cGMP for Radiopharmaceuticals and Radionuclides in Late-Phase Trials: Short Half-Lives and Dosimetry

cGMP for radiopharmaceuticals and radionuclides in late-phase trials: how FDA handles short half-lives and dosimetry

Radiopharmaceuticals sit at an awkward intersection of drug manufacturing and nuclear physics. The product decays while you make it, sterility results arrive after the dose is gone, and the "dose" that matters clinically is measured in absorbed radiation, not milligrams. FDA's current good manufacturing practice (cGMP) expectations for these products are built around those realities. This article lays out which rules apply as a program moves into late-phase (Phase 2b/3) clinical trials, how the agency accommodates short half-lives in release testing, and what it expects for radiation dosimetry in both diagnostic and therapeutic settings.

Which cGMP framework applies

The first question for any radioactive drug is which set of regulations governs it, and the answer splits on whether the product is a PET drug.

PET drug products are regulated under the PET-specific cGMP regulations in 21 CFR part 212 ¹⁶¹⁵. Compliance with part 212 became required on December 12, 2011, except for investigational and research PET drugs ¹⁶¹⁵¹³⁷.
Non-PET radioactive drugs (for example, SPECT agents and therapeutic radioligands) do not fall under part 212. Human drugs that do not meet the definition of a PET drug must be manufactured under the general finished-drug cGMP regulations in 21 CFR parts 210 and 211 ¹⁶¹⁵¹².
For investigational PET drugs and research PET drugs, producers have a choice: they may follow part 212 or produce the drug in accordance with USP Chapter <823>, which FDA describes as the minimum cGMP requirement for those investigational and research products in lieu of part 212 ¹⁶¹⁵.

That last point matters for trial sponsors. During clinical development, a PET producer can operate to USP <823> rather than the full part 212 framework, but the endpoint (a marketing application) is built against part 212 and the PET drug application content expectations. Late-phase programs are effectively closing the gap between the flexible investigational path and the fully validated commercial state.

What "late phase" means for radiopharmaceutical cGMP

FDA does not publish a standalone "Phase 3 radiopharmaceutical cGMP" rule. Instead, late-phase expectations are the maturation of the same controls toward the state a marketing application must document. The clearest articulation of that endpoint is the process-verification framework FDA applies to PET drugs:

If every batch undergoes full finished-product testing to confirm it meets all specifications, process verification is not required ³².
If an entire batch is not fully verified by finished-product testing (or only the initial sub-batch in a series is tested), the producer must show the process is reproducible and capable of producing a product meeting predetermined acceptance criteria, with documented verification including date, signature, monitoring and control methods and data, and qualification of major equipment ³²¹⁸.
For new processes or significant changes, FDA recommends verification before distribution, generally including at least three consecutive acceptable production runs ³².
Established facilities may support process verification using historical batch records under a written protocol reviewing accumulated production, testing, and control data ³².
Concurrent verification with batch distribution is possible only if consistent with 21 CFR 314.70 and the approved application, justified in writing, run under a written protocol and QA procedures, with each batch fully tested except sterility and released only if it meets all requirements ³².

For marketing applications, FDA expects release and stability data for a minimum of three consecutive batches ²¹². A late-phase program should therefore be generating the batch history, process-verification data, and validated methods that turn into that package.

Core cGMP building blocks

Underneath the framework, the day-to-day expectations for radiopharmaceutical production track ordinary drug cGMP, adapted to radiochemistry.

Production and process controls. FDA expects a step-wise description of the synthesis and purification procedure, including the amount of each reactant, reagent, and solvent used and the radiochemical yield(s) obtained, and covering the preparation and manipulation of sub-batches ²⁴⁷. The process should be governed by monitoring specified process controls and parameters to ensure continued product quality, and those controls should be documented in the master production and control records ²⁴⁷. For drug product production, the master production and control records should provide the specific procedures used and ensure full traceability and accountability of all components, materials, equipment, and operators for each batch ²⁴⁷.

Quality control and finished-product testing. FDA expects written test procedures for finished products and, where applicable, components and in-process materials ¹¹¹³. Specifications must ensure identity, strength, quality, and purity and, where appropriate, sterility and pyrogens ³⁴³⁵. Analytical methods should be suitable for their intended use, sufficiently sensitive, specific, reproducible, and accurate, and new methods should be validated with documented data ¹¹¹³. For PET products relying on finished-product testing alone, each batch must be tested for conformity to all specifications ²⁰⁶.

Components, containers, and closures. Written procedures should describe in sufficient detail the receipt, identification, storage, handling, sampling, testing, and approval or rejection of components and drug product containers and closures, which should be handled and stored to prevent contamination ²⁴⁸. Any lot liable to objectionable microbiological contamination should be subjected to microbiological testing before use ²⁴⁸.

Sterility and aseptic processing

Most injectable radiopharmaceuticals are produced by aseptic processing, and FDA's sterility-assurance expectations are extensive because there is no terminal sterilization step to fall back on.

Aseptic processing should be validated with process simulations (media fills) that closely simulate actual production, including aseptic manipulations and interventions, product-contact surfaces, critical environments, and process steps up to release ²¹⁷. For PET drugs specifically, FDA recommends three successful media fill runs to qualify a new operator, followed by one media fill annually for each operator ²¹⁴²¹⁵²²⁴²³¹.
Sterilizing filtration is treated as a critical procedure for injectable PET radiopharmaceuticals ⁸⁷⁸⁹. Filtration conditions such as pressure, flow, and volume should be identified in the batch record and not exceeded, and the sterilizing filter should undergo post-filtration integrity testing after each run and before release ⁸⁷⁸⁸⁸⁹.
The aseptic workstation should provide a suitable environment (aseptic hood, isolator, or other system), with air classification stated using standard nomenclature such as ISO or U.S. Fed. Std. 209E ⁸⁷⁸⁹. In general aseptic-processing terms, the critical area should be Class 100 (ISO 5) ²²³, and for isolators the interior should meet Class 100 (ISO 5) with a commonly used Class 100,000 (ISO 8) background ²²⁶. For PET aseptic workstations, FDA recommends HEPA integrity testing at installation and at least every six months ²²⁵²²⁷.
Environmental monitoring should include periodic microbiological testing of the aseptic environment, with described sampling methods, frequency, culture media, and incubation conditions; acceptable methods include swabs/contact plates for surfaces and settle plates or dynamic air samplers for air, plus airborne non-viable particle counting ⁸⁷⁸⁹²³⁴.

The short half-life problem

This is where radiopharmaceutical cGMP diverges most sharply from conventional drugs. A fluorine-18 product has a shelf life measured in hours; a shorter-lived tracer may be gone before a conventional sterility test could ever read out. FDA accommodates this by decoupling distribution from final release, while preserving hard stops before human administration.

Conditional distribution before sterility is complete. A PET drug may be released for distribution, but not administered, while sterility testing is still pending; all finished-product tests other than sterility must be completed or in progress at the time of shipment ³⁸³⁷. Transportation deadlines are recognized as a legitimate reason to distribute before the full sterility result is available ³⁸³⁷³⁴³⁵.
All specifications except sterility before final release. Before final release for human administration, FDA requires a determination that the batch (or sub-batches, if applicable) meets specifications except for sterility ³⁸³⁷¹¹⁶¹¹⁷.
Endotoxin before administration. A product may be distributed after the bacterial endotoxin test is initiated, but the endotoxin result must meet its acceptance criteria before the product is administered to humans ³⁸³⁷.
A later positive sterility test. If a sterility test subsequently shows microbial growth, FDA expects an immediate, documented, complete investigation with prompt corrective actions ³⁸³⁷. The Phase 1 investigational-drug guidance makes the same point: if positive sterility results are obtained after a short-shelf-life product has been released, an investigation should be performed ⁵⁷.
Same-day sub-batches. For very short half-life PET drugs produced in multiple sub-batches on the same day, finished-product testing of the initial sub-batch can support release of subsequent sub-batches, provided enough sub-batches have already demonstrated a product meeting acceptance criteria and the initial sub-batch meets all criteria ³⁸³⁷.
Parametric release is narrow. Parametric release (substituting process-parameter and load-monitor data for end-product sterility testing) applies to terminally moist-heat sterilized products and may be used only when the sterilization process is in control and the documented conditions are met, and, for application products, only when set out in the approved application ⁴⁹²³⁶²³⁸²⁴⁰. It is not a general workaround for aseptically produced radiopharmaceuticals.

Stability and expiration dating are similarly compressed. FDA expects stability data based on batches prepared at the upper range of the proposed radioactive concentration, stored in the same container/closure used for production (with the vial stored inverted in the sample formats), and expiration dating expressed in hours from end of synthesis (EOS) ¹¹⁴⁸⁶. At the end of the proposed dating period, the sample formats evaluate appearance, radiochemical purity, radionuclidic purity, pH, and stabilizer concentration where present ¹¹⁴⁸⁶. The NDA sample format shows complete release and stability data for three batches; the ANDA format shows release data for three batches plus stability data on one of the three ¹¹⁴.

CMC content for PET drugs in clinical trials

FDA's guidance on Investigational New Drug Applications for PET drugs sets out the CMC package that supports a clinical trial, and it is more granular than many sponsors expect for an investigational product.

Drug substance (the radioactive molecule, including the radionuclide). General information (name, structure, physical/chemical/biological properties); manufacturing information (identity and responsibility of each facility involved in the radionuclide, nonradioactive intermediate/precursor, and radioactive drug substance; flow diagram and process description; batch formula and equipment); control of starting materials, reagents, solvents, and auxiliary materials; controls of critical steps and intermediates; characterization; and reference standard information ²⁶²¹¹.
Drug product. Description and composition (components, grade, amounts, function, diluent, container closure); manufacturing information and controls; control of excipients; and control of the drug product, including tests performed, tentative acceptance criteria, and analytical data from three batches made similarly to the clinical batches ²¹¹.
Test methods. Analytical procedures for the specification tests of both drug substance and drug product, with impurities discussed ²¹¹.
Stability. Data to support the expiration dating period, performed at the upper range of radioactive concentration produced, plus stability and storage information for the precursor and reference standards ¹³⁶²⁶²¹¹.
Environmental assessment. PET drug products generally qualify for a categorical exclusion from an environmental assessment under 21 CFR 25.31, and sponsors may request that exclusion ¹³⁶¹⁹⁰.

Externally sourced drug substance (for example, fluoride F-18 obtained from an outside producer) carries its own control expectation: the application should include the acceptance procedures used to accept and release each lot for PET drug product production ¹⁹⁰, and specifications and test procedures controlling identity, purity, and quality for the starting/target material ²¹.

Dosimetry: diagnostic radiopharmaceuticals

For imaging agents, dosimetry is a core part of the safety and clinical-pharmacology package, and FDA is specific about the calculation.

Sponsors should submit enough animal or human data to allow a reasonable calculation of radiation absorbed dose to the whole body and to critical organs on administration to a human subject ⁶⁸.
At minimum, provide absorbed dose estimates for all organs and tissues in standardized anthropomorphic phantoms established in the literature, such as MIRD phantoms, and calculate effective dose as defined by ICRP Publication 60 ⁶⁸.
For pediatric development, provide absorbed dose for all intended pediatric age groups using standard phantoms (newborn, 1-year-old, 5-year-old, 10-year-old, and 15-year-old) ⁶⁸.
Use standardized internal-dose methods such as the absorbed fraction method described by MIRD and ICRP, and specify the methodology, including body models, equations, assumptions, sample calculations, and any software used ⁶⁸⁶⁵⁶⁶.
Collect organ and tissue distribution data over time, with human pharmacokinetic and biodistribution studies performed during Phase 1 to characterize the disposition of radioactivity and optimize imaging protocols ⁶⁸⁹⁷.
Administer the smallest radiation absorbed dose practical that still provides an adequate diagnostic examination ⁶⁶¹⁹³.

Dosimetry: therapeutic radiopharmaceuticals

For radioligand and radionuclide therapy, dosimetry drives dose selection, and FDA's expectations have sharpened in recent draft guidance on dosage optimization.

Do dosimetry early. Phase 1 studies of radioactive drugs must include studies that obtain sufficient data for dosimetry calculations, and each novel molecular entity should have dosimetry studies performed early in clinical development to characterize organ absorbed dose and support dosage optimization ⁷⁰. Where an IND is supported by prior studies that did not collect dosimetry, FDA expects dosimetry to be obtained during the IND trial (for example, via a dedicated imaging study or a lead-in cohort) to verify the assumptions used in dosage selection ⁷⁰.
Optimize dosage, not just find an MTD. Dosages intended to support a future marketing application should be based on the totality of data across a range of dosages and not solely on the maximum tolerated dose or normal-organ tolerances ⁷⁶.
Prespecify cumulative limits. Protocols should pre-specify a limit on cumulative administered activity and the corresponding absorbed doses to critical organs, justified by available organ-tolerance data, including a maximum cumulative absorbed dose accounting for prior and current therapy ⁸².
Treat external-beam limits as a starting point only. EBRT organ-tolerance literature should be used as a starting point, but its transferability is limited by differences in dose rate and radiation distribution, so optimized radiopharmaceutical therapy dosages may be higher or lower than an EBRT-limited dose; organ-tolerance doses for systemically administered radiopharmaceuticals are not directly comparable to EBRT ⁷¹⁶²⁷².
Account for radiobiology of alpha emitters. Because alpha particles have higher biological effectiveness, FDA recommends applying a relative biological effectiveness (RBE) of 5 when using EBRT-derived organ-tolerance data for alpha-emitting therapeutic radiopharmaceuticals ¹⁴¹⁵.
Prefer direct imaging. Direct imaging of the therapeutic product is preferred when possible; surrogate-analog dosimetry may supplement direct imaging when direct imaging is not possible or has substantial limitations ⁷⁰.

The RDRC pathway: basic research without an IND

Not every human study of a radioactive drug proceeds under an IND. Under 21 CFR 361.1, the Radioactive Drug Research Committee (RDRC) pathway allows human research with radioactive drugs to proceed without an IND when the work is basic science research, not intended for immediate therapeutic or diagnostic purposes and not to determine safety or effectiveness for those purposes, and is approved by an FDA-approved RDRC and an IRB before initiation ¹²⁴²⁵⁰. The dose-related conditions are strict: the dose must be known not to cause any clinically detectable pharmacological effect in humans; the total radiation dose must be the smallest practical to perform the study without jeopardizing its benefits; and the dose must be within specified limits ¹²⁴²⁴⁹. If the drug produces a pharmacological effect or the study has therapeutic or diagnostic intent, an IND is required instead ¹²⁴²⁵⁰.

Practical takeaways

For a late-phase radiopharmaceutical program, the through-line is convergence on the commercial control state:

Confirm your framework early: part 212 for PET, parts 210/211 for non-PET radioactive drugs, with USP <823> available as the investigational-PET path ¹⁶¹⁵¹².
Build the process-verification record now, aiming at three consecutive acceptable runs for new or changed processes and release/stability data on at least three consecutive batches for the application ³²²¹².
Lock down aseptic controls: operator media-fill qualification, filter integrity testing before release, defined air classification, and environmental monitoring ²¹⁷²¹⁴⁸⁷⁸⁹.
Treat the short half-life as a documented, bounded exception, not a relaxation: distribute conditionally, but hold administration until all non-sterility specifications and endotoxin criteria are met, and have a written response to a later positive sterility result ³⁸³⁷⁵⁷.
Generate dosimetry on the schedule FDA expects, whole-body and critical-organ absorbed dose plus effective dose for diagnostics, and early, prespecified, cumulative-limit dosimetry for therapeutics ⁶⁸⁷⁰⁸².

These are the areas a reviewer will probe first, and each maps to a specific FDA guidance a sponsor can cite in its own development plan.