Conditional release

Conditional release is the regulated act of releasing a batch for use before every required finished-product test has completed. It exists because some products — most prominently PET radiopharmaceuticals — cannot wait. By the time the standard 14-day sterility test is complete on an F-18 FDG batch (half-life 109.77 minutes), the drug has been administered, decayed below useful activity, or both. Holding the batch until classical finished-product testing finishes is not just commercially unviable — it makes the product physically impossible to use.

Conditional release is not a workaround, a shortcut, or a regulatory grey area. It is an explicit provision written into 21 CFR 212.70(d) for PET, backed by tightly defined in-process testing, a contractual recall posture for the missing tests, and a signed-and-attributable release decision that names exactly which tests were complete at the moment of release and which were not.

The mechanism is narrow. It applies only where the rule names it, and it carries obligations — recall plan, deferred-test capture, contingent product hold — that persist for the entire shelf life of the batch (which for PET is often hours, but for other conditional-release contexts can be days).

The mechanism is product-specific and rule-specific. The major recognised contexts are:

Because finished-product sterility cannot be confirmed before release in a PET context, the entire assurance burden shifts onto in-process controls performed during synthesis and on the final dose before it leaves the cyclotron facility. 21 CFR 212.70(d) is explicit: conditional release requires that all in-process testing that CAN be completed before release IS completed and meets specification.

For a typical F-18 FDG batch, the in-process control battery completed before release looks like:

Sterility (USP <71>, 14-day incubation) and sometimes the full LAL confirmation are the tests deferred. Everything else has to be on the page, signed, in-spec at the moment the release button is pressed.

1. Master Batch Record approved with conditional-release as an authorised disposition pathway for this product (Part 212).
2. All in-process controls that can be completed pre-release listed in the MBR, with specs and methods.
3. Each in-process result captured in the batch record, signed, with reviewer signature, before the release decision.
4. An identified and documented set of tests that will complete after release — sterility, LAL if deferred — with named methods and acceptance criteria.
5. A written recall plan tied to the specific units released from this batch (every dose, every receiving site, every prescribing physician), executable within hours if a deferred test fails.
6. A signed, attributable release decision recording who released, on what basis, with what tests still pending, and what the recall plan reference is.
7. An out-of-specification (OOS) procedure that explicitly covers deferred-test failures and integrates with the recall plan rather than running in parallel to it.

The conditional-release decision itself is one of the most consequential signatures in regulated manufacturing. Done wrong, it puts patients at risk and the facility on the front page. Done right, it is a 30-second event with every piece of evidence already in front of the releaser.

A defensible release-decision screen — whether on paper or in software — presents:

• Batch ID, product, activity at calibration, dose count.
• Every in-process test result in a single table: spec, result, pass/fail, who performed, when, reviewer signature.
• Explicit list of deferred tests with method, due date, recall trigger.
• Recall plan reference and dose-recipient summary (number of doses, number of sites).
• The releaser's e-signature with the meaning string "Conditional release per 212.70(d) — deferred tests as listed."
• Optionally a reviewer signature (some sites require independent two-person sign-off for conditional release; the rule does not mandate it but many SOPs do).

If any of those elements is missing or arrives after the signature, the release is procedurally indefensible — even if the batch turned out fine.

Conditional release places a contingent recall obligation on the batch from the moment of release. The obligation persists until every deferred test has reported and met specification. If a deferred test fails — most commonly sterility positive at 7 or 14 days — the response sequence must execute fast:

1. Stop further dispense of any remaining material from the batch immediately.
2. Notify the receiving clinical sites by phone, not just email. Confirm receipt of the notification.
3. Quarantine any undispensed material at the receiving sites.
4. Notify each dispensed-dose recipient (or their prescribing physician) per the recall plan.
5. Open a deviation referencing every dose dispensed from the batch.
6. Report adverse events through MedWatch if any occur.
7. Investigate root cause; CAPA if systemic.
8. Notify FDA per 21 CFR 314.81 (NDA field alerts) or 600.14 (biologics) as applicable, generally within 3 working days.
9. Document everything in the batch record so the eventual inspection narrative is self-contained.

The recall posture is not a theoretical concern. Every PET facility that has been operating for more than a year or two has needed to act on it at least once, usually for a non-pathogenic environmental sterility positive that turned out to be a method false positive. Speed of contact matters more than depth of investigation in the first hour — the goal is to stop additional patient exposure while you figure out whether the failure was real.

The conditional-release concept exists in both jurisdictions but with structurally different responsibility models.

Sites that operate in both jurisdictions usually adopt the stricter framing (QP-style personal accountability) as the operational default and document the US release per 212.70(d) on top.

Conditional release is one of the most heavily scrutinised events in any FDA PET inspection. The inspector will want to see, for a randomly chosen batch from the last 90 days:

• Every in-process test result, with timestamp and operator identity.
• Reviewer signature on each in-process result before the release signature.
• Release signature with timestamp, identity, and meaning.
• Deferred test list captured at the moment of release (not edited after the fact).
• Deferred test results when they landed, in the same batch record.
• If any test failed, the deviation, the recall actions taken, the field alert filed.
• Audit trail evidence that no record was edited after release without a controlled change.

ALCOA+ applies in full. The release decision must be Attributable to a specific named person, Legible, Contemporaneous to the release (signature timestamp within seconds of the decision), Original (not a transcription), and Accurate (matching what actually happened). Conditional release does not relax data-integrity expectations — if anything, it raises them, because the safety net of finished-product testing is not there.

• Treating conditional release as routine final release — deferred tests get forgotten in the workflow, and a sterility positive lands 14 days later with no recall plan attached to the batch.
• Recall plan held only as a separate SOP, with no per-dose recipient list attached to the specific batch — when a failure occurs, the first 90 minutes are spent finding out who got the doses.
• Releaser does not have full visibility of which in-process tests have passed at the moment of decision — signatures happen before reviewer sign-off lands.
• Deferred test failures handled as a clinical-pharmacy issue rather than a batch event — the deviation never opens, the field alert never files.
• Audit trail of the conditional-release decision held outside the batch record (an email, a paper log, a separate system).
• Release meaning string saying "Released" rather than "Conditional release per 212.70(d)" — same signature, different defensibility.
• Sterility deferred but LAL also quietly deferred without it being on the release-decision screen.
• No procedure distinguishing the release decision from the post-release deferred-test capture — they merge into one event in the system and inspectors cannot tell what was known when.

Although §212.70(d) is the formal mechanism most often discussed, similar logic appears elsewhere under different names. Understanding the distinctions matters because using the wrong terminology in front of an inspector implies the wrong rule-set.

These mechanisms share the principle that classical finished-product testing is not the only valid assurance pathway. They differ sharply in what is substituted, what regulatory rule grants the substitution, and what residual obligations remain.

21 CFR 212.70(d) exists because PET radiopharmaceuticals decay faster than the sterility test can complete. F-18 has a 109.8-minute half-life; Ga-68 is 68 minutes; C-11 is 20.4 minutes; O-15 is 122 seconds. By the time a USP <71> 14-day sterility test reads out, an F-18 dose has decayed through more than 180 half-lives and is no longer detectable as a radiopharmaceutical, let alone administrable. The dose has been administered, the scan has been read, and the clinical decision has been made. The regulatory architecture must therefore allow release on the strength of fast tests (Bubble Point + LAL + radiochemical purity + appearance + pH + activity) with the sterility test deferred and the radiochemical-purity-bracketed retention sample held for the investigation that almost never happens.

• F-18 FDG — the dominant PET tracer. Six-hour useful clinical window from end-of-synthesis. Conditional release is the only economically viable model.
• Ga-68 PSMA / DOTATATE — therapy-adjacent imaging. Five-half-life window forces release within roughly two hours of synthesis.
• C-11 choline / PIB — single-site, single-shift production with the cyclotron physically adjacent to the scanner. Released within minutes.
• Lu-177 therapy doses — long half-life (6.6 days) means conditional release is unnecessary; standard release applies, often QP-style.
• Actinium-225 alpha therapy — long enough half-life (9.9 days) for full release; the rate-limit is isotope supply, not test turn-around.

The result is that the conditional-release mechanism is a PET-domain norm but vanishingly rare in commercial pharma outside PET. Cell-therapy products (CAR-T, TCR-T, TIL) negotiate analogous out-of-specification-after-release frameworks under their IND/BLA, but those are bespoke commitments rather than a blanket regulatory carve-out.

PET sterility positives are uncommon — most reported events are method false positives traced to handling artefacts during incubation — but every one triggers a non-discretionary cascade. The 21 CFR 314.81 NDA Field Alert Report (FAR) clock starts the moment the investigator confirms the OOS reading, not when the lab opens the deviation. Sites that misread the start of the clock typically miss the three-working-day window.

1. T+0: lab confirms positive growth on the deferred sterility flask. The radiopharmaceutical scientist on call is paged.
2. T+30min: deviation opened against the originating WO; the production records, environmental monitoring, operator training and aseptic-process simulation history are pulled into the investigation file.
3. T+2h: contact the treating institution(s) that administered the dose. Identify patients and notify the responsible physicians per the firm's pharmacovigilance SOP.
4. T+24h: complete the initial method-vs-true-positive triage. Repeat the test from the retention sample if available; review identification of the organism; cross-check against environmental-monitoring isolates from the same week.
5. T+3 working days: file FORM FDA 3331a (NDA Field Alert) if the investigation has not unambiguously identified a method false-positive root cause. Always file rather than defer — late FARs are an inspection finding in their own right.
6. T+15 working days: MedWatch (FORM FDA 3500A) for any patient who experiences an adverse event judged related to the dose.
7. Closure: full investigation report, CAPA, environmental-monitoring trend review, aseptic-process-simulation review, and a decision on continued conditional release for the affected product/line.

Most PET sites have never filed a true-positive FAR. Most have filed at least one false-positive FAR over a multi-year operating window. The reporting decision is binary and conservative: when in doubt, file.

V5 ships an industry-specific conditional-release workflow for tenants whose execution profile includes PET production. The mechanism is opt-in at the product level — only products with `release_mode = 'conditional'` follow the deferred-test gate, so the bulk of the catalog continues to release through the standard QC pathway.

• Per-product release mode (standard | conditional | parametric) configured against the Product Master and snapshotted into the WO at release-planning time so a mid-batch reconfiguration cannot weaken the active-batch controls.
• Two-bucket QC checklist on the WO Release screen: required-before-release (LAL, Bubble Point, radiochemical purity, appearance, pH, activity) and deferred-allowed (USP <71> sterility). Release authority is enabled only when every required-before-release test is signed-off PASS.
• Activity-time calculation on the WO using the standard A(t) = A0·e^(−λt) decay formula against the captured calibration time and the isotope master half-life. The release record carries the calibration time and the at-release activity in becquerels.
• Deferred-test follow-up task on the originating WO with an SLA equal to the method's incubation window (14 days for USP <71>); the WO cannot close until the deferred result is in and signed.
• On deferred-test fail: an alert fires to the QA distribution list, the originating WO is flagged in red on every dashboard view, and the FAR / MedWatch task generators open with the relevant context pre-populated.
• BMR rendering: the regulated PDF carries the conditional-release marker on the cover, lists the deferred test under a clearly labelled 'tests outstanding at release' section, and is re-rendered automatically when the deferred result is signed.