Biomarkers of synaptic and neuronal dysfunction in relation to mild behavioral impairment: a study of dementia‐free older adults

Abstract

AbstractBackgroundMild behavioral impairment (MBI) represents a high‐risk state for incident cognitive decline and dementia, capturing later‐life emergent and persistent neuropsychiatric symptoms in five domains: 1) decreased motivation; 2) affective dysregulation 3) impulse dyscontrol; 4) social inappropriateness; and 5) psychosis. We examined cross‐sectional associations between MBI and cerebrospinal fluid levels of markers of synaptic/neuronal dysfunction including 1) alpha‐synuclein (α‐syn); 2) growth‐associated protein 43 (GAP‐43); 3) neurogranin (NGRN); 4) synaptosomal‐associated protein 25 (SNAP‐25); 5) visinin‐like protein 1 (VILIP‐1); and 6) chitinase‐3‐like protein 1 (YKL‐40). Longitudinally, we determined associations between these biomarkers and incident MBI or dementia.MethodAlzheimer’s Disease Neuroimaging Initiative participants with normal cognition and mild cognitive impairment were included (n = 853). MBI scores were derived from the Neuropsychiatric Inventory using a published algorithm. Participants were classified as MBI+ if MBI symptoms were present at ≥2 consecutive visits.Cross‐sectional associations between MBI total score and each log‐transformed biomarker were modelled using multivariable‐adjusted linear regression. Biomarkers with cross‐sectional associations meeting a pre‐defined p‐value threshold (p<0.1) were then incorporated as exposures into adjusted longitudinal Cox proportional hazard regression models (n = 111) of 1) MBI and 2) dementia as the outcomes, the latter of which also included interactions between MBI status and biomarker levels.ResultDemographics are summarized in Table 1. Cross‐sectionally, only MBI associations with VILIP‐1 (beta = 3.3%, 95%CI:+0.2% to +6.5%, p = 0.04) and SNAP‐25 (beta = 2.0%, 95%CI:‐0.4% to 6.3%, p = 0.09) met the pre‐defined p‐value for inclusion in Cox models. Longitudinally, higher VILIP‐1 levels were associated with a 3.02‐fold greater rate of incident MBI (95%CI:1.43‐6.40, p = 0.004) and a 3.83‐fold greater rate of incident dementia (95%CI:1.22‐12.03, p = 0.021). The relationship between VILIP‐1 and dementia did not depend on participants having MBI. MBI status predicted incident dementia with an HR = 4.75 (95%CI:2.36‐9.55, p<0.001) and HR = 4.06 (95%CI:2.01‐8.21, p<0.001) after adjusting for VILIP‐1.ConclusionElevated VILIP‐1 levels are associated with greater incidence of MBI and greater incidence of dementia, suggesting that synaptic dysfunction traditionally related to cognitive decline and dementia may also contribute to behavioral changes. Treatments aimed to mitigate MBI symptoms and dementia may explore VILIP‐1—a neuronal calcium‐sensor protein—as a therapeutic target, though more research is needed.