Autopsy studies have shown that adults with Down's syndrome aged 40 years and above almost universally exhibit Alzheimer's disease neuropathology (Reference MannMann, 1988), a fact that is supported by neuro-imaging studies (Reference Deb, de Silva and GemmellDeb et al, 1992). Saunders et al (Reference Saunders, Strittmater and Schmechel1993) reported an association between APOE ϵ4 and late-onset Alzheimer's disease. A number of studies have produced inconsistent support for APOE ϵ4 as a risk factor and APOE ϵ2 as a protective factor against Alzheimer's disease in people with Down's syndrome (Reference Prasher, Chowdhury and RowePrasher et al, 1997; Reference Tyrell, Cosgrave and HawiTyrell et al, 1998). Our study examined whether the age specific manifestation of Alzheimer's disease in adults with Down's syndrome was influenced by risk factors such as age, APOE status, and genotype for an intronic polymorphism in the PS-1 gene, which is also reputedly associated with late onset Alzheimer's disease (Reference Kehoe, Williams and HolmansKehoe et al, 1996). We examined the APOE polymorphism in three groups: (a) a population-based sample of adults with dementia with Down's syndrome aged 35 years and over; (b) an elderly group of non-demented adults with Down's syndrome drawn from the same population-based sample; and (c) a group of normal non-demented adults selected for intelligence, collected from the same geographical area as the first two groups. We also carried out a meta-analysis comprising all published studies of association between the APOE gene and Alzheimer's disease in adults with Down's syndrome, incorporating the data from the current study.
METHOD
Subjects
The names of all known adults with Down's syndrome over the age of 35 years were collected from the clinicians and staff of Community Learning Disability Teams in all five Health Districts in South Wales. Subjects were assessed for Alzheimer's disease using the ICD-10 (World Health Organization, 1992) criteria in accordance with the guidelines produced by an international consensus panel established under the auspices of the Ageing Special Interest Group of the International Association for the Scientific Study of Intellectual Disabilities (IASSID) (Reference Aylward, Burt and ThorpeAylward et al, 1997). The diagnosis of dementia was further supported by using two observer-rated questionnaires, namely the Dementia Questionnaire for Mentally Retarded Persons (DMR) (Reference EvenhuisEvenhuis, 1992) and the Dementia Scale for Down's syndrome (DSDS) (Reference GedyeGedye, 1995). Both questionnaires are reported to have good inter-rater reliability and internal validity (Reference EvenhuisEvenhuis, 1992; Reference GedyeGedye, 1995). Deb & Braganza (Reference Deb and Braganza1999) have recently reported a positive correlation between diagnosis of dementia by a clinician in adults with Down's syndrome and diagnosis carried out according to the DSDS criteria (specificity=0.89, sensitivity=0.85), and the DMR criteria (specificity=0.92, sensitivity=0.92).
The age of onset of dementia was determined by asking the carers of adults with Down's syndrome when clinical symptoms of dementia were first noticed. Diagnosis of Down's syndrome was based on the characteristic clinical features and no karyotyping was carried out. The level of intellectual disability was determined either on the basis of IQ scores or the assessment of adaptive behaviours.
We excluded adults with Down's syndrome who showed medical, psychiatric, neurological or laboratory characteristics not related to Alzheimer's disease, but which might explain their mental deterioration, in particular severe hearing loss, untreated hypothyroidism and depression (n=2). In all, 24 adults with Down's syndrome who had a diagnosis of Alzheimer's disease were included in this study. We ascertained a second group of 33 individuals, identified as the oldest adults from a list of non-demented adults with Down's syndrome. As the prevalence of dementia in adults with Down's syndrome tends to increase with age (Reference Lai and WilliamsLai & Williams, 1989), it is anticipated that those who have lived longer without manifesting clinical dementia have least risk factors for developing dementia, and therefore provides an appropriate group for comparison. Finally, a control group of 164 non-demented adults representative of the local population and unselected for intelligence were used (54% male, mean age 38.27 years, s.d.=12.16 years).
Genotyping
Both the adults with dementia and the non-demented adults with Down's syndrome, and the non-demented, non-learning disabled control group were genotyped for both the APOE and PS-1 polymorphisms using standard techniques (Reference Wenham, Price and BlundellWenham et al, 1991; Reference Wragg, Hutton and TalbotWragg et al, 1996). For a detailed description of PS-1 data see Deb et al (Reference Deb, Braganza and Owen1998), but data in relation to a multivariate analysis of risk factors in relation to Alzheimer's disease in Down's syndrome will be presented in this paper. The study was performed with the approval of the local research ethics committee and written, informed consent was obtained from participants or carers where appropriate.
Statistical analysis
The χ2 and Fisher's exact tests were used to test for association between APOE and Alzheimer's disease in Down's syndrome. Multiple logistic regression analysis was also carried out to estimate the relative influence of risk factors such as age, gender, APOE and PS-1 genotype on the development of Alzheimer's disease in adults with Down's syndrome. The Woolf method (Reference WoolfWoolf, 1955) was used to perform a meta-analysis of association between APOE genotype and Alzheimer's disease in Down's syndrome. A Mann-Whitney U-test was used to test for a relationship between APOE and the age of onset of Alzheimer's disease in Down's syndrome. A probability of less than one in 20 (P<0.05) was regarded as significant in all statistical analyses.
RESULTS
Current study
The age range of the total cohort of adults with Down's syndrome was between 35 and 72 years (mean 51, s.d.=7.8 years). Twenty-three (40.4%) were aged between 35 and 49 years, and 34 (59.6%) were 50 years or older. The age range of Down's syndrome adults with dementia was between 35 and 71 (mean 54, s.d.=7.24), and those without dementia was between 35 and 72 years (mean 49, s.d.=7.55). Thirty-one (54.4%) were male and 26 (45.6%) were female. Thirteen (22.8%) had mild (IQ 70-50), 37 (65%) moderate (IQ 50-35), and seven severe (12.2%) intellectual disability (IQ<35). There was no statistically significant difference in the proportion of subjects with different degrees of severity of learning disability between the adults with dementia and the non-demented adults with Down's syndrome.
The distribution of APOE alleles among the three study groups is shown in Table 1, while the distribution of APOE genotypes is presented in Table 2. No statistically significant differences were observed in any of the inter-group comparisons shown in Table 1 and Table 2. However, a higher frequency of ϵ4 allele (17% v. 9%; χ2=1.4, P=0.2) and a lower frequency of ϵ2 allele (0% v. 4.5%) (P=0.26: Fisher's exact test) were observed among adults with Down's syndrome with dementia compared with those without (see Table 1).
Table 1 APOE allele frequency in different groups
| ϵ 2, n (%) | ϵ 3, n (%) | ϵ 4 n (%) | |
|---|---|---|---|
| Adults with dementia with Down's syndrome (n=24) | 0 | 40 (83.0) | 8 (17.0) |
| Non-demented adults with Down's syndrome (n=33) | 3 (4.5) | 57 (86.5) | 6 (9.0) |
| All adults with Down's syndrome (n=57) | 3 (2.6) | 97 (85.0) | 14 (12.3) |
| Non-learning disabled control group (n=164) | 23 (7.0) | 264 (80.0) | 41 (13.0) |
Table 2 APOE genotypes in different groups
| ϵ 4ϵ4 n (%) | ϵ 4ϵ2 n (%) | ϵ 4ϵ3 n (%) | ϵ 3ϵ3 n (%) | ϵ 3ϵ2 n (%) | ϵ 2ϵ2 n (%) | |
|---|---|---|---|---|---|---|
| Adults with dementia with Down's syndrome (n=24) | 0 | 0 | 8 (33.3) | 16 (66.7) | 0 | 0 |
| Non-demented adults with Down's syndrome (n=33) | 0 | 0 | 6 (18.2) | 24 (72.7) | 3 (9.1) | 0 |
| All adults with Down's syndrome (n=57) | 0 | 0 | 14 (24.6) | 40 (70.2) | 3 (5.2) | 0 |
| Non-learning disabled control group (n=164) | 3 (1.8) | 3 (1.8) | 32 (19.5) | 108 (65.8) | 16 (9.7) | 2 (1.2) |
Adults with Down's syndrome were divided up into two age groups (below 50 years, and 50 years and older), and two gender groups (male and female). This was done because of Sekijama et al's (Reference Sekijima, Ideda and Tokuda1998) finding that the frequency of ϵ4 allele in Down's syndrome adults with Alzheimer's disease under 50 years was significantly higher (28.6%), and Schupf et al's (Reference Schupf, Kapell and Nightingale1998) finding of earlier onset of Alzheimer's disease in men with Down's syndrome. The APOE ϵ 4 allele frequency between the adults with dementia and the non-demented adults with Down's syndrome according to the age- and gender-groups are presented in Table 3. None of the inter-group comparisons in Table 3 was statistically significant.
Table 3 Frequency of APOE ϵ4 and ϵ2 alleles in adults with Down's syndrome according to age and gender group
| ϵ 4 frequency | ϵ 2 frequency | |||
|---|---|---|---|---|
| Demented subjects, n (%) | Non-demented subjects, n (%) | Demented subjects, n (%) | Non-demented subjects, n (%) | |
| Age <50 years (n=23) | 2 (25) | 2 (33.3) | 0 | 1 (33.3) |
| Age ≥50 years (n=34) | 6 (75) | 4 (66.6) | 0 | 2 (66.7) |
| Male (n=31) | 3 (37.5) | 3 (50) | 0 | 3 (100) |
| Female (n=26) | 5 (62.5) | 3 (50) | 0 | 0 |
A multiple logistic regression analysis was carried out in the whole cohort of adults with Down's syndrome using the presence of dementia as a dependent variable and age, gender, APOE status, and PS-1 polymorphism as convariates. The presence of dementia was only significantly related to age in adults with Down's syndrome.
To assess the influence of APOE ϵ4 on the age of onset of dementia in adults with Down's syndrome we compared the mean age of onset between those who had an ϵ4 allele and those who did not. The mean age of onset of dementia among ϵ4 positive cases (at least one ϵ4 allele) was 51 years and 53 years for those who did not have an ϵ4 allele. A comparison of ages of onset in those with and without an ϵ4 allele was not statistically significant.
Meta-analysis
Prasher et al (Reference Prasher, Chowdhury and Rowe1997) presented a meta-analysis of the data collected from all the known published papers in addition to their own data in relation to APOE status in adults with Down's syndrome with and without Alzheimer's disease. We have added data from two other recent studies (Reference Sekijima, Ideda and TokudaSekijama et al, 1998; Reference Tyrell, Cosgrave and HawiTyrell et al, 1998), and our current study to those analysed by Prasher et al (Reference Prasher, Chowdhury and Rowe1997) and this is presented in Table 4. However, we excluded Wisniewski et al's (Reference Wisniewski, Morelli and Wegiel1995) data, which were included in Prasher et al's (Reference Prasher, Chowdhury and Rowe1997) meta-analysis, because unlike all the other studies they diagnosed Alzheimer's disease on the basis of neuropathological findings alone in the absence of any clinical data. The previous meta-analysis (Reference Prasher, Chowdhury and RowePrasher et al, 1997) did not show a statistically significant difference in the distribution of APOE alleles between the adults with dementia and Down's syndrome and the non-demented adults with Down's syndrome. However, the meta-analysis in the current study showed a statistically significant excess of APOE ϵ4 in Down's syndrome cases with Alzheimer's disease compared with those without (odds ratio=2.02, 95% CI 1.33-3.07, χ2=10.83, P=0.001) with no evidence of heterogeneity (P=0.2). We did not observe a significantly lower rate of APOE ϵ2 allele frequency in Down's syndrome adults with Alzheimer's disease (odds ratio=0.69, 95% CI 0.35-1.37). The odds ratios along with 95% CIs of those individual studies where a statistically significant excess of APOE ϵ4 or reduction of ϵ2 allele frequency was observed among adults with Down's syndrome with Alzheimer's disease are presented at the bottom of Table 4.
Table 4 Meta-analysis of pooled data from reports on APOE allele frequency in adults with and without dementia (updated from Prasher et al's (Reference Prasher, Chowdhury and Rowe1997) data)
| Study | Adults with Down's syndrome with dementia (allele frequency) n (%) | Adults with Down's syndrome without dementia (allele frequency) n (%) | ||||||
|---|---|---|---|---|---|---|---|---|
| Total | ϵ 2 | ϵ 3 | ϵ 4 | Total | ϵ 2 | ϵ 3 | ϵ 4 | |
| Royston et al (Reference Royston, Mann and Pickering-Brown1994) | 34 | 1 (3.0) | 25 (73.5) | 8 (23.5) | 10 | 5 (50.0) | 5 (50.0) | 0 |
| van Gool et al (Reference van Gool, Evenhuis and van Duijin1995) | 52 | 3 (5.8) | 42 (80.8) | 7 (13.4) | 52 | 6 (11.0) | 41 (78.8) | 5 (9.6) |
| Martins et al (Reference Martins, Clarnette and Fisher1995)1 | 12 | 1 (8.3) | 7 (58.3) | 4 (33.4) | 34 | 5 (14.7) | 28 (82.4) | 1 (2.9) |
| Lambert et al (Reference Lambert, Perez-Tur and Dupire1996) | 16 | 1 (6.2) | 13 (81.3) | 2 (12.5) | 54 | 7 (13.0) | 40 (74.0) | 7 (13.0) |
| Schupf et al (Reference Schupf, Kapell and Lee1996)2 | 26 | 0 | 18 (69.2) | 8 (30.8) | 138 | 12 (8.7) | 109 (79.0) | 17 (12.3) |
| Prasher et al (Reference Prasher, Chowdhury and Rowe1997) | 34 | 4 (11.8) | 28 (82.3) | 2 (5.9) | 166 | 11 (6.6) | 135 (81.3) | 20 (12.1) |
| Tyrell et al (Reference Tyrell, Cosgrave and Hawi1998) | 62 | 0 | 51 (82.3) | 11 (17.7) | 120 | 10 (8.3) | 97 (80.8) | 13 (10.9) |
| Sekijima et al (Reference Sekijima, Ideda and Tokuda1998)3 | 32 | 2 (6.3) | 24 (74.9) | 6 (18.8) | 174 | 4 (2.3) | 158 (90.8) | 12 (6.9) |
| Current study | 48 | 0 | 40 (83.3) | 8 (16.7) | 66 | 3 (4.5) | 57 (86.4) | 6 (9.1) |
| Total group | 316 | 12 (3.8) | 248 (78.5) | 56 (17.7) | 814 | 63 (7.7) | 670 (82.3) | 81 (10) |
DISCUSSION
The current study
We observed a higher frequency of the APOE ϵ4 allele among subjects with dementia and Down's syndrome compared with those without dementia, although this trend was not statistically significant. Similarly a slightly lower frequency of the protective ϵ2 alleles was observed among the adults with dementia and Down's syndrome. The small cohort size of the current study reduced the statistical power for this study (24% power at 5% level) to detect an effect size of that reported previously. However, in this study the age-matching of two groups of patients with Down's syndrome one of which had dementia, was done in such a way so as to increase its sensitivity and thus lessen the influence of this statistical weakness.
Meta-analysis
To increase the possibility of detecting small effect sizes, we collated data from all known published studies of association between APOE status in Down's syndrome and Alzheimer's disease. This was the method employed recently by Prasher et al (Reference Prasher, Chowdhury and Rowe1997) that showed no evidence to support an involvement of APOE. However, the meta-analysis reported here, showed a statistically significant excess of the APOE ϵ 4 allele among the subjects with dementia when compared with the non-demented group of adults with Down's syndrome. However, we did not observe a significant reduction of APOE ϵ2 alleles among the adults with dementia and Down's syndrome.
Certain factors may have influenced the outcome of the meta-analysis in the current study as well as in Prasher et al's (Reference Prasher, Chowdhury and Rowe1997) study. It is likely that the age difference in the cohorts used in the previous studies (Table 4) is a factor that may have introduced errors into the results of meta-analysis. The age range of subjects included in these studies varied, in that some used age 18 whereas others used age 35, 40 and 50 respectively as the minimum age for inclusion in the study. If APOE is responsible for earlier age of onset of Alzheimer's disease, this differential rate of age range among the different cohorts will make the interpretation of meta-analysis difficult. The lack of appropriately matched control group is another likely source of bias. Only the current study, Tyrell et al's (Reference Tyrell, Cosgrave and Hawi1998) study and van Gool et al's (Reference van Gool, Evenhuis and van Duijin1995) study matched the dementia group with the non-dementia group. Prasher et al's (Reference Prasher, Chowdhury and Rowe1997) study while not originally matched demonstrated no statistically significant difference between the two groups in the age and the gender distribution. The cohort size is also a likely source of error. Apart from the current study, only two other studies (see Table 4) included 20 or more adults with dementia in their cohort. Another source of error is the use of different diagnostic criteria for defining dementia in different studies. Some used formal tools like the DMR scale (Reference EvenhuisEvenhuis, 1992) and Adaptive Behaviour Scale (Reference Nihira, Foster and ShellhaasNihira et al, 1974), whereas others made their diagnosis on the basis of clinical findings alone.
Age of onset of Alzheimer's disease
In the current study, a regression analysis unequivocally showed an influence of ageing on the occurrence of Alzheimer's disease in adults with Down's syndrome. Age is a well-known risk factor for the development of Alzheimer's disease in both the general population and in the adults with Down's syndrome. This effect seems more pronounced among the adults with Down's syndrome. Some have suggested that age of death among adults with Down's syndrome is influenced by the APOE status. In Hardy et al's (Reference Hardy, Crook and Perry1994) autopsy study, the Down's syndrome cases showing Alzheimer's disease neuropathology and who had an APOE ϵ4 (n=10) allele, tended to die at a younger age (age of death ranged between 48 and 60 years, mean 54, s.d.=6) than those who had an ϵ2 allele (n=2) (age of death 69 and 76 years respectively). A similar trend was also reported by both Royston et al (Reference Royston, Mann and Pickering-Brown1994) and Mann et al (Reference Mann, Pickering-Brown and Siddons1995) based on their small cohort studies. Some suggested that APOE status also influences the age of onset of Alzheimer's disease in the general population (Reference Corder, Saunders and StrittmaterCorder et al, 1993). However, the findings of the current study do not suggest such trend among cases of Down's syndrome. Considering the small cohort size included for the analysis of age of onset data in the current study, a Type II error is likely to influence the out-come. As early symptoms of Alzheimer's disease could be difficult to detect in adults with Down's syndrome, it is difficult to be precise about the exact age of onset of Alzheimer's disease in this cohort.
The frequency of APOE ϵ4 allele distribution among the adults with dementia and Down's syndrome in different studies mentioned in Table 4 varied between 12.5% and 33.4%, apart from Prasher et al's (Reference Prasher, Chowdhury and Rowe1997) study, which showed a much lower (5.9%) frequency than that expected in the general population. In contrast, the frequency of APOE ϵ2 alleles varied widely between 0% in the current study, and that of Schupf et al's (Reference Schupf, Kapell and Lee1996) and Tyrell et al's (Reference Tyrell, Cosgrave and Hawi1998), to the 11.8% observed in Prasher et al's (Reference Prasher, Chowdhury and Rowe1997) study, which is higher than expected even for the general population. The frequency of ϵ2 allele among the non-demented subjects with Down's syndrome was on average much higher than that expected in the general population, with a wide range between 2.3% in Sekijima et al's (Reference Sekijima, Ideda and Tokuda1998) study and 50% in Royston et al's (Reference Royston, Mann and Pickering-Brown1994) study. Overall these data are indicative of APOE ϵ4 being a risk factor for the manifestation of Alzheimer's disease in adults with Down's syndrome.
APOE ϵ4 and Alzheimer's disease neuropathology in Down's syndrome
The findings of autopsy studies of patients with Down's syndrome, however, show a somewhat unclear relationship between APOE ϵ4 and Alzheimer's disease neuropathology in this population. For example, in Wisniewski et al's (Reference Wisniewski, Morelli and Wegiel1995) autopsy study of brains of 40 subjects with Down's syndrome (of whom 15 showed Alzheimer's disease neuropathology), only one 21-year-old subject had an APOE ϵ4 allele (ϵ3ϵ4 genotype) and his brain did not show the Alzheimer's disease neuropathology. In Mann et al's (Reference Mann, Pickering-Brown and Siddons1995) study of 20 brains of subjects with Down's syndrome, all of whom showed Alzheimer's disease neuropathology, only 35% (n=7) had APOE ϵ4 allele (one with ϵ2ϵ4 and six with ϵ3ϵ4). Similarly, in Hardy et al's (Reference Hardy, Crook and Perry1994) series of 22 autopsy brains of subjects with Down's syndrome, all of whom showed Alzheimer's disease neuropathology, 45% (n=10) had APOE ϵ 4 (all with ϵ3ϵ4 allele). Further research is needed to clarify this apparent discrepancy perhaps by using immunoneuropathological techniques. It is also important to explore the exact mechanism by which APOE ϵ4 may influence the clinical manifestation of Alzheimer's disease in Down's syndrome subjects. However, it is well known that despite the increased risk associated with the ϵ4 allele, the presence of ϵ4 is neither necessary nor sufficient for the development of Alzheimer's disease (Reference Blacker and TanziBlacker & Tanzi, 1998).
Clinical Implications and Limitations
CLINICAL IMPLICATIONS
-
▪ APOE ϵ4 allele influences the manifestation of Alzheimer's disease in adults with Down's syndrome.
-
▪ The role of APOE ϵ2 allele in relation to Alzheimer's disease in adults with Down's syndrome is not clear.
-
▪ The influence of APOE ϵ4 allele on the age of onset of Alzheimer's disease in adults with Down's syndrome is not clear.
LIMITATIONS
-
▪ The studies included in the meta-analysis used cohorts of different age groups.
-
▪ The adults with Down's syndrome without Alzheimer's disease were not always matched in these studies.
-
▪ Criteria used for diagnosing Alzheimer's disease varied in different studies.
ACKNOWLEDGEMENTS
We thank all the subjects and their carers for taking part in the study, Dr Frank Dunstan for his advice on statistical analysis and Mrs J. Wheeler for typing the manuscript.
eLetters
No eLetters have been published for this article.