I find it surprising that histological grading for two of the most common adult malignancies--colon and lung--lacks any kind of consensus criteria for grading elements let alone clinical validation as a prognostic marker. I have previously blogged on proposals for grading for lung adenocarcinoma but the February 2012 issue of American Journal of Surgical Pathology has an intriguing proposal for criteria for grading of colorectal cancer by Hideki Ueno et al. (abstract).
I am most familiar with Dr. Ueno's previous work on tumor budding in colon cancer (you should be too!) and this work builds on and expands on this work. In contrast to tumor budding, defined as invasive single cells and/or clusters with <5 cells, this work describes and defines the significance of "poorly differentiated clusters" for grading. Poorly differentiated clusters are defined by these authors as "cancer clusters in the stroma composed of >5 cancer cells and lacking a gland-like structure." These clusters are identified at low power and then counted per 20X (1-mm diameter) field. The color illustrations included in the text are particularly well-selected and beautifully reproduced. Grading was also assessed independently by assessing quantitatively the area of the tumor involved by poorly differentiated clusters and by tumor budding.
Although this is a single institution study, it included 500 consecutive patients with stage 2 or stage 3 CRC who underwent curative resection and had follow-up on 97% (!) of these patients (483 patients) for a median of 68 months for the 378 surviving patients.
Three things stand out from the results. First, there was excellent separation in 5-Year DFS using quantity of poorly differentiated clusters as the criterion for grading: grade 1--96%, grade 2--85%, grade 3--59%. Second, tumor grading based on poorly differentiated clusters, T and N stages, and venous invasion were independent prognostic parameters related to survival on mutlivariate analysis. FInally, the authors measured intra- and inter-observer variability in tumorg grading based on poorly differentiated clusters.
This is an interesting and status quo-challenging article that should provoke some good discussion amongst GI specialist and general pathologists.