The last few years have seen some dramatic changes in the field of drug discovery, but there is an urgent need to make R&D more cost effective, according to Alex Matter, Ph.D., director of the Novartis Institute for Tropical Diseases (Singapore). Dr. Matter was addressing the 8th annual "MipTec" meeting in Basel, Switzerland, last month.
"We are in an age of revolution with the arrival of new technologies and tools, such as proteomics, gene expression profiling, structural drug design, and noninvasive imaging," he said.
However a gap is opening between R&D costs and the productivity of the pharmaceutical industry. "If this trend continues, R&D will become unsustainable. Either productivity must increase or R&D costs must come down. The new technologies are beautiful, but they must start to deliver."
Companies that can either shorten the drug discovery and development timeline or reduce the rate of attrition will have a significant advantage, he added. The rate-determining steps that need to be worked on include selecting the right targets and good leads, developing predictive biological screens, having the "right" patients for clinical trials, and meaningful biomarkers, as well as smarter clinical trial strategies.
Where targets are concerned, more information on molecular epidemiology is needed, i.e., to show the role of the target both in the disease and among the population. It is also important to work on those targets that are essential to the disease, but not to the host.
When it comes to druggability of compounds, properties like potency, selectivity, cell permeability, pharmacokinetic parameters and others need to be taken into account.
Novartis took on many of these issues with Gleevec, its successful anticancer compound, which was one of the first truly targeted drugs. Gleevec acts on three separate tyrosine kinases and exploration of these different targets has allowed the extension of the indications for Gleevec beyond chronic myelogenous leukemia to other cancers, while AMN107, a more potent version, is under development.
Another important issue in drug discovery is unmet medical need. Ninety percent of R&D in the pharmaceutical industry is devoted to just 10% of the world's disease burden, which means that the three killer diseases in the tropics (HIV/AIDS, malaria, and TB) have been neglected.
For instance, there has been no new TB drug for 40 years, yet multidrug resistant TB is a growing problem. New funding models are needed for all stages of drug discovery and development, said Dr. Matter, and costs must be brought down to about one tenth of their current level. He is optimistic that this can be achieved.
"There is a new willingness to do something about unmet medical need that was not there ten years ago," he continued.
Biomarkers were a key theme at the meeting. Molecular profiling is now finding its way into preclinical and clinical development in the context of the discovery of useful biomarkers that may allow early diagnosis or prediction of an individual's response to treatment.
Nicholas Dracopoli, Ph.D., vp of oncology at Bristol-Myers Squibb (BMS; Princeton, NJ), predicted that cancer patients will have much to gain from profiling of their tumors, because of the importance of prescribing optimal therapy at the time of diagnosis.
Transcription factor profiling can show whether related types of tumors respond differently to the same therapy. Work at BMS has already demonstrated clustering by sets of genes in clinically similar cancers, one of which appears to respond best to therapy in early proof-of-concept studies.
Heart disease remains a major killer worldwide, and Robert Gerszten, M.D., of Massachusetts General Hospital (MGH), described how proteomics is leading to the identification of useful biomarkers. Researchers carried out a procedure known as a therapeutic myocardial infarction on a group of patients with enlarged hearts and studied the resulting proteome.
One protein identified in this study was alpha-1-acid glycoprotein (AGP); elevated levels are also seen among patients who do not respond to clotbusting therapy after heart attack, providing a measure of validation for the experimental finding. The team is now looking for markers of thrombosis, necrosis, inflammation, and unstable hemodynamics after a heart attack.
"This will help because there is a need to start treatment early for these patients," said Dr. Gerszten. "A protein profile can show the changes in the levels of circulating proteins."
The work is being carried out via a joint venture between MGH and Thermo Electron (Waltham, MA) known as the Biomarkers Research Initiatives in Mass Spectrometry (BRIMS).
"We realized that Thermo cannot work in isolation. We need to understand our customers' needs in biomarker discovery," explained Leo Bonilla, Ph.D., director of BRIMS.
The analysis of plasma samples is challenging because there is such a wide range of different protein concentrations. And each biomarker discovery program re-quires a customized discovery strategy.
Interest in the Peptidome
Meanwhile, there is interest in the peptidome's role in human disease, because many native peptides have been shown to be involved.
"We have so far failed to find, in medicine, any indication where peptides fail to play an important role," said Peter Schulz-Knappe, CSO of BioVisioN (Hannover, Germany). "This justifies the development of technologies to detect the peptidome."
BioVisioN has produced a 2-D representation of the peptides in a sample, with a third dimension that represents their intensity. From this, it is possible to find the increase or decrease of peptides in disease. Ideally, one would analyze tissue because this is closest to the disease, but this is only readily available in dermatology or oncology indications.
There is intense interest in the blood proteome, but it has to be kept in mind that the important proteins may already have been proteolyzed. Serum is highly dynamic, with its protein composition changing over time so there is an issue over whether biomarkers do actually originate from the patient and the disease.
Platelet-poor plasma may prove to be the best medium for biomarker discovery. BioVisioN is doing peptidomic studies in neurodegenerative disease, oncology, and metabolic syndrome (a typical peptide disease).
Other new techniques are aiding biomarker discovery. Jol Rossier, Ph.D., of DiagnoSwiss (Monthey, Switzerland), described a novel electrophoretic pre-fractionation method, off-gel, which is used to separate proteins and peptides according to their isoelectric point.
The technique has been applied, in collaboration with Geneva Hospital, to the study of neurodegenerative markers in cerebrospinal fluid in ante-mortem and post-mortem samples. So far, the scientists have identified a number of unique proteins, which are being validated through conventional Western blotting.
Meanwhile, LC/MALDI is rapidly increasing in importance as a proteomics tool, with advantages over gels or LC alone when it comes to the analysis of complex or difficult mixtures.
"Separation is superior on LC/MALDI," pointed out Denis Mller, Ph.D., of Novartis Pharma (Basel, Switzerland). For example, LC/MALDI analysis of one band in an E. coli membrane preparation yielded 55 proteins as opposed to only seven using standard methods.
Another band yielded 28 proteins compared to four. Many were integral membrane proteins, which are hard to separate with traditional 2-D PAGE analysis and included maltoporin, the protein with the most transmembrane domains (18) in E. coli.
The search for genes or proteins with altered expression in disease may lead to the identification of novel targets and new biomarkers. According to Stefan Evers of the Roche Center for Medical Genomics (Basel), there is some overlap between targets and biomarkers. The breast cancer drug herceptin has a markerthe overexpressed Her-2 geneas its target.
"Research aimed at biomarkers and target identification can and should be done in parallel," emphasized Dr. Evers.
Roche has been looking for biomarkers indicating the earliest stages of diabetes using a rat insulinoma model. While no new targets were found in this study, several biomarkers, falling into five different classes, were identified.
One, called PBCFM1, is increased in diabetics compared to controls and is linked to failure of the insulin-producing beta cells. In insulin-resistant individuals, E-selectin has been identified as a biomarker and this has been confirmed from a recent report from the Nurses' Health Study.
The identification and validation of protein targets is probably the most important aspect of drug discovery. Alex Crawford, of the Flanders Interuniversity Institute for Biotechnology (Leuven, Belgium), described the Institute's Zebrafish Morpholino Screen Initiative. This involves genome-wide, function-based target discovery in a zebrafish model using morpholino antisense technology.
Microinjection of the antisense oligos into zebrafish embryos produces readily identifiable phenotypic change, which leads to much more rapid discovery of gene function than is possible using knockout mice. So far, the initiative has analyzed 3,000 genes involved in vascular development, hematopoiesis, neurogenesis and other areas.
Current work is focused on angiogenesis, with a view to discovering novel targets with several genes now undergoing further validation. The platform can also be applied to drug discovery, looking for phenotypic changes in the fish on exposure to a small molecule.
For example, one compound, VIB003, affects vascular development and has been shown to reduce melanoma growth in mice.
"Zebrafish can be used for both target discovery and drug screening. This is one of the technologies of the future for drug discovery," said Crawford.
Assumptions Being Challenged
However, no one should get carried away by the idea that there are, post-genome, many thousands of new targets, according to Christopher Lipinski, Ph.D., senior research fellow at Pfizer (Groton, CT). This is just one of the many assumptions about drug discovery that needs to be challenged.
It is anybody's guess about the true number of drug targets, but it is probably in the low hundredsat least when it comes to looking at a single mechanism involving a single target for an oral drug. However, target opportunities do expand when nonoral drugs and polypharmacy are considered.
"Every clinically useful CNS drug is poly-pharmacological rather than selective," added Dr. Lipinski.
Low-dose oral drugs are also important, because no drug with a dose of less than 10 mg has ever been taken off the market. Unfortunately, in the present state of knowledge, finding such drugs is a matter of luck rather than logic.
"It is not really correct that drugs are becoming more potent. In fact, there has been little progress in getting low-dose drugs," noted Dr. Lipinski, who also pointed to the role of phenotypic screening, which was successful in the 1970s, before the advent of in vitro screening The move now to cell-based screening is welcome.
The combination of HTS and combichem has not been good for the industry and, as for genomics, it may have a positive impact on productivity a few years down the line, but it could also make things worse, according to Dr. Lipinski.
The rate of finding ligands that bind protein targets is lowthere have been only 24 new ligands over the last eight years and few of these have emerged from combichem.
What might help is to have academia, which is not forced to make money, work on targets that might be risky for commercial ventures.
Dr. Lipinski has been a pioneer in setting the ground rules of medicinal chemistry and does not think that the concept of "chemical diversity" is useful to the industry.
"The best way to never discover a hit is to rely on a chemically diverse library," he said.
After all, the targets that ligands bind are not chemically diverse; on the contrary there are a limited number of protein folds and only a few ways in which a protein molecule can form a site where a ligand can bind.
Of course, HTS does find some hits but this is not because of chemical diversity in libraries.
True diversity does not actually existexcept in silicofor various reasons. Only a finite number of compounds can actually be made because of the laws of chemistry and the availability of reagents.
And medicinal chemists are themselves biased, because of their individual experience. Library sizes are now going down (typically to below a hundred) and this has led to a better success rate in generating hits.