Polygenic risk scores for Alzheimer's disease are related to dementia risk in APOE ɛ4 negatives

Abstract Introduction Studies examining the effect of polygenic risk scores (PRS) for Alzheimer's disease (AD) and apolipoprotein E (APOE) genotype on incident dementia in very old individuals are lacking. Methods A population‐based sample of 2052 individuals ages 70 to 111, from Sweden, was followed in relation to dementia. AD‐PRSs including 39, 57, 1333, and 13,942 single nucleotide polymorphisms (SNPs) were used. Results AD‐PRSs (including 39 or 57 SNPs) were associated with dementia (57‐SNPs AD‐PRS: hazard ratio 1.09, confidence interval 1.01–1.19, P = .03), particularly in APOE ɛ4 non‐carriers (57‐SNPs AD‐PRS: 1.15, 1.05–1.27, P = 4 × 10–3, 39‐SNPs AD‐PRS: 1.22, 1.10–1.35, P = 2 × 10–4). No association was found with the other AD‐PRSs. Further, APOE ɛ4 was associated with increased risk of dementia (1.60, 1.35–1.92, P = 1 × 10–7). In those aged ≥95 years, the results were similar for the AD‐PRSs, while APOE ɛ4 only predicted dementia in the low‐risk tertile of AD‐PRSs. Discussion These results provide information to identify individuals at increased risk of dementia.


BACKGROUND
In addition to the apolipoprotein E (APOE) gene, Alzheimer's disease (AD)-associated single nucleotide polymorphisms (SNPs) have been identified through genome-wide association studies (GWASs). 1-3 These variants are usually combined into polygenic risk scores (PRSs), representing the accumulated effect of variants associated with AD. 4 PRSs for AD (AD-PRS) have been associated with AD and dementia in clinical [5][6][7][8][9] and population-based samples. [10][11][12][13] Thus far, population-based studies mainly included individuals around age 60 years and the number of participants above age 95 years was relatively low.
Previous studies examining the effect of APOE genotype 14 15 The Rotterdam study reported an effect of AD-PRS (including 23 SNPs) on dementia, with the strongest effects in APOE ɛ4 carriers. 10 However, none of these studies investigated the effect of AD-PRSs and the possible interaction with APOE genotypes on dementia risk in very old individuals.
We examined the effect of AD-PRSs and APOE genotype on incident dementia, in a large population-based study of individuals aged 70 to 111 years. We studied if the AD-PRSs, the APOE genotype, and the interaction of these, predicted risk of dementia by analyzing the full age spectrum, as well as subgroups aged 70 to 94 years and 95 years or older.

Study population
We used a population-based sample of 3612 participants (1257 men, 2355 women), aged 70 to 111 years, with genotyped data from the Gothenburg H70 Birth Cohort Studies (including the H70, H75, H85, and H95+ studies, and the Prospective Population Study of Women).
The participants were all residents of Gothenburg, Sweden, and were born 1901-1911, 1914, 1918, 1922-1924, 1930, or 1944. They were systematically selected from the Swedish Population Registry based on specific birth dates to yield representative samples at the ages studied. [19][20][21][22][23] All participants were examined at least once between 2000 and 2016. Age at first examination ranged from 70 to 100 years.
In total, 3467 of 3612 had genotyped data after performing quality control (QC), of which 3449 (1196 men and 2253 women) had data on dementia status (see assessment procedures, sections 2.6 and 2.7).
After exclusion of 266 persons with dementia at baseline, 3183 participants were eligible for the present follow-up study. Of these, 1118 were excluded due to having cross-sectional information only and 13 died within a year of baseline, leaving 2052 participants for the analytic sample ( Figure 1).
The total sample was stratified based on age at blood sampling (70-94 years and ≥95 years). In those aged 70 to 94 years, 1717 were followed in relation to incident dementia and 387 developed dementia.
Among those aged ≥95 years, 335 were followed in relation to incident dementia and 219 developed dementia ( Figure 1).

Statistical analysis
All analyses were done in R ( Mini-Mental State Examination (MMSE) score was obtained through semi-structured interviews at age of blood sampling.
Cox regression models using age as time scale were used to analyze the effect of the AD-PRSs and APOE genotype on incident dementia, presented as HR and 95% CI in a model including the following covariates: age at blood sampling, birth year, sex, and 10 PCs to correct for population stratification. Participants were censored at the date of (1) dementia diagnosis, (2) death, or (3)  We also examined the interaction between APOE ɛ4 status and the two AD-PRSs in relation to mortality.

DISCUSSION
In a population-based sample of individuals aged 70 to 111 years, AD-PRSs (including 39 or 57 SNPs) were associated with incident dementia, particularly in APOE ɛ4 non-carriers and in those aged 95 years or older. However, no association was found between the wider AD-PRSs (including >1000 SNPs, P values ≥ 1e -3 ) and incident dementia. APOE non-carriers. 10 We found an association between AD-PRSs (including 39 and 57 SNPs) and dementia among APOE ɛ4 non-carriers. One reason for the discrepancy could be that our study had a rather high mean age at baseline (80 years), whereas the Rotterdam Study had a considerably lower age at inclusion (mean 67.5 years). 10 APOE ɛ4 carriership is related to earlier age of dementia onset 29 as well as premature death. 30 In the current study, the inclusion of older participants renders a selection of healthier APOE ɛ4 carriers who may carry additional genetic variants preventing them from developing dementia at the ages observed. Further, our AD-PRSs included 39 SNPs and 57 SNPs, while the population-based Rotterdam Study, reporting a higher effect among APOE ɛ4 carriers, used an AD-PRS including 23 SNPs. 10 However, analysis using this 23 SNPs AD-PRS in our sample showed similar results as for the 39 SNPs AD-PRS (data not shown).
We found no effect of the wider AD-PRSs (including more than In conclusion, we found an effect of AD-PRSs (including 39 SNPs or 57 SNPs) and APOE genotype on dementia risk in the general population up to very old ages, especially among APOE ɛ4 non-carriers. Results from this type of study could provide additional information to identify individuals at increased risk of dementia for implementation of potential preventative strategies before dementia pathology starts to accumulate.

ACKNOWLEDGMENTS
The authors thank the participants in the Prospective Population