As a member of Translational Medicine at Wyeth, Orest Hurko is helping to build one of the most innovative discovery and development processes in operation today. Wyeth has been dramatically reshaping its organizational structure and methods to reduce drug failure rates. This effort is focused on leaving less to chance and establishing better connections between discovery researchers and their colleagues working on clinical trials. In this interview, Hurko answers the question of what's working with Wyeth's new approach and more. 

PharmaWeek: Why does there have to be a change in how drug discovery and development is done? 

Orest Hurko:  One thing that influenced us was the observation that there is still a huge disconnect between what we already understand and what we have yet to learn about basic biology and human disease. Next, you have the failure rates.  That problem is now shifting so that less failure is taking place in Phase I, but more is happening in Phase II, where you are looking at efficacy.  Finally, as in most other major companies, a large proportion of our new drugs address novel targets, and those have a much higher failure rate than established targets.   

PharmaWeek: What's the major change with the new approach? 

Hurko: One big difference is the judicious application of biomarkers.  Probably the best known example of these are response markers:  It is now increasingly realized that there are different molecular defects in tumors, and that's why drugs like Herceptin work beautifully on those who have a certain marker, but don't work on other patients. Iressa, too, works well only in a subgroup of lung cancer patients.  

But I'm not just talking about the kind of biomarkers that you would include in a registration study being submitted to the FDA. Many markers can replace indirect and noisy clinical endpoints for internal decision-making, but it takes a much longer time to validate surrogate markers for a registration filing.

Even if they can't be used in a submission, these other kinds of biomarkers can also be very useful.  For example, you can often use biomarkers to get a quicker fix on whether the drug is altering the biology of a human subject in a way that will help deal with the disease.  The idea is to test more compounds, but also to find out more early on.  By doing small, early clinical trials with biomarkers, you can make much more meaningful decisions and save a lot of time and money. 

PharmaWeek:  People have been talking about biomarkers for several years now.  What's actually working besides those few obvious examples, such as Herceptin? 

Hurko:  We and our competitors are seeing interesting things, and that's starting to be reflected in the literature. For example, one type of biomarker, which I consider the most mundane, is the kind that lets you measure how much of the drug gets to the target.  A popular emerging technique for this is [positron emission tomography] PET.  It's used mostly in neurological and psychiatric studies. We've already seen a number of published examples whereby through direct measurement via PET, people can find the right dose to occupy the drug receptor for a long time and sufficient to achieve saturation.  Knowing early on how much drug to give is a huge advantage. Before we had that technique, the dose could only be guessed at. 

PharmaWeek:  What does Wyeth do better than anyone else? 

Hurko:  If I had to pick one thing, it would be transcriptional profiling. We are better at measuring RNA levels than just about anybody. We were the first to submit expression data as part of the FDA's voluntary genomic data submission. Through our acquisition of Genetics Institute we became frontrunners in this field, and we've greatly broadened our program. 

Because of that early lead, we've learned ahead of many others what RNA profiling is good for, and what it's not good for.  For example, we have used it for a specific application in toxicogenomics, and now, both industry and academics have discovered how useful this technique is for understanding differences between tumors.  It is very clear that for the clonal diseases–such as certain leukemias and breast and prostate cancers--this is a very powerful tool for distinguishing subgroups. 

There is little doubt that these 'omic approaches have demonstrated their utility. 

PharmaWeek:  Are you just as enthusiastic about genetic variations as markers? 

Hurko:  Not yet. For one thing, I think those genetic studies are more relevant to Phase IV failure than to improving Phase II outcomes, which is what I'm focused on. In addition, I'm not sure if pharmacogenetics is going to prove to be as useful as some of the traditional tools we already have.  Those sorts of mutations are usually low frequency.  If you only screen 20 to 30 people, chances are you won't find anyone with the mutation even if you have a good test. 

Of course some variations, such as the major drug-metabolizing genes, are relevant, but that is on a different scale than most.  Many companies are looking at DNA polymorphisms, but to me the jury is still out.  It will be very interesting in the next few years to compare notes and see if it's really worthwhile.  

 © Copyright 2006, Cambridge Healthtech Institute.
 All Rights Reserved.