Patient tumor samples in BATTLE, which were acquired fresh after enrollment, were analyzed for the presence of 11 different biomarkers, including mutations in genes such as EGFR and KRAS, excessive copies (or amplification) of Cyclin D1, or the expression of proteins such VEGF.
Trial Makes Push Toward More Individualized Lung Cancer Therapy
Preliminary findings from the first clinical trial in lung cancer to use molecular analysis of tumor biopsy samples and an “adaptive” design to direct patients to a specific targeted therapy were presented at the AACR annual meeting in Washington, DC. In the study, dubbed BATTLE, patients had their tumor samples tested for specific biomarkers and were enrolled into one of four treatment arms, each one testing a different targeted therapy, based on that analysis. (See the box below.)
The first 97 patients to enroll in BATTLE were randomly assigned to one of the four treatment arms as would traditionally be done. After that point, new patients were assigned to one of the arms based on a statistical model called an adaptive Bayesian model. In addition to using the result of the biomarker analysis, the model made the assignment based on findings from patients who had already undergone treatment, which, as the trial proceeds, are fed back into the model.
“Our hope is that this type of study will streamline development because it will identify efficacy signals earlier,” said Dr. Edward Kim from the University of Texas M. D. Anderson Cancer Center, who presented the results on Sunday. “So instead of doing large phase III trials where we’re looking for a 1.5-month improvement in overall survival, we can find a marker that represents a strong signal for the drug and makes the overall efficacy much better” in a smaller, more tightly defined trial population.
But, Dr. Kim acknowledged, BATTLE, which was designed in 2004, is only a beginning. “This type of trial model is highly dependent on having good biomarkers and good treatments,” he said, both of which are severely lacking for patients with advanced disease.
Like BATTLE, the recently launched I-SPY2 trial for breast cancer has an adaptive design statistical model. But I-SPY2 involves a healthier population of patients and is studying the delivery of treatments prior to surgery.
More than 300 patients were enrolled in BATTLE. The results presented at the AACR meeting were based on data from 244 patients. The trial’s primary endpoint was control of patients’ disease at 8 weeks. While overall survival for patients in the trial was 9 months, patients whose disease was well controlled at 8 weeks had a median overall survival of 11 months, compared with 7.5 months for patients whose disease was not well controlled at that point. No progression-free or overall survival data for each of the four trial arms are yet available.
Interpreting the limited findings from BATTLE is difficult, said Dr. Giuseppe Giaccone, chief of the Medical Oncology Branch and head of the Thoracic Oncology Section in NCI’s Center for Cancer Research. That is due in large part because of some significant limitations in the study’s design. For example, Dr. Giaccone explained, 8-week disease control is not a validated endpoint in lung cancer clinical trials. And the study lacked a control arm, making it difficult to judge the efficacy of the individual drugs, of which only erlotinib (Tarceva) is approved by the FDA for treating lung cancer.
During the plenary session presentation, Dr. Paul Bunn, director of the University of Colorado Cancer Center, lauded the ability to conduct such a complex trial but also touched on some concerns, including the rationale of using erlotinib in combination with bexarotene (Targretin), which is approved to treat cutaneous T-cell lymphoma but has not shown activity in lung cancer.
While it is a phase II study and has its limitations, Dr. Kim stressed that BATTLE underscores the importance of getting adequate tissue samples from patients with non-small cell lung cancer (NSCLC). Currently, Dr. Kim said, tumor samples are available for only 20 to 30 percent of patients.
Uniformly obtaining suitable tissue samples will greatly aid research, Dr. Kim noted. And as studies emerge pointing to biomarkers that may predict response to approved and investigational treatments—such as EGFR mutations for erlotinib and the presence of the EML4-ALK gene fusion for investigational ALK inhibitors—samples will be available to test for those markers and offer patients a potentially effective treatment.
Based in part on their experience with BATTLE, M. D. Anderson Cancer Center researchers will publish papers in the future that identify “good targets” for metastatic or primary tumor sites “to optimize the amount of tissue you can get from a biopsy,” Dr. Kim said.
Two similar adaptive-design trials, BATTLE 2 and BATTLE 3, are already moving ahead, Dr. Kim said. BATTLE 2 will have a similar patient population to BATTLE, but with a number of modifications to the design, including less reliance on prespecified biomarkers. BATTLE 3 will involve patients undergoing initial therapy for advanced NSCLC.