Professor Peter O'Brien – Faculty of Pharmacy, University of Toronto

How did you first get interested in toxicology?

I think I was always interested . However I came from an environmental toxin point of view (particularly metals) and went to Professor Sten Orrenius’s lab on sabbatical which is where I got introduced to growing live hepatocytes and determining what quinones were poisonous and what were not poisonous. That is probably what started me, so I accredit it to going on sabbatical at the Karolinska Institute in Stockholm.

You are currently working on drug induced liver toxicity. Can you tell us about want you are doing?

I was mostly originally working on, as I was saying, poisons that cause liver toxicity like an overdose of acetaminophen because it is metabolized to a toxic quinoneimine , using traditional ways of working on liver toxicity. But then I was shocked to hear that therapeutic doses of some drugs (often new drugs) can cause idiosyncratic liver toxicity which did not show up during animal testing or in clinical trials. Occasional signs of liver injury in clinical trial were found but were reversible – and so then I became interested in that subject even though it is relatively rare that people die from these drugs. We don't know how this all occurs, and obviously it would be nice to be able to tell somebody that they can't take this drug because they will be susceptible to it, and at the moment we don't even know who would be susceptible. Then, of course, there is this paper that came out saying that taking drugs are the 4th leading cause of death in North America, which was another shock. I am sure that was highly exaggerated, but there was a lot of statistics that were used to create that paper.

So are you currently looking at methods to predict which drugs are going to be toxic to the liver before they get to the clinical stage?

Yes. Combinatorial chemistry has become a very popular way of creating new agents. The trouble is that it is producing them so fast that nobody can test their effectiveness and toxicity very easily. This is because it costs such a lot of money to test this agents in live animals. Once you have started this, it is only done under very close supervision and you can't change the protocols or anything, so it is important that we start doing this screening in vitro beforehand. At the moment my job is to be sure that what we are doing with our in vitro is consistent with what is found in vivo studies models of toxicity in rodents. Then we have to cross the species gap, nobody can predict what is going on in human beings without doing clinical trials and, of course, if you think the non chemical entity is toxic you are not going to put it into clinical trial. You are not going to take that risk so you have to use animal models all the time. It is quite a difficult problem and a very expensive problem. I think it has been estimated that up to 4% of all drugs are withdrawn from the market.

Do you think virtual screening will be of any use in predicting which drug will be toxic?

Well I would like to think that by putting together all these different approaches onto a different platform we would be able to do that. But I can't guarantee that when you are doing animal testing you have very highly inbred strains of animals that you are using, so you know the genetics/diet pretty well, and then to extrapolate that to human beings is a horrendous jump. Human beings, of course, have the freedom to eat whatever they want to, they are not all eating rat chow and they are not inbred and finally you can't do any experiments that could in any way risk their health. I do think the answer lies with QSAR analysis of the toxicity of chemical structures to validate using in vitro toxicology testing and QSAR analysis and that we should concentrate more on that than rather than shoving huge amounts of compound into animals which is the way we do animal testing now.

What are you hoping to get out of your research?

Obviously I would love to be able to predict which drugs, drug candidates and non-chemical entities are going to cause toxicity in animals. I am not sure I can go as far as ever predicting it for human beings, but if we could maybe get transgenic rats with humanized genes, that will be a more acceptable way of testing for human toxicity . As soon as the drug started to cause adverse effects in a human being in a clinical trial, you have to stop the treatment and you have got to tell the patient, these are the ethics you have to live by, even though the toxicity could be reversible.

I believe that you do not use highthroughput technologies at all, but use students instead. Are there actually any advantages to doing that, do you look at things more in depth?

We have to because in academia you are not supposed to be screening things, you are supposed to be discovering scientific mechanisms. Students, as soon as they get a result are going to complain that they don't understand how they got that and so you have to create a hypothesis. I think it keeps us on track in terms of understanding what we are looking at. Whereas, I think in industry they don't have time for that, they just want to pull one out and say develop this. They are not interested in things they don't understand, they just want to pull out what looks safe. I think industry and academia can work very well together. In fact, you will notice with all of the high technology that you hear about, it is not possible to understand any of it without going back to the research that was carried out in academia actually by graduate students. This old fashioned type of slow research in the past has built up a knowledge of understanding of why things are toxic.

Do you think we are sacrificing 'good data' for speed?

Guaranteed. In industry the bottom line is almost everything and getting the drug out as soon as possible. It is already far harder to get drugs onto the market now than it ever used to be. That has put pressure in favor of getting a drug to market rather than doing good research.

Further Information: http://phm.utoronto.ca/graduate/index.jsp