A33: Cognitive Performance Scores for the Pediatric Automated Neuropsychological Assessment Metrics in Childhood-Onset Systemic Lupus Erythematosus.

Background/Purpose: Children with SLE (cSLE) can experience neuropsychiatric SLE (NPSLE), commonly manifesting as neurocognitive dysfunction which can interfere with normal development. Formal neurocognitive testing (FNCT) is the most accepted test for diagnosing neurocognitive deficits (NCD). Access to it is limited and costly, and it is time-consuming. The Pediatric Automated Neurophysiological Assessments Metrics (PedANAM) is a computerized battery of 10 subtests measuring various aspects of cognitive ability. Concurrent validity of PedANAM test scores with FNCT has been demonstrated. However, the PedANAM generates several measures of accuracy (% of correct responses), processing speed and efficiency, and it is unclear how they can be used in a clinical setting. The usefulness of the PedANAM as a screen for NCD would be enhanced by the development of a PedANAM Cognitive Performance Score (PedANAM-CPS) to represent aspects of PedANAM task performance that are sensitive to NCD in cSLE. The purpose of this study was to develop a PedANAM-CPS for use in cSLE, using statistical methods. Methods: cSLE patients (pts) and age plus sex-matched healthy controls enrolled in a study of cognitive functioning and neuroimaging were studied. At the time of enrollment (visit 1-V1) and 18 months later (visit 2-V2), subjects completed the PedANAM and FNCT. Three candidate PedANAM-CPS measurement approaches were explored via 3 statistical methods: 1) Simple mean-of all subtest accuracy scores; 2) Logit score-via a logistic regression model; 3) PCA score-using a Principal Component Analysis (PCA) method. The latter 2 methods assigned in a different way a statistical weight to each subtest accuracy score. Fixed effect models were used to compare performance scores between groups. Receiver operating characteristic (ROC) curves were used to assess the accuracy of the CPS as predictors of NCD as determined by FNCT. Results: 77 children (female = 68%) were evaluated at V1; Nine cSLE pts with NCD, 31 with cSLE and no NCD, and 37 control with no NCD as per FNCT. At V1, age (values are mean ± standard deviation) of children was 13.6 ± 2.4 years. For cSLE pts, disease activity (SLEDAI) was 4.9 ± 4.4, and 77.5% were on oral prednisone (19.8 ± 17.4 mg). Table summarizes the PedANAM-CPS for all methods at V1. The Logit score best discriminated the groups, especially contrasting the NCD group against the other groups. The Logit and PCA scores showed g 82% area under the ROC curve during the validation stage using V2 data. [ABSTRACT FROM AUTHOR]