Are empirically-derived subtypes of mild cognitive impairment consistent with conventional subtypes?

Given the importance of identifying dementia prodromes for future treatment efforts, we examined two methods of diagnosing mild cognitive impairment (MCI) and determined whether empirically-derived MCI subtypes of these diagnostic methods were consistent with one another as well as with conventional MCI subtypes (i.e., amnestic, non-amnestic, single-domain, multi-domain). Participants were diagnosed with MCI using either conventional Petersen/Winblad criteria (n = 134; >1.5 SDs below normal on one test within a cognitive domain) or comprehensive neuropsychological criteria developed by Jak et al. (2009) (n = 80; >1 SD below normal on two tests within a domain), and the resulting samples were examined via hierarchical cluster and discriminant function analyses. Results showed that neuropsychological profiles varied depending on the criteria used to define MCI. Both criteria revealed an Amnestic subtype, consistent with prodromal Alzheimer's disease (AD), and a Mixed subtype that may capture individuals in advanced stages of MCI. The comprehensive criteria uniquely yielded Dysexecutive and Visuospatial subtypes, whereas the conventional criteria produced a subtype that performed within normal limits, suggesting its susceptibility to false positive diagnostic errors. Whether these empirically-derived MCI subtypes correspond to dissociable neuropathologic substrates and represent reliable prodromes of dementia will require additional follow-up.