L M Wang1, R Guy2, E Fryer1, C Kartsonaki3, P Gill1, C Hughes1, A Szuts1, R Perera1, R Chetty1, N Mortensen2. 1. Department of Cellular Pathology, John Radcliffe Hospital, University of Oxford, Headington, Oxford, UK. 2. Department of Colorectal Surgery, Churchill Hospital, University of Oxford, Headington, Oxford, UK. 3. CR-UK/MRC Oxford Institute for Radiation Oncology, Department of Oncology, John Radcliffe Hospital, University of Oxford, Headington, Oxford, UK.
Abstract
AIM: Early pT1 polyp colorectal cancers (CRCs) present challenges for accurate pathology substaging. Haggitt and Kikuchi stages depend on polyp morphology and are often difficult to apply due to suboptimal orientation or fragmentation, or absence of the muscularis propria in polypectomy or submucosal resection specimens. European guidelines for quality assurance suggest using Ueno's more objective approach, using depth and width measurements beyond muscularis mucosae. We have investigated interobserver variation using Ueno's approach. METHOD: Ten consecutive pT1 polyp CRCs were identified and the slides assessed by six gastrointestinal pathologists for depth and width of invasion. A further 60 polyps were studied by a group of specialist and general pathologists. Agreement was assessed by analysis of variance. A polyp CRC is classified as high risk if it has a depth ≥ 2000 μm or a width ≥ 4000 μm and low risk with a depth < 2000 μm or a width < 4000 μm. Concordance for the dichotomized values was assessed using the kappa statistic. RESULTS: The intraclass correlation coefficient (ICC) for depth was 0.83 and for width 0.56 in the 10-polyp group. The ICC for the 60-polyp CRCs was 0.67 for depth and 0.37 for width. In both groups, when polyp CRCs are divided into high- and low-risk categories based on depth, there was substantial and moderate agreement (κ = 0.80 and 0.47) but only fair agreement when based on width (κ = 0.34 and 0.35). CONCLUSION: Ueno's method has the advantage of being independent of polyp morphology. Our study shows better concordance for depth measurement and reproducibility in nonfragmented specimens, with poorer agreement when based on width. Colorectal Disease
AIM: Early pT1polyp colorectal cancers (CRCs) present challenges for accurate pathology substaging. Haggitt and Kikuchi stages depend on polyp morphology and are often difficult to apply due to suboptimal orientation or fragmentation, or absence of the muscularis propria in polypectomy or submucosal resection specimens. European guidelines for quality assurance suggest using Ueno's more objective approach, using depth and width measurements beyond muscularis mucosae. We have investigated interobserver variation using Ueno's approach. METHOD: Ten consecutive pT1 polyp CRCs were identified and the slides assessed by six gastrointestinal pathologists for depth and width of invasion. A further 60 polyps were studied by a group of specialist and general pathologists. Agreement was assessed by analysis of variance. A polyp CRC is classified as high risk if it has a depth ≥ 2000 μm or a width ≥ 4000 μm and low risk with a depth < 2000 μm or a width < 4000 μm. Concordance for the dichotomized values was assessed using the kappa statistic. RESULTS: The intraclass correlation coefficient (ICC) for depth was 0.83 and for width 0.56 in the 10-polyp group. The ICC for the 60-polyp CRCs was 0.67 for depth and 0.37 for width. In both groups, when polyp CRCs are divided into high- and low-risk categories based on depth, there was substantial and moderate agreement (κ = 0.80 and 0.47) but only fair agreement when based on width (κ = 0.34 and 0.35). CONCLUSION: Ueno's method has the advantage of being independent of polyp morphology. Our study shows better concordance for depth measurement and reproducibility in nonfragmented specimens, with poorer agreement when based on width. Colorectal Disease