Computerized image analysis vs semiquantitative scoring in evaluation of kidney allograft fibrosis and prognosis.

Chronic morphological changes in the kidney allograft predict long-term graft function, but there are few studies comparing different methods in assessing chronic lesions. In the present study, we evaluated allograft cortical interstitial fibrosis, and compared semiquantitative assessment with computerized image analysis of Sirius red-stained collagen in prediction of graft prognosis. Sections were obtained from a series of 1-year protocol living donor kidney graft biopsies (n = 33) and their corresponding baseline specimens (n = 32). At light microscopy, the biopsies were scored for interstitial fibrosis as a percentage of involved tubulointerstitium according to the Banff schema. Quantitation of cortical fractional interstitial fibrosis volume (Vint) was performed with computerized image analysis on coded sections stained with Sirius red. The results were correlated with kidney function at 8-10 years after transplantation, and with late graft loss. There was a significant correlation between the semiquantitative and quantitative methods for measuring cortical interstitial fibrosis in all the biopsies (n = 65, percentage area vs Vint: R = 0.439, P = 0.0003). The correlation further improved when analysing the baseline specimens separately (n = 32, R = 0.704, P<0.0001) and was still significant, but less precise for the 1-year biopsies (n = 33, R = 0.384, P = 0.0274). One-year semiquantitative fibrosis (percentage area) was correlated to serum creatinine at 8-10 years (P = 0.010) and to late graft loss (P = 0.0445). The 1-year Vint values for interstitial fibrosis showed a similar trend but did not reach statistical significance in prediction of long-term graft function. Image analysis quantitation of interstitial collagen with Sirius red corresponded well to light microscopic semiquantitative assessment of interstitial fibrosis. In prediction of long-term graft function, the semiquantitative method was superior, indicating that accumulation of matrix molecules other than fibrillary collagens, oedema and inflammation are also important in graft prognosis.