Factors modifying the association between blood‐based biomarkers for Alzheimer’s disease and amyloid status

Abstract

AbstractBackgroundRecent evidence suggests high potential for implementation of plasma biomarkers for Alzheimer’s disease(AD) such as amyloid‐beta42/40(Aβ42/40)‐ratio, tau‐phosphorylated‐at‐threonine‐181(P‐tau181), neurofilament‐light‐chain(NfL) and glial‐fibrillary‐acidic‐protein(GFAP). However, it is crucial to understand which factors could affect the diagnostic value of plasma biomarkers before they are implemented in clinical care. Therefore, we aimed to determine which factors influence the association between blood‐based biomarkers for AD and amyloid‐status(Aβs) in a clinically relevant manner.MethodThis study included 1199 participants from the Amsterdam Dementia Cohort. Baseline diagnoses were subjective cognitive decline(n = 323), mild cognitive impairment(n = 283), AD‐dementia(n = 320), frontotemporal dementia(n = 162) and dementia with Lewy‐bodies(n = 111). Plasma Aβ42/40, P‐tau181, GFAP and NfL were measured using Single‐Molecule‐Array‐technology(neurology‐4‐plex‐E, P‐tau181‐V2,Quanterix,USA). Aβs was determined using amyloid PET(n = 309) or CSF P‐tau181/Aβ42(n = 890) based on cut‐offs published in previous studies. The possible influencing factors were recorded during clinical dementia work‐up or from medical records. We applied logistic regression using plasma biomarker as predictor and Aβs as outcome in the total cohort. In separate models adjusted for age, possible influencing factors were added as covariate to assess confounding and biomarker*factor interactions to assess effect‐modification. We used ROC‐curves to estimate stratified area under the curve and concordance percentage between plasma biomarkers and traditional biomarkers to determine the clinical relevance based on plasma biomarker cut‐off values using Youden’s Index in the total cohort.ResultsStroke(pinteraction<0.002) and antipsychotics(pi<0.05) modified the association between plasma Aβ42/40 and Aβs(Figure 1). Hypertension, hypercholesterolemia(both pi<0.02), body‐mass‐index[BMI], alcohol‐consumption, Charlson‐Comorbidity‐Index[CCI], antihypertensives, and blood‐pressure(all pi<0.002) modified the association between plasma P‐tau181 and Aβs. Stroke, antidepressants, renal‐function(all pi<0.01), hypercholesterolemia‐medication(pi<0.05), creatinine(pi<0.005) and CCI(pi<0.001) modified the association between plasma GFAP and Aβs. Hypertension and antihypertensives(both pi<0.005) modified the association between plasma NfL and Aβs while BMI, creatinine, haemoglobin, erythrocytes and CCI were confounders. The concordance between plasma Aβ42/40 and amyloid PET or CSF P‐tau181/Aβ42 changed>5% in individuals with stroke and antidepressants‐users(Table 1). Similarly, the concordance of GFAP changed in individuals with stroke.ConclusionMultiple factors were statistically significant factors in the association between these plasma biomarkers and Aβs. However, in most cases minimal clinically relevant change was detected except for stroke and antidepressant‐use in the association between Aβ42/40 and Aβs and stroke in the association between GFAP and Aβs.