Am I Right? New Tools Will Revolutionize CRC Diagnoses

David J. Kerr, CBE, MD, DSc, FRCP, FMedSci


September 05, 2019

This transcript has been edited for clarity.

Hello. I'm David Kerr, professor of cancer medicine from University of Oxford.

I would like to discuss a study published very recently in the Journal of Clinical Oncology by the excellent SACURA collaborative group in Japan on the prognostic impact of tumor budding in stage II colon cancer.[1]

This well-conducted, well-analyzed large multicenter trial takes the phenomenon of tumor budding (which is recognized morphologically by colorectal cancer pathologists as a sign of the epithelial-mesenchymal transition on the biological level) and looks at the prognostic impact of this pathologic feature. The authors think that tumor budding would be a very useful addition to any of the Union for International Cancer Control (UICC) classifications on prognosis.

To an extent, I agree with their very carefully conducted study. The statistics demonstrate in multivariate modeling that tumor budding does provide prognostic benefits.

The problems I have with it are threefold.

First, while the stratification is reasonable, the hazard ratios (HRs) are in the range of 1.5 to 2.5—not very exciting. The P values are entirely reasonable, but the HRs don't provide a lot of discrimination. If you look at the different survival characteristics, again there is not a huge discrimination or scatter that would allow me to say that I definitely would or would not give chemotherapy or even intensify chemotherapy. The survival figures are too closely clustered to be able to use tumor budding in a clinically important way, and to make significant treatment decisions in partnership with our patients.

Second, this study was done beautifully and was meticulously quality-controlled. All of us recognize and understand the quality of work that comes out of Japanese pathology laboratories. I think it would be more difficult to maintain that degree of quality assurance in pathology laboratories all over the world. While we salute and herald the quality of the study, there remains a question mark as to whether we would be able to maintain that degree of discrimination if we tried to make it more widely available.

Third, I have a feeling that these morphologic tests are going to be overtaken by digital pathology and artificial intelligence (AI). I have seen presentations recently (not yet published) in which extraordinary work is being done using neural algorithms and deep learning. Computers are being trained using AI to recognize pixels in conventional H&E-stained sections and making very large, prognostically separate groups on the basis of digital scanning of an H&E section. This is something that can be applied objectively, potentially in every laboratory in the world, with central quality assurance driven by software digital analysis and central collection of the digital data. I feel that we're standing at the cusp of a revolution using digital pathology and AI. I think it's here, and we will see real-world tools within the next 12-24 months that will revolutionize how the clinician provides information to our patients on the basis of simple H&E stains. It's remarkable.

This is a nice piece of work from Japan; it is of its time. But I don't feel that there is strong discrimination for me to be able to use as a clinician.

I have a feeling that tumor budding will possibly be overtaken soon by different digital pathology tools. I am interested to know whether you think digital pathology and AI are moving forward. If I'm being Nostradamus and predicting the future, do you agree with me? Or, do you think it's baloney, that it's a false dawn?

As always, thanks for listening. If you want to post or say anything, I'd be happy to listen and respond. Medscapers, ahoy.

Follow Medscape on Facebook, Twitter, Instagram, and YouTube


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: