Background
The neurodegenerative disorder Alzheimer’s disease (AD) caused the most of dementia in the elderly [
1]. Previous findings indicated that the incidence increased from 1% in 65–69 year-olds to about 50% in 85–95 year-olds [
2]. Many genetic and environmental risk factors contribute to the degenerative progress of AD, such as family history, low income and education, exposure to aluminium in drinking water, dietary habits, smoking, physical activity, hypertension, diabetes and genetic variations [
3]. Molecular genetics researches have shown that AD was a class of complex polygenic diseases with genetic heterogeneity. Several genes have been reported to associate with AD. Beta-amyloid precursor protein (APP) and presenilin 2 played major role in early-onset familial AD [
4,
5]. The death-associated protein kinase 1(DAPK1) [
6] and ATP-binding cassette subfamily A member 7 (ABCA7) [
7] have been mainly implicated with late-onset AD. The ϵ4 allele of apolipoprotein E (
APOEϵ4) was the only verified risk factor for sporadic AD [
8]. However, the presence of variants for these genes and the
APOEϵ4 allele was neither necessary nor sufficient for AD development. About 50% of AD patients did not have mutations in the genes mentioned above or carry the
APOEϵ4 allele, and not everyone who has the mutations of the genes will acquire AD [
9], suggesting that it is necessary to identify additional genetic or non-genetic factors which modulate the AD susceptibility.
The main histopathologic features of AD are neurofibrillary tangles and Neuritic plaques which consist of hyperphosphorylated tau protein and amyloid peptides, respectively. Cathepsin D (
CTSD), an intracellular acid protease, contributed to the proteolytic cleavage of APP and the clearance of the β-amyloid (Aβ) from the central nervous system [
10,
11]. As such,
CTSD might involve in the pathogenesis of AD. Variants of
CTSD gene might impede the functions of proteolytic degradation, thus increasing the risk of AD. A
CTSD C224T polymorphism(C-to-T) in exon 2 can bring about amino acid change (Ala38-to-Val), increase pro-CTSD secretion and alter intracellular maturation [
12]. It has been proved that this polymorphism was significantly associated with the general intelligence of healthy elderly [
13].
Recently, numerous studies have focused on the correlation between the
CTSD C224T polymorphism and AD risk [
14‐
36]. Unfortunately, the results of these studies were contradictory. Five previous studies reported that the
T allele of the
CTSD-C/T polymorphism was a high-risk factor for developing AD [
14‐
18]; however, other relevant studies yielded contradictory results [
19‐
36]. Furthermore, the results of previous meta-analysis which research the association between the
CTSD polymorphism and AD risk were contradictory as well. Bertram et al. [
37] and Ntais et al. [
38] did not find any significant association, whereas Schuur [
18] reported that
T allele increased the risk of AD in Caucasians. Possible reasons for these contradictory results include the small sample size of the Ntais study; the absence of an Asian population in the Schuur study; and the fact that the Bertram study only compared alleles
T and
C. Considering that those factors could contribute to bias in the final result, we updated the present meta-analysis which included a larger sample size to provide a more reliable correlation between
CTSD C224T and AD.
Methods
Search strategy
The relevant studies were identified by searching PubMed, Embase, Web of Science, Google Scholar and Wan fang electronic databases in July 2013 for all the articles regarding the correlation between CTSD C224T polymorphism and AD risk. The key words of search strategy as follow: “Alzheimer’s disease or AD”, “CTSD or cathepsin D”, and “polymorphism, mutation or variant”. References listed in reviews and retrieved articles were also screened. There were no language or country restrictions. When multiple articles researched the same cohort, the one with the largest population was included. When a publication reported more than one subpopulation, we regarded every subpopulation as a separate study.
Selection criteria
The eligible studies were requested to agree with the inclusion criteria: (1) a case–control study; (2) research of the correlation between CTSD C224T polymorphism and AD susceptibility; (3)inclusion of the sample size and distribution of alleles and genotypes; (4) AD diagnosed according to the criteria of the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA), or the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV). Exclusion criteria of our study were followed as: (1) duplicated literature, reviews, or animal studies; (2) genotype frequency and distribution were not included; (3) not enough information for data extraction.
Two reviewers (Cuiju Mo and Jingzhe Sui) extracted the information independently. If there was a disagreement, the data was checked again, and a third reviewer (Xue Qin) was invited to check the data. Information collected from each eligible study was included: first author, year of publication, country, ethnicity, genotyping method, AD diagnosis, control sources, sample sizes, age of onset, and genotype distribution in cases and controls.
Statistical analysis
All analysis was conducted using Stata version 12.0 software (Stata Corp, College Station, TX). The association was assessed by pooled odds ratio (OR) together with the corresponding 95% confidence interval (CI). Only heterozygote comparison model (TC vs. CC) and the dominant genetic model (TT + TC vs. CC) were analysed. Furthermore, we evaluated the effect in different subgroup stratified by ethnicity (Asian vs. Caucasian) and age of onset. Early-onset AD (EOAD) was defined as age at onset <65 years, and age at onset ≥65 years was considered as late-onset AD (LOAD). To evaluate the interaction of the CTSD with the APOEϵ4 allele, we compared the dominant genetic model (TT + TC vs. CC) between case and control subjects stratified by the APOEϵ4 allele status. Similarly, the relationship of the APOEϵ4 allele with AD risk was investigated between the patients carrying the T allele or not.
The x2-test based Q-statistic and I2 statistic was used to evaluate the heterogeneity among the studies. The DerSimonian–Laird random-effects model was used to assess pooled OR when a significant heterogeneity (PQ < 0.1 or I2 ≥ 50%) was observed. Otherwise, the Mantel–Haenszel fixed-effects model was used. The publication bias was detected by funnel plot and Egger’s test. An Egger’s test P value <0.05 was considered as statistically significant. The genotype distribution of the control population was used to evaluate Hardy–Weinberg Equilibrium (HWE) by a goodness-of-fit Chi-square test. P <0.05 (two-side) was considered as statistically significant.
Discussion
The effects of genetic sequence variants in complex human traits are not readily detectable in population samples. However, meta-analysis that accumulates published data from small single research is a valuable tool in identifying disease genes. The functions of CTSD are to hydrolyse APP protein and clear Aβ from the central nervous system [
10,
11]. In AD patients, CTSD was expressed in the core of neuritic plaques [
45], and cellular and cerebrospinal levels are elevated [
46]. The variants of this gene might impede the proteolytic cleavage of APP and the degradation and clearance of Aβ, the synthesis of which is a supposed pivotal event in the pathogenesis of AD. Therefore, our motivation for the present study was to determine the association between
CTSD polymorphism and AD risk from abundant data over 16,651 genotype cases and controls.
As far as we know, the present meta-analysis involving 5602 cases and 11,049 healthy controls was the most comprehensive to date to investigate the relation between the
CTSD C224T polymorphism and AD susceptibility. Our finding indicated that the
C224T polymorphism was not associated with the AD risk both in Asian and Caucasian populations, which were in accord with the results of the previous meta-analysis [
38] and inconsistent with Schuur’s results [
18]. Compared to the previous study, our meta-analysis has some particular strength. First, we had the largest sample size; we added four Asian population studies, the absence of which in the Schuur study might have caused a deviation in the final result; and ten new case–control studies were added compared to the Ntais study, which might have effectively altered the overall results. Second, because nearly half of the eligible studies did not detect the homozygous
TT polymorphism, and the proportion of
TT was very small, as is usual in common polymorphisms, heterozygote might be responsible for the significant difference in frequency; therefore, we only compared the CT vs. CC and the dominant CT + TT vs. CC models. Lastly, no significant publication bias was observed in any of the studies analyzing by Egger’s test and Begg’s funnel plot. Thus, based on the above factors, the results of our meta-analysis were more reliable than those of previous studies.
Our results from the CT vs. CC and dominant CT + TT vs. CC comparison models suggested that no significant correlation was existed between the
CTSD C224T polymorphism and AD risk. Given that the control genotypes of two case-control studies [
22,
29] were out of HWE, they might have contributed some bias to our summary OR. When we excluded those two studies, the summary OR was not effectively altered, showing that our result was reliable. A great degree of heterogeneity among studies was identified for CT vs. CC (x
2 = 39.65, P
Q = 0.023) and CT + TT vs. CC (x
2 = 44.23, P
Q = 0.007) in the overall populations. Several factors might contribute to the heterogeneity. First, AD is a complicated and multi-genetic disease. Second, clinical heterogeneity, such as gender, age of onset, and diagnosis criteria, were factors. The different studied populations, such as ethnicity, might also explain the discrepancy. In subgroup analysis stratified by ethnicity and age of onset, heterogeneity only existed in the Caucasian subgroup, indicating that age was the major contributor to the existence of all heterogeneity.
Considering the impact on the summary OR of different ethnicities, we further performed subgroup analysis based on ethnicity. Those results indicated no significant association between the
CTSD C224T polymorphism and AD risk either in Asian or in Caucasian population, which was inconsistent with the previous meta-analysis[
18]. Similarly, the results did not change when the two studies that violated HWE [
22,
29] were excluded. The number of samples in the Asian subgroup was dramatically less than those in the Caucasian subgroup, which may weaken the conclusions. Our results also differed from the Schurr study after excluding the Mateo study in Caucasian population. While after excluding Albayrak [
39] and Mateo [
33] study, a significant association was found in the dominant CT + TT vs. CC genetic model(OR = 1.201, 95% CI = 1.004-1.436, P = 0.045). The principal cause for the difference with our results was the inclusion of the Albayrak [
39] study. The Albayrak study reported that the
CTSD C224T polymorphism increased AD risk in men only which might cause the false-negative result. As no study has clarified gender-specific differences regarding lysosomes or its components and the characteristic lesions in AD, therefore, future study with larger samples to investigate the gender-specific is necessary. When stratified by age of onset, we found no significant differences both in EOAD and LOAD subsets. Possible explanations for these findings might be the small sample sizes for analysis; the same control source, without strict age matching, and missing age information in some studies. Given these factors may affect the statistical power. Further research is required to assess the gene effects and validate our findings.
To evaluate the interaction of
CTSD polymorphism and
APOEϵ4 allele on AD, ten studies which provided genotype distribution data of
APOEϵ4 status were chosen for further study, and of which only four showed evidence of an association [
14,
15,
20,
25]. The results of our study showed non-significant relation between the
C224T polymorphism and AD risk in
APOEϵ4 carriers and non-carriers. The association of
CTSD T allele with AD risk between
APOEϵ4 carriers and non-carriers in Caucasians was quite similar, contrary to the Schuur result. Due to the lack of an Asian population in the Schuur study, sample size and ethnicity might have contributed to some bias in the final result. While the ORs of
APOEϵ4 were greater in the
T allele carriers group than the subjects without the
T allele. Because of the extensive overlap in two effect sizes and the remarkably small group of subjects who carry both the
APOEϵ4 and
CTSD T alleles, the association between the
CTSD T and
APOEϵ4 alleles should be interpreted cautiously.
There were some limitations that merit attention. First, some of the eligible studies lacked sufficient information for detailed and deep analysis. In some studies, the controls were not uniformly defined as matched by age and gender; and it may lead to some negative correlation. Second, we mainly focused on the C224T polymorphism, discounted the potential linkage disequilibrium with another mutation of this gene, and ignored the interactions between gene and gene or gene and environment. Third, the data of our meta-analysis was unadjusted; the suspected factors could be analysed, such as, gender, diet, lifestyle habit, and environmental factors. Fourth, we included the English or Chinese publications only; the lack of unpublished data and data published in other languages might contribute some bias. There were only four articles in the Asian subgroup, with small sample size, which may cause low statistical power.
Competing interests
All authors state no conflict of interest.
Authors’ contributions
CM, QL, XQ, SL originally designed the study, collected the literature data and drafted the manuscript. JS, JW, YD performed data extraction and statistical analysis. LX, TL, YH executed literature search and checked the results. All authors have read and agreed with the final manuscript.