The Cost-Effectiveness of Axicabtagene Ciloleucel as Second-Line Therapy in Patients with Large B-Cell Lymphoma in the United States: An Economic Evaluation of the ZUMA-7 Trial

Abstract

Axicabtagene ciloleucel (axi-cel) was found to have superior clinical outcomes compared to standard of care (SOC; salvage chemoimmunotherapy, followed by high-dose therapy with autologous stem cell rescue for responders) for second-line large B-cell lymphoma (2L LBCL) in the pivotal ZUMA-7 trial. The aim of this analysis was to evaluate the cost effectiveness of using axi-cel compared to the current standard 2L LBCL therapy. A 3-state partitioned-survival model estimated the cost effectiveness and budget impact from a payer perspective in the United States. Clinical outcomes were extrapolated based on the pivotal trial. The model calculated expected quality-adjusted life years (QALYs), total costs (in United States dollars [USD], and the incremental cost-effectiveness ratio (ICER), along with the budget impact. Sensitivity and scenario analyses were performed. The proportion alive at 10 years was estimated as 48% for axi-cel and 38% for SOC; median overall survival was estimated at 59 and 24 months for axi-cel and SOC, respectively. Over a lifetime horizon, the model estimated a total of 5.56 and 7.08 QALYs for SOC and axi-cel, respectively, of which 41% and 74% were in the event-free state, respectively. Incremental QALYs and costs were 1.51 and $100,366 USD, resulting in an ICER of $66,381 USD per QALY for axi-cel versus SOC. Despite crossover to subsequent CAR T in the SOC arm, second-line CAR T use was found to improve the quality and length of life compared to SOC. Cost offsets due to subsequent CAR T use led to a limited incremental cost difference. Treatment with axi-cel is a cost-effective option that addresses an important unmet clinical need for patients with LBCL who relapse or are refractory to front-line therapy.

Large B-cell lymphoma (LBCL) is the most common subtype of non-Hodgkin lymphoma. In 2016, around 28,000 new cases were diagnosed in the United States (US) [1]. Overall, 10% to 15% have primary refractory LBCL, and an additional 20% to 25% of patients with LBCL will ultimately relapse from first-line therapy after an initial response [2]. Before chimeric antigen receptor (CAR) T-cell therapies, potentially curative treatments for patients with second-line LBCL who would receive salvage immunochemotherapy followed by high-dose chemotherapy included autologous stem cell transplantation (auto-SCT), and then allogeneic-SCT for those in whom auto-SCT fails. However, event-free survival (EFS) and overall survival (OS) for second-line LBCL patients treated with salvage chemotherapy are poor: less than 20% of SCT-intended patients are event free at 2 years, and around 30% are alive at 2 years, with a median survival of 11 months [3].

Multiple CAR T therapies have recently been approved for the treatment of third-line (3L) LBCL in the US, and in relapsed/refractory LBCL, after 2 or more prior lines, CAR T-cell therapies have shown high response rates and prolonged survival [4–6]. Recently presented long-term follow-up data from the pivotal phase 2 ZUMA-1 trial found that treatment with axicabtagene ciloleucel (axi-cel, Kite Pharma) resulted in 43% alive at 5 years, further supporting the potentially curative treatment option for many patients [7]. The comparative SCHOLAR-1 study also found that patients treated with axi-cel had a 73% reduction in the risk of death (hazard ratio [HR] = 0.27; 95% confidence interval [CI], 0.00-0.38) compared to salvage regimens [8]. Previous cost-effectiveness analyses of axi-cel in the 3L setting have found that treatment with axi-cel is associated with incremental gains in survival over chemotherapy and is a cost-effective use of healthcare resources [9,10].

The pivotal ZUMA-7 (NCT03391466) trial is the first and largest phase 3 randomized study in 2L LBCL evaluating the efficacy of a CAR T-cell therapy versus standard of care (SOC) (salvage chemotherapy followed by high-dose therapy with auto-SCT rescue for responders) [11]. At a median follow-up of 24.9 months, the primary analysis found that axi-cel demonstrated a 60.2% improvement in EFS compared with patients receiving SOC (HR = 0.398; 95% CI, 0.32 to 0.51; P< .0001) [11]. In addition, despite 56% of patients in the SOC arm receiving a subsequent CAR T-cell therapy, a trend toward improved OS was observed (HR = 0.73; 95% CI, 0.53 to 1.01; P= .03) [7]. These data supported axi-cel’s recent approval by the Food and Drug Administration in the US for the treatment of adult patients with LBCL who relapse or are refractory to first-line chemoim-munotherapy [12]. However, these benefits come at a higher upfront cost and must be examined formally in an economic evaluation. To inform healthcare decision-making, we conducted an economic evaluation of axi-cel compared to SOC, to investigate the long-term economic and humanistic implications of administering axi-cel in the 2L setting for LBCL.

MATERIAL AND METHODS

Cost-Effectiveness Model Overview

A partitioned survival model based on the patients enrolled in the ZUMA-7 trial (n = 11) was developed in Microsoft Excel in line with best practice guidelines [13]. The model follows a theoretical cohort of 2L LBCL patients over time, as they move between the mutually exclusive health states as event-free, post-event, or dead. The model estimates the proportion of the cohort in each health state at a given time based on the extrapolated survival curves and assigns costs and utilities based on the health state occupied. Reported Kaplan-Meier data from the trial informed EFS and OS inputs (OS in Supplementary Figure S1), and time to event outcomes beyond the study observation period were extrapolated using mixture cure models (MCM). Associated direct medical expenditures (costs) and quality of life with the health states were included over a lifetime time horizon. The analysis was conducted from a US third-party commercial payer perspective. Modeled outcomes include costs, life years (LYs) and quality-adjusted life years (QALYs) (all discounted at 3.0% per year) and the incremental cost effectiveness ratio (ICER). Model input values and data sources are summarized in Table 1, with additional detail provided in the below sections and in the Supplementary Materials.

Table 1.

Model Input Values, Distributions, Variations and Data Sources

Model input	Point Estimate	Distribution	Variation	Source
Health state utility values
Axi-cel, on-treatment	0.740	Beta	0.031	Roth 2018 [9]
SOC, on-treatment	0.673	Beta	0.038	Roth 2018 [9]
Pre-event, off-treatment	0.823	Beta	0.041	Sullivan 2006 [22]
Post-event	0.710	Beta	0.036	NICE (TA567) [23]
Direct costs*
Axi-cel acquisition	$399,000	N/A	$79,800	List price
Lisocabtagene maraleucel acquisition	$410,300	N/A	$82,060	RED BOOK [18]
Tisagenlecleucel acquisition	$373,000	N/A	$74,600	RED BOOK [18]
CAR T-cell administration†	$72,977	Gamma	$14,595	Liu 2021 [24]
Leukapheresis, cost	$1,173	Gamma	$235	RED BOOK [18]; CMS 2021 [19]
Bridging/conditioning chemotherapy	$5,231	N/A	N/A	RED BOOK [18]; CMS 2021 [19]
R-ICE	$7,873	N/A	N/A	RED BOOK [18]; CMS 2021 [19]
R-ESHAP	$6,950	N/A	N/A	RED BOOK [18]; CMS 2021 [19]
R-GDP	$7,035	N/A	N/A	RED BOOK [18]; CMS 2021 [19]
R-DHAP	$6,973	N/A	N/A	RED BOOK [18]; CMS 2021 [19]
HDT	$10,569	N/A	N/A	RED BOOK [18]; CMS 2021 [19]
Chemotherapy inpatient administration	$3,642	Gamma	$728	Dasta 2005 [25]
Chemotherapy outpatient administration	$1,183	Gamma	$237	CMS 2021 [19]
Stem cell harvest	$78,358	Gamma	$15,672	Pelletier 2018 [26]
Auto-SCT	$44,506	Gamma	$8901	Pelletier 2018 [26]
Auto-SCT follow-up†	$13,949	Gamma	$2790	Pelletier 2018 [26]
Allo-SCT(+)	$355,387	Gamma	$71,077	Broder 2017 [27]
Allo-SCT follow-up§	$313,495	Gamma	$62,699	Broder 2017 [27]
End of life care	$19,406	Gamma	$3881	Kutikova 2006 [28]
Pre-event resource utilization	$2,217	Gamma	$443	See Supplementary Table S4
Post-event resource utilization‖	$2,429	Gamma	$486
BIM inputs
Axi-cel market uptake¶	2%-6%	N/A	N/A	Data on file
Year 1	9%-13%
Year 2	12%-16%
Year 3	14%-18%
Year 4	16%-20%
Year 5
Incidence of LBCL	54/1,000,000	N/A	N/A	SEER [29]
Primary refractory disease	15%	N/A	N/A	Sehn 2021 [2]
Relapsed disease	25%	N/A	N/A	Sehn 2021 [2]
Proportion relapsed within 12 months	70%			Maurer 2014 [30]
Proportion intended for auto-SCT	50%	N/A	N/A	Friedberg 2011 [31]

Allo-SCT indicates allogeneic stem cell transplantation; BIM, budget impact model; CMS, Center for Medicaid Services[CT, computerized tomography; HDT, high-dose therapy; R-DHAP, rituximab, dexamethasone, cytarabine, cisplatin; R-ESHAP, rituximab, etoposide, cytarabine, cisplatin; R-GDP, rituximab, gemcitabine, dexamethasone, cisplatin; R-ICE, rituximab, ifosfamide, carboplatin, etoposide.

^*All costs reported in 2021 USD.

^†Inpatient administration costs include the management of all AEs with the exception of B-cell aplasia.

^‡Auto-SCT follow-up applied monthly up to a maximum of 24 months.

^§Only as a subsequent treatment; allo-SCT follow-up applied as a one-off cost.

^‖Post-event resource utilization reverts to $0 at 5 years.

^¶SOC is assumed to account for the remainder of the market share.

Model Structure

The partitioned survival model included three mutually exclusive health states: event-free, post-event and death (Figure 1). Sub-states were used to model time on or off treatment to account for treatment related costs and adverse events. In the event-free state, patients were on-treatment with either 2L axi-cel or SOC. Treatments in the post-event state were based on third or higher lines observed in the ZUMA-7 clinical trial [14]. Model cycle length was 1 month, and half-cycle corrections were applied.

Figure 1.

Model structure. Mutually exclusive health states are depicted by the circles. Arrows represent transitions between or within states; death is an absorbing state. Transition between states was driven by extrapolations of time to event data from the ZUMA-7 trial.

Time to Event Outcomes

EFS, time to next treatment (TTNT), and OS were fit independently and extrapolated over the lifetime horizon from the Kaplan-Meier curves for respective trial arms (Figure 2) [15]. Based on previous research on axi-cel in LBCL, MCMs were deemed appropriate to predict the shape of the curves [16]. The method for selecting the base case curves was based on best statistical fit and clinical plausibility. Additional details on model selection and MCM methods are described in the Supplementary Materials.

Figure 2.

Survival plots (A) axi-cel, (B) standard of care, and (C) modeled extrapolated survival. TTNT indicates time to next treatment.

Costs

All costs are reported in 2021 US dollars (USD) and were inflated using the Consumer Price Index in the US [17]. Costs of drugs at list price were taken from RedBook [18], current as of December 2021; costs associated with the administration of treatments were obtained from the Center for Medicaid Services Physician Fee Schedule [19].

Cost of 2L Therapies

The cost of treatment with axi-cel included acquisition cost, infusion-related costs, leukapheresis, bridging corticosteroids, and conditioning chemotherapy (Table 1). Costs for 2L therapies in the SOC arm included salvage chemotherapy, high-dose chemotherapy, and auto-SCT-related costs. These were applied as a weighted average based on the frequencies of 2L treatment received in the ZUMA-7 trial. Of the 179 patients in the SOC arm, 168 received one of the four allowed salvage chemotherapies available in the trial (Supplementary Table S1). Of the of 80/179 patients who responded to salvage chemotherapy, 64/179 received high-dose therapy, and 62/179 underwent transplant in the second-line [14]. Additional detail is provided in the Supplementary Material.

Cost of Subsequent Therapies

Distribution and costs of subsequent therapies are presented in Supplementary Table S2. Costs were applied as a one-off cost at time of initiation of the post-event subsequent treatment, based on the TTNT curve. Duration of treatment was not modeled and an average number of treatment cycles were estimated. All patients receiving a CAR T-cell therapy as a subsequent treatment were costed with the same assumptions as in the 2L (Table 1). Patients who did not have a TTNT event by 5 years were assumed to achieve longterm remission, thus not requiring subsequent treatment.

Cost of Adverse Events

Grade 3 and greater adverse events (AEs) associated with axi-cel treatment, with the exception of B-cell aplasia, were assumed to occur during the initial treatment and hospitalization period and were thus not costed separately. The incidence of grade 3 and greater AEs for the SOC arm were informed by the ZUMA-7 trial (Supplementary Table S3). Because safety data were not available for all subsequent treatments, the cost of managing treatment-related AEs for subsequent treatments, including CAR T-cell therapies, was not considered. Costs for managing AEs are presented in Supplementary Table S3.

Health State Resource Utilization

Resource utilization included annual hospitalization, physician visits, lab testing and disease management costs (Supplementary Table S4) and were assumed to be equivalent for axi-cel and SOC (Supplementary Table S4) [14]. A one-time cost for end of life care was assigned to patients upon death. It was assumed that no LBCL-related resource use was incurred for those patients who remained event free after 5 years.

Health-Related Quality of Life

Utility values for on-treatment, pre-event off-treatment, and post-event are presented in Table 1 with additional detail in the Supplementary Materials. It was assumed that utility values revert back to that of the age-matched general population for patients who remain event-free for 5 years.

Sensitivity and Scenario Analyses

Input uncertainty was evaluated through both one-way and probabilistic sensitivity analyses. One-way sensitivity analyses, where costs and utilities were varied by ±20%, are presented as tornado diagrams, displaying the most influential parameters on model results. Probabilistic sensitivity analyses, where all parameters are simultaneously varied across predetermined distributions (gamma for costs, beta for bounded estimates), are presented as a cost-effectiveness plane. The probabilistic sensitivity analysis was run until model convergence. Alternate distributions of time-to-event data were explored in scenario analyses, including more optimistic and more pessimistic curves. A scenario analysis specific to subsequent treatments used from participating US clinical sites in the ZUMA-7 trial was performed. Subsequent treatment data were reanalyzed based on treatments used only by US patients to account for potential impacts on costs (Supplementary Table S5).

Budget Impact Model

A budget impact model was built using the cost-effectiveness model engine and included the cost of 2L treatments, adverse events in the 2L, as well as 3L treatment-related costs. The budget impact was calculated based on the difference of a future practice where CAR T-cell therapy is included as a treatment option for patients in the 2L versus the current practice where no CAR T-cell therapy is included in the 2L. Epidemiological and projected treatment pattern inputs are presented in Table 1, with additional detail in the Supplementary Material.

RESULTS

Cost-Effectiveness Analysis

In the base case analysis, the model estimated 5-year EFS to be 36.1% and 15.4% for axi-cel and SOC, respectively. The 10-year OS in the model was estimated as 46% for axi-cel and 38% for SOC (Figure 2) with a median OS of 59 and 24 months for axi-cel and SOC, respectively. Based on results from the MCM fitting to the OS curve, 53% and 42% of axi-cel and SOC patients, respectively, were considered long-term responders and of patients alive at 10 years, patients treated with axi-cel were 94% more likely to be event-free. Total discounted treatment-related costs (including both 2L and subsequent treatment) were $546,704 USD for axi-cel and $440,099 USD for SOC (Supplementary Table S6). Costs for subsequent therapy were higher among those receiving SOC in the 2L (Supplementary Figure S2). Among patients receiving 2L SOC, the highest proportion of subsequent treatment costs was for CAR-T therapies; among patients receiving 2L axi-cel, the highest proportion of subsequent treatment costs were for SCT. Further, use of CAR-T therapies were a primary constituent of costs in every line of therapy following SOC in the 2L (Supplementary Figure S3). Over the lifetime time horizon, incremental total costs were $100,366 USD and the model estimated 7.08 and 5.56 discounted QALYs for axi-cel and SOC, respectively, for a difference of 1.51. Treatment with axi-cel resulted in an ICER of $66,381 USD per QALY gained (Table 2).

Table 2.

Cost-Effectiveness Results (Discounted) in Base Case, 2021 USD

	AXI-CEL	SOC	Difference
Life years	9.14	7.59	1.55
Event-free	6.48	2.86	3.63
Post-event	2.66	4.73	−2.07
Quality-adjusted life years	7.08	5.56	1.55
Event-free	5.23	2.28	2.95
Post-event	1.84	3.28	−1.44
Total costs	$635,794	$535,428	$100,366
Second-line treatment-related	$449,786	$95,319	$354,467
Subsequent treatment-related	$96,917	$344,779	−$247,862
Disease management*	$85,658	$77,282	$8376
ICER, axi-cel versus SOC			$66,381

^*Includes terminal care.

One-way deterministic sensitivity analyses found that the ICER was most sensitive to the axi-cel treatment cost (varied by ±20%), subsequent treatment patterns in the SOC arm, the number of inpatient days for axi-cel administration and post-event utilities (Supplementary Figure S4). The probabilistic sensitivity analysis found that, at a willingness to pay threshold of $150,000 USD per QALY, axi-cel was cost-effective versus SOC in 75% of the simulations (Supplementary Figures S5, S6).

When the gamma curve (a less favorable but still clinically plausible extrapolation) was selected for the extrapolation of axi-cel OS, the model estimated 10-year OS was 44.9%, and the ICER increased to $73,503 USD per QALY (Supplementary Table S7). The Gompertz curve (a more optimistic) and loglogistic curve (less optimistic for efficacy with a cure fraction of 44%) resulted in ICERs of $60,709 USD and $118,171 USD per QALY, respectively. When only considering US clinical practice treatment patterns, incremental costs of axi-cel versus SOC were reduced to $91,374 USD and axi-cel remained cost-effective with an ICER of $60,434 USD per QALY.

Budget Impact Analysis

It was estimated that 9 patients in a million member plan would be eligible for axi-cel treatment as 2L LBCL patients. The introduction of axi-cel to 2L treatment of patients with LBCL leads to a total 5-year cumulative budget impact ranging of $0.03 USD per member per month (PMPM) for the lower bound to $0.07 USD PMPM for the upper bound (Supplementary Table S8).

DISCUSSION

Patients with LBCL in whom a front-line therapy has failed have few curative intent treatment options, with associated limited long-term remission and prolonged survival. This analysis suggests increased overall survival from treatment with axi-cel as observed with 10-year estimated OS of 46.4% and 37.5% for axi-cel and for SoC, respectively, and a higher proportion of long-term responders for patients treated with axicel compared to SoC (53% versus 43%). This finding is consistent with the higher proportion of patients that received a potentially curative treatment option in the axi-cel arm of the trial and is reflected by the increased incremental QALYs of treatment with axi-cel. Furthermore, treatment with axi-cel in the second-line resulted in a budget impact of ~$0.01 PMPM due to the small population of patients eligible for treatment and cost-offsets of subsequent treatment.

This analysis estimated incremental costs were $103,808 USD higher with axi-cel; however, there were important offsets compared to SOC. Although 2L treatment costs were lower for the SOC arm due to the low cost of salvage chemotherapy, subsequent treatment costs represented nearly 80% of the total treatment-related costs for the SOC arm, including $266,778 USD from CAR T-cell therapy use as a subsequent treatment. Additionally, the higher proportion of long-term responders after treatment with axi-cel limited the disease management costs in the post-event health state, further reducing the difference in cost between arms. As with the cost-effectiveness analysis, as 2L CAR T uptake was increased in the budget impact analysis, the costs associated with CAR T therapies as a subsequent treatment decreased.

We found the ICER to be $66,381 USD per QALY gained, suggesting that 2L treatment with axi-cel was highly cost-effective versus SOC in the US. These results were maintained over a wide range of sensitivity and scenario analyses. Given uncertainty around long-term survival, when a more conservative yet still clinically plausible curve (i.e., demonstrating a benefit for axi-cel treatment vs. SoC) was used to extrapolate OS, the ICER increased to $73,503 USD per QALY, still below the accepted threshold for cost-effectiveness in the US [20]. Sensitivity analyses found that the biggest cost drivers of the cost-effectiveness and budget impact analysis were subsequent treatment patterns in the SOC arm, including the use of CART-cell therapies and the cost of auto-SCT.

The increase in length and quality of life from treatment with axi-cel resulted in incremental QALYs of 1.51, a large and meaningful improvement. Furthermore, 74% of the QALYs were in the EFS state for the axi-cel cohort compared to 41% for SoC, suggesting that patients treated with axi-cel spend a greater proportion of their lives with the higher quality-of-life benefit of EFS. These gains in efficacy estimated in this economic analysis align with what was observed in the ZUMA-7 trial, including prolonged EFS, contributing to incremental QALYs. Although the utilities in this economic model were taken from the literature, as observed in the ZUMA-7 trial, treatment with axi-cel results in a quicker return to baseline in quality of life compared to SOC [21]. In addition to the benefit observed in EFS health state, a trend of improved OS was observed. This trend is likely influenced by the receipt of CAR T therapies outside the protocol in the ZUMA-7 trial, which may have confounded the OS analysis [11]. Notwithstanding, the benefit of treatment with axi-cel translates into important incremental long-term benefits for patients.

The ZUMA-7 trial data used to conduct this economic evaluation is the first randomized controlled trial to compare a CAR T-cell therapy versus SOC, with the largest sample size to date and longest duration of follow-up (median 24.9 months). Given that both CAR T and auto-SCT are potentially curative and require upfront costs, this cost-effectiveness analysis may support clinical decisions for determining treatment pathways and timing for treatments with curative intent. The ZUMA-7 trial suggests that 64% of transplant-intended patients did not receive auto-SCT; furthermore, at least 56% of patients in the SoC arm received a subsequent treatment with curative intent. This results in both increased overall costs and a higher risk of death before patients can receive 3L therapy.

This economic analysis has limitations. Although all available data from the trial were used to inform the economic models, the cost of managing adverse events and related disutilities for subsequent treatments were not included in this economic analysis. Thus the ICER presented could be conservative, as the impact on patients due to delayed or reduced need for subsequent treatments was not accounted for. The OS curves selected for extrapolation were based on statistical criteria and clinical plausibility, however, as with many economic analyses, survival benefits may be driven by the selection in curve. Scenario analyses where survival along with other parameters extrapolations were varied found that treatment with axi-cel remained cost-effective in the majority of simulations. It should also be noted that the use of CAR T therapies as a subsequent treatment in the model were based on observed use in the ZUMA-7 trial and may not be reflective of real-world treatment patterns, which would vary due to logistical, socio-economic, or provider considerations.

CONCLUSIONS

This economic analysis compared 2L axi-cel versus SOC, where we found that 2L axi-cel use improved survival and increased time without disease progression, resulting in a substantial gain in QALYs. Furthermore, 2L CAR T has limited financial impact due to the offset of subsequent CAR T use and reduced disease progression costs. Treatment with axi-cel for patients with LBCL who relapse or are refractory to front-line therapy is cost-effective compared to a less dependable strategy of SOC, where despite completing therapy, a minority of patients make it to definitive therapy in 2L and many patients proceed to 3L CAR-T.