Association of Plasma Protein Risk Score and Risk of Cognitive Impairment in the Framingham Heart Study

Abstract Background Plasma‐based biomarkers are emerging as a new diagnostic tool for Alzheimer’s disease (AD) with high accuracy and low cost. Whether aggregated plasma protein risk score (PPRS) derived from a single sample could be useful for diagnosis of incident mild cognitive impairment (MCI) and AD is uncertain. This study aims to understand the association between PPRS and future risk of AD related outcomes. Method We included Framingham Heart Study (FHS) Offspring cohort participants with plasma aptamer based SOMAscan proteomics assay data at exam 5. After excluding subjects who were have missing data for AD risk factors (age, sex, education and APOE ɛ4) and had other‐types of dementia, 1523 participants (cognitive normal (CN)=1258, MCI=137, AD=128) were considered for the primary analyses. The Cox model with LASSO penalty was used to build the PPRs for AD and MCI. The association of PPRS for AD risk was tested for outcomes of MCI/AD incidence (mean follow‐up=23.06 years), brain volume, and blood AD biomarkers. Result We identified 279 proteins for inclusion in the MCI PPRS and 103 proteins for inclusion in the AD PPRS. The MCI PPRS had a hazard ratio (HR) of 1.79 [4.99, 7.37] and the discriminating power (C‐index) was 0.86, and after including risk factors C‐index was 0.88. The AD PPRS had a HR of 1.79 [4.86,7.33] and the C‐index was 0.89 and after including the risk factors the power of the model was improved (C‐index=0.92). The chance of developing AD for a CN participant at age 80 years with in 20 years with high AD PPRS was 85% and with low AD PPRS was 15%. Brain volume distribution using 8 variables was significantly decreasing with increasing level of MCI PPRS (p‐value<0.001 for 5 variables) and AD PPRS (p‐value<0.001 for all 8 variables). The increased levels of MCI PPRS (p‐value<0.001) and AD PPRS (p‐value<0.001) were also significantly associated with increased plasma p‐tau181. Conclusion This study suggests that aggregated protein risk scores from plasma samples can be used to predict future risk of MCI and AD, thus serving as a potential screening tool for early intervention to prevent AD or slow its progression.