Absence of association between Alzheimer disease and the −491 regulatory region polymorphism of APOE

Neuroscience Letters 250 (1998) 189–192

Absence of association between Alzheimer disease and the −491 regulatory region polymorphism of APOE Y.-Q. Song a,b, E. Rogaeva a,b, S. Premkumar c, N. Brindle a,b, T. Kawarai a,b, An. Orlacchio a,b, G. Yu a,b, G. Levesque a,b, M. Nishimura a,b, M. Ikeda a,b, Y. Pei a,b, C. O’Toole a,b, R. Duara d, W. Barker d, S. Sorbi e, M. Freedman f, L. Farrer c, P. St. George-Hyslop a,b,* a

Centre for Research in Neurodegenerative Diseases, Tanz Neuroscience Building, University of Toronto, 6 Queen’s Park Crescent West, Toronto, Ontario, M5H 3H2, Canada b Department of Medicine, The Toronto Hospital, 399 Bathurst Street, Toronto, Ontario M5T 2S8, Canada c Genetics Program, Boston University School of Medicine, 80 East Concord Street, Boston, MA 02118-2394, USA d Department of Neurology, Wien Centre, Mount Sinai Medical Centre, 4300 Alto Road, Miami Beach, FL 33140, USA e Department of Neurology and Psychiatry, University of Florence, viale Morgani 85, 50134 Firenze, Italy f Department of Behavioural Neurology, Baycrest Hospital, 3560 Bathurst Street, Toronto, Ontario, M6A 2E1, Canada Received 6 April 1998; received in revised form 2 June 1998; accepted 2 June 1998

Abstract A novel polymorphism (−491 A/T) within the regulatory region on the apolipoprotein E gene has recently been reported to be associated with risk for Alzheimer disease (AD). To test this association in an independent data set, we have examined this polymorphism in a sample of 88 well-characterized AD cases and compared the allele frequency and genotype frequencies for this polymorphism with those observed in 112 cognitively normal subjects drawn from the same ethnic group. These results suggest that in the current data set at least, the −491 A/T polymorphism is not associated with risk for AD, but may be in partial linkage disequilibrium with the APOE e2/e3/e4 polymorphism.  1998 Elsevier Science Ireland Ltd. All rights reserved

Keywords: Alzheimer disease; − 491 A/T polymorphism; APOE e2/e3/e4 polymorphism

The e4 allele (Cys110Arg) polymorphism of the apolipoprotein E (APOE) gene has been associated with increased risk for both sporadic and familial forms of Alzheimer disease (AD) in multiple different ethnic groups and in multiple different data sets [3,7]. However, despite the robustness of the genetic association between e4 and AD, the biological mechanism by which the e4 allele might promote increased susceptibility to AD is unclear. In the absence of a robust biological explanation for the association, there has been a lingering concern that the e4 allele may be a genetic marker in linkage disequilibrium with a second nearby sequence change which is the actual biological effector. Such a putative biological effector could be a sequence change either * Corresponding author. Fax: +1 416 9781878; e-mail: [email protected]

within other domains of APOE, or in another nearby gene. Homozygosity for another common APOE sequence variant (A/T polymorphism at position −491 in the regulatory region) has now been reported to be associated with increased risk for AD [2]. This association between AD and the −491 A/A genotype was highly significant (P = 0.00003) in a group of Spanish AD cases and controls, but was much less significant (P = 0.012) in a group of American subjects [2]. Although the effect of the −491 A/ T polymorphism on APOE expression in vivo is unknown, a luciferase reporter assay in transfected human hepatoma cells suggests that the −491 A allele causes a higher level of APOE promoter activity than does the −491 T allele [2]. We have examined an independent cohort of ‘sporadic’ AD cases and controls to assess the contribution of the −491 A/T polymorphism to the risk for AD.

0304-3940/98/$19.00  1998 Elsevier Science Ireland Ltd. All rights reserved PII S0304- 3940(98) 00470- 4

190

Y.-Q. Song et al. / Neuroscience Letters 250 (1998) 189–192

Inspection of the −491 (A/T) allele and genotype frequencies in the total sample revealed that all three genotypes have equivalent frequencies among cases and controls, and that the frequencies of the individual alleles also do not differ between cases and controls (Table 1). Allele and genotype frequencies were nearly identical among cases and controls in the portion of subjects lacking the APOE e4 allele. Among e4 carriers, the −491 A allele, and the A/A genotype in particular, were slightly, but not significantly more frequent in controls than AD cases. It is noteworthy that the allele and genotype frequencies of our North American controls are intermediate between those reported for the controls from USA and for controls from Spain [2]. This observation suggests that the frequency of the −491 A/T polymorphism may vary considerably within different ethnic groups. On the other hand, the frequency of the A/A genotype in our AD cases, which was similar to the frequency in our controls, is less than the corresponding frequencies in the Spanish and USA cases by 13% and 16%, respectively. An explanation for the differences between AD cases from different data sets is unclear. Logistic regression analysis was carried out to assess the influence of −491 (A/T) genotype on risk of AD adjusting for age, sex and APOE genotype [1]. Following the scheme of Bullido et al. [2], the A/T and T/T genotypes were combined into a single referent group. For the APOE polymorphism, subjects were collated into two groups defined by the presence or absence of the e4 allele. Models were evaluated using the logistic procedure in SAS [6]. This analysis failed to reveal any significant association between the A/A genotype and risk of AD (Table 2), although there was a strong effect of the e4 allele on AD risk (OR = 4.6, 95% CL = 2.3–9.3). These conclusions were unchanged even when subgroups stratified by e4 status. Cumulatively, our results confirm the association between AD and the e4 allele of APOE but do not support an association between AD and the −491 A/T polymorphism. The failure to detect an association with the −491 A/T polymorphism cannot be ascribed to lack of statistical power because our sample of 88 AD cases and 112 controls has the power to detect an odds ratio of at least 2.7 assuming

Eighty-eight Caucasian patients with clinically diagnosed AD established by NINCDS/ADRDA criteria [4,5] were collected through Memory Disorders Clinics at the University of Miami and at the University of Toronto. One-hundred and twelve cognitively normal (control) Caucasian subjects were recruited from the same communities either through participation in normative studies of aging or as spousal controls for the AD patients. The control subjects were assessed to be normal by a structured interview including neurological mental status examination, category fluency test, Folstein Mini-Mental Status Examination (MMSE) and two tests of delayed recall (for spousal controls), or by a battery of neuropsychological tests of memory, language and semantic function, visuospatial and constructional praxis, attention and mood, and, in some cases, regional cerebral blood flow single photon-positron emission computed tomography (rCBF-SPECT) (for subjects enrolled in normative studies of aging). The mean age at onset of the AD patients was 76.1 ± 7.8 years and the mean age at examination of the controls was 69.6 ± 12.5 years. DNA was extracted from buffy coat leukocytes and the APOE e2, e3, e4 genotype was determined using previously described methods [7]. The genotype at the −491 A/T locus was ascertained using the method in the original report describing this polymorphism [2]. Thus, 100 ng of genomic DNA were amplified in a reaction volume of 10 ml containing 10 pmol of each primer, 250 mM dNTPs, 1 mM MgCl2, 5% dimethyl sulfoxide and 0.5 U Taq polymerase. For the first round of PCR (forward primer 5′-CAAGGTCACACAGCTGGCAAC-3′; reverse primer 5′-TCCAATCGACGGGTAGCTACC-3′), 35 cycles were performed with denaturation at 94°C × 30 s, annealing at 65°C × 30 s and extension at 72°C × 30 s, then a final extension at 72°C for 7 min. These PCR products were then used to perform nested PCR (forward primer 5′-TGTTGGCCAGGCTGGTTTTAA-3′; reverse primer 5′-CTTCCTTTCCTGACCCTGTCC-3′) for 35 cycles of denaturation at 94°C for 30 s, annealing at 55°C for 30 s and extension for 30 s. The nested PCR products were digested with 10 U of DraI overnight and resolved on 8% non-denaturating polyacrylamide gels and exposed to autoradiographic film. Table 1 APOE − 491 A/T allele and genotype frequencies Allele frequency (%) Group Alzheimer disease Controls Alzheimer disease Controls Alzheimer disease Controls

N Overall 88 112 APOE e4(+) subjects 52 36 APOE e4(−) subjects 36 74

Genotype frequency (%) A

T

A/A

A/T

T/T

81.2 80.4

18.8 19.6

67.0 65.2

28.4 30.3

4.6 4.5

82.7 90.3

17.3 9.7

69.2 80.6

26.9 19.4

3.9 0

79.2 76.4

20.8 23.6

63.9 59.4

30.6 33.8

5.5 6.8

191

Y.-Q. Song et al. / Neuroscience Letters 250 (1998) 189–192 Table 2 Odds ratios for AD by APOE − 491 A/T genotype and e4 statusa All subjects Genotype A/A A/T,T/T a

b

Odds ratio 0.72 1.00

e4(+) subjects c

95% CI (0.35–1.48) Referent

e4(−) subjects

Odds ratio 0.29 1.00

95% CI (0.07–1.12) Referent

Odds ratio 1.10 1.00

95% CI (0.4–2.7) Referent

All odds ratios are adjusted for age at examination and sex. bAdjusted for APOE genotype. cCI, Confidence interval.

a significance level (a) of 0.05, power (1 − b) of 0.80, and an exposure frequency of 0.65 in controls. By comparison, the odds ratio for the A/A genotype in the Spanish sample studied by Bullido et al. was 3.8 [2]. Three potential explanations can be envisaged for the discrepancy between the current result and the previous report of an association between the −491 A/T polymorphism and AD. The first explanation is that the association is true, but only in certain ethnic groups. This explanation is intuitively unlikely since neither this, nor the previous study examined unique population isolates in which a rare genetic trait might be enriched. The alternate explanation is that the ethnic and genetic origins of the AD and control groups are not equivalent in one or more of the three data sets. Allelic association studies are notoriously unreliable under such circumstances, yielding either false positive or false negative associations. It is conceivable therefore that our failure to detect an association may reflect obscuration of a true association due to sampling of a mixed population in which the −491 A allele might be associated with AD in one sub-population, while the −491 T allele might be associated with AD in the other sub-population. Although we cannot definitively exclude this possibility, we would note that if this explanation were correct, it would still negate a biological role for the −491 A/T polymorphism per se because it would imply that the −491 A/T polymorphism would have different effects in different populations. In view of the fact that only one of the previous data sets showed truly significant results (i.e. P , 0.01), an equally plausible explanation is that the allelic association in the previous study is spurious.

The third explanation for the discrepancy between this and the previous report is that, in our data set (and possibly also in that of Bullido et al. [2]), there is evidence for disequilibrium between the APOE −491 A/A genotype and heterozygosity for APOE e4. Specifically, in our control group, 80.6% of e4 carriers but only 59.4% of non-e4 carriers are A/A (P = 0.058). Similarly, in the Spanish control group of Bullido et al. [2], 64.7% of e4 carriers are A/A while only 53.0% of non-e4 carriers are A/A. In their USA controls, 81.2% of e4 carriers are A/A and 66.9% of non-e4 carriers are AA. Haplotype analysis of the data from the 88 AD cases and 112 controls in our study revealed that the two APOE polymorphisms are not in Hardy–Weinberg equilibrium in AD cases (x2 = 10.4; d.f. = 5; P , 0.07) or controls (x2 = 11.9; d.f. = 5; P , 0.04). Table 3 shows that most of the disequilibrium is attributable to under-representation of the T/e4 haplotype and over-representation of the A/e3 and A/e4 haplotypes. Such disequilibrium, if present in the populations studied by Bullido et al. [2], may also have led to a spurious association between the −491 A allele and AD. This work was supported by grants from the Medical Research Council of Canada, The Canadian Genetic Diseases Network, The Alzheimer Association of Ontario, The Howard Hughes Medical Research Foundation, the EJLB Foundation, the National Institutes of Health (AG09029), the Peter Burgess Fellowship (E.R.), the Alzheimer Society of Canada Fellowship (G.L.), the Helen B. Hunter Fellowship (G.Y.), the National Institutes of Health (T32-AG00115) (S.P).

Table 3 Distribution of APOE haplotypes in AD cases and controls AD cases A/e2 Observed Expected

Observed Expected

A/e3

A/e4

T/e2

T/e3

T/e4

Unknown

3 85 3.6 76.3 Normal controls

41 40.2

2 0.8

15 17.7

2 9.3

28*

A/e2

A/e3

A/e4

T/e2

T/e3

T/M4

Unknown

7 11.3

129 120.2

30 26.0

0 2.8

30 29.3

0 6.3

24**

*Includes 14 A.T/e3.e4 subjects; **includes 5 A.T/e2.e3, 1 A.T/e2.e4 and 6 A.T/e 3.e4 subjects.

192

Y.-Q. Song et al. / Neuroscience Letters 250 (1998) 189–192

[1] Breslow, N.E. and Day, N.E., The analysis of case control studies: statistical methods in cancer research, Vol. 1, IARC Scientific Publications 32, Lyon, 1980. [2] Bullido, M.J., Artiga, M.J., Recuero, M., Sastre, I., Garcia, M.A., Aldulo, J., Lendon, C., Han, S.W., Morris, J.C., Frank, A., Vasquez, J., Goate, A. and Valdivieso, F., A polymorphism in the regulatory region of APOE associated with risk for Alzheimer’s Disease, Nat. Genet., 18 (1998) 69–71. [3] Farrer, L.A., Cupples, L.A., Haines, J.L., Hyman, B., Kukull, W.A., Mayeux, R., Myers, R.H., Pericak-Vance, M.A., Risch, N. and van Duijn, C.M., Effects of age, sex, and ethnicity on the association between apolipoprotein E genotype and Alzheimer’s Disease: a meta-analysis (APOE and Alzheimer’s Disease Meta Analysis consortium), J. Am. Med. Assoc., 278 (1997) 1349–1356.

[4] Khachaturian, Z.S., Diagnosis of Alzheimer Disease, Arch. Neurol., 42 (1985) 1097–1105. [5] McKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D. and Stadlan, E.M., Clinical diagnosis of Alzheimer Disease: Report of the NINCDS-ADRDA Work Group under the auspices of the Dept. of Health and Human Services Task force on Alzheimer Disease, Neurology, 34 (1984) 939–944. [6] SAS User’s guide: Statistics. SAS Institute, Cary, NC, 1992. [7] Saunders, A., Strittmatter, W.J., Schmechel, S., St. GeorgeHyslop, P., Pericak-Vance, M., Joo, S.H., Rosi, B.L., Gusella, J.F., Crapper-McLachlan, D., Growden, J., Alberts, M.J., Hulette, C., Crain, B., Goldgaber, D. and Roses, A.D., Association of apoliprotein E allele e4 with the late-onset familial and sporadic Alzheimer Disease, Neurology, 43 (1993) 1467– 1472.