Performance of plasma p-tau217 for the detection of amyloid-β positivity in a memory clinic cohort using an electrochemiluminescence immunoassay

Abstract

Background: Plasma p-tau217 has emerged as the most promising blood-based marker (BBM) for the detection of Alzheimer Disease (AD) pathology, yet few studies have evaluated plasma p-tau217 performance in memory clinic settings. We examined the performance of plasma p-tau217 for the detection of AD using a high-sensitivity immunoassay in individuals undergoing diagnostic lumbar puncture (LP).

Methods: Paired plasma and cerebrospinal fluid (CSF) samples were analysed from the TIMC-BRAiN cohort. Amyloid (Aβ) and Tau (T) pathology were classified based on established cut-offs for CSF Aβ₄₂ and CSF p-tau181 respectively. High-sensitivity electrochemiluminescence (ECL) immunoassays were performed on paired plasma/CSF samples for p-tau217, p-tau181, Glial Fibrillary Acidic Protein (GFAP), Neurofilament Light (NfL) and total tau (t-tau). Biomarker performance was evaluated using Receiver-Operating Curve (ROC) and Area-Under-the-Curve (AUC) analysis.

Results: Of 108 participants (age: 69 ± 6.5 years; 54.6% female) with paired samples obtained at time of LP, 64.8% (n = 70/108) had Aβ pathology detected (35 with Mild Cognitive Impairment and 35 with mild dementia). Plasma p-tau217 was over three-fold higher in Aβ + (12.4 pg/mL; 7.3-19.2 pg/mL) vs. Aβ- participants (3.7 pg/mL; 2.8-4.1 pg/mL; Mann-Whitney U = 230, p < 0.001). Plasma p-tau217 exhibited excellent performance for the detection of Aβ pathology (AUC: 0.91; 95% Confidence Interval [95% CI]: 0.86-0.97)-greater than for T pathology (AUC: 0.83; 95% CI: 0.75-0.90; z = 1.75, p = 0.04). Plasma p-tau217 outperformed plasma p-tau181 for the detection of Aβ pathology (z = 3.24, p < 0.001). Of the other BBMs, only plasma GFAP significantly differed by Aβ status which significantly correlated with plasma p-tau217 in Aβ + (but not in Aβ-) individuals. Application of a two-point threshold at 95% and 97.5% sensitivities & specificities may have enabled avoidance of LP in 58-68% of cases.

Conclusions: Plasma p-tau217 measured using a high-sensitivity ECL immunoassay demonstrated excellent performance for detection of Aβ pathology in a real-world memory clinic cohort. Moving forward, clinical use of plasma p-tau217 to detect AD pathology may substantially reduce need for confirmatory diagnostic testing for AD pathology with diagnostic LP in specialist memory services.

Keywords: Alzheimer disease; Amyloid; Blood-based markers; Cerebrospinal fluid; Diagnosis; P-tau217.

Fig. 1

P-tau217 Exhibits Excellent Performance for the Detection of Aβ Pathology in Individuals with MCI/Dementia. Plasma p-tau217 was measured in 108 individuals undergoing diagnostic lumbar puncture for the detection of Alzheimer Disease pathology. Paired plasma samples were analysed for p-tau217. A (i) P-tau217 was nearly four-fold higher in Aβ + vs Aβ- individuals (Mann–Whitney U = 230; p < 0.001). The red dotted line indicates the Youden optimised cut-off. (A) (ii) p-tau217 exhibited excellent performance in the detection of Aβ + status (Area-Under the Curve [AUC]: 0.91; 0.86–0.97). B (i) P-tau217 was significantly elevated in T + vs T- individuals. (ii) Performance of p-tau217 for detection of T + pathology alone gave an AUC of 0.83 (0.75–0.90) which was significantly lower than that for Aβ positivity (DeLong test, p = 0.04). C (i) Significant differences were not seen in concentrations of p-tau217 between A- T- and A- T + individuals or between A + T- and A + T + individuals supporting the role of p-tau217 as a marker of amyloid positivity. (ii)For Aβ + individuals, concentrations were significantly higher (p = 0.03) in individuals with dementia vs MCI due to AD. D (i) CSF p-tau217 was significantly higher in individuals with Aβ positivity. (ii) CSF p-tau217 had lower performance than plasma p-tau217 with an AUC 0.83 (0.75–0.91) with a trend for statistical significance (p = 0.05). (iii) Significant correlations were observed between CSF and plasma p-tau in individuals with Aβ positivity. ****p < 0.0001, ***p < 0.001, ** < 0.01, *p < 0.05, ns: non-significant; AUC: Area-Under-the-Curve

Fig. 2

Exploration of Two-Point Thresholds for Plasma p-tau217. A In order to examine different thresholds of sensitivity and specificity, we considered performance of plasma p-tau217 at three thresholds: (i) 90% sensitivity and 90% specificity; (ii) 95% sensitivity and 95% specificity; (iii) 97.5% sensitivity. Those above these specificity and below these sensitivity cut-offs were judged to have high risk and low risk of CSF-determined Aβ positivity respectively. Shaded areas indicate those in the intermediate category, with scores above the specified sensitivity cut-off but below the specificity cut off. B Tabular results obtained by applying these cut-offs indicating low, intermediate and high risk of CSF-determined Aβ, presented by CSF-defined Aβ status. C Positive Predictive Accuracy (PPA), Negative Predictive Accuracy (NPA) and Overall Percent Agreement for p-tau217 positive and negative participants are provided at each threshold

Fig. 3

P-tau217 Outperforms p-tau181 for the Detection of Aβ Positivity in Individuals with MCI/Dementia. Plasma samples were analysed for both p-tau181 and p-tau217. A P-tau181 was 1.7 times higher in Aβ + vs Aβ- individuals (Mann–Whitney U = 661; p < 0.001). B p-tau181 exhibited inferior performance than p-tau217 in the detection of Aβ + status (Area-Under the Curve [AUC]: 0.91; 0.86–0.97 for p-tau217 vs 0.72; 0.62–0.83, DeLong test p < 0.001). C Significant differences were not seen in concentrations of p-tau181 between A- T- and A- T + individuals or between A + T- and A + T + individuals. D There were no significant differences between dementia and MCI for either Aβ + or Aβ – groups. E Significant correlations were seen between plasma p-tau181 and plasma p-tau217 in Aβ + (Spearman’s R = 0.60, p < 0.001) but not Aβ – (Spearman’s R = 0.16, p = 0.34) individuals. F Significant correlations were seen between CSF p-tau181 and plasma p-tau217 again in Aβ + (Spearman’s R = 0.63, p < 0.001) but not Aβ – (Spearman’s R = 0.10, p = 0.60) individuals. ****p < 0.0001, ***p < 0.001, ** < 0.01, *p < 0.05, ns: non-significant; AUC: Area-Under-the-Curve

Fig. 4

Plasma P-tau217 is Significantly Correlated with Plasma GFAP, Plasma Total-tau and CSF Total-tau in Aβ + Individuals. Plasma and CSF samples were additionally analysed for Glial Fibrillary Acidic Protein (GFAP), Neurofilament Light (NfL) and Total tau (t-tau). A (i) Plasma GFAP significantly differed between Aβ + vs Aβ- individuals (U = 811, p = 0.004) (i-vi) None of the additional markers differed in Aβ + vs Aβ – individuals. B (i) Significant correlations were observed between plasma GFAP in Aβ + but not Aβ – individuals. (ii, v) No correlations were seen between plasma or CSF NfL and plasma p-tau217. (iv) CSF GFAP did not correlate with plasma p-tau217. (iii, vi) Significant correlations were seen between plasma p-tau217 and both plasma and CSF t-tau in Aβ + but not Aβ – individuals. ****p < 0.0001, ***p < 0.001, ** < 0.01, *p < 0.05, ns: non-significant