Reproducibility of the Banff schema in reporting protocol biopsies of stable renal allografts

Background. There is evidence that biopsy of stable renal allografts may be of value in predicting chronic allograft nephropathy, the main cause of graft loss. However, the reproducibility of such histological evaluation has not been tested in this setting. We tested the reproducibility of the Banff schema for this purpose. Methods. We rated acute and chronic changes in 184 protocol biopsies. Individual pathologists at two different Canadian transplant centres reported independently. Results. There was agreement in 73.53, 42.86, and 77.08% of cases in assigning a diagnosis of acute rejection. borderline changes (as defined in the schema), and no acute rejection, respectively. Applying kappa statistics, there was very good agreement in making the diagnosis of acute rejection vs no acute rejection (kappa 0.77). There was good inter-observer agreement in scoring glomerulitis, intimal arteritis, interstitial infiltrates, tubulitis, and arteriolar hyalinosis. Rating chronic changes also gave good interobserver agreement (kappa=0.53, 0.65, and 0.62, respectively, for mild, moderate, and severe chronic allograft nephropathy). Agreement on transplant glomerulopathy was, however, poor. Conclusions. We conclude that the Banff classification provides a reproducible method for the histological assessment of protocol renal allograft biopsies in stable grafts. Such biopsies may be valuable in detecting subclinical rejection and early chronic allograft nephropathy and may also be used as surrogate end-points in the evaluation of therapy to prevent the latter.