21 CFR 212FDA cGMP for PET drugs

21 CFR Part 212 is the FDA's cGMP rule for PET drugs. It was finalised by the agency on 10 December 2009 (74 FR 65409) and became effective on 12 December 2011, with a single-year compliance extension to 12 December 2012 for facilities that had filed an NDA or ANDA by the original effective date. Before Part 212, the agency had attempted to apply 21 CFR Part 211 (the general drug GMP) to PET production, and the practical result was unworkable — Part 211 requires completed finished-product testing before release (§211.165) on a timescale that simply cannot be met for radiopharmaceuticals with half-lives measured in minutes.

Part 212 keeps every substantive cGMP concept — personnel qualification, facility and equipment controls, components, production and process controls, laboratory controls, finished-product release, distribution and complaint handling, records and reports — but the structure is built around the operational reality of producing PET drugs in cyclotron facilities, academic medical centres, hospital PET pharmacies and commercial PET distributors. The simplifications relative to Part 211 (no long-term stability programme, no classical shelf-life testing, conditional release permitted) are compensated by tighter in-process controls because the post-release safety net of finished-product testing is not available in real time.

Part 212 has eight subparts (A through H) covering, in order: general provisions and definitions; personnel and resources; quality assurance; facilities and equipment; control of components, in-process materials and finished products; production and process controls; laboratory controls; and finished-product controls and acceptance, plus records and reports under subpart H. Subpart F (production and process controls) contains §212.70 — the section that includes the conditional-release provision at §212.70(d). Subpart H specifies the per-batch production record content and the retention period (records to be retained for at least 1 year past the expiration date of the batch, or 2 years from the date of full release, whichever is longer).

The subpart structure deliberately mirrors Part 211 to make cross-reading easier for operators trained on Part 211, but each subpart is materially shorter — Part 212 in total runs to roughly half the length of Part 211 because the PET-specific simplifications remove large sections (the classical stability programme is the most visible deletion).

The simplifications come with one major addition: the requirement for very tight in-process and pre-release controls, because the post-release safety net of finished-product testing cannot rescue a bad batch. Every facility operating under Part 212 invests heavily in the synthesizer-module qualification, the dose-calibrator calibration, the HPLC and GC method qualification, and the radiochemical-purity / pH / appearance / half-life-confirmation in-process tests that drive the conditional-release decision.

Section 212.70(d) is the operational keystone of Part 212. It explicitly permits a PET drug to be released for human use before every finished-product test is complete, provided four conditions are met: (1) the in-process controls and finished-product tests that have completed all meet specifications based on the available data, (2) any test that has not completed is performed afterwards, (3) the responsible-individual (typically the production manager or a designated qualified individual) signs the release decision documenting the in-process and completed-finished-product results that form its basis, and (4) if any deferred test subsequently fails the affected batch and any doses dispensed from it must be evaluated and the patient's prescribing physician (and the patient where appropriate) must be notified.

In practice the conditional-release flow looks like this. Synthesis completes; the activity is verified at calibration time on the dose calibrator; the in-process controls — radiochemical purity by TLC or HPLC, residual solvents by GC, pH by paper or meter, appearance by visual inspection, half-life confirmation by repeated activity measurement — are completed and in spec; the unit doses are drawn; the drug is released. The longer endotoxin test (typically 15–60 minutes with rapid LAL) is usually completed before release; the sterility test (14 days minimum) runs in parallel and the release record is updated with the result when it completes. If the sterility test fails, the recall is operationally meaningless (the doses have decayed and been administered) but the deviation, the adverse-event monitoring, and the corrective-action workflow on the synthesis or aseptic-handling process all engage.

Every activity-bearing record in a PET facility must carry a calibration timestamp because the activity value is meaningless without it. A vial at 100 mCi at 09:00 is at 50 mCi at 10:50 if it is F-18 (T½ 110 min), or at 50 mCi at 09:20 if it is C-11 (T½ 20 min). A system that does not natively handle this math cannot operate a PET facility safely — the patient dose, the dispensing record, the disposal manifest, the survey-meter background subtraction, the radwaste decay-in-storage calculation, all depend on knowing the activity at a specific time and projecting it forward or backward by the isotope's half-life.

The governing equation is A(t) = A₀ · e^(−λt) where λ = ln(2) / T½ and T½ is the physical half-life of the isotope. For mixed isotopes (rare in PET production but common in research), the decay is computed per-isotope and summed. V5 Ultimate carries this math natively — every activity-bearing record has a captured value, a captured timestamp, and the system computes the current activity on display from the isotope half-life on the lot's isotope record. The 17-isotope master in V5 carries the half-life for every commonly-produced PET, SPECT and therapeutic radionuclide; new isotopes are added with a controlled change record.

• Assay results — captured at the assay time, with the isotope identified, the half-life referenced for any time projection.
• Dispensing — the operator enters the patient calibration time (the time the dose is administered or the scanner uptake time), and the system computes the required draw activity at the dispense time to deliver the target activity at calibration.
• Dose calculation — body-weight-based or fixed-dose, computed in MBq or mCi at calibration, with the chronological projection from dispense time captured.
• RG 1.86 release surveys — survey-meter background subtracted with the isotope-specific decay correction applied when the survey is performed after the activity has substantially decayed.
• Waste decay-in-storage — the disposal date is computed forward from the receipt date using the isotope half-life and the regulatory release threshold (10 half-lives is the common operational rule for low-activity waste).

Section 212.50 specifies the content of the per-batch production record. The record must be prepared for each batch and must contain enough information to provide a complete history of the production and control of the batch. The required content includes:

• Batch identity — batch number, synthesis date, target isotope, calibration time, intended use (clinical / investigational / research).
• Components — precursor identity and lot, target water or gas, reagents, vials, sterile filters, lot numbers and supplier identities.
• Equipment — cyclotron identity and beam-on/off times, synthesizer module identity, dose calibrator identity (with calibration certificate currency), HPLC and GC identity, pH meter identity, sterile-handling hood identity.
• In-process results — radiochemical purity at end of synthesis, chemical purity, pH at end of synthesis, appearance, residual solvent (typically ethanol, acetonitrile), half-life confirmation by repeated activity measurement, total activity at calibration time.
• Conditional-release decision — if applicable, the responsible-individual e-signature, the time of decision, the basis (specific tests completed and specific tests deferred), the dispensed-dose range covered, the action plan if a deferred test fails.
• Deferred tests — endotoxin (typically rapid LAL completed before release; full LAL or kinetic LAL may be deferred), sterility (14-day minimum), with the scheduled completion time and the eventual result captured.
• Yield and reconciliation — activity at end of synthesis vs activity at end of bombardment, expressed as decay-corrected yield; total dispensed activity vs total available activity at calibration; waste-stream activity reconciliation.
• Reviewer signature — independent QA reviewer (the quality unit under §212.20) signs the batch review prior to final release with a Part 11 §11.50 manifest.
• Distribution — every dispensed dose with the receiving institution or prescribing physician, the patient identifier or scan identifier (per institutional convention), the activity at calibration, the time of dispense.
• Complaint and adverse-event linkage — any complaint or AE later associated with the batch is appended to the record by reference.

Part 212 governs cGMP for the drug. It is not the only rule a PET facility operates under, and inspectors will check the boundaries explicitly. The principal adjacent regimes are:

• Radiation safety — 10 CFR Part 20 (NRC Standards for Protection Against Radiation) and the facility's NRC or Agreement-State licence conditions govern occupational dose, public dose, posting and labelling, and personnel monitoring. Part 212 does not.
• Radwaste release — NRC Regulatory Guide 1.86 (now superseded for many purposes by 10 CFR 20.2002 and state-level rules) governs the surface-contamination and removable-contamination limits for unrestricted release of materials and equipment. Part 212 does not.
• Transport — 10 CFR Part 71 (NRC) and 49 CFR (DOT) govern the packaging, shipping papers, marking, labelling, placarding and accident-response requirements for radioactive material transport. Part 212 does not.
• Investigational PET drugs — 21 CFR Part 312 (IND) governs the use of investigational drugs in clinical trials; investigational PET drugs operate under both Part 312 and Part 212 (or under USP <823> in the academic-research setting prior to IND).
• Compounded PET — USP General Chapter <823> covers the compounding, investigational and research use of PET drugs in academic and hospital settings where commercial Part 212 production does not apply. <823> is operationally similar to Part 212 but has a different legal basis (USP-NF rather than CFR).
• Patient dose responsibility — the prescribing physician and the institution administering the dose are responsible for the dose decision and the patient-safety verification at the point of administration. The PET facility is responsible for the drug; the administration is regulated separately.

A complete radiopharma operation runs all of these simultaneously. V5 Ultimate covers the cGMP side under Part 212 (and where applicable USP <823>); we sell honestly alongside the customer's existing rad-safety system and transport documentation rather than attempting to replace those domains. The integration point is the calibration-time-anchored activity record, which feeds the cGMP view and is the same record the rad-safety system consumes.

1. Conditional release treated as routine bypass. The responsible individual signs the §212.70(d) release without genuinely reviewing the available in-process data, because every batch goes that path and the form has become wallpaper. Fix: kiosk forces explicit acknowledgement of each in-process result and each deferred test before the signing prompt; signing is keyed on the result panel, not on a separate page.
2. Activity records without calibration timestamps. An assay result captured as '100 mCi' with no time annotation is operationally meaningless — the value at the moment of dispense depends on the half-life elapsed since the assay. Fix: every activity-bearing record requires a captured time; the system carries the isotope half-life and computes current activity on display.
3. Deferred sterility results not linked back to dispensed doses. A sterility test fails at day 14; the doses dispensed from the batch are not identified; physician notification does not happen because the data link is broken. Fix: every dispensed dose is referenced from the batch record; deferred-test failure auto-opens a deviation listing every dispensed dose with prescribing-physician contact.
4. In-process methods not validated. Radiochemical purity by TLC method that has no validation against the specific tracer; pH paper used where a calibrated meter is required by the spec; residual-solvent GC without method validation on the actual matrix. Fix: every in-process method is itself a controlled validated procedure with a method-qualification record.
5. Dose-calibrator calibration overdue. The constancy / accuracy / linearity / geometry tests required by NRC and best-practice (daily / quarterly / annual respectively) are skipped. Activity records become unreliable. Fix: instrument-calibration tasks are scheduled with hard due-dates; instruments lock out at the kiosk when calibration is overdue.
6. Records retention not meeting §212.110. Records retained for one year flat, ignoring the 'or two years from full release whichever is longer' clause. Fix: retention policy enforced by storage lifecycle, defaulted to the longer of one-year-past-expiry or two-years-from-release.

• 21 CFR Part 212 — the FDA cGMP rule for PET drugs (this page).
• 21 CFR Part 11 — every electronic record and electronic signature under Part 212 is also a Part 11 record.
• 21 CFR Part 312 — investigational PET drugs operate under Part 312 (IND) in addition to Part 212.
• USP General Chapter <823> — the compendial standard for compounding, investigational and research PET drugs in academic and hospital settings where commercial Part 212 production does not apply.
• 10 CFR Part 20 — NRC Standards for Protection Against Radiation. Personnel monitoring, dose limits, posting, labelling.
• 10 CFR Part 35 — NRC Medical Use of Byproduct Material. Authorisations for medical use of radiopharmaceuticals.
• 10 CFR Part 71 / 49 CFR — NRC and DOT transport rules for radioactive material.
• NRC Regulatory Guide 1.86 (and 10 CFR 20.2002) — surface and removable contamination limits for unrestricted release of materials.
• FDA Guidance — PET Drugs: cGMP Small Entity Compliance Guide (2011) — the agency's interpretive guidance on Part 212 implementation.
• FDA Guidance — PET Drug Applications, Content and Format for NDAs and ANDAs — the application-side counterpart.
• State agreement-state licensing and inspection — many states have NRC Agreement-State status and are the primary radiation-safety inspector for PET facilities in their jurisdiction; FDA remains the cGMP inspector under Part 212 nationally.

• Conditional-release record completeness — share of §212.70(d) signings where the in-process results, the deferred-test list, the responsible-individual signature, and the action plan are all captured. World-class: 100%.
• Deferred-test on-time completion — share of deferred sterility / endotoxin tests completed and reviewed within the planned window. World-class: 100%; any miss is a deviation.
• Deferred-test failure rate — share of deferred tests that subsequently fail. World-class: low single digits per 1,000 batches; any failure triggers the dispensed-dose review and physician notification workflow.
• Activity-record calibration-timestamp coverage — share of activity-bearing records with a captured calibration time and a linked isotope. World-class: 100%.
• Dose-calibrator calibration currency — share of daily constancy / quarterly accuracy / annual linearity-and-geometry tests completed on schedule. World-class: 100%.
• In-process method qualification currency — share of in-process methods (RCP, residual solvent, pH, half-life confirmation, appearance) with a current method-qualification record. World-class: 100%.
• Two-person batch-review completion — share of batches with both the conditional-release signer and the independent quality-unit reviewer captured. World-class: 100% on every commercial batch.
• FDA Part 212 inspection findings — mature sites: zero. Any finding here is a structural process gap, not a one-off operator error.

1. Isotope master carries 17 commonly-produced PET, SPECT and therapeutic nuclides with half-life, decay mode and the regulatory references. New isotopes are added through controlled change-control with QA sign-off.
2. Every activity-bearing record (lot, assay, dispense, survey, waste) carries a captured value, captured timestamp, and isotope reference; the system computes current activity on display using A(t) = A₀·e^(−λt).
3. The per-batch production record under §212.50 is a structured object covering synthesis, in-process tests, conditional-release decision, deferred-test schedule, distribution log and quality-unit batch review — all rendered as one Part-11-compliant PDF through @react-pdf/renderer into the regulated-reports bucket.
4. Conditional-release signing under §212.70(d) is an explicit screen — the responsible individual sees every in-process result, the deferred-test list, and the action plan on the same panel before the Part 11 §11.50 manifest is captured.
5. Deferred tests (sterility, full LAL) are scheduled with hard due-dates, surface on the QA controller dashboard, and on failure auto-open a deviation listing every dispensed dose with prescribing-physician contact for the institutional notification workflow.
6. Dose-calibrator and other in-process instruments are on per-instrument calibration schedules (constancy daily, accuracy quarterly, linearity-and-geometry annual); the kiosk locks the instrument out when calibration is overdue.
7. In-process methods (RCP by TLC or HPLC, residual solvent by GC, pH, appearance, half-life confirmation) are themselves controlled documents with method-qualification records; an out-of-currency method blocks the in-process test execution.
8. Two-person sign-off under §212.20 — the conditional-release signer and the independent quality-unit reviewer are enforced as distinct individuals on every commercial batch.
9. Retention is indefinite by default with the per-tenant policy always meeting or exceeding the §212.110 minimum (one year past expiry or two years from full release, whichever is longer).
10. Mobile-safe ≤390px rendering for every operator surface — kiosk batch tile, conditional-release signing screen, deferred-test schedule, dispense screen, survey screen — handheld and tablet operators get the same Part 212 controls as desktop reviewers, with no horizontal scroll.