Background and aim: The outcome for some colorectal cancer patients correlates poorly with classical prognostic factors, like tumor stage. Tumor budding (TB) is a promising and intensely studied new prognostic factor. We aimed to evaluate the reliability of bud counting on Hematoxylin-Eosin (HE)-stained and immunohistochemically (IHC)-stained scanned slides.
Materials and methods: We evaluated 21 cases of robotic surgery colorectal cancer specimens that were submitted to the Department of Pathology, Emergency County Hospital, Timişoara, Romania. TB was assessed by one experienced (R3) and two junior pathologists (R1, R2), in 10 circular areas at 20× (0.785 mm²) on scanned HE-stained and IHC-stained [cytokeratin (CK) AE1∕AE3] slides. Interobserver agreement (Cohen's kappa) and intraclass correlation coefficient (ICC) were calculated.
Results: In the case of HE-stained slides, the inter-item correlation matrix showed values between 0.632 and 0.84, while the ICC on average measures for consistency showed very good correlation [ICC: 0.887, 95% confidence interval (CI): 0.765-0.95)]. The inter-item correlation matrix for IHC-stained slides comprised values between 0.864 and 0.921, while the ICC for average measures for consistency yielded an excellent value (ICC: 0.95, 95% CI: 0.896-0.978). We identified higher values for budding scores on IHC-stained slides, in comparison to the HE-stained slides: in 19∕21 cases for R1 (average increase of 234.85%), 16∕21 cases for R2 (average increase of 114.14%), and 20∕21 cases for R3 (average increase of 66.92%).
Conclusions: We consider the method of buds counting in 10 microscopic fields on scanned slides to be reliable and valuable. TB counts are higher on IHC-stained slides and associate a better interobserver agreement.