FDA Replication-Competent Virus Testing and Long-Term Follow-Up for Gene Therapy Products

FDA expectations for replication-competent virus testing and long-term follow-up of gene therapy products

FDA handles replication-competent virus (RCV) risk and delayed adverse events in gene therapy through a small set of CBER guidances, most finalized in January 2020, together with product-specific decisions at BLA approval. The guidances set numeric testing expectations for retroviral, lentiviral and adenoviral vectors. They also set a risk-based long-term follow-up (LTFU) framework that ranges from no follow-up for low-risk modalities to 15 years for integrating vectors. In approvals, CBER has applied that framework mainly through 15-year postmarketing requirement (PMR) registries for products with integrating vectors, and through comparable registries for gene-edited products. The AAV products reviewed here have generally carried voluntary long-term follow-up rather than LTFU safety PMRs.

Key takeaways

RCR/RCL lot testing is quantitative. Test at least 5% of vector supernatant (enough volume for a 95% probability of detecting 1 RCR per dose equivalent), and 1% or 10^8 cells (whichever is less) of end-of-production and ex vivo transduced cells, by co-culture on a permissive cell line ²⁹⁹³⁰⁰.
RCA has a numeric ceiling. The 2020 CMC guidance recommends a maximum of 1 RCA in 3 × 10^10 viral particles, tested on each production lot ¹⁶⁷.
The 2020 RCR guidance sets a default patient monitoring schedule. Sample before treatment, then at 3, 6 and 12 months, then yearly for up to 15 years. Yearly sampling can stop if all first-year assays are negative ³⁹³. Approvals have varied from this default.
LTFU duration depends on the vector. Integrating vectors get 15 years. Genome editing, latent herpesvirus vectors and persistent microbial vectors get up to 15 years. AAV gets up to 5 years. Plasmid, RNA, poxvirus and adenovirus products are listed as not requiring LTFU ³³²³³⁴³³⁵³³⁶³³⁷³³⁸.
Approval precedent centers on 15-year PMR registries for products using integrating vectors, such as Kymriah, Zynteglo, Lyfgenia, Lenmeldy, Tecelra and Aucatzyl ³²¹³²²⁸¹⁸⁵⁸⁶⁸⁹⁹². The gene-edited product Casgevy has a sponsor-proposed post-licensure registry consistent with the 15-year recommendation ⁸³⁸⁴.

The core guidance framework

Guidance	Status / date	What it governs
Testing of Retroviral Vector-Based Human Gene Therapy Products for Replication Competent Retrovirus During Product Manufacture and Patient Follow-up	Final, January 2020	RCR testing during manufacture (material, amount, methods) and patient follow-up. The same expectations apply to gammaretroviral RCR and lentiviral RCL ¹⁷¹⁸¹
Chemistry, Manufacturing, and Control (CMC) Information for Human Gene Therapy INDs	Final, January 2020	RCR testing at multiple production points, RCA limits, replication-competent virus acceptance criteria in Phase 1 specifications ¹⁹¹⁶⁵¹⁶⁶¹⁶⁷
Long Term Follow-Up After Administration of Human Gene Therapy Products	Final, January 28, 2020	Risk-based decision on whether LTFU is needed and how long it should run, protocol content, post-licensure planning ¹⁵²¹⁵³¹⁷²
Considerations for the Development of Chimeric Antigen Receptor (CAR) T Cell Products	Final, January 31, 2024	15-year follow-up for CAR T cells with an integrated transgene ¹⁵⁸
Design and Analysis of Shedding Studies for Virus or Bacteria-Based Gene Therapy and Oncolytic Products	Final, August 2015	Shedding study design, with timing driven by replication competence ⁴⁰⁴²⁴³
Postapproval Methods to Capture Safety and Efficacy Data for Cell and Gene Therapy Products	Draft (not for implementation)	Collection of postapproval long-term safety and durability data, and follow-up continuity if a sponsor stops operating ¹³¹

The 1998 Guidance for Human Somatic Cell Therapy and Gene Therapy is still the source for several assay design principles for replication-competent adenovirus (RCA) ¹⁶⁸¹⁷¹.

Replication-competent virus testing during manufacture

Retroviral and lentiviral vectors (RCR/RCL)

The RCR guidance covers both retroviral vector supernatant given directly to patients and ex vivo transduced cells ¹⁷¹⁸. The CMC guidance recommends RCR testing at multiple points in retroviral vector production and refers sponsors to the RCR guidance for the details ¹⁹.

Material	FDA recommendation
Vector producer cell master cell bank (MCB)	Test both cells and supernatant on a cell line permissive for the relevant RCR. If an ecotropic retroviral vector was used to derive the producer cells, test for ecotropic as well as amphotropic RCR ²⁹⁷²⁹⁸
Vector supernatant	Test at least 5% of total supernatant by amplification on a permissive cell line. The volume tested should give a 95% probability of detecting RCR at 1 RCR per dose equivalent. The older 300 mL alternative for lots over 6 liters may not apply to current manufacturing ²⁹⁹
End-of-production cells and ex vivo transduced cells	Test 1% or 10^8 cells, whichever is less, by co-culture with a permissive cell line ³⁰⁰

Culture-based amplification has typically used at least five passages. Fewer passages are acceptable if they give similar assay sensitivity. Amplified material is then read out with an indicator assay or with methods such as PCR ³⁰¹³⁰². For ex vivo transduced cell lot release, PCR may replace a culture-based method, particularly when time is short. Any alternative method should be developed in consultation with CBER and supported by sensitivity, specificity and reproducibility data ³⁰³²⁹⁶.

The RCR guidance also says manufacturing and release RCR testing continues after licensure ¹⁴⁶.

Adenoviral, AAV and other vectors

The CMC guidance recommends testing non-replicating viral vectors for replication-competent, parental or wild-type virus at the manufacturing stage where it is most likely to be detected. Phase 1 IND specifications should include acceptance criteria for replication-competent virus safety testing ¹⁶⁵¹⁶⁶.

RCA: Qualify the master viral bank for RCA and test drug substance or drug product from each production lot. The recommended maximum is 1 RCA in 3 × 10^10 viral particles ¹⁶⁷. The 1998 guidance prefers a cell-culture cytopathic-effect assay on final vector. It recommends validating sensitivity by spiking in decreasing amounts of wild-type adenovirus and using a quantitative assay that can report RCA per patient dose ¹⁶⁸. Older FDA guidance, predating the 2020 CMC recommendation, called for no more than 1 plaque-forming unit of RCA, or equivalent, per patient dose ¹⁶⁹.
Replication-competent AAV (rcAAV): The 2020 CMC guidance identifies helper-virus-dependent rcAAV as a possible contaminant of AAV preparations, with possible immune-response risks. The passages reviewed did not give an rcAAV-specific numeric limit ¹⁷⁰.
Other vectors: Master viral banks for replication-defective or replication-selective vectors should be shown to be free of replication-competent virus ¹⁷¹.

In approvals, the Elevidys review states that the product is tested for rcAAV and that all batches had tested negative. The passage does not give a numeric acceptance limit ¹¹⁰. The Papzimeos (zopapogene imadenovec) review describes design features intended to lower the risk of generating replication-competent adenovirus in a non-replicating gorilla adenovirus product ¹¹¹¹¹².

Patient monitoring for RCR/RCL

For retroviral vector products, the RCR guidance recommends:

Schedule: a pre-treatment sample, then samples at 3, 6 and 12 months after treatment, then yearly for up to 15 years. If all post-treatment assays in the first year are negative, yearly sampling can stop and a yearly medical history review will generally be enough. Sponsors can also propose stopping active testing, with support from accumulated product-specific monitoring data ³⁹³.
Assay choice: serology for RCR-specific antibodies, or PCR of peripheral blood mononuclear cells for RCR-specific DNA. PCR may be preferable when vector or producer cells are given directly, or when the patient's immunocompromise may limit antibody production ³⁹⁴³⁹⁵.
Positive results: confirm every positive with a direct culture assay to obtain and characterize an infectious isolate. Report positive patient results immediately as an IND safety report, and expand follow-up in consultation with CBER ³⁹⁶³⁹⁷.
Event-driven testing: collect and test relevant samples when an adverse event suggests retrovirus-associated disease. Assay pertinent neoplastic or autopsy tissue within the 15-year window ¹³⁷³⁹⁸.

After licensure, a product-specific risk assessment may support dropping periodic patient RCR testing. Testing should still be done when adverse events suggest RCR ¹⁴⁶.

How CBER applied RCL testing in approvals

All of the reviewed lentiviral approvals treated RCL as a theoretical manufacturing risk. They relied on release testing of vector and cells, and none reported detecting RCL in the clinical program. The depth of patient monitoring varied by product.

Product	RCL handling described in FDA review documents
Kymriah (tisagenlecleucel)	Vector tested by co-culture. Final transduced cells tested by co-culture or PCR. No RCL detected in trials ²²⁹²³⁰. Routine post-infusion RCR sampling was not planned for the registry. Instead, tissue from any secondary malignancy was to be analyzed for vector persistence ³²³³²⁴³²⁵
Abecma (idecabtagene vicleucel)	Final vector and production cells tested before release. No RCL detected in clinical vector lots or transduced product ²³⁴
Carvykti (ciltacabtagene autoleucel)	Co-culture testing of vector material before release. Final product absence-of-RCL assay. A qualified patient PBMC monitoring assay ²³⁵²³⁶²³⁷
Zynteglo (betibeglogene autotemcel)	Vector-derived RCL was a study safety endpoint, with screening at months 3, 6, 12 and 24. No RCL was detected in 16 subjects with enough follow-up ²⁴⁰²⁴¹
Skysona (elivaldogene autotemcel)	Co-culture testing of the final Lenti-D vector and production cells. The LTFU schedule includes RCL assessments ²³⁸²³⁹
Aucatzyl (obecabtagene autoleucel)	Patient blood monitored with a validated method. No evaluable patient was positive at the June 9, 2023 data cutoff ²⁴⁴
Kresladi (marnetegragene autotemcel)	Vector and production cells tested before release. Peripheral blood tested at least three times. No RCL cases observed ²⁴⁶

None of the reviewed approval documents explicitly set an RCL-specific postmarketing requirement ²¹⁶²⁰⁴²⁰⁹²¹⁹²²⁰. Postmarketing obligations for these products have targeted insertional oncogenesis and secondary malignancy rather than RCL itself.

Long-term follow-up: the risk-based framework

The 2020 LTFU guidance does not set one follow-up period for every gene therapy. It ties LTFU to an assessment of delayed risk, and low-risk products generally do not need it ¹⁵²¹⁵³.

Deciding whether LTFU is needed

The guidance sets out a decision sequence ³⁴⁰³³⁰³⁴¹:

Does the product use genome editing? If yes, include LTFU.
If not, is the vector used only to modify cells ex vivo? If not, do preclinical data show the product persists? If it does not persist, delayed-event risk is low and LTFU may not be needed.
If the product persists, or is used only ex vivo, do vector sequences integrate or otherwise alter the genome? If not, can the product go latent and reactivate?
If the product integrates, alters the genome, or can go latent and reactivate, LTFU is needed.

Other inputs include evidence of delayed toxicity, persistence and duration of transgene expression, route of administration and biodistribution, and the disease and patient population, including background event rates and comorbidities ³⁴²¹⁶¹¹⁵⁹¹⁶⁰.

Recommended durations

Product type	LTFU duration
Integrating vectors (gammaretroviral, lentiviral, transposon)	15 years ³³⁵
CAR T cells with an integrated transgene	15 years ¹⁵⁸
Genome-editing products	Up to 15 years ³³⁴³⁴⁴
Herpesvirus vectors or oncolytics capable of latency	Up to 15 years ³³⁶
Microbial vectors that establish persistent infection	Up to 15 years ³³⁷
AAV vectors	Up to 5 years. Need is product-specific ³³²³³³
Plasmid, RNA, poxvirus, adenovirus	Listed as not requiring LTFU ³³⁸³³⁹

The August 19, 2026 Frequently Asked Questions: Developing Potential Cellular and Gene Therapy Products guidance repeats that observation may last as long as 15 years depending on product type. It says genome editing calls for longer monitoring because of potential cancer or off-target effects ¹⁵⁷.

Protocol content

For the first five years, or longer where the product calls for it, the LTFU protocol should cover ⁵¹⁵²⁵³:

case histories of exposure to mutagenic agents and other medicines
new malignancies, new or worsening neurologic, rheumatologic or other autoimmune disorders, new hematologic disorders, and infections that may be product-related
scheduled visits with history, physical examination and laboratory testing as needed, plus a way to report delayed events and hospitalizations

After that, subjects should be contacted at least once a year. Telephone or written questionnaires can replace office visits unless the protocol calls for extra screening ⁵⁴.

For genome editing, the safety plan should assess off-target editing, unintended consequences of on-target editing, and adverse events tied to chromosomal changes, immunogenicity and tumorigenicity ³⁴⁵.

Integrating vectors call for PCR of surrogate samples at least every six months for the first five years, then at least yearly for the next ten years, or until vector sequences are no longer detectable. Integration site analysis is triggered when at least 1% of cells in the surrogate sample are PCR-positive ⁴²⁰. If a predominant clone is found, determine the integration site and repeat the clonality analysis within three months. Compare the site against human genome and oncogene databases ⁴²¹.

Informed consent: subjects who receive a product with delayed-event risk should consent to the LTFU protocol. If LTFU runs under a separate protocol, they should consent to it before entering the initial trial ⁴²³⁴²⁴⁴²⁵.

Post-licensure follow-up: guidance and approval precedent

The LTFU guidance expects a post-licensure LTFU or surveillance plan for products with long-term risk. Clinical trials are usually too small and too short to detect delayed events ¹³⁸¹³⁹. FDA recommends:

submitting a pharmacovigilance plan with the BLA, which can absorb an ongoing IND LTFU study ¹³⁹¹⁴⁰
using new or existing registries, where appropriate, to track treated patients and adverse events ¹⁴¹
including the protocol, statistical analysis plan, milestones, adverse events of interest and observation period for any proposed or required study in the BLA ¹⁴²¹⁴³
clearly presenting immediate and long-term RCR risks in labeling for retroviral products ¹⁴⁶

Integrating and gene-edited products: 15-year PMRs as the norm

Product	Post-licensure LTFU at approval
Kymriah (2017)	PMR observational study B2401: at least 1,000 pediatric and young adult B-ALL patients, enrolled over 5 years and followed 15 years after infusion, mainly for secondary malignancy ³²¹³²²
Zynteglo	PMR registry REG-501 follows commercial recipients for at least 15 years for insertional oncogenesis and secondary malignancy. Trial patients are followed in LTF-303 ⁸¹⁸²
Lyfgenia	PMR REG-503: a 15-year observational safety study of 250 sickle cell disease patients. Trial patients are followed in LTF-307 ⁸⁵
Skysona	Safety PMR REG-502 for post-licensure patients, addressing insertional oncogenesis ⁸⁰. Labeling calls for lifelong malignancy monitoring. For the first 15 years, that means a CBC at least twice a year, and integration site analysis or other clonality testing at least twice in year one and then annually ¹⁹²¹⁹³¹⁹⁴
Lenmeldy	PMR prospective study of at least 17 children followed 15 years for secondary malignancy and long-term safety ⁸⁹⁹⁰⁹¹
Tecelra	PMR registry study following synovial sarcoma patients up to 15 years. Interim analyses at 5, 10 and 15 years post-approval ⁸⁶⁸⁷⁸⁸
Aucatzyl	15-year PMR safety registry AUTO1-LT2, plus trial follow-up in AUTO-LT1 (quarterly visits in year 1, every six months through year 5, then annually) ⁹²
Casgevy	Sponsor-proposed registry VX22-290-101 (target 400 patients, with allogeneic transplant comparator cohorts) and continued trial follow-up in VX22-290-131, consistent with the 15-year recommendation for gene editing. Endpoints include secondary malignancy and off-target editing ⁸³⁸⁴

CAR T cell class labeling for T-cell malignancies

In 2024, FDA treated T-cell malignancies after BCMA- or CD19-directed CAR T cell therapy as a class risk. It added the risk to boxed warnings and other labeling ³⁷¹³⁷²³⁷³:

The Abecma (April 2024) and Tecartus labeling directs lifelong monitoring for secondary malignancies, with instructions for reporting and sample collection ³⁷⁴³⁷⁵³⁷⁷³⁷⁸.
For Carvykti, FDA relied on pharmacovigilance and the existing postmarketing LTFU registry, which follows patients for up to 15 years. No new safety PMR was added with that supplement ³⁸³³⁸⁴³⁸⁵.

Lifelong monitoring in the label and 15-year registry follow-up therefore run side by side as separate tools ³⁸⁹³⁹⁰³⁹¹.

AAV products reviewed: voluntary follow-up, generally without LTFU safety PMRs

FDA's reviews describe AAV vectors as lacking a propensity to integrate or to reactivate after latency. That means a lower risk of delayed adverse events, though not zero, since low-level random integration can occur ²⁷⁴²⁷⁵²⁷⁶. Approval outcomes reflect this:

Luxturna: an ongoing study follows 41 trial participants for up to 15 years, and a proposed registry follows 40 patients for 5 years. The reviewer found no safety PMR or PMC was warranted ²⁸⁸²⁸⁹²⁹⁰²⁹¹.
Roctavian: 5 years of trial follow-up plus a 10-year extension (15 years in total), along with two voluntary observational studies. No long-term-follow-up safety PMR or PMC was required ²⁷⁰²⁷¹.
Hemgenix: a voluntary 15-year observational study of 250 patients. The safety PMRs concerned anti-AAV5 antibody assay work, not long-term follow-up ²⁷².
Beqvez: a planned 15-year post-authorization safety study meant to cover the up-to-5-year AAV follow-up FDA recommends ²⁷³²⁵⁰.

Shedding and replication competence

Replication competence also shapes shedding expectations. For replication-competent and oncolytic products, FDA recommends starting clinical shedding data collection in Phase 1, sampling for a possible second peak after in vivo amplification, and monitoring for longer ⁴²⁴⁴⁴⁵. Positive nucleic acid results for replication-competent products should be followed with infectivity assays. If a sponsor relies on qPCR alone, FDA will assume positive shed material is infectious ⁴⁶⁴⁷.

Practical implications for sponsors

Plan RCR/RCL assay capacity early. The 5% supernatant and 1% or 10^8 cell testing amounts, and the 95% detection-probability standard, drive sample volumes and assay sensitivity. PCR-based release for transduced cells needs agreement with CBER ²⁹⁹³⁰⁰³⁰³.
Expect 15-year follow-up for integrating or gene-edited products. Whether it is a PMR or a sponsor-proposed registry is product-specific (Casgevy's registry was sponsor-proposed). Recent PMR registries show a consistent pattern: secondary malignancy as the primary endpoint, a defined registry size, and interim analysis milestones ⁸¹⁸⁵⁸⁶⁹²⁸³.
Justify how much patient RCL monitoring you plan. The guidance schedule is the default, but approvals show variation, from scheduled screening (Zynteglo) to malignancy-triggered tissue analysis (Kymriah) ²⁴⁰³²³³²⁴.
For AAV, ground the follow-up plan in the 5-year guidance benchmark. Many sponsors have voluntarily extended follow-up to 15 years ³³²²⁷⁰²⁷²²⁷³.

Product-specific questions, such as the exact PMR milestone dates for a given BLA, integration site analysis results in individual reviews, or how the draft postapproval data guidance may change registry design, can be researched further against the underlying review documents.