| |
“The findings and conclusions in this review are those of the author(s) and do not
necessarily represent the views of the funding agency.” |
This HuGE Review was published in the Am J of Epidem 2004;159:527-536 |
|
|
Christos Ntais 1, Anastasia Polycarpou 1 and John P. A. Ioannidis 1,2,3
1 Clinical and Molecular Epidemiology Unit, Department of Hygiene and Epidemiology, University of Ioannina School of Medicine, Ioannina, Greece.
2 Biomedical Research Institute, Foundation for Research and Technology-Hellas, Ioannina, Greece.
3 Institute for Clinical Research and Health Policy Studies, Department of Medicine, Tufts-New England Medical Center, Tufts University School of Medicine, Boston, MA. Received for publication August 5, 2003. Accepted for publication October 10, 2003.
|
ABSTRACT |
A C-to-T polymorphism in exon 2 of the cathepsin D gene encoding cathepsin D ( CTSD ) has been implicated as a risk factor for Alzheimer's disease. The authors performed a meta-analysis of 14 studies (16 comparisons) with CTSD genotyping (3,174 Alzheimer's disease cases and 3,298 controls). Overall, the random effects odds ratio for the T versus the C allele was 1.17 (95% confidence interval (CI): 0.95, 1.44), with some between-study heterogeneity ( p < 0.01). There was significant between-study heterogeneity but no evidence of a significant association when the first hypothesis-generating study was excluded from the calculations (odds ratio (OR) = 1.11, 95% CI: 0.91, 1.35; p = 0.29). The summary odds ratio for T carriers versus T noncarriers was similar in subjects carrying or not carrying an apolipoprotein E
ε4 allele ( APOE * 4 ). The increased susceptibility to Alzheimer's disease conferred by APOE * 4 carriage tended to be more prominent in the presence of the T allele (random effects OR = 6.07, 95% CI: 4.19, 8.79, and OR = 4.09, 95% CI: 3.15, 5.31, in T carriers and noncarriers, respectively). The meta-analysis shows that the CTSD polymorphism is not a major risk factor for Alzheimer's disease, although a small effect or an enhancement of the APOE * 4 effect cannot be excluded.
Keywords: Alzheimer disease; cathepsin D; CTSD; epidemiology; genetics; meta-analysis; polymorphism (genetics)
Abbreviations: CI, confidence interval; OR, odds ratio.
 |
GENE |
Cathepsin D, an intracellular acid protease which exhibits beta-secretase activity in vitro, has been implicated in the processing of the amyloid precursor protein and the tau protein (1, 2). The cathepsin D gene ( CTSD ) is located on the short arm of chromosome 11 (11p15.5) and consists of nine exons. The synthesis of beta-amyloid peptide is a putative key event in the pathogenesis of Alzheimer's disease. Beta-amyloid derives from its precursor protein via proteolytic cleavage by secretases. Therefore, it has been postulated that variants in the genes coding for enzymes involved in the proteolytic cleavage of amyloid precursor protein or in the degradation and clearance of beta-amyloid from the central nervous system may be potential risk factors for Alzheimer's disease.
|
GENE VARIANTS |
| The CTSD gene contains a polymorphic C-to-T transition site at position 224 in exon 2. This polymorphism results in an Ala38-to-Val substitution in the cathepsin D profragment (3). The polymorphism has been associated with increased secretion and altered intracellular maturation of the cathep-sin D profragment in one study (3). Moreover, the T allele has been associated with a 50 percent decrease in beta-amyloid peptide 1-42 levels in the cerebrospinal fluid of patients with Alzheimer's disease (4). Finally, this polymorphism was recently reported to be significantly associated with general intelligence in healthy elderly people (5).
Molecular epidemiologic studies have presented seemingly contradictory results concerning a potential role of CTSD polymorphism in Alzheimer's disease (6-19). There is also controversy on whether this polymorphism may interact with the apolipoprotein E ε4 allele ( APOE * 4 ), which is the best known genetic determinant for sporadic Alzheimer's disease (20). Single studies may have been underpowered to detect interactions or even overall effects. Given the amount of accumulated data, we deemed it important to perform a quantitative synthesis of the evidence using rigorous methods. Thus, we conducted a comprehensive meta-analysis of all available studies relating the CTSD polymorphism to the risk of Alzheimer's disease.
|
DISEASE |
Alzheimer's disease is the most common cause of progressive cognitive impairment in the elderly, with an annual incidence of approximately 1 percent at 65-69 years increasing up to 40 percent in the very elderly (>85 years of age) (21, 22). Mutations in the amyloid precursor protein, presenilin 1, and presenilin 2 genes account for 5 percent of the cases and result in an autosomally dominant pattern that is expressed with complete penetrance and early manifestations (23). Alzheimer's disease is probably slightly more common in females (24). Other proven or postulated risk factors include head injury (in particular among males) (25), as well as family history, low income, low education, low occupational status, depression, exposure to aluminum in drinking water, hypertension, and Down's syndrome (26 , 27). Conversely, the use of nonsteroidal antiinflammatory drugs to treat arthritis has been associated with a reduced risk of Alzhei-mer's disease, as has been estrogen use by postmenopausal women (27). Physical activity, diets with high levels of vitamins B6 , B12 , and folate, and red wine in moderate quantities may be protective (27). The prevalence of Alzheimer's disease varies considerably among different population groups (28). At least a few dozens of polymorphisms have been examined in relation to sporadic Alzheimer's disease, and published reviews are available (23, 29). Among them, there is conclusive evidence from several studies and meta-analysis thereof that APOE * 4 is a strong risk factor for developing Alzheimer's disease for both male and female subjects and for both early onset (<65 years) and late-onset disease (20), with an approximately fivefold increase in the odds of developing Alzheimer's disease. Single studies have also implicated other polymorphisms as being important, although the reported odds ratios are much smaller than those seen for APOE * 4, and attempts at replication in subsequent research have not been conclusive. Meta-analyses for some other polymorphisms have already appeared in the literature (30-35). They suggest no significant overall associations for several of these polymorphisms, including the myeloperoxidase gene promoter polymorphism (30), intronic or promoter region polymorphisms of presenilin 1 (for late-onset disease) (31), an insertion-deletion polymorphism or a missense mutation in the alpha-2 macroglobulin gene (32), and several polymorphisms of the protein tau gene (33). Associations of modest effect size (odds ratios (ORs) = 1.30-1.35) have been claimed in meta-analyses of the low density lipoprotein receptor-related protein gene exon 3 polymorphism (34) and of an insertion-deletion polymorphism in the angiotensin-converting enzyme I gene (35).
|
META-ANAYLSIS METHODS |
Identification and eligibility of relevant studies
We considered all studies that examined the association of the CTSD polymorphism with Alzheimer's disease. Sources included MEDLINE and EMBASE (from January 1994 to September 2003). The search strategy was based on combinations of "Alzheimer's disease," " CTSD," "cathepsin D," "polymorphism," "allele," and "genetics." References of retrieved articles were also screened.
Case-control studies were eligible if they had determined the distribution of CTSD genotypes in Alzheimer's disease cases (regardless of age of onset) and in a concurrent control group of dementia-free subjects using a molecular method for genotyping. Cases with Alzheimer's disease were eligible regardless of whether they had a family history of Alzheimer's disease or not. However, we excluded family-based studies of pedigrees with several affected cases per family, because their analysis is based on linkage considerations.
Data extraction
Two investigators independently extracted data and reached consensus on all items. The following information was sought from each report: authors, journal and year of publication, country of origin, selection and characteristics of Alzheimer's disease cases and controls, demographics, ethnic group of the study population, eligible and genotyped cases and controls, and number of cases and controls for each CTSD genotype. For studies including subjects of different ethnic groups, data were extracted separately for each ethnicity, whenever possible. Furthermore, we examined whether matching had been used, whether there was specific mention of blinding of the personnel who performed the genotyping to the clinical status of the subjects, and whether the genotyping method had been validated.
Meta-analysis
The primary analysis compared Alzheimer's disease cases with controls for the contrast of T versus C alleles. This analysis aims to detect overall differences. We also examined the contrast of extremes (homozygotes), T / T versus C / C . Finally, we examined the contrast of T / T versus ( C / T + C / C) and the contrast of ( C / T+ T / T) versus C / C. These contrasts correspond to the recessive and dominant effects, respectively, of the T allele.
The odds ratio was used as the metric of choice. For each genetic contrast, we estimated the between-study heterogeneity across all eligible comparisons using the chi-square-based Q statistic (36). Heterogeneity was considered significant for p < 0.10. Data were combined using both fixed-effects (Mantel-Haenszel) and random-effects (DerSimonian and Laird) models (37). Random effects incorporate an estimate of the between-study variance and tend to provide wider confidence intervals, when the results of the constituent studies differ among themselves. In the absence of between-study heterogeneity, the two methods provide identical results. Random effects are more appropriate when heterogeneity is present (37).
We also performed cumulative meta-analysis (38) and recursive cumulative meta-analysis
(39, 40) to evaluate whether the summary odds ratio for the T versus C contrast changed over time as more data accumulated and whether the strength of the association changed when the first hypothesis-generating study was excluded from the calculations (41). Inverted funnel plots and the Begg-Mazumdar publication bias diagnostic (nonparametric  correlation coefficient) (42) evaluated whether the magnitude of the observed association was related to the variance of each study, that is, whether large studies gave different results compared with smaller ones
(43). Finally, we evaluated whether the summary results were different when the analysis was limited to studies with more intensive efforts to exclude Alzheimer's disease from controls (those that clearly performed neuropsychological testing for all controls).
Previous investigations have alluded to the possibility that the T allele may interact with the APOE * 4 allele in conferring susceptibility to Alzheimer's disease (7,9). Thus, we also evaluated the effect of T allele carriage on the risk of Alzheimer's disease separately for APOE * 4-positive and APOE * 4 -negative subjects. Moreover, we evaluated the genetic effect conferred by the presence of APOE * 4 separately in subjects carrying the T allele and those not carrying the T allele. Odds ratios were combined with fixed and random effects models, as described above. When these data were not reported, we communicated with the primary investigators to obtain this information, whenever possible.
Analyses were performed with SPSS 11.0 (SPSS, Inc., Chicago, Illinois) and Meta-Analyst (Joseph Lau, Boston, Massachusetts) software. Whenever there were 0 values in a 2 x 2 table, we added 0.5 to all four cells, so that an odds ratio could be calculated. All p values are two tailed.
|
META-ANAYLSIS RESULTS |
Eligible studies
Fourteen studies probing the relation between the CTSD polymorphism and Alzheimer's disease susceptibility were identified (6-19) and are profiled in Table 1. Two of the eligible studies (6, 15) contained subjects of two different ethnic groups, so a total of 16 separate comparisons were considered. There was considerable diversity of ethnic groups. Eleven studies (6-15, 19) selected Alzheimer's disease cases according to criteria from the National Institute of Neurological and Communicative Disorders and Stroke-Alzheimer's Disease and Related Disorders Association (NINCDS-ADRDA), two studies (16, 18) selected Alzhei-mer's disease cases according to Diagnostic and Statistical Manual of Mental Disorders: DSM-IIIR and DSM-IV criteria, and one study (17) did not clarify the exact criteria used for the diagnosis of Alzheimer's disease. Five studies (13-17) also included autopsy-confirmed Alzheimer's disease cases. Two studies (7, 16) mentioned that they included cases with a family history of Alzheimer's disease, eight studies (9-14, 18, 19) specifically excluded such patients, and the remaining did not clarify the background of family history. One study (6) mentioned that 61 percent of Alzheimer's disease cases had late-onset disease (age of onset: >65 years), five studies (8, 12-14, 19) specifically included only late-onset Alzheimer's disease cases, one study (18) specifically excluded such patients, and the remaining did not clarify the age at onset. Controls did not have a diagnosis of Alzheimer's disease, but the amount of additional screening (general physical and neurologic examination, psychiatric interview, neuropsychological testing, blood and cerebrospinal fluid studies, computed tomography scan, and Mini-Mental State Examination score) to exclude Alzheimer's disease differed substantially across studies.
|
Table 1: Characteristics of studies included in the meta-analysis
|
Specific matching for age was described in five studies (10, 12-15). One study also matched for sex (10). Only one study (7) specifically mentioned blinding of the personnel who performed the genotyping. Appropriate molecular methods for genotyping were used. All studies used polymerase chain reaction, and two studies (14, 18) also used dynamic allele-specific hybridization.
Meta-analysis database
The eligible studies summarized in Table 2 included a total of 3,175 cases with Alzheimer's disease and 3,334 controls, of whom 3,174 and 3,298, respectively, had genotype data. The T allele was more highly represented among controls of American descent (overall prevalence of 8.6 percent, 95 percent confidence interval (CI): 7.1, 10.1) than in controls of European (7.8 percent, 95 percent CI: 7.0, 8.6) or Asian (0.9 percent, 95 percent CI: 0.3, 1.5) descent. There was significant heterogeneity in the prevalence rates of the T allele even across the control subjects of European descent, with a rate of 14.3 percent among Hispanic Americans, 10.4 percent in Italy, 10.1 percent in Spain, and lower rates in northern European countries (7.6 percent in the United Kingdom, 6.3 percent in Sweden, 6.1 percent in Germany, and the lowest prevalence rate of 4.5 percent in a Polish population). Overall, the prevalence of T / T homozygosity was 0.3 percent, 0.8 percent, and 0.2 percent in control subjects of American, European, and Asian descent, respectively. The respective prevalence rates of C / T heterozygosity were 16.6 percent, 14.0 percent, and 1.5 percent, and the respective rates for C / C homozygosity were 83.1 percent, 85.2 percent, and 98.3 percent. The distribution of genotypes in control groups was consistent with Hardy-Weinberg equilibrium in all studies.
|
Table 2: Distribution of CTSD alleles among Alzheimer’s disease cases and controls in the included studies
|
Overall effects
There was a trend suggesting that the T allele may confer increased susceptibility to Alzheimer's disease
(Figure 1). As shown in Table 3 , the summary odds ratio was 1.17 by random effects ( p = 0.14), and there was significant heterogeneity among the 16 study comparisons ( p < 0.01 for heterogeneity). We found no evidence of an association of the T / T genotype with the risk of Alzheimer's disease relative to the C / C genotype. There was no significant between-study heterogeneity. No evidence of an association with Alzheimer's disease was discerned also when the recessive model was examined for the effect of T, while a trend for an association was seen in the dominant model (by random effects, OR = 1.19, 95 percent CI: 0.97, 1.47; p = 0.10). There was no between-study heterogeneity in the recessive model contrast, while significant heterogeneity ( p < 0.05) was still seen for the dominant model contrast. Subgroup analysis of studies with cases and controls of European descent yielded similar results (15 comparisons (11,436 alleles): OR = 1.18, 95 percent CI: 0.96, 1.46; p = 0.12) ( p < 0.01 for heterogeneity).
|
Table 3 : Summary odds ratios for various contrasts
|
Bias diagnostics
In cumulative meta-analysis and recursive cumulative meta-analysis, the magnitude of the summary odds ratio had not been stable over time, and it had changed considerably per year with an apparent dissipation of the postulated effect (by random effects, summary OR for T vs. C : 2.05 at the end of 1999, 1.41 at the end of 2000, 1.16 at the end of 2001, 1.14 at the end of 2002, and 1.17 in 2003). Excluding the first hypothesis-generating study (9), we found that the summary odds ratio became 1.11 (95 percent CI: 0.91, 1.35; p = 0.29) with significant between-study heterogeneity ( p < 0.05), and there was no evidence of any association even in the comparison of the homozygotes ( T / T vs. C / C: OR = 0.96, 95 percent CI: 0.48, 1.90; p = 0.91), with no between-study heterogeneity. There was no relation between the effect size and the variance of each study, suggesting that larger studies agreed with the results of smaller studies. Analyses limited to studies with more intensive efforts to exclude Alzheimer's disease from controls yielded similar results (11 comparisons (10,382 alleles): OR = 1.22, 95 percent CI: 0.96, 1.54; p = 0.10) ( p = 0.02 for heterogeneity).
Interaction with the APOE genotype
Nine studies (7-9, 11-14, 16, 18) obtained data on both CTSD and APOE genotypes. With one study (8) separating male and female subjects, 10 comparisons became available.
These nine comparative studies tended to gave a slightly inflated effect for the T allele, as compared with the full meta-analysis of 14 comparisons (summary OR by random effects = 1.23 vs. 1.17 for the complete meta-analysis); thus, inferences should be cautious. Among carriers of the high-risk APOE * 4 allele, T allele carriers had a random-effects odds ratio of 1.38 (95 percent CI: 0.89, 2.15) for Alzheimer's disease compared with subjects not carrying the T allele. Among subjects not carrying the APOE * 4 allele, the respective odds ratio was 1.13 (95 percent CI: 0.90, 1.42). There was significant between-study heterogeneity in the APOE * 4-positive group ( p = 0.07). The two effect sizes overlapped widely.
Among carriers of the T allele, the presence of APOE * 4 increased the risk of Alzheimer's disease 6.07-fold (95 percent CI: 4.19, 8.79), with no between-study heterogeneity. Among subjects without the T allele, the presence of APOE * 4 increased the risk of Alzheimer's disease 4.09-fold (95 percent CI: 3.15, 5.31), with significant between-study heterogeneity ( p < 0.01). The two estimates overlapped widely in individual studies and overall, but typically the odds ratios were larger in the group of T allele carriers (Figure 2).
|
| DISCUSSION |
This meta-analysis includes data from 14 case-control studies with over 6,000 genotyped Alzheimer's disease cases and controls, and it proves that the CTSD polymorphism is not a strong risk factor for Alzheimer's disease. The current evidence cannot exclude that the T allele of the CTSD polymorphism may increase modestly the risk of Alzheimer's disease, but there was significant heterogeneity in the results of various studies. However, we found that bias might exist, with a decreasing effect size, as more data accumulated over time. Moreover, the meta-analysis also cannot exclude the possibility that the presence of the CTSD T allele may enhance the increased susceptibility toward Alzheimer's disease conferred by APOE * 4. However, even if this is true, it would pertain to the relatively small group of subjects who carry both the APOE * 4 and CTSD T alleles. In all, the impact of the T allele on a population level would be small, if present at all.
The meta-analysis findings may be interpreted against the postulated biologic context of the CTSD polymorphism. Cathepsin D is an intracellular acid protease with beta-secretase activity in vitro (1, 2) that can cleave amyloid precursor protein and the tau protein to generate fragments with intact microtubule-binding domains (44), which might play a role in the pathogenesis of paired helical filaments. One study has shown that the CTSD polymorphism is associated with increased secretion and altered intracellular maturation of procathepsin D (3). It should be noted that it is not clear whether the CTSD polymorphism has functional consequences for the mature form of the enzyme. Although we cannot exclude a biologic effect for the CTSD polymorphism, our findings are in accordance with the results of a recent full genome scan showing no significant linkage of Alzheimer's disease to the short arm of chromosome 11, the region where the CTSD gene is located (45-47).
Attention should be given to the design of individual studies. The results of meta-analyses may be affected by methodological problems and potential biases in the designs of the constituent studies. Nondifferential misclassification errors may dilute the strength of an observed association. Alzheimer's disease cases were generally selected according to appropriate criteria. However, some young control subjects may have developed Alzheimer's disease at older ages. The choice of an appropriate age window for assessing a postulated genetic risk factor for Alzheimer's disease is difficult. Studies of younger subjects may be more suitable for identifying risk factors that result in early onset Alzheimer's disease. Conversely, selection of younger subjects may be less appropriate, if the influence of the postulated genetic risk factor is more important in late-onset Alzheimer's disease.
Subgroup effects and effect modification (e.g., differential effects of a genetic polymorphism on early vs. late-onset Alzheimer's disease cases or familial vs. sporadic disease) or complex interactions with other genes may also need to be considered (48). Our analyses addressed interactions with APOE * 4, the major known genetic determinant of Alzhei-mer's disease. Interactions with other genetic or environmental factors have not been studied. The trend for a stronger effect of APOE * 4 in the presence of T allele carriage is interesting in the light of data suggesting that T carriage may affect the general intelligence (5). However, subgroup and interaction analyses should be interpreted cautiously, since differences between subgroups may occur by chance (49) and their validation would require studies with even larger sample sizes than the several thousand included in this meta-analysis. Finally, population stratification may theoretically have affected the results of the constituent studies in the meta-analysis (50), since we documented that the frequency of the T allele varied considerably among the different ethnic groups or even among the different ethnic groups of European descent. However, most studies were strictly ethnically defined, so the population stratification effect is unlikely to have been of any considerable magnitude.
Because of the increasing prevalence of Alzheimer's disease worldwide, it is crucial to identify genetic risk factors for this neurodegenerative disease. Thus, the list of identified polymorphisms that may influence the risk of Alzheimer's disease is continuously expanding
(23, 29), but most of the reported associations of candidate genes to date remain nonreplicated or at least controversial after subsequent investigation. Early and small genetic association studies may come up with spurious findings (41, 51, 52). Even when genetic associations are replicated, usually they do not have a major public health impact that would lead to screening recommendations (53). Nevertheless, such knowledge could improve our understanding about the pathogenesis of complex diseases such as Alzheimer's disease.
|
LABORATORY TESTING |
The methods used for CTSD genotyping in the analyzed studies are straightforward and include polymerase chain reaction (9) and dynamic allele-specific hybridization (14). The error rate due to misclassification is likely to be small. Future studies should nevertheless ensure and clearly report that assessment of genotyping has been performed while blinded to the clinical status of the patient.
|
POPULATION TESTING |
| To date there has been no population testing of the CTSD polymorphism. Based on the results of the meta-analysis, such testing would not be indicated given the currently available data.
|
ACKNOWLEDGMENTS |
The authors thank Jonathan A. Prince, Patrizia Mecocci, and Anthony J. Brookes for providing additional data from their studies.
|
NOTES |
Correspondence to Dr. John P. A. Ioannidis, Department of Hygiene and Epidemiology, University of Ioannina School of Medicine, Ioannina 45110, Greece (e-mail: jioannid@cc.uoi.gr )
|
TABLES AND FIGURES |
|
|
REFERENCES |
List of References
|
| INTERNET SITES |
|
|
|
|
Provides link to non-governmental sites and does not necessarily represent the views of the Centers for Disease Control and Prevention. |
|
|
Page last reviewed: June 8, 2007 (archived document)
Page last updated: November 2, 2007
Content Source: National Office of Public Health Genomics |
|
(last accessed 4/2007)
|