Basic Science and Pathogenesis

Area of Science:

• Neuroscience
• Genetics
• Biostatistics

Background:

• Alzheimer's disease (AD) genetic studies typically exclude individuals with mild cognitive impairment (MCI) due to diagnostic uncertainty.
• This exclusion may limit the power to detect genetic associations relevant to AD development.

Purpose of the Study:

• To develop and validate a method for estimating the probability of AD conversion in MCI individuals.
• To enhance genetic association analyses by incorporating MCI individuals with estimated probabilities of disease into case-control studies.

Main Methods:

• A multi-step approach was proposed: 1. Identify a predictive model for AD using logistic regression on established AD cases and cognitively unimpaired (CU) controls. 2. Estimate AD probability (p_i) for MCI individuals using this model. 3. Incorporate MCI individuals into genetic association tests via thresholding (p_i > t) or probabilistic resampling.
• The APOE-e4 allele association with AD was assessed in 1420 participants (448 AD, 714 CU, 258 MCI) as a proof of principle.

Main Results:

• The benchmark analysis of APOE-e4 in AD cases vs. CU controls yielded an odds ratio (OR) of 3.25 (Z=9.088).
• Including all MCI individuals as cases reduced the OR to 1.97 (Z=6.628).
• A threshold model (t=0.8) incorporating clinical, demographic, and biomarker data (pTau-181, CDR memory score, age, sex) slightly improved the test statistic (OR=3.22, Z=9.174) by adding 24 MCI individuals as probable cases. The resampling approach was less effective than the threshold method.

Conclusions:

• Clinical, demographic, and biomarker data can effectively impute disease probability ('caseness') in MCI individuals for genetic association studies.
• Even a small number of imputed MCI cases significantly improved statistical test performance.
• This imputation strategy holds promise for enhancing genetic discovery in larger Alzheimer's disease datasets.