On May 23rd 2017, the cancer research community applauded a landmark accomplishment when for the first time the U.S. Food and Drug Administration (FDA) approved Merck’s pembrolizumabfor use in tumors located in any tissue. While cancers are traditionally organized and treated based on their tissue of origin, tumor sequencing has demonstrated that there are important genetic and molecular similarities between tumors across tissue types. These data prompted researchers to hypothesize that molecularly targeted therapies could be effective independent of a disease’s tissue of origin or histology. This means that drugs that have been approved for one cancer type may be effective in others as well as long as the molecular target is present, leading to exciting new avenues of treatment across all cancers.
In the past, cancer drug approval in different diseases was accomplished by carrying out individual clinical trials for each tumor type. The FDA has approved almost 70 targeted therapeutic agents, many for use in multiple cancers, including trastuzumab (approved in both breast cancer and adenocarcinoma of the stomach) and imatinib (first approved in leukemia and later approved in four additional cancers). However, the traditional clinical trial paradigm is prohibitive to high-throughput drug development; standard clinical trials are designed to test the efficacy of one drug in one disease at a time, a process that can take between 5-10 years per tissue type.
While the same drug can be approved for different diseases through multiple, time consuming, iterations of this process, researchers are designing new, broader, clinical trial paradigms to speed up the approval of drugs in other diseases. One of the most exciting of these new methodologies is the basket trial, which allows clinicians to investigate multiple drugs or diseases under the same clinical trial framework, each in their own “basket.” By analyzing a patient’s tumor response in the context of their basket, the trial can fluidly evolve to match how patients respond; baskets that do not respond well can be discontinued, while those demonstrating a good drug response can be immediately expanded to test the drug in a larger group. If a drug is shown to be effective regardless of tumor location, it may be approved for use in drug-target containing tumors located in any tissue, even those not included in the basket trial (tissue-agnostic use).
The first cancer drug to garner FDA approval for tissue-agnostic use is pembrolizumab. Pembrolizumab was first approved for treatment of patients with melanoma, and later for use in lung and several other solid tumor cancers, each after an independent traditional clinical trial. To investigate pembrolizumab in additional tissue sites, Merck initiated a basket trial investigating pembrolizumab’s efficacy in patients with 15 different cancer types, all expressing pembrolizumab’s target, PD-1. Based on the positive results of these trials, the FDA granted pembrolizumab accelerated approval, meaning it will undergo further study but is immediately available to patients whose disease will progress without it. This tissue-agnostic drug approval has been met with tremendous enthusiasm from the cancer research community, and may be the next step toward delivering on the promise of personalized medicine. The advent of next generation sequencing has made it easier, cheaper, and faster to sequence a patient’s tumor as a routine test when performing biopsies, and identification of a tumor’s exact genetic landscape means that the most important mutations and targets can be identified. This allows doctors to create an individualized treatment regimen according to a patient’s tumor-specific mutations, and the approval of drugs for tissue-agnostic use significantly increases the number of drugs available for targeted treatment.
Increased implementation of the basket trial design will benefit patients that are the underserved by classic clinical trials. Often it is difficult to recruit enough participants for clinical trials testing rare mutation-targeting drugs leaving patients with these mutations without effective treatment options. Basket trials can expand the population of patients eligible for a trial, benefiting those with less common diseases. Moreover, trials can be conducted in a shorter time, with fewer patients, and at lower cost than traditional trial designs.
Despite these benefits, basket trials do have disadvantages. First, the presence of a molecular target in a tumor does not guarantee it plays an important role or that inhibition of that target will have a beneficial effect for the patient. This was recently demonstrated with the BRAF inhibitor vemurafenib. While it is highly effective in BRAF mutation-carrying, metastatic, melanoma patients, it is ineffective in colorectal cancer with mutated BRAF. Basket trials also increase the statistical complexity of a clinical trial. Each basket of the trial carries its own chance of resulting in a false positive, which combine to increase the rate of false positives for the trial as a whole. Finally, it’s nearly impossible to simplify tumors down to one or two targetable mutations. The biological context of mutations impacts the role that a mutation has in the tumor, and basket trials will only show that a treatment works if the tumor’s survival relies on the targeted pathway.
We remain on the cusp of a true paradigm shift in personalized genomics and medicine, and the approval of pembrolizumab for tissue-agnostic use is the first glimpse of this future. Forty three percent of all tumors share more genetic similarities to tumors from different tissues than tumors from their own tissue of origin, and yet we still classify and treat cancers based on their tissue of origin. This is a rudimentary construct for what are incredibly complicated diseases. Basket trials are the first clinical reflection of the need to change this, and will set the foundation for more innovative clinical trial design to fulfill the hopes of personalized cancer treatment.
Originally published in the AACR Cancer Policy Monitor