What Does the FDA's CAR-T Boxed Warning on Secondary Cancers Actually Show?

In November 2023, the FDA opened an investigation into reports of secondary T-cell cancers in patients treated with CAR-T therapy, and by early 2024 had required a class-wide boxed warning — the agency's most serious label designation — across all six approved CD19- and BCMA-directed CAR-T products. The announcement triggered genuine alarm, since these are living, gene-modified cell therapies already approved for aggressive blood cancers with few other options. Two years on, the largest datasets — a 724-patient Stanford cohort, a 5,517-patient meta-analysis, FDA's own adverse-event mining, and a French national registry — have converged on a consistent picture: the signal that triggered the warning was real and worth flagging, but the absolute risk of a CAR-driven secondary cancer has turned out to be low, and it has not changed the calculus that CAR-T remains the better option for most patients who qualify for it.

What Triggered the FDA's Action

On November 28, 2023, the FDA issued a safety communication disclosing that it had received reports of T-cell malignancies, including CAR-positive lymphomas, in patients treated with BCMA- or CD19-directed autologous CAR-T products. The agency was explicit that it did not yet know whether the therapies caused the cancers or whether the finding would change the products' overall risk-benefit profile, but it opened a formal investigation. By January 25, 2024, the FDA had gathered enough detail to request a class-wide boxed warning, disclosing in a companion New England Journal of Medicine Perspective (authored by FDA officials Nicole Verdun and Peter Marks) that it was aware of 22 cases of T-cell cancer following CAR-T treatment, identified against a backdrop of more than 27,000 CAR-T doses administered in the United States since the first approval in 2017 — roughly 0.08% of treated patients, though the true rate could be somewhat higher given underreporting to adverse-event databases. Of the 22 cases, 14 had enough data for analysis: about half arose within the first year post-infusion, and in a subset where genetic sequencing was performed, the CAR transgene itself was found integrated into the malignant clone — evidence that, in at least some instances, the engineered cells were mechanistically implicated.

What the Boxed Warning Actually Requires

The FDA finalized the labeling changes in spring 2024, across all six approved products: Abecma, Breyanzi, Carvykti, Kymriah, Tecartus, and Yescarta. The boxed warning itself flags the risk of T-cell malignancy, and the FDA required accompanying updates to the Warnings and Precautions, Postmarketing Experience, and Patient Counseling sections, plus the Medication Guide given to every patient. Practically, this means prescribers must now monitor patients for secondary malignancies for life following CAR-T infusion, report any new cancer diagnosis to the manufacturer and to FDA's MedWatch system, and, if a secondary T-cell malignancy occurs, collect a specimen for testing to determine whether it carries the CAR transgene. Notably, the FDA did not withdraw or restrict any of the six therapies' approved indications — the warning is a monitoring and disclosure mandate, not a finding that the products' benefit-risk balance had shifted unfavorably.

What the Largest Studies Now Show

Since 2024, several much larger datasets than the FDA's initial case series have reported. A Stanford Medicine cohort of 724 CAR-T patients followed a median of three years, published in NEJM in June 2024, found a 6.5% cumulative incidence of any secondary cancer — comparable to rates seen after allogeneic stem cell transplantation — but only a single case of T-cell lymphoma, which on genetic analysis did not carry the CAR transgene and was molecularly distinct from the infused cells. A University of Pennsylvania series of 449 patients found 16 secondary cancers (3.6%) over a median 10-month follow-up, again with only one T-cell lymphoma. The largest analysis to date, a systematic review and meta-analysis of 5,517 lymphoma and myeloma patients published in Clinical Cancer Research (AACR) in 2024, found that 5.8% developed a second primary cancer over roughly 22 months of follow-up — but T-cell malignancies accounted for only about 1.5% of those secondary cancers, and the overall secondary-cancer rate was essentially identical to that seen with conventional (non-CAR-T) treatment of the same diseases (roughly 5% versus 4.9%). FDA's own mining of its adverse-event database identified 19 T-cell malignancy reports among 12,394 total CAR-T adverse-event reports — about 0.1%. A subsequent French national registry (DESCAR-T, 3,066 patients) again found just one confirmed CAR-positive T-cell lymphoma. Taken together, the field's working explanation has shifted: rather than insertional mutagenesis from the CAR vector being the primary driver, most researchers now point to pre-existing clonal hematopoiesis (mutations such as TET2 and DNMT3A already present in a patient's blood-forming cells before CAR-T), lymphodepleting chemotherapy, and the underlying disease itself as the dominant contributors, with vector insertion an occasional but not universal cofactor.

Weighing the Signal Against the Benefit

Context matters here on both sides. Patients eligible for CAR-T have already failed multiple lines of therapy for aggressive lymphoma, myeloma, or leukemia, diseases where median survival without effective salvage options is often measured in months. CAR-T has produced durable remissions, and in some cases apparent cures, that were not achievable before these therapies existed. Against that benefit, a secondary-cancer signal in the low single digits of a percent — and a CAR-attributable T-cell lymphoma rate that appears to be a small fraction of even that — is a real but modest addition to risk in a population that already carries elevated cancer and treatment-related risk from prior chemotherapy, radiation, and the malignancy itself. The honest description is a genuine, monitorable risk that has not, on the best current evidence, reversed CAR-T's favorable risk-benefit profile for its approved indications.

What This Means in Practice

The boxed warning does not mean patients should decline CAR-T therapy when it is indicated; oncology societies and the FDA have continued to support its use. It does mean informed consent conversations should explicitly cover the risk, that patients need lifelong post-treatment cancer surveillance rather than being considered "finished" once their original cancer is in remission, and that any new malignancy after CAR-T should be reported and, where feasible, genetically tested to determine whether the CAR transgene is present — data that will keep refining these risk estimates as registries mature.

Bottom Line

The FDA's boxed warning was a proportionate response to a real, if still incompletely understood, safety signal: at least a few dozen documented cases, in at least some of which the CAR vector was found inside the malignant cells. But the largest cohorts and meta-analyses assembled since the warning — covering thousands more patients than the original case series — have not found a secondary-cancer rate meaningfully higher than what's seen with comparable cancer treatments, and CAR-associated T-cell lymphoma specifically appears to affect a small fraction of one percent of treated patients. This is a story where the regulatory caution and the reassuring data are both true at once: patients need lifelong monitoring and honest disclosure, and CAR-T remains, for now, a therapy whose benefits clearly outweigh this particular risk for the relapsed and refractory cancers it treats.