FDA Controls for Retroviral/Lentiviral CAR-T Vectors

What FDA has required for retroviral and lentiviral CAR-T vectors to control insertional oncogenesis (2021 to 2026)

Autologous CAR-T cell therapies genetically modify a patient's T cells with an integrating gammaretroviral or lentiviral vector. Because the vector inserts its payload into the host genome, every product in this class carries a theoretical risk of insertional mutagenesis, insertional oncogenesis, and generation of a replication-competent retrovirus or lentivirus (RCR/RCL). Over the last five years, FDA (CBER) has approved and supplemented a group of T-cell products built on both vector platforms: the lentiviral products Abecma (idecabtagene vicleucel, BLA 125736), Breyanzi (lisocabtagene maraleucel, BLA 125714), and Carvykti (ciltacabtagene autoleucel, BLA 125746), and the gammaretroviral products Yescarta (axicabtagene ciloleucel, BLA 125643) and Tecartus (brexucabtagene autoleucel, BLA 125703), the latter two of which received new indications and label changes in this window.

This article summarizes what FDA actually required across those BLA reviews to control unsafe genetic modification and insertional oncogenesis. The requirements cluster into four layers: (1) vector design, (2) manufacturing release testing (RCR/RCL and vector copy number), (3) nonclinical genotoxicity/integration assessment, and (4) postmarketing long-term follow-up and labeling for delayed malignancy. The common theme is that FDA does not rely on any single control; it requires a layered, defense-in-depth package and defers the residual, unquantifiable insertional-oncogenesis risk to mandatory 15-year follow-up.

1. Vector design as the first line of defense

FDA's reviews treat vector architecture as the primary engineering control against insertional mutagenesis, and the agency documented the design features in the CMC and pharmacology/toxicology reviews.

For Breyanzi, FDA described the vector as a replication-incompetent, self-inactivated (SIN) lentiviral vector ¹²⁹¹³⁰ that was "designed to remove any known viral enhancer elements," a feature the reviewers explicitly cited as the mitigation for insertional mutagenesis risk ¹²². FDA agreed with the sponsor that the vector integrated similarly to wild-type HIV-1, that integration correlated with expected genomic features, and that analysis of tumor suppressor genes did not show preferential integration at loci associated with higher transformation risk ¹²³. The agency also judged the likelihood of recombination to a replication-competent virus "extremely low" and stated it had not been detected ¹²⁹¹³⁰.

For Abecma, FDA likewise recorded that the lentiviral vector used a self-inactivating design "to remove any known viral enhancer elements," identifying that architecture as the mitigation for insertional mutagenesis, and noted that insertion-site analysis showed no areas of increased or preferred integration ¹⁷⁴¹⁷⁶. Notably, FDA stated that lot-release testing limited the average integrated vector copy number per transduced cell to the range supported by clinical-trial experience ¹⁷⁴¹⁷⁶.

For Carvykti, FDA confirmed the use of a lentiviral vector to modify autologous T cells ⁸⁸⁹² and that the likelihood of RCL generation was reduced by the vector design ². The released Carvykti review text did not clearly confirm the SIN status, LTR enhancer/promoter deletions, or the internal promoter identity (the promoter is referenced but redacted) ⁸⁸, so the design-level detail is thinner in the public record than for Abecma and Breyanzi.

A consistent limitation across all three lentiviral products: the internal promoter driving CAR expression is redacted in the released reviews, so the public record documents the SIN/enhancer-deletion strategy and FDA's safety conclusions but not the full construct map.

2. Manufacturing release testing: RCR/RCL and vector copy number

Replication-competent virus (RCR/RCL)

FDA's expectations here are anchored to its gene-therapy guidance and are applied at three levels: the vector, the drug substance/product, and the treated patient. The specifics vary by product, which is instructive for sponsors.

Carvykti: The final lentiviral vector is tested for RCL by co-culture "in accordance with current FDA guidance" before release into manufacturing, and absence of RCL in the final product is confirmed by a defined assay (identified in the CMC review as test method TV-TMD-33726) ¹². FDA reported no RCL detected in vector lots, drug product, or treated subjects at data cutoff ²⁴⁵. The clearest stated acceptance criterion is "absence of RCL in the final product"; no numeric limit is given in the released text ¹.
Abecma: The final anti-BCMA lentiviral vector and the production cells were both tested for RCL before release "in accordance with current FDA guidance," and no RCL was detected in vector lots, the transduced product, or patient blood ⁷⁸⁹¹⁰. The released rows do not state a separate numeric release criterion beyond the requirement to test ⁷⁸.
Breyanzi: FDA treated RCL as a theoretical manufacturing concern reduced by process controls, and it confirmed correct vector identity before transduction ¹¹. Importantly, the CMC review states that testing for RCL in the final drug product was not included in the commercial specification ¹³¹⁴¹⁵. The RCR/RCL control for Breyanzi was instead shifted to patient long-term follow-up (below).
Yescarta (gammaretroviral): FDA did not require routine RCR sample collection in the postmarketing registry, but the ZUMA-1 long-term follow-up protocol included testing for development of RCR alongside product persistence through 15 years; no RCR or insertional mutagenesis events were reported in the BLA ²⁰²¹²².
Tecartus (gammaretroviral): FDA accepted the absence of RCR testing on the final cellular product, explicitly stating "the absence of RCR testing on the final cellular product is acceptable" under FDA guidance, because a product sample retention program allowed retrospective RCR analysis if needed; the vector master/working cell banks and clinical lots were negative for RCR ²³. Patient blood was monitored for RCR post-infusion, with PBMC/RCR samples collected through Month 24 ²⁶²⁸³³.

The Breyanzi patient-level RCR/RCL schedule is the most explicit in the reviews and reflects the FDA retroviral-vector follow-up paradigm: follow-up studies must monitor for RCL/RCR and test for persistent vector sequences (PVS) and RCR; if both are undetectable at 5 years the patient can stop in-person assessments (health-status contact continues to 15 years), and if either is detectable at 5 years the patient continues annual PVS and RCR testing to 15 years or until both are undetectable. The reviewer states this schedule is in accordance with FDA guidance for retroviral vector-based gene therapy products ¹¹.

Vector copy number (VCN)

VCN is the manufacturing lever most directly tied to insertional-oncogenesis risk, because integrated copies per cell scale the number of insertion events. The released reviews show FDA using it but frequently redact the numeric limit:

Abecma: FDA required that lot-release testing limit the average integrated VCN per transduced cell to the clinical-trial experience ¹⁷⁴¹⁷⁶. This is the clearest example in the dataset of a VCN-based release control tied to insertional risk.
Yescarta: FDA limited the average vector copy number to less than a specified value to mitigate insertional mutagenesis risk, but the numeric limit is redacted in the released review ⁴⁸.
Carvykti, Breyanzi, Tecartus: No VCN limit or specification appears in the released review pages retrieved ⁴³⁴⁴⁴⁷⁷³⁷⁴. Absence in the public text is not evidence FDA imposed none; VCN specifications are routinely part of the release panel but the numeric criteria are commonly redacted.

3. Nonclinical integration and clonality assessment

None of these sponsors ran traditional in vitro or in vivo genotoxicity/carcinogenicity studies; FDA instead accepted a weight-of-evidence package built on integration-site analysis and growth-transformation assays, supported by the published literature on vector-transduced T cells.

Yescarta: FDA noted the applicant did not conduct genotoxicity/carcinogenicity studies and relied on published nonclinical and clinical information on retroviral vector-transduced T cells; no insertional mutagenesis events were reported in the BLA ⁵⁷⁶³⁶⁵.
Tecartus: FDA reviewed an integration-site analysis using NGS on transduced and non-transduced T cells from three healthy donors, quantifying each integration site. The review found no dominant integration sites suggesting a clonal-expansion advantage, but did note low-frequency integration in exons that could indicate gene disruption; functional assays for proliferative-advantage clones were not performed. FDA judged the malignant-transformation risk low based on the analysis plus a literature review ⁷³⁷⁴.
Carvykti: FDA assessed insertional risk with a weight-of-evidence approach combining lentiviral integration-site analysis on preinfusion product and a cytokine-independent growth assay. The data showed the vector does not preferentially integrate at genomic sites of concern for oncogenic transformation and showed no uncontrolled growth without exogenous cytokine, supporting the conclusion that insertional events from the transduction method carry minimal oncogenic-transformation risk ⁸⁵⁸⁶⁸⁹.
Abecma: Integration-site analysis across 20 clinical lots found no promoter-region or oncogene-proximal integration preference, and an independent growth assay showed no malignant transformation; on that basis FDA did not recommend a dedicated post-approval insertional mutagenesis study ⁹⁹¹⁰⁰. In the clinical dataset, integration analysis showed clonal heterogeneity that did not support monoclonality ⁴⁵⁴⁶.
Breyanzi: FDA's mitigation for insertional-mutagenesis and secondary-malignancy risk was framed largely as postmarketing pharmacovigilance (registry, long-term follow-up safety study, targeted questionnaires, labeling, and a transgene assay service) rather than a mandated pre-approval clonality-monitoring program; in one case insertion-site analysis did not show integration near oncogenic loci ⁴⁷.

The clearest example of FDA requiring integration-site work as an ongoing obligation is Carvykti's clinical follow-up: FDA required post-treatment monitoring for subsequent malignancies with collection of tumor/blood/bone marrow/autopsy specimens when feasible, DNA/RNA/protein analysis for lentiviral elements, and lentiviral integration-site analysis triggered when at least 1% of cells in a blood sample are positive for vector sequence, with close monitoring for persistent monoclonality or clonal expansion ⁴³⁴⁴. Tecartus's pharmacovigilance plan similarly commits to tumor-tissue collection and analysis for retroviral vector/RCR, with insertional mapping if a T-cell malignancy is suspected ²⁶¹²⁶²²⁶³.

4. Mandatory 15-year long-term follow-up

Across every product in this class, FDA imposed a postmarketing requirement (PMR) for long-term follow-up of approximately 15 years, reflecting its long-term follow-up guidance for gene therapies with integrating vectors. This is the mechanism through which FDA manages the residual insertional-oncogenesis risk that manufacturing controls and nonclinical data cannot fully exclude.

Carvykti: 15-year follow-up after the last infusion, monitoring new/secondary malignancies, insertional mutagenesis (blood/biopsy workup and vector integration-site analysis if vector-positive), and RCL, plus neurologic, autoimmune, hematologic, and infection events. FDA determined a safety PMR under FDAAA section 505(o) was required because spontaneous reporting and Sentinel would be insufficient to characterize the secondary-malignancy risk. The registry is a prospective, multicenter, observational study of at least 1,500 patients followed 15 years (study completion June 30, 2041; final report June 30, 2042), alongside a clinical-trial long-term follow-up study collecting new malignancies and RCL data ¹⁵¹¹⁵²¹⁵³¹⁵⁶¹⁵⁷¹⁵⁹¹⁶²¹⁶³¹⁶⁹.
Abecma: 15-year PMR registry of at least 1,500 patients for secondary malignancies and long-term safety, addressing theoretical insertional-mutagenesis and product-related secondary-malignancy concerns; also captures CRS, neurotoxicity, prolonged cytopenia/rescue transplant, and HLH/MAS (final study completion June 30, 2041; final report June 30, 2042) ¹⁷⁴¹⁷⁶¹⁸⁴¹⁸⁵¹⁸⁷¹⁹⁰¹⁹⁵. FDA noted no routine RCL sampling was planned for Abecma, so it did not confirm an explicit RCR-monitoring obligation in the retrieved rows ¹⁹³.
Breyanzi: PMR 017001, a long-term follow-up registry of 1,500 DLBCL patients for 15 years, justified by the lack of long-term safety data; the retrieved rows describe it in general long-term safety terms rather than enumerating insertional-event or RCR endpoints ²⁰¹²⁰². Patient-level RCR/PVS testing is handled through the follow-up schedule described in section 2 ¹¹.
Yescarta: 15-year PMR registry of at least 1,500 subjects with secondary malignancy as the primary endpoint; delayed risks to monitor include secondary malignancy, insertional mutagenesis, and RCR, with fresh tumor specimens obtained to determine whether a malignancy is product-related. No routine RCR sample collection was planned in the registry itself ²⁰³.
Tecartus: 15-year long-term follow-up registry with subsequent neoplasms as the primary objective; RCR/retroviral vector sequence testing is requested if a patient develops a subsequent neoplasm, and insertional-mutagenesis toxicity is the stated rationale for the study, a PMR safety study under section 505(o) ²⁰⁶²⁰⁷²⁰⁸²⁰⁹²¹⁰²¹¹²¹²²¹³²¹⁴²¹⁵²¹⁶²¹⁷.

5. Labeling: the class-wide secondary (T-cell) malignancy action

The most consequential recent regulatory action is the class-wide labeling change following FDA's identification of T-cell malignancies, including CAR-positive tumors, after treatment with BCMA- and CD19-directed autologous CAR-T therapies. Every product in this class now carries a boxed warning for secondary hematological malignancies that specifically states T-cell malignancies have occurred, may present as soon as weeks after infusion, and may be fatal. The labeling also requires lifelong monitoring for secondary malignancies and directs clinicians to contact the manufacturer for reporting and instructions on collecting patient samples for testing (which, per the pharmacovigilance plans, feed retroviral/lentiviral integration and RCR analysis).

Carvykti: Boxed warning for secondary hematological malignancies (including MDS and AML) plus the T-cell malignancy statement; lifelong monitoring; report to Janssen at 1-800-526-7736 with sample collection; supported by REMS and the PMR registry ²¹⁸²²²²²⁹²³⁰²³²²³³.
Abecma: Boxed warning updated to include secondary hematological malignancies with the T-cell malignancy statement; lifelong monitoring; report to Bristol-Myers Squibb at 1-888-805-4555 with sample collection. FDA noted no T-cell lymphoma was seen in the Abecma study itself, but the class signal triggered a class safety labeling change into the boxed warning ²³⁴²³⁵²³⁶²³⁷.
Breyanzi: 2024 boxed warning for secondary hematological malignancies with the T-cell malignancy statement; report to Bristol-Myers Squibb at 1-888-805-4555; long-term follow-up/registry to 15 years ²⁴¹²⁴³²⁴⁵²⁴⁶.
Yescarta and Tecartus: Boxed warning for secondary hematological malignancies with the T-cell malignancy statement (including CAR-positive tumors that may be fatal); lifelong monitoring; contact Kite at 1-844-454-KITE (5483) for sample collection ²⁴⁷²⁴⁸²⁵²²⁵⁵²⁵⁶²⁵⁷²⁵⁸²⁵⁹²⁶⁰.

Practical takeaways for sponsors

FDA expects a layered control strategy, not a single test. Vector design (SIN architecture, enhancer/promoter deletions), RCR/RCL release testing per FDA guidance, VCN control, nonclinical integration-site and growth-transformation data, and 15-year long-term follow-up are treated as a package, and weakness in one layer raises the bar on the others.
RCR/RCL testing is negotiable in location but not in principle. FDA accepted the absence of final-product RCR testing for Tecartus (given sample retention) and did not put final-product RCL in Breyanzi's commercial spec, but in each case it required a defensible alternative (retention for retrospective testing, or patient-level PVS/RCR monitoring on the guidance-defined schedule).
VCN is a recognized insertional-risk lever. Where disclosed, FDA tied release-level VCN caps to clinical-trial experience (Abecma) or a specified maximum (Yescarta); expect a justified average-VCN specification even though numeric limits are usually redacted publicly.
Integration-site analysis is increasingly an ongoing obligation, not just a filing deliverable. Carvykti's 1%-vector-positive trigger for integration-site analysis and Tecartus's tumor-tissue integration mapping on suspected T-cell malignancy show FDA operationalizing clonality surveillance into the follow-up phase.
The 15-year follow-up and the class-wide T-cell malignancy boxed warning are now fixed expectations for any integrating-vector T-cell therapy, and sponsor sample-collection infrastructure must be able to support integration and RCR analysis on any post-treatment malignancy.

A useful follow-up to ask Rhizome directly: pull the exact PMR milestone tables and the specific RCR/RCL and VCN acceptance criteria (where unredacted) for a single product, or compare the gammaretroviral (Yescarta/Tecartus) versus lentiviral (Abecma/Breyanzi/Carvykti) RCR follow-up schedules side by side.