87 Markers of neurodegeneration in subjective cognitive decline Correlations between N biomarkers The different N biomarkers were modestly to moderately correlated (range r -0.28 – 0.58, Figure 2A). Serum markers NfL and GFAP correlated most strongly (r 0.58, p <0.01). P-tau and t-tau, representing different AT(N) biomarker groups (T and N respectively), were very strongly correlated (r 0.89, p <0.01). Overall, the correlation coefficients between the different biomarkers for N were in a similar range as the correlation coefficients between the different biomarkers for N on the one hand, and biomarkers for A and T on the other hand (r -0.43 – 0.33, excluding the correlation between p-tau and t-tau). After adjusting for age and sex, drastically lower coefficients were observed (Figure 2B). Figure 1. Correlations between N biomarkers Heatmaps showing correlations between different biomarkers. A. Correlation coefficients (Pearson), B. Correlation coefficients (partial correlation, adjusted for age and sex). Phosphorylated tau (p-tau), total tau (t-tau), neurofilament light (NfL), and glial fibrillary acidic protein (GFAP) were log-transformed. HV = hippocampal volume; MTA = medial temporal atrophy. Risk of progression to MCI or dementia We investigated the predictive value of the different N biomarkers using Cox proportional hazards analyses. The mean follow-up duration was 3.8 years (± 2.8 years). In uncorrected models, t-tau, MTA, HV, NfL and GFAP all predicted clinical progression to MCI or dementia (Table 2, model 1). After adding covariates in model 2 (age and sex), 3 (Aβ, age and sex) and 4 (Aβ, p-tau, age and sex), hazard ratios were attenuated. Model 4 showed that HV, NfL and GFAP added predictive value to Aβ and p- Aβ. T-tau also predicted MCI or dementia in models 1 to 3, but was not entered in model 4 due to collinearity between t-tau and p-tau. In an additional explorative 4
RkJQdWJsaXNoZXIy MjY0ODMw