NCRI Cancer Conference 2010, 7 November 2010, Liverpool

Professor Will Steward – University of Leicester, UK

Personalising medicine in colorectal cancer

Yesterday afternoon we had a session relating to personalising medicine in colorectal cancer and the points that I tried to make were that at the moment colon cancer makes up about 800,000 to a million new cases per year, about 37,000 in a country like the United Kingdom and consumes a massive amount of resource because a vast proportion of those patients require quite expensive treatment. The problem with the treatment is it, generally speaking, has considerable toxicity. If we look at the large proportion of patients who require adjuvant chemotherapy, in other words attempt to cure, then you have to treat about 100 people for every six or seven who will benefit, so that approximately 90 patients will get prolonged, expensive, toxic treatment and gain no benefit at all. And the advanced setting which makes up a large proportion of patients, over half, also ultimately will require treatment for advanced disease, then again we are using all these drugs with a lot of toxicity, massive expense and approximately 50% of patients will gain some benefit. The problem with that benefit is it may be quite transient, it may be only a year’s prolongation of survival, and if you have to spend half that year having treatment and toxicity then you have to ask is it worth it?

How could we identify which patients will respond to treatment?

What we’ve got to do is be more sophisticated than we’ve been to date. The way we have done it so far has been quite crude: we’ve looked at the extent of disease pathologically, the biological state of the patient, to some extent age, although we discussed that at length yesterday. What we need to do now is take advantage of the new era proteomics, genomics, to try to see if we can’t find the links between giving certain drugs and those people who benefit. And so we talked about looking at, for example, mismatch repair gene; looking at genes which might predict how you metabolise a drug; looking at things like do people actually stick with protocols and give patients holidays from drugs, maybe that impacts on whether they do well or not. What we came to a consensus was that at the present time, despite massive amounts of research, we still don’t have good molecular predictors of who is going to do well or not. But what we could look to… we have one, we have KRAS which can predict if you will respond or not to cetuximab and EGFR inhibitors; we don’t have anything that could predict whether we could respond or not to bevacizumab, another very expensive agent. So at the present time we are still pretty much stuck in our old ways and we’ve really got to intensify looking at things like circulating tumour cells, looking at whole genomic screens to see if we can pick up associations between response and not.

Are there any other ways to potentially predict patient response?

What we felt we needed was perhaps to start looking at pre-operative exposure to drugs so that you could take a situation where patients are going to have surgery. We could give them some of the newer agents; there are many new agents which are not routinely used in colorectal cancer, brief exposure to those in the pre-operative setting. What we would then do is we would get a pre-exposure biopsy and a post-exposure biopsy at the time of resection surgery and look to see whether there are alterations in downstream targeting of those drugs in all individuals, in some individuals. If we could see differences in that situation, because that’s not been done before, we’ve not looked at pre- and post-exposure tissue specimens. We could look at that, see whether there are changes in certain individuals, perhaps correlate that with certain genomic changes then perhaps we can start at last to find a signature of genomic changes associated with certain chances of responding to certain drugs.

How would these advances benefit the patient?

There are several reasons the patient might be interested. First of all perhaps we can get a signature for that individual which might indicate that certain drugs might benefit them. We might do something a bit negative but it does still help the patient: we could say unfortunately none of those drugs are going to benefit you. That’s not a good news initially but at the very least it avoids that patient being exposed to toxicity, exposed to being out away from their home for the last months, years of their life. Of course the other thing it does is it has a massive impact on the health care system by markedly improving cost effectiveness of drugs if we don’t give them to everybody, we only focus on some people. If we can make the drugs more cost effective then NICE, the health care providers, are much more likely to fund the drugs which we don’t currently have for anybody at the present time.