Healthcare resource utilization and costs in patients with multiple myeloma administered ciltacabtagene autoleucel in outpatient versus inpatient settings after one to three prior lines of therapy

In this observational analysis of patients with RRMM treated with cilta-cel after 1–3 prior LOT in real-world clinical practice, OP administration of cilta-cel was associated with significantly lower IP days and lower imputed cost estimates compared with IP administration of cilta-cel. In the first 30 days post-infusion, OP administration of cilta-cel was associated with a mean cost savings of approximately $40,000 due to a lower number of IP days, offsetting the slight increase of $800 due to a higher number of OP days. In the first 90 days post-infusion, we found a mean monthly cost savings of approximately $18,000 in the OP cohort due to a lower number of IP days, and an increase of $500 due to a higher number of OP days. Based on these monthly cost savings, we estimated an overall cost savings of approximately $53,000 per patient with OP administration of cilta-cel over the first 3 months post-infusion relative to IP administration, with the vast majority of HCRU and associated cost savings being MM-related. These findings reinforce the potential HCRU and economic value of OP cilta-cel delivery, align with emerging evidence on OP administration of CAR-T therapy [8,9], and provide early insights into a growing trend toward OP CAR-T administration.

The observed reduction of approximately 10 IP days within 30 days post-infusion among patients with OP versus IP administration of cilta-cel may support a meaningful shift in post-cilta-cel care. The magnitude of reduction was consistent with prior single-center preliminary evaluations of OP cilta-cel administration, where the median number of IP days during the first 30 days post-infusion was 15 days shorter for OP versus IP administration (OP: 4 days; IP: 19 days) [11]. Similarly, in a retrospective claims-based analysis of patients receiving cilta-cel after ≥4 prior LOT, OP administration of cilta-cel was associated with approximately 9 fewer IP days in the first 30 days post-infusion, and a substantial proportion of patients in the OP cohort did not require IP admission during the early post-infusion period, consistent with our findings despite differences in LOT [12]. Studies of CAR-T therapies across hematologic malignancies have also shown that OP administration reduced overall early IP-related HCRU post-infusion despite the occurrence of some unplanned IP stays among patients with OP CAR-T administration, which were primarily driven by anticipated symptoms of CRS [13]. Indeed, previous studies reported a mean length of IP stays among patients with OP CAR-T administration two-to-three-times shorter than those among patients with IP CAR-T administration [13]. This difference likely reflected the structured monitoring protocols integral to OP CAR-T delivery, which emphasize proactive symptom management and selective escalation to IP care when needed, therefore resulting in a reduction in overall IP-related resource use [13].

Consistent with structured monitoring protocols, patients in the OP cohort in our study had more OP days during the first month post-infusion than those in the IP cohort. This pattern aligns with the expected timing of CAR-T-related adverse events such as CRS and ICANS, which typically occur within the first 2 weeks following infusion [5,6,16]. Our findings indicate that, in this cohort, 40.5% of patients with OP administration of cilta-cel did not require an IP admission within the first 30 days post-infusion. These results were consistent with a prior study in patients with ≥4 prior LOT, where 32% of patients treated with OP cilta-cel did not require IP admission during the first 30 days post-infusion [12].

Although costs in this analysis were extrapolated using literature-based estimates, the observed cost savings were consistent with previously reported real-world evaluations of OP and IP CAR-T administration. In our study, OP administration of cilta-cel was associated with an estimated cost savings of $40,000 per patient within the first 30 days post-infusion and $53,000 over the first 90 days. Prior cost analyses have reported total savings of $30,000–50,000 per patient in the first 6 months post-infusion with OP versus IP CAR-T administration, largely attributable to fewer IP days [13], while a recent conference proceeding reported more modest cost savings of approximately $20,000 in the first 30 days among patients receiving OP cilta-cel compared with a matched IP cohort [26]. However, the latter analysis matched patients on severity and timing of subsequent CRS and ICANS – key cost drivers – which likely attenuated post-infusion cost differences between cohorts. Accordingly, OP cilta-cel administration may help offset the upfront costs of CAR-T infusion by lowering post-infusion resource utilization and overall costs of care.

The present study builds upon previous findings in patients with ≥4 prior LOT [12] by evaluating HCRU and costs among patients who received cilta-cel after 1–3 prior LOT, a population in which cilta-cel adoption has increased following its expanded FDA approval in 2024. By focusing on earlier LOT, this study provides new insights into the potential cost efficiencies that may be achieved through implementation of OP CAR-T programs for patients with RRMM. This is particularly relevant as cilta-cel use extends to less heavily pretreated populations. Collectively, this study adds to the growing body of evidence that supports the continued expansion of OP cilta-cel programs to optimize value-based care delivery in RRMM. Shifting post-infusion monitoring care from an IP to OP setting may improve health-system efficiency, alleviate capacity constraints, and expand access to CAR-T therapy [8,9]. Institutions adopting OP administration models may also benefit from improved patient experience and faster throughput for advanced therapies [8,9].

Findings from this study should be interpreted considering some limitations. First, we relied on EMR data, which provides a more comprehensive view of patient encounters within the EMR network relative to claims data which includes claims billed for insurance purposes. However, EMR does not capture care delivered out of network and may have varying degrees of data completeness across EMR sources. To mitigate this, patients were required to have a known setting of cilta-cel administration, which may have ensured a minimum level of data completeness across patients. However, misclassification of LOT status due to incomplete capture or coding of healthcare encounters was still possible, and MM-related HCRU may have been underestimated, particularly if treatment-related complications were miscoded [16]. To mitigate the potential for bias, we additionally evaluated all-cause HCRU to capture IP and OP utilization that may not have been explicitly coded as MM-related, and findings were consistent across all-cause and MM-related HCRU. Additionally, IP utilization included admissions associated with the initial CAR-T infusion for the IP cohort; therefore, differences in total IP days partly reflect the structure of the treatment pathway itself that accounts for post-infusion monitoring of patients while hospitalized. The dataset did not capture the intent of admissions, and some OP pathways may include protocol-driven or scheduled IP monitoring; therefore, results did not distinguish between planned and unplanned care. Furthermore, cost estimates were imputed, using methodologies described in prior research on cost imputation [20], based on current literature-derived costs [21,23] and may not reflect variation in payer-paid amounts or intensity of IP and OP services, and may not have accounted for differences in costs associated with virtual or remote care. For example, IP days involving higher-intensity care may incur greater costs than routine monitoring days; per-day cost estimates may not fully capture heterogeneity within care settings and may result in over- or underestimation of absolute costs. Furthermore, follow-up time was limited, precluding assessment of long-term HCRU and costs. As this was a real-world retrospective study, patients were classified into cohorts based on observed site of administration rather than randomized assignment; thus, differences in patient characteristics and system-level factors may have influenced cohort classification. Adjusted regression analyses were conducted to account for observed differences between cohorts, but residual confounding due to unmeasured factors not captured in the EMR (e.g., institutional characteristics, practice patterns) may remain. The relatively small cohort sizes limited the extent of covariate adjustment to avoid overfitting and precluded the use of more complex methodologies (e.g., propensity score-based approaches or clustering to account for variation across centers). In addition, only limited facility-level information was available; therefore, variation in center and regional factors may not be fully captured. Future studies with larger sample sizes are warranted to more comprehensively characterize and adjust for factors influencing OP versus IP administration. Nonetheless, the consistency of our findings with prior studies across multiple data sources supports the robustness of the observed patterns.

In conclusion, in this observational study, OP administration of cilta-cel was associated with lower IP resource utilization and reduced imputed costs during the first 3 months post-infusion compared with IP administration. In this cohort, 40.5% of patients receiving OP cilta-cel did not require an IP admission in the first 30 days post-infusion. These findings reflect early experience with an evolving shift toward OP CAR-T administration and underscore the potential economic value of OP cilta-cel delivery, while highlighting an opportunity for further evaluation of OP CAR-T models in supporting high-value, patient-centered care for the eligible population with RRMM.

All authors have made substantial contributions to the conception or design of the study, or the acquisition, analysis, or interpretation of data, drafting the manuscript and revising it critically for important intellectual content, and have provided final approval of this version to be published and agree to be accountable for all aspects of the work.

This study was funded by Johnson & Johnson (NJ, USA), and Legend Biotech USA Inc. (NJ, USA).

M Janakiram reports research for Johnson & Johnson, BMS, Legend, FATE Therapeutics, and advisory board with Johnson & Johnson and BMS. S Ghosh, V Alegria, T Bixby, SP Nagar and ZP Qureshi are employees and stockholders of Johnson & Johnson. M Perciavalle is an employee and stockholder of Legend Biotech. B Emond and J Maitland are employees of Analysis Group ULC, a consulting company that has provided paid consulting services to Johnson & Johnson, which along with Legend Biotech USA Inc., funded the development and conduct of this study. DR Bakaloudi has no conflicts of interest to disclose. D Dima reports consulting for Karyopharm, Caribou, KITE and Pfizer. The authors have no other competing interests or relevant affiliations with any organization or entity with the subject matter or materials discussed in the manuscript apart from those disclosed.

Editorial and analytical support was provided by Alvi Rahman, Ph.D., Jingyi Liu, M.BI. and Alessio Palladino, B.Eng., employees of Analysis Group ULC, a consulting company that has provided paid consulting services to Johnson & Johnson, which, along with Legend Biotech USA Inc., funded the development and conduct of this study.

This study was considered exempt from review by an institutional review board as it involved only the secondary use of data that were de-identified in compliance with the Health Insurance Portability and Accountability Act (HIPAA). This research study was conducted in accordance with the principles stated in the Declaration of Helsinki.

The authors certify that this manuscript reports the original results of a real-world evidence study. No prespecified study protocol or preregistration was reported for this study. The data that support the findings of this study are available from Loopback Analytics. Restrictions apply to the availability of these data, which were used under license for this study. The data are available through requests made directly to Loopback Analytics, subject to their respective requirements for access.

This work is licensed under the Attribution-NonCommercial-NoDerivatives 4.0 Unported License. To view a copy of this license, visit https://creativecommons.org/licenses/by-nc-nd/4.0/