Premium
[P1–251]: CLINICAL PROFILE OF PATIENTS WITH NORMAL CSF Aβ42 AND Aβ42/Aβ40 RATIO, YET ABNORMAL CSF 181P‐TAU: A SNAP STORY
Author(s) -
PoucletCourtemanche Hélène,
Nguyen TriBao,
Schraen Susanna,
Pasquier Florence,
Dumurgier Julien,
Paquet Claire,
Lebouvier Thibaud
Publication year - 2017
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2017.06.267
Subject(s) - memory clinic , concomitant , cognition , cerebrospinal fluid , medicine , c9orf72 , psychology , cognitive impairment , dementia , pathology , gastroenterology , frontotemporal dementia , psychiatry , disease
Background: Amyloid beta 1-42 (Ab42), total Tau (Tau) and phosphorylated tau (Ptau) are abnormal in Alzheimer’s disease (AD) and are widely used to diagnose AD. Our aim was to develop an easily clinical applicable decision tree for the diagnosis of AD that combines these markers using Classification and Regression Tree (CART) analysis. Second, we studied if the decision tree could improve predicting progression to AD in subjects with mild cognitive impairment (MCI) over conventional cut-points. Methods:We included patients from the memory clinic based Amsterdam Dementia Cohort with defined diagnosis of AD (n1⁄41004), subjective cognitive decline (SCD n1⁄4 442) and MCI (n1⁄4363). A decision tree was modeled in a randomly selected training set (AD n1⁄4500; SCD n1⁄4223) and validated in an independent test set (AD n1⁄4504; SCD n1⁄4219). Predictor variables entered in the CART analysis were Ab42, Tau, Ptau, age, gender and ApoE status. The conventional cutpoints, to which the performance of the tree was compared, were 640 pg/ml for Ab42 and 375 pg/ml for Tau. Results: The CART analysis selected Ab42 and Tau, but not Ptau, age, gender or ApoE status for the decision tree. Patients were classified as AD if Ab42 concentrations were <523 pg/ml irrespective of Tau values or when abnormal Ab42 concentrations were between 523 and 748 pg/ml and Tau values 375 pg/ml (Figure 1). The decision tree lead to an accuracy of 87% in the validation set with a sensitivity of 88% [85-91], a specificity of 84% [7987], positive predictive value (PPV) of 93% [90-95] and negative predictive value (NPV) of 76% [71-80]. The accuracy applying the conventional method was 77% (sensitivity 71% [67-75], specificity 93% [88-96], PPV 96% [93-97] and NPV 58% [5561]) . The accuracy in predicting conversion to AD in MCI patients (average follow-up 2.5 6 1.5) was comparable to the conventional method (75%). Conclusions: The CART analysis yielded an easily clinical applicable decision tree with two cutoff points. One with low Ab42 regardless of Tau levels, and