In medicinal chemistry, getting a read on toxicity earlier on in the drug discovery process is critical, said Jefferson Tilley, senior director, discovery chemistry at Hoffmann-LaRoche (www.rocheusa.com). “Toxicity is an issue, consequently there is a high attrition rate.”
“It takes 12 to 15 years and billions of research dollars to move a compound through the process, and 50% of these fail in the animal portion of the study,” noted James McKim, president and CSO of CeeTox(www.ceetox.com). “Historically,the focus was on the target and druggable attributes. There was little focus on liability and toxicity until late in the process.”
Druggable Target Space
There are only about 600–1,500 targets in the druggable genome that are viable, according to Tilley. “One way to expand this repertoire is to reconsider the opportunities presented by protein-protein interactions engaged in cellular signaling.”
Tilley not only sees opportunity but also challenges. “What interests me is personalized medicine—how drugs are metabolized in the body and how patient response differs from one individual to another. Within the next 10 years, I think there will be better ways to tell who is suitable and who is not suitable for a given regimen of drugs.”
Tilley said that, in principle, medicinal chemists can already do this. His presentation assessed the suitability of a target protein-protein interaction for modulation and its application to the design of inhibitors of the p53-MDM2 interaction for the treatment of cancer tumors. Such inhibitors are only expected to be effective for approximately 50% of patients whose solid tumors have wild type p53. The Roche work relied on a combination of high-throughput screening (HTS) with NMR and protein x-ray analysis of complexes between MDM2 and candidate inhibitors.
Fragment screening is another approach to discover lead modulators of protein-protein interactions. “Instead of looking at a collection of compounds, this strategy encourages fragment screening, looking at smaller pieces first to see if you can find a signal of any kind, then attempting to understand the binding mode in detail,” said Tilley. “And that can be a starting point.”
Tilley also raised the possibility of creating designed fragments. There is a low probability of a fit when comparing HTS to fragment screening in the case of a complex molecule. “With the fragment, there is simple binding interaction, but it is weak and needs elaboration. Combining fragment screening with x-ray crystallization is a method that is gaining ground,” said Tilley. “There are many possible targets, and we’re fishing in a large pond with some hope of success.”
In Vitro-In Vivo Correlations
Philip Burton, CEO and CSO of Admetrx (www.admetrx.com), presented a basic overview of ADMET and a case study addressing the need for in vitro data to have relevance in vivo. Discussion of in vitro-in vivo correlations stressed the value and limitations of the data.
The FDA Critical Path Initiative, published in 2004, states that in 2000, 8% of compounds passed the clinical trial process, as opposed to 14% historically. “We generate more data, certainly,” Burton noted, “but I believe that in the process of advancing candidates, we have thrown out a lot of good compounds. The number of candidates submitted for trial is keeping pace with spending, but a smaller percentage of these survive late-stage development. Overall, productivity is declining.”
The challenge is not to advance more candidates, but to advance the best of them, “and I don’t think that is what is happening,” said Burton. “A better understanding of the data is needed and the decision-making process needs to be improved. We need better in vitro toxicity tools. We need better data-integration tools, recognizing it is the composite profile of a molecule that dictates its success rather than any individual property. The industry is waiting for the development of these tools.”
And some of these advances will have immediate impact. “Better prediction of safety in larger populations, better understanding of drug-drug interactions, and a better understanding of efficacy are the tools needed to accomplish these things, and as they come on line the drug discovery process will improve,” said Burton. “But the process takes so long to begin with that you may not see the results for a couple of years after these tools are introduced.”
McKim noted that there are multiple parameters that both biologists and chemists need to track. “A chemist needs to have the ability to evaluate whether increasing the potency of a compound also increases the toxicity,” McKim said. “In addition, in vitro biochemical assays can provide important information on subcellular targets and potential mechanisms of toxicity long before the compound even enters an animal study.
“And if you don’t have a rapid screening system with an in vivo correlate (cell-animal bridge) that allows the ability to track new drug attributes as well as toxicity, you will have a hard time knowing whether the compound is going to be successful prior to the animal safety studies.”
McKim said that regardless of the methodology used to collect in vitro data, the key to using the information is to link changes observed in the cell models to effects in the animal model. He uses an algorithm that analyzes multiple endpoints over a broad exposure range to provide a predictive value for toxicity in animals. “This is where cytotoxicity programs have failed in the past.
“We’re focused now on trying to identify biochemical profiles associated with unexpected toxicity of approved drugs in small patient populations. There have been a number of noteworthy drugs that have passed clinical trials, only to be pulled from the market because of severe toxic effects in patients.” In the future, McKim said, “we will see those red flags sooner.”
The future will also focus on speed, efficiency, and reducing animal usage. Developing more robust data sets early so that better decisions for compound advancement can be made is the goal. And that, McKim noted, will depend on how drug discovery is approached.
“It is not enough to look at any one parameter—you can’t just look at efficacy, solubility, or ADME. You also need to look at toxicity and the intended therapeutic use to understand whether or not the compound in question has a tolerable risk profile. All of these things need to go into the decision-making process.”
The blood-brain barrier, which provides protection for the sensitive neural microenvironment, also creates a challenge for medicinal chemists, said Steven Hitchcock, director of medicinal chemistry at Amgen (www.amgen.com).
Blood-Brain Barrier Challenge
“You are dealing with a formidable barrier, active transport, and metabolizing enzymes, so how do you design a molecule that can not only get around this barrier but connect with its intended target and have a high degree of safety and solubility?”
Hitchcock noted that while great progress has been made in the neuropsychiatric pharmaceuticals arena, there remains much to be done.
“We are still making headway on disorders of the central nervous system and Alzheimer’s,” said Hitchcock. “The attrition rates are high—only seven percent of compounds oriented toward CNS gain approval, as opposed to 15 percent for all other applications. Clinical trials take longer, and as of right now, we don’t have anything to stop underlying disease, only improve the symptoms.”
One of the most exciting targets discovered in the CNS field is the BACE aspartyl protease target, which was identified in 1999. However, no clinical candidate has yet been identified by the pharmaceutical industry, said Hitchcock.
To permeate the blood-brain barrier, a drug requires specific properties, Hitchcock said. “It needs to be small, have the ability to bind to specific brain tissue, and have biological activity. As recently as five years ago, we weren’t able to get the compound to its target in the brain. Using computational tools and placing a high premium on in vivo models, we’ve worked hard to design a process to eliminate the efflux substrates.”
Understanding the nature and role of brain-blood barrier uptake and efflux transporters remains a developing area of research, said Hitchcock. “A key challenge for medicinal chemists resides in deciding which data provides the most relevant information to drive structural changes aimed at optimizing or minimizing blood-brain barrier permeability.”
Building Blocks for Drug Discovery
An important part of the drug discovery process is having ready access to the chemical building blocks necessary to create the compounds for drug development. Frank Kerrigan, research manager at Maybridge, part of Thermo Fisher Scientific (www.maybridge.com), said that his laboratory’s approach to developing and synthesizing novel heterocyclic building blocks is but one of many different aspects of the drug discovery timeline.
“Heterocyclic chemistry is key to the armamentarium of the drug discovery chemist,” said Kerrigan. “It provides key pharmacophoric diversity to the development of drug molecules. Of the top 100 drugs prescribed in the U.S. in 2005, 64 contain heterocyclic moieties.”
Heterocyclic reactive intermediates are at the core of Kerrigan’s group’s building-block research. “We have made quite a few over the years but we are constantly looking to expand on their flexibility and utility, for example, by expanding the range of functionality and adding solubility-enhancing groups,” said Kerrigan. “The key things scientists look for are building blocks that they can use with confidence, that react predictably, and that can be stored stably for long periods of time.”
A challenge that Kerrigan’s group faces is that some heterocyclic systems appear incompatible with the more reactive functional groups such as carboxylic acid chlorides. This particularly becomes an issue for acid-sensitive, strongly basic, or nucleophilic moieties within the systems. “These reactive intermediates can sometimes be synthesized on a small scale and used immediately,” said Kerrigan, “but we need to work at 50- to 100-g scale and store for extended periods of time for off-the-shelf availability.”
In the acid chloride case, Kerrigan’s group found a solution in activated pentafluorophenol (PFP) esters. “What we tried was an experiment for a customer that wanted expansion libraries based on hits it had obtained from our fragment screening collection. One of the hits, a pyrrole-containing carboxylic acid, did not work well with common amide coupling methodologies and the acid chloride was extremely unstable,” said Kerrigan.
“We applied the stable PFP ester to this library and found that it reacted effectively—it yielded a 70 percent success rate with 85 to 95 percent purity, as opposed to the 15 percent success rate shown with other coupling reagents. This was a huge advance, and the methodology behind it was actually simple.”
PFP esters appear to be a viable alternative to acid chlorides in many cases. “We have been expanding on this area, and so far we have trialed 50 novel PFP ester reactive intermediates,” said Kerrigan, “We believe these versatile novel heterocyclic reactive intermediates will be a great resource for medicinal and synthetic chemists.”