Reliability of chronic allograft nephropathy diagnosis in sequential protocol biopsies.

Background: Chronic allograft nephropathy (CAN) progresses rapidly during the first few months and slowly thereafter. Although the presence of CAN in protocol renal biopsies is a predictor of outcome, the reliability of this diagnosis according to Banff criteria has not been characterized.
Methods: Renal lesions were evaluated according to the Banff criteria in sequential protocol biopsies performed at 4 and 14 months in 310 biopsies obtained from 155 patients.
Results: CAN progressed from 40 to 53% (P=0.001) while serum creatinine remained stable (146 +/- 44 vs. 147 +/- 48 micromol/L, P=NS). Graft survival in patients with and without CAN in the first biopsy was 74 versus 91% (P < 0.05), and in the second biopsy 75 versus 94% (P < 0.05). In 54 patients (35%) no CAN was present in both biopsies, 39 (25%) showed progression to CAN, 19 (12%) showed regression of CAN, and 43 (28%) showed CAN in both biopsies. Graft survival was: 100%, 81.6%, 82.6% and 69.4%, respectively (P < 0.01). Assuming that CAN does not regress and sampling error is normally distributed, we estimated that 25% of biopsies cannot be properly classified.
Conclusions: The increase in the incidence of CAN between the 4th and 14th month is lower than the proportion of misclassified biopsies. Thus, monitoring the progression of CAN by means of two sequential biopsies at 4 and 14 months is inaccurate. We suggest that progression of scarring be monitored by means of a donor and a protocol biopsy performed during the first year evaluated with a quantitative approach.