UC San Diego

Abstract: Background: In Parkinson’s disease (PD), concomitant Alzheimer’s disease (AD) pathologic change (ADNC) is common and results in altered motor and cognitive phenotypes. However, detection of PD with AD (PD+AD) using biofluid markers is challenging. While decreased cerebrospinal fluid (CSF) β‐amyloid 1‐42 (Aβ42) strongly reflects β‐amyloid burden, PD subjects typically harbor lower CSF phosphorylated tau 181 (p‐tau181) and total tau (t‐tau) levels than healthy controls, which complicates detection of tau tangles and neurodegeneration. We previously tested PD‐specific application of the β‐amyloid/tau/neurodegeneration framework (ATNPD); combining CSF Aβ42, CSF p‐tau181, and serum neurofilament light (NfL) in a living PD cohort. ATNPD, using a lower CSF p‐tau181 cutpoint, predicted cognitive decline. However, ATNPD cutpoints still must be validated against autopsy assessments of ADNC as gold‐standard. Here, we compare biomarker strategies in all available autopsy‐confirmed PD from the Parkinson’s Progression Markers Initiative (PPMI). Methods: Eighteen PD participants with autopsy‐confirmed Lewy body disease and antemortem biofluid were available for analysis (Table 1). PD+AD included high/intermediate ADNC (n=9); PD without AD (PD; n=9) included not/low ADNC. Cerebral cortical atrophy determined neurodegeneration (mild/moderate vs. none). CSF was assayed for Aβ42 (n=14), p‐tau181 (n=17), and t‐tau (n=17) using Roche cobas e 601; p‐tau181/Aβ42 and t‐tau181/Aβ42 ratios were calculated. Serum NfL was assayed using Simoa Quanterix (n=18). Biofluid measurements closest to autopsy were selected. Receiver operating characteristic (ROC) analyses with bootstrapping tested discrimination of PD+AD from PD using CSF biomarkers, and of neurodegeneration from not using CSF t‐tau and serum NfL. Results: ROC cutpoints for CSF Aβ42, p‐tau181, and serum NfL were equivalent to ATNPD cutpoints, while p‐tau181 and t‐tau were lower than published AD‐cutpoints (Table 2). CSF p‐tau181/Aβ42, t‐tau181/Aβ42, Aβ42 and serum NfL had high area under the curve (AUC>0.80; Table 2A,2B). In contrast, CSF p‐tau181 and t‐tau demonstrated poor discrimination (Table 2A) and no difference between groups (Table 1), potentially due in part to low sample size. A chi‐square test confirmed classification is improved using ATNPD and AD‐cutpoints (χ2=14, p=0.0015; Figure 1). Conclusions: PD‐specific biomarker strategies/cutpoints are needed to maximize detection of concomitant ADNC, but must be validated in larger autopsy cohorts.