Metallothionein and neurodegenerative diseases.

Mild cognitive impairment (MCI) is widely regarded as an intermediate stage of cognitive decline that falls between normal cognition and dementia (Scheltens et al., 2021; Frisoni et al., 2022). The condition is typically characterized by a relatively circumscribed memory deficit in the context of preserved mental status and independence in day-to-day functioning (Petersen et al., 2009; Petersen et al., 2013; Dubois et al., 2021; Mengel et al., 2025). Alzheimer’s disease (AD) is a common presumed cause of MCI, although it can have other neuropathological (e.g., dementia with Lewy bodies, limbic-predominant age-related TDP-43 encephalopathy [LATE], primary age related tauopathy [PART]), medical (e.g., cerebrovascular disease, infections, sleep disorders, thyroid disorders), or psychiatric (e.g., depression) causes. Numerous clinical trials have attempted to alleviate MCI and slow progression to dementia, with limited success (Huang et al., 2020, 2023; Cummings et al., 2021). Factors contributing to past failed trials may include variability in the underlying cause of MCI across trial participants or difficulty evaluating treatment effects due to cognitive heterogeneity involving impairment of language, executive functions, attention, or visuospatial abilities, either alone or in conjunction with impaired memory. Current clinical trials at the MCI stage of AD (i.e., prodromal AD) often address etiological heterogeneity by employing cerebrospinal fluid (CSF) or plasma AD biomarkers to confirm the presumed underlying cause of impairment (Dubois et al., 2021; Altomare et al., 2023). While this reduces one aspect of heterogeneity in the targeted cohort, it may not reduce variability in the cognitive abilities most affected during the MCI phase of AD. The impact of cognitive variability might be reduced by clinically sub-categorizing MCI into amnestic and non-amnestic types with single or multiple cognitive domain subtypes (Winblad et al., 2004), but this level of clinical subtyping can be challenging because it relies on subjective interpretation of overlapping patterns of performance on cognitive tests, potentially leading to unreliable categorization. More sophisticated and objective subtyping of MCI can be achieved through data-driven clustering techniques that analyze natural groupings and similarities within a panel of neuropsychological test scores (Clark et al., 2013; Bondi et al., 2014; Eppig et al., 2017; Edmonds et al., 2021, 2024). When applied to cohorts of individuals with clinically-diagnosed MCI, the data-driven approach provides greater diagnostic precision than clinical subtyping, and results in stronger associations with AD biomarkers and more accurate prediction of decline (Bondi et al., 2014; Ferreira et al., 2020; Krell-Roesch et al., 2021; Chen et al., 2023). Previous studies using data-driven approaches to subtype MCI have not restricted their cohort to those with biomarker confirmed AD, although they have shown stronger associations with biomarkers than conventional diagnostic schema (Clark et al., 2013; Bondi et al., 2014; Eppig et al., 2017; Edmonds et al., 2021, 2024; Shand et al., 2023). Thus, the nature of cognitive variability in a more circumscribed biomarker-positive prodromal AD cohort—and its possible implications for rate of cognitive decline, changes in neuroimaging, and ability to monitor treatment effects—remains unknown. In a past study that applied data-driven subtyping in a cohort with mild to moderate dementia due to AD, we identified two cognitive clusters: a “typical” AD cognitive profile with greater impairment to memory than to other cognitive abilities and an “atypical” AD cognitive profile in which impairment to attention/executive functions was greater and impairment to memory was less than in the typical pattern (Qiu et al., 2019). In a subsample with autopsy-confirmed AD, 79.6% of patients had the “typical” AD cognitive profile, and 20.4% had the “atypical” profile. Atypicality was associated with younger age, fewer apolipoprotein E (APOE) ε4 alleles (a variant known to confer genetic risk for AD), lower levels of neurofibrillary tangle pathology (i.e., Braak stage), less severe global dementia, higher depression scores, and slower cognitive decline. The typical or atypical cognitive profile persisted over four years of annual testing, suggesting that distinct cognitive subtypes may develop early and last throughout the course of the disease.