Emerging Platform Technology for Cancer: Nuclear Imaging

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


September 06, 2018

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

I'm never one not to be self-promoting. That's a gentle joke. Most of you are used to me talking about gastrointestinal tumors, but I wanted to stray into other areas because I think this is a really interesting platform technology.

I thought I'd talk a little about some work that has just been published by my friends and colleagues from the Cancer Institute in Oslo. Håvard Danielsen is a great friend of mine. I am one of the coauthors, so there is a massive conflict of interest; nonetheless, let me talk about it. Loosely speaking, the group looked at nucleotype and nuclear texture as a marker of chromatin structure, function, and entropy and how it serves as a prognostic factor across a range of different gynecologic cancers.[1]

We know that genomic instability drives a lot of the hallmarks of cancer. We know that chromatin structure is complex. It is highly regulated, and, of course, it maps onto genetic instability and transcriptional control, so it's at the "center" of how the cell works. As we continue to take a reductionist approach, looking for individual markers that tailor individual prognostic paths, this is a step back. It is using nuclear imaging—a very highly sophisticated and digital image analysis technique to look at nuclear texture. The mathematics is way beyond me, but Håvard and his gang have put together a really fantastic group to look at the bioinformatics of nuclear texture.

A very large, well-conducted study of over 1000 patients showed for uterine sarcoma, ovarian cancer, and endometrial cancer that nuclear typing—digital microscopic assessment of nuclear texture—serves as an independent prognostic marker, with hazard ratios of 2 to 3, which are believed to be within the clinically valuable range for each of these different tumor types.

Why do I mention this? I have already talked a bit about another study recently published in [Annals of Oncology][2] in which we saw something very similar in a large series of colorectal cancer. It seems possible that nucleotyping could be a pan-prognostic marker across a range of different tumor types. This is exciting. We are entering a new age of digital pathology, machine learning, artificial intelligence, and so on. It seems to be that large-scale image analysis of this sort is going to generate very interesting, clinically correlated data.

I think we are seeing the emergence of a new platform technology. For the time being, it's related to prognosis, but who knows? With access to the right clinical trial specimens, we might be able to machine-learn some predictive markers out there too without having to sequence every single gene and without having to understand every single RNA transcript. We might be able to step back a little and look at the degree of organization or disorganization of chromatin itself.

Have a look at the paper and see what you think. For the time being, Medscapers, over and out. Thank you.


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: