Authors (including presenting author) :
Chan PL (1), Shea YF (2), Wong YL (1), Lam CW (1)(3)
Affiliation :
(1) Division of Chemical Pathology, Department of Pathology, Queen Mary Hospital, (2) Department of Medicine, Queen Mary Hospital, Li Ka Shing Faculty of Medicine, University of Hong Kong, (3) Department of Pathology, Li Ka Shing Faculty of Medicine, University of Hong Kong
Introduction :
Phosphorylated tau (pTau) 217 has emerged as an important biomarker in Alzheimer’s disease, with higher levels in plasma samples associated with underlying pathology. To investigate the distribution of pTau217, ninety-eight patients were recruited from the Memory Clinic at Queen Mary Hospital. These patients underwent plasma pTau217 measurements using the S-PLEX® assay. Traditional clinical assessment remains an integral part of patient care, but a complementary data-driven approach can provide new insights into subgroup identification.
Objectives :
The primary objective of this study was to apply unsupervised clustering analysis to identify potential subgroups of patients with similar plasma pTau217 levels, without relying on any clinical background information. The secondary objective was to explore the reference interval based on the “normal” subgroup.
Methodology :
A Gaussian Mixture Model (GMM) was applied to the plasma pTau217 data to identify naturally occurring clusters without pre-specified labels. The Bayesian Information Criterion (BIC) was used to determine the optimal number of clusters, and the Shapiro-Wilk test was used to assess the normality of each identified subgroup.
Result & Outcome :
Two distinct clusters were identified as best fitting the dataset from the BIC analysis, with both clusters being normally distributed as assessed by the Shapiro-Wilk test (Cluster 0: p=0.27, Cluster 1: p=0.09). Cluster 0 (n=52) had a mean plasma pTau217 level of 6.20 pg/mL (range 0.62 – 11.50 pg/mL). Its 2.5th–97.5th percentile range of 2.04 – 11.19 pg/mL suggests a putative “normal” reference interval. In contrast, Cluster 1 (n=46) had a significantly higher mean pTau217 level of 26.25 pg/mL (range 11.88 – 52.62 pg/mL). The corresponding 2.5th–97.5th percentile range of 12.20 – 46.88 pg/mL is consistent with possible Alzheimer’s disease pathology. A single cut-off value at 11.69 pg/mL separated these two clusters. These findings highlight two homogeneous subgroups within the cohort, providing a data-driven analytical perspective that complements traditional clinical assessment for interpretation of plasma pTau217 results and future clinical decision-making.
