Data-informed optimization of CAR T-cell therapy long-term follow-up

Abstract

Following administration of chimeric antigen receptor (CAR) T-cell therapy, extensive long-term follow-up (LTFU) requirements and complex data collection processes have posed significant challenges for patients and providers. To reassess whether a 15-year LTFU period remains scientifically justified, we convened a multistakeholder working group that included representatives from patient advocacy groups, academia, industry, and government. This analysis incorporates newly aggregated primary data on composite percentages of adverse events (AEs) reported by year for five Food and Drug Administration-approved CAR T-cell therapies. Combined with previously published research, the findings indicate that AEs are infrequently reported after 3 years post-infusion, and secondary T-cell malignancy—the AE of primary concern based on the mechanism of action of CAR T-cell therapy—has predominantly been reported within the first 2 years. Based on current cumulative safety data, a 5-year follow-up period may be scientifically sufficient in both clinical trial and commercial settings. Additionally, we propose a streamlined process that leverages technological advancements to automate the transfer of focused safety data from electronic health records into a third-party database. To facilitate implementation, we recommend feasibility testing of this updated data collection approach using an established platform. We also outline regulatory policy considerations to most effectively enable adoption of these recommendations.

For the graphical abstract of this paper see figure 1.

Figure 1. Graphical Abstract.

Background

When the first chimeric antigen receptor (CAR) T-cell therapy was approved by the Food and Drug Administration (FDA) in 2017,¹ long-term safety sequelae were unknown. Since then, more than 30,000 patients have been treated with CAR T-cell therapy, and 7 CAR T-cell therapies have been approved. The totality of data over 15 years since cancer CAR T clinical trials began supports a positive benefit/risk profile and, in some cases, the curative potential of CAR T-cell therapy. This positive CAR T-cell therapy experience to date, coupled with the current technology-enabled environment, requires a re-examination of the current long-term follow-up (LTFU) process and regulatory requirements to ensure they remain both scientifically justified and patient-centered. In this paper, we argue that the accumulated data support shortening and streamlining the LTFU requirements, and we provide recommendations on how to do so.

Current LTFU data collection requirements and process

CAR T-cell therapies, which currently use integrating viral vectors to introduce the CAR gene into T cells, have a theoretical risk of insertional oncogenesis. Due to a concern that patients treated with CAR T-cell therapy may develop second primary malignancies (SPMs), or new cancers that are independent of an original malignancy, the FDA to date has required 15 years of LTFU data collection for all participants receiving the CAR T-cell therapies in clinical trials and for a significant subset of patients receiving CAR T-cell products in the commercial setting. More recently, the FDA stated in November 2023 that patients and clinical trial participants receiving CAR T-cell therapy products should be monitored for new malignancies for life.² As further detailed below, the current LTFU data collection requirements and process are extensive and complex. The online supplemental materials (supplementary methods) detail how the following information on treatment site procedures and staffing was collected.

Clinical study administration

For participants receiving CAR T-cell therapy products in clinical trials, FDA guidance³ indicates the investigator should prepare and maintain adequate and accurate case histories on each participant, which should include information from scheduled visits. Physical examinations with a healthcare provider (HCP) are recommended for the first 5 years after infusion, and investigators are recommended to contact participants at a minimum of once a year for the full 15-year duration. Investigators are to report LTFU data to the study sponsor, who in turn reports data to the FDA (figure 2).

Figure 2. Typical process for collecting LTFU data from clinical trial participants. The process begins with research personnel obtaining informed consent and proceeds through investigator-led patient evaluations, data collection, and sign-off. The research personnel must then enter the data into electronic case report forms before sharing with the sponsor or designated vendor who reviews/monitors the data. Reports are submitted annually to the FDA to ensure regulatory compliance. *Data typically include AEs or serious AEs; health status information from a physical examination, answers to questionnaires, telephone or email communications; and information on targeted concomitant medications, subsequent therapy, survival status and, for pediatric studies, growth, development, and sexual maturity status. The persistence of modified/CAR T cells is monitored at least annually until vector sequences become undetectable. **Sponsors also provide reports to FDA when expedited reporting of qualified events is required or in response to requests for information. AEs, adverse events; CAR, chimeric antigen receptor; FDA, Food and Drug Administration; LTFU, long-term follow-up; SPMs, second primary malignancies.

Commercial CAR T-cell product administration

In the commercial setting, safety monitoring is fulfilled via post-marketing requirements (PMRs) that the FDA issues on a product-by-product basis. To date, all approved CAR T-cell therapies have PMRs that require study of significant numbers of patients for 15 years—up to 1,500 patients for initial product approvals and between 300 and 1,500 additional patients for new indications of a product (table 1). Products receiving approval for earlier lines of therapy have typically required either no additional PMRs or a small number of additional patients (ie, 200) to be studied for 15 years.

Table 1. FDA post-market requirements for FDA-approved CAR T-cell therapies.

Approval date	Product generic name (brand name)	Indication	Number of patients in post-approval safety study
12/4/25	Lisocabtagene maraleucel (Breyanzi)	Marginal zone lymphoma	PMR for new indication: 300
11/08/24	Obecabtagene autoleucel (Aucatzyl)	B-cell precursor acute lymphoblastic leukemia	500
5/30/24	Lisocabtagene maraleucel (Breyanzi)	Mantle cell lymphoma	PMR for new indication: 300
5/15/24	Lisocabtagene maraleucel (Breyanzi)	Follicular lymphoma	PMR for new indication: 300
4/05/24	Ciltacabtagene autoleucel (Carvykti)	Multiple myeloma, second line	PMR for earlier line of therapy: 200
4/04/24	Idecabtagene vicleucel (Abecma)	Multiple myeloma, third line	PMR for earlier line of therapy: 200
3/21/24	Lisocabtagene maraleucel (Breyanzi)	Small lymphocytic and chronic lymphocytic leukemia	PMR for new indication: 300
6/24/22	Lisocabtagene maraleucel (Breyanzi)	Large B-cell lymphoma, second line	PMR for earlier line of therapy: 200*
5/27/22	Tisagenlecleucel (Kymriah)	Follicular lymphoma	PMR for new indication: 300
4/01/22	Axicabtagene ciloleucel (Yescarta)	Large B-cell lymphoma, second line	PMR for earlier line of therapy: 0
2/28/22	Ciltacabtagene autoleucel (Carvykti)	Multiple myeloma, fifth line	1,500
10/01/21	Brexucabtagene autoleucel (Tecartus)	Adult acute lymphoblastic leukemia	PMR for new indication: 500
3/26/21	Idecabtagene vicleucel (Abecma)	Multiple myeloma, fifth line	1,500
3/05/21	Axicabtagene ciloleucel (Yescarta)	Follicular lymphoma	PMR for new indication: 300
2/5/21	Lisocabtagene maraleucel (Breyanzi)	Large B-cell lymphoma, third line	1,500
7/24/20	Brexucabtagene autoleucel (Tecartus)	Mantle cell lymphoma	500
5/01/18	Tisagenlecleucel (Kymriah)	Diffuse large B-cell lymphoma	PMR for new indication:1,500
10/18/17	Axicabtagene ciloleucel (Yescarta)	Large B-cell lymphoma, third line	1,500
8/30/17	Tisagenlecleucel (Kymriah)	B-cell precursor acute lymphoblastic leukemia, patients up to 25 years of age	1,000

Source: Regulatory review documents (Summary Basis for Regulatory Action documents, Approval Letters, and Clinical Review Memos) available on individual product pages linked from the FDA’s Approved Cellular and Gene Therapy Products main page.43

^*This value was provided by BMS and reflects information discussed between BMS and FDA during the review and incorporated into an updated registry protocol submitted to the IND after sBLA approval.

BMS, Bristol Myers Squibb; CAR, chimeric antigen receptor; FDA, Food and Drug Administration; IND, investigational new drug application; PMR, post-marketing requirement; sBLA, supplemental biologics license application.

For patients receiving commercial CAR T-cell therapy products within a post-approval safety study, the process for collecting LTFU data is similarly extensive and complex. Treatment sites typically report follow-up data to a patient registry, with the primary patient registry run by the Center for International Blood and Marrow Transplant Research (CIBMTR). Per the LTFU guidance document, the FDA may recommend use of a patient registry as the mechanism for data collection.³

Data managers at treatment sites must manually locate data in a patient’s electronic health record (EHR) and enter that data into a series of up to approximately 14 CIBMTR forms, each of which contains between 7 and 522 questions. While a few of the forms are specific to certain types of cancer or certain subsequent occurrences after infusion (eg, secondary malignancy, pregnancy), many of the forms need to be completed for all CAR T-cell therapy recipients, ranging from one time for some forms (eg, pre-infusion baseline data) to every follow-up visit (eg, post-cellular therapy follow-up at day 100, 6 months, 1 year and annually thereafter). Figure 2 provides an overview of the typical process for collecting post-market registry study data.

Figure 3. Typical process for collecting LTFU data for patients treated with commercial products. After patient recruitment to a PMR study, CIBMTR assigns a unique recipient ID (CRID) that is entered by the data manager to trigger required data forms. HCPs collect relevant data during routine care and enter it into the electronic health record from which data managers extract the necessary data points and enter them into CIBMTR forms. Sponsors periodically request this data from CIBMTR and submit annual reports to the FDA for up to 15 years post-treatment. The safety data reported to the FDA include the development of subsequent neoplasms, cause of death for patient deaths, and the incidence and severity of cytokine release syndrome, neurologic toxicities, serious infections, prolonged cytopenias, hypogammaglobulinemia, and pregnancy outcomes in females of childbearing potential. CAR, chimeric antigen receptor; CIBMTR, Center for International Blood and Marrow Transplant Research; FDA, Food and Drug Administration; HCP, healthcare provider; LTFU, long-term follow-up; PMR, post-market requirement; SPMs, second primary malignancies.

Extended safety surveillance

The 15-year LTFU and PMR requirements are designed to collect long-term safety and integration data for regulatory purposes. These obligations are distinct from clinical recommendations for extended patient monitoring, which reflect ongoing patient safety considerations rather than formal regulatory data collection requirements.

In addition to the 15-year LTFU and PMR studies, FDA stated in a November 2023 Safety Communication² that patients and clinical trial participants receiving B-cell maturation antigen-directed or CD19-directed autologous CAR T-cell therapy products should be monitored life-long for secondary malignancies, which is performed through standard medical care. The agency indicates providers should report the event to the manufacturer and obtain instructions for collecting patient samples to test for the presence of the CAR transgene. Providers should also report all suspected adverse events (AEs) to MedWatch, the FDA Safety Information and Adverse Event Reporting Program. Reports of these events are stored in the FDA Adverse Event Monitoring System database.

This Safety Communication² was prompted by rare cases of secondary T-cell malignancies reported in patients following CAR T-cell therapy. Although rare instances showed the presence of the CAR transgene, a causal role for CAR T cells in the development of T-cell malignancy has yet to be demonstrated. In addition, patients receiving these products for B-cell malignancies often have risk factors, such as prior treatments including chemotherapy, which can increase the background risk of T-cell malignancy. Importantly, in November 2024, Peter Marks, then Director of the Center for Biologics Evaluation and Research (CBER), noted publicly that this background incidence was not formally considered when the safety communication was issued.

Challenges of the current LTFU data collection requirements and process

The length and complexity of the current LTFU process place significant burdens on patients and providers.

Patient burden and loss to follow-up

A recent Emily Whitehead Foundation and Catalyst Healthcare Consulting study⁴ of nearly 100 CAR T-cell therapy recipients observed significant patient attrition over the follow-up period. Notably, 20% of CAR T-cell therapy recipients who received their infusion more than a year ago had stopped going to follow-up visits. Moreover, of respondents who were treated recently (within a year), more than one-third (38%) did not see themselves following up for 15 years, most of whom did not see themselves following up for more than 8 years.

The top challenges identified in attending follow-up visits per the survey were travel-related, including distance to the treatment site and travel costs. Over half of respondents (53%) indicated that they lived more than 2 hours away from their original treatment site, with nearly one-third (31%) living more than 6 hours away.

Provider/investigator burden: duration and cost

In both trial and commercial settings, the LTFU process involves multiple steps and is resource intensive for providers and investigators.

In investigator-sponsored trials, the investigator/sponsor typically does not receive research funding to cover the costs of LTFU data collection and needs to identify alternative funding sources. Academic medical centers must hire additional staff to perform the manual collection and entry of long-term safety data. In the commercial setting, treatment centers often require several data managers/coordinators for CAR T-cell therapy data collection and entry. For example, providers interviewed for this paper employ up to 6 full-time equivalent data managers/coordinators.

These staffing costs far exceed the amount of compensation the providers receive from the registry. Furthermore, when patients choose to receive follow-up care with a community oncologist near home, academic medical centers may lack the resources to retrieve data from community EHRs and report it to registries.

In recent years, there has been growing momentum to administer CAR T-cell therapy in community settings, with the goal of increasing patient access to what may be a life-saving treatment.⁵ Community-based oncologists are unlikely to have the financial resources to obtain data managers, nor do they have the time needed to complete the extensive forms in addition to providing patient care. Thus, there is heightened urgency to promote streamlined approaches to data collection as CAR T moves into the community setting.

Solutions to address challenges

Reassess duration of follow-up based on evidence

The FDA’s guidance on LTFU after administration of gene therapy products³ states that the objective of LTFU is “to identify and mitigate the long-term risks to the patients receiving the [gene therapy] product.” In alignment with this goal, our analysis considers risk-based factors when evaluating the appropriate duration of LTFU for integrating vectors. Accordingly, our assessment focuses on AEs that are most clinically and mechanistically relevant. Specifically, we assess whether the theoretical risk of insertional mutagenesis has materialized by assessing the frequency of SPMs following CAR T-cell therapy relative to standard-of-care (SOC) treatments, and whether a causal relationship to the product has been demonstrated.

Further elucidated below, our analyses show that AEs in general tend to be reported within the first 3 years and rarely after 5 years; the vast majority of SPMs have occurred within the first 5 years6,8; and T-cell malignancies, in particular, have been reported within the first 2 years after infusion.^{9 10} Moreover, no direct causal association via insertional oncogenesis between CAR T-cell therapy and SPMs, including T-cell malignancies, has been demonstrated.11,14

Aggregated AE reporting for five CAR T-cell therapy products

To begin to assess the timing of AEs after CAR T-cell therapy administration, we collected pivotal clinical trial reporting data from developers of 5 FDA-approved CAR Ts. Long-term safety reporting is more robust within the context of a clinical trial, wherein follow-up periods and assessments are mandated for sponsors and AEs are solicited, than AE data collected within the post-marketing setting. Therefore, for this paper, developers of these 5 FDA-approved CAR Ts have interrogated their global safety databases to assess the reporting ratios of AEs reported within their pivotal clinical trials and LTFU protocols (figure 4). The data were anonymized and aggregated to protect product-level confidentiality. This assessment is an event-level (not patient-level) evaluation that describes the proportion of AEs reported each year after infusion relative to the total number of events for each product. More detailed methods are available in the Supplementary Materials (online supplemental methods).

Figure 4. Range of percentages of total AEs reported in each year after infusion across five CAR T products. Annual low-end (blue line) and high-end (red line) ranges of percentages of AEs reported for the first 7 years post-infusion, derived from global safety data from the pivotal clinical trials for 5 FDA-approved CAR T-cell therapy products. The shaded area highlights the full range of AEs reported per year. The data shows a significant drop in AEs after the first year post-infusion and event reporting reaching or approaching zero after year 5. AEs, adverse events; CAR, chimeric antigen receptor; FDA, Food and Drug Administration; SPMs, second primary malignancies.

With median follow-up times of 24–64 months, the majority (53%–88.6%) of AEs were reported in the first year after infusion (figure 4). A drop in the percentage of the total reported AEs was observed after the first year—5.5%–18% in year 2; 3.1%–24% in year 3; 0.6%–6.5% in year 4; and 0%–8.5% in year 5. After year 5, reporting of AEs declined sharply, reaching 0%–2% in year 6 and 0%–0.2% in year 7. Therefore, AEs tend to be reported within the first 3 years and rarely after 5 years.

The few AEs that continued to be reported beyond year 5 were predominantly infections and SPMs. These events are consistent with expectations for an immunosuppressed, pretreated population receiving CAR T-cell therapy. Infections are common in these underlying disease settings, and both infections and SPMs are recognized risks associated with other SOC treatments used in these indications. Consequently, the background incidence of SPMs following SOC therapy should also be considered, as discussed further below.

It is important to note that these data may be limited in interpretability due to potential differences in data collection and reporting across global safety databases among sponsors, notable differences in duration of follow-up, patient attrition over time (lost to follow-up, death, consent withdrawn), and the evolving nature of global safety database reporting.

Literature synthesis

Since chemotherapy and/or radiation therapy can cause SPMs, we reviewed the literature for SPM development after CAR T-cell therapy relative to SPM development after SOC therapy. Across reports that had up to 6 years of follow-up, the frequency of SPMs after CAR T-cell therapy (0.4%–6.2%)613,22 was similar to the frequency of SPMs after receiving SOC treatments (4%–6.7%)15,1723 (table 2). When median follow-up was not reported, mean follow-up was provided and denoted accordingly in table 2.

Table 2. Frequency of SPMs after CAR T versus standard of care^*.

Study	After CAR-T			After SOC for blood cancers
	Frequency of SPMs	Number of patients infused	Median follow-up time (months)	Frequency of SPMs	Number of patients treated	Median follow-up time (months)
Rodriguez-Otero, et al, 202318	6%	225	18.6	4%	126	18.6
San-Miguel, et al, 202317	5.1%	176	15.9	6.7%	208	15.9
Abramson et al, 202316	4%†	147	17.5
Jadlowsky et al, 20257	2.3%	783	18.7
John et al, 202522	0.4%‡	703	28
Ghilardi et al, 202419	3.6%	449	10.3
Hamilton et al, 202414	3.5%§	724	15
Thieblemont et al, 202421	6.2%	97	53
Tix et al25	6%	5517	21.7
Pasquini et al, 202020	0.8%‡	255¶	13.4
Laetsch et al, 202323	1.3%	79	38.8
Miret et al, 202326				6%	7807	21.2††
Sahebi et al, 201824				4.2%	3204**	58.6‡‡

^*The majority of studies reported frequencies rather than incidence rates, with the exceptions noted below. Median follow-up times should be considered when interpreting these results, and studies cannot be directly compared given heterogeneities across studies, including in standard of care comparators and follow-up durations. SPM frequency was calculated using infused/treated patients (safety population) as the denominator.

^†SPMs are reported for patients after CAR T infusion, including both patients initially randomized to CAR T and SOC patients who crossed over.

^‡Excludes ALL to AML lineage switch.

^§Excludes non-melanoma skin cancers.

^¶Children and young adults with ALL.

^**After autologous hematopoietic stem cell therapy.

^††Mean follow-up; median follow-up not reported.

^‡‡Study additionally reports a 72-month cumulative incidence of 5.3%.

ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; CAR, chimeric antigen receptor; SOC, standard of care; SPMs, second primary malignancies.

To complement these frequency reports, we provide cumulative incidence estimates from the literature at a consistent follow-up duration of 5 years. Because these studies use different reporting terminology, we distinguish between SPMs—defined as histologically distinct, independent primary cancers—and subsequent malignant neoplasms (SMNs), a broader term often used to capture a wider range of post-treatment events, including therapy-related myeloid neoplasms. In a mixed adult and pediatric cohort, the 5-year cumulative incidence of SPMs after CAR T-cell infusion was 2.8% (95% CI 1.7% to 4.5%).⁶ A separate study of pediatric and young adult patients with B-cell acute lymphoblastic leukemia (B-ALL) reported a 5-year cumulative SMN incidence of 1.5% (95% CI 0.4% to 4%),⁷ (which included cases that may fall outside strict “independence” criteria for an SPM). For context, the 6-year cumulative incidence of SPMs after autologous transplantation for multiple myeloma in adults has been reported at 5.3% (95% CI 4.4% to 6.3%).²⁴

Although the overall frequency is low, hematological malignancies (ie, myelodysplastic syndrome, acute myeloid leukemia, T-cell lymphoma) have been reported to occur at a slightly higher frequency after CAR T-cell therapy (0.8%–2.2%)1316,20 23 than after SOC treatments (0%–0.9%)^{16 17 25 26} (table 3).

Table 3. Frequency of hematologic malignancies after CAR T versus standard of care^*.

Study	After CAR-T			After SOC
	Frequency of hematological SPMs	Number of patients infused	Median follow-up time (months)	Frequency of hematological SPMs	Number of patients treated	Median follow-up time (months)
Rodriguez-Otero et al17	1%	225	18.6	0%	126	18.6
San-Miguel et al16	1.7%	176	15.9	0%	208	15.9
Ghilardi et al18	1.1%	449	10.3
Hamilton et al13	1.9%†	724	15‡
Thieblemont et al20	2.1%	97	53
Tix et al25	2.2%	5517	21.7
Pasquini et al19	0.8%§	255¶	13.4
Laetsch et al22	1.3%	79	38.8
Joelsson et al, 202226				0.7%	32,100	85
Sahebi et al, 201824				0.9	3,204**	58.6††

^*The majority of studies reported frequencies rather than incidence rates, with the exceptions noted below. Median follow-up times should be considered when interpreting these results, and studies cannot be directly compared given heterogeneities across studies, including in standard of care comparators and follow-up durations. SPM frequency was calculated using infused/treated patients (safety population) as the denominator.

^†Excludes non-melanoma skin cancers.

^‡ Study additionally reports a cumulative incidence of 6.5% at 3 years post-infusion.

^§Excludes ALL to AML lineage switch.

^¶Children and young adults with ALL.

^**After autologous hematopoietic stem cell therapy.

^††Study additionally reports a cumulative incidence of 1.4% at 72 months post-infusion.

ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; CAR, chimeric antigen receptor; SOC, standard of care; SPMs, second primary malignancies.

The higher frequency of hematological malignancies observed in patients treated with CAR T-cell therapy relative to SOC treatments is likely attributable to patient factors rather than the CAR T-cell therapy itself.^{27 28} The relapsed/refractory population eligible for CAR T-cell therapy is typically older and has received at least one, and often multiple, prior lines of systemic therapy. For example, patients with primary mediastinal B-cell lymphoma commonly receive intensive frontline therapy, such as four to six cycles of R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone), sometimes followed by mediastinal radiation, or six cycles of DA-EPOCH-R (dose-adjusted etoposide, prednisone, vincristine, cyclophosphamide, doxorubicin, and rituximab), which generally does not require radiation.²⁹ In a retrospective analysis of 511 patients with non-Hodgkin’s lymphoma who were treated with CAR T-cell therapy between September 2017 and August 2023, Gazeau and colleagues reported that patient age and number of prior therapies were statistically significant risk factors for the development of treatment-related myeloid neoplasms (eg, acute myeloid leukemia and myelodysplastic syndromes).²⁷ In another recent report, Farina and colleagues reported that most patients who developed secondary myeloid malignancies following CAR T-cell therapy exhibited high-risk mutations (del7 and TP53).²⁸ Genetic alterations were seen in some patients prior to therapy, suggesting an additional adverse prognostic factor for SPM development.¹²

Due to the mechanism of action of CAR T-cell therapy, which involves integration of the CAR gene into the T cell’s genome, T-cell malignancies are the SPM of primary concern. Significantly, FDA leaders concluded last year that T-cell malignancies after CAR T-cell therapy have been rare.⁹ The FDA reported that as of the end of 2023, it had become aware of 22 cases of T-cell malignancies after treatment with 5 of the 6 therapies approved at the time, but that the small sample size and variation in CAR T-cell product precluded conclusions about a possible association. Importantly, FDA leaders stated, “With more than 27 000 doses of the six approved products having been administered in the United States, the overall rate of T-cell cancers among people receiving CAR T-cell therapies appears to be quite low, even if all reported cases are assumed to be related to treatment.”⁹

The studies that have reported rare cases of T-cell malignancies after CAR T-cell therapy¹⁴ have frequencies well below 1% (0.09%–0.5%).^{6 13 16 18 23} Several other studies have reported no cases of second T-cell malignancy.20,2230

In the FDA report, 3 cases of T-cell malignancies for which genetic sequencing had been performed showed the presence of the CAR transgene.⁹ However, the presence of neoplastic CAR-positive T cells does not itself prove that the CAR is responsible for the malignancy,¹¹ and a causal role for CAR T cells in the development of T-cell malignancy has yet to be demonstrated.^{12 13}

When new malignancies occur after CAR T-cell therapy, they tend to occur relatively early. A 2025 study⁶ of 783 participants receiving T-cell therapy in clinical trials for cancer or HIV-1 infection found that the vast majority of new malignancies occurred within 5 years of infusion, except a single case of papillary thyroid cancer that occurred at year 14 in a subject with a history of radiation therapy to the neck. Of note, 22% (172) of the participants in this study had a follow-up duration of more than 5 years (5–15+years).

In another study of 420 patients receiving CD19 CAR T-cell therapy, the median time to diagnosis of a second malignant neoplasm was 3.2 years (range, 0.6–8.2 years), leading to a 5-year cumulative incidence of subsequent malignant neoplasms of 1.5%.⁷ According to CIBMTR data on 11,345 recipients of commercial CAR T as of February 2024, of whom 8,060 are enrolled in post-authorization safety studies, the median time from CAR T-cell infusion to the first subsequent neoplasm was 9 months.⁸ In the FDA report, the 14 cases of T-cell malignancies with adequate data occurred within 2 years (range, 1–19 months) of CAR T administration.⁹

The theoretical risk for SPMs was the initial catalyst for lengthy follow-up of products with integrating viral vectors. Since a causal role for CAR T cells has yet to be demonstrated, and the vast majority of SPMs have occurred within the first 5 years of CAR T-cell infusion, we are recommending a decrease in the duration of LTFU data collection to 5 years. These clinical findings align with FDA’s historical willingness to adjust LTFU requirements based on evolving evidence, providing a regulatory precedent for a risk-based approach.

Regulatory precedent for a risk-based approach

FDA has previously revised LTFU expectations in response to emerging safety data. In the 2006 FDA guidance document, Gene Therapy Clinical Trials—Observing Subjects for Delayed Adverse Events,³¹ the agency recommended a minimum of 15 years of LTFU for adeno-associated viral (AAV) vectors, citing the potential for delayed immune responses for AAV vectors exhibiting persistent transgene expression. However, subsequent clinical experience has demonstrated a generally favorable safety and tolerability profile for AAV-based therapies.³² Reflecting these data, the 2020 final guidance Long Term Follow-Up After Administration of Human Gene Therapy Products concluded that LTFU observations from trials conducted since 2006 support a lower risk profile for certain gene-delivery modalities, including AAV vectors, and therefore reduced the recommended LTFU duration for AAV vectors to up to 5 years.³ Both guidance documents underscore that evolving vector modifications and accumulating clinical data may justify sponsor reassessment of LTFU recommendations. Together, these actions demonstrate FDA’s long-standing willingness to adapt LTFU recommendations for specific modality types in response to emerging evidence and a risk-based approach.

Modernizing the process: technological tools and standards for data collection

In addition to reductions in the duration requirements of LTFU, the process for post-market data collection could be streamlined and updated. A promising long-term solution would be to enable deidentified real-world data (RWD) from EHRs collected by any treating HCP to be directly accessible to, and usable by, an FDA-partnered third-party database. Such an approach would remove several steps from the current reporting process and improve data capture when patients receive follow-up care with their local HCPs. Although establishing a centralized FDA-partnered database would require sustained multistakeholder collaboration, the growing use of real-world evidence (RWE) by sponsors, along with ongoing advances in EHR interoperability and data standardization, provides a foundation that could be expanded to support this goal.

First, sponsors are beginning to use EHR-derived RWD on a voluntary, product-specific basis. According to the fiscal year 2024 annual report, there were 10 regulatory submissions to CBER containing EHR data as a source of RWE, compared with none in fiscal year 2023, when FDA first began tracking RWE submissions.³³ Of the 22 submissions containing RWE in fiscal year 2024 (FY2024), 17 used RWE to satisfy PMRs or post-marketing commitments.

Second, significant progress is being made toward addressing implementation limitations, including variations in how data are collected and recorded across EHR vendors, inconsistent labeling of data elements, and privacy concerns. Interoperability, or the ability of different EHR systems to communicate and exchange patient information, is central to modernizing the LTFU process. Recent technological and data standardization efforts have improved EHR interoperability, and these advances are being translated into practice. For example, Epic Systems Corporation, the largest EHR vendor,³⁴ has a longstanding health information exchange platform, Care Everywhere, that has enabled EHR interoperability since 2008.

More recently, in 2022, the Assistant Secretary for Technology Policy’s Office of the National Coordinator for Health Information Technology completed the establishment of the Trusted Exchange Framework and Common Agreement (TEFCA), which is a nationwide framework for health information sharing. As of 2023, more than 60% of hospitals reported plans to participate in TEFCA.³⁵ Recent standardization of clinical study data from RWD sources has also furthered interoperability (ie, the Fast Healthcare Interoperability Resources technical standard³⁶ and the US Core Data for Interoperability standardized data elements).³⁷ These and other standardization efforts are also helping to reduce gaps in data capture and ensure that critical information is consistently recorded, including for CAR T-cell therapy.³⁸

An important aspect of the proposed process is direct data exchange between databases supporting the EHR and an FDA-partnered third-party database, which would greatly decrease burden for providers in finding and entering data into a registry, and for sponsors in reporting to the FDA. Third-party platforms have started to leverage EHR interoperability to enable direct data exchange for other treatment modalities. For example, the Data Transformation Initiative of CIBMTR is able to provide direct data exchange from a transplant center to the registry for certain fields following hematopoietic cell transplantation.³⁸ Using the CIBMTR Reporting App, available in the Epic App Market, data managers can push a button to transmit data directly from the EHR to CIBMTR, where the data is ingested into the CIBMTR Outcomes Database.

The most pertinent example of this capability is FDA’s Biologics Effectiveness and Safety (BEST) Initiative. In 2017, the FDA CBER launched the BEST Initiative to enhance post-market AE reporting and serve as an active surveillance program for biologics. Through its Innovative Methods (IM) Initiative, the BEST team developed a prototype of the exchange platform which uses automated detection of potential AEs from EHRs, using highly sensitive data elements, followed by semi-automated validation and reporting.³⁹ To pilot the platform, the BEST IM team leveraged the non-profit health information exchange network, eHealth Exchange, to query and receive data from the network partners. In 2023, CBER presented the results of one of its pilots, in which the platform was used to retrieve clinical data for post-vaccination AEs from 11 health provider partners using Epic EHRs.⁴⁰ The results indicated the overall data quality met general requirements for regulatory grade data quality, thereby successfully validating that an exchange platform is a feasible means to automate the exchange of AE data.

Consistent with the Health Insurance Portability and Accountability Act (HIPAA) Privacy Rule, under BEST, data providers retain control over their data, which remain behind data partners’ local firewalls, and study results are returned to eHealth Exchange via a web portal in an aggregated format with all identifiers removed.⁴¹ When individual-level information is required, all individual identifiers such as names, addresses, phone numbers, and other identifying data elements are removed before information is shared with the database.

The FDA indicates that artificial intelligence (AI) may also play a role in the identification, evaluation, and processing for reporting post-marketing adverse experience information.⁴² Separate from the post-vaccination AE pilot, the BEST Initiative has used AI in other previous studies and continues to explore its use in current pilot studies.

We recommend in the long term that a third-party platform, like the one used by the BEST pilot, be used to transfer LTFU data points to a centralized FDA-partnered third-party database. A step toward this goal would be for CBER’s BEST Initiative to prioritize a feasibility assessment of the use of an automated data exchange platform for CAR T-cell therapy specifically, or more broadly for a related gene therapy product class. Such a study could inform the operationalization of a platform to streamline the collection of post-market safety data for these products. In addition, we recommend that LTFU data collection be refined to focus on AEs that have more theoretical potential to be related to CAR T-cell therapy by using more precise AE definitions.

Recommendations

In summary, we are proposing the following recommendations:

1. Shorten the LTFU requirement for CAR T-cell clinical trial studies and post-approval safety studies for marketed CAR T-cell products to 5 years.

For clinical trial participants, we recommend the FDA revise its guidance in the document, Long Term Follow-Up After Administration of Human Gene Therapy Products, to state that in general, the recommended duration of an LTFU protocol is 5 years for delayed AEs for CAR T-cell therapies using integrating vectors. We also propose incorporating this 5-year LTFU recommendation into the draft guidance document issued on September 24, 2025 on its finalization—Methods and Approaches for Capturing Post-Approval Safety and Efficacy Data on Cell and Gene Therapy Products.

Similarly, for post-approval safety studies for marketed CAR T-cell products (if required and as determined on an individual product basis), we recommend a 5-year duration for monitoring of all delayed events, including SPMs. In addition, we encourage the FDA to consider decreasing the number of patients to be studied for PMRs because post-market product data can be supplemented by data from clinical trials. Because current PMRs for CAR T-cell therapies specify 15 years of follow-up, a future change in guidance recommending shorter LTFU in clinical trials could inform the duration of PMRs for newly approved products or indications, subject to FDA agreement. Continued passive monitoring through standard HCP spontaneous voluntary reporting of AEs to MedWatch could be recommended for both clinical trial participants and patients receiving a product in the commercial setting.

• 2. 1Streamline the AE data collection process by enabling the automated exchange of EHR data to a central third-party database, such as CIBMTR or eHealth Exchange.

In the long term, we recommend the development of a platform, similar to that used in the BEST Initiative, to enable the automated secure transfer of patient and clinical trial participant data between HCPs and regulators through a central third-party database run by an FDA partner. As an initial step toward this goal, we suggest that CBER’s BEST team continue its efforts with a feasibility assessment of the use of its automated data exchange platform for CAR T-cell therapy or a related gene therapy class.

Due to the challenges of the current manual process of data collection and entry into registries, we also propose that the FDA allow sponsors the flexibility to select their data collection mechanism as another interim step toward this goal. We recommend the FDA update the LTFU guidance document to delete the statement that the FDA “may recommend that [a sponsor] establish a registry or use an existing patient registry…” for data collection.³ We encourage the FDA to give sponsors the choice of data collection mechanism, such as the use of a third-party database and/or new technologies (like AI), to enable the development and use of less resource-intensive methods. This recommendation would result in a streamlined process for data collection of post-market safety data to consist of the following steps:

• Sites obtain patient consent.

• An HCP assesses the patient.

• HCP reports the results in the EHR.

• Automated data transfer from the EHR to a central third-party database occurs.

• The sponsor requests and receives data from the third-party database to complete and submit FDA reports for 5 years

• 3. 1Streamline AE data collection requirements.

Currently, the long-term toxicities that need to be reported are within the following general categories: SPMs, new or exacerbated neurologic disorders and autoimmune disorders, new hematologic disorders, infections and any AEs or conditions that may have a reasonable relationship with CAR T-cell therapy. The use of more precise AE definitions by the entity collecting data would better capture relevant safety information, while decreasing the reporting burden on stakeholders, by collecting only data on more precisely defined AEs. We propose definitions for the scope of AE collection for each AE category in table 4. We recommend that data collection entities, in collaboration with other stakeholders, narrow the number of questions on data forms to reflect a more focused, relevant scope of AE data collection.

Table 4. Proposed scope of AE data collection.

AE reporting categories	Proposed scope of AE data collection
Second primary malignancies	Newly diagnosed malignancy only; recurrence or progression of an existing malignancy would not qualify
New or exacerbated neurologic disorders	New incidence or exacerbation of a pre-existing serious neurologic disorder and any condition requiring neurological consult and examination
New or exacerbated autoimmune disorders	New incidence or exacerbation of a pre-existing autoimmune disorder
New hematologic disorders	New incidence of serious hematologic disorder, including hypogammaglobulinemia, B-cell aplasia, and prolonged cytopenia
Infections and other AEs or conditions that may have a reasonable relationship to therapy	For patients with disease progression and initiation of subsequent treatment, only serious infections considered possibly related by the treating physician need to be reported

AE, adverse event.

Conclusion

In the 15 years since the first-in-class clinical trials began and with over 30,000 patients treated, CAR T-cell therapy has proven to be an effective, life-saving treatment with curative potential and an established safety profile. The current 15-year LTFU requirement is not justified in light of the long-term safety data summarized herein. Optimizing the process and reducing the duration of LTFU requirements would retain appropriate safety monitoring while minimizing unnecessary burden for patients, clinicians, sponsors, and regulators.