P2‐451: RECOGNITION MEMORY PERFORMANCE AS A COGNITIVE MARKER OF PRODROMAL ALZHEIMER'S DISEASE

Abstract

The utility of recognition memory indices for identifying persons with biological evidence of Alzheimer's disease is unclear, as studies have diagnosed mild cognitive impairment due to AD using clinical criteria rather than CSF measures of central amyloidosis and tauopathy. We evaluated whether recognition memory indices of discriminability and response bias derived from signal detection theory differentiate persons with amnestic MCI with positive CSF evidence of AD (prodromal AD) vs. non-prodromal amnestic MCI. 133 participants in the VAScular ContribUtors to ProdromaL AlzheimeR's disease (VASCULAR) study were clinically diagnosed at their screening evaluations with amnestic MCI using ADNI criteria including a subjective memory complaint, impaired performance on Logical Memory, 0.5 on the CDR memory box, and relatively preserved IADLs (FAQ). At their baseline visit, they received the Hopkins Verbal Learning Test-Revised which assesses learning and recall of 12 words, followed by delayed recall and recognition (12 targets, 12 foils). Participants also underwent LPs, and AD biomarkers were measured (ALZBIO3) in Dr. Leslie Shaw's lab (U.Penn). Participants were classified as prodromal AD if they had a tau/amyloid ratio of ≥0.45 using previously published cutoffs (Shaw et al., Alzheimers Dement 10:698-705, 2018). 34 (26%) MCI participants had prodromal AD. They were significantly older (72.2±6.8 years) and had lower MoCA scores (20.9±2.6 points) than the non-prodromal group (64.6±8.5 years; 22.4±2.8 points), but did not differ in education (15.8±2.9 years vs. 15.5±3.1 years). ANCOVAs controlling for age and MoCA scores demonstrated differences in hits (Prodromal: 9.5±2.9, non-Prodromal: 11.2±1.2, p=.004) but not false alarms (Prodromal: 2.1±1.9; non-Prodromal 1.3±1.4, p=.14). A discriminability measure (d’) which simultaneously considers hits and false alarms was lower in the prodromal (2.0±1.9) vs. non-prodromal (5.1±2.2) groups, p<.001. A cutoff of 5.0 had 71% sensitivity and 70% specificity, with 93/133 participants correctly classified. There were no significant differences in response bias (Prodromal: .12±1.3; non-Prodromal: −.42±1.2, p=.10). Individuals with amnestic MCI due to prodromal AD are less accurate in distinguishing targets from foils. Discriminability may serve as a marker of prodromal AD and could be utilized with other memory indices to facilitate differential diagnosis of amnestic MCI due to AD vs. non-AD etiologies.