Today there was a lot of buzz surrounding the release of results from the BATTLE trial (Biomarker-integrated Approaches of Targeted Therapy for Lung cancer Elimination), sponsored by the US Department of Defense. This trial, started in 2006 attempted to group patients by predominant biologic features of their cancer, including those that can be characterized by genetic changes in EGFR, KRas, RXR/CyclinD1 or VEGF (all predominant defining molecular signatures lung cancer) and see if they could match these patients with the optimal treatments choosing from four treatment regimens.
A nice piece from the Wall Street Journal distils the results here. But the story had an odd typo last I checked. See if you can find it.
The first phase of the ~250 participant trial matched patients with treatments with a roughly equal randomization, then as more data accumulated, they began to optimize the assignment, a process they called adaptive randomization, essentially learning as they go. Oddly, the treatments which were largely developed based on defects in specific genetic pathways had never been tested directly against just the cancers known to have the defects. This is one of the first trials to take early biopsy results and follow up information and feed it back into the trial design.
The results are, as one might expect, mixed. Median survival went up to nearly a year from seven months. They did highlight some unknown aspects of the disease and the treatments, however, such as a way to subgroup EGFR mutations that might be more responsive to a drug called Erlotinib. They also highlighted the drug Sorafenib which appeared to have good outcome for most groups of patients including those with Kras mutations, which have been particularly hard to treat. It’s too early to make clinical recommendations based on these phase II trials says Edwin Kim, the principal investigator from MD Anderson Cancer Center in Texas. But they show that this adaptive randomization approach can work in trials.
Currently the slides and the recorded talk can be found here although I don’t know how long they’ll be available.
In a press conference after the data were presented, the lead investigators introduced two patients who had participated in the trial. Ervin Lobo was 37 when he was diagnosed with lung cancer. Jeffrey Wigbels, 60, who was diagnosed with lung cancer the day before his daughter’s birth was surprised, as was I that taking tumour biopsies to optimize treatment is currently so rare. “I’ve not met anybody whose had a biopsy for the purpose of treatment,” he said. His daughter is now three.
Kim and other study leaders stressed that for such studies to be successful, taking good, representative tumour biopsies, which can be difficult for both doctors and patients, will be important for capturing as much information in the clinical setting as possible.