Association between ApoE polymorphism and type 2 diabetes: A meta-analysis of 59 studies

(1) Aims Due to the ever increasing incidence of T2DM, it is estimated that only half of the 79 million adults with type 2 diabetes (T2DM) will have adequate access to insulin by 2030 if the current levels of access is not improved. It is urgent to identify the important risk factors for T2DM and develop effective strategies to address the problem of T2DM. Our study aimed to evaluate the association between apolipoprotein E (ApoE) genetic polymorphism and type 2 diabetes, and to provide clues for the etiology of T2DM and even molecular marker of targeted therapy for the treatment of T2DM. (2) Methods Case-control studies of ApoE polymorphism and T2DM, which were included in PubMed, Web of Science, Medline, WanFang, VIP, and CNKI databases, were selected and evaluated according to criteria of inclusion and exclusion. Eligible data were extracted and pooled, and were analyzed and assessed using R soft-ware (version 3.4.3). Random-effect models were used when heterogeneity existed in between-study, and fixed-effect models were applied otherwise. (3) Results A total of 59 studies that consisted of 6,872 cases with T2DM and 8,250 controls were selected. Alleles and genotypes of ApoE between cases and controls were compared. For ApoE alleles, we observed the contrast of ε4 versus ε3 allele yielding a pooled OR of 1.18 (95% CI: 1.09-1.28; P<0.001). For ApoE genotypes, compared with ε3/ε3 genotype, ε2/ε2 genotype showed a possible association with T2DM (OR=1.46; 95% CI: 1.11-1.93; P=0.007), ε3/ε4 genotype had a 1.11-fold risk of developing T2DM (OR=1.11; 95% CI: 1.01-1.22; P=0.039), and ε4/ε4 genotype had a 1.71-fold risk of developing T2DM (OR=1.71; 95% CI: 1.33-2.19; P<0.001). (4) Conclusions There is an association between ApoE polymorphism and T2DM: allele ε4 and genotypes (ε2/ε2, ε3/ε4, and ε4/ε4) are associated with the increased risk for the development of T2DM, and they may be risk factors for T2DM.


Introduction
It is estimated that only half of the 79 million adults with type 2 diabetes will have adequate access to insulin by 2030 if the current levels of access is not improved (BASU et al. 2019). Moreover, one of the significant causes of worldwide mortality and morbidity is diabetes (2016), especially type 2 diabetes mellitus, which is also the major cause of substantial global economic burden (BOMMER et al. 2017). Therefore, there is an urgent need to identify the important risk factors for T2DM and develop effective strategies to address the problem of T2DM.
It is well accepted that genetic factor, environmental factors, and lifestyle contribute to the development of T2DM. Complex interactions between multiple genes and a range of environmental factors are involved in the onset and progression of type 2 diabetes (SCHEUNER et al. 2008). A better understanding of the contribution of genetic factors in the etiology of T2DM will facilitate the development of effective preventive strategies to reduce the ever increasing incidence of T2DM (DAVIES and THIRLAWAY 2013), it will also improve the effectiveness and precision of treatment and prevention strategies (O'RAHILLY et al. 2005).
Much of the recent research has studied the association between the ApoE gene polymorphism and the risk of T2DM, however, there are inconsistencies between the results of the different studies. The inconsistency may result from the difference of included population, sample size, and genotyping methods. Moreover, 18 new papers (CHEN 2006;TANG et al. 2007;ERDOGAN et al. 2009;AL-MAJED et al. 2011;CHAUDHARY et al. 2012;MUSTAPIC et al. 2012;GE et al. 2013;RONG et al. 2013;SUN et al. 2013a;XIONG et al. 2013;ALHARBI et al. 2014;LIU 2014;WANG et al. 2014;ATTA et al. 2016;LIU et al. 2016;LUO et al. 2016;MEHMET et al. 2016;LIANG et al. 2017) have been published since the publication of latest meta-analysis of the association between ApoE gene polymorphism and T2DM in 2014 (YIN et al. 2014). Therefore, we enrolled these new published articles, and performed a further meta-analysis to investigate whether ApoE polymorphism is associated with the increased risk of T2DM.

Search strategy
We performed this meta-analysis by extensive literature search in PubMed, Web of Science, Medline, WanFang, VIP, and CNKI databases (last search on November 19, 2018). The terms used for searching were ("ApoE" OR "Apolipoprotein E") AND ("polymorphism, Genetic" OR ''variant" OR "mutation") AND ("type 2 diabetes mellitus" OR "type 2 diabetes" OR "T2DM" OR "non-insulin dependent diabetes" OR "NIDDM"). The equivalent Chinese terms were used in the Chinese databases. In addition, we retrieved related articles that had not been identified in the initial search to replenish literatures.

Inclusion/exclusion criteria
Studies included in this meta-analysis were based on the following criteria: (1) casecontrol studies; (2) assessing the association between ApoE polymorphism and type 2 diabetes. The exclusion criteria met the follows: (1) duplicate articles; (2) no healthy controls; (3) lack of sufficient information on genotype or allele frequencies.

Data extraction
We extracted the main characteristics of each eligible study, including first author's last name, date of publication, region, population's ethnicity, genotyping method, number of cases and controls, and counts of the ApoE genotype or allele. We collected and calculated Hardy-Weinberg equilibrium (HWE) among the controls.

Quality assessment
The Newcastle-Ottawa scale (NOS) was used to evaluate quality of each article through a "star" rating system consisting of selection, comparability, and exposure. We allocated a score of 1 point for each condition a study met, and no point (0 score) if the condition or requirement was lacking. We calculated the total Quality Score of each study. Two authors (Jikang Shi and Shuping Ren) assessed the quality of included studies independently, When inconformity existed between the two authors, the results were requested to discuss with the third investigator (Dawei Chen). To avoid selection bias, studies with poor quality score were not excluded.

Statistical analysis
Allele and genotype frequencies of ApoE were calculated for each study to evaluate the HWE using Goodness of fit Chi-square test among control groups, and P<0.05 was considered as a significant deviation from HWE. The strength of association between ApoE polymorphisms and type 2 diabetes susceptibility was assessed using odds ratios (OR) and 95% confidence intervals (95% CI) because outcome variable was binary.
Heterogeneity was evaluated by the Chi-square test based Q-statistic and quantified by I 2 -statistic (HIGGINS et al. 2003). Random-effect models (DerSimonian and Laird methods) were used to calculate OR and 95% CI when P value of Q test was more than 0.10 or I 2 value was more than 50%; otherwise, fixed-effect models (Mantel and Haenszel methods) were applied (I 2 ≥50% considered heterogeneity existed in betweenstudy in this meta-analysis). Subgroup analyses stratified by ethnicity, quality score and Hardy-Weinberg equilibrium were performed to identify main sources of heterogeneity and to observe the association between ApoE polymorphisms and type 2 diabetes in different groups. Publication bias was assessed using funnel plots, and quantified by the Begg's and Egger's tests (P<0.05 considered statistically significant publication bias) (BEGG and MAZUMDAR 1994). Sensitivity analysis was performed to examine stability of results by omitting each study in each turn. All data management and statistical analyses were used R soft-ware (version 3.4.3), P-value < 0.05 was considered statistically significant.

Study Characteristics
Our meta-analysis initially collected 791 published articles, including 782 papers identified using our search strategy and 9 papers identified through the references. After abstracts and full texts were scanned according to the inclusion and exclusion criteria, 59 eligible articles with 6,872 cases and 8,250 controls were finally included in this paper. The protocol of the process for literature identification and selection is listed in Figure 1, and the baseline characteristics of the included studies are summarized in Table 1.

Association between alleles of ApoE and type 2 diabetes
There was significant heterogeneity in the comparison of ApoE ε2 with ε3 allele (I 2 =62%), and the pooled OR was 1.16 (95% CI: 0.98-1.37; P=0.079) when ApoE ε2 was compared with ε3 using the random-effects model ( Figure 2); however, there was not heterogeneity in the comparison of ApoE ε4 with ε3 allele (I 2 =36%), and the pooled OR was 1.18 (95% CI: 1.09-1.28; P<0.001) when ApoEε4 was compared with ε3 using the fixed-effects model (Figure 3), suggesting that ApoE ε4 allele may be a risk factor for type 2 diabetes.

Subgroup analysis
We conducted subgroup analysis stratified by ethnicity, quality score and Hardy-Weinberg equilibrium in order to identify main sources of heterogeneity. There were significant heterogeneity in the comparison of ApoE ε2 with ε3 allele (I 2 =62%) and the comparison of ε2/ε3 genotype with ε3/ε3 genotype (I 2 =55%) in our paper; however, we did not investigate sources of heterogeneity and there was no significant association between ApoE polymorphisms and type 2 diabetes in different subgroups (Supplementary Figure S1-S3).

Publication bias
Publication bias was assessed by funnel plots and quantified by Begg's and Egger's tests. All the funnel plots for ApoE allele and ApoE genotypes seemed symmetrical (Supplementary Figure S4-S5), and the results of Begg's and Egger's tests showed that there was no publication bias for the association between ApoE allele and type 2 diabetes and for the association between the ApoE genotypes and type 2 diabetes (all P>0.05).

Sensitivity analysis
Our results of sensitivity analysis showed that none of individual study influenced on the corresponding pooled ORs and 95% CIs in the comparison of ApoE ε4 with ε3 allele or in the comparison of ApoE ε2/ε3, ε2/ε4, and ε4/ε4 with genotype ε3/ε3 genotype

Discussion
In this meta-analysis, we included 59 literatures with 6,872 cases and 8,250 controls to explore the association between the ApoE gene polymorphism and type 2 diabetes mellitus. The major findings of our study are that allele ε4 and genotypes (ε2/ε2, ε3/ε4, and ε4/ε4) are associated with the increased risk for the development of T2DM, however, allele ε2 and genotypes (ε2/ε3 and ε2/ε4) are not associated with T2DM.
The strengths of the present study are that, 1) we included all the published literatures on the association between ApoE gene polymorphism and T2DM regardless of regions or ethnicities; 2) we had a large sample size. There are 18 new published papers discussing the association between ApoE gene polymorphism and T2DM since the last meta-analysis published in 2014, all of them are included in our present meta-analysis, which will provide more convincing evidence to the association of ApoE gene polymorphism with T2DM; 3) the results of our sensitivity analysis demonstrate that the conclusion of the present study is very stable; 4) the results of publication bias analysis reveal that the conclusion of our study is absent of publication bias. However, our study also has several weaknesses, 1) presence of heterogenicity in our study. We did the subgroup analysis on HWE, genotyping methods and ethnicities, but we did not trace the source of heterogenicity; 2) since the present study is a case-control study, the findings of our study cannot provide the causal relationship between ApoE gene polymorphism and T2DM, only the association of ApoE gene polymorphism with T2DM.
The findings of our meta-analysis are in accordance with the previous studies (ANTHOPOULOS et al. 2010;QIU XU 2010;AIMEI LONG 2013;YIN et al. 2014), showing that both ApoE ε4 allele and the genotypes (ε3/ε4 and ε4/ε4) were associated with increased risk of T2DM. Subjects carrying the ε4 alleles had higher plasma total cholesterol levels compared to subjects carrying the ε3/ε3 genotype, and HDL cholesterol was significantly lower in the ε3/ε4 than in the ε3/ε3 individuals (DALLONGEVILLE et al. 1992); individuals carrying the ε2/ε2 genotype had about 31% lower mean LDL than those with the ε4/ε4 genotype (BENNET et al. 2007). Insulin resistance is known to be strongly associated with metabolic dyslipidemia and the correlation of lipid profiles with diabetic phenotypes is significant. Therefore, ApoE ε4 allele and the genotypes (ε3/ε4 and ε4/ε4) were associated with an increased risk of T2DM through affecting the lipid metabolism.
We found the genotype ε2/ε2 was associated with increased risk of T2DM, but not allele ε2 or genotype ε2/ε3; which are not in agreement with the results of previous meta-analyses (YIN et al. 2014 which also coincides with the finding that the higher frequency of the ε2/APOE allele might be primarily related to T2DM (ERRERA et al. 2006).
The significance of the present study is that we identified significant association between ApoE gene polymorphism and T2DM, which will provide clues for the etiology of T2DM and even molecular marker of targeted therapy for the treatment of T2DM. However, it is essential to further investigate the interaction between gene and gene as well as the gene and environment since T2DM is the result of interaction between genetic and environmental factors.
In conclusion, there is an association between ApoE polymorphism and T2DM: allele ε4 and genotypes (ε2/ε2, ε3/ε4, and ε4/ε4) are associated with the increased risk for the development of T2DM, and they may be risk factors for T2DM.