Association of the learning ratio with AD biomarkers after application of actuarial diagnostic criteria to ADNI cognitively normal participants

Abstract

AbstractBackgroundEarly and accurate diagnosis is critical as disease‐modifying therapies for Alzheimer’s disease (AD) become available and clinical trials are moving increasingly into early disease stages. The Learning ratio (LR) has demonstrated the highest sensitivity towards changes in memory, relative to other learning slope metrics. The current study examines the performance of LR within the ADNI cognitively normal population.MethodOur analyses included 672 participants aged 55 to 90 years, diagnosed in ADNI as cognitively intact. After applying Jak/Bondi actuarial diagnostic criteria to the sample1, 2, 56 participants met actuarial criteria for MCI (actuarial‐MCI), 253 had normal cognition with subjective memory complaints (actuarial‐SMC), and 363 were cognitively normal without subjective complaints (actuarial‐CN). We compared LR derived from the Rey Auditory Verbal Learning Test (RAVLT)3 and the Word Recall subtest of the ADAS‐Cog4 between the diagnostic groups, and studied the effects of LR, diagnosis, and the interaction between LR and diagnosis on bilateral hippocampal volume and β‐amyloid pathology.ResultActuarial‐MCI performances on RAVLT LR and Word Recall LR were significantly worse than actuarial‐SMC and actuarial‐CN (ps<.005). No differences were seen between LRs for actuarial‐CN and actuarial‐SMC (ps>.05). The association between Word Recall LR and hippocampal volume was strongest in actuarial‐MCI followed by actuarial‐SMC, both of which displayed a positive relationship between the two variables, while actuarial‐CN displayed a negative relationship (R2 = .209, p<.001, Figure 1A). Adding the interaction term improved model performance (R2 change = .007 p = .013). Similar trends were observed between RAVLT LR and diagnostic group predicting hippocampal volume, though the interaction was not significant (R2 = .208, p<.001, Figure 1B; R2 change = .001 p = .28). No main or interaction effects were observed between LR measures, diagnostic groups, and amyloid status (ps>.05, Figure 2).ConclusionA subset of ADNI cognitively intact participants displayed MCI after applying Jak/Bondi actuarial norms, suggesting that ADNI diagnostic criteria may have under‐pathologized memory loss in a subset of patients. The actuarial‐MCI group displayed worse learning slopes and stronger relationships with hippocampal volume than CN participants with and without SMC. Applying the actuarial criteria to AD clinical trials – in conjunction with LR – may yield greater sensitivity in detecting those in the early stages of AD.