Running a basket trial operationally requires infrastructure that treats each tumor type cohort as a distinct operational unit within a shared governance framework: cohort-specific site selection (not study-level), a central IRB operating under a master protocol, biomarker screening designed for the sample type and testing requirements of each tumor type, and cross-cohort safety oversight that can assess signals at both cohort and study level simultaneously. Sponsors who apply a single-indication operational model to a basket trial encounter predictable and avoidable failures at site selection, screening logistics, and safety governance.
—
Why Is a Basket Trial More Demanding Than Running Multiple Separate Trials?
A basket trial — evaluating one therapy across multiple tumor types based on a shared molecular target — is one of the most operationally demanding designs in clinical oncology. The scientific logic is elegant: if the drug’s mechanism is driven by a molecular alteration rather than a tissue of origin, it should work across cancer types. The operational consequence is that you are running multiple parallel enrollment programs simultaneously, under one protocol, with shared governance but distinct site and patient populations for each cohort.
Sponsors who approach basket trial execution with the same operational model they would apply to a single-indication study encounter predictable and avoidable problems.
—
Why Is Governance the Non-Optional Starting Point?
The operational challenge of basket and umbrella trials is fundamentally a governance challenge. These designs require a protocol that creates a bridge across all of the multiple stakeholders — regulatory agencies, potentially multiple sponsors, genomic test providers, academic investigators, and clinical sites.
Without explicit governance design — a steering committee with defined authority, an escalation path for cross-cohort decisions, and a central IRB operating under a master protocol — the coordination overhead becomes the primary source of delay, not the science and not the sites.
Similarly, “a central IRB is the best way to ensure efficient implementation of these studies,” specifically because local IRB review “can significantly prolong the time to trial activation” when applied cohort by cohort (Sitero oncology program data).
Kyle Hanson, Director of Clinical Operations, Sitero:
“You have to separate the reason for the imbalance before you apply a solution — a cohort stalling due to low prevalence is a completely different problem from one stalling due to site capability gaps or screening failures. For prevalence-driven stalls, the lever is typically site network expansion with targeted selection — adding sites with documented patient populations for that tumor type, often community oncology networks that see higher volume for specific histologies than academic centers. For capability-driven stalls, the intervention is operational: site coaching, re-training on biomarker workflows, or activating a regional CRA specifically for that cohort. The engagement risk no one talks about: sites that enroll heavily in one cohort but have patients in slow cohorts start to deprioritize the trial overall when they feel their biomarker-positive patients for the slow cohort ‘never go anywhere.’ Proactive site communication about cohort status and realistic enrollment projections is what keeps that from becoming a site relationship problem.”
—
Why Must Site Selection Be Done Per Cohort, Not Per Study?
In a single-indication trial, site selection is done once across a relatively homogeneous patient population. In a basket trial, each tumor type cohort requires its own site selection logic — different patient throughput requirements, different biomarker testing infrastructure, and potentially different geographic distribution based on where the relevant patient population is concentrated.
Treating site selection as a study-level activity in a basket trial produces two predictable failure modes:
- Sites are activated that can enroll one cohort quickly but have no pathway to the patient population for another cohort
- Resources are concentrated in sites for fast-filling cohorts while rare tumor type cohorts stagnate
Sitero’s approach to basket trial execution treats each tumor type cohort as a distinct operational unit within the shared study framework — site selection conducted per cohort, with targeted feasibility assessments for each indication’s specific requirements (Sitero oncology program data).
—
Basket vs. Single-Indication Trial: Operational Comparison
| Operational Dimension | Single-Indication Trial | Basket Trial |
|---|---|---|
| Site selection | Study-level, one patient population | Per-cohort; distinct feasibility for each tumor type |
| Biomarker screening | One assay, one lab, one sample type | May differ by cohort; assay and sample type optimization per tumor type |
| IRB model | Local IRB standard | Central IRB strongly preferred (Sitero oncology program data) |
| Safety oversight | Study-level aggregate | Both cohort-level and cross-cohort aggregate required simultaneously |
| EDC design | Single eligibility workflow | Cohort-specific eligibility logic with shared governance data structures |
| Governance complexity | Sponsor + sites + IRB | Adds steering committee, cross-cohort IDMC, potential multi-sponsor structure |
| Enrollment variance | Relatively predictable | Cohort-specific variance; some cohorts may fill while others stagnate |
—
How Should Biomarker Screening Work Across Multiple Tumor Types?
Basket trials require biomarker screening across multiple cancer types, each potentially with different tissue requirements, testing platforms, and turnaround expectations. Two models exist:
Unified screening model: Single central lab, single assay panel across all cohorts. Operationally simpler, but may not be optimal for all tumor types — particularly rare histologies with specific testing requirements or tissue constraints.
Cohort-specific screening model: Different labs or assays per cohort. More complex to coordinate, but may deliver better performance for rare tumor types. Increases EDC design complexity (multiple lab sources, multiple assay results, cohort-specific eligibility logic) and requires explicit data management planning from the start.
The design decision must be resolved in protocol development — it cannot be left to CRO discretion at startup. And the data management implications must be designed into the EDC from the start, not retrofitted after the assay architecture is determined.
—
How Does Cross-Cohort Safety Oversight Work?
Safety signals in one tumor type cohort have implications for the full basket. An unexpected toxicity in the colorectal cancer cohort may affect the benefit-risk assessment for the non-small cell lung cancer cohort in the same basket. The pharmacovigilance infrastructure must support both cohort-level case processing and study-level aggregate safety assessment — and the DSMB or safety monitoring committee must have visibility across all cohorts simultaneously.
Kyle Hanson, Director of Clinical Operations, Sitero:
“Signal attribution. In a basket trial, you have the same therapeutic agent across molecularly defined but histologically diverse populations. When a safety signal emerges, the immediate question is whether this is a drug effect, a tumor-type effect, or an interaction effect driven by prior treatment history in a specific cohort. That question is genuinely hard to answer quickly. The operational requirement that makes this manageable is a pre-specified cross-arm safety review framework in the DSMB charter — not just a general statement that safety will be monitored, but explicit rules about what triggers a cross-cohort safety review, what data will be presented, and what the decision thresholds are. Without that pre-specification, you end up in ad hoc committee discussions with incomplete data packages and no shared framework for what ‘a signal’ means. The safety medical monitor also needs visibility across all arms in real time — which sounds obvious but often isn’t reflected in how monitoring contracts are actually scoped.”
Three pharmacovigilance infrastructure requirements specific to basket trials:
- Case processing system that captures tumor type cohort as a required field on every serious adverse event
- Aggregate safety analysis plan that addresses cross-cohort signal assessment explicitly
- DSMB charter that specifies the committee’s authority to pause individual cohorts vs. the full basket based on safety findings
Learn how Sitero manages basket and umbrella trial execution: Basket & Umbrella Trial Execution
Frequently Asked Questions
Q: How do you manage cohort enrollment imbalance — when some tumor types fill quickly and others stagnate?
Cohort-level enrollment tracking with pre-defined response protocols is the operational answer. The steering committee should have a pre-specified decision framework for cohort-specific interventions: targeted site additions for slow cohorts, enrollment cap procedures for fast cohorts. Without this framework, steering committee bandwidth is consumed by enrollment management rather than scientific governance.
Q: How does a basket trial handle a safety signal in a tumor type cohort where the benefit-risk profile may differ from other cohorts?
The IDMC or safety monitoring committee needs explicit charter language for cohort-specific benefit-risk assessment — the ability to pause or close one cohort based on safety findings without triggering automatic suspension of all cohorts. This requires that the adaptive design’s cohort independence be pre-specified in both the protocol and the safety charter.
Q: What is the minimum governance infrastructure required to start a basket trial?
At minimum: a steering committee charter with decision authority documented, an IDMC with adaptive trial experience and a pre-specified charter, a central IRB relationship established before first site activation, and a protocol amendment process with defined timelines from IDMC recommendation to IRB approval to site notification. All of these must be in place before the first patient is screened.
—
Planning an oncology trial with a basket trial design?
Sitero has supported 200+ oncology studies across 67+ countries. Talk to an oncology trial expert to discuss your protocol.
—
References
- Sitero. Oncology Program Operational Data. Internal dataset. 200+ oncology studies across 67+ countries. sitero.com/oncology/
- Hanson K. Director of Clinical Operations, Sitero. Expert interview conducted for this article. April 2026.
Evaluating a CRO for a biomarker-driven oncology trial? Use these questions on site selection, central lab coordination, EDC design, and screening failure management.
In an adaptive oncology trial, the RTSM must do four things that a conventional randomization system is not required to do: enforce configurable cohort and dose assignment rules in real [...]
