P279 - ACCURATE DISCRIMINATION OF BRAIN AMYLOID STATUS IN THE MULTI-CENTRIC A4 STUDY BY PLASMA AΒ42/AΒ40 MEASURED WITH A NOVEL HPLC-MS/MS METHOD. (ID 875)
Abstract
Aims
To retrospectively explore the ability of Aβ42/Aβ40 plasma ratio, as measured with a novel antibody-free HPLC-MS/MS method (ABtest-MS, Araclon Biotech), to predict the brain amyloid status in a subset of cognitively unimpaired individuals (CU) from the screening visit of the A4 study.
Methods
Aβ40 and Aβ42 plasma levels from 731 CU participants were quantitated with ABtest-MS. Plasma samples were obtained in 59 recruitment sites across USA, Canada and Australia. Logistic regression modeling and ROC curve analysis were carried out to assess the ability of Aβ42/Aβ40, as measured with ABtest-MS, to predict amyloid brain status. Amyloid positivity for 18F-Florbetapir PET brain imaging was established for standardized uptake value ratios (SUVR) ≥ 1.15.
Results
Plasma Aβ42/Aβ40 values were significantly lower in the Aβ-PET positive than in the Aβ-PET negative group (p=3.7·10-35). Log(Aβ42/Aβ40) showed a significant negative correlation with 18F-Florbetapir-PET SUVR values (rho=-0.440; p=2.9·10-36).
Plasma Aβ42/Aβ40 discriminated Aβ-PET positivity with an AUC [95% CI] of 0.78 [0.75-0.86]. After inclusion of recruitment site as a covariate in the regression model, AUC increased up to 0.86 [0.83-0.89].
Noteworthy, a full regression model including ratio, recruitment site, age, gender and number of APOE4 alleles , yielded an AUC of 0.88 [0.86-0.91] (accuracy 81.3%).
Conclusions
ABtest-MS accurately identifies amyloid brain deposition in this subset of CU individuals from the A4 study. These results suggest that despite the standardization of the pre-analytical variables, the effect of the recruitment site is not negligible. Therefore, an extensively validated, robust and centralized sample analysis would be highly desirable to minimize this additional variability.