Table S1. Antibodies used in immunohistochemical staining experiments in the present study.

Table S2. Results of univariate and multivariate regression analyses to determine predictors of cognitive decline.

Table S3. Results of sensitivity analyses done by excluding certain subjects based on their comorbidty pathologies and genetic features. Results are age and sex adjusted.

Table S4. Results of logistic regression analysis of the MAPT 3'UTR region between the PART and low AD groups using PLINK, adjusted for sex and age.

Table S5. Results of logistic regression analysis of the MAPT 3'UTR region between the PART and high AD groups using PLINK, adjusted for sex and age.

Table S6. Results of Haploview haplotype analysis of the MAPT 3'UTR region between the PART and low AD groups.

Table S7. Results of Haploview haplotype analysis of the MAPT 3'UTR region between the PART and high AD groups.

Table S8. Comparison of common MAPT haplotypes between the PART and low AD groups.

Table S9. Comparison of common MAPT haplotypes between the PART and high AD groups.

Table S10. Frequency of dementia in the study groups when excluding LATE-NC with HS, diffuse neocortical and limbic predominant LRP and small cortical infarcts (n=152 remaining).

Figure S1. Line chart showing how MMSE scores changed in follow-up, using MMSE at baseline in 1991 as the reference. In the background the change in MMSE scores over time are seen for individual study participants, whereas group medians are highlighted.

Figure S2. Standardised AD genetic risk scores in the PART, low AD and high AD groups.

Figure S3. Standardised AD genetic risk scores without APOE in the PART, low AD and high AD groups.