By Kevin Davies
July 20, 2005 | “Personalized medicine is about optimizing available therapies, not developing personalized medicines for everyone who walks through the [doctor’s] door,” said Nick Dracopoli, vice president of clinical discovery technology for Bristol-Myers Squibb (BMS). Dracopoli was the featured speaker in the two-day conference program, Advances in Genomic Medicine, at the 2005 Bio•IT World Conference + Expo.
Why are there so few examples of genomics-based drugs? “Because drug development takes 8 to 15 years. Genomic drugs are just now moving into development,” Dracopoli said. Moreover, a major question must be answered: “What is the drug doing in the patient? We must improve decision making, develop a better understanding at earlier stages” of the pipeline, Dracopoli said. The problem is exacerbated by the sheer number of patients, which can escalate into the thousands for certain oncology Phase III trials.
Dracopoli presented two cancer drugs that are exciting examples of clinical pharmacogenomics. For Erbitux, originally developed by ImClone Systems, BMS is searching for proteome and plasma biomarkers in mouse tumors that could provide a predictor of patient response. As for the novel breast cancer drug Ixabepilone, BMS is comparing gene expression in tumors and pretreatment biopsies in an effort “to correlate genes with drug sensitivity and resistance.”
Gene Discovery
One of the most important resources for gene mapping and discovery is the HapMap. According to Perlegen Sciences’ vice president Kelly Frazer, there are more than 7 million documented single nucleotide polymorphisms (SNPs). A subset of “250,000 tagging SNPs is a powerful tool for whole genome scanning,” Frazer said. Perlegen and Pfizer have collaboratively discovered four SNPs in the CETP gene associated with high HDL cholesterol. Perlegen has also identified SNPs in several unidentified genes associated with complex traits.
Rather than focus on candidate genes or gene association studies, Alisoun Carey described how U.K.-based Oxagen has focused on “the cream of the crop” — the roughly 800 members of the G-protein-coupled receptor family. Carey, director of target analysis, used expression studies to whittle that number down to 200 for genetic analysis, mostly for autoimmune diseases. Genotyping using TaqMan, Pyrosequencing, and RASH revealed several attractive receptor targets, including a strong candidate for Graves’ disease.
A unique resource that is helping many organizations in gene discovery, including Perlegen, is the Utah Population Database (UPDB), which contains medical and genealogical records on some 7 million Mormons. The database is now managed by a nonprofit organization called LineaGen Research Corporation. Chief operating officer Michael Paul noted that the database contains the records of dozens of multigenerational reference pedigrees that have been a staple for genetic studies for a quarter century, helping track genes for breast cancer, epilepsy, heart disease, and many other diseases.
The database is so powerful that in studies on autoimmune disease, Paul noted that LineaGen was able to “electronically mine the Utah Genetic Reference Project (UGRP) to identify three large pedigrees and eventually identify a disease locus — without doing a single wet lab experiment.” Collaborations are ongoing with Celera Diagnostics, Battelle, and Amgen.
Meanwhile, during the conference, IBM vice president of information-based medicine Michael Svinte announced a new collaboration in which IBM will produce an integrated clinical environment for both the UPDB and UGRP, designed to accelerate the dissection of complex disease genetics and target discovery (see “LineaGen and IBM to Collaborate,” page 14).
While delighted with Aspreva Pharmaceuticals’ recent public offering (see “Social Medicine, Canadian Style,” April 2005 Bio•IT World, page 12), Michael Hayden devoted most of his talk to Xenon Pharmaceuticals, which follows an expression coined by William Bateson, “Treasure your exceptions.” Xenon looks at rare genetic disorders that offer “desired phenotypes.” An example is sclerosteosis, a condition of increased bone density, which led to the identification of sclerostin, and a promising approach to osteoporosis for Amgen. Xenon’s biggest discovery has come in the field of obesity. The identification of the SCD1 mutation, which causes striking body fat reduction in mice, as well as lead molecules, resulted in a deal with Novartis potentially worth up to $150 million.
There are two reasons Jason Johnson of Rosetta Inpharmatics and colleagues are continuing to study the total complement of human genes. The first is to ensure that Rosetta microarrays have the “complete set of genes.” The other reason is to see “how complete must the parts list be for predictive network models.”
Johnson noted the “dark matter — intergenic regions with no ESTs [expressed sequence tags] or gene predictions.” While many of these may ultimately prove to be exons associated with known genes, there could be as many as 2,000 to 3,000 unknown genes lurking in the human genome. Rosetta has also a trio of pathway tools — Ingenuity Systems’ Ingenuity Pathways Analysis, Ariadne Genomics’ PathwayAssist, and GeneGo’s MetaCore — to produce a complete network database to help scientists generate and test hypotheses.
Biomarker Discovery
As noted by Pfizer’s senior director of clinical pharmacogenomics Keith Johnson, “Every day shortened in drug development is worth big bucks” — in the case of a blockbuster such as Lipitor, as much as $25 million a day. Genotyping at Pfizer embraces a raft of high-throughput technologies, including Applied Biosystems, Perlegen, GE Healthcare, and Sequenom’s MassARRAY, which runs off a genotype every 1.5 seconds. The problem, Johnson said, is that “there is no standard data-handling platform.” And the lack of a simple device to handle complex data in the clinic “is stopping the implementation of genetics in the clinic.”
John Quackenbush, who recently left The Institute for Genomic Research (TIGR) to join the Dana-Farber Cancer Institute, presented compelling case studies of how microarray analysis has identified a molecular basis for survival rates in types of colon cancer. Based on recently published microarray data [1], his team predicted a patient’s outcome. Future work will apply the microarray results to identify potential target molecules.
According to Emanuel Petricoin of George Mason University, protein microarrays offer improved assessment of molecular pathways and target molecules than their DNA counterparts. Another new study [2] predicts the response of breast cancer patients to specific treatments based on protein microarray data.
Finally, Zak Zimmerman of Alnylam Pharmaceuticals described potential therapeutic applications for RNA interference. Alnylam’s treatment for age-related macular degeneration should begin Phase I clinical trials later this year, and trials for a respiratory synctial virus drug are expected to begin early in 2006. Because RNAi probably evolved to fight viruses, Alnylam is looking to target other viruses in the future.