ASTRO 2015
Investigating genetic vulnerabilities to radiation therapy
Dr Mohamed Abazeed – Cleveland Clinic, Cleveland, USA
What research were you here to present at ASTRO 2015?
Our work is premised on the fact that we can eventually predict response to radiation. We know from several studies in the last decade that cancer is complex, it’s heterogeneous. What we find in our studies is that when you irradiate cells on plastic from even within single lineages, for example non-small cell lung cancer, adenocarcinomas of the lung, that you find a significant heterogeneity of response to radiation. So you have underlying genetic diversity of cancer and then you have an underlying genetic diversity in response to radiation and we try to determine what these determinants are of response. We think we can learn a lot about the interplay between the cancer genome and radiation response.
How many patients do or don’t respond to radiation?
We all know about the bell curve, the Gaussian distribution, so it seems remarkably that when you look at most of the lineages you have a Gaussian distribution of response. That is, a vast majority of patients have a moderate response but a significant subset of patients have an exquisitely nice response to radiation and some are very, very resistant to radiation. The extent of that curve, of course we need to profile more patients, but we think that there are going to be significant tails on both ends.
Is response to radiation linked to the tumour type?
Absolutely. So we know that tumours that undergo rapid apoptosis or programmed cell death respond exquisitely well to radiation. We know tumours like endometrial cancer respond really well; gliomas, breast cancers, pancreatic cancers tend not to respond well and that’s mimicked in our profiling efforts. So there is a correlation between what we see clinically and what we’re profiling in plastic on a dish.
Can you outline the presentation you will be giving?
The presentation that I’ll be giving is essentially describing this complex heterogeneous response to radiation. We look at distributions and then we apply a bioinformatics algorithm that allows us to determine what are genetic changes that correlate with resistance and sensitivity. Then we go on to show biological plausibility of those determinants. For example in lung cancer specifically we have identified specific mutations in the BRAF oncogene. These mutations fall within the kinase domain previously thought to perhaps be responsive to anti-BRAF or anti-MEK inhibitors but nobody has really identified how well they function. We know that there is a significant subset of melanoma patients with these BRAF mutations but for the first time we’re showing these mutations outside of the traditional V600E that appear to be conferring resistance to radiation.
What is the significance of this finding?
So we think that by profiling patients, which we predict will ultimately be a standard of care in the next few years in oncology, that we’re going to extract significant genetic information from all of our patients. We already profile, for example, non-small cell lung cancer, we profile patients for EGFR mutations, ALK translocations. We believe that there’s going to be more comprehensive genetic data that’s available to the clinician and the goal of our effort is essentially to be able to determine which of these genetic determinants may classify you as a potential responder or non-responder to radiation. That allows us to stratify patients in clinical studies and be able to fine tune those, for example the responders perhaps don’t need a significant dose. But more importantly for those that are resistant then we can maybe target those pathways of resistance with drugs.
What other data from the Cleveland Clinic are being presented?
The holistic presentation talks about across all lineages so really probably too much information to talk about in this small session. But we essentially look at major determinants of resistance to radiation across 26 different cancer types. I’ll give you just a couple of examples: for example, in colorectal cancer, endometrial cancer and ovarian cancer the degree of genetic changes in the tumour, the more unstable tumours the more likely you are have to be resistant to our therapy; those are called somatic copy number alterations. So we think this could eventually translate into a diagnostic where we can then identify patients with colorectal, endometrial or ovarian cancer that are more or less likely to respond. So you can see the general theme here which are molecular diagnostics so that we can instead of a priori irradiating patients with the same dose perhaps we can learn more about their potential to respond and then fine tune our radiation specifically for those patients, really in line with this effort generally in medicine a precision or personalised therapy.
How will you translate your findings into the clinical situation?
So you have cells on plastic and then you have patients. So for a subset of the findings that I mentioned we’ve actually done clinical correlative studies where we’ve taken tissue from patients and we’ve identified patients that have these alterations and those that don’t have these alterations and we generally ask how well do they respond. There is correlation in a specific subset of genes that I talked about in which we see that. But I think we need to go beyond that and in particular my lab is going to be and has invested significantly in an effort called patient-derived xenografts. You may be familiar with it, but patient-derived xenografts essentially are when you take a tumour from a biopsy or surgical specimen, you inject it into an immunocompromised mouse and that tumour grows. When you profile that tumour and you compare it to the original tumour taken directly from the patient those tumours look somewhat very, very similar. They’re quite identical actually. The genome is similar, but that occurs also in the cell lines, but more importantly the RNA expression, the differences in the RNA, almost match exactly, with R2 values of about 0.8, that means a very tight correlation. So we think these are great avatars of human tumours that we can then analyse in a translational setting where we don’t have to necessarily look only in plastic. But the plastic studies really allow us to generate this significant amount of data that then you can test in a more finite amount of xenografts and then also a finite amount of patients. So it really directs future work.
Do you have a take home message from the research presented?
The take home message is what we’ve learned that is a nice contribution to the literature is that the response is really heterogeneous and of course this presents a lot of complexity in terms of predictive models and how you begin to take a multivariable complex distribution, which is the Gaussian distribution, and determine who is resistant and who is sensitive. But I think this also presents an opportunity, the fact that our patients respond with a differential response to our treatments, suggests that we can potentially identify and predict those patients. We can predict a lot of complex things, some better than others. It’s very hard to predict how the economy is going to do or if an earthquake is going to happen in this city at a particular period in time. I know probably the weather actually we do a pretty good job of predicting, I can tell you with some level of probability whether there’s going to be a thunderstorm the next day or not. But can we bring this type of predictive potential to radiation oncology, radiation therapy and we think a priori that this should be feasible, the question is can we generate enough information to do this in a reasonable amount of time?
