Therapies and outcomes following extramedullary relapse of pediatric and young adult ALL after CD19 CAR T-cell therapy

TO THE EDITOR:

CD19-targeted chimeric antigen receptor (CAR) T-cell therapy (CAR19) produces remarkably high remission rates in relapsed/refractory pediatric B-cell acute lymphoblastic leukemia (B-ALL); however, approximately 50% of children and young adults experience another relapse within 2 years after infusion.^1-4 Although the majority of post-CAR19 relapses are isolated to the bone marrow (BM), extramedullary disease (EMD) is not uncommon. Up to 34% of post-CAR19 relapses include EMD, either in isolation or in combination with BM disease.^5,6 Outcomes after post-CAR19 relapse are poor overall,^6,7 but data regarding outcomes after EMD relapse are limited. To address this gap, we sought to describe the incidence, patterns, salvage strategies, and long-term outcomes of post-CAR19 EMD relapses in patients treated at our center.

We conducted a retrospective review of patients aged <30 years treated on 5 CAR19 clinical trials or with commercial tisagenlecleucel (Kymriah, Novartis) for B-ALL at Children’s Hospital of Philadelphia (CHOP) between 2012 and 2022. Patients treated on clinical trials received either the murine CD19/4-1BB CAR construct, CTL019 (US Food and Drug Administration approved as tisagenlecleucel; ClinicalTrials.gov identifiers: NCT01626495, NCT02906371, NCT02228096, and NCT02435849), or the humanized CD19/4-1BB CAR construct, humanized CD19-targeted CAR T-cell therapy (NCT02374333).^4,8-10 The EMD relapse cohort included patients who experienced a relapse with central nervous system (CNS) and/or non-CNS EMD involvement after CAR19 infusion but before consolidative hematopoietic cell transplant (HCT) or other antileukemia therapy. CNS relapse was defined as cerebrospinal fluid (CSF) with ≥5 white blood cells per μL and positive for blasts or parenchymal or cranial nerve involvement. Non-CNS EMD was defined as focal bony, soft tissue, or mass lesions diagnosed by biopsy or characteristic imaging findings. To note, before CAR19 infusion, all patients had CSF sampling, but only those with a history of or clinical suspicion for EMD underwent imaging. Data were abstracted from clinical trial databases and electronic health records using REDCap electronic data capture tools hosted at CHOP.¹¹ This retrospective study was reviewed and considered exempt by the institutional review board of CHOP. The primary objective was to determine the overall survival (OS) rate from post-CAR19 EMD relapse. Secondary objectives were to evaluate the incidence of EMD relapse and to describe salvage regimens and their associated outcomes. OS was calculated using Kaplan-Meier methods from the day of post-CAR19 relapse identification to the time of death or censored at the last follow-up, with a data cutoff of 1 September 2023. Statistical analysis was performed in GraphPad Prism 10.0.0 (Boston, MA) and Stata, version 14.0 (StataCorp, College Station, TX).

Of 300 patients treated with CAR19 across 5 clinical trials or with commercial tisagenlecleucel, 267 achieved a complete remission (Figure 1A). With a median follow-up of 66 months after infusion, 105 (39%) experienced a subsequent relapse, of which 24 (23%) had EMD involvement as established by tissue biopsy, CSF sampling, or, in some cases, imaging alone (Table 1; supplemental Table 1, supplemental Appendix). In the EMD relapse cohort, the median age was 10.2 years (range, 0.7-26.5) at infusion and 12.1 years (range, 0.8-28.1) at post-CAR19 relapse. The median time from CAR19 infusion to relapse was 18.2 months (range, 1.4-59.4). Prior immunotherapy exposures included HCT (n = 16 [67%]), blinatumomab (n = 2 [8%]), inotuzumab (n = 2 [8%]), and other CAR products (n = 2 [8%]).

Sites of post-CAR19 relapse were categorized as isolated CNS (n = 7 [29%]), combined BM and CNS (n = 4 [17%]), isolated non-CNS EMD (n = 10 [42%]), and combined BM and non-CNS EMD (n = 3 [12%]). Non-CNS EMD sites included breast (n = 4), peri-orbital/orbital (n = 3), scalp/forehead (n = 2), mediastinum (n = 1), kidney (n = 1), maxillary bones and sinus (n = 1), and diffuse bones/liver (n = 1). Sixteen of 24 patients (67%) with EMD relapses had EMD involvement before CAR19 infusion. Of those 16, the post-CAR19 EMD relapse site was the same as the preinfusion site for 9 patients (56%). In contrast, in 15 of 24 (63%), EMD relapses occurred in new sites. Most relapses were CD19⁺ (n = 18 [75%]); 3 (13%) were CD19^–; 1 (4%) had both CD19⁺ and CD19^– leukemic cells, and 2 (8%) were unknown. At the time of post-CAR19 relapse, 18 patients (75%) had functional CAR T-cell persistence as evidenced by ongoing B-cell aplasia, including 12 of 18 with CD19⁺ relapses. Five patients (21%) had B-cell recovery, all of whom experienced CD19⁺ relapses. B-cell aplasia status was unknown in 2 patients (8%).

With a median follow-up of 47 months from post-CAR relapse, the median OS was 11.5 months (95% confidence interval [CI], 6-56) (Figure 1B). OS rates at 2 and 3 years after relapse were 44% (95% CI, 23-62) and 38% (95% CI, 19-58), respectively. OS was similar for CNS and non-CNS relapses (median OS, 11.5 months vs 13.0 months; P = .49). After relapse, 21 patients (88%) received salvage treatment with curative intent (1 received palliation only, and 2 were unknown; Table 1). A variety of treatment approaches were used. For patients with CNS involvement (n = 10), first-line therapies included intrathecal chemotherapy alone (n = 1) or in combination with radiation therapy (n = 2), systemic chemotherapy (n = 4), inotuzumab (n = 1), or additional CAR19 infusions (n = 2). Ultimately, 5 patients were successfully bridged to HCT; 2 are in ongoing remissions, and 3 experienced another relapse, of whom 2 died of progressive disease and 1 is disease free again after multiple salvage therapies. For patients with non-CNS EMD involvement (n = 11), first-line therapies included focal radiation (n = 3), CAR19 reinfusion (n = 2), radiation with CAR19 reinfusion (n = 1), re-treatment with a new CAR19 product (n = 1), CAR22 (n = 1), systemic chemotherapy (n = 2), or inotuzumab (n = 1). Ultimately, 9 patients were bridged to definitive therapy (HCT, n = 5; additional CAR treatments, n = 4); 3 are in ongoing remission, 3 died of nonrelapse mortality, and 3 experienced another relapse and died of progressive disease.

Our report demonstrates that almost 1 in 4 post-CAR19 relapses involve EMD sites. More than 70% of these are isolated EMD relapses. Although diverse salvage approaches are used, overall outcomes remain poor with a median survival of 11.5 months from relapse. EMD relapse locations vary, but more than half are outside the CNS. EMD relapses are also frequently reported after other immunotherapies, which may apply selective pressure that drives leukemic recurrence in extramedullary sites.^12-14 

A growing body of literature reports on the impact of pre–CAR19 infusion EMD^14-18; however, minimal data about post-CAR19 extramedullary failure are available. These studies on pre-CAR19 EMD demonstrate that CAR19 can clear EMD, but non-CNS EMD is more challenging to eradicate and frequently recurs. Indeed, in our study, two-thirds of patients with post-CAR19 EMD relapses had a history of EMD before CAR19 treatment. Still, one-third had no known history of EMD. Among patients with preinfusion EMD, postinfusion relapse occurred in different locations in 44%. Although it is possible not all preinfusion EMD was detected, it is likely that extramedullary relapse occurs in the absence of prior disease. To note, this cohort only included patients treated with 4-1BB–based CAR19 constructs. It is unknown whether the efficacy of CD28-based constructs demonstrated in lymphoma could translate to fewer B-ALL EMD events after CAR19.

The mechanisms for post-CAR19 EMD relapses are not fully elucidated, but the patterns observed suggest challenges with penetration of the tumor microenvironment. Three-quarters of the EMD relapses were CD19⁺, which are typically associated with limited CAR T-cell persistence.^10,19,20 Yet, 12 of 18 patients with CD19⁺ relapses had functional persistence at relapse, which highlights that functional CAR T cells may not adequately surveil all extramedullary sites.

After post-CAR19 EMD relapse, we report poor outcomes. The median OS was <1 year from relapse, which is inferior to cohorts consisting of patients with primarily medullary post-CAR19 CD19⁺ relapses.^6,7 We found a heterogeneous approach to salvage therapies with varying success rates. Many patients received additional CAR treatments, either reinfusions of their original product or re-treatment with a novel product. Some achieved durable remissions, highlighting that extramedullary failure does not preclude further CAR treatment. Overall, 14 of 24 patients were successfully bridged to definitive therapy, but only 5 remain in long-term remissions.

In summary, almost 25% of post-CAR19 relapses occur in extramedullary sites. Most patients have evidence of functional CAR T-cell persistence at relapse, yet they still relapse with CD19⁺ EMD, highlighting the tumor microenvironmental barriers. The inferior survival outcomes emphasize the importance of optimizing salvage therapies for this unique, but not uncommon, population.

Acknowledgments: This work was supported by the Children’s Hospital of Philadelphia Frontier Program, National Cancer Institute 5-K08-CA-277013 (R.M.M.), and an American Society of Hematology Scholar Award (R.M.M.).

We dedicate this article to Barbara Friedes, our colleague, mentee, and friend, who tragically passed away shortly after finishing this manuscript. The impact she had on the field of pediatric oncology in her career, sadly cut too short, is tremendous. This important, foundational work that will contribute to improvement of outcomes for children with refractory leukemias is just 1 piece of her professional legacy.

Contribution: B.D.F., A.M.D., and R.M.M. conceptualized, designed, and planned the study; B.D.F., E.I., and R.M.M. collected data; B.D.F. and R.M.M. performed statistical analysis; and all authors reviewed the analyses, contributed to interpretation of results and writing of the manuscript, and approved the final version of the submitted report.

Conflict-of-interest disclosure: S.A.G. receives research funding from Cellectis, Kite (a Gilead company), Novartis, Servier, and Vertex; consults for Adaptive Biotech, Eureka Therapeutics, Estrella Immunopharma, Jazz Pharmaceuticals, and Novartis; has advised for Allogene, Cabaletta Bio, Jazz Pharmaceuticals, Verismo, Novartis, and Vertex; and has patents managed according to the University of Pennsylvania and the Children's Hospital of Philadelphia policies. S.R.R. has received consulting fees from AbbVie and Pfizer. S.L.M. has received clinical trial support from Novartis and Wugen; has served on advisory and/or study steering committees for Novartis, Wugen, and Syndax; and has a patent pending and licensed to Novartis Pharmaceuticals without royalty for PCT/US2017/044425: Combination Therapies of CAR and PD-1 Inhibitors. The remaining authors declare no competing financial interests.

Barbara D. Friedes died on 17 July 2024.

Correspondence: Regina M. Myers, Children’s Hospital of Philadelphia, Hub for Clinical Collaboration, Office 3545, 3500 Civic Center Blvd, Philadelphia, PA 19104; email: myersrm@chop.edu.