Jun 24, 2014
A new approach to clinical trial design determines the optimal targeted therapy for patients who harbor the associated genetic target. In so-called “basket” trials, next-generation sequencing (NGS) is used to identify patients with a specific genetic alteration or mutation, regardless of the type of cancer they have. These patients can then be enrolled in a trial to assess a particular molecularly targeted therapy.
David B. Solit, MD, of Memorial Sloan Kettering Cancer Center, described the development of this concept, at his institution and elsewhere, during the Pre-Annual Meeting Seminar, “Genetics and Genomics for the Practicing Clinician,” held immediately before the ASCO Annual Meeting.
Dr. Solit said the purpose of a basket trial is to test the hypothesis that a particular mutation predicts response to a given drug. “The challenge in the clinic right now is to identify patients who have alterations in a particular target so that you can match them up with a drug,” Dr. Solit said.
Dr. Solit pointed out that the traditional method to identify drug targets was by analysis of data from large, retrospective studies. If a substantial subset of a study population has a particular genetic alteration associated with a response, a drug can be developed to target that alteration. Clinical trials can then be conducted to verify the association. Dr. Solit noted the example of the identification of the BRAF mutation, and subsequently, RAF inhibitors have become part of the standard of care.
Rather than use this genotype-to-phenotype approach, basket trials use a phenotype-to-genotype strategy. This is helpful to identify therapies for rarer, harder to characterize cancers.
As an example, Dr. Solit described an extraordinary responder in a trial of the mTOR inhibitor everolimus for metastatic bladder cancer. In the trial as a whole, the drug was associated with no improvement in progression-free survival, the primary endpoint. But several patients responded to the drug, including one patient with metastatic bladder cancer, who has experienced a complete response for more than 4 years with everolimus treatment. Whole genome sequencing of that patient’s tumor identified two mutations that, together, activate mTORC1, the target of everolimus.
“We didn’t know that molecular information going into the clinical trial,” Dr. Solit said. “We now have an institution-wide effort to identify these extraordinary responders, and there is talk of a nationwide effort supported by the National Cancer Institute.”
The paradigm for creating a basket study is to first run a phase II trial of a drug in an unselected population. If a few patients benefit, NGS is performed to identify molecular characteristic that distinguish those patients. After laboratory work validates the functional importance of the alteration, a test is designed to prospectively screen patients. Those who have the target alteration can be enrolled in a study restricted to patients with that particular gene or mutation.
“So have patients with many cancer types all in the same study, and statistically it becomes a group of parallel, phase II studies,” Dr. Solit said. This is the design of the vemurafenib basket study (VE-BASKET), he noted.
Basket Study Design
One approach to the use of genotyping in study design is to center on a particular cancer. All participants are genotyped, and those with mutations associated with a targeted drug (EGFR, BRAF, etc.) are allocated to separate arms of the study to receive the appropriate drug. These studies are difficult to organize, Dr. Solit said, and involve multiple drugs from multiple companies. Study design can take years, and by the time the study begins some of the chosen drugs may no longer be best in class. In addition, the number of patients with any one mutation is generally low, and thus this design may not identify sufficient numbers of patients with rare mutations to test whether these mutations correlate with drug response.
By contrast, in a basket study, patients with multiple types of cancer are enrolled as long as they have the particular molecular alteration of interest.
“The best types of targets are alterations associated with many different diseases, but in each case, altered in only a small percentage of patients,” Dr. Solit said. He gave the example of the ERBB2 gene, which is amplified or mutated in multiple cancer types.
“An analysis of more than 7,000 exomes found ERBB2 mutations across many types of cancers,” he said. “It was most common in bladder cancer, but only in 6% of samples in that disease, and only 1% to 2% in many other diseases. You wouldn’t want to open up a trial in every one of these cancers, but you can open up one study that allows patients with any of those cancers to enroll.”
An advantage of the basket trial methodology is that it allows testing of a defined biologic hypothesis: Do patients with this mutation respond to this drug? Tissue can be collected to determine the basis for heterogeneity of response. This design also allows flexibility. By including an “other” cohort, arms can be added for diseases not previously anticipated. Cohorts for specific mutations can be added and the trial can be amended to allow a specific combination therapy for a particular disease.
Challenges with this approach include criticisms that the trial may fail to identify patients who may potentially respond to a drug, but lack the biomarker being tested. In addition, Dr. Solit said, there is the “sad fact” that getting multiple disease teams to work together can be difficult. Most of all, the primary hurdle is identifying patients who are candidates for the study. The screening protocol should be separated from the treatment protocol, Dr. Solit said, although this is a polarizing concept.
Initiative to Find Patients
As part of a commitment to the identification of extraordinary responders, Dr. Solit said, Memorial Sloan Kettering Cancer Center has committed to screening every patient with metastatic disease in the next 12 months-—estimated to be 10,000 to 12,000 patients—using an NGS assay, the Integrated Mutation Profiling of Actionable Cancer Targets (IMPACT). The assay, which probes for 341 cancer genes, will be used to screen for somatic mutations through the analysis of tumor and normal pairs.
“The goals are to identify a driver alteration in every single patient and to identify germline variants that could contribute to cancer initiation,” Dr. Solit said. The major hurdle is financial, as genotyping of any kind is not standard of care for most cancers. “We believe there are only nine diseases we can bill for, so we have rallied a lot of philanthropic and institutional support for doing other cancers. The endpoint will be to determine whether this profiling has an impact on our care of patients.”