Fig. 5

Identification of a pediatric SLE proliferative nephritis predictive signature. a Least absolute shrinkage and selection operator (LASSO) was used on a training set of ELISA and clinical measurements from 41 new-onset pSLE patient samples as a variable selection and linear regression method to construct a predictive model of proliferative nephritis. A separate test set of 23 new-onset pSLE patient samples was used to evaluate the performance of the model, and the nephritis scores obtained for each patient with the model are shown. Unfilled black circles indicate patients who were biopsied and found to be class II, and unfilled black triangles indicate class V nephritis patients. The model was also used to calculate nephritis scores for samples from new-onset pSLE patients who were suspected to have nephritis, but were not confirmed by biopsy (unknown). b Five pSLE patients from the test set were selected for longitudinal analysis. ELISA was used to determine each serum sample’s anti-dsDNA and anti-C1q IgG levels, and a chart review was performed to collect all applicable clinical data. The LASSO model based on the training set was used to calculate nephritis scores for each patient at each time point. Vertical dashed lines indicate when biopsies were performed, and the horizontal dashed line indicates the nephritis score cutoff. ELISA enzyme-linked immunosorbent assay, IgG immunoglobulin G, pSLE pediatric systemic lupus erythematosus, SLE systemic lupus erythematosus