By Jim Shrine


BioNumerik Pharmaceuticals Inc. quietly has developed a ¿mechanism-based¿ drug-discovery platform it says cuts the process from synthesis to human testing by two-thirds or more.

The fact that the privately held San Antonio company is taking its third compound into the clinic in about six years may attest to the efficiency of the platform, which, according to BioNumerik, integrates supercomputing with quantum physics, synthetic chemistry, and pharmaceutical sciences.

BioNumerik¿s new drug, karenitecin, was designed to address metabolism, bioavailability, toxicity and resistance problems associated with other camptothecins.

Karenitecin was developed in 24 months. The company¿s two other drugs in Phase I trials ¿ MDAM, from the methotrexate class, and BNP7787 ¿ took only 18 months each. The process is speeded by the use of two Cray supercomputers made by Silicon Graphics Inc., of Mountain View, Calif.

¿This is a high-risk business,¿ said Frederick Hausheer, BioNumerik¿s founder, chairman and CEO. ¿This technology is best suited to reduce the risk of failure at various stages where critical evaluation is required. A lot of things we look at won¿t work. If you know it won¿t work, you can abandon it quickly. We get to failure quicker.¿

The idea is that points of possible failure ¿ from chemistry to formulation, toxicology, delivery, metabolism and manufacturing ¿ can be dealt with concurrently, rather than as separate steps in the development process, Hausheer said.

The Cray supercomputers are capable of conducting 12.8 billion operations per second and hold more than 1 trillion bytes of data. BioNumerik uses the computers through a beta-test-site arrangement with Silicon Graphics.

In producing karenitecin, the company simulated more than 12 trillion possible drug candidates in less than one year. It used computer simulations to construct a detailed computational model involving the molecular interactions with cancer cell DNA and topoisomerase I, an enzyme involved in DNA synthesis and replication.

The resulting product represents a new class of drug ¿ one that contains silicon, and one that is highly lipophilic and poorly water soluble, thus expected to penetrate tissue better than other camptothecins. The drug also was engineered to have less interpatient variability and less toxicity.

The first patient in the Phase I trial is being treated at the University of Chicago Medical Center. The study is an accelerated dose-titration design in which dose levels will increase with each patient until Grade II toxicity or higher is seen. BioNumerik intends to follow the intravenous, single-agent study with a study of oral karenitecin and then one in children.

The Phase I study of BNP7787 is so promising, Hausheer said, that the company plans to go directly to a Phase III registration-supportive study in about four or five months. That drug was designed to increase the therapeutic index of taxane- and platinum-based cancer drugs.

¿It looks like it prevents clinically significant neurotoxicity,¿ he said. ¿We think there is reasonable evidence of neuroprotection, and we don¿t see the kidney toxicity. There have only been two or three patients [out of 14] with Grade I neurotoxicity. Several of them, maybe eight, have had four to more than six cycles without any clinically significant neuropathy. That is very striking, in our view.¿

BioNumerik has about seven other drugs in various preclinical stages. Nearly all are in oncology because that¿s where the company has expertise, but Hausheer said the mechanism-based platform probably ¿could be applied to any area of medicine amenable to pharmaceutical intervention.¿

Focus On Oncology, Not Services

For now, anyway, BioNumerik plans to keep the focus on oncology and not diversify into becoming a service-oriented business. A few pharmaceutical companies, such as Merck & Co., of Whitehouse Station, N.J., and Bristol-Myers Squibb Co., of New York, use supercomputers in their development efforts. But BioNumerik has a lot of proprietary expertise, with about 35 issued patents and more than 100 applications pending worldwide.

¿Our strategy has been not to license things in the preclinical stage,¿ Hausheer said. ¿Generally, we¿re more interested in getting good Phase I data, then commencing discussions with potential partners. We want to demonstrate that we¿re moving a high-quality drug through the process, not only in shorter time, but [a drug] that has promise in addressing large, unmet medical needs.¿ n