Researchers at the Alzheimer Center Amsterdam have applied the newly published clinical criteria for limbic-predominant age-related TDP-43 encephalopathy (LATE) to a working memory clinic population for the first time, and report that 298 of 3,606 eligible patients, 8.2 percent, met one of the LATE categories. In the retrospective analysis of the Amsterdam Dementia Cohort, published in Neurology on 16 April 2026 and preprinted on medRxiv in October 2025, Colin Groot, Rik Ossenkoppele and colleagues classified 56 individuals (1.6 percent) as Probable LATE, 115 (3.2 percent) as Possible LATE and 127 (3.5 percent) as co-occurring LATE and Alzheimer's disease (LATE-AD). A further 1,675 (46.5 percent) met criteria for Alzheimer's disease without LATE and served as the comparison group. The records covered visits from September 1997 to March 2024 in people older than 50 with mild cognitive impairment or dementia; mean age at baseline was 66 years and 49.2 percent were women.
The criteria being tested were proposed by David Wolk, Peter Nelson and a large consensus panel in Alzheimer's and Dementia in January 2025. They matter because there is still no biomarker with molecular specificity for TDP-43 pathology in living people. Instead the criteria work from a clinical picture, a progressive amnestic syndrome with limbic-predominant atrophy on MRI, combined with amyloid and tau biomarkers used largely to rule Alzheimer's disease in or out. The Amsterdam question was whether a definition built that way, by phenotype and by exclusion, picks out a group that behaves distinctly.
On the study's own measures it did, in opposite directions. Compared with the Alzheimer's group, Probable LATE declined more slowly on the Mini-Mental State Examination (standardised beta 0.12, standard error 0.05, p = 0.02) and on memory, attention, executive function and visuospatial domains, and had lower mortality risk (hazard ratio 0.70, 95 percent confidence interval 0.49 to 0.99, p = 0.04). The LATE-AD group declined faster across cognitive domains and had higher mortality risk (hazard ratio 1.25, 1.01 to 1.53, p = 0.04). On imaging, Probable and Possible LATE showed the expected limbic-predominant signature, a higher inferior temporal to hippocampus ratio (standardised beta 0.59 and 0.40, both p 0.05).
The authors set clear boundaries around this. No participant had autopsy confirmation, so the labels remain clinical, not pathological. The cohort's mean age of 66 meant the criterion of age generally above 75 could not be applied, and the authors note that the low mean age may explain why the 8.2 percent figure sits below the 25 to 40 percent seen in older clinicopathological series. Because the Amsterdam cohort favours thorough initial workup over long follow-up, the criteria had to be operationalised on cross-sectional data rather than on the slow course the criteria describe. Notably, the Probable LATE group was not older than the Alzheimer's group, which the authors say points to a more complicated relationship between age and clinically defined LATE than expected. Neurology published a companion editorial on diagnosing Alzheimer's disease, LATE and both together in the same issue.