Background and objectives: Limbic-predominant age-related TDP-43 encephalopathy (LATE) is a neurodegenerative condition often overlapping clinically with Alzheimer disease (AD). No established in vivo biomarkers exist for LATE. The amygdalar atrophy scale (AAS) is a visual MRI rating tool scoring the amygdala as AAS₀ (no atrophy), AAS₁ (mild-to-moderate atrophy), or AAS₂ (severe atrophy) and is associated with TDP-43 neuropathology in limbic regions. This study describes the clinical, neuroimaging, and longitudinal cognitive profiles of individuals classified using a proposed framework that combines the AAS with established AD biomarkers.

Methods: This retrospective study included individuals attending the Geneva Memory Center (2014-2022) who underwent T1-weighted MRI, as well as clinical and neuropsychological assessments. Participants were categorized with an etiologic grouping defined as No-AD/No-LATE (AAS₀ , negative AD biomarkers), AD (AAS₀ , positive AD biomarkers), LATE (AAS_1-2 , negative AD biomarkers), or AD/LATE (AAS_1-2 , positive AD biomarkers). Group differences were compared using Kruskal-Wallis or χ² tests. Differences in MRI brain volumes and cortical thickness between groups were examined with analysis of covariance. Cognitive trajectories using Mini-Mental State Examination (MMSE) scores over 30 months were estimated with linear mixed-effects models accounting for baseline age and education.

Results: The analytic sample included 469 participants (mean age 71.2 ± 8.5 years, 53% female). The No-AD/No-LATE group exhibited the highest MMSE scores (N = 181, 27.9 ± 2.2), outperforming AD (N = 146, 24.2 ± 4.9), LATE (N = 36, 26.2 ± 3.2), and AD/LATE (N = 106, 22.5 ± 5.3) groups (p < 0.001). On the Free and Cued Selective Reminding Test, the No-AD/No-LATE group showed the highest scores (15.1 ± 1.8); the LATE and AD groups performed similarly (13.3 ± 3.5 and 12.5 ± 4.0), both exceeding the AD/LATE group (10.6 ± 4.1) (p < 0.001). LATE and AD/LATE groups showed lower volumes/thicknesses in TDP-43-related regions compared with AD and No-AD/No-LATE groups. Longitudinally, the LATE group maintained cognitive stability as the No-AD/No-LATE group (β = -0.010, 95% CI -0.096 to 0.075, p = 0.813), whereas both AD (β = -0.136, 95% CI -0.184 to -0.088, p < 0.001) and AD/LATE (β = -0.139, 95% CI -0.196 to -0.082, p < 0.001) groups showed faster decline over time.

Discussion: Patients identified as LATE by the integration of AAS with AD biomarkers had generally mild amnestic cognitive impairment, significant limbic atrophy, and slow decline over time, all features consistent with previous autopsy series. Moreover, AD/LATE cases were more impaired than those with LATE or AD alone. Prospective and neuropathologic validation is warranted.