AACR 2016
Technologies for personalised cancer immunotherapies
Prof James Heath - California Institute of Technology, Pasadena, USA
We’ve been focussing on trying to understand what are the what are called antigens, what are the antigens that are recruiting T-cells into the tumour. The essence of immunotherapy is to activate T-cell killing in the tumour and there has to be something to draw them in. Once they’re drawn in things like checkpoint inhibitors can be activated. Something that has emerged over the past couple of years, this is probably the most rapidly developing field of cancer research that exists and certainly has existed in my life, is that antigens, which are fragments of proteins that are maybe only ten or eleven amino acids long that contain mutations, so they’re fragments of mutated proteins, are very, very important for immunotherapy, largely because they don’t look like self because they have mutations. So if you can trigger a T-cell response that is associated with these mutated antigens, then that T-cell response will just affect the tumour, nowhere else. So we’ve been developing technologies that allow us on a per patient basis to identify what neoantigens – they’re called neo if they have a mutation – what neoantigens are drawing T-cells into the tumour and then what are the functional properties of those T-cells once they’re drawn in to the tumour.
How do you identify a neoantigen?
It’s an interesting concept because the neoantigens themselves are being viewed as potential vaccines and the T-cells that are attracted to them are viewed as candidates for engineering T-cell drugs. The interesting concept behind this is that you actually have to start with the entire genome of the patient; everybody’s different. So you analyse the patient’s genome to identify what are the mutated proteins, you look at the RNA expression data to identify which of those proteins are actually being made in the tumour and then you take those proteins that are being made and also just computationally you chop them up as if they were being enzymatically digested. And for all those fragments then you calculate, once again it’s all in silico, which of those will fit into what are called major histocompatibility complexes which is how they’re displayed for T-cells to see. So you do this all computationally and you come up with a list of maybe two hundred or a hundred neoantigens in a given tumour that might be responsible for drawing T-cells in. Then you have to test them and testing them, that’s the challenge that we’ve been developing. We’ve developed a nanotechnology microfluidics approach that allows us to build basically a test for all the entire list. And we find quite a few of them there and we’re able to even follow those. It turns out those T-cell populations, it recognises neoantigens are in the tumour but they also are in the blood. So we’re able to follow the same populations in the blood just over the course of the therapy.
How effective has this development been?
We’re looking at patients that are on trials and what we’ve found, now these are melanoma patients that are responding pretty well to anti-PD1 therapies. What we’ve found is that something like half of the what are called CD8 T-cells, which are the major effectors, half of those T-cell populations are specific to neoantigens. So that’s a big number, it tells you overwhelmingly that these neoantigen specific T-cells are what’s killing the tumour.
What are the costs associated with treating a patient?
It was completely unthinkable a few years ago; right now to get a genome sequenced, maybe it’s two thousand bucks. In two years there are companies that tell me it’ll be down to four hundred bucks. Four hundred bucks, it almost makes sense for everybody to have their genome sequenced. Our analysis… so, for example, we were doing this entire analysis of a large library of neoantigens to try to figure out which ones were important and analysing the patients’ blood. I had a post-doc who started doing an analysis Friday afternoon, I landed here in New Orleans on Saturday morning, I took an unpleasant red-eye flight, and I had the data. So the technologies we developed, and that was searching through on the order of a hundred neoantigens, so for sure it’s an expensive thing right now but it’s getting cheaper very, very quickly.
Is the treatment still very personalised?
Everybody’s different so the challenge, and you mentioned the cost, the challenge is can we identify these candidate vaccines or candidate T-cell receptor genes and turn those into a therapy in a meaningful timeframe for a patient. I don’t see any barrier to doing that at the moment; there’s a lot of stuff we have to develop but one thing associated with technology that people get misled by is that you tend to look at what is the technology now and then you extrapolate linearly. So if I get cancer in three years am I going to be able to manage this technology? But technology evolves exponentially and so trying to project where we are now and where you’re going in a linear fashion doesn’t work. Let’s say it costs us $50,000 to do the entire thing, labour, everything for a patient now, in a couple of years that might be down to $1,500 and that’s very doable, that’s not ridiculous. That’s just how technology evolves.
Any final thoughts?
We started working in this field with my colleague, Tony Ribas something like twelve years ago. At the time it was a backwater and there were hints of success, Steve Rosenberg’s group was doing well. We had trials that we were involved in where you had patients respond but it wasn’t durable. Over the past three years, of course, it’s become the forefront of cancer therapy and it’s just amazing, it’s just amazing. Once you have signal, once you have patients that really have durable responses, you can always make it better. That’s the thrust of the field now, is to extend those durable responses to larger and larger patient populations. I see no reason why that’s not going to happen.
