Clinical utility of an antibody-free LC-MS method to detect brain amyloid deposition in cognitively unimpaired individuals from the screening visit of the A4 Study

Abstract

Introduction: This study explored the ability of plasma amyloid beta (Aβ)42/Aβ40 to identify brain amyloid deposition in cognitively unimpaired (CU) individuals.

Methods: Plasma Aβ was quantified with an antibody-free high-performance liquid chromatography tandem mass spectrometry method from Araclon Biotech (ABtest-MS) in a subset of 731 CU individuals from the screening visit of the Anti-Amyloid Treatment in Asymptomatic Alzheimer's (A4) Study, to assess associations of Aβ42/Aβ40 with Aβ positron emission tomography (PET).

Results: A model including Aβ42/Aβ40, age, apolipoprotein E ε4, and recruitment site identified Aβ PET status with an area under the curve of 0.88 and an overall accuracy of 81%. A plasma-based pre-screening step could save up to 42% of the total number of Aβ PET scans.

Discussion: ABtest-MS accurately identified brain amyloid deposition in a population of CU individuals, supporting its implementation in AD secondary prevention trials to reduce recruitment time and costs. Although a certain degree of heterogeneity is inherent to large and multicentric trials, ABtest-MS could be more robust to pre-analytical bias compared to other immunoprecipitation mass spectrometry methods.

Highlights: Plasma amyloid beta (Aβ)42/Aβ40 accurately identified brain Aβ deposition in cognitively unimpaired individuals from the Anti-Amyloid Treatment in Asymptomatic Alzheimer's (A4) Study.The inclusion of the recruitment site in the predictive models has a non-negligible effect.A plasma biomarker-based model could reduce recruitment costs in Alzheimer's disease secondary prevention trials.Antibody-free liquid chromatography mass spectrometry methods may be more robust to pre-analytical variability than other platforms.

Keywords: Alzheimer's disease; Anti‐Amyloid Treatment in Asymptomatic Alzheimer's (A4) Study; Aβ42/Aβ40; amyloid beta; blood biomarkers; clinical trials; mass spectrometry; pre‐screening.

FIGURE 1

Distribution of plasma biomarkers between Aβ PET(−) and Aβ PET(+) groups and correlation of plasma biomarkers with brain amyloid deposition. Distribution of plasma Aβ40 (A), Aβ42 (B), and Aβ42/Aβ40 (C) between Aβ PET(−) and Aβ PET(+) groups. Group comparisons were carried out using a Mann–Whitney test. Correlations for Aβ40 (D), Aβ42 (E), and Aβ42/Aβ40 (F) and ¹⁸F‐Florbetapir SUVR values. Dashed lines represent 95% confidence intervals. ^*** P < 0.001. Aβ, amyloid beta; n.s., non‐significant; PET, positron emission tomography; SUVR, standardized uptake value ratio.

FIGURE 2

Predictive ability of different regression models for identifying Aβ PET status. ROC curves for discriminating Aβ PET status. Five regression models from the 12 shown in Table 2 are selected for representation. The model with highest AUC and accuracy, as well as lowest AIC, includes Aβ42/Aβ40, age, number of APOE ε4 alleles, and recruitment site. AIC, Akaike information criterion; APOE, apolipoprotein E; AUC, area under the ROC curve; PET, positron emission tomography; ROC, receiver operating characteristic.

FIGURE 3

Concordance plots and distributions of model‐derived probabilities between Aβ PET(−) and Aβ PET(+) groups. A, Concordance plot between ¹⁸F‐Florbetapir SUVR and derived probabilities of the regression model including Aβ42/Aβ40, age, number of APOE ε4 alleles, and recruitment site. Dashed lines represent cutoffs for SUVR (vertical) or probability at Youden maximum index (horizontal). B, Distribution of predicted probabilities between Aβ PET(−) and Aβ PET(+) groups for the model described above (A). Group comparisons were carried out using a Mann–Whitney test (^*** P < 0.001). Values exceeding median value ± 1.5 x interquartile range (IQR) are displayed as outliers. C, Concordance plot between ¹⁸F‐Florbetapir SUVR and derived probabilities of the regression model including Aβ42/Aβ40, age, and number of APOE ε4 alleles. Dashed lines represent cutoffs as in (A). D, Distribution of predicted probabilities between Aβ PET(−) and Aβ PET(+) groups for the model described above (C). Group comparisons were carried out using a Mann–Whitney test (^*** P < 0.001). Values exceeding median value ± 1.5 x IQR are displayed as outliers. Aβ, amyloid beta; APOE, apolipoprotein E; PET, positron emission tomography; SUVR, standardized uptake value ratio.

FIGURE 4

Heat maps showing predicted probability of being Aβ PET(+) according to age (horizontal axis) and Aβ42/Aβ40 plasma values (vertical axis) for those subjects bearing at least one APOE ε4 allele (left) and APOE ε4 non‐carriers (right). Predicted probabilities are displayed in percentages. 95% confidence intervals are indicated between brackets. Aβ, amyloid beta; APOE, apolipoprotein E; PET, positron emission tomography.