AUSTRALIAN FRONTIERS OF SCIENCE, 2005

Walter and Eliza Hall Institute of Medical Research, Melbourne, 12-13 April

Session 4: Discussion

Question – I guess this is a question for all of you. A lot of cancer biologists have commented that there are a lot of effective therapies for murine cancer but often they don’t translate so well to human cancer. I would be really interested in knowing what are the latest developments in new animal models for, if you like, initial testing.

Ricky Johnstone--One of the things that we like to do is to actually use immunocompetent mice as our models. As you said, often it was quite easy to cure mice that had human tumours, because they were using not-SCIDS mice or nude mice that had huge xeno-tumours sticking out the side, and really that didn’t reflect what would happen in a clinical situation. The transgenic models and the genetic models that are coming through in mice genetically replicate what happens in the humans, in terms of the oncogenic effect – and the Eμ-myc model is a very nice example of that, where we can replicate what happens genetically for that tumour to arise in the first place, and model then on what happens in the human. If we use those models, and also use immunocompetent mice for the therapy studies, that actually starts to better replicate what we would, hopefully, see in the human situation, the clinical situation. I think we and others are moving more towards that sort of situation for our models.

Question (continued) – So still with mice but better mice?

Ricky Johnstone – Yes, ones that, hopefully, better replicate what you would see in a human scenario.

Chair – And there are a lot better mice, such as mice in which you can induce a cancer because of the genetics in the same progressive way in which it will happen in the human – the blood flows, the size of the tumours, it is all very similar now. And you could probably treat the mice over a longer period of time. Getting rich quick by treating a xenograft and curing it in 30 days is not what Grant is normally faced with in a human patient. So the side effects and all those things can be brought to play.

Furthermore – I will just mention this and Grant can follow on – the National Cancer Institute (NCI) is asking for the drugs to be trialled in what is called a phase 0 manner, doing the type of work that Grant is doing: getting the most out of the initial use of the drug in humans, to see what is actually happening to the mechanism when it goes into humans, the metabolic responses, the growth factor responses, the target responses, the pharmacology, so that in the first eight to 20 patients you learn more about the drug than you ever knew before, before you start having hundreds of patients on a trial.

Grant McArthur – I would certainly agree there are a lot of limitations with xenografts, and things are slowly changing. Industry and the FDA in the US still like to see things happen to human cells, unfortunately. So I think that is still a bit of a barrier that we need to keep chipping away at, to convince them that in fact you get more bang for the buck from looking at the syngeneic immune-competent models such as Ricky has referred to.

What Tony said is absolutely correct. I think that by investing more heavily in some sophisticated trials very early on, firstly in mice and then translating that very rapidly into humans, is going to be the way to speed up what has been a very slow process.

I would say, though, that there are a few targets for which you could almost say you don’t need mice. I think BCR-ABL is a good example of that, in chronic myeloid leukaemia. Yes, there were some cell lines of that Harvard BCR-ABL that showed some good responses in vitro, but that drug virtually went straight into humans without good animal xenograft data at all. And it was proved to be very effective because it was a really good target and it was a really good drug.

Question – Ricky, I suppose the combination of HDAC and 5-azacytidine gives some added tumour specificity because of cell division. Can you pre-treat to lock things up as the beginning of S phase with something and then put in your treatment?

Ricky Johnstone – Yes. We haven’t done that, and I think Steve Baylin is actually trying those sorts of experiments at the moment. I haven’t really done any combination therapies with 5-azacytidine myself, other than just to do a few little things to show that we could reactivate genes that were hypermethylated. But adding the drugs in specific phases of the cell cycle I think would be a really good idea, and certainly 5-azacytidine at that time would be a great idea.

Question – Is another answer to James’s question the variability you get in a group of patients in a phase II or phase III trial, in the stage of their tumour and all the rest of it, versus 500 mice in a laboratory situation, all with the same tumour, in the same housing conditions, that are absolutely under control. Isn’t that one of the major differences in going from mouse to man, or to human – just the total variability of a group of 100 patients in a phase II trial?

Grant McArthur – Yes, there is no doubt that that is the case – you have got a highly controlled experimental situation. But there are also other very profound differences between xenografts and human tumours. For angiogenesis, for example, every angiogenesis inhibitor works wonderfully in a xenograft, because they are really rapidly growing tumours that have to be highly angiogenic, where human tumours are a lot more weakly angiogenic. So yes, there are a lot of differences other than just that variability factor, but you are quite right: that is one of the challenges that you face when you move to human tumours.

Question – Ricky, I would like to ask a question about the biology of the HDAC inhibitors. If I understood you right, up to 30 per cent of the gene’s transcription might be affected. But you made a little summary that the apoptotic genes go up and the anti-apoptotic genes go down. If 30 per cent are affected, that is a remarkable coincidence for the drug to actually choose those ones that you want to go up and the ones that you want to go down. What is the biology of the drug that would have allowed that to happen?

Ricky Johnstone – I should point out that we took a selection of that 30 per cent and I showed the apoptotic genes, and it just came out that the balance was certainly in favour of a pro-apoptotic response. Whether that truly is the key for the induction of apoptosis in those tumour cells we don’t really know, and we really need to test that.

I guess one of the other points we need to get at is: why are they tumour-selective, and are the gene expression changes different in a tumour cell versus a normal cell, and does that underpin the selective apoptosis that we can see in the tumour cells? We are actually doing those experiments now, where we are taking isogenic matched normal and tumour cells and going through, array them, just to see whether we get differences in the transcription responses in those matched cells. It could be that they transcriptionally respond the same way but it is the effect at the next level of activation that is different. So the trigger might be the same, if you like, for apoptosis, but the way they ultimately respond to that is different in the tumour versus a normal cell. We really need to get at that. It is a major question in the field.

Question (continued) – I presume that the HDACi were discovered because they were inhibitors, and not because they were HDAC inhibitors. In other words, they had a biology that might have led you to them. Is that correct?

Ricky Johnstone – Yes. Certainly the butyrates we know were used and were known to induce differentiation, and it wasn’t till later that it was understood that they regulated these enzymes. So, yes, history is being reinvented a little bit in terms of those compounds.

Question – Given that HDAC is actually PDAC, then I thought that blocking that function could have had marked side effects. Are you anticipating that?

Ricky Johnstone – It appears from the early phase clinical trials that there really aren’t huge side effects in the patients beyond the usual sorts of things: nausea, fatigue. But certainly a lot of the compounds have passed through phase I – Are they safe? Yes – into phase II, and are showing some remarkable activity. So I don’t understand why that is the case, but as I said to Tony, it could be that the trigger is the same in a normal versus a tumour but that the response to that trigger might be different. And so a normal cell might see all these gene expression changes as being some sort of DNA damage and respond to that, perhaps, by arresting its cell cycle. We know that if you block cells at G1, for example, HDACi don’t kill cells. So there has got to be some link between regulation of cell cycle and induction of apoptosis that we don’t yet understand.

Question (continued) – But with the combination therapy are you at more risk of such effects?

Ricky Johnstone – Possibly. And I think with the 5-azacytidine combination therapy there is more risk that you are going to reactivate more genes and perhaps have greater toxicities. I think the idea with the Apl, where you have got a fusion protein that you are perhaps targeting and those cells that express the fusion protein should primarily respond to the retinoic acid and the HDAC inhibitor, is a much better idea and a rational way to go after these things. It sort of fits with the Gleevec sort of model, I guess, where you have got a nice tumour-specific target that you are going after.

Question – Grant, it is not intuitively obvious that, if you inhibit VEGF, you would increase oxygen flow to the tumour – which is what you showed us. Perhaps you would just explain to us why that is the case.

Grant McArthur – There is not a definitive answer to that question, but I will give you the best working hypothesis at the moment. If you do scanning electron micrographs of tumour vasculature that is driven by VEGF versus more normal vasculature, the vasculature is very disordered, there is certainly evidence that the vasculature is leakier and has higher interstitial fluid pressures. So, potentially, by taking away the VEGF-driven abnormality in the vasculature you end up with a vasculature that in fact is more normal, and actually allows oxygenation and transfer of small molecules more effectively to the tumour. And that could mean drugs and oxygen, things that might be quite might be quite useful for trying to kill the tumour.

Question – This is a question for Ricky, related to Tony’s and Jamie’s point earlier on. As was pointed out, one of the key things is understanding our target and validating the drug target. The problem I see with the HDAC inhibitors is the pleotrophy effects and the multiple targets. My question is: how do you see it best to validate your target? What is the most practical way of doing it? And would that be an important issue in further development of this class of drugs?

Ricky Johnstone – Yes. I think one of the ways that certain drug companies are starting to get around that is by trying to develop compounds that might specifically target one HDAC, or at least one class of HDAC. That might get around some of the problems that are occurring at the moment. As I said, most of the compounds that are in use actually target all eleven HDACs, and so there isn’t true target specificity there, although it does target a class of proteins and not a single protein. So that is one way to get around it.

Another thing that we and others are doing is trying to delete different HDACs within the cell and then seeing whether that has any response – so, is there a specific HDAC that actually is important and is the specific target molecule that these drugs are hitting. Until we really get to that stage of specificity in our in vitro assays, I don’t think we are going to be able to answer those sorts of questions.

You are right, they are pleotrophic. But maybe that has an advantage for some of these compounds. We have found it is very hard to actually get resistance to these compounds in terms of upstream effects. So if you overexpress Bcl-2 then you can confer resistance, but a lot of the classic drug resistance mechanisms don’t hold true for HDAC inhibitors. [inaudible] a protein, for example: doesn’t affect it. Knock out a p53: doesn’t affect it. So maybe it is a bit of a balance, because they are somewhat broad, they have a bit of a multi-effect, I guess, on inducing apoptosis, and that is a good thing. The bad thing could be, of course, that it can bring about some toxicities that we might see down the track.