Cognitive Variability Predicts Incident Alzheimer's Disease and Mild Cognitive Impairment Comparable to a Cerebrospinal Fluid Biomarker

Abstract

Background: Alzheimer's disease (AD) biomarkers are emerging as critically important for disease detection and monitoring. Most biomarkers are obtained through invasive, resource-intense procedures. A cognitive marker, intra-individual cognitive variability (IICV) may provide an alternative or adjunct marker of disease risk for individuals unable or disinclined to undergo lumbar puncture.

Objective: To contrast risk of incident AD and mild cognitive impairment (MCI) associated with IICV to risk associated with well-established biomarkers: cerebrospinal fluid (CSF) phosphorylated tau protein (p-tau181) and amyloid-β 42 (Aβ42) peptide.

Methods: Dispersion in cognitive performance, IICV, was estimated with a published algorithm, and included Trail Making Test A and B, Rey Auditory Verbal Learning Test (RAVLT), and the American National Adult Reading Test (ANART). CSF biomarkers were expressed as a ratio: p-tau181/Aβ42, wherein high values signified pathognomonic profiles. Logistic regression models included longitudinal data from 349 Alzheimer's Disease Neuroimaging Initiative (ADNI) participants who completed lumbar puncture. All subjects were cognitively healthy (n = 105) or diagnosed with MCI (n = 244) at baseline. We examined odds of conversion associated with baseline elevations in IICV and/or ratio of CSF p-tau181/Aβ42.

Results: When included in models alone or in combination with CSF p-tau181/Aβ42, one standard IICV unit higher was associated with an estimated odds ratio for incident AD or MCI of 2.81 (95% CI: 1.83-4.33) in the most inclusive sample, and an odds ratio of 3.41 (95% CI: 2.03-5.73) when restricted to participants with MCI. Iterative analyses suggested that IICV independently improved model fit even when individual index components were included in comparative models.

Conclusions: These analyses provide preliminary support for IICV as a marker of incident AD and MCI. This easily-disseminated, non-invasive marker compared favorably to well-established CSF biomarkers.

Keywords: Alzheimer’s disease; amyloid beta-protein; biological markers; cerebrospinal fluid; cognition; cognitive dysfunction; incidence studies; mild cognitive impairment; tau protein.

Figure 1

CONSORT Diagram – Sample derivation for logistic regression model including subjects with complete longitudinal and APOE e4 and CSF data. Analytic model included baseline age, years of education, APOE e4 status, years of follow-up, and baseline diagnosis. ^aThe distributions of cognitive test scores from 1324 participants were used to derive IICV estimates. ^bOne subject died during length of follow-up. The subject was included; brain autopsy was used for the final diagnosis instead of clinical examination.

Figure 2

IICV vs. Conversion status with local smoothing trend (LOESS). Data are jittered around conversion status to better illustrate data densities. Figure 2a includes participants with MCI and those who were cognitively healthy at baseline (N=349). Figure 2b includes only those individuals with MCI at baseline (N=244). Notes: IICV: Intra-Individual Cognitive Variability. Used raw IICV score for comparison. Conversion is a dichotomous variable: Converters vs. Non-converters. The trend line is smoothed using LOESS in order to illustrate the relationship between conversion and IICV score without linearity assumptions. Additionally, IICV scores are jittered to more clearly see the number of participants obtaining individual IICV scores.

Figure 3

Ratio of CSF p-tau₁₈₁/Aβ₄₂ vs. Conversion status with local smoothing trend (LOESS). Data are jittered around conversion status to better illustrate data densities. Figure 3a includes participants with MCI and those who were cognitively healthy at baseline (N=349). Figure 3b includes only those individuals with MCI at baseline (N=244). Notes: Conversion is a dichotomous variable: Converters vs. Non-converters. The trend line is smoothed using LOESS in order to illustrate the relationship between conversion and CSF biomarkers without linearity assumptions. Additionally, values for the ratio of CSF analytes were jittered to more clearly see the number of participants obtaining individual ratio values.