Matching-adjusted indirect comparison of efficacy outcomes for ciltacabtagene autoleucel in CARTITUDE-1 versus idecabtagene vicleucel in KarMMa for the treatment of patients with relapsed or refractory multiple myeloma

Abstract

Objective

This study estimated the comparative efficacy of ciltacabtagene autoleucel (cilta-cel) versus the approved idecabtagene vicleucel (ide-cel) dose range of 300–460 × 10⁶ CAR-positive T-cells for the treatment of patients with relapsed or refractory multiple myeloma (RRMM) who were previously treated with a proteasome inhibitor, an immunomodulatory drug, and an anti-CD38 monoclonal antibody (i.e. triple-class exposed) using matching-adjusted indirect treatment comparisons (MAICs).

Methods

MAICs were performed with individual patient data for cilta-cel (CARTITUDE-1; NCT03548207) and published summary-level data for ide-cel (KarMMa; NCT03361748). Treated patients from CARTITUDE-1 who satisfied the eligibility criteria for KarMMa were included in the analyses. The MAIC adjusted for unbalanced baseline covariates of prognostic significance identified in the literature and by clinical expertise. Comparative efficacy was estimated for overall response rate (ORR), complete response or better (≥CR) rate, duration of response (DoR), progression-free survival (PFS), and overall survival (OS).

Results

Cilta-cel was associated with statistically significantly improved ORR (odds ratio [OR]: 94.93 [95% confidence interval [CI]: 21.86, 412.25; p < .0001]; relative risk [RR]: 1.34), ≥CR rate (OR: 5.49 [95% CI: 2.47, 12.21; p < .0001]; RR: 2.21), DoR (hazard ratio [HR]: 0.50 [95% CI: 0.29, 0.87; p = .0137]), and PFS (HR: 0.37 [95% CI: 0.22, 0.62; p = .0002]) when compared with ide-cel. For OS, the results were in favor of cilta-cel and clinically meaningful but with a CI overlapping one (HR: 0.55 [95% CI: 0.29, 1.05; p = .0702]).

Conclusions

These analyses demonstrate improved efficacy with cilta-cel versus ide-cel for all outcomes, highlighting its therapeutic potential in patients with triple-class exposed RRMM.

Keywords:

• Relapsed or refractory multiple myeloma
• ciltacabtagene autoleucel
• CARTITUDE-1
• idecabtagene vicleucel
• KarMMa
• indirect treatment comparison

View correction statement:

Correction

Introduction

Multiple myeloma (MM) is a highly heterogenous cancer that accounts for 10% of all hematological malignancies¹. MM is characterized by the overproduction of monoclonal (M) protein and proliferation of malignant plasma cells, resulting in a variable clinical course often associated with anemia, lytic bone disease, kidney dysfunction, progressive debility, and ultimately mortality¹.

Available treatment options for MM include proteasome inhibitors (PIs), immunomodulatory drugs (IMiDs), and monoclonal antibodies (MoABs)². Despite many advancements in therapy with improved responses and survival, MM remains an incurable malignancy and most patients ultimately experience relapse and succumb to their disease. New, more efficacious treatment options are required for relapsed or refractory multiple myeloma (RRMM) to extend survival, delay disease progression, and improve quality of life.

To address this unmet need, two novel chimeric antigen receptor T-cell (CAR-T) products have emerged for RRMM patients. Ciltacabtagene autoleucel (cilta-cel; JNJ-68284528) is a CAR-T therapy with two B-cell maturation antigen (BCMA)-targeted, single-domain antibodies designed to confer avidity³. Cilta-cel is being evaluated in the CARTITUDE-1 trial (NCT03548207) for the treatment of RRMM patients who received an IMiD, a PI, and an anti-CD38 MoAB as part of previous therapy (i.e. triple-class exposed)⁴. The Food and Drug Administration (FDA) granted Breakthrough Therapy Designation for cilta-cel in the treatment of RRMM patients and recently accepted the Biologics License Application for priority review⁵. Idecabtagene vicleucel (ide-cel; bb2121), another BCMA-directed CAR-T therapy, has been evaluated for the treatment of triple-class exposed patients with RRMM in the KarMMa trial (NCT03361748)⁶. Ide-cel was recently approved by the FDA as a one-time infusion of 300–460 × 10⁶ CAR-positive (CAR+) T cells for adult patients with MM who are triple-class exposed and have received at least four prior therapies⁷.

The comparative efficacy of the two novel CAR-T products, cilta-cel and ide-cel, has not been assessed in a head-to-head clinical trial in the treatment of patients with triple-class exposed RRMM. In the absence of head-to-head clinical trials, indirect treatment comparisons (ITCs) can provide insight into the comparative efficacy of non-randomized patient cohorts. However, a naïve comparison of outcomes from both trials may be biased due to confounding that arises from imbalances in prognostic characteristics of the cohorts being compared. Using individual patient-level data (IPD) for CARTITUDE-1 and published summary-level results for KarMMa, the matching-adjusted indirect comparison (MAIC) method⁸ can adjust for confounding bias and the effect size of the outcomes can be predicted as if the population of KarMMa received cilta-cel. The objective of this study was to estimate the comparative efficacy of cilta-cel versus ide-cel for the treatment of patients with RRMM who were triple-class exposed and had received at least three prior lines of therapy (LOTs) using MAICs.

Methods

Study design

CARTITUDE-1 (NCT03548207) is a phase 1 b/2 trial evaluating the safety and efficacy of cilta-cel in adult patients with RRMM³. Patients diagnosed with MM according to International Myeloma Working Group (IMWG) diagnostic criteria were eligible if they were triple-class exposed, and had received at least three prior LOTs or were double-refractory to an IMiD and a PI³. Full CARTITUDE-1 inclusion and exclusion criteria have been previously published³. Patients were enrolled in CARTITUDE-1 between 16 July 2018 and 7 October 2019, and the treated population consisted of 97 patients from the United States cohort who received an infusion of cilta-cel³. The data cutoff for CARTITUDE-1 was February 2021 and the median follow-up duration was 18 months⁹.

KarMMa (NCT03361748) is a phase 2 trial designed to assess the efficacy and safety of ide-cel in adult patients with RRMM¹⁰. Eligible patients were diagnosed with MM according to IMWG diagnostic criteria and failed at least three prior LOTs including a PI, an IMiD, and an anti-CD38 MoAB¹⁰. Full KarMMa inclusion and exclusion criteria have been previously published¹⁰. Patients were enrolled in KarMMa between 13 December 2017 and 13 November 2018¹⁰. The population of interest for the present study consisted of 124 patients who underwent leukapheresis for the 300 × 10⁶ and 450 × 10⁶ CAR + T cells dose cohorts and received an initial infusion of ide-cel¹⁰. Although an additional four patients were treated in the 150 × 10⁶ CAR-T cells cohort¹⁰, this dose level has not been approved and was therefore not included. The data cutoff for KarMMa was January 2020 and the median follow-up duration was 13.3 months¹⁰.

A feasibility assessment was conducted a priori to assess cross-trial imbalances in available patient or disease characteristics of prognostic significance. CARTITUDE-1 and KarMMa are both open-label, multi-center, single-arm, non-randomized trials that uses the same lymphodepleting chemotherapy (300 mg/m² of cyclophosphamide and 30 mg/m² of fludarabine)^3,4,6,10. Furthermore, bridging therapy for disease control was allowed in both studies provided it was restricted to treatments that the patients had previously received^3,10. The studies differed in their site locations, and CAR-T cell type and dose. Regarding patient characteristics, both trials included patients with triple-class exposed MM and required patients to have measurable disease, adequate organ function, and be ambulatory and able to complete work activities (i.e. Eastern Cooperative Oncology Group [ECOG] score of 0 or 1)^3,10. Although only KarMMa required patients to be refractory to their last LOT, all but one patient in CARTITUDE-1 also satisfied this criterion^3,10. An overview of the study designs and patient populations of CARTITUDE-1 and KarMMa is shown in Supplemental Appendix Table A.1. Based on the feasibility assessment, a MAIC was determined to be an appropriate approach to compare the treated populations of CARTITUDE-1 and KarMMa.

Outcomes

Five efficacy outcomes were assessed in this analysis: overall response rate (ORR), complete response or better (≥CR) rate, duration of response (DoR), progression-free survival (PFS), and overall survival (OS). In both CARTITUDE-1 and KarMMa, evaluation of ORR, ≥CR rate, DoR, and PFS was based on IMWG criteria and adjudicated by the independent review committee (IRC)^3,10. The outcome definitions for the treated populations were aligned in both trials: ORR was defined as the proportion of patients who achieved a partial response or better; ≥CR rate was defined as the percentage of subjects who achieved complete response or stringent complete response; DoR was defined as the time from initial documentation of a partial response or better to the date of disease progression, or death due to any cause, whichever occurred first; PFS was defined as the duration from the date of infusion to progressive disease, or death from any cause, whichever occurred first; and OS was defined as the time from infusion to the date of death due to any cause^3,10.

Statistical analyses

Matching-adjusted indirect comparison

Given the lack of a common comparator, an unanchored MAIC^8,11 was used to estimate the relative treatment effect of cilta-cel versus ide-cel by leveraging IPD from CARTITUDE-1 and published summary-level data from KarMMa^10,12,13. Simulated IPD were derived from summary-level data for KarMMa by reconstructing reported data for ORR and ≥ CR rate, and by using the Guyot method^14–16 to digitize published Kaplan–Meier curves for DoR, PFS, and OS.

The CARTITUDE-1 population was adjusted to match the eligibility criteria and distribution of prognostic factors in KarMMa. Infused patients from CARTITUDE-1 who satisfied the eligibility criteria of KarMMa were reweighted to adjust for imbalances in baseline characteristics of prognostic significance (see Identification and Rank Ordering of Prognostic Factors). Weights were based on a generalized method-of-moments propensity score algorithm, which guaranteed a close balancing of covariates between the CARTITUDE-1 and KarMMa populations^17,18. The effective sample size (ESS) was calculated to reflect the impact of weighting on the available information in the IPD¹⁷. The weighted IPD for CARTITUDE-1 were combined with the simulated IPD for KarMMa.

Relative efficacy for cilta-cel versus ide-cel was determined for the binary endpoints ORR and ≥ CR rate by estimating the odds ratios (ORs) and their 95% confidence intervals (CIs) using weighted logistic regression. Furthermore, relative risks (RRs) were calculated as the ratio of response rates for cilta-cel versus ide-cel. For the time-to-event endpoints DoR, PFS, and OS, hazard ratios (HRs) and their 95% CIs were estimated using a weighted Cox proportional hazards model¹⁹. Validity of the proportional hazards assumption for DoR, PFS, and OS was assessed using the Grambsch-Therneau test²⁰ (with a p-value less than .05 considered to indicate a violation of the assumption).

All analyses were conducted using SAS 9.4 based on the methods developed by Signorovitch et al.⁸, and in line with the National Institute of Health and Care Excellence (NICE) Evidence Synthesis Technical Support Document Series¹¹.

Identification and rank ordering of prognostic factors

Confounding driven by differences in baseline patient and disease characteristics that are prognostic of the outcomes of interest leads to biased comparative efficacy estimates if left unadjusted. Fifteen potentially important prognostic factors, which were identified based on a literature review for clinical outcomes in triple-class exposed RRMM and by consulting independent clinical experts, were ranked in order of importance prior to conducting the analysis (Supplemental Appendix Table B.1). Four factors were identified to be the most important prognostic variables and were included in the base case analysis: refractory status, cytogenetic profile, revised International Staging System (R-ISS) stage, and all plasmacytomas. A series of sensitivity analyses were conducted whereby all commonly available rank-ordered variables (i.e. number of prior LOTs, time since MM diagnosis, age, prior stem cell transplant, ECOG status, and sex) were sequentially and incrementally adjusted for.

Research ethics statement

The CARTITUDE-1 trial protocol was reviewed and approved by an independent ethics committee/institutional review board (IEC/IRB) at all participating sites. All patients participating in the trial provided written informed consent. Similarly, the KarMMa trial protocol was approved by local or independent IRBs or ethics committees at participating sites and all patients provided written informed consent¹⁰. The current analyses were conducted in accordance with a protocol and statistical analysis plan developed prior to the start of data analysis.

Results

Population alignment

Prior to adjustment, the observed baseline characteristics (with the exception of refractory status, cytogenetic profile, R-ISS stage, and all plasmacytomas) were similar between the 97 treated patients in CARTITUDE-1 and the 124 patients who underwent infusion for the 300 × 10⁶ and 450 × 10⁶ CAR + T cells dose cohorts in KarMMa (Table 1). Of the two populations, CARTITUDE-1 had a greater proportion of patients who were penta-refractory (to at least two IMiDs, at least two PIs, and an anti-CD38 MoAB), whereas KarMMa had a greater proportion of patients with high-risk cytogenetics, R-ISS stage II or III, and all plasmacytomas. In the adjusted analysis, the one patient in CARTITUDE-1 who did not satisfy KarMMa’s inclusion criteria for refractoriness to last LOT was excluded. Further exclusions were not required, as all other key eligibility criteria were similar between the trials (see Supplemental Appendix Table C.1). After reweighting the 96 treated patients from CARTITUDE-1 to align with the population of interest in KarMMa, all baseline characteristics of interest were similar (Table 1). In the comparison that adjusted for the base case variables, the means of the adjusted covariates aligned exactly across the two populations, accompanied by a 46% reduction in ESS for the cilta-cel treated population. Furthermore, the sensitivity analysis which adjusted for all variables improved the overall population alignment compared to the base case; however, it reduced the ESS by 57% for the cilta-cel treated population. The distribution of weights for the base case adjusted analysis and the all variables adjusted sensitivity analysis are shown in Supplemental Appendix D.

Table 1. Differences in baseline characteristics.

	Categories	Unadjusted comparison		Base case adjusted^a comparison	All variables adjusted^b comparison (sensitivity)
		KarMMa	CARTITUDE-1	CARTITUDE-1	CARTITUDE-1
		%	%	%	%
		N = 124	N = 97	N (ESS)=96 (52)	N (ESS)=96 (42)
All Variables
Base Case Variables
Refractory status	Non triple-class refractory	16	12	16	16
	Triple or quad refractory only^c	58	45	58	58
	Penta-refractory^d	26	42	26	26
Cytogenetic profile	High-risk^e	36	24	36	36
	Standard risk/ unknown	64	76	64	64
R-ISS stage	I	12^f	34^g	12	12
	II	71^f	59^g	71	71
	III	17^f	7^g	17	17
Presence of all plasmacytomas	No	60	80	60	60
	Yes	40^h	20	40	40
Number of prior LOTs	Median	6	6	5	6
Years since MM diagnosis	Median	7	6	6	7
Age (years)	Median	61	61	62	61
Prior stem cell transplant	No	6^i	10	8	6
	Yes	94^i	90	92	94
ECOG status	0	44	40	33	44
	1	56	60	67	56
Sex	Female	42	41	42	42
	Male	58	59	58	58

^aPatients from CARTITUDE-1 who satisfied KarMMa’s eligibility criteria were adjusted for refractory status, cytogenetic profile, R-ISS stage, and all plasmacytomas in the base case analysis.

^bPatients from CARTITUDE-1 who satisfied KarMMa’s eligibility criteria were adjusted for refractory status, cytogenetic profile, R-ISS stage, all plasmacytomas, number of prior LOTs, years since MM diagnosis, age, prior stem cell transplant, ECOG status, and sex in the sensitivity analysis.

^cRefractory to two IMiDs and one PI; or two PIs and one IMiD; or two IMiDs and two PIs.

^dRefractory to at least two IMiDs, two PIs, and an anti-CD38 MoAB.

^eHigh-risk defined as at least one of del17p, t(4;14), or t(14;16).

^fPercentages are out of 121 (3 patients from KarMMa with missing R-ISS were excluded).

^gIn CARTITUDE-1, R-ISS was derived, rather than prospectively collected.

^hDefined as “paraskeletal soft-tissue masses, soft-tissue masses spreading outside the bone marrow, or both”.

ⁱAutologous stem cell transplant.

Abbreviations. ECOG, Eastern Cooperative Oncology Group; ESS, effective sample size; IMiD, immunomodulatory drug; LOTs, lines of therapy; MM, multiple myeloma; MoAB, monoclonal antibody; PI, proteasome inhibitor; R-ISS, revised International Staging System.

Source: The baseline characteristics for CARTITUDE-1 were derived using individual patient data from the February 2021 data cut. KarMMa baseline patient characteristics from 2021 NEJM publication by Munshi et al. [10].

Overall response rate and ≥ complete response rate

For both the unadjusted and adjusted comparisons, the observed response rates, and the relative treatment effect estimates (RRs and ORs) are summarized in Table 2.

Table 2. Response rates and comparative efficacy for cilta-cel versus ide-cel.

	Unadjusted comparison			Base case adjusted^a comparison			All variables adjusted^b comparison (sensitivity)
	Observed response rate		OR^c (95%CI), p-value	Adjusted response rate	OR^c (95%CI), p-value	Relative risk^d	Adjusted response rate	OR^c (95%CI), p-value	Relative risk^d
	Cilta-cel (%)	Ide-cel (%)	Cilta-cel vs Ide-cel (%)	Cilta-cel (%)	Cilta-cel vs Ide-cel	Cilta-cel vs Ide-cel	Cilta-cel (%)	Cilta-cel vs Ide-cel	Cilta-cel vs Ide-cel
Overall Response Rate	97.9	74.2	16.52 (3.85, 70.93), p = .0002	99.6	94.93 (21.86, 412.25), p < .0001	1.34	99.5	70.53 (12.23, 406.82), p < .0001	1.34
Complete Response or Better Rate	80.4	33.1	8.31 (4.45, 15.54), p < .0001	73.0	5.49 (2.47, 12.21), p < .0001	2.21	67.4	4.18 (1.86, 9.43), p = .0006.	2.04

^aPatients from CARTITUDE-1 who satisfied KarMMa’s eligibility criteria were adjusted for refractory status, cytogenetic profile, revised International Staging System stage, and all plasmacytomas in the base case analysis.

^bPatients from CARTITUDE-1 who satisfied KarMMa’s eligibility criteria were adjusted for refractory status, cytogenetic profile, revised International Staging System stage, all plasmacytomas, number of prior lines of therapy, years since multiple myeloma diagnosis, age, prior stem cell transplant, Eastern Cooperative Oncology Group status, and sex in the sensitivity analysis.

^cOdds ratios (ORs) >1 are in favor of cilta-cel versus ide-cel.

^dRelative risks are equal to the ratio of the adjusted response rates for cilta-cel versus the observed response rates for ide-cel.

Abbreviations. CI, confidence interval; ESS, effective sample size; OR, odds ratio.

Prior to adjustment, 98% of patients who received cilta-cel responded and 80% achieved ≥ CR. In comparison, 74% of patients treated with ide-cel responded and 33% achieved ≥ CR. After base case adjustment, patients treated with cilta-cel were 1.3-fold more likely to respond (OR: 94.93 [95% CI: 21.86, 412.25; p < .0001]; RR: 1.34) and 2.2-fold more likely to achieve ≥ CR (OR: 5.49 [95% CI: 2.47, 12.21; p < .0001]; RR: 2.21) compared to patients treated with ide-cel.

Duration of response

For both the unadjusted and adjusted comparisons, the relative treatment effect estimates (HRs) for DoR are summarized in Table 3.

Table 3. Estimated medians and comparative efficacy for cilta-cel versus ide-cel.

	Unadjusted comparison			Base case adjusted^a comparison		All variables adjusted^b comparison (sensitivity)
	Median, months (95% CI)		HR^c (95% CI), p-value	Median, months (95% CI)	HR^c (95% CI), p-value	Median, months (95% CI)	HR^c (95% CI), p-value
	Ide-cel	Cilta-cel	Cilta-cel vs Ide-cel	Cilta-cel	Cilta-cel vs Ide-cel	Cilta-cel	Cilta-cel vs Ide-cel
Duration of Response^d	11.1 (9.1, 11.3)	21.8 (21.8, NR)^e	0.33 (0.21, 0.52), p < .0001	21.8 (12.9, NR)^e	0.50 (0.29, 0.87), p = .0137	15.9 (12.9, NR)^e	0.58 (0.33, 1.02), p = 0.0600
Progression-free Survival	9.2 (6.0, 12.1)	22.8 (22.8, NR)^e	0.26 (0.17, 0.40), p < .0001	22.8 (14.6, NR)^e	0.37 (0.22, 0.62), p = .0002	16.8 (1.0, NR)^e	0.43 (0.25, 0.73), p = .0016
Overall Survival^f	19.5 (18.9, NR)	NR (23.6, NR)	0.39 (0.23, 0.66), p = .0004	NR (22.8, NR)	0.55 (0.29, 1.05), p = .0702	NR (22.8, NR)	0.62 (0.31, 1.23), p = .1680

^aPatients from CARTITUDE-1 who satisfied KarMMa’s eligibility criteria were adjusted for refractory status, cytogenetic profile, revised International Staging System stage, and all plasmacytomas in the base case analysis.

^bPatients from CARTITUDE-1 who satisfied KarMMa’s eligibility criteria were adjusted for refractory status, cytogenetic profile, revised International Staging System stage, all plasmacytomas, number of prior lines of therapy, years since multiple myeloma diagnosis, age, prior stem cell transplant, Eastern Cooperative Oncology Group status, and sex in the sensitivity analysis.

^cHazard ratios (HRs) <1 are in favor of cilta-cel.

^dFor KarMMa, estimates were calculated using the number of responders (N = 92).

^eMedians should be interpreted with caution, as reached when few patients were still at risk and may be an underestimate.

^fFor KarMMa, estimates were calculated using publicly available data, which excluded one patient who received a non-confirming product that did not meet product release specifications for ide-cel (N = 123).

Abbreviations: CI, confidence interval; HR, hazard ratio; NR, not reached.

Prior to adjustment, the HR for DoR for cilta-cel versus ide-cel was 0.33 (95% CI: 0.21, 0.52; p < .0001) (Figure 1). The effect of cilta-cel was statistically significantly superior to ide-cel after base case adjustment (HR: 0.50 [95% CI: 0.29, 0.87; p = .0137]).

Figure 1. Observed (unadjusted) and adjusted (base case) Kaplan–Meier plots of a duration of response. Note. Base case results adjusted for refractory status, cytogenetic profile, revised International Staging System stage, and all plasmacytomas. Abbreviations: ESS, effective sample size.

The Grambsch-Therneau test²⁰ for violation of the proportional hazards assumption was non-significant for the base case adjusted analysis (p = .12), indicating that the proportional hazards assumption was not violated.

Progression-free survival and overall survival

For both the unadjusted and adjusted comparisons, the relative treatment effect estimates (HRs) for PFS and OS are summarized in Table 3. PFS data was available for all 124 patients in KarMMa who underwent infusion for the 300 × 10⁶ and 450 × 10⁶ CAR + T cells dose cohorts, whereas OS data was only available in the public domain for 123 of these patients (the available data removed one patient who received a non-conforming product that did not meet the product release specifications for ide-cel)¹².

Prior to adjustment, the HR for cilta-cel versus ide-cel was 0.26 (95% CI: 0.17, 0.40; p < .0001) for PFS (Figure 2) and 0.39 (95% CI: 0.23, 0.66; p = .0004) for OS (Figure 3). The effect of cilta-cel for PFS was statistically significantly superior to ide-cel after base case adjustment (HR: 0.37 [95% CI: 0.22, 0.62; p = .0002]). For OS, the estimated treatment effect was in favor of cilta-cel (HR: 0.55 [95% CI: 0.29, 1.05; p = .0702]), but with widened CIs overlapping one.

Figure 2. Observed (unadjusted) and adjusted (base case) Kaplan–Meier plots of progression-free survival. Note. Base case results adjusted for refractory status, cytogenetic profile, revised International Staging System stage, and all plasmacytomas. Abbreviations: ESS, effective sample size.

Figure 3. Observed (unadjusted) and adjusted (base case) Kaplan–Meier plots of overall survival. Note. Base case results adjusted for refractory status, cytogenetic profile, revised International Staging System stage, and all plasmacytomas. Abbreviations: ESS, effective sample size.

The Grambsch-Therneau test²⁰ for violation of the proportional hazards assumption was non-significant for both outcomes in the base case adjusted analysis (PFS: p = .62; OS: p = .19), indicating that the proportional hazard assumption was not violated.

Sensitivity analyses

The sensitivity analyses that incrementally adjusted for additional baseline characteristics (i.e. number of prior LOTs, time since MM diagnosis, age, prior stem cell transplant, ECOG status, and sex) (Supplemental Appendix Table E.1) only had a limited impact on the base case comparative estimates presented in Table 2 and Table 3, and confirmed the conclusions of superior outcomes for cilta-cel versus ide-cel. Comparative efficacy estimates for the all variables adjusted comparison were as follows: ORR: OR = 70.53 (95% CI: 12.23, 406.82; p < .0001); RR = 1.34; ≥CR rate: OR = 4.18 (95% CI: 1.86, 9.43; p = .0006); RR = 2.04; DoR: HR = 0.58 (95% CI: 0.33, 1.02; p = .0600); PFS: HR = 0.43 (95% CI: 0.25, 0.73; p = .0016); and OS: HR = 0.62 (95% CI: 0.31, 1.23; p = .1680). For DoR, the results were in favor of cilta-cel but with a CI overlapping one.

Discussion

Despite ten new drug approvals in the past decade^1,7,21–24, MM remains a difficult disease to treat with no proven curative therapy. One of the unmet needs is the treatment of RRMM and specifically the patients who relapse after being triple-class exposed; having received IMiDs, PIs, and anti-CD38 MoABs. Recently three novel drug combinations have been approved for use in triple-class exposed RRMM; selinexor in combination with dexamethasone, belantamab mafodotin, and melphalan flufenamide^22–24. These agents have shown fair ORRs in refractory patients, ranging between 26% and 34% but with a limited median PFS of less than 5 months^25–27. The two novel CAR-T products compared above have shown high response rates in triple-class exposed patients; ORR of 97.9% and 74.2% for cilta-cel and ide-cel, respectively. These results are unprecedented and even more impressive are the median PFS values of 22.8 and 9.2 months for cilta-cel and ide-cel, respectively. Notably, cilta-cel has a unique BCMA CAR-T construct that includes two binding domains to confer avidity, whereas ide-cel contains a single domain for the BCMA antigen^3,10. Clinical data for these treatments suggests that not all CAR-T therapies are the same and even small differences in the construct or final drug product may lead to differences in clinical performance^28,29.

The present analysis assessed the comparative efficacy between cilta-cel and ide-cel in the absence of a direct head-to-head comparative trial. Published summary-level data from the treated population of KarMMa and IPD from the treated population of CARTITUDE-1 enabled the correction of cross-trial imbalances in patient characteristics through the conduct of unanchored MAICs. Although both CAR-T cell products have shown impressive activity in triple-class exposed RRMM patients, the results from this analysis demonstrate that cilta-cel had superior efficacy compared with ide-cel. The base case analysis adjusting for the most clinically important factors (i.e. refractory status, cytogenetic profile, R-ISS stage, and all plasmacytomas) demonstrated superior and statistically significant results in favor of cilta-cel for ORR, ≥CR rate, DoR, and PFS, compared with ide-cel. Patients treated with cilta-cel were 1.3-fold more likely to respond and 2.2-fold more likely to achieve a ≥ CR when compared with ide-cel, demonstrating a higher likelihood of a deeper response with cilta-cel. Furthermore, cilta-cel reduced the risk of disease progression or death by approximately 50% in responders and 63% in all patients. Sensitivity analyses adjusting for additional prognostic factors showed consistent statistically significant estimates for ORR, ≥CR rate, and PFS in favor of cilta-cel, showing robustness of findings from the base case analysis. For the sensitivity analysis with DoR, the results were in favor of cilta-cel and clinically meaningful but with a CI overlapping one.

In the evaluation of OS, data in the public domain was available only for 123 of the 124 patients who underwent infusion for the 300 × 10⁶ and 450 × 10⁶ CAR + T cells dose cohorts in KarMMa. In the base case adjusted comparison, cilta-cel showed a reduction in the risk of death by approximately 45% compared to ide-cel; however, a limited number of events resulted in low power for the OS outcome.

After a median follow-up duration of 18 months, median DoR and median PFS for cilta-cel were 21.8 months (95% CI: 21.8, NR) and 22.8 months (95% CI: 22.8, NR), respectively; however, these were expected to be underestimated, as the median values for both endpoints were reached at a timepoint where few patients remained at risk. Overall, these conservative estimates exceed the median DoR of 11.1 months and the median PFS of 9.2 months observed for patients treated with ide-cel. Due to the lack of longer follow-up data, the median OS was not reached for patients treated with cilta-cel, while the median OS for patients treated with ide-cel was 19.5 (95% CI: 18.9, NR) months, although this result is not fully mature.

Depth of response is an important indicator of prognosis for patients with RRMM, having been associated with improvements in both PFS and OS^30,31. ≥CR rate, a common measure of the depth of response, was assessed in the present study. The observed ≥ CR rate in CARTITUDE-1 (80.4%) far exceeded that observed for patients in the 300 × 10⁶ and 450 × 10⁶ CAR + T cells dose cohorts in KarMMa (33.1%). After population adjustment in the present study, cilta-cel demonstrated an increased likelihood of a deeper response, as well as a decreased risk of progression or death. Another measure of the depth of response is a minimal residual disease (MRD), which evaluates the presence of MM cells in the bone marrow³⁰. In CARTITUDE-1, MRD was measured for all patients with evaluable samples at day 28 and months 6, 12, 18, and 24, of whom 92% achieved MRD-negative status at a sensitivity level of 10⁻⁵, corresponding to 58% of the treated population⁹. MRD was measured in KarMMa for all patients with evaluable samples at screening and months 1, 3, 6, and 12, after which MRD was assessed at months 18 and 24 only in patients who achieved a very good partial response or better or who were MRD-negative at their prior assessment¹⁰, making cross-trial comparison with CARTITUDE-1 difficult. MRD-negative status at a sensitivity level of 10⁻⁵ was achieved in 26% of the treated KarMMa population, corresponding to 79% of patients who achieved ≥ CR and 100% of evaluable ≥ CR patient samples¹⁰.

A notable strength of this analysis was the rigorous qualitative assessment of cross-trial heterogeneity conducted a priori. This feasibility assessment evaluated the suitability of performing valid ITCs based on differences in study characteristics, patient eligibility criteria, outcome characteristics, and baseline patient characteristics between KarMMa and CARTITUDE-1. An unanchored MAIC was determined to be an appropriate approach for estimating the relative effects as it can provide valid estimates by reducing cross-trial differences in baseline patient characteristics between CARTITUDE-1 and KarMMa. Furthermore, although MAICs rely on published summary-level data from the comparator trial, the simulation of IPD from published Kaplan–Meier plots for DoR, PFS, and OS from KarMMa using the validated Guyot method^14–16 enabled the reliable reconstruction of the observed IPD. The Kaplan–Meier curves based on simulated IPD were nearly identical to the published Kaplan–Meier curves, with potential slight discrepancies in terms of the exact timings of censoring expected to have only a minimal impact on the estimated HRs.

As in any comparative analysis using non-randomized data, the potential for residual confounding after adjusting the patient population must be acknowledged. Some clinically relevant prognostic factors were not reported in KarMMa (including hemoglobin, time to progression on the last regimen, and type of MM); however, the present study was able to account for the most important factors, which were identified in close consultation with multiple independent clinical experts.

Future analyses using real-world data generated from CAR-T registries can be used to validate the findings from the current study. In addition, such studies may provide insight into the safety profile of cilta-cel versus ide-cel, which was outside of the scope of the present study. Briefly, although CAR-T therapies show substantial promise for patients with triple-class exposed RRMM, the same immune activation that provides clinical benefits has also been associated with cytokine release syndrome (CRS)³². Most CRS events in CARTITUDE-1 and KarMMa were not severe^9,10. In CARTITUDE-1, 95% of patients experienced CRS, with a median time to onset of 7 days, median duration of 4 days, and the majority of cases (95%) being grade one or two⁹. Similar results were reported in the 300 × 10⁶ and 450 × 10⁶ dose cohorts of KarMMa, wherein 85% of patients experienced CRS, with a median time to onset of 1–2 days, median duration of 4–7 days, and 79% of cases being grade one or two¹⁰.

Conclusion

Cilta-cel demonstrated clinically superior results for all outcomes studied (ORR, ≥CR rate, DoR, PFS, and OS), and these were robust across sensitivity analyses. Based on these findings, cilta-cel offers substantial clinical benefits for patients with triple-class exposed RRMM compared to ide-cel.

Transparency

Declaration of funding

This work was supported by Janssen Pharmaceuticals and Legend Biotech.

Declaration of financial/other relationships

TM is a consultant for GlaxoSmithKline and receives research funding from Janssen, Amgen, and Sanofi. SZU receives grant support from Pharmacyclics; receives grant support and consulting fees from Amgen, Bristol-Myers Squibb, Celgene, Sanofi, Seattle Genetics, Janssen, Takeda, SkylineDx, Merck, and GlaxoSmithKline; and consulting fees from AbbVie, Karyopharm, and Mundipharma. JMS, MV, WD, HT, TY, AB, SVS, JD, and SV are employed by Janssen and have restricted stock units and/or stock options. CCJ is employed by Janssen and is a consultant physician at the Memorial Sloan Kettering Cancer Center. LP and AG are employed by Legend Biotech, USA. AH, CC, and IAS are employed by EVERSANA, Canada, which was contracted by Janssen to work on this project. CC is also a shareholder of EVERSANA, Canada. A peer reviewer on this manuscript has received speaking fees from Takeda, Ono, BMS, Jansen, and Sanofi. The remaining reviewers have no other relevant financial relationships or otherwise to disclose.

Author contributions

All authors were responsible for study conception and design. JD, SVS were responsible for acquisition and analysis. AH and IAS were responsible for the initial draft of the manuscript. All authors were responsible for interpretation of data and revising it critically.

Acknowledgements

The CARTITUDE-1 study and these analyses were funded by Janssen Research & Development, LLC, and Legend Biotech, Inc. Medical writing support was provided by EVERSANA and funded by Janssen Global Services, LLC. The authors acknowledge the contribution of Eloquent Scientific Solutions for the editorial support. The authors thank Marie-Kristin Leisten for her review and comments. She was employed by Janssen during the conduct of the study. The authors thank Krista Tantakoun and Emily Rosta for their editorial assistance. Both were employed by EVERSANA, Canada.

Data availability statement

Requests for access to the CARTITUDE-1 trial study data may be submitted through the Yale Open Data Access (YODA) Project site at http://yoda.yale.edu. The data sharing policy of Janssen Pharmaceutical Companies is available at https://www.janssen.com/clinical-trials/transparency. For KarMMa study, a data sharing statement is available with the full text of the primary publication at NEJM.org¹⁰.