FDA's 2026 oncology eligibility guidance: preserve the clinical rationale in the study build

TL;DR

FDA's July 2026 final guidances on laboratory values, performance status, and washout periods and concomitant medications share a direction: eligibility restrictions should be justified by the drug, disease and available evidence, rather than copied forward automatically. Their application requires drug-specific justification and the authorized protocol; editing an electronic eligibility form alone provides no authority to change an active study. These are nonbinding recommendations unless they cite a specific legal requirement. [1] [2] [3]

For sponsors and CROs, the consequential implementation risk is a loss of meaning between the clinical rationale and the executable rule. “Do not require every electrolyte to be normal” can become an unsafe blanket allowance. “In adult oncology trials, include ECOG PS2 unless a justified safety concern argues otherwise” can become an automatic assignment to an exploratory cohort. “Use recovery criteria instead of arbitrary washout periods” can become the deletion of a scientifically necessary interval.

Useful implementation starts with those distinctions before deciding how many fields to collect. For each eligibility rule, the study build should preserve three distinct things: the approved criterion, the evidence used to evaluate it, and the authorized eligibility decision. This is our implementation framework, not a database design prescribed by FDA. Its value is that a medical reviewer can trace what the system did without confusing a software flag with clinical permission to enroll.

The change is in the justification, not a universal numerical threshold

The laboratory-values guidance recommends exclusionary laboratory criteria only when clearly necessary to mitigate potential safety concerns. It links restrictions to the investigational drug's metabolism, excretion, pharmacology and anticipated toxicity, and calls for reassessment as clinical experience accumulates. It also preserves a role for conventional precautions in first-in-human studies of first-in-class drugs or novel platforms whose human pharmacology and toxicity are not yet known. [1]

That is a reason to inspect inherited criteria, not to assume that a broader criterion is always safer or scientifically adequate. A renal-function restriction may be central to exposure control for one drug and poorly justified for another. Even within a therapeutic area, the evidence needed to change a criterion can differ. A protocol amendment should make that reasoning visible before it becomes a software change request.

An implementation review can ask a sharper question than “Is this threshold too restrictive?” Ask which risk the threshold controls, which evidence supports its direction and magnitude, and what additional information is required to apply it. Implementing a reference-range multiple requires the relevant laboratory's upper limit of normal and consistent units. Interpreting a permitted repeat test requires its timing and relationship to the initial result.

These are operational implications of the guidance, not new FDA field specifications. Different data models must preserve the same clinical meaning. A rule that appears correct in a requirements spreadsheet may still be wrong when applied to a different laboratory, a retest, or a participant assessed under a different protocol version.

Laboratory eligibility: avoid replacing one blunt rule with another

FDA gives a useful example involving drugs that may prolong QTc. Low potassium, calcium or magnesium may increase arrhythmia risk. Requiring every result to lie within the normal range can unnecessarily exclude a slightly above-range value with no increased QTc-prolongation risk. The guidance uses that example to argue for precise protocol language aligned with the particular risk. [1]

The safety assessment remains electrolyte-, value- and treatment-specific; higher values require their own assessment rather than automatic acceptance above the lower limit. Other clinical risks and protocol criteria still matter. The safe translation is to implement the approved, analyte-specific criterion and its clinical context—not convert an illustrative lower-bound argument into a universal no-upper-bound rule.

The same care applies to absolute neutrophil count. FDA notes that the Duffy-null phenotype is associated with lower ANC without greater infection risk in that context and suggests considering broader ANC criteria for people with that phenotype. The discussion concerns phenotype, with the ANC criterion and safety assessment still dependent on the particular anticancer treatment and myelosuppressive risk. Drug-specific safety assessment remains necessary. [1]

A study build should therefore not derive Duffy-null status from a demographic category. If an approved protocol distinguishes a subgroup, the eligibility process needs the evidence and assessment specified for that distinction. A proposed subgroup rule must enter the authorized protocol before it becomes an operational eligibility rule in EDC.

FDA also says that, where appropriate for particular analytes, sponsors can include the ability to repeat testing within a specified period in the protocol. Use the protocol's permitted repeat-testing and result-selection rules. [1] Retain the initial and repeat results, their dates, relevant units and reference information, and the basis for selecting the result used in the eligibility decision. An overwritten initial result can make a legitimate retest indistinguishable from selective reporting.

Consider a hypothetical protocol that permits one repeat within a defined screening window. A useful acceptance test includes a repeat within the window, a repeat outside it, an initial result with a different unit, and an attempt to overwrite the original result. The expected outcome should be derived from the approved protocol and data-handling rules, not supplied by a generic “broaden eligibility” switch.

Performance status: inclusion and analysis membership are different decisions

For adult oncology trials, the performance-status guidance recommends inclusion of patients with ECOG PS2, or KPS 60–70, unless a scientific or clinical rationale for exclusion is justified by established safety considerations. Its recommendations concern adults; pediatric application requires separate justification even where general considerations are relevant. It recommends collecting baseline performance status to characterize the enrolled population and considers PS as a possible stratification factor where there is a broad baseline range. It also acknowledges the subjectivity of PS assessment and the potential value of complementary functional assessments. [2]

Crucially, the guidance does not say that every PS2 participant belongs outside the primary analysis. When there are concerns in the particular trial context, it discusses alternative designs such as prespecified low-PS cohorts excluded from the primary analysis, generally small and exploratory, potentially enrolled incrementally with safety stopping rules. That is a design option, not the default destination for all PS2 patients. The analysis approach should be determined during study design and discussed with FDA when appropriate. [2]

This distinction creates three separate questions for implementation: is the participant eligible, which treatment assignment process applies, and which analysis population includes the participant? One recorded PS value may inform all three decisions, each with its own applicable rules. An EDC field that silently sets an “exploratory” flag whenever PS equals 2 would be wrong for a protocol that includes those participants in the main randomized population.

The proposed control is to document cohort and analysis rules independently from the observed PS assessment. For a protocol with an exploratory cohort, test the actual eligibility, enrollment limits and safety-review process for that cohort. For a protocol without that design, ensure that no inherited exploratory-cohort logic remains in the build. The statistical analysis plan, protocol and randomization specifications should agree on the participant's treatment and analysis context.

A further implementation hazard is treating longitudinal function as interchangeable with baseline PS. A patient-reported physical-function instrument or activity tracker may complement clinician assessment; replacing the protocol's PS criterion requires appropriate validation and authorization. FDA recommends considering additional assessments without making them exclusionary absent a justified safety rationale. [2] Adding an instrument should answer a scientific question and include its own administration and interpretation requirements, rather than create a hidden new enrollment barrier.

Washout: separate recovery, interaction risk and attribution

The washout and concomitant-medications guidance recommends using relevant clinical and laboratory parameters, based on the preceding therapy, in place of time-based washout periods to address safety considerations. When a time-based interval is used, it should have a scientific justification that considers relevant pharmacokinetics and pharmacodynamics. The guidance also recognizes that delayed antitumor effects of prior therapy can justify a washout period when estimating the investigational treatment's antitumor effect. [3]

Assess resolved toxicity and the potential for prior treatment to confound response as separate questions. A recovery criterion addresses one concern; a justified interval may address another. Deleting all time-based rules because a recovery assessment exists can remove a control that serves a different purpose.

FDA recommends recovery from clinically significant adverse events from the most recent anticancer therapy or intervention depending on their relevance to the investigational drug, including potentially overlapping toxicity. It does not prescribe a universal CTCAE grade-at-or-below-1 criterion for every prior event. Nor does it state that an unresolved event should be ignored whenever a calendar interval has elapsed. [3]

A practical design separates the prior therapy's identity and dates, the relevant residual toxicity assessment, and the approved eligibility rule. The implementation must then combine them as the protocol requires. For a hypothetical trial requiring both a justified interval and recovery of a specified toxicity, testing should include a participant who satisfies only the interval and another who satisfies only the recovery condition. Both expose a common logical error: implementing an “and” requirement as “or.”

For concomitant medications, FDA recommends exclusions based on clinically relevant known or predicted interactions and overlapping toxicities affecting participant safety. It discusses dose or regimen modification as an alternative in some circumstances, with the approach justified and clearly specified in the protocol and study materials and adequately communicated to participants and caregivers. [3]

Any dose modification requires the protocol's medical and operational decision process. A medication interaction flag should follow the approved rule and direct the issue to the authorized clinical process. An excluded drug, regimen change or unresolved interaction must go through that process, with the authorized clinical decision recorded explicitly. For certain complex biological products, including cell or gene therapies, FDA notes that other considerations may apply and should be discussed with the appropriate CBER office. The development approach still requires a scientific justification. [3]

A rule-to-evidence matrix for the amendment meeting

The most useful deliverable is a small set of decisions that clinical, statistical and operational owners can review together. The following matrix is an editorial implementation aid, not a required FDA form.

The matrix deliberately separates the data from the decision. It avoids asking sites to enter an unsupported rationale for every normal result while still preserving the rationale where an eligibility exception, repeat or clinical assessment matters. Keep the data proportionate to the approved trial design and sufficient to interpret the decision.

Amend the protocol before activating a different rule

An active trial's protocol changes through the applicable amendment process, separately from publication of final guidance. For US IND studies, 21 CFR 312.30(b) covers phase 1 changes significantly affecting participant safety and phase 2 or 3 changes significantly affecting safety, scope or scientific quality. For those changes, implementation ordinarily follows FDA submission and responsible IRB approval; the regulation separately addresses changes intended to eliminate an apparent immediate hazard. The applicable process must be assessed for the particular change and jurisdiction. [4]

Operationally, distinguish a proposed criterion, an approved amendment, a released system configuration and a site's authorization to use it. They can have different dates. If sites activate an amendment at different times, a single unqualified “current eligibility rule” can apply the wrong criterion to a screening participant. The implementation record should retain the rule version used for the assessment and make the effective operational scope clear.

The applicable procedure depends on the actual change, including whether it is an editorial clarification or a substantive amendment. Medical and regulatory owners determine that category before the technology release implements it. A correction to a misimplemented existing rule and a scientifically broadened eligibility criterion are different change requests, even if both alter the same comparison operator in code.

Release evidence should show more than a successful eligible case. Include expected ineligible and indeterminate cases, missing information, permitted repeats, protocol-version transitions and the relevant cohort or interaction boundary. For each, the reviewer should be able to explain why the displayed status follows the approved rule and what human assessment remains necessary. An automated result must leave any unresolved medical decision visible.

What to ask an EDC and study-delivery partner

The relevant demonstration is a trace from protocol wording to configured rule, source evidence, assessment and downstream use. Ask the partner to show a representative amendment and a boundary case rather than only a completed screening form. If randomization or an analysis population consumes the eligibility result, include that handoff in the demonstration.

EClinCloud's EDC and study services are relevant places to discuss study build and data-management support. Clinical eligibility decisions and regulatory acceptance depend on the responsible decision-makers and applicable evidence, beyond the product's technical role. A useful implementation enquiry supplies the protocol criteria, proposed amendment, affected assessments and interfaces, so that the required configuration and review work can be defined.

The implementation aim is to preserve scientific justification as eligibility becomes an operational rule: reviewers should be able to trace the criteria used, supporting evidence and each participant's treatment and analysis context. The system's job is to preserve those decisions faithfully, not to create new ones.

Sources

1. FDA. Cancer Clinical Trial Eligibility Criteria: Laboratory Values. Final guidance, July 2026. Sections II–III, printed pages 2–6: drug-specific rationale, QTc/electrolyte example, Duffy-null phenotype, retesting and reassessment across development.

2. FDA. Cancer Clinical Trial Eligibility Criteria: Performance Status. Final guidance, July 2026. Sections II–IV, printed pages 2–7: adult-population scope, inclusion, justified exclusions, alternative cohort designs and complementary assessments.

3. FDA. Cancer Clinical Trial Eligibility Criteria: Washout Periods and Concomitant Medications. Final guidance, July 2026. Sections II–III, printed pages 2–4, including the footnote on complex biological products.

4. Electronic Code of Federal Regulations. 21 CFR 312.30—Protocol amendments. Paragraph (b), significant protocol changes and implementation conditions. Accessed October 11, 2026.