Stability of amnestic mild cognitive impairment: Cantab paired associate learning as a predictor of a consistent diagnosis

Background: The initial clinical diagnosis of mild cognitive impairment (MCI) can be unstable, with the degree of cognitive impairment rapidly worsening in some people, and apparently resolving in others. This study used a computerised measure of paired associate learning (Cantab PAL) to predict stable versus transient MCI from baseline cognitive scores.

Methods: Participants were recruited from a longitudinal study of neuropsychological function in community-residing adults with suspected MCI. Inclusion criteria were: (i) presence of a subjective cognitive complaints (ii) preserved general cognition; (iii) independent daily functioning. From initial recruitment (n ¼ 286), 200 participants were classified as either amnestic MCI (aMCI: impairments including one or more visual or verbal memory tasks), non-amnestic MCI (na-MCI: impairments in one or more non-memory task) or unimpaired (no cognitive scores <10th percentile relative to normative sample). At 10 month follow-up (n¼122), a number of MCI cases had resolved. The final analysis includes 21 unimpaired controls, 43 with naMCI, 17 with a-MCI, 21 with resolved a-MCI and 43 with resolved na-MCI. Participants underwent Cantab PAL testing as part of their neuropsychological testing at baseline and 10 months.

Results: At both baseline and follow-up, PAL performance was significantly worse in a-MCI patients compared to all other groups, which were not significantly different from each other. This included participants initially diagnosed with an initial diagnosis of a-MCI which then resolved. There was a significant interaction between group (a-MCI vs others) and time, indicating that the rate of decline in the a-MCI group was greater than other groups. Logistic regression showed that both PAL impairment at baseline and change in PAL scores predicted a-MCI group membership, even controlling for age, education and gender.

Conclusions:The Cantab PAL task is a sensitive and specific indicator of amnestic MCI. Patients who perform badly on this test at baseline are a stable group with a clinically-relevant impairment that worsens steadily over time.