Quantification of Interstitial Fibrosis by Image Analysis on Routine Renal Biopsy in Patients Receiving Cyclosporine

Abstract

Renal interstitial fibrosis (IF) is the main histopathological feature of chronic allograft injury. IF is currently assessed by semiquantitative analysis, but automatic color image analysis may be more reliable and reproducible. We performed a retrospective analysis to calculate IF on routine renal biopsies performed at 1 year posttransplant. Data were obtained from MO2ART, a prospective multicenter trial in which cyclosporine A dose was adjusted based on C2 level. One-year routine biopsies were assessed from 26 patients from two centers. For each biopsy, a section was analyzed by a program of color segmentation imaging, which automatically extracts green color areas characteristic of IF. Results were expressed as percentage of IF and grade (grade 1: <25%, grade 2: 25-50%, and grade 3: >50%). Mean IF score was 0.35+/-0.04. Quantitative IF grade 1 was observed in 9 biopsies (34.6%), grade 2 in 12 (46.1%), and grade 3 in 5 (19.2%). Diabetes and cytomegalovirus infection were significantly associated with a higher percentage of IF. There was no correlation between the group of randomization and IF. We found a statistical significant correlation between Banff 05 chronic lesions classification and the IF index (P<0.02). Repeated analysis of variance demonstrated an association between high grade of automated IF and a worsening of creatinine clearance (Modification of Diet in Renal Disease) between 1 and 3 years. Automatic quantification of IF on routine renal biopsies at one year posttransplant is predictive of long-term allograft function and may assist early diagnosis of the interstitial lesions of chronic allograft injury.