Imaging tumour habitats - life at the edge

Share :
Published: 26 Apr 2016
Views: 2942
Robert Gillies - Moffitt Cancer Center, Tampa, USA

Dr Robert Gillies talks to ecancertv at AACR 2016 about novel imaging techniques in tumour habitats.

Considering the stalling durable response rate of many targeted-therapy patients due to tumour evolution, Dr Gillies discusses the value of multiple imaging techniques to determining tumour habitat and response potential.

He also discusses the impact of data libraries on treating patients with regard to understanding tumour biology and heterogeneity.

AACR 2016

Imaging tumour habitats - life at the edge

Robert Gillies - Moffitt Cancer Center, Tampa, USA

The main topic of the session is on imaging and metabolism and genomics of tumours and my particular subject is imaging of tumour habitats. The main reason or the motivation for what I do is that targeted therapies are failing and there are very few durable responses to these targeted therapies and the reason why targeted therapies fail is because tumours evolve and they have an ecology. Evolution occurs not only when you have genomic or phenotypic plasticity across the tumour but they also have to have a selection pressure. So the microenvironment exerts a selection pressure. You can change the microenvironment by adding drugs and what you’re doing is you’re just selecting for cells that have resistance to those drugs.

The ecology aspect has more to do with how this all came about in the first place. The ecology is a characterisation of the physiological habitats that exist within tumours and these are the microenvironmental selection pressures that cause the outgrowth of the cells with different phenotypic properties. So in different regions of the tumour you’re going to have different cellular cancer cell and stromal cell phenotypes and we know this just by examining tumours ex vivo but more importantly we can now image this. So we do multi-parametric MR imaging or you can do PET plus CT imaging, we use this information to create areas that have specific combinations of these imaging features. So importantly some MR imaging, one of the techniques we use is called diffusion and that’s sensitive to cell density. T2 imaging is sensitive to the presence of metals, if you will; you can make it sensitive to the presence of metals which can be either in the mitochondria or it could be macrophages will cause a change in T2. We look at contrast enhancement so that’s a measure of blood flow in the tumour and these are all regional. So basically we can identify with these different combinations of imaging, we can define specific regions that have specific combinations of these factors and these are what we call habitats. These habitats, we propose, within them have a specific phenotype of tumour cell. We’ve seen this, the important other thing about imaging is you can capture the data longitudinally in a patient, so before they start therapy and during the course of therapy and we can use these habitats to monitor the response and so we’re doing this with immune therapy and we’re doing it with chemotherapy, we’re doing it with targeted therapy as well.

How have you used big data?

There’s always a limitation when you base your information on a biopsy, just because of the heterogeneity I talked about before. So you don’t know where within the tumour you’re capturing the biopsy. There was a very important paper in PNAS maybe three years ago where they sampled different regions of a brain tumour and eight out of the nine biopsies gave one diagnosis, one out of the nine gave a different diagnosis and that diagnosis was actually the one that you would base therapy on. So there are challenges when you’re limiting your information to a single point in time. If you can capture these data, though, on thousands of patients then those issues tend to be washed out and you can see some patterns emerge. But you need thousands of patients, it doesn’t do much for the individual patient though. So that’s why I’m committed to imaging because we see the whole tumour so we’re not subject to sampling bias and we can extract big data from images. Images are not pictures, images are data and we can extract big data from images and we can get images from thousands of patients because it’s standard of care, every patient gets imaged. So here are data that are routinely captured in the care and the diagnosis of cancer patients and we’re not analysing it to the extent that we need to. So this is a whole field called radiomics which is conversion of images to minable data that there are a number of people in the world that are doing.

Do patients have access to this data?

Yes, it’s a matter of degree because this is done already. Images are part of the patient record, oftentimes in the treatment of a patient the oncologist will show the images to the patient to point out the tumour. But for the most part oncologists are limited to just looking at the size of the tumour and we know that that’s misguided, that in many cases the size of the tumour is not reflective of how well it’s responding to therapy.

What can we expect to see next?

I do think it is an exciting time in cancer research, primarily because of immune therapy and the promise that this has and how much we’re learning what we know and what we don’t know. So we have identified a number of known unknowns about immune therapy. The exciting thing about it is it’s cancers that were previously so heterogeneous that they could not be treated by any targeted therapy are now being targeted with immune therapy. So I think it’s an exciting time, I think we have a lot more to promise our patients. But we are still, for the most part this is at the cutting edge of cancer care, many patients are still being treated with drugs that are thirty, forty, fifty years old and that’s not where we need to be. So I think we still have a lot of work to do.

Where the imaging fits into this is, like I said, images are data that can and should be used in the diagnosis and the individualisation, the personalisation. We talk about precision medicine all the time, well here’s a huge piece of data that should and could be incorporated into the treatment decision models for individual patients.

What is your take home message?

Where we are right now is conveying this message that I’m conveying right now that images are data and we can use these data in very meaningful ways to improve the care of individual patients. Where we’re going with this is to make this a reality, to collect these large data sets, to build prognostic and predictive models and then to get people to use them as biomarkers. So imaging provides very important biomarkers.

The thing that’s most important about it is that these are routinely captured on every patient so you don’t have to do anything extra. It doesn’t cost extra money, we just have to do a better job of analysing them.