Evaluation of pancreas transplant needle biopsy: reproducibility and revision of histologic grading system

Tissue samples for the diagnosis of pancreatic allograft rejection are now obtained routinely through the application of the percutaneous needle biopsy technique. The availability of biopsy material (89% adequate for diagnosis in our setting) presents a challenge for pathologists who are asked to provide a fast and accurate diagnosis of rejection and its severity, while at the same time being able to differentiate rejection from other causes of graft dysfunction.

To differentiate rejection from other pathologic processes, 26 histologic features were assessed in 92 biopsies performed for confirmation of clinical diagnosis of rejection and the results were compared with 31 protocol biopsies, 12 allograft pancreatectomies with non-rejection pathology, and 30 native pancreas resections with various disease processes.

Based on these comparisons, a constellation of findings relating to the vascular, septal, and acinar inflammation was identified for the diagnosis of rejection. Application of these features led us to revise our scheme for grading rejection (ranging from 0-normal to V-severe rejection) to include the categories of "inflammation of undetermined significance" and "minimal rejection." The scheme was used by five pathologist to grade 20 biopsies independently of any clinical data and the interobserver level of agreement was highly significant (kappa=0.83, P<0.0001). This grading scheme was applied blindly to all (183) biopsies from 77 patients with 6-52 months of follow-up. The correlation of the highest degree of rejection on each patient and ultimate graft loss (0% for grades 0-I, 11.5% for grade II, 17.3% for grade III, 37.5% for grade IV, and 100% for grade V) was highly statistically significant (P<0.002). The fraction of grafts lost due to pure immunologic causes increased proportionally to the grade of rejection (0, 50, 66, and 100% for grades II, III, IV, and V, respectively).

This study provides strong support for the proposed pancreas rejection grading scheme and confirms its potential for practical use.