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

Abstract

Aim

The aim of this study was to clarify the quantitative and qualitative differences of tumour budding identification between haematoxylin and eosin (H&E) and immunohistochemical 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 and immunohistochemical staining for cytokeratin. Tumour budding counts based on cytokeratin‐stained slides were strongly correlated with those based on H&E‐stained slides, having higher detection and reproducibility. Based on receiver operating characteristic (ROC) 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 immunohistochemical 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 high‐ and low‐budding, respectively), and similar results were observed for immunohistochemical staining (79.9% and 91.7%, respectively). The Akaike Information Criterion value for RFS in H&E staining was favourable compared with that in immunohistochemical staining.

Conclusions

Tumour budding counts based on cytokeratin‐stained slides demonstrated higher detection and better reproducibility, but did not present as satisfactory clinical impacts as those based on H&E.

This article is protected by copyright. All rights reserved.

http://bit.ly/2sSAcuA