Prior to analysis, we transformed the plasma protein levels and AD‐related endophenotype values by z‐score normalization using the “normalize()” function in the R som package (v0.3‐5.1). We then determined the associations between the normalized protein levels and clinical phenotypes (ie, AD or MCI vs CN), adjusting for age, sex, history of cardiovascular disease (ie, heart disease, hypertension, diabetes mellitus, and hyperlipidemia), and BMI using the following linear regression model:
where β is the weighted coefficient for the corresponding factors and ε is the intercept of the linear equation. Similarly, the associations between normalized protein levels and AD‐related endophenotypes were determined using the following linear regression model:
We considered plasma proteins with a false discovery rate (FDR)‐adjusted p‐value of <0.05 as being significantly associated with AD or AD‐related endophenotypes.
where β is the weighted coefficient for the corresponding factors and ε is the intercept of the linear equation. Similarly, the associations between normalized protein levels and AD‐related endophenotypes were determined using the following linear regression model:
We considered plasma proteins with a false discovery rate (FDR)‐adjusted p‐value of <0.05 as being significantly associated with AD or AD‐related endophenotypes.
