Cognitively impaired patients whose hippocampal volume falls in the lowest quartile are disproportionately likely to carry limbic-predominant age-related TDP-43 encephalopathy (LATE) alongside Alzheimer's disease, according to a retrospective imaging study led by Nidhi S. Mundada, Sandhitsu R. Das and David A. Wolk at the University of Pennsylvania, published in Alzheimer's & Dementia on 13 December 2025 and posted as a preprint on 14 August 2025. In the study's autopsy validation sample, drawn from Alzheimer's Disease Neuroimaging Initiative (ADNI) participants with intermediate or high Alzheimer's neuropathologic change, 13 of 15 people in the bottom hippocampal quartile had autopsy-confirmed LATE neuropathologic change, against 13 of 39 in the top two quartiles (Fisher's exact test, odds ratio 13.8, p less than 0.001). The average interval between the MRI scan and death was 7.2 years, with a standard deviation of 3.5 years.
The main analysis was observational and retrospective. It covered cognitively impaired ADNI participants who had a T1 structural MRI plus amyloid and tau PET: 115 were classified as amyloid-positive with a hippocampal volume in the top two quartiles (above 2861.4 cubic millimetres after adjustment for age and head size), 58 as amyloid-positive in the bottom quartile (below 2502.2 cubic millimetres), and 20 as amyloid-negative in the bottom quartile. Volumes were taken from the smaller of the two hippocampi, because LATE changes are often one-sided. Of 386 participants in the thickness analysis, which included matched amyloid-negative unimpaired controls, 304 had repeat MRI, an average of 3.9 scans each over follow-up spans of 0.5 to 9.8 years. The bottom-quartile amyloid-positive group showed more anterior hippocampal and amygdala atrophy than the amyloid-positive comparison group, worse memory and language scores, and faster cognitive decline.
The interest in a crude structural marker comes from a gap: there is no validated biomarker for TDP-43 in living people, and LATE co-pathology is common in older patients who also meet criteria for Alzheimer's disease. Clinical criteria for LATE published by Wolk and colleagues in January 2025 already point to the bottom quartile of hippocampal volume as a supporting feature. The counterpoint sits inside this paper. Patients in the bottom quartile also carried more tau, measured by temporal and global Tau-MaX, which the authors say indicates greater Alzheimer's progression in that group, so a small hippocampus is not by itself specific to LATE; the authors suggest that comparing hippocampal atrophy against medial temporal tau on PET might improve specificity, and note that a limbic-predominant form of Alzheimer's disease can look similar.
The authors frame the cut-off as a way to select or exclude probable mixed-pathology cases in trials and in clinics, which bears on who is enrolled in anti-amyloid studies and how their outcomes are read. No treatment was tested in this work, and it does not show whether LATE co-pathology changes the effect of any therapy. The authors list their own limits: a modest autopsy cohort, a cut-off derived from single-timepoint scans, and a highly selected research cohort in ADNI whose findings may not carry over to community populations. Follow-up analyses of structural and cognitive trajectories using the same quartile scheme have so far appeared only as conference abstracts, most recently in January 2026.