S210 Agreement and Reproducibility in the Rediagnosis of Sessile Polyps

Abstract

Introduction: Most serrated polyps (SPs) fall into two categories: hyperplastic polyps (HPs) and sessile serrated polyps (SSPs), the former lacking precancerous potential while the latter is precancerous. These subtypes can be difficult to diagnose histologically. Inaccuracy in SP diagnosis can lead to incorrect colonoscopy surveillance recommendations that can increase the risk of colorectal cancer. In this study, we present GI pathologists with previously diagnosed SPs and aim to quantify the frequency of diagnostic change of SPs and diagnostic agreement. Methods: Polyp pathology data was utilized from a colonoscopy quality database from colonoscopies performed from 2012-2020. 167 serrated polyps (either HP or SSP) were selected for analysis balanced on previous histology, size and location, excluding those found in patients with known polyposis syndromes. Polyp specimens underwent re-diagnosis from a five member GI pathology team, their experience ranging from fellow to experienced attending. Each pathologist reviewed pathological specimens independently without knowledge of the original diagnosis and made their diagnosis of SSPs following WHO definition. Effort was made to keep diagnostic conditions like real-world practice. Statistical analysis was performed in SPSS. Kappa analysis was performed for inter-observer agreement. Kappa values were grouped as poor (< 0.2), fair (0.21-0.40), moderate (0.41-0.60), good (0.61-0.80), and perfect ( >0.80). Means were compared using independent samples t-test. A p value less than 0.05 was considered significant. (Table) Results: On average, the five GI pathologists re-diagnosed 163 SPs matching the previous diagnosis about 74.1% of all polyps. The mean kappa value for variability in SSP diagnoses between original diagnosis and each GI pathologist was 0.497. Kappa value for polyps less than 1 cm was 0.313 versus 0.692 for polyps greater than 1 cm (p=0.006). The mean kappa value between all pathologists in their re-diagnosis was 0.669. There was no significant difference in kappa when stratified by proximal versus distal colon. Conclusion: Re-diagnosis of SSPs resulted in only moderate level agreement between GI pathologists and the previous diagnosis. Interestingly, inter-observer agreement among pathologists was at a good level, and there was increased agreement for larger polyps, but not for location. While moderate to good level of agreement is above what is reported in literature, our study highlights the need for improved diagnostic reproducibility of SSPs. Table 1. - Bolded values being statistically significant with p-value < 0.05 Number of polyps matching previous diagnosis/Number of polyps re-diagnosed Kappa Kappa for polyps less than 1 cm Kappa for polyps greater than 1 cm Kappa for polyps in proximal colon Kappa for polyps in distal colon Pathologist A vs original diagnosis 84/102 (82.4%) 0.661 0.471 0.835 0.768 0.555 Pathologist B vs original diagnosis 57/78 (73.1%) 0.466 0.166 0.589 0.386 0.625 Pathologist C vs original diagnosis 122/163 (74.8%) 0.518 0.287 0.725 0.563 0.450 Pathologist D vs original diagnosis 125/162 (77.2%) 0.539 0.326 0.579 0.600 0.449 Pathologist E vs original diagnosis 12/19 (63.2%) 0.300 -0.176 0.229 0.500 0.167 Mean 0.3125 0.682 0.579 0.520 Pathologist A vs B 0.744 0.602 0.885 0.615 0.903 Pathologist A vs C 0.696 0.604 0.850 0.694 0.695 Pathologist A vs D 0.785 0.723 0.842 0.850 0.724 Pathologist A vs E 0.635 0.556 0.514 0.556 0.690 Pathologist B vs C 0.604 0.572 0.514 0.661 0.489 Pathologist B vs D 0.651 0.517 0.771 0.664 0.623 Pathologist B vs E - - - - - Pathologist C vs D 0.651 0.592 0.663 0.715 0.592 Pathologist C vs E 0.543 0.556 0.341 0.556 0.535 Pathologist D vs E 0.700 0.737 0.538 0.500 0.833 Mean 0.685 0.606 0.697 0.700 0.676