WASHINGTON, DC—Biomarkers can be identified that predict tumor responses to specific biological drugs, according to results of the Phase II BATTLE (Biomarker-integrated Approaches of Targeted Therapy for Lung Cancer Elimination) trial in pretreated non-small cell lung cancer patients, presented here at the AACR Annual Meeting as Late-Breaking Abstract #1.
“Identifying proper biomarkers for a favorable population is a step towards personalizing therapy,” said Edward S. Kim, MD, Associate Professor in the Department of Thoracic/Head and Neck Medical Oncology at the University of Texas M. D. Anderson Cancer Center, who said that the study represents a new research paradigm of real-time tumor profiling of current disease status.
The hypothesis-driven, prospective trial was designed to assess each patient's tumor characteristics and then to direct novel biologic treatments in hopes of improving outcomes.
Unique to this trial, patients were first required to undergo fresh core biopsies to test for lung cancer markers. From these results, five biomarker groups were formed according to their signaling pathways.
On the treatment side, four regimens were selected. Patients were initially selected randomly to receive one of the four treatments, but after the first third had been treated, subsequent patients were treated based on responses of earlier patients, matching drugs to biomarkers.
The trial's primary endpoint was disease control at eight weeks. Trial patients' median age was 62 (range of 26 to 84), and all were heavily pretreated with chemotherapy.
“Patients with stable brain metastases were allowed in the trial; thus we were including lung cancer patients with the worst prognosis,” he said.
Bayesian Adaptive Randomization
Using an unusual randomization method, the researchers treated the first 97 patients (38% of the 244 evaluable patients) by randomly selecting them to receive erlotinib at 150 mg daily; vandetanib at 300 mg daily; erlotinib at 150 mg plus bexarotene at 400 mg/m2; and sorafenib at 400 mg twice daily.
In that first stage, 37% of patients achieved disease control by eight weeks, Dr. Kim reported.
In the second stage, treatment for the subsequent 158 patients was based on a Bayesian adaptive method of randomization. Each patient's treatment was based on outcomes of previous trial patients who received a particular drug regimen and who had specific biomarkers.
In effect, as the study progressed, each patient was more likely to receive a drug that had worked for earlier patients with the same tumor biomarkers.
For example, Dr. Kim said, patients with VEGFR overexpression responded well to vandetanib, but those with KRAS mutation did not.
Dr. Kim reported that 42% of patients achieved disease control at eight weeks when treated through adaptive randomization.
Eight week disease control does predict outcome, Dr Kim said. He said median overall survival for all patients was nine months.
“However, if a patient had disease control at eight weeks, median overall survival improved to 11.3 months, vs 7.5 months for those whose disease was controlled at eight weeks.”
Dr. Kim said that Phase II trials using adaptive randomization in the future will lead to Phase III trials requiring smaller sample sizes and taking less time to complete, as biomarkers help identify a potential patient population.
At eight weeks, better disease control was seen in patients who had the EGFR mutation biomarker and were treated with erlotinib. Patients with the markers Cyclin D1, IHC positivity, and EGFR FISH A achieved better disease control when treated with erlotinib plus bexarotene.
On the other hand, when patients were treated with sorafenib, better disease control was seen when EGFR mutation or high polysomy were absent.
Six patients with EGFR mutation and FISH A biomarkers had 100% disease control with erlotinib or erlotinib plus bexarotene, but no disease control with sorafenib.
Patients with KRAS mutation responded better with sorafenib (61% disease control at eight weeks) compared with the other three regimens (32%).
And patients with KRAS Cys amino acid substitution had worse progression-free survival compared with patients with wild type or all other KRAS amino acid substitutions.
Other biomarkers were high polysomy, VEGF/R2, RXR alpha nuclear staining, and Cyclin D1.
The only toxicity reported was pneumothorax, at a rate of 11.5%, with one Grade 3 patient hospitalized for pneumothorax. There were no other complications, Dr. Kim said.
Paul Bunn: Novel & Important Trial, but Future Trials Should be Simplified
The Discussant for the presentation, Paul A. Bunn Jr, MD, Professor at the University of Colorado Cancer Center, began by congratulating the M. D. Anderson team for completion of a novel and important trial.
Then he dug in, first noting that “adoptive randomization did not appear to influence outcome.” While 42% of patients in the adaptive randomization stage achieved disease control at eight weeks, he noted, those undergoing standard randomization achieved only slightly less, at 37%.
Dr. Bunn said this was likely because of non-optimal biomarkers and “population drift,” when patients enrolled late in the trial were more likely to have poor performance status and fewer favorable mutations.
The trial was too complicated, Dr. Bunn said, with four treatments and five biomarker groups. He said future trials of this type should be simplified.
He thought the biomarkers used were suboptimal as they were not validated individually or by grouping. And 18% of patients had no biomarker identified at all.
He also questioned the choice of treatment. Erlotinib, he said, was the only pure inhibitor among the regimens, and he did not see the rationale for combining it with bexarotene.
“And there is little evidence that vandetanib was better than erlotinib,” he said.
Nonetheless, Dr. Bunn said the trial results suggested that sorafenib is worthy of further study for these patients. And he said the trial showed that biopsies biomarker results are feasible.
“Adoptive randomization can be improved and used in future studies, but follow-up trials with simplified treatment and marker groups are indicated,” Dr. Bunn said.
In a media conference after the oral presentations, coauthor Roy S. Herbst, MD, PhD, Professor and Chief of the Section of Thoracic Medical Oncology at M. D. Anderson, agreed that the trial protocol was complicated.
“But those were necessary assumptions that one had to do in order to get the trial under way,” Dr. Herbst said. “We've learned from this and we won't be bundling the markers, and we'll try to avoid population drift.”
Future BATTLE Trials
Dr. Herbst said the BATTLE-2 trial, now being planning with 320 patients, will use adaptive randomization but with only one biomarker, the KRAS mutation, and only three treatments: erlotinib, erlotinib plus an AKT pathway inhibitor, and an AKT inhibitor plus an MEK pathway inhibitor.
Dr. Kim, who paid tribute to Waun Ki Hong, MD, head of M. D. Anderson's Division of Cancer Medicine and principal investigator on the BATTLE grant from the US Department of Defense, as the originator of the BATTLE studies, added that also in the planning stages is BATTLE-3, as well as ultimately, a BATTLE-Prevention study.