Ciltacabtagene autoleucel in high-risk smoldering multiple myeloma: the CAR-PRISM phase 2 trial

High-risk smoldering multiple myeloma (HR-SMM) carries an increased risk of progression to multiple myeloma, making it an ideal setting to test whether chimeric antigen receptor (CAR) T cell therapy can achieve curative outcomes. Here in this phase 2 study, patients with HR-SMM received ciltacabtagene autoleucel (cilta-cel) at 0.3–0.5 × 10 6 or >0.5 × 10 6 viable CAR + T cells per kilogram without induction or bridging therapy. Patients with >40% marrow involvement were excluded. Primary endpoints were dose-limiting toxicities (DLTs) and treatment-emergent adverse events; secondary endpoints included response and minimal residual disease (MRD) negativity. As of 11 February 2026, 20 patients had been treated. The trial met the prespecified endpoints. No DLTs occurred. Adverse events included transient cytopenias (90% grade 3/4) and cytokine release syndrome (100% grade 1/2). Non-immune effector cell-associated neurotoxicity syndrome neurologic toxicities (NINTs) occurred in seven patients, with four comprising cranial nerve palsies that completely resolved. Three patients had persistent grade 1 symptoms. At a median follow-up of 15.3 months, all patients achieved MRD negativity 10 −6 by 2 months and have remained MRD negative. Sixteen patients with follow-up >6 months achieved a complete response; no progression or deaths were observed. Cilta-cel produced rapid, deep, sustained MRD-negative responses in HR-SMM without induction therapy. Toxicities were consistent with the safety profile of cilta-cel. ClinicalTrials.gov: NCT05767359 .

Ciltacabtagene autoleucel (cilta cel), a B cell maturation antigen (BCMA)-directed CAR T cell therapy, is approved for patients with relapsed or refractory multiple myeloma (RRMM) who have received one prior line of therapy and are lenalidomide refractory 1 . Although cilta-cel can produce durable responses beyond 5 years in a subset of heavily pretreated patients with RRMM 2 , most patients ultimately have disease progression, with a median progression-free survival (PFS) of approximately 34 months in late-line disease 3 .

Clinical data suggest that efficacy may be greater when CAR T cell therapy is administered earlier in the disease course than in later lines, potentially reflecting greater T cell fitness and a less immunosuppressive host environment 4 . These observations have generated interest in evaluating CAR T cell therapy in earlier disease states, and ongoing trials are assessing cilta-cel in newly diagnosed multiple myeloma 5 , 6 .

HR-SMM is associated with a substantial risk of progression to symptomatic multiple myeloma with end-organ damage 7 , 8 , 9 , 10 , 11 , 12 , 13 . Genomic studies indicate that HR-SMM is often a biologically mature malignancy that resembles overt myeloma although with less genetic heterogeneity, lower burden of disease and, in many cases, better-preserved immune function 14 , 15 , 16 .

Over the past decade, multiple early-intervention strategies have been tested in HR-SMM with the goals of delaying progression and, more recently, achieving deep remissions that may intercept disease evolution 17 , 18 . Single-agent daratumumab is the first therapy approved by the US Food and Drug Administration (FDA) for HR-SMM 19 . However, multiagent regimens adapted from symptomatic myeloma (triplets and quadruplets) have achieved MRD negativity in only approximately half of patients, suggesting that alternative approaches may be required to achieve durable, treatment-free remissions in this precursor setting.

We hypothesized that cellular therapy could produce deep and durable responses in HR-SMM because of lower tumor burden, reduced genomic complexity and higher T cell fitness, thereby enhancing the benefit from T-cell-redirecting therapies and raising the possibility of cure 20 . We, therefore, conducted a phase 2 trial to evaluate the efficacy and safety of cilta-cel in patients with HR-SMM and to explore biomarkers associated with response and toxicity.

Between 15 April 2023 and 3 July 2025, 23 patients were enrolled; 20 underwent leukapheresis and received a single cilta-cel infusion (Fig. 1a ). Planned dosing included an initial safety run-in (six patients; standard 3 + 3 design) to evaluate escalating dose levels (0.5 × 10 6 and 0.5–1.0 × 10 6 viable CAR + T cells per kilogram), followed by dose expansion at the protocol-specified target dose (0.5–1.0 × 10 6 viable CAR + T cells per kilogram) (Fig. 1b ). Due to emerging data that peak absolute lymphocyte count (ALC) may be associated with increased risk of neurologic toxicities, along with low threshold for non-ICANS neurologic toxicities (NINTs) in this population, the protocol was amended and the final seven patients received 0.3 × 10 6 viable CAR + T cells per kilogram. All patients had successful manufacturing of cilta-cel, and no patient discontinued the study prior to cilta-cel infusion. At the data cutoff for this report (11 February 2026), all 20 treated patients remained in the primary analysis population.

a , CONSORT flow diagram. Numbers of participants screened, enrolled, treated and included in analyses are shown, with reasons for exclusion and allocation to each protocol phase. b , Trial design and dose modifications. The study included an initial safety run-in with three treated participants. After completion of safety assessment and FDA review, an additional three participants were treated before opening the expansion cohort at a target dose of 0.5–1.0 × 10 6 CAR + T cells per kilogram. After identification of a safety signal, the protocol was amended to reduce the target dose to 0.3 × 10 6 CAR + T cells per kilogram and to introduce prophylactic dexamethasone, triggered by ALC > 3,000 per microliter. Squares denote individual participants and are positioned at the delivered (received) dose. ALC, absolute lymphocyte count; BM, bone marrow; DLT, dose-limiting toxicity; PET/CT, positron emission tomography/computed tomography.

Among treated patients, the median age was 58 years (range, 37–78), and the median time between initial diagnosis and enrollment was 16.1 months (interquartile range (IQR), 8.4–36.2) (Table 1 ). Across the cohort, the median bone marrow plasma cell infiltration was 20% (IQR, 20–30), consistent with the protocol-defined eligibility criterion limiting marrow involvement to 40% or less. Thirteen patients (65%) harbored high-risk cytogenetic abnormalities, defined in this study as del(17p), t(4;14), t(14;16), del(13q) or gain or amplification of chromosome 1q. M-protein levels, serum free light chain ratios and other tumor surrogate markers were within ranges typical for SMM. Thirteen patients (65%) met the 20–2–20 definition of high risk (Supplementary Table 1 ).

The median time from screening to leukapheresis was 15 days (IQR, 12–27), and the median time from leukapheresis to CAR T cell infusion was 56 days (IQR, 52–76), reflecting a protocol-specified 1-month stagger implemented for the initial cohort. Bridging therapy before cilta-cel infusion was not permitted per protocol. According to established risk stratification tools developed in the relapsed myeloma CAR T cell setting, including CAR-HEMATOTOX and MyCARE, all patients in this cohort were at low risk for survival or toxicity events 21 , 22 .

The primary endpoint of this study was safety. Adverse events were reported in all 20 treated patients (Table 2 and Supplementary Table 2 ). No patients met protocol-defined DLT criteria in the safety run-in cohort.

Cytokine release syndrome (CRS) occurred in all patients, with 17 (85%) cases of grade 1, three (15%) cases of grade 2 and no grade 3 or higher CRS events. Tocilizumab was administered in 17 patients (85%) and dexamethasone in 13 patients (65%) using a low clinical threshold.

Neutropenia was common, with grade 4 observed in 11 (55%) patients and grade 3 observed in seven (35%) additional patients. These events were predominantly short-lived, with a median duration of 7 days (IQR, 4−13) (Extended Data Fig. 1a ). Accordingly, all patients were classified as low risk by immune effector cell-associated hematotoxicity (ICAHT) grading (Extended Data Fig. 1b ) 23 . Severe thrombocytopenia or anemia (grade 3 or 4) was uncommon, occurring in three and two patients, respectively (Extended Data Fig. 1c,d ). Two cases of severe thrombocytopenia had features consistent with immune thrombocytopenia; one of these patients also developed hemolytic anemia, possibly related to fludarabine. These events were transient and responded to corticosteroids and intravenous immunoglobulin (IVIG). No persistent hematologic toxicity was observed.

Infectious adverse events were all grade 1−2, primarily affecting the upper respiratory tract (25%). All patients received IVIG prophylaxis.

Seven patients developed NINTs, four comprising cranial nerve palsies (three grade 1 and one grade 2), with objective improvement after 2 weeks and full resolution within 4−8 weeks (Table 3 ). Five of seven patients with NINTs received 0.5−0.8 × 10 6 CAR + T cells per kilogram dose. Median time to onset was 21 days (range, 10−52). The only grade 2 event included bilateral cranial nerve VII palsies with mononeuritis multiplex in a patient with possible predisposing risk factor of diabetes mellitus and was treated with dexamethasone and IVIG, resulting in complete resolution. Grading of NINTs was based on symptom severity, and management was deliberately proactive and, in some cases, escalated out of caution rather than due to higher-grade neurologic deficits.

Three patients had other ongoing NINTs (non-cranial nerve palsies) at the data cutoff after doses of 0.8 × 10 6 , 0.7 × 10 6 and 0.3 × 10 6 viable CAR + T cells per kilogram (Extended Data Fig. 2 ). Patients 14 and 18, both treated at ≥0.5 × 10 6 viable CAR + T cell dose, had grade 1 movement and neurocognitive symptoms, including one of the following: tremor, bradykinesia or bradyphrenia, gait disturbance or postural instability, dysarthria or cognitive or personality changes. Patient 18 had deposition of phosphorylated α-synuclein on skin biopsy, suggesting a potential underlying predisposition to Parkinson’s disease 24 . Despite these symptoms, both patients remained fully functional, did not require assistive devices and continued to work with minimal interference with daily activities. Patient 25, treated at 0.3 × 10 6 viable CAR + T cell dose, had grade 1 intention tremor during fine motor tasks without interference with daily living activities. Longitudinal CAR T cell measurements in peripheral blood and cerebrospinal fluid (CSF) demonstrated heterogeneous kinetics: patient 18 had persistent CAR T cell detection in peripheral blood and CSF through month 15 (predominantly CD4 CAR + T cells); patient 14 cleared CAR T cells from peripheral blood by day 56 with no detection in CSF; and patient 25 had CAR T cells detectable in peripheral blood through day 100 with limited CSF assessment (day 52 cytology negative; flow cytometry not performed). Brain magnetic resonance imaging was normal for patients 18 and 14. Therapeutic approaches varied and included high-dose corticosteroids and IVIG (all three), symptomatic agents (amantadine and/or dopaminergic therapy) and, in patient 18, escalation to intrathecal chemotherapy, low-dose ruxolitinib and cyclophosphamide. None of the patients developed symptoms exceeding grade 1 per Common Terminology Criteria for Adverse Events (CTCAE). The initiation of interventions was not prompted by worsening neurologic toxicity but, rather, reflected a preemptive management strategy to prevent further symptom progression. Clinical responses were mixed: tremor resolution and improvement in bradykinesia and bradyphrenia in patient 14; improved but persistent symptoms in patient 18; and improvement in intention tremor in patient 25.

No immune effector cell-associated hemophagocytic lymphohistiocytosis-like syndrome (IEC-HS) syndrome was observed. One patient developed transient grade 3 aspartate aminotransferase (AST) and alanine aminotransferase (ALT) elevation after cilta-cel infusion in the setting of recent grade 1 CRS, with concurrent grade 3 hypertriglyceridemia, and improved after dexamethasone. A second patient had transient grade 3–4 hypertriglyceridemia after grade 1 CRS without concurrent clinically significant transaminitis or other features suggestive of IEC-HS.

Gastrointestinal adverse events were limited to transient grade 1–2 diarrhea and vomiting. These events were temporally associated with lymphodepleting chemotherapy, cilta-cel infusion and/or low-grade CRS, and no cases were consistent with characteristics of immune effector cell-associated enterocolitis (IEC-enterocolitis).