Uncovering Cognitive Subtypes in Essential Tremor: A Data-Driven Clustering Approach in Cognitively Normal Individuals

TL;DR

The study identifies two cognitive subtypes in essential tremor patients, one with memory impairment not seen in healthy controls.

Contribution

A data-driven clustering approach reveals a memory-impaired cognitive subtype in essential tremor with potential for early intervention.

Findings

Healthy controls formed a single high-functioning cognitive cluster.

Essential tremor patients split into two clusters, one with significant memory impairment.

The memory-impaired cluster may represent a vulnerable subgroup for cognitive decline.

Abstract

Different profiles of cognitive functioning have been demonstrated in ET subjects, also in patients with normal cognition. However, the prognostic significance of these profiles remains still debated. In this study, we aimed to explore different cognitive patterns among cognitively normal ET subjects and their relationship with the cognitive profiles of healthy subjects.

We enrolled 50 cognitively normal subjects (26 ET and 24 age-, sex-, and education-matched healthy subjects), which scored within normal ranges individually in all tests of a comprehensive neuropsychological battery covering memory, executive function, attention, visuospatial abilities, and language. Unsupervised clustering was applied separately within each group. Cluster membership was validated by post-hoc comparisons using ANOVA and Bonferroni-corrected pairwise tests to compare the variables among the clusters.

All HC clustered together into a single high-functioning cognitive profile. On the contrary, we found two different clusters within ET, C1 (n = 14), showing high performance across all domains, and C2 (n = 12) which exhibited significantly poorer performances in the RAVLT-IR (p < 0.0001), RAVLT-DR (p = 0.0002), and Digit Span Forward (p = 0.015) than both ET-C1 and HC subjects. Other domains showed no significant differences across ET clusters.

This study demonstrates a cognitive heterogeneity in ET and reveals a memory-impaired subgroup absent among HC. The ET cluster with lower memory performance likely reflects a pattern of vulnerability for longitudinal decline or progression to mild cognitive impairment. The identification of this profile has relevant translational implications for prognosis and identification of early intervention strategies.

A data-driven clustering approach was applied to cognitive variables in HC subjects and ET patients. HC formed a homogeneous cluster. ET were divided into two cognitive subgroups: one cluster with high performance, and one memory-impaired cluster, significantly diverging from both the intact ET subgroup and HC. This may represent a cognitively vulnerable ET subgroup, with strong implications for targeted screening, early neuroprotective interventions and personalized clinical management.

Figures