Plasma biomarkers discrimination accuracy of biologically defined Alzheimer’s disease in a memory clinic setting

Abstract Background Blood‐based biomarkers offer a non‐invasive and cost‐effective means for Alzheimer's disease (AD) detection. In this study, we performed a direct comparison of these novel biomarkers in a memory clinic population to facilitate their implementation into clinical practice. Method We included a total of 208 patients with cognitive complaints from the BIODEGMAR cohort at Hospital del Mar (Barcelona, Spain). CSF was used as standard‐of‐truth and patients were categorized as having an AD CSF profile with two approaches: Lumipulse CSF Aβ42/p‐tau181<10.25 (N=208) or Elecsys® p‐tau181/Aβ42>0.022 (N=157). The following biomarkers were measured in paired plasma and CSF samples: Aβ42 and p‐tau181 (Lumipulse CSF<CE>, plasma<RUO>, Fujirebio), Aβ42 and p‐tau181 (NeuroToolKit (NTK) a panel of robust prototype assays (Roche Diagnostics International Ltd), and p‐tau181, p‐tau217, p‐tau231 and MAP‐T (NULISA, Alamar). P‐value and effect size of the group comparison (AD vs non‐AD CSF profiles) were calculated using a Mann‐Whitney U test. ROC curve analysis evaluated plasma biomarkers accuracy to discriminate AD from non‐AD CSF profiles. Spearman test was used to assess the correlation between plasma and CSF biomarkers. Result Demographic information of the study cohort is reported in Table 1 for both classification thresholds. All plasma biomarkers were significantly different between the AD and non‐AD CSF profile groups, but the effect size varied among them. Specifically, NULISA p‐tau217, Lumipulse p‐tau181 and NULISA p‐tau231 reported the greatest effect sizes for the Lumipulse threshold and NULISA p‐tau217, NTK p‐tau181, Lumipulse p‐tau181 and NULISA p‐tau231 reported the greatest effect sizes for the Elecsys® threshold (Table 1). ROC curve analysis performed consistently over the two thresholds with the following discrimination accuracy (AUC) values (from highest to lowest): NULISA p‐tau217, NTK p‐tau181, NULISA p‐tau231, Lumipulse p‐tau181, NULISA p‐tau181 (Figure 1). In comparison to the p‐tau assays, Aβ42 reported modest effect size and AUCs. NULISA p‐tau217 and NULISA p‐tau231 had the highest correlation between plasma and CSF (Spearman ρ=0.75 and ρ=0.60, respectively, with both FDR corrected p‐values<0.001), Figure 2. Conclusion In a memory clinic population, various plasma Aβ and tau plasma biomarkers, targeting different epitope and using different platforms, demonstrated high performance in distinguishing patients with biologically defined AD from those without.