Tumor budding in colorectal carcinoma: An institutional interobserver reliability and prognostic study of colorectal adenocarcinoma cases.

BACKGROUND: Colorectal carcinomas are one of the most commonly diagnosed malignancies. There are many prognostic factors relating to clinical course and disease progression, including tumor stage, metastasis, and tumor budding. In 2016, the International Tumor Budding Consensus Conference (ITBCC) created a system to uniformly assess tumor budding. This system includes a 3-tier system for the grading of tumor budding. In the past, there lacked uniform consensus, however the general grading practice was based on a 2-tiered system. Given that tumor budding is considered to have prognostic value, the accuracy and reproducibility of its assessment is vital. Our study aims to look at interobserver agreement in the scoring of tumor budding.
DESIGN: A total of 233 cases of colorectal carcinoma diagnosed in our health system were retrospectively analyzed and routine H&E stained slides of these cases were collected. A representative slide for tumor budding was selected per case. Four investigators with different levels of experience and expertise evaluated the selected slide of each case for tumor budding. Scoring was based on the ITBCC protocol. Clinico-pathological data was collected for each case and analyzed with tumor budding scores. Tumor budding scores per individual investigator and consensus tumor budding score were compared to patient and tumor characteristics including patient survival, tumor grade, tumor stage, and lymph node status.
RESULTS: Inter-observer agreement was calculated using Gwet's Agreement Coefficient (AC1) and associated 95% confidence intervals was used to compare the ratings made by 4 pathologists. Overall, there was variation among pathologists in tumor budding score (Gwet's agreement coefficient = 0.25 and 0.326 for 3-tier and 2-tier grading system, respectively). Results show higher reliability with the 2-tier system compared to the 3-tier system. Tumor stage was significantly associated with budding score for all individual investigators and the consensus value (p value < 0.001).
CONCLUSION: There is low inter-observer agreement in the assessment of tumor budding in colorectal carcinoma. This suggests that it is difficult to uniformly grade tumor budding and that our classification system needs improvement. We found that the older 2-tier system (Hase et al.) results in slightly higher inter-observer agreement than the recently proposed 3-tier grading system (ITBCC, 2016), though both systems lead to suboptimal agreement. Worth noting is that observers with subspecialty GI training and more work experience had higher inter-observer agreement. Our results showed that subspecialty training tends to increase agreement more than overall work experience. In addition, our exploratory results showed that there is an association of tumor budding score to tumor stage. While increasing refinement in classification, the 3-tiered system resulted in decreased agreement in tumor budding assessment. Clearly, there is more work to be done in the identification and quantification of tumor buds.