Bladder Cancer Drug R&D Falls Short
The situation reflects multiple factors, including insufficient funding for research on bladder cancer despite its prevalence and enormous management costs which rank among the top for all cancers, he told the recent Society of Urologic Oncology meeting in Bethesda, Md.
An important and generic obstacle to more rapid progress in this field is the general limitation of drug research, added Dr. Theodorescu, a professor in the department of molecular physiology at the University of Virginia in Charlottesville, where he also is director of the Mellon Prostate Cancer Institute.
Despite steadily increasing spending on research and development, the output of new approved drugs has not kept up with these expenditures, he reported. According to data from Price Waterhouse Coopers, 33 new chemical entities (NCEs) became available in 1999, down from 44 in 1997 and 52 in 1991. Drug development is an expensive, lengthy, and high-failure process, he said.
As a result, 99.98% of compounds screened in high throughput systems never win approval as therapies. For every drug that enters the market, between 5,000 and 10,000 compounds have been screened, 250 have entered preclinical testing, and five are tested in patients. This translates into roughly 14 years and $880 million in R&D costs per drug.
This cumbersome, inefficient approach continues to be used because the fundamental dilemma in drug re-search has not been solved: the lack of predictability of drug action in humans. Compounds that work against cancer cells in vitro or in animals may not work in patients and vice versa. So thousands of compounds are subjected to a battery of in vitro assays and only those passing with flying colors continue down the pipeline to more expensive, involved phases of research.
This embodies the drug development concept of “failing early, failing cheap.” Unfortunately, potentially effective drugs for bladder cancer may never reach preclinical or clinical testing with this model. Conversely, poor drugs (i.e., those that have no effect in patients) may get through.Another problem: The standard National Cancer Institute (NCI) drug screening cell panel called the “NCI-60” contains various neoplastic cell lines including breast, ovarian, leukemia, kidney, and prostate, but no bladder cancer cell lines. Therefore, the standard antineoplastic drug screening tool does not specifically assay activity against bladder cancer.
Genetics to the rescue
Nevertheless, new therapeutic approaches based on emerging technologies are on the horizon and stand to drastically alter the practice of oncology, Dr. Theodorescu said. In the last decade, enormous strides have been made in mapping the human genome, determining the function of genetic information, and processing massive quantities of data with bioinformatics.
The field of pharmacogenetics integrates multiple techniques with the aim of “developing drug therapies to compensate for genetic differences in patients and tumors (in the case of cancer), which cause varied responses to a single therapeutic regimen,” according to the NIH's MedlinePlus web site. These advances may well point researchers toward a solution to the fundamental problem of pharmaceutical unpredictability.
One answer lies in the “COXEN Principle” (CO-eXpression ExtrapolatioN), a complex algorithm developed to predict efficacy of a drug specifically against bladder cancer. Once a screened drug is found to be generally active against cancer cells in vitro via the NCI-60 panel (which as you recall includes no bladder cells), co-activity against bladder cell lines would be extrapolated using advanced computational methods around common biomarkers, or COXEN.
Any drug that passes the initial screen would undergo a second round of analysis integrating genetic data from an individual patient's tumors. Drugs that emerged with COXEN scores predictive of efficacy against bladder cancer are good candidates for clinical testing.The process has been validated on several drugs used against metastatic bladder cancer, and comparisons have been made to either in vitro activity or expected clinical response. Cisplatin and paclitaxel were analyzed for activity against bladder cancer using COXEN, and these results were compared with in vitro activity against 40 human bladder cell lines in BLA-40 panel. With 85% accuracy, COXEN correctly predicted which bladder cancer cell lines would respond to the drugs.
But can COXEN predict the effectiveness of marketed drugs in pa-tients? Eight drugs used to treat bladder cancer were evaluated using COXEN and results were compared with known clinical response rates. Interestingly, COXEN-predicted re-sponse rates vs. actual clinical re-sponse rates (complete plus partial responses) were closely related and suggested COXEN scores as surrogates of clinical response in patients.
Predicting sensitivity 79% of the timeTo causally demonstrate this association, the study looked at whether COXEN could prospectively predict patient treatment outcomes to specific therapies, such as in breast cancer clinical trials. Tumors removed from breast cancer patients were tested with gene microarray techniques in vitro and those data were integrated into the COXEN algorithm. Treatment outcomes were compared to COXEN scores; the COXEN system predicted tumor sensitivity to doxetaxel with excellent sensitivity.
To summarize, these studies revealed that COXEN has promise in accurately predicting efficacy of known bladder cancer drugs in vitro and in vivo, and predict treatment responses of known cancer drugs in patients in clinical trials, Dr. Theodorescu said, noting that it follows that COXEN might be used to discover new drugs for bladder cancer.