ABSTRACT: Subjective memory decline is associated with neurodegeneration and increased risk of cognitive decline in participants with no or subjective cognitive impairment, while in patients with mild cognitive impairment or Alzheimer's-type dementia, findings are inconsistent. Our aim was to provide a comprehensive overview of subjective memory decline changes, relative to objective memory performances, and of their relationships with neurodegeneration, across the clinical continuum of Alzheimer's disease. Two hundred participants from the Imagerie Multimodale de la maladie d'Alzheimer à un stade Précoce (IMAP+) primary cohort and 731 participants from the Alzheimer's Disease Neuroimaging Initiative (ADNI) replication cohort were included. They were divided into four clinical groups (Imagerie Multimodale de la maladie d'Alzheimer à un stade Précoce/Alzheimer's Disease Neuroimaging Initiative): controls (n = 67/147, age: 60-84/60-90, female: 54/55%), patients with subjective cognitive decline (n = 30/84, age: 54-84/65-80, female: 44/63%), mild cognitive impairment (n = 50/369, age: 58-86/55-88, female: 45/44%) or Alzheimer's-type dementia (n = 36/121, age: 51-86/61-90, female: 41/41%). Subjective and objective memory scores, and their difference (i.e. delta score reflecting memory awareness), were compared between groups. Then, voxelwise relationships between subjective memory decline and neuroimaging measures of neurodegeneration [atrophy (T1-MRI) and hypometabolism (18F-fluorodeoxyglucose-PET)] were assessed across clinical groups and the interactive effect of the level of cognitive impairment within the entire sample was assessed. Analyses were adjusted for age, sex and education, and repeated including only the amyloid-positive participants. In Imagerie Multimodale de la maladie d'Alzheimer à un stade Précoce, the level of subjective memory decline was higher in all patient groups (all P < 0.001) relative to controls, but similar between patient groups. In contrast, objective memory deficits progressively worsened from the subjective cognitive decline to the dementia group (all P < 0.001). Accordingly, the delta score showed a progressive decline in memory awareness across clinical groups (all P < 0.001). Voxelwise analyses revealed opposite relationships between the subjective memory decline score and neurodegeneration across the clinical continuum. In the earliest stages (i.e. patients with subjective cognitive decline or Mini Mental State Examination > 28), greater subjective memory decline was associated with increased neurodegeneration, while in later stages (i.e. patients with mild cognitive impairment, dementia or Mini Mental State Examination < 27) a lower score was related to more neurodegeneration. Similar findings were recovered in the Alzheimer's Disease Neuroimaging Initiative replication cohort, with slight differences according to the clinical group, and in the amyloid-positive subsamples. Altogether, our findings suggest that the subjective memory decline score should be interpreted differently from normal cognition to dementia. Higher scores might reflect greater neurodegeneration in earliest stages, while in more advanced stages lower scores might reflect decreased memory awareness, i.e. more anosognosia associated with advanced neurodegeneration.
