Moving Beyond Tumor Origin: The Rise of Tissue-Agnostic Oncology

The result is a more layered understanding of oncology. A patient may have a cancer defined by anatomy, histology, stage, prior treatment, and biomarker status at the same time. In some cases, the biomarker may open a pathway to a therapy or clinical trial that would not be visible through tumor origin alone. In others, the absence of a relevant biomarker may reinforce the need for conventional tumor-specific options. Tissue-agnostic oncology therefore does not erase the older map of cancer. It adds a molecular coordinate system that can reveal new treatment categories across traditional boundaries.

The difficulty of rare cancer has always been partly numerical. When a cancer affects relatively few people each year, the evidence base tends to be smaller, diagnosis can be delayed or difficult, and fewer trials may be available. Rare cancers are often harder to prevent, diagnose, and treat than more common cancers, and the small number of cases can make research difficult.³ Those challenges affect patients, clinicians, trialists, sponsors, regulators, and development partners.

The scale of rare cancer can be easy to underestimate because each individual tumor type may be uncommon. Viewed collectively, rare cancers represent a substantial burden. Approximately one-quarter of U.S. cancer diagnoses each year are rare cancers by the NCI definition.⁴ Another analysis summarized by the American Cancer Society found that about one in five cancer diagnoses in the United States is a rare cancer when using the fewer-than-six-per-100,000 threshold, and it noted that the proportion of cancers considered rare is likely to grow as molecular markers are increasingly used to classify cancers.⁷ That last point is central to the tissue-agnostic era. Molecular markers do not simply add detail to cancer classification; they can subdivide existing categories into smaller and more specific populations.

Historically, cancer classification has relied heavily on tissue type and primary site. That structure is clinically logical because tumors that arise in different organs often have different biology, symptoms, natural histories, standards of care, and outcome measures. For rare cancers, however, an organ-based framework can intensify fragmentation. Each rare diagnosis may have too few patients for large conventional trials, too little commercial attention for extensive development programs, and too little accumulated clinical experience to generate robust treatment pathways.

Biomarker-driven oncology introduces a different possibility. Instead of asking only whether a patient has a rare tumor type, clinicians and researchers can also ask whether the tumor belongs to a rare molecular subgroup with an available therapy, an investigational agent, or a basket trial. This shift does not make rare cancer less rare, nor does it make development straightforward. It does make some rare cancers more visible within a broader molecular landscape. A tumor that was once isolated within an uncommon anatomical category may become part of a cross-tumor population defined by the same actionable alteration.

Tissue-agnostic therapy is sometimes described as treatment independent of tumor origin, but that shorthand can be misleading if it is taken too far. A tissue-agnostic therapy is not a drug that works in every cancer. It is a therapy used across cancer types that share the specific genetic mutation, biomarker, or molecular pathway targeted by the drug.^1,2 The biomarker is not incidental. It is the reason the same therapy can be considered across multiple tumor types.

The U.S. Food and Drug Administration (FDA) has described tissue-agnostic oncology drug development as a distinct development approach because it raises issues that generally do not arise in more traditional tissue-specific programs.² In a conventional development model, a sponsor may study a therapy in one tumor type, define eligibility around that disease, and generate evidence within a relatively bounded clinical setting. In a tissue-agnostic model, the same molecular alteration may appear across several cancers, each with different prevalence, prior-treatment expectations, natural history, and standard-of-care context.

That complexity creates a central scientific question: when is a molecular alteration sufficiently important across tumor types to justify a tissue-agnostic development strategy? Some biomarkers may function as strong therapeutic drivers across multiple cancers. Others may be context-dependent, influenced by tumor lineage, co-occurring alterations, immune environment, disease stage, or prior therapy. Tissue-agnostic oncology therefore requires careful evidence generation rather than simple extrapolation.

The approved tissue-agnostic therapy landscape shows how selective this approach remains. The National Cancer Institute’s agnostic cancer therapies summary lists FDA-approved tissue-agnostic therapies that include pembrolizumab, dostarlimab, entrectinib, larotrectinib, repotrectinib, dabrafenib with trametinib, selpercatinib, pemigatinib, and trastuzumab deruxtecan.⁸ That list is meaningful, but it is ultimately a defined set of biomarker-based indications rather than a broad replacement for tumor-specific oncology.

For rare cancer, the practical value lies in the way tissue-agnostic therapy can create a treatment category that spans otherwise separate diseases. A patient with a rare tumor type may become eligible for a therapy because the tumor shares a biomarker with a larger, cross-tumor population. Conversely, a patient with a common tumor type may belong to a rare molecular subgroup. In both cases, rarity shifts from a purely anatomical concept to a more operational one: who has the alteration, who can be tested for it, who meets the label or trial criteria, and who can access the therapy.

The tissue-agnostic concept became clinically tangible through regulatory approvals that defined eligibility by molecular features rather than by a single tumor origin. Pembrolizumab’s approval for microsatellite instability–high (MSI-H) or mismatch repair–deficient (dMMR) solid tumors was a key early example. In 2017, the FDA granted accelerated approval to pembrolizumab for adult and pediatric patients with unresectable or metastatic MSI-H or dMMR solid tumors whose disease had progressed after prior treatment and who had no satisfactory alternative treatment options.^8,9 That approval was based on data from five uncontrolled, multicohort, multicenter, single-arm clinical trials that included 149 patients with MSI-H or dMMR cancers.⁸

Additional approvals extended the tissue-agnostic model across other biomarkers and therapeutic modalities. Larotrectinib received accelerated approval in 2018 for adult and pediatric patients with solid tumors harboring neurotrophic receptor tyrosine kinase (NTRK) gene fusions, provided the tumors met the specified clinical criteria in the label.¹⁰ Pembrolizumab later received accelerated approval for tumor mutational burden–high (TMB-H) solid tumors, with TMB-H defined in the FDA notice as at least 10 mutations per megabase by an FDA-approved test, in adult and pediatric patients with unresectable or metastatic disease after progression and with no satisfactory alternative treatment options.¹¹ Selpercatinib received approval for locally advanced or metastatic RET fusion–positive solid tumors in adults whose disease progressed after prior systemic treatment or who had no satisfactory alternative treatment options.¹² Fam-trastuzumab deruxtecan-nxki received accelerated approval in 2024 for adult patients with unresectable or metastatic HER2⁺ solid tumors who had received prior systemic treatment and had no satisfactory alternative treatment options.¹³

These approvals are not identical. They involve different biomarkers, drug classes, eligibility criteria, cancer settings, and evidentiary packages. Some are accelerated approvals, which carry specific postmarketing expectations. Some indications are restricted by prior treatment, disease status, test requirements, or absence of satisfactory alternatives. Nonetheless, they share a common organizing principle: the clinically relevant population is defined at least in part by a molecular feature that can appear across tumor types.

That matters because regulatory categories influence development behavior. Once a biomarker-defined, cross-tumor indication becomes possible, sponsors can consider development strategies that do not begin and end with one tumor origin. They can ask whether a molecular alteration is present across cancers, whether a drug’s mechanism plausibly maps to that alteration, whether evidence can be generated across multiple cohorts, and whether a diagnostic approach can reliably identify eligible patients. In rare cancer, that framework may create a path for groups that are too small to support conventional evidence generation within a single tumor type.

Tissue-agnostic oncology depends on the ability to identify the relevant alteration. Biomarker testing looks for genes, proteins, and other substances that can provide information about a person’s cancer, and it may help clinicians choose treatment.¹⁴ Some cancer treatments, including targeted therapies and immunotherapies, may work only for people whose cancers have certain biomarkers.¹⁴ In a tissue-agnostic indication, testing is not merely descriptive. It can determine whether the patient belongs to the treatment-eligible population.

The language around testing can vary. Biomarker testing may also be called tumor testing, tumor genetic testing, genomic testing, genomic profiling, molecular testing, molecular profiling, somatic testing, or tumor subtyping. The practical point is that the test must produce information that can guide treatment, trial matching, or both. Some tests evaluate a single biomarker. Others assess multiple genes or markers at once. Some are designed for one cancer type, while others can look for biomarkers found across many cancers.¹⁴

This creates both opportunity and friction. Biomarker testing can identify a therapy approved for the patient’s cancer type, a therapy approved for another cancer type with the same biomarker, an off-label option, or a clinical trial. It can also return findings that do not guide treatment, reveal a biomarker with no available matched therapy, or identify a treatment that is inaccessible because of coverage, geography, clinical status, or trial availability. Testing can be limited by insufficient tissue, inability to obtain a biopsy, tumor heterogeneity, changes in biomarkers over time, and the fact that even a matched biomarker does not guarantee response.¹⁴

For rare cancer, these issues can be more consequential. A rare cancer patient may have fewer standard options, fewer disease-specific trials, and less established guidance for sequencing therapies. If a tissue-agnostic therapy is available for the tumor’s molecular profile, testing may be the pathway to finding it. If testing is not performed, is delayed, is too narrow, or does not capture the relevant alteration, the patient may never be recognized as part of the actionable population.

Clinical guidance has begun to reflect this reality. The American Society of Clinical Oncology (ASCO) addresses the use of tumor genomic testing in patients with metastatic or advanced solid tumors and states that site-agnostic approvals for high TMB, mismatch repair deficiency, and NTRK fusions provide a rationale for genomic testing for all solid tumors.¹⁵ That statement should not be stretched into a claim that comprehensive genomic profiling is mandatory for every patient in every setting. However, it does support the narrower and more useful point that tissue-agnostic approvals have made genomic testing relevant beyond tumor-specific biomarker paradigms.

The most important conceptual shift in tissue-agnostic oncology may be the emergence of rare actionable populations. A rare cancer has historically been defined by how many people are diagnosed with that cancer type within a population. Biomarker-driven oncology adds a second kind of rarity: how many patients, across one or many cancer types, harbor a specific alteration that can guide therapy.

Those two forms of rarity overlap but are not the same. A tumor type can be rare by incidence and yet contain a targetable alteration. A tumor type can be common, while a particular molecular subset within it is rare. A biomarker can be uncommon within each tumor type, but meaningful when patients are pooled across tumor types. This is where tissue-agnostic oncology begins to reshape the practical meaning of rare cancer without changing its formal definition.

The American Cancer Society report summary makes this point especially relevant by noting that the proportion of cancers considered rare is likely to grow as the use of molecular markers to classify cancers increases.⁷ That statement does not say that molecular markers have replaced incidence-based rare-cancer definitions. It does suggest that classification itself is becoming more granular. As more cancers are subdivided by molecular features, the field may identify more groups that are small, clinically distinct, and potentially actionable.

This has consequences for research strategy. In an organ-based model, a rare cancer program may struggle because each tumor type is isolated from others. In a biomarker-based model, patients with different tumor origins can sometimes be studied together if they share the relevant alteration. That does not remove the need to understand tumor-specific context. It does create a way to ask whether the molecular feature is strong enough to support treatment across cancers.

It also changes the patient journey. A diagnosis may still begin with the anatomical tumor type, but the treatment question can quickly become molecular: does this tumor have MSI-H or dMMR status, high TMB, an NTRK fusion, a RET fusion, a BRAF V600E mutation, HER2 expression, or another relevant feature? The answer can place the patient within a group that cuts across the usual categories. For rare cancer patients, that group may be the difference between an exhausted disease-specific pathway and an available biomarker-defined option.

Tissue-agnostic oncology requires trial designs that can follow the biomarker across tumor types. Traditional trials often begin with one disease and evaluate a therapy within that population. Tissue-agnostic development may need to evaluate activity across multiple cancers, some common and some rare, while still accounting for differences in disease course, prior treatment, response assessment, and available alternatives.

Master protocols and basket trials are part of this evolution. FDA master protocol guidance covers trial designs that evaluate more than one investigational drug and/or more than one cancer type within one overall trial structure.¹⁶ Biomarker testing can also support trial matching, and the NCI describes basket trials as studies that enroll people based on biomarkers in their cancer rather than where in the body the cancer started growing.¹⁴ These approaches are particularly relevant when the molecular alteration is uncommon within any one tumor type but appears across several.

NCI-MATCH provides one example of genomically guided trial design. The trial used genomic sequencing to help plan treatment for people with advanced cancer and enrolled 1,201 people across 38 treatment arms.¹⁷ That kind of design reflects a broader shift toward matching treatment to alterations rather than relying solely on tumor origin. It also illustrates the logistical and analytical complexity of precision oncology: patients must be screened, alterations must be interpreted, treatment arms must be available, and outcomes must be assessed across biologically and clinically diverse cohorts.

The ROAR trial offers a rare-cancer-specific example. ROAR evaluated dabrafenib plus trametinib in BRAF V600E–mutated rare cancers and included several rare tumor cohorts, such as anaplastic thyroid cancer, biliary tract cancer, gastrointestinal stromal tumor, small-intestine adenocarcinoma, gliomas, germ-cell tumors, hairy cell leukemia, and multiple myeloma.¹⁸ ROAR was a multicenter, single-arm, open-label phase II basket study conducted across community and academic cancer centers in multiple countries, and its findings supported a tumor-agnostic approach to BRAF V600E inhibition in diverse rare cancer cohorts.¹⁸

These trial models can help address fragmentation, but they do not eliminate it. A basket trial still needs sufficient enrollment, reliable assays, meaningful endpoints, careful cohort design, and interpretable evidence across tumor contexts. A response signal in one tumor type may not automatically apply to all others with the same alteration. Regulators, sponsors, and clinicians still need to evaluate whether the biomarker is acting as a consistent therapeutic driver and whether the available evidence supports the proposed population.

Tissue-agnostic oncology has practical implications well beyond trial nomenclature. When a development program is organized around a molecular alteration rather than a single tumor origin, the sponsor may need to plan for a patient population that is biologically defined, geographically dispersed, and unevenly distributed across cancer types. This can affect clinical strategy, diagnostic planning, site selection, patient identification, supply forecasting, and evidence generation.

The FDA guidance states that tissue-agnostic drug development raises issues that generally do not arise in more traditional development approaches.² From a development perspective, those issues can begin early. A sponsor must determine whether the target alteration is sufficiently well understood, whether it is present across multiple tumor types, whether the therapy has a plausible mechanism across those cancers, and whether a testing strategy can identify eligible patients. The development plan may also need to account for small cohorts, variable standards of care, and different lines of therapy across tumor types.

This creates a different type of operational pressure. Biomarker-defined populations may be small, but they can span many sites, countries, and cancer centers. Trial enrollment may require broad screening, careful coordination with testing laboratories, and protocols that can accommodate multiple tumor cohorts. Clinical supply plans may need flexibility because enrollment can be uneven across cohorts and may change as additional tumor types or biomarkers are added. These realities can influence how sponsors choose development partners, contract research organizations, diagnostic collaborators, and contract development and manufacturing organizations (CDMOs).

For CDMOs, the most relevant implications are not unique to tissue-agnostic oncology, but the tissue-agnostic model can intensify them. Programs may need to move quickly from early signals to expanded cohorts or additional indications. Sponsors may need small-batch clinical supply, scalable manufacturing strategies, robust analytical methods, and comparability planning that can support an evolving clinical program. For biologics, antibody–drug conjugates, targeted small molecules, and immunotherapies, the development pathway may require close coordination between clinical evidence, regulatory strategy, diagnostic eligibility, and chemistry, manufacturing, and controls.

The rare-cancer connection heightens these needs. Smaller populations can make timelines, enrollment, and supply assumptions less predictable. At the same time, successful biomarker-defined development can expand a program beyond the first tumor type or cohort. A sponsor may therefore need a development model that preserves optionality: the ability to support narrow initial use, generate additional evidence, and scale or adapt if the biomarker-defined opportunity grows.

Tissue-agnostic oncology is powerful, but its limits are as important as its promise. It does not mean that a biomarker always overrides tumor context. It does not mean that every patient with a target alteration will respond. It does not mean that every tissue-agnostic approval has the same evidence base, the same durability of response, or the same confirmatory requirements. It also does not mean that rare cancer has become a purely molecular category.

Biomarker testing itself has limitations. Testing may not help every patient, and the NCI describes several reasons why, including inability to safely obtain tissue, insufficient tumor material, lack of a matched available therapy, coverage barriers, or inability to participate in a relevant clinical trial. A matched biomarker does not guarantee clinical benefit, because other features of the cancer or the patient can affect treatment response. Tumors may also be heterogeneous, and biomarkers can change over time, meaning a prior test may not always reflect the current disease state.¹⁴

These caveats matter because tissue-agnostic oncology can be oversimplified. A biomarker-defined approval is not the same as universal molecular determinism. A therapy may have strong activity in one biomarker-defined population and more variable activity in another. Even when the same alteration is present, tumor lineage and biological context may influence response. Because of that, tissue-agnostic development requires evidence that is broad enough to support cross-tumor use and specific enough to avoid assuming that all cancers with the same alteration behave identically.

The formal definition of rare cancer also remains intact. Rare cancer is still defined by incidence thresholds, and those thresholds vary across frameworks.^3,5 What is changing is the practical landscape in which rare cancers are diagnosed, subdivided, and treated. Molecular markers can create smaller categories within already rare cancers and can identify rare actionable subgroups within common cancers. That makes rarity more complex, not obsolete.

The rise of tissue-agnostic oncology points toward a more layered future for cancer classification and development. Tumor origin, histology, stage, and clinical context remain central. Biomarker status adds another layer that can define treatment eligibility, trial access, and regulatory strategy. For rare cancer, this additional layer may be particularly valuable because it can connect patients across otherwise separate diagnoses.

The most precise way to describe the shift is not that tissue-agnostic oncology has redefined rare cancer, but that it has reframed how rarity can be acted upon. Incidence-based definitions still describe population burden. Molecular definitions can identify treatment-relevant subgroups. When those two frameworks are combined, rare cancer becomes both an epidemiologic category and a development challenge shaped by biology, testing, trial design, and access.

That reframing has implications for every stakeholder in oncology drug development. Clinicians need to know when and how to test. Patients need access to appropriate testing and biomarker-informed trials. Sponsors need development strategies that can follow actionable alterations across tumor types without ignoring tumor-specific context. Regulators need evidence packages that support cross-tumor claims while preserving appropriate standards for safety and effectiveness. CDMOs and other development partners need operational models that can accommodate smaller, more adaptive, and potentially expanding programs.

Tissue-agnostic oncology does not simplify cancer. In many ways, it makes the field more complex by adding molecular eligibility to an already intricate system of diagnosis and treatment. Yet, that added complexity can be clinically productive. It can reveal treatment opportunities that tumor origin alone may obscure, and it can create development pathways for patient populations that are too fragmented when viewed only through the lens of anatomy.

For rare cancer, the promise is not that every rare tumor will suddenly have a targeted option. The promise is that some patients may be found through a different kind of classification: not only what cancer they have, but what molecular feature their cancer carries. As tissue-agnostic therapies, biomarker testing, and cross-tumor trial designs continue to mature, the definition of a treatable rare cancer population may increasingly depend on both where the cancer began and what makes it vulnerable.

1. “Definition of Tissue-Agnostic Therapy.” National Cancer Institute Dictionary of Cancer Terms. Accessed 24 Jun. 2026.

2. Tissue Agnostic Drug Development in Oncology: Draft Guidance for Industry. U.S. Food and Drug Administration. 17 Oct. 2022.

3. “Definition of Rare Cancer.” National Cancer Institute Dictionary of Cancer Terms. Accessed 24 Jun. 2026.

4. “Rare Cancer Webinar Series.” National Cancer Institute Epidemiology and Genomics Research Program. 8 Sep. 2025.

5. “Definition of Rare Cancers.” European Society for Medical Oncology. Accessed 24 Jun. 2026.

6. “Rare Cancer Classification — SEER Recodes.” Surveillance, Epidemiology, and End Results Program. Accessed 24 Jun. 2026.

7. “One in Five Cancers Diagnosed in the United States Is a Rare Cancer.” American Cancer Society. 19 May 2017.

8. “Agnostic Cancer Therapies (PDQ^®) — Health Professional Version.” National Cancer Institute. 12 Feb. 2025.

9. Jaber, Nadia. “FDA Approves Pembrolizumab for Tumors with Specific Genetic Features.” National Cancer Institute. 20 Jun. 2017.

10. “FDA Approves Larotrectinib for Solid Tumors with NTRK Gene Fusions.” U.S. Food and Drug Administration. 14 Dec. 2018.

11. “FDA Approves Pembrolizumab for Adults and Children with TMB-H Solid Tumors.” U.S. Food and Drug Administration. 17 Jun. 2020.

12. “FDA Approves Selpercatinib for Locally Advanced or Metastatic RET Fusion-Positive Solid Tumors.” U.S. Food and Drug Administration. 21 Sep. 2022.

13. “FDA Grants Accelerated Approval to Fam-Trastuzumab Deruxtecan-nxki for Unresectable or Metastatic HER2-Positive Solid Tumors.” U.S. Food and Drug Administration. 5 Apr. 2024.

14. “Biomarker Testing for Cancer Treatment.” National Cancer Institute. 14 Dec. 2021.

15. Chakravarty, Debyani, et al. “Somatic Genomic Testing in Patients With Metastatic or Advanced Cancer: ASCO Provisional Clinical Opinion.” Journal of Clinical Oncology. 40: 1231–1258 (2022).

16. Master Protocols: Efficient Clinical Trial Design Strategies to Expedite Development of Oncology Drugs and Biologics Guidance for Industry. U.S. Food and Drug Administration. 2 Mar. 2022.

17. “NCI-MATCH Precision Medicine Clinical Trial.” National Cancer Institute. 14 Dec. 2023.

18. Subbiah, Vivek, et al. “Dabrafenib plus Trametinib in BRAFV600E-Mutated Rare Cancers: The Phase 2 ROAR Trial.” Nature Medicine. 29: 1103–1112 (2023).