Abstract
The ATP-binding cassette transporter A7 (ABCA7) was identified as a known risk factor for Alzheimer’s disease (AD). However, the relation between ABCA7 and AD was still inconsistent across these studies. Here, our meta-analysis aimed at confirming the association of ABCA7 with AD. Finally, 16 case-control studies (63747 versus 85833) were retrieved from PubMed and other databases. Three common loci were confirmed to increase the risk of AD (rs3764650: OR = 1.20, 95% CI = 1.16–1.24; rs3752246: OR = 1.13,95% CI = 1.08–1.19; rs4147929: OR = 1.17, 95% CI = 1.10–1.24), but the associations varied among the different races. Furthermore, ABCA7 loss-of-function (LOF) mutations conferred a higher risk for AD than did the above variants (LOF: OR = 1.78, 95% = 1.43–2.22). In conclusion, ABCA7 genetic variants, especially the LOF mutations, were significantly associated with the risk of AD.
INTRODUCTION
To date, genome-wide association studies (GWAS) have reported ABCA7 (the ATP-binding cassette, subfamily A, member 7 gene) (rs3764650, rs3752246, rs4147929) as late-onset Alzheimer’s disease (LOAD) susceptibility loci in Caucasian populations [1–3], which were widely replicated by following studies [2, 4–6]. A meta-analysis of 74,046 individuals showed evidence of the association between ABCA7 (rs4147929) and AD in Caucasians [5]. But these significant associations of ABCA7 with AD were not successfully replicated by other Caucasian studies [7, 8]. In addition, our group and other investigators also investigated the relation between ABCA7 loci (rs3764650, rs3752246) and AD in Asian populations, and did not find the relation between them [9–12], although the links were detected in other studies [13]. Furthermore, the diverse results about the associations of ABCA7 variants with AD were reported in African populations [14, 15]. Therefore, the relation between ABCA7 and AD was still inconsistent across these studies, and it was necessary to detect the association of ABCA7 with AD in a meta-analysis.
Recently, the next generation sequencing of the GWAS-identified loci was conducted to ascertain whether novel mutations of these loci contributed to the risk of AD, and it detected that ABCA7 loss-of-function (LOF) mutations are strongly associated with a higher risk of AD, which was also replicated by the next studies [15–22]. LOF mutations, including nonsense, frameshift and canonical splice-site mutations, resulted in the loss of the biological function [15], and they were treated as a single allele for research [20]. Our study also explored the association of ABCA7 LOF mutations with the risk of AD.
Here, we conducted a meta-analysis that pooled previous data to investigate the association between ABCA7 genetic variants and the risk of AD. Furthermore, our study attempted to determine this association among different populations by dividing the data into three ethnic subgroups (Caucasians, Asians, and Africans).
MATERIALS AND METHODS
Search strategies
All of the potential qualified studies were found through literature search of PubMed, EMBASE, and Medline databases up to August 2017 and were supplemented by manual retrieval of the references. “ABCA7” or “ATP binding cassette subfamily A member 7” was used to search for related articles. Additional researches were searched manually from the reference materials in each of the qualified studies. The titles and abstracts of the previous search were estimated, and full texts of selected studies were included.
Study selection
Inclusion criteria were: 1) research must be based on human beings; 2) the sample size of the study is clearly defined; 3) odds ratios and its 95% confidence interval should be provided or can be calculated; 4) studies are obliged to evaluate the relevance between polymorphism of ABCA7 and the susceptibility to AD; 5) the diagnosis of AD should be consistent with the clinical criteria set by the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA), and all patients were diagnosed with definite, probable or possible Alzheimer’s disease on the basis of the NINCDS-ADRDA guideline.
Data extraction
A predesigned data aggregation table was used by two reviewers independently to collect the following data: name of first author, year of publication, country and ethnicity of studied participants, total number of case and control groups, minor allele frequency, odds ratios and its 95% confidence interval.
Statistical analysis
The method of generic inverse variance was used to pool odds ratios (OR) and 95% confidence intervals (95% CI) to evaluate the association of ABCA7 polymorphisms with AD. To explore the association in different populations, the ethnicity-based subgroup meta-analyses were also conducted (Caucasians, Asians, and Africans). The degree of heterogeneity was assessed by a χ 2-test combined with the I2 method, and I2>50% or p < 0.10 was considered as statistically significant heterogeneity [23]. If significant heterogeneity is observed, then a random-effects model is appropriate for analysis [24]; otherwise, a fixed-effects model is suitable for analysis. Begg’s adjusted rank correlation test [25] and Egger’s regression test [26] were used to detect potential publication bias [27]. Data were analyzed with R-3.2.0software.
RESULTS
A total of 178 studies were retrieved through literature search, and 31 published articles were included by manual searching. Finally, 16 full-text studies were chosen for extraction, evaluation and meta-analysis. The characteristics of the included studies are listed in Table 1.
Characteristics of included studies for ABCA7 and AD
Characteristics of included studies for ABCA7 and AD
A total of 11 studies (50,729 versus 67,369) were included in the meta-analysis of the association of ABCA7 rs3764650 polymorphism with AD, including 6 studies in Caucasians [2, 28], 4 studies in Asians [9–11, 13], and 1 study in Africans [14]. ABCA7 rs3764650 was significantly associated with an increased risk of AD with no evident heterogeneity (OR = 1.20, 95% CI = 1.16–1.24, I2 = 0% , p = 0.60) (Fig. 1). In addition, subgroup analyses by ethnicity indicated that rs3764650 increased the risk of AD in Caucasian, Asian and African populations (Caucasian: OR = 1.22, 95% CI = 1.17–1.26; Asian: OR = 1.10, 95% CI = 1.01–1.19; African: OR = 1.27, 95% CI = 1.04–1.55) (Supplementary Figure 1). No heterogeneity existed in any subgroup (I2 = 0% ), and the funnel plots showed no publication bias (p = 0.414).
Results of the meta-analysis for ABCA7 and Alzheimer’s disease.
Five studies (22,918 versus 21,304) investigated the association of ABCA7 rs3752246 polymorphism with AD, including 3 studies in Caucasians [4, 29], 1 study in Asians [9], and 1 study in Africans [14]. ABCA7 rs3752246 significantly increased the risk of AD without heterogeneity (OR = 1.13, 95% CI = 1.08–1.19, I2 = 0% , p = 0.62) (Fig. 1). Moreover, the pooled results of Caucasians showed rs3752246 was a risk factor for AD (Caucasians: OR = 1.15, 95% CI = 1.09–1.21), while the association of rs3752246 and AD did not achieve the significant level in Asians and Africans (Asians: OR = 1.05, 95% CI = 0.89–1.24; Africans: OR = 0.82, 95% CI = 0.53–1.27) in subgroup analyses (Supplementary Figure 2). In addition, no publication bias was observed (p = 0.150).
Only 1 study (8572 versus 11312) explored the relation between ABCA7 rs4147929 polymorphism and AD in Caucasians [5] (Supplementary Figure 3). ABCA7 rs4147929 was associated with an increased risk of AD (OR = 1.17, 95% CI = 1.10–1.24) (Fig. 1). There was no heterogeneity across studies (I2 = 0% , p = 0.65) and no publication bias (p = 0.835).
Totally, 3 studies (4115 versus 6480) assessed the correlation between LOF mutations in ABCA7 and AD, including 2 studies in Caucasians [20, 22] and 1 study in Africans [15]. ABCA7 LOF mutations strongly increased the risk of AD without significant heterogeneity (OR = 1.78, 95% CI = 1.43–2.22, I2 = 0% , p = 0.42) (Fig. 1). In addition, subgroup analyses by ethnicity indicated that LOF mutations increased the risk of AD in both Caucasian and African populations (Caucasians: OR = 2.24, 95% CI = 1.40–3.60; Africans: OR = 1.67, 95% CI = 1.30–2.14) (Supplementary Figure 4). No publication bias existed here (p = 0.481).
DISCUSSION
Our meta-analysis provided insights into the association between ABCA7 variants and the susceptibility to AD. The pooled results revealed a notable association between the three ABCA7 common variants (rs3764650, rs3752246, rs4147929) and the increased risk of AD. Moreover, subgroup analyses by ethnicity were conducted to investigate the relationships between these variants and AD in different populations. Rs3764650 was found to be significantly associated with the increased risk of AD in Caucasian, Asian and African populations, whereas the effect of rs3752246 on the risk of AD was inconsistent among the different populations. These results in Asian and African populations can probably be explained by a limitation of small sample size, the difference in the minor allele frequency (Caucasians: MAF = 0.19; Asians: MAF = 0.31; Africans: MAF = 0.41) [30] and the underpowered studies included. Due to the limitation of the included studies, we failed to make a convincing conclusion on the relationship between rs4147929 and AD in various populations. Furthermore, our results showed that ABCA7 LOF mutations were strongly associated with the increased risk of AD, which conferred a higher risk to develop AD than did the other common variants. In addition, the effect of ABCA7 LOF mutations on AD was consistent in Caucasians and Africans.
Loss-of-function (LOF) (nonsense, frameshift and canonical splice-site mutations) was recently found to play a role in AD by facilitating the clearance of amyloid-β [15]. Besides, these functional variants were also predicted to cause premature protein termination via LOF mechanism, which increases the risk of AD. But the mechanism of ABCA7 in AD still needs to be elucidated by further studies.
The results of our meta-analysis were consistent with previous meta-analyses, which confirmed the association of ABCA7 with AD [3, 31]. By comparison with these analyses, we added several new studies [4, 29] in this meta-analysis, which increased the sample size to improve theaccuracy of outcomes. Moreover, previous studies only described the relationship between ABCA7 and AD in Caucasians, while our study divided the population into three subgroups to analyze the influence of ABCA7 variants on the risk of AD in different ethnic groups, and found that although the majority of studies reported a positive correlation between ABCA7 variants and the risk of AD, the associations varied in subgroup analyses. Finally, the associations of ABCA7 LOF mutations with the risk of AD were summarized here, and ABCA7 LOF mutations seemed to be a higher risk factor for AD.
Conclusions
Our meta-analysis confirmed that ABCA7 common variants (rs3764650, re3752246, rs4147929) and LOF mutations significantly increased the risk of AD, and ABCA7 LOF mutations seemed to be a higher risk factor for AD than the common variants. Furthermore, the majority of the included studies were performed in Caucasians, which indicated that large-scale studies in Asians and Africans are needed to confirm the associations of ABCA7 and AD in non-Caucasian populations.
Footnotes
ACKNOWLEDGMENTS
This work was supported by grants from the National Key R&D Program of China (2016YFC1305803), the National Natural Science Foundation of China (81701253), Qingdao Key Health Discipline Development Fund, Qingdao Outstanding Health Professional Development Fund, and Shandong Provincial Collaborative Innovation Center for Neurodegenerative Disorders.
