A basket trial is a type of clinical trial that tests a single drug across multiple types of cancer, as long as those cancers share the same genetic mutation or biomarker. Instead of organizing patients by where their cancer started (lung, breast, colon), a basket trial groups them by what’s driving their cancer at the molecular level. This represents a fundamental shift in how cancer treatment is studied: treating the mutation, not the organ.
How Basket Trials Work
In a traditional cancer trial, everyone enrolled has the same type of cancer. A lung cancer drug is tested in lung cancer patients, a breast cancer drug in breast cancer patients. A basket trial flips that logic. Patients with lung cancer, thyroid cancer, sarcoma, and a dozen other tumor types might all enroll in the same trial, provided their tumors carry a specific genetic change that the drug is designed to target.
Each cancer type forms its own “basket” within the trial. Everyone receives the same treatment, and researchers track how well the drug works in each basket separately. Some baskets may show strong responses while others show little benefit, even though the same mutation is present. This structure lets researchers quickly see which cancers are most responsive to targeting that particular mutation.
Why Genetic Testing Made This Possible
Before 2005, testing a tumor for genetic changes was slow and narrow. Clinicians could only check for one mutation at a time, so tumors were typically tested only for biomarkers already validated for that specific cancer type. A lung tumor might be checked for known lung cancer mutations, but nobody was looking for rare mutations shared across unrelated cancers.
That changed with next-generation sequencing, which can evaluate hundreds of genes simultaneously. Commercial sequencing panels now cover more than 500 cancer-associated genes in a single test. When a patient’s report comes back, the key question becomes: does this tumor have a mutation that a drug could target? If so, a basket trial may be an option, regardless of where the cancer originated.
Real Trials and What They Found
One of the landmark basket trials tested a drug called vemurafenib in patients with a specific BRAF mutation across multiple non-melanoma cancers. (Vemurafenib was already approved for melanoma with this mutation.) The results, published in the New England Journal of Medicine, showed meaningful responses in several cancer types: a 42% response rate in non-small-cell lung cancer and 43% in two rare diseases, Erdheim-Chester disease and Langerhans cell histiocytosis, which had no approved therapies for adults at the time. But colorectal cancer patients with the exact same mutation showed zero responses, illustrating that the organ of origin still matters.
The NCI-MATCH trial, one of the largest basket-style efforts ever conducted, screened nearly 6,000 patients and assigned 1,473 to one of 38 treatment arms, each matching a drug to a specific genomic change. Running from 2015 to 2023, it served as a massive test of the principle that molecular profiling could guide treatment across cancer types.
Perhaps the most striking success story involves a class of drugs targeting NTRK gene fusions. Trials of the drug larotrectinib enrolled 244 patients across 25 different tumor types, including soft tissue sarcoma, thyroid, lung, and salivary gland cancers. The drug worked so consistently across cancer types that it received a “tissue-agnostic” approval, meaning it can be prescribed for any cancer carrying an NTRK fusion, regardless of where it started. That kind of approval was essentially unheard of before basket trials.
How Basket Trials Differ From Umbrella Trials
Both basket and umbrella trials use molecular screening to match patients with treatments, but they’re structured in opposite directions. A basket trial takes one drug and tests it across many cancer types that share a molecular target. An umbrella trial takes one cancer type and tests multiple drugs, each matched to a different molecular subgroup within that cancer. Think of it this way: basket trials organize around a single treatment, umbrella trials organize around a single disease.
The Tissue-of-Origin Problem
The core promise of basket trials is that a shared mutation means a shared vulnerability to the same drug. In practice, that assumption often breaks down. Response rates can vary significantly between tumor types even when the same mutation is present. Breast and ovarian cancers, for example, tend to respond better than sarcomas or head and neck cancers in many basket trial settings.
This happens because cancer biology is more complex than any single mutation. The tissue where a cancer develops brings its own signaling pathways, immune environment, and resistance mechanisms. A mutation might be the primary growth driver in one cancer type but a secondary player in another. Researchers now account for this when designing trials, sometimes excluding tumor types with predicted resistance from specific baskets.
Not all mutations are equally good targets, either. Mutations that develop early in a tumor’s evolution (called truncal alterations) tend to be present in virtually all of the cancer cells, making them better drug targets. Mutations that appear later may only exist in a small fraction of cells, limiting how much damage a targeted drug can do. Even different mutations within the same gene can respond differently to the same drug, a nuance that trial designers increasingly build into their eligibility criteria.
Why Basket Trials Matter for Rare Cancers
For patients with rare cancers or uncommon mutations, basket trials solve a practical problem that traditional trial designs cannot. If a mutation appears in only a small percentage of any single cancer type, there simply aren’t enough patients to run a traditional trial focused on one organ. A basket trial pools patients across many cancer types, reaching enrollment numbers that would be impossible disease by disease.
This has been transformative for genuinely rare diseases. The NTRK fusions targeted by larotrectinib are enriched in conditions like infantile fibrosarcoma and secretory breast carcinoma, diseases so uncommon that dedicated trials would never have been practical. Without the basket trial framework, effective treatments for these cancers might never have been identified or approved.
How Results Are Analyzed
Basket trials face a statistical challenge: each basket may contain only a handful of patients, making it hard to draw reliable conclusions from any single group. To address this, many basket trials use a Bayesian statistical approach, which allows information to be shared across baskets. If several cancer types are responding similarly to the drug, that collective evidence strengthens the conclusions for each individual basket.
This “information borrowing” works by assuming the baskets may have related response rates and then letting the data reveal how similar or different they actually are. When baskets behave similarly, the shared analysis produces more stable estimates than analyzing each tiny group alone. When one basket clearly diverges, the statistical model recognizes that and treats it more independently. The primary measure of success in most basket trials is the objective response rate: the percentage of patients whose tumors shrink meaningfully on imaging.