Ariel Hippen and John Kauwe, Biology

Alzheimer’s disease is the leading cause of dementia in the elderly and the third most common cause of death in the United States. A vast number of genes regulate Alzheimer’s disease, including Presenilin 1 (PSEN1). It is possible that novel variants in the PSEN1 gene have an effect on Alzheimer’s disease status. A recent study suggested that one of these variants, PSEN1 E318G, significantly affects Alzheimer’s disease status in a large case-control dataset, particularly in connection with the APOEε4 allele [1].

Our study looks at the same variant, using samples from the Cache County Study on Memory Health and Aging. When the study began in 1994, the 5092 subjects therein represented approximately 90 percent of all residents age 65 and older in Cache County, Utah. Case-control status was determined through a series of clinical dementia evaluations and cognitive assessments, including the Modified Mini-Mental State Exam-Revised [2]. Individuals with dementia only showed symptoms of AD with no other comorbid forms of dementia. Cognitively normal individuals were free of any symptoms of AD and other comorbid forms of dementia at all stages of screening.

We genotyped the PSEN1 E318G locus for 3420 individuals using a custom TaqMan assay. Of those, 478 were clinically ascertained Alzheimer’s disease cases and 2942 were cognitively normal controls. We used Fisher’s exact test to calculate odds ratios for effect of APOEε4 status on AD risk in E318G carriers and E318G non-carriers. To determine if the effect of E318G on AD risk was different for APOEε4 homozygotes versus heterozygotes we calculated the odds ratios for each combination of E318G carrier/non-carrier status and their number of APOEε4 alleles (0, 1 or 2).

We used the odds ratios described above to calculate the synergy factor between E318G status and APOEε4 status. Synergy factors are the ratio of the observed and expected odds ratios for the two interacting SNPs. Assuming there is no synergy between the SNPs, the expected odds ratio then equals the product of the individual odds ratios. The synergy factor is calculated by dividing the observed odds ratio by the expected odds ratio for the interacting SNPs. A synergy factor that deviates from 1 suggests a statistical interaction between the SNPs [3],[4].

We ran a logistic regression model using case/control status as the response, with E318G, age, gender, and APOEε4 positive or negative status as covariates. We also included an interaction between APOE and E318G to test the finding by Benitez et. al. that carriers of both E318G and APOEε4 were more at risk of developing AD than carriers of either mutation alone.

To determine if we had sufficient statistical power to observe an effect of E318G on AD status in connection to APOEε4, we calculated the minimum discernible effect size of E318G in all individuals in the Cache County Study using a power analysis [5]. We used sample size, E318G exposure, and AD status to find the smallest effect size we could detect with 80 percent power. Then, to determine if we had enough power to see an effect of E318G on AD status independent of APOEε4, we calculated the minimum discernible effect size of E318G in APOEε3 homozygotes.

APOEε4 carriers with an E318G allele have slightly higher risk for AD that those without the allele (3.3 vs. 3.8). Comparisons of odds ratios for AD risk based on E318G and number of APOEε4 alleles revealed that E318G carriers with one APOEε4 allele were at a higher risk for AD (OR=3.5, CI=1.0-12.7, p=0.043) compared to APOEε4 heterozygotes who did not carry the E318G allele (OR = 3.0, CI = 2.4 – 3.7, p=0.78).

In our logistic regression model, which considered all individuals in the dataset and included an interaction term between APOEε4 and E318G, we found a positive but non-significant main effect for E318G (p=0.895). The interaction term between E318G and APOEε4 was also non-significant (p=0.689). Our power analysis indicated that, with a sample size of 3420, we had 80 percent power to detect an odds ratio of 1.9 of bigger if one existed (alpha = 0.05). For the second logistic regression model, with a sample size of 1916 APOEε3 heterozygotes, we had 80 percent power to detect an odds ratio of 2.4 or larger (alpha = 0.05).

While our calculations indicate that this study is adequately powered, we failed to detect significant associations with E318G and AD as described by Benitez et al. However, the overall trends of our results are supportive of the results reported by Benitez et al; the E318G allele increases risk for AD in APOEε4 carriers. Differences in ascertainment and population may present challenges for replication of this finding in the Cache County Study. As such, we believe that further analyses using additional samples are necessary to determine the significance of the interaction between the APOEε4 allele and E318G on Alzheimer’s disease risk.

References

1. B. A. Benitez, C. M. Karch, Y. Cai, S. C. Jin, B. Cooper, D. Carrell, et. al. “The PSEN1, p.E318G variant increases the risk of Alzheimer’s disease in APOEε4 carriers.” PLoS Genetics. 2013 Aug; 9(8).
2. J. T. Tschanz, K. A. Welsh-Bohmer, B. L. Plassman, M. C. Norton, B. W. Wyse, J. C. Breitner. “An adaptation of the modified Mini-Mental State Examination: analysis of demographic influences and normative data: The Cache County Study.” Neuropsychiatry Neuropsychol Behav Neurol, 2002 March; 15(1):28-38.
3. M. Cortina-Borja, A. D. Smith, O. Combarros, D. Lehmann. “The synergy factor: A statistic to measure interactions in complex diseases.” BMC Research Notes, 2009 Jun 15.
4. O. Combarros, M. Cortina-Borja, A. D. Smith, D. Lehmann. “Epistasis in sporadic Alzheimer’s disease.” Neurobiology of Aging, 2009 September; 30(9):1333-49.
5. E. Demidenko. “Sample size determination for logistic regression revisited.” Statist. Med, 2007 Aug 15; 26(18):3385-97.