Age-Adjusted Serum Neurofilament Predicts Cognitive Decline in Parkinson's Disease (MARK-PD).

Abstract

We read with great interest the study by Aamodt and colleagues reporting that plasma neurofilament light (NfL) chain levels are associated with UPDRS III (Unified Parkinson's Disease Rating Scale, Part III) score but not the Mattis Dementia Rating Scale (DRS-2), whereas plasma NfL predicted cognitive decline (DRS-2 change) but not motor worsening (UPDRS III ≥5 points).1 Interestingly, these findings are in line with our results that serum NfL was associated with UPDRS III but not Montreal Cognitive Assessment (MoCA) score, whereas serum NfL predicted cognitive decline (decrease in MoCA score >2 points) but not motor worsening (increase in UPDRS III >4 points) in the Biomarkers in Parkinson's Disease (MARK-PD) study.2 Numerous studies have shown that blood NfL predicts cognitive decline in patients with Parkinson's disease, but cutoff values determined by receiver operating characteristic (ROC) analysis revealed very different results.1-3 Given that serum NfL levels increase by 3.35% per year in PD patients and double in individuals above 70 years compared with those below 50 years in the normal population, age is the most important covariate of NfL variability.4, 5 Consequently, we hypothesize that age-adjusted NfL percentiles might reveal more consistent results among cohort studies compared with absolute NfL values for the prediction of disease progression in PD patients. Therefore, we evaluated if the plasma NfL cutoff value of 14.6 pg/mL determined by Aamodt and colleagues also predicts cognitive decline in the Mark-PD study.6 Furthermore, we analyzed if age-adjusted serum NfL above the 95th percentile predicts cognitive decline in MARK-PD. In MARK-PD, serum NfL and cognitive outcome (initial and follow-up MoCA scores) were available for 106 patients with PD (age: 64.6 ± 8.8 years, men: 60.4%, disease duration: 12 [7.5, 15.5] years, follow-up: 512 [365, 680] days). During follow-up, we registered 21 patients with cognitive decline (decrease in MoCA score >2 points). In a Kaplan–Meier analysis, we observed no association between serum NfL levels and faster cognitive decline using a cutoff value of 14.6 pgl/mL (log-rank test: P = 0.322) (Fig. 1A). However, patients with NfL levels above the age-adjusted 95th percentile revealed a significantly faster cognitive decline (P = 0.026) than those below (Fig. 1B). This difference remained significant after further adjustment for age, sex, disease duration, and initial MoCA score (P = 0.017) (Fig. 1C). In summary, both studies using different cognitive scales (ie, DRS-2 and MoCA) show that increased NfL levels predict cognitive decline in PD patients. We suggest that NfL levels above the age-adjusted 95th percentile are independently associated with cognitive decline. We propose that age-adjusted NfL cutoffs might reveal more consistent results across different cohort studies and therefore facilitate the establishment of NfL as a clinically useful blood-based biomarker in patients with PD. The data that support the findings of this study are available from the corresponding author upon reasonable request.