Skip to main content
. 2024 Feb 16;9(5):1369–1378. doi: 10.1016/j.ekir.2024.02.1397

Figure 2.

Figure 2

Unsupervised analysis of IgAN (primary IgAN and Cirrhosis-IgAN). (a) Partitional clustering using a k-mean algorithm (k = 3) on our 46 cases of cirrhosis-related IgAN mix-up with 83 biopsy-proved primary IgAN nephropathy diagnosed in the same period. The variables included in the analysis were qualitative variables: mesangial hypercellularity (M1), endocapillary hypercellularity (E1), focal segmental glomerulosclerosis (S1), crescent, interstitial fibrosis >25% (T), interstitial infiltrate, presence of ischemic glomeruli, arteriolar hyalinosis (ah), fibro-intimal thickening (FI_thick), thrombotic microangiopathy (TMA), double contour, acute tubular necrosis (ATN), IgA parietal deposit (parietal) and intrinsic AKI occurrence (without confounding factor); along with clinical quantitative variable: uPCR and preexisting eGFR. (b) Visualization of the K-means center values for the 3 identified clusters across the variables, enabling to annotate them as chronic involvement (cluster 1), mild disease (cluster 2), and MPGN-like pattern (cluster 3) respectively. (c) Frequency of each cluster within the primary IgAN and cirrhosis-related IgAN. (d) Frequency of each cluster in cirrhosis-related IgAN according to the history of decompensated cirrhosis or not. IgAN, IgA nephropathy; MPGN, membranoproliferative glomerulonephritis; uPCR, urinary protein-to-creatinine ratio.