.04|0.968 |
All | GG + GA vs. AA | 2.08|0.037 | 2.90|0.009 |
Asian | GG + GA vs. AA | 1.09|0.276 | 2.48|0.035 |
All | GG vs. AA | 2.87|0.004 | 2.72|0.014 |
Asian | GG vs. AA | 1.56|0.119 | 0.73|0.482 |
All | GA vs. AA | 1.72|0.085 | 3.18|0.005 |
Asian | GA vs. AA | 1.40|0.161 | 2.59|0.029 |