Amyloid status prediction by combining plasma p‐tau181 and NfL cut‐offs

Abstract

AbstractBackgroundPlasma biomarkers have emerged as a promising tool for screening Alzheimer's disease(AD). However, no externally defined cut‐off values for any plasma biomarkers yet to exist to predict amyloid β (Aβ) status.MethodTaking advantage of the different plasma p‐tau181 and NfL increase profiles, we used the two plasma biomarkers to establish preliminary cut‐off values (two cut‐offs for p‐tau181, one for NfL) for screening Aβ positivity. Participants with a clinical diagnosis of AD(N=20), mild cognitive impairment(MCI)(N=12), other neurodegenerative disorders(N=19), psychiatric disorders(N=3 with depression, N= 1 with delusional disorder), and healthy controls(N=37) were recruited from the Memory Clinic at Keio University Hospital. Static 18F‐Florbetaben (FBB) PET was conducted on all the participants. A trained neuroradiologist visually assessed the acquired images to determine Aβ positivity. Each participant's p‐tau181 value was plotted against NfL values. After visualizing their distribution stratified by the Aβ status, we conducted an exhaustive search to predict Aβ positivity. We first created a ROC curve for p‐tau181 alone and then made a 'quasi‐ROC' curve by combining two p‐tau181 cut‐offs(i.e., the higher and lower cut‐offs) and one NfL cut‐off. The optimal cut‐off values for each ROC curve were determined by the Youden Index.ResultOf the 92 participants, 30(32.6%) were Aβ‐positive(Aβ+). Mean ages were 68.7 years(SD8.2) for Aβ‐ and 72.4(SD9.3) for Aβ+ group (p=0.02). Mean p‐tau181 values were 1.72pg/ml (SD0.75) for Aβ‐ and 3.58pg/ml (SD1.09) for Aβ+ group (p<0.001), whereas mean NfL values were 24.47pg/ml(SD13.77) vs 27.24(SD9.96) (p=0.01). The ROC curves yielded each found the optimal cut‐off values as follows: for plasma p‐tau181 alone, ptau‐181 value, true positive rate(tpr) and false positive rate(fpr) were 5.8 pg/ml, 0.767 and 0.065, respectively. For a combination of higher and lower p‐tau181 cut‐offs and one NfL cut‐off, each biomarker value and the corresponding tpr and fpr were 3.0pg/ml, 2.0/g/ml, 35pg/ml, 0.933, and 0.177, respectively.ConclusionOur results suggest combining two cut‐off values for plasma p‐tau181 and one cut‐off for NfL may yield an excellent discriminant capacity for Aβ positivity. Further research is needed to validate the result.