Subjective estimation of cognitive function in mild cognitive impairment: relationship with neurodegenerative and non-degenerative factors.

BACKGROUND: Subjective cognitive complaints are poor predictors of neurodegenerative disease and future dementia. Errors in metacognition, positive or negative differences between actual and perceived performance, may partially explain this. We aimed to assess whether hypothesized indicators of underlying neurodegenerative factors (e.g. hippocampal atrophy) in mild cognitive impairment (MCI) were associated with overestimation of actual cognitive performance, and hypothesized non-degenerative factors (e.g. depression) were associated with underestimation of performance. METHODS: Metacognitive error was estimated from paired subjective and objective cognitive assessments using the Multifactorial Memory Questionnaire and Addenbrooke's Cognitive Examination - Revised, respectively. A normative model was developed with cognitively healthy older adults (n = 36), and applied to individuals with suspected MCI due to Alzheimer's disease (AD) or MCI with Lewy bodies (total n = 88). Theorized predictors of subjective overestimation or underestimation of performance (metacognitive error) were assessed, including demographics, AD biomarkers, and mental and physical ill health. Metacognitive error was also assessed as a predictor of conversion to dementia. RESULTS: Underestimation of cognitive function was associated with depressive symptoms, anxiety, and self-reported autonomic symptoms. Overestimation of cognitive function was associated with age, hippocampal atrophy, plasma glial fibrillary acidic protein, and subsequent dementia conversion. CONCLUSIONS: Underestimation of cognitive function may reflect functional cognitive changes linked to mental and physical ill health, while overestimation of function may be a marker of neurodegenerative changes. Quantifying metacognitive error may provide a noninvasive screening tool for progressive MCI, requiring investigation in an independent sample.