CAR-T Cell Therapy: Engineering Remission in Refractory Myeloma

Abstract

Chimeric antigen receptor (CAR) T cells have rapidly reshaped the treatment of relapsed and refractory multiple myeloma (MM), progressing from an experimental salvage option to a therapy with a demonstrated overall survival advantage and durable, potentially curative remissions in a subset of patients. This review surveys the current evidence supporting their use, the biology underlying their successes and failures, established and emerging toxicities, and novel approaches that will define the field in the coming years. Approved products directed against B cell maturation antigen (BCMA) induce rapid and deep responses even in heavily pretreated patients, and long-term follow-up now suggests a meaningful fraction of patients achieve prolonged remission after a single infusion without maintenance therapy. These benefits are accompanied by distinctive toxicities that shape the risk-benefit calculus as therapy moves to earlier lines and even to treatment naïve patients. We discuss the determinants of response and relapse, as well as the mechanisms of antigen-positive and antigen-negative escape that next-generation approaches aim to overcome. These include novel antigen targets, multi-antigen and armored constructs, rapid manufacturing, and in vivo CAR generation, all with the goal of extending durable remissions to every patient.

Harnessing the immune system has long been a core part of the treatment of multiple myeloma (MM),¹ beginning with immune stimulants such as interferon² and allogeneic hematopoietic stem cell transplantation.³ These early approaches were limited in efficacy and carried significant toxicity, yet rare reports of long-term remissions established the possibility that immune-directed therapy might one day be curative in MM. Over the last 3 decades, the advent of immunomodulatory agents and monoclonal antibodies has steadily advanced this goal. None of these advances, however, has been as dramatic as the recent development of engineered cellular therapies, chief among them chimeric antigen receptor (CAR) T cells. CAR-T cells are unique among MM therapies in that they are a living drug, and in the case of autologous products, a custom therapeutic manufactured for each individual patient.

Structurally, a CAR is a synthetic recombinant fusion protein in which an extracellular antigen-recognition domain, most often a single-chain variable fragment (scFv), is joined through a hinge and transmembrane region to intracellular signaling modules and is transduced typically with a lentiviral vector into target cells.⁴ In the second-generation constructs that define all current MM products, the intracellular modules pair the CD3ζ chain of the normal T cell receptor with a 4-1BB costimulatory domain. This structure enables the engineered cell to recognize target antigens independent of MHC presentation and to induce a rapid and self-amplifying response.

As we review here, these therapies are no longer theoretical and their ability to induce rapid and deep responses, even in otherwise refractory disease, is well established. The challenge now is one of engineering and biology, namely, how to ensure that treatment induces deep and durable remissions, or even cures, for every patient while limiting toxicity.

THE BCMA BACKBONE

While a number of antigenic targets exist and have been tested in MM, the 2 best characterized are B cell maturation antigen (BCMA), encoded by the TNFRSF17 gene, and G protein-coupled receptor class C group 5 member D (GPRC5D). Two anti-BCMA CAR-T cell products are currently FDA-approved, and multiple additional agents directed at one or both of these antigens are in advanced clinical development (Tables 1 and 2).

TABLE 1.

Selected BCMA-Directed CAR-T Cell Products in Multiple Myeloma

Product (Construct)	Trial Phase (NCT)	Line of Therapy / setting	Status
Approved autologous
Idecabtagene vicleucel (ide-cel); murine scFv, 4-1BB	KarMMa — Ph 2 (NCT03361748)	RRMM, ≥3 prior lines (triple-class exposed)	Pivotal; FDA approved 20215
	KarMMa-3 — Ph 3 (NCT03651128)	RRMM, 2-4 prior lines, triple-class exposed	Positive; basis for ≥2-line expansion (2024)6
	KarMMa-2 — Ph 2, multicohort (NCT03601078)	Cohort 2a: early relapse <18 mo of frontline ASCT (2nd line); Cohort 2c / 3: NDMM, inadequate response to / not in CR after frontline ASCT (first-line consolidation)	Ongoing / reported7
	KarMMa-9 — Ph 3 (NCT06045806)	NDMM, suboptimal response to ASCT; ide-cel + len vs len maintenance	Enrollment discontinued (2026)
Ciltacabtagene autoleucel (cilta-cel); 2× camelid VHH, 4-1BB	CARTITUDE-1 — Ph 1b/2 (NCT03548207)	RRMM, ≥3 prior lines	Pivotal; FDA approved 20228
	CARTITUDE-4 — Ph 3 (NCT04181827)	RRMM, 1-3 prior lines, lenalidomide-refractory	Positive (OS benefit); ≥1-line approval (2024)9
	CARTITUDE-2 — Ph 2, multicohort (NCT04133636)	Cohort A: 1-3 prior lines, len-refractory; Cohort B: early relapse ≤12 mo (2nd line); Cohort D: NDMM <CR after frontline ASCT ± len (first-line consolidation)	Ongoing / reported10
	CARTITUDE-5 — Ph 3 (NCT04923893)	First-line NDMM, transplant not planned; VRd→cilta-cel vs VRd→Rd	Ongoing
	CARTITUDE-6 / EMAGINE — Ph 3 (NCT05257083)	First-line NDMM, transplant-eligible; D-VRd→cilta-cel vs D-VRd→ASCT	Ongoing
Emerging autologous
Anitocabtagene autoleucel (anito-cel); D-Domain synthetic binder, 4-1BB	Phase 1 (NCT04155749)	RRMM, ≥3 prior lines	Reported (median PFS ∼30 mo)11
	iMMagine-1 — Ph 2 (NCT05396885)	RRMM, ≥3 prior lines (triple-/penta-class refractory)	Registrational; reported12
	iMMagine-3 — Ph 3 (NCT06413498)	RRMM, 1-3 prior lines (incl. anti-CD38 + IMiD) vs SOC	Ongoing
Durcabtagene autoleucel (PHE885); fully human scFv, 4-1BB; T-Charge	Phase 1 (NCT04318327)	RRMM	Reported (ORR 98%; median PFS 12 mo)13
	Phase 2 (NCT05172596)	RRMM	Enrollment completed
Dual-target (BCMAxCD19)
GC012F (BCMA/CD19 dual; FasTCAR next-day manufacture)	Phase 1, RRMM (NCT04236011)	RRMM	Reported14
	Phase 1, frontline (NCT04935580)	NDMM, transplant-eligible, high-risk (first-line)	Reported (100% ORR / MRD-)14
	Phase 1b/2, RRMM (NCT05850234)	RRMM	Ongoing
Allogeneic
ALLO-715 (allogeneic; TRAC/CD52 KO, TALEN)	UNIVERSAL — Ph 1 (NCT04093596)	RRMM	Reported15
P-BCMA-ALLO1 (allogeneic; TRAC/β2-microglobulin KO, Cas-CLOVER; Tscm-enriched)	Phase 1 (NCT04960579)	RRMM	Ongoing16
In vivo
ESO-T01 (nanobody-directed, immune-shielded lentiviral vector; anti-BCMA)	Phase 1 (NCT06791681)	RRMM	First-in-human; stopped early17
KLN-1010 (envelope-engineered lentiviral vector; anti-BCMA)	inMMyCAR — Ph 1 (NCT07075185)	RRMM	First-in-human; early MRD-neg responses18

ASCT indicates autologous stem cell transplant; D-VRd, daratumumab–bortezomib–lenalidomide–dexamethasone; IMiD, immunomodulatory drug; MRD, minimal residual disease; NDMM, newly diagnosed multiple myeloma; ORR, overall response rate; OS, overall survival; Rd, lenalidomide–dexamethasone; RRMM, relapsed/refractory multiple myeloma; VHH, variable domain of heavy-chain-only antibody.

TABLE 2.

Selected Non-BCMA Directed CAR-T Cell Products in Multiple Myeloma

Product (Construct)	Trial — Phase (NCT)	Line of Therapy / Setting	Status
GPRC5D
Arlocabtagene autoleucel (arlo-cel; BMS-986393); GPRC5D-directed, autologous	Phase 1 — CC-95266-MM-001 (NCT04674813)	RRMM, ≥3 prior lines (with separate 1-3 prior-line cohort)	Reported (ORR: 87%; CR: 53%; PFS: 18.3 mo)19
	QUINTESSENTIAL — Ph 2 (NCT06297226)	RRMM, ≥3 prior lines / ≥4 classes incl. prior anti-BCMA	Ongoing (registrational)
	QUINTESSENTIAL-2 — Ph 3 (NCT06615479)	RRMM, lenalidomide-exposed; vs. standard regimens	Ongoing
MCARH109 (GPRC5D; same construct as arlo-cel)	Phase 1 (NCT04555551)	RRMM, ≥3 prior lines (incl. prior BCMA-directed therapy)	Reported (first-in-human GPRC5D CAR; cerebellar toxicity at highest dose)20
OriCAR-017 (GPRC5D; autologous)	POLARIS — Ph 1 (NCT05016778)	RRMM (incl. prior BCMA CAR-T)	Reported (ORR 100%)21
BMS-986453 (BCMAxGPRC5D; autologous)	Phase 1 (NCT06153251)	RRMM	Ongoing
SLAMF7 (CS1)
CARAMBA (SLAMF7 CAR; virus-free Sleeping Beauty transposon)	Ph 1/2A (NCT04499339)	RRMM	Ongoing22
CD19
huCART19 / CART-BCMA (BCMA alone or in combination with CD19)	Phase 1 (NCT03549442)	Consolidation in patients responding to therapy	Reported (no benefit to CD19)23
CD138 (syndecan-1)
CART-138 (CD138; autologous)	Phase 1 (NCT03672318)	RRMM	Ongoing24

ASCT indicates autologous stem cell transplant; CS1, cell-surface glycoprotein CD2 subset 1; GPRC5D, G protein-coupled receptor class C group 5 member D; RRMM, relapsed/refractory multiple myeloma; SLAMF7, SLAM family member 7.

Idecabtagene vicleucel (ide-cel, bb2121) is a BCMA-targeting CAR, employing a murine scFv and a 4-1BB costimulatory domain, that was first FDA-approved in 2021. In the pivotal KarMMa-1 study, 128 infused patients with heavily pretreated disease achieved an overall response rate (ORR) of 73%, with a complete response (CR) or better in 33%; at the 450×10⁶ cell target dose, the ORR was 82% with a median progression-free survival (PFS) of 12.1 months, without any maintenance therapy.^5,25 Grade 3 or higher hematologic toxicities occurred in the majority of patients, and roughly 3-quarters developed cytokine release syndrome (CRS), although most events were grade 1 or 2 and severe neurological toxicity, including immune effector cell-associated neurotoxicity syndrome (ICANS) was uncommon. These findings were confirmed in the randomized phase 3 KarMMa-3 study, which compared ide-cel with standard-of-care regimens in patients who had received 2 to 4 prior lines.⁶ A total of 386 patients were randomized 2:1. At the final PFS analysis (median follow-up 30.9 mo), median PFS was 13.8 months with ide-cel versus 4.4 months with standard regimens [hazard ratio (HR), 0.49].²⁶ The ORR was 71% versus 42%, and CR rate 44% versus 5%, respectively. Overall survival (OS) was confounded by crossover, as 56% of control patients received ide-cel upon progression; the unadjusted interim survival HR was 1.01, but a prespecified crossover-adjusted analysis favored ide-cel (adjusted HR, 0.72).²⁶ On the basis of these data, the ide-cel indication was expanded in 2024 to patients with 2 or more prior lines of therapy, including an IMiD, a proteasome inhibitor, and an anti-CD38 antibody.

Ciltacabtagene autoleucel (cilta-cel) is an anti-BCMA CAR that differs from ide-cel in its use of 2 camelid single-domain (VHH) BCMA binders in a bivalent configuration that increases avidity for the target. This difference in binding is believed to underlie the observed differences in both response and toxicity. In the phase 1b/2 CARTITUDE-1 registration trial, 97 patients received cilta-cel at the recommended phase 2 dose of 0.75×10⁶ CAR-T cells/kg, producing an ORR of 97% and CR or better in 67% of patients.⁸ Long-term follow-up has been striking: at a median follow-up of 61.3 months, the median OS was 60.7 months and one-third of patients (32 of 97) remained alive and progression-free for at least 5 years after a single infusion, without any maintenance therapy.²⁷ Among a subset undergoing serial assessments, all evaluable long-term responders were minimal residual disease (MRD)-negative and imaging-negative for at least 5 years. While relapses can and do occur, and the tail of the curve is not yet defined, this large fraction of patients remaining disease-free and off all therapy has buoyed hopes for a potentially curative treatment. In CARTITUDE-1, CRS occurred in nearly all patients (95%), with grade ≥3 in 6%, and neurotoxicity including ICANS was seen in 21%, with grade ≥3 events in only 2 patients.

These findings were extended by the phase 3 CARTITUDE-4 study, which randomized 419 lenalidomide-refractory patients exposed to 1 to 3 prior lines 1:1 to cilta-cel (n=208) or standard of care (n=211).⁹ Cilta-cel reduced the risk of progression or death by 74% (HR: 0.26) at the primary analysis. With longer follow-up (median 33.6 mo), median PFS was not reached with cilta-cel versus 11.8 months with standard of care (HR: 0.29), the ORR was 84.6% with CR or better in 69.2%, and MRD negativity at 10^-5 was achieved in 62% versus 18% of patients.²⁸ Most importantly, cilta-cel produced a significant OS benefit (HR: 0.55; P=0.0009; 30 mo OS 76.4% vs. 63.8%), the first time any therapy has demonstrated an overall survival advantage over conventional treatment in this setting, despite many control patients subsequently receiving BCMA-directed therapy.²⁸ These results supported the 2024 expansion of the cilta-cel indication to patients with at least one prior line including a PI and an IMiD who are refractory to lenalidomide.

The toxicities of the approved CAR products, while generally manageable, are distinctive and warrant emphasis. Beyond CRS and ICANS, cilta-cel in particular has been associated with delayed non-ICANS neurotoxicity, including cranial nerve palsies, Parkinsonian movement and neurocognitive syndrome (MNT),^29,30 and gastrointestinal toxicity in the form of immune effector cell-associated enterocolitis.^31,32 Immune effector cell-associated hemophagocytic lymphohistiocytosis-like syndrome (IEC-HS) has also been described.^33,34 These events, while individually uncommon, together impact ∼10% of patients treated with cilta-cel and can lead to severe morbidity or death.^30,35 They occur more commonly in patients treated with cilta-cel compared with ide-cel and their frequency appears to correlate with the degree of CAR-T cell expansion and disease burden, making the patients most in need of highly efficacious therapies like CAR-T the ones at highest risk.^35,36 This has motivated strategies to reduce tumor burden before infusion and to recognize toxicities early so that interventions can be initiated before the development of severe morbidity.³⁷ Prolonged and recurrent cytopenias, hypogammaglobulinemia, and a substantial infection burden are common to both products and represent another leading cause of non-relapse mortality.^38–41 Finally, secondary malignancies, including secondary myeloid malignancies (almost 10% of patients in CARTITUDE-1 developed myelodysplasia or acute myeloid leukemia) and rare cases of CAR-positive T cell lymphoma clonally derived from the engineered product, have prompted an FDA boxed warning and ongoing investigation of T cell malignancy risk following treatment with CAR-T cell therapies generally.^35,41–48 Collectively, these toxicities shape the risk-benefit calculus especially as CAR-T moves to earlier lines and a major area of study is the identification of patients at the highest risk of severe toxicity and the development of mitigation strategies.

While no direct head-to-head randomized comparison of ide-cel and cilta-cel has been performed, and none is likely ever to be conducted, a growing body of real-world evidence has evaluated their efficacy and toxicity outside of the trial setting.³⁵ Registry analyses from the US Multiple Myeloma Immunotherapy Consortium,³⁵ the French DESCAR-T registry,⁴⁹ and the CIBMTR⁵⁰ have consistently reproduced the pivotal-trial results even though the majority of real-world recipients would not have met original trial eligibility. For ide-cel, real-world ORR and median PFS have closely mirrored the KarMMa experience.⁵¹ Retrospective comparisons have demonstrated higher response rates and longer PFS with cilta-cel than with ide-cel (ORR: 86% vs. 60%;⁵² median PFS not reached vs. 7.5 mo), although ide-cel was more often selected for older or frailer patients with adverse features, confounding any direct interpretation and underscoring the role of patient selection.³⁵ Longer follow-up will be needed to determine whether the long tail of disease-free survival seen in CARTITUDE-1 is recapitulated in the real-world setting.

NEW PRODUCTS, NEW TARGETS, AND NEXT-GENERATION PLATFORMS

Multiple new BCMA-targeting CARs are in clinical development, alongside concerted efforts to engineer products with less toxicity, more rapid turnaround, and properties such as improved persistence that might translate into deeper and more durable responses. Although complete loss of BCMA is relatively uncommon after CAR-T, downregulation can occur and other antigens such as GPRC5D appear to be lost more commonly.^53,54 This has motivated the development of dual-antigen and even triple-antigen targeting constructs. Allogeneic “off-the-shelf” products, and in vivo-generated CARs are in various stages of clinical development, and together they represent the likely trajectory of the field over the coming decade. Other approaches, such as armored CARs and the use of other immune cells, such as NK cells (both autologous and allogeneic), are much earlier stages of development and have a less clear route to the clinic.

Next-Generation BCMA CARs

Anitocabtagene autoleucel (anito-cel, CART-ddBCMA) is an anti-BCMA CAR that employs a novel, compact synthetic binder known as a D-Domain in place of a conventional scFv.¹¹ Its small size permits high surface expression, and its binding kinetics characterized by a rapid off rate are thought to reduce tonic signaling and preserve T cell fitness. In the registrational phase 2 iMMagine-1 trial, at ∼12.6 months of median follow-up, the ORR was 96%, with CR or better in 68% and MRD negativity at 10^-5 in 93% of evaluable patients and an estimated 18-month PFS of 66%.¹² In the earlier phase 1 experience, the median PFS was 30.2 months.¹² Notably, across the anito-cel program, there have been no reports of delayed or non-ICANS neurotoxicity and no cases of MNT or cranial nerve palsies. Together, these data suggest that anito-cel may achieve efficacy approaching that of cilta-cel while avoiding much of its distinctive neurological toxicity. The phase 3 iMMagine-3 study is underway and initial regulatory approval in late-line patients is expected in the next year.

GC012F is a dual-targeting CAR directed against both BCMA and CD19, produced on the FasTCAR next-day manufacturing platform.⁵⁵ While CD19 is not expressed on most malignant plasma cells, it has been proposed to mark a less differentiated, disease-propagating compartment, providing a rationale for its inclusion. Prior efforts using anti-CD19 CARs alone or co-infused with BCMA-targeting CARs have yielded mixed results. Despite this, early reports of activity with GC012F, including in newly diagnosed patients,¹⁴ have been promising, and multiple trials in a variety of disease settings are now underway. The specific contribution of the CD19-targeting component and of the rapid manufacturing itself to the observed responses remains difficult to isolate.

Other efforts to enhance the potency and persistence of the CAR product have also focused on shortened manufacturing to reduce ex vivo expansion and T cell differentiation. Durcabtagene autoleucel (durca-cel, PHE885) is a fully human anti-BCMA CAR produced using the T-Charge platform, which compresses ex vivo processing to under 2 days.¹³ This rapid process shortens vein-to-vein time and preserves a less differentiated, stem-cell-memory phenotype. Durca-cel demonstrated robust in vivo expansion and long-term polyclonal persistence, with functional CAR T cells detectable beyond 6 months in roughly half of patients.¹³ Whether this favorable cellular phenotype translates into deeper or more durable clinical responses remains to be established.¹³ Several other groups have pursued similarly abbreviated manufacturing platforms, including the NEX-T process (BMS-986354) and the FasTCAR discussed above, reflecting a broad interest in reducing culture time both to improve access and to select for a more favorable T cell product.⁵⁶

A New Target: GPRC5D

Arlocabtagene autoleucel (arlo-cel; BMS-986393) and the related product MCARH109 is a GPRC5D-directed CAR.²⁰ In its multicohort phase 1 study of 84 patients, arlo-cel produced deep and durable responses across heavily pretreated patients, with an ORR 87%, a CR rate 53%, and a median PFS of 18.3 months.¹⁹ The phase 3 QUINTESSENTIAL-2 study is now underway.⁵⁷ Because GPRC5D expression is independent of BCMA, arlo-cel offers the prospect of activity in patients who have relapsed after BCMA-directed therapy.⁵⁸ In addition to general CAR-associated toxicities including CRS and ICANS, it carries a distinct on-target/off-tumor toxicity profile reflecting GPRC5D expression in keratinized tissues⁵⁸ (dysgeusia, rash, and nail changes) and cerebellum leading to sometimes-debilitating vertigo, dysarthria, and ataxia.^19,20 While skin toxicities are typically less severe with the CAR compared with the T cell engager Talquetamab, the cerebellar toxicities occur more frequently.⁵⁹ The complementary target biology and nonoverlapping toxicity of GPRC5D and BCMA targeting agents set the stage for combinatorial and dual-targeting strategies.

Dual-Antigen and Multiantigen Targeting

The most direct approach to preventing antigen escape is to target more than one antigen simultaneously.⁶⁰ BMS-986453, a bispecific CAR targeting both BCMA and GPRC5D, is in phase 1 dose escalation (NCT06153251). A variety of construct designs are being explored, including tandem and loop configurations and armored bispecifics incorporating VHH binders engineered to resist inhibition by soluble BCMA.⁶¹ Extending this logic further, preclinical trispecific, armored CARs targeting BCMA, GPRC5D, and CD38 with novel VHH binders have demonstrated superior expansion and tumor killing relative to mono-specific or bispecific constructs.⁶² Whether the added complexity of these designs will impair manufacturing reliability and will yield superior durability without excess toxicity remains an open and important question.

Armored CARs

Beyond antigen redundancy, “armoring” refers to engineering CAR T cells to secrete or express additional factors that enhance their function or remodel the immunosuppressive tumor microenvironment. Strategies under investigation include CARs engineered to secrete supportive cytokines such as IL-15⁶³ or IL-18,⁶⁴ to express a dominant-negative TGF-β receptor that renders them resistant to TGF-β-mediated suppression,⁶⁵ or to deliver secreted checkpoint-blocking molecules locally.⁶⁶ These approaches are aimed squarely at the twin problems of T cell exhaustion and limited persistence that underlie antigen-positive relapses, and although largely preclinical and not yet tested in MM, they represent a rational next step.

Allogeneic CARs

Allogeneic “off-the-shelf” CARs would obviate the delay, cost, and manufacturing variability of autologous products and provide immediate availability for patients with rapidly progressive disease. Products such as ALLO-715 (with TRAC and CD52 knockout enabling anti-CD52-based lymphodepletion)¹⁵ and P-BCMA-ALLO1 (with TRAC and β2-microglobulin editing to limit host rejection)¹⁶ have shown proof-of-concept activity, but their promise has been tempered by limited CAR persistence and the need for intensive lymphodepletion that leaves patients vulnerable to atypical infections. Allogeneic CARs will likely only become viable once complex genetic editing to enable long-term persistence and activity without harsh immunosuppression becomes possible.

In Vivo CARs

By generating CAR-T cells directly within the patient, many of the challenges associated with CAR, including leukapheresis, ex vivo manufacturing, and lymphodepleting chemotherapy, can be bypassed altogether. Two first-in-human efforts in myeloma have now been reported. ESO-T01, a nanobody-directed, immune-shielded lentiviral vector encoding an anti-BCMA CAR, was administered as a single intravenous infusion without lymphodepletion to 5 heavily pretreated patients.^17,67 The approach proved feasible, albeit with severe infusion-related reactions, and the trial was stopped early.¹⁷ KLN-1010, an envelope-engineered lentiviral vector delivering a fully human anti-BCMA CAR, produced MRD-negative responses in the first patients treated on the phase 1 inMMyCAR study.¹⁸ While these results are preliminary and questions of durability and long-term safety remain to be evaluated, they establish the biological feasibility of in vivo CAR generation. If borne out, this approach will likely reshape the logistics and accessibility of cellular therapy for MM and become the foundation on which further advanced constructs are built.

DETERMINANTS OF RESPONSE AND RELAPSE

Although CAR-T cells induce responses in most patients, the depth and durability of those responses vary considerably. The determinants of response are increasingly understood to be as much a function of the host and the T cell product as of the construct itself. Response and persistence are strongly influenced by the fitness of the starting T cell population, which is negatively impacted by both the immunosuppressive effects of neoplastic plasma cells and by cumulative exposure to cytotoxic agents.⁶⁸ Long-term responders in CARTITUDE-1, for example, were enriched for central-memory CAR-T cell phenotypes, and real-world experience suggests that fitter T cells collected earlier in the disease course yield better outcomes.^27,35,68 In contrast, no difference was seen in the benefit of cilta-cel among patients who had received only one prior line of therapy as compared with those who had received 2–3 lines.²⁸ High tumor burden, extramedullary disease, and high-risk cytogenetics at the time of infusion all predict an inferior likelihood of long-term remission.⁶⁹

Relapse arises through 2 broad and nonexclusive routes: (i) antigen-positive relapse, driven by loss of CAR-T cell persistence, T cell exhaustion, an immunosuppressive microenvironment, or highly proliferative MM capable of outgrowing the CAR, which account for the majority of failures; and (ii) antigen-negative relapse, driven by biallelic loss, mutation, or silencing of BCMA (or GPRC5D), which is less common but increasingly relevant as antigen-targeted therapies are used sequentially.^53,70–73 These patterns may also change as products are engineered to produce longer persistence, leading to continuous selective pressure on antigen-expressing cells, akin to what is seen during treatment with T cell engaging antibodies where antigen loss is more common.^53,74 These mechanisms of treatment failure are what the next-generation strategies discussed above, including improved binders, multiantigen targeting, armoring, earlier collection, and in vivo production, are designed to address.

INTEGRATING MINIMAL RESIDUAL DISEASE AND MAINTENANCE

A defining feature of CAR-T cell therapy in myeloma is the frequency with which it drives minimal residual disease (MRD) negativity. MRD negativity, and especially sustained MRD negativity, is the strongest available correlate of durable benefit and functions as both a prognostic marker and a key surrogate endpoint.⁷⁵ In addition, it is a requisite feature of the new International Myeloma Society definition of cure. The approved and emerging products all achieve high rates of MRD negativity, 62% at 10^-5 with cilta-cel in CARTITUDE-4 and 93% with anito-cel.^12,28 Importantly, among the exceptional long-term responders in CARTITUDE-1, those who may represent the cure fraction, all evaluable patients have remained MRD-negative at 10^-6 and imaging-negative for 5 years or more, making this the high bar over which other products must reach.²⁷

The question of consolidation and/or maintenance after CAR-T cell therapy remains unresolved and is, in some respects, unique to this modality. Unlike other MM therapies, which are given continuously until progression, the approved CAR products are administered as a single infusion with no subsequent maintenance. The durability of cilta-cel, in particular, demonstrates that a meaningful fraction of patients require nothing further. Yet most patients still relapse, providing a rationale for post-CAR consolidation or maintenance aimed at deepening responses and eliminating residual disease. The strongest biological preclinical rationale exists for the immunomodulatory agents, which can enhance T cell function and CAR-T cell persistence. Early data support a potential benefit to lenalidomide maintenance following CAR infusion^76–78 and the use of newer CELMoDs as consolidation/maintenance is under active investigation.^79,80 Maintenance adds infectious, hematologic, and potentially second-malignancy risk, along with cost and the burden of continuous therapy, to a treatment whose appeal lies partly in being finite. An MRD-adapted strategy in which maintenance/consolidation is reserved for patients who remain MRD-positive or who carry high-risk features offers the most rational path to avoiding excess toxicity for those who do not need it. Other approaches in preclinical development include administering agents that reinvigorate existing CAR-T cells, so-called CAR enhancers, that can be given after initial CAR expansion to prolong persistence and improve efficacy.⁸¹ Ongoing studies will determine the ideal approach.

CONCLUSIONS

In a remarkably short time span, CAR-T cell therapy has moved from an experimental salvage option to a therapy with a demonstrated overall survival advantage and a durable, potentially curative tail in a subset of patients. These therapies are now advancing rapidly into earlier lines and toward the frontline setting in trials such as CARTITUDE-5, CARTITUDE-6, and DURGA-2. While the promise is great, the remaining challenges are clear: extending durability from a fortunate subset to all patients, overcoming the logistical and access barriers of autologous manufacturing, mitigating the distinctive toxicities of these living drugs, and integrating them rationally with MRD monitoring to identify patients who might benefit from consolidation and/or maintenance. CAR-T cell therapy has already transformed the treatment of myeloma and now it remains to be seen whether the limited long-term remissions seen thus far are the baseline for engineering a cure.