SAN ANTONIO -- A novel way to speed the testing of cancer drugs and quickly separate winners from duds has yielded its first big result: an experimental medicine that shows promise against a hard-to-treat form of breast cancer.

The method involves studying drugs in small groups of people and using statistical techniques to analyze the results as they come in, instead of waiting for all the data to arrive.

Whether the drug, veliparib, makes it to market remains to be seen, but it has shown potential to advance to final-phase testing aimed at Food and Drug Administration approval.

Bringing a new drug to market usually takes more than a decade and tests in thousands of patients, and costs more than $1 billion. Companies can't afford many studies like that, and patients can't wait years for potentially life-saving medicines, said Don Berry, a biostatistician at the University of Texas MD Anderson Cancer Center.

He helped design the novel analytical method discussed Friday at the San Antonio Breast Cancer Symposium.

Researchers testing a drug usually don't see results until they're all in, to prevent biasing the study. But several years ago, an unusual partnership decided to try a new way. It involves the National Cancer Institute, the FDA, drug companies, dozens of cancer research centers and charitable foundations.

The study, called I-SPY 2, puts small groups of women on experimental drugs or combinations, then gives them surgery to see what effect the medicines had. The best result is a complete response, where no signs of cancer remain.

Each patient's results are analyzed as they come in, and advanced statistical methods are used to calculate probabilities that the drug would help in various situations, depending on which women had a complete response.

"This allows us to learn and adapt from each patient as the study goes on," and results on early participants guide treatment that later ones get, said Dr. Hope Rugo of UC San Francisco. When enough evidence indicates a high probability of success, the drug "graduates" to final-phase testing.