Skip to main content
. 2021 Apr 8;6(7):e144779. doi: 10.1172/jci.insight.144779

Figure 5. Podocyte morphometric signature identifies ANCA-GN patients.

Figure 5

(A) Podocyte-morphometric analysis (podometrics; per glomerulus) showing a pattern of podocyte loss and hypertrophy in glomeruli classified as “normal” (without lesion) in ANCA-GN patients. (B) Principal component analysis (PCA) using Pareto scaling to rows. Probabilistic PCA was used to calculate principal components, confirming that normal glomeruli in ANCA-GN patients represent a transitional state between normal glomeruli in controls and lesions in ANCA-GN patients. In A and B, n = 722 normal glomeruli for controls and n = 373 glomeruli for ANCA-GN patients; Kruskal-Wallis with Dunn’s multiple-comparisons tests were performed. (C) Receiver operating characteristic (ROC), precision-recall curves, and confusion matrices of patient classification with a logistic regression using leave-one-out cross-validation based on eGFR and on a morphometric signature of podocyte depletion (PD), which combines morphometric data from every available glomerulus per biopsy per patient. (D) ROC, precision-recall curves, and confusion matrices for eGFR and PD signature as classifiers. In C and D, n = 48 patients for controls and n = 58 patients for ANCA-GN with PD signature; n = 62 patients for ANCA-GN with eGFR only. In violin plots, each gray dot represents 1 glomerulus, red lines represent medians, and blue lines represent IQRs. ANCA-GN, antineutrophil cytoplasmic antibody–associated glomerulonephritis; eGFR, estimated glomerular filtration rate; TPR, true positive rate; FPR, false positive rate. ****P < 0.0001, ***P < 0.001, and *P < 0.05.