Rules and filters and their impact on success in chemical biology and drug discovery

Published on April 2, 2014   78 min

Other Talks in the Series: Small Molecule Drug Discovery

0:00
Hello. My name is Chris Lipinski. And I'm currently a scientific adviser to Melior Discovery. Melior Discovery is a little start-up company that's located in Pennsylvania. The offices and vivarium are in Pennsylvania. I actually work out of my home office, in Waterford, Connecticut. And the title of today's talk is, Rules and Filters and Their Impact on Success in Chemical Biology and Drug Discovery.
0:32
The second slide is an outline of the topics I'd like to cover in this talk. It essentially consists of two parts. The first part, I'll be talking about academic targets and the translational gap. I'll talk about chemistry and nutrition, how it's getting worse with time. And reductionism, genomics, HTS, are they to blame? And in this first half, some comments about screening diverse compounds. And an important point is that screening diverse compounds really is the worst way to discover a drug. In the second part of the talk, I'll become a little bit more detailed, talking about what is a medicinal chemist? I'll try to get the point across that I firmly believe that medicinal chemistry is a pattern recognition discipline. I'll go on to some discussion of biology and chemistry networks analysis. And then I'll finish up with some comments about ligand efficiency and selectivity.
1:34
I'll be talking about drivers for discovery changes. So we all know that there's a problem in terms of productivity in the drug discovery process. And it's helpful to try to analyze where the problems occur, especially in the early discovery phase. And so one can consider the issue of attrition. So the loss of compounds in sort of three buckets. So the chemistry bucket, a safety bucket, and an efficacy bucket. Now the chemistry bucket is everything about a compound that might become a drug potentially, that can be predicted just from the chemistry structure alone. And that's the bucket where, a priori, we have the best success rate at predicting which are going to be the good compounds, and which are going to be the compounds that we probably want to stay away from. And this is the area where rules and filters come into play. For example, rules and filters having to do with physical chemical properties, or structural features in a compound that we might want to stay away from. And, overall, we're quite successful with that. 2/3 of the time, approximately, we can a priori predict this is going to be a good compound, or not such a good compound. Now the caveat here is that the predictivity, the ADME predictivity-- so that's absorption, distribution, metabolism, excretion-- that's what ADME acronym stands for. It gets worse as the compounds lie further and further outside a RO5 space. So this 2/3 success rate is for the-- let's call it the traditional compounds with good physical chemical properties. If you're very, very high in molecular weight, or very, very high in lipophilicity or extremely polar, then the chemistry predictivity drops off markedly. Now in a safety bucket, it's still reasonable, not quite as good as the chemistry bucket. But the safety bucket is everything that can go wrong in, say pre-clinical, in vitro assays or animal toxicity, and even reaching into the clinic phase. And one reason why we're relatively good at this is that we have quite a good handle on the major target organ toxicity. So for example, we have a lot of experience, and we know a lot about pre-clinical assays and animal tests for hepatotoxicity and renal toxicity. And those would be the two most common causes of toxicity. Now where you run into a problem, is if you have toxicity in some area where either the experience internally or the literature precedent is not very good. And that's why it's only at about 50%. Now the area that really contributes to loss of compounds in the discovery process is efficacy. And of course, we only find out about efficacy, once we get into the clinical phase, usually Phase 2B. And this is atrocious. It's not better than 10%. And in some areas, it's a lot of worse than 10%. And this is where the majority of compounds are lost. And this issue of inability to predict clinical efficacy, reliably, really is not getting any better. And so, it is such a bad situation that it really has, for example, senior executives in drug discovery tearing their hair out. How are they going to handle it? And so what has happened is, one solution that's in play right now is to tackle efficacy using academic collaborations. So the idea is, we probably will have the greatest success in clinical efficacy, if we work in an area where our biology knowledge is rich, where we know as much about the target as possible, potential target, as much about the disease as possible. Well, where are we going to find those people? Well, you're going to find them in academia, because they're supported in the US by the National Institutes of Health. And so, in that sense, having very close collaborations with academic, primarily biology experts, is a real advantage. And the alternatives, really in theory, there are alternatives. But in practice, there aren't. So for example, many people believe that systems biology, the knowledge of how signaling networks actually exist in a human, in a disease, if we really understood that, then we could rationally, for example, choose targets and have a much higher probability of efficacy. But despite, you know, now multiple decades of work, it's a very, very complex problem. And we're really not there yet. So this bottom line of academic collaborations, many people believe that target quality is most likely from rich biology. And that really means collaborations, either with the academics, or potentially with a small biotech start-up that has some-- maybe it's a spin out from an academic profession.
Hide

Rules and filters and their impact on success in chemical biology and drug discovery

Embed in course/own notes