Plasma pTau 217/β-amyloid 1-42 ratio for enhanced accuracy and reduced uncertainty in detecting amyloid pathology

Blood biomarkers have the potential to revolutionize Alzheimer's disease diagnosis, offering advantages over CSF and PET due to their accessibility, scalability and cost-effectiveness. This study evaluated the effectiveness of individual plasma biomarkers, such as phosphorylated Tau (pTau) 217, as well as biomarker combinations, with a focus on the pTau 217/β-Amyloid (Aβ) 1-42 ratio to predict amyloid positivity. To improve clinical utility, a dual threshold approach was applied to maximize predictive values and positive likelihood ratios while minimizing the proportion of indeterminate results. Plasma samples from 208 participants (including seven with subjective cognitive decline, 150 with mild cognitive impairment, 12 with Alzheimer's disease dementia and 39 with other cognitive conditions) from three cohorts (BioFINDER2, BIOCARD and MissionAD) were analysed to measure Aβ1-42, Aβ1-40 and pTau217 levels using the Fujirebio LUMIPULSE® G1200 platform. Amyloid status was determined by FDA-cleared PET imaging and/or CSF biomarker ratios. Logistic regression modelling evaluated biomarkers either individually or in combination to identify those that best distinguished amyloid positivity. Clinically applicable thresholds were established through likelihood ratio analysis and further evaluated based on predictive values. When assessing the ability of individual plasma biomarkers to differentiate between amyloid-positive and amyloid-negative participants, plasma pTau217 (P < 0.001) and plasma Aβ1-42 (P = 0.0056) demonstrated significant discriminative power, whereas Aβ1-40 (P = 0.30) did not. Notably, the integration of these biomarkers into the plasma pTau217/Aβ1-42 ratio demonstrated enhanced classification performance (P < 0.001). Using a two-threshold approach based on positive and negative likelihood ratios (PLR/NLR) targets of 14/20, respectively, the plasma pTau217/Aβ1-42 ratio achieved a positive predictive value (PPV) of 94.44% and negative predictive value (NPV) of 94.28%, in the parametric model, comparable to plasma pTau217 alone (PPV: 94.44%, NPV: 94.28%), but yielded fewer indeterminate results (26.5% versus 38.6%). Using a non-parametric model, the plasma ratio achieved a PPV and NPV of 94.62% and 91.78%, respectively, while plasma pTau217 alone achieved 92.41% and 92.86%; the ratio once again reduced the proportion of indeterminate results (20.2% versus 35.1%). The plasma pTau217/Aβ1-42 ratio demonstrated superior performance in identifying amyloid pathology and reduced the frequency of indeterminate results compared to plasma pTau217 alone. These findings support the evaluation of the clinical utility of the plasma pTau217/Aβ1-42 ratio as a tool for identifying amyloid pathology in patients presenting with cognitive complaints.