What a Tumor-Agnostic Approval Means

A tumor-agnostic, or tissue-agnostic, approval defines an oncology treatment by a molecular feature found across multiple cancers rather than by one organ alone. The idea is powerful: an NTRK gene fusion can create a drug target in a salivary-gland tumor, sarcoma, thyroid cancer, or another solid tumor. The approval is not borderless. It is a carefully written indication with biomarker, stage, prior-treatment, age, and testing conditions.

The first US site-agnostic approval came in 2017, when pembrolizumab received accelerated approval for unresectable or metastatic solid tumors with microsatellite-instability-high or mismatch-repair-deficient status under label conditions. Larotrectinib followed for solid tumors with an NTRK gene fusion and additional qualifications. These precedents changed trial design without abolishing tumor biology.

Traditional approvals start with the organ#

Oncology has historically classified disease by anatomy and histology: breast carcinoma, lung adenocarcinoma, melanoma, and many narrower entities. Trials enroll one such disease, often at a defined stage and treatment line, then compare outcomes within that population.

Molecular oncology showed that some driver alterations recur across organs. A drug designed against the driver might work wherever that biology is dominant, and when the alteration is rare within each individual cancer, running a separate large trial for every organ can be impractical. Tumor-agnostic development reverses the organizing principle. The shared biomarker becomes the entry point, while tumor origin becomes a subgroup and source of heterogeneity.

The FDA definition is narrower than the phrase#

FDA's 2022 draft guidance defines a tissue-agnostic oncology drug as one targeting a specific molecular alteration or related alterations across multiple cancer types. “Across” requires evidence in more than one tumor context. “Specific” requires a biomarker that can be identified reproducibly.

The approval may still limit the indication to solid tumors, unresectable or metastatic disease, absence of satisfactory alternatives, progression after prior treatment, or a defined age. A resistance mutation can exclude use even when the headline fusion is present. So read the current prescribing information itself rather than a list of tumor-agnostic drugs. Indications can convert, expand, narrow, or be withdrawn over time.

Biomarkers can represent different biology#

Some biomarkers are oncogenic drivers, such as an NTRK fusion that creates constitutive kinase signaling, and others indicate a genomic state, such as mismatch-repair deficiency causing many mutations and potential immune recognition. Tumor mutational burden is a quantitative measure with assay and threshold complexity.

Two alterations in the same gene may not be equivalent. An activating mutation, amplification, fusion, and passenger variant can have different consequences. A report saying “NTRK mutation” is not the same as a qualifying NTRK fusion.

Pathway context matters too. A driver can be central in one tumor and bypassed by alternative signaling in another. Co-mutations can create primary resistance or shorten response.

The test is part of the treatment pathway#

A biomarker-defined indication works only when the biomarker can be measured accurately. Testing can use tumor DNA, RNA, immunohistochemistry, polymerase chain reaction, next-generation sequencing, or circulating tumor DNA, depending on the alteration and approved context.

Specimen age, fixation, tumor percentage, decalcification, low shedding into blood, and assay coverage can cause false negatives. DNA-only panels can miss some fusions that RNA methods detect. Immunohistochemistry can screen efficiently but may need confirmation.

FDA maintains a list of approved companion diagnostics for drugs where a specific test is essential to safe and effective use. Some labels permit broader validated testing. The oncology team and molecular laboratory should match the report method to the current label.

A negative result can be technically uninformative#

If a report says “not detected,” that means the assay did not find the alteration in the specimen it analyzed, within its limits. It does not always mean the tumor lacks it, and a small biopsy can contain little tumor, a blood test can have too little circulating DNA, or the assay may not cover the relevant intron or fusion partner.

Look on the report for specimen adequacy, the limit of detection, which genes and alteration types were covered, and any quality-control limitations. Orthogonal testing may be reasonable when the tumor type and clinical pattern make a missed alteration plausible. Repeating broad testing without a defined question can waste tissue and time. The decision turns on whether a different result would change your treatment.

Basket trials solve a rarity problem#

A basket trial enrolls participants with different tumor types that share a biomarker. Everyone may receive the same targeted treatment, or the protocol may include multiple cohorts and drugs. This aggregates rare patients who would be scattered across organ-specific trials.

The design can detect a strong, durable signal efficiently. Larotrectinib development enrolled adults and children with multiple NTRK fusion cancers and found high response rates across several histologies, supporting accelerated approval.

Efficiency creates statistical challenges. One common response rate can be dominated by more common baskets, while several tumor types contribute only a few participants, and a zero of two result and a five of six result are both imprecise.

Pooling requires biological justification#

Investigators must decide whether effects are similar enough to combine. Complete pooling assumes one common effect. Separate analyses discard shared information and can be unstable. Hierarchical models borrow information while allowing tumor-specific differences.

Borrowing should be prespecified and calibrated so a highly responsive tumor does not overwhelm a resistant group. The report should show overall and tumor-specific numbers, response duration, confidence intervals, and shrinkage assumptions. FDA guidance asks sponsors to address the strength of biological rationale, number and diversity of tumor types, consistency, and whether certain tumors should be excluded or studied separately.

Objective response is useful but incomplete#

Rare biomarker trials often lack a randomized control and use objective response rate. Independent radiology review may classify complete or partial tumor shrinkage using prespecified criteria. Duration of response asks how long that shrinkage lasts.

A high response rate with durable responses is compelling when historical spontaneous regression is rare; it still does not directly estimate overall survival benefit, symptom improvement, or comparative effectiveness against another treatment. Stable disease can matter but is more difficult to interpret without a control because some tumors naturally grow slowly. Time-to-event outcomes in single-arm studies also depend on patient selection and assessment timing.

Accelerated approval is a conditional evidence pathway#

US accelerated approval can rely on a surrogate or intermediate clinical endpoint reasonably likely to predict clinical benefit for a serious condition with unmet need. Oncology indications often use response rate and duration.

The sponsor must complete confirmatory studies to verify benefit. FDA's Project Confirm tracks oncology accelerated approvals, conversions, and withdrawals. If benefit is not confirmed, an indication can be changed or withdrawn. Tumor-agnostic and accelerated are not synonyms. A tumor-agnostic indication can have regular approval, and an organ-specific drug can receive accelerated approval.

The 2017 pembrolizumab precedent#

Pembrolizumab's 2017 indication for MSI-H or dMMR solid tumors was the first FDA cancer approval based on a shared biomarker rather than tumor site; it pooled evidence from several single-arm studies and tumor types in heavily treated disease.

Mismatch-repair deficiency impairs correction of DNA errors, producing microsatellite instability and many mutations. This can increase neoantigens recognized by the immune system, providing biological rationale for PD-1 blockade.

Response was not uniform across every tumor and patient. Immune-related adverse effects, organ transplant, autoimmune disease, performance status, and prior therapy still require clinical assessment. A biomarker gives you a probability, not a certainty.

Larotrectinib illustrates a driver target#

NTRK fusions join an NTRK gene with another gene, producing an active TRK fusion protein. They are common in a few rare tumors and uncommon in many common cancers.

FDA granted accelerated approval in 2018 for adult and pediatric patients with qualifying solid tumors that were metastatic or where surgery was likely to cause severe morbidity, had no known acquired resistance mutation, and lacked satisfactory alternatives or had progressed. Those conditions show why “NTRK positive” on your report is not a complete eligibility statement. Fusion type, resistance, resectability, treatment alternatives, and age all belonged to the decision.

Tissue of origin still shapes response#

Tumors with the same biomarker can differ in co-mutations, gene expression, immune environment, drug delivery, growth rate, and dependence on the target. Colorectal cancers with BRAF V600E, for example, require pathway combinations different from melanoma because feedback biology differs.

Anatomy determines whether surgery or radiation can cure or relieve symptoms. Organ function affects dosing and toxicity. Brain metastases raise central-nervous-system activity questions. Pediatric tumors require developmental and long-term safety considerations. Tumor-agnostic does not mean context-agnostic. It changes the first classification axis while preserving the others.

Resistance can be primary or acquired#

A tumor may carry the target but fail to respond because another pathway drives growth, the alteration is not functional, or drug concentration is inadequate, and after an initial response, a secondary target mutation or bypass pathway can restore signaling.

Repeat tissue or liquid biopsy at progression can sometimes identify resistance and support another targeted drug or trial. It can also fail because of low tumor content or spatial heterogeneity. Resistance evidence belongs in the label and treatment sequence. Reusing the same target class after a known resistance mutation requires drug-specific support.

Rare biomarkers make denominators important#

Reporting 75 percent response in a cohort of four means three people responded. The confidence interval is wide, and one additional patient would change the percentage sharply.

Waterfall plots show maximum tumor change but can omit patients without evaluable scans. Swimmer plots show response duration but are sensitive to follow-up. Look for a clear denominator beside both, plus the reasons for nonevaluable status and the data cutoff. That cutoff matters more than it looks: publication updates can lengthen apparent durability as follow-up matures, so two reports of the same cohort are only comparable at the same date.

Safety can differ across tumor populations#

Basket trials combine people with different prior treatments, organ function, ages, and disease burdens. A toxicity may be more consequential after thoracic radiation, in a child, or with liver involvement.

Small cohorts may miss rare serious harms. Postmarketing surveillance and confirmatory studies expand the safety database. Drug interactions and class-specific effects remain relevant regardless of biomarker.

The absence of a tumor-organ restriction does not remove contraindications, monitoring, or supportive care.

Molecular tumor boards help with uncertain reports#

Complex findings can include variants of uncertain significance, low allele fractions, multiple potential drivers, clonal hematopoiesis in blood, and biomarkers not validated for the tumor. A molecular tumor board brings oncology, pathology, genetics, laboratory science, pharmacy, and trial expertise together.

The board should distinguish approved use, guideline-supported off-label use, clinical trial, and hypothesis, and should document test validity and the exact evidence population; that help is worth most when a commercial report prints a list of therapies next to the variants without saying what evidence stands behind each one.

A label-first reading checklist#

Read the current indication word for word. Confirm the molecular alteration and the acceptable test. Check the tumor-type exclusions, the age, the stage, the prior treatment, the satisfactory alternatives, the resectability, and the resistance language.

Then go to the evidence: how many tumor types, how many participants per basket, the response rate, the duration, the follow-up, the nonevaluable cases, and whether the approval is accelerated. Look for the confirmatory status and any current safety updates.

Finally, put the result back inside organ-specific care. The biomarker can open a treatment door for you. It does not decide the whole sequence behind it.

References#

1. FDA draft guidance on tissue-agnostic oncology development
2. NCI list of drugs approved for biomarker-defined solid tumors
3. FDA account of the first site-agnostic approval
4. FDA larotrectinib approval for NTRK fusion tumors
5. FDA Project Confirm accelerated-approval tracker
6. FDA companion diagnostic list

For your own health, talk with your clinician.*