MARBLE, a fully automated in vivo MRI-based marker of LATE-NC developed using machine learning on brain MRI and neuropathology data from 918 community-based older adults, achieved an area under the curve of 0.92 for in vivo prediction of LATE-NC in an independent test group.
MARBLE achieved high diagnostic accuracy for in vivo prediction of LATE-NC in an independent test group.
Higher MARBLE scores were significantly associated with higher LATE-NC neuropathologic stages.
Higher MARBLE scores were associated with lower levels and accelerated decline of cognition.
MARBLE was developed using machine learning methods applied to brain MRI and neuropathology data from a large community-based cohort.
LATE-NC is common in older adults but currently can only be definitively diagnosed at autopsy.
This research describes the development of a new automated brain imaging tool called MARBLE that can detect a specific type of brain disease called LATE-NC (limbic-predominant age-related TDP-43 encephalopathy neuropathologic change) while a person is still alive. LATE-NC is a condition found commonly in older adults where a protein called TDP-43 builds up abnormally in brain regions involved in memory, but until now it could only be confirmed by examining brain tissue after death. The researchers used MRI brain scans and post-mortem brain tissue data from 918 older community members to train a machine learning algorithm to recognize patterns associated with this disease. When tested in a separate group of older adults who had MRI scans within three years of their death, MARBLE correctly identified LATE-NC with very high accuracy, achieving an area under the curve of 0.92 (where 1.0 would be perfect). The tool also showed that people with higher MARBLE scores had more severe disease stages confirmed at autopsy, and experienced greater memory problems and faster cognitive decline during life — particularly in the ability to recall specific past events (episodic memory), recall general knowledge (semantic memory), and process information quickly (perceptual speed). This research suggests that MARBLE could be a practical tool for doctors and researchers to identify LATE-NC in living patients, which has previously been impossible without autopsy. This matters because LATE-NC can look similar to Alzheimer's disease but may require different treatments. Having a reliable in vivo marker could help ensure that clinical trials for Alzheimer's or dementia therapies enroll the right participants, and could eventually support the development of treatments specifically targeting LATE-NC.
Tazwar M, Evia A, Ridwan A, Leurgans S, Wagner M, Bennett D, et al.. (2026). MARBLE: An in vivo MRI marker of limbic-predominant age-related TDP-43 encephalopathy neuropathologic change (LATE-NC).. Alzheimer's & dementia : the journal of the Alzheimer's Association. https://doi.org/10.1002/alz.71832