P284. Agreement of central readers in the evaluation of histopathological disease activity in ulcerative colitis
M.H. Mosli1,2, B.G. Feagan2,3, G. Zou2,3, K. Geboes4, C. Langner5, M. Peterson6, R. Pai7, W.J. Sandborn2,8, G. D'Haens2,9, M.K. Vandervoort2, C. Behling10, K. Kaplan11, L.M. Shackelton2, J. Jones2, R. Khanna2, B.G. Levesque2,8, 1King Abdulaziz University, Department of Medicine, Jeddah, Saudi Arabia, 2Robarts Clinical Trials Inc., Robarts Research Institute, Western University, London, Ontario, Canada, 3Western University, Department of Epidemiology and Biostatistics, London Ontario, Canada, 4University Hospital of KU Leuven, Department of Pathology, Leuven, Belgium, 5Medical University of Graz, Institute of Pathology, Graz, Austria, 6Western Washington Pathology and Multicare Health System, Pathology, Tacoma, United States, 7Cleveland Clinic, Department of Anatomic Pathology, Cleveland Ohio, United States, 8University of California, San Diego, Division of Gastroenterology, La Jolla, United States, 9University of Amsterdam, Academic Medical Center, Amsterdam, Netherlands, 10Pacific Rim Pathology Medical Corporation, Pathology, San Diego California, United States, 11Carolinas Medical Center, Medicine, Charlotte North Carolina, United States
Histologic assessment is an evolving outcome measure in ulcerative colitis (UC). Many histological indices, including the Geboes score (GS; Figure 1, top panel) and modified Riley Score (MRS; Figure 1, bottom panel) have been used to grade disease activity and served as endpoints for clinical trials.
The operating properties of these indices, including agreement have not been fully evaluated. We assessed inter- and intra-rater agreement of central readers for histopathological evaluation of UC disease activity.
Five central readers individually evaluated 50 slides of colonic biopsies taken from patients with UC on three separate occasions at least two weeks apart using the GS, MRS and a global rating of severity based on a 100 mm visual analogue scale (VAS). Inter- and intra-rater agreement was measured using intraclass correlation coefficients (ICC) for each grading system and for components of acute and chronic inflammation.
Intra-rater ICCs (95% confidence intervals) for the total GS, MRS and VAS scores were 0.83 (0.78, 0.87), 0.72 (0.65, 0.79) and 0.79 (0.74, 0.85), respectively. Corresponding inter-rater agreement ICCs were 0.57 (0.45, 0.68), 0.49 (0.38, 0.61) and 0.61 (0.51, 0.72) (Table 1).
The correlation between each central reader's VAS with GS and MRS descriptors were 0.60 (0.44, 0.76) and 0.63 (0.47, 0.77), respectively. The items with lowest ICC (highest disagreement) were granuloma (0.01), patchiness (0.20), lamina propria fibrosis (0.21), lamina propria eosinophilia (0.26), crypt abscesses (0.32), crypt destruction (0.34), surface epithelium integrity (0.35), and lamina propria neutrophils (0.37).
|Inter-rater agreement||Intra-rater agreement|
|Geboes-Structural||0.70 (0.60–0.79)||0.80 (0.74–0.86)|
|Geboes-Chronic inflammatory infiltrate||0.64 (0.54–0.74)||0.81 (0.75–0.86)|
|Geboes-Lamina propria eosinophils||0.26 (0.18–0.37)||0.59 (0.52–0.66)|
|Geboes & Modified Riley-Lamina propria neutrophils||0.37 (0.27–0.49)||0.59 (0.51–0.68)|
|Modified Riley-Neutrophils in epithelium||0.47 (0.37–0.59)||0.71 (0.64–0.78)|
|Geboes-Crypt destruction||0.34 (0.24–0.47)||0.61 (0.54–0.69)|
|Geboes-Erosion or ulceration||0.56 (0.45–0.67)||0.78 (0.73–0.84)|
There is “substantial” to “almost perfect” intra-rater agreement among histopathologists in the assessment of disease activity in UC. Inter-rater agreement was less satisfactory. The results indicate that a central reader is highly reliable for the assessment of UC histologic disease. However, differences in interpretations of the scores among readers requires further study and standardization in order to determine the optimal instrument for use in UC clinical trials. A Delphi process has been initiated to further characterize the most important sources of disagreement.