Differential clinical impacts of tumour budding evaluated by the use of immunohistochemical and haematoxylin and eosin staining in stage II colorectal cancer

AIMS: The aim of this study was to clarify the quantitative and qualitative differences in tumour budding identification between haematoxylin and eosin (H&E) staining and immunohistochemical (IHC) staining for cytokeratin, and to estimate the respective clinical impacts in stage II colorectal cancer. METHODS AND RESULTS: We retrospectively examined 314 surgically resected cases of stage II colorectal cancer, and assessed tumour budding on serial section slides with H&E staining and IHC staining for cytokeratin. Tumour budding counts based on cytokeratin-stained slides were strongly correlated with those based on H&E-stained slides, and had higher detection and reproducibility. On the basis of receiver operating characteristic analyses, the optimal cut-off values of budding counts for relapse-free survival (RFS) were 7 and 16 in a ×200 microscopic field with H&E and IHC staining, respectively. With these cut-off values, tumour budding based on H&E staining had a significant correlation with RFS (80.3% and 93.2% of 5-year RFS in the high-budding group and the low-budding group, respectively), and similar results were observed for IHC staining (79.9% and 91.7%, respectively). The Akaike Information Criterion value for RFS with H&E staining was favourable as compared with that with IHC staining. CONCLUSIONS: Tumour budding counts based on cytokeratin-stained slides showed higher detection and better reproducibility, but did not have as satisfactory clinical impacts as those based on H&E staining.