Identifying patients with mild cognitive impairment at high risk of transitioning to Alzheimer's disease using routinely collected data

Abstract

Background

Mild cognitive impairment (MCI) is a heterogeneous condition with variable progression to Alzheimer's disease (AD). Identifying MCI individuals at high risk for progression typically requires cerebrospinal fluid (CSF) biomarkers, magnetic resonance imaging (MRI), which are costly and invasive.

Objective

This study aimed to develop a cost-effective approach using routinely collected clinical data to identify a subgroup of MCI individuals at high risk for AD progression.

Methods

Analyses were conducted using the UK Biobank dataset, focusing on 1019 participants identified as having MCI, using the ICD-10 code F06.7 (mild neurocognitive disorder due to known physiological condition) in the absence of a dedicated diagnostic code for MCI. Participants (mean age = 71.7 years; 44% women) were characterized using routinely recorded demographic, comorbidity, and lifestyle data. A mixed-data clustering model was applied to classify individuals into subgroups. Clinical relevance of each cluster was evaluated using Kaplan-Meier survival analysis of MCI-to-AD progression over an average follow-up of 4.5 years.

Results

Three subtypes were identified with distinct progression risks: high-risk (HR), medium-risk (MR), and low-risk (LR). The HR subtype had significantly higher prevalence of hypertension (98%), cardiovascular disease (89%), diabetes (48%), and high cholesterol (67%) than MR and LR (p < 0.05). The HR group was younger on average but had greater comorbidity burden and higher likelihood of AD progression.

Conclusions

This study demonstrates the feasibility of using routinely collected data to identify high-risk MCI individuals. This approach offers a practical preliminary screening tool to prioritize individuals for targeted interventions and further specialized assessments.