Julia Thornton Snider, PhD; Michelle Brauer, BS; Rebecca Kee, BA; Katharine Batt, MD, MSc; Pinar Karaca-Mandic, PhD; Jie Zhang, PhD; and Dana P. Goldman, PhD
Although there have been major advances in treatments for hematologic cancers such as pediatric acute lymphoblastic leukemia (pALL) and diffuse large B-cell lymphoma (DLBCL),1-3
efficacious treatments have historically remained limited for the population with relapsed or refractory disease.3,4
However, chimeric antigen receptor (CAR) T-cell therapies, such as tisagenlecleucel and axicabtagene ciloleucel, offer a possible cure for these patients.5-9
A recent review by the Institute for Clinical and Economic Review (ICER)10
concluded that tisagenlecleucel for pALL and axicabtagene ciloleucel for DLBCL are cost-effective treatments with incremental costs per quality-adjusted life-year (QALY) of $45,971 and $136,078, respectively.
Despite the recent approval of breakthrough therapies using CAR T cells in the United States, patients have faced barriers to treatment, including manufacturing challenges and a lack of formal coverage policies for CAR T-cell therapy (CAR T) in an inpatient setting,11,12
with delays as long as 90 days.11
Given the aggressive nature of relapsed/refractory disease, patients eligible for CAR T may have to settle for less efficacious third- or fourth-line therapies10,13
or even die while waiting for CAR T reimbursement approval.11
Cost-effectiveness analyses, like the ICER report, are useful for informing how resources may be allocated to treatments with the greatest QALY gains; however, stakeholders must consider the trade-off between treatment access today and incentivizing future treatment innovation. Social value analyses can complement cost-effectiveness analyses by shedding light on the access/innovation trade-off. Both types of analyses can inform coverage decisions, but they provide insight into different trade-offs that decision makers must weigh.
In this study, we measured the social value of treating pALL and DLBCL with CAR T in the United States and the social value lost from treatment delays as reported in the media.11,12,14-16
Social value analyses are used to quantify a therapy’s economic value from a societal perspective17
and determine the share of that value accruing to the manufacturer and patients. Expanded patient access and greater health benefits increase social value, whereas a greater requirement of society’s resources to produce the therapy (ie, higher production costs) reduces it. The higher the share of social value accruing to the manufacturer, the stronger the incentives for innovation. However, when treatment is delayed, social value is lost for both patients and manufacturers: Patients lose access to health gains from the treatment, and manufacturer profit is reduced.
An economic framework for therapy valuation was used. Specifically, we measured social value as the sum of consumer surplus and manufacturer profit.17
In the health context, consumer surplus measures the difference between the value of the health gains from a therapy and its incremental cost to the patient. It also accounts for indirect costs and benefits to patients. We calculated the economic benefit of tisagenlecleucel for pALL relative to standard of care (SOC), clofarabine monotherapy, and of axicabtagene ciloleucel for DLBCL relative to salvage chemotherapy.10
In each case, social value was estimated for 20 incident cohorts in the United States over a lifetime horizon. In each year, a new incident cohort entered the model and the existing prevalent cohorts aged an additional year. Each cohort’s survival followed that of the average patient for each treatment. We explain our calculations using tisagenlecleucel as an example; calculations for axicabtagene ciloleucel were similar, unless otherwise noted.
We obtained clinical and cost parameters from the literature and ICER’s assessment of CAR T (Table 1 [part A and part B]10,18-26
ICER reported that 400 incident cases of relapsed or refractory pALL occur annually and estimated that the average patient with pALL treated with CAR T would gain 7.9 discounted life-years (12.1 undiscounted) and 7.2 discounted QALYs (10.9 undiscounted) over SOC. Costs to the patient of tisagenlecleucel and SOC were obtained from the literature.20
Other treatment-associated costs were obtained from ICER’s report and can be found in Table 1.10,18-26
We measured patient value, also known as consumer surplus, which is the difference between how much a consumer is willing to pay for a good or service and its price. We first estimated the health value that patients obtained, calculated as the value of QALYs gained with tisagenlecleucel compared with SOC, valuing each QALY gained at $150,000 (a midrange value from the literature).18
To obtain patient value, the incremental cost of tisagenlecleucel relative to SOC was subtracted from the health value.
Next, we estimated productivity gains from tisagenlecleucel. Because the value of QALYs gained to patients includes the value of the labor and leisure they afford, some of the health value from tisagenlecleucel is attributable to productivity gains. Nationally representative data on employment and wages by age and sex from the Bureau of Labor Statistics and the US Census Bureau were used to calculate productivity gains.21,22
The manufacturer profits were calculated next. We considered a range of production costs from $100,000 to $300,000 to reflect uncertainty and likely changes in the production process over time. The midpoint value, $200,000, was taken as the base-case value for sensitivity analysis. Given uncertainty about the future price of tisagenlecleucel with loss of exclusivity and competitor entry, we simplistically assumed a 30% price reduction in 2030 based on estimates from the literature.27
Finally, total social value was calculated by summing the patient value and manufacturer profit.
We calculated social value lost from treatment delays for the first pALL cohort. We examined the first cohort rather than all 20 cohorts given uncertainty around the extent of treatment delays in the future. We assumed that patients would take the SOC treatment while waiting for tisagenlecleucel and would initiate tisagenlecleucel treatment if they survived long enough to receive it. Survival of patients treated with clofarabine monotherapy was obtained from clinical trial data.28
For the first cohort, life-years, QALYs, and productivity were calculated conditional on patients receiving treatment after 1, 2, and 6 months of delay. Incremental value lost was calculated as the difference between the value obtained by patients who were treated immediately (0-month delay) and those who experienced delays.
The steps above were repeated to measure the value of axicabtagene ciloleucel for the treatment of 20 incident cohorts of patients with DLBCL, using a clinical trial for salvage chemotherapy13
as the SOC comparator in the treatment delays analysis. Table 110,18-26
contains the parameters used in the calculations for patients with DLBCL. All health and monetary values were discounted at a rate of 3.0% and costs were inflated to 2017 US dollars. Additional detail is available in the eAppendix
(available at ajmc.com
We ran sensitivity analyses to test how consumer surplus, manufacturer profit, and social value changed by varying key model inputs. In 1-way sensitivity analyses, we adjusted the number of patients eligible for treatment, economic value of a QALY, price of CAR T, production costs, and future reduction in CAR T price individually from minimum to maximum values. We also varied life-year and QALY gains concurrently by ±50% and potential patient income by ±20%. In multiway sensitivity analyses, we conducted 1000 Monte Carlo simulations to vary each of the above parameters concurrently by selecting values of each parameter from its distribution, which measured the sensitivity of social value, manufacturer profit, and consumer surplus to the model assumptions. Each parameter was assumed to follow a beta distribution.
In the population with pALL (n = 20 × 400 = 8000), considering production costs of $100,000, $200,000, and $300,000 and a price of $475,000, we found that the total social values of tisagenlecleucel at each production cost were $6.5 billion, $5.8 billion, and $5.2 billion, respectively (Figure 1
). The value accruing to patients was $4.4 billion regardless of production costs, representing 68.9%, 76.1%, and 85.0% of total social value, respectively. This translates to 48,485 life-years, 44,010 QALYs (worth $6.6 billion), and $352.0 million in productivity (worth 5.3% of QALY gains). The remaining 15.0% to 31.1% of total social value accrued to manufacturers.
Assuming no treatment delays, patients with pALL in the first cohort gained 2872 total QALYs. The value of those QALY gains totaled $430.8 million, of which $23.0 million (5.3%) was attributable to added patient productivity from employment gains. Accounting for the cost of acquiring CAR T, the total patient value was $271.2 million and the total social value was $381.2 million. This translates to 7.2 QALYs (worth $1.1 million), $57,423 in added productivity, and a social value of $952,991 per patient.
However, with 1, 2, or 6 months of treatment delay (assuming $200,000 production costs), the first pALL cohort lost 9.8%, 36.2%, and 67.3% of social value, respectively, relative to no treatment delays. Contributing to this were losses of 311, 1146, and 2128 total life-years; 282, 1040, and 1932 total QALYs; and $2.3 million, $8.3 million, and $15.4 million in total productivity, respectively. Each patient lost 0.8, 2.9, and 5.3 life-years; 0.7, 2.6, and 4.8 QALYs; $5638, $20,796, and $38,622 in productivity (Figure 213,28
); and $93,560, $345,133, and $640,967 in social value, respectively. The loss of social value stems primarily from a high mortality rate in patients receiving SOC while awaiting treatment with tisagenlecleucel.28
In the population with DLBCL (n = 20 × 5902 = 118,040), given production costs of $100,000, $200,000, and $300,000 and a price of $373,000, the total social values of axicabtagene ciloleucel were $34.8 billion, $25.8 billion, and $16.7 billion, respectively (Figure 3
). The value accruing to patients was $13.5 billion regardless of production costs, which represents 38.7%, 52.2%, and 80.5% of total social value, respectively. This translates to gains of 372,617 life-years, 306,595 QALYs (worth $46.0 billion), and $12.5 billion in productivity (worth 27.3% of QALY gains). The remaining 19.5% to 61.3% of total social value accrued to manufacturers.
The first cohort of patients with DLBCL gained 20,008 total QALYs, assuming no treatment delays. The value of those QALY gains totaled $3.0 billion, of which $818.9 million (27.3%) was attributable to added patient productivity from employment gains. Accounting for the cost of acquiring CAR T, the total patient value was $659.5 million, and the total social value was $1.68 billion. This translates to 3.39 QALYs (worth $508,500), $138,742 in added productivity, and a social value of $284,743 per patient.
However, with 1, 2, or 6 months of treatment delay (assuming $200,000 production costs), the first DLBCL cohort lost 4.2%, 11.5%, and 46.0% of social value, respectively, relative to no treatment delays. Contributing to this were losses of 1021, 2796, and 11,185 total life-years; 840, 2301, and 9204 total QALYs; and $34.4 million, $94.2 million, and $376.7 million in total productivity, respectively. Each patient lost 0.2, 0.5, and 1.9 life-years; 0.1, 0.4, and 1.6 QALYs; and $5827, $15,955, and $63,821 in productivity (Figure 213,28
), resulting in losses of $11,959, $32,745, and $130,982 in social value, respectively.
In the pALL analysis, results of 1-way sensitivity analyses showed that social value was most sensitive to the discount rate, value of a QALY, and survival gains (Table 2
). When key parameter assumptions were varied simultaneously to test the sensitivity of the model to those inputs, the social value and patient value results were most sensitive to the discount rate, value of a QALY, and survival gains (eAppendix Figures 1
), and the manufacturer profits were most sensitive to the production costs, discount rate, and number of patients eligible for tisagenlecleucel (eAppendix Figure 3
In the DLBCL analysis, results of 1-way sensitivity analyses indicated that social value was most sensitive to the survival gains, value of a QALY, and production costs of axicabtagene ciloleucel (Table 2). These findings are similar in the multiway sensitivity analysis of social value (eAppendix Figure 4
). Meanwhile, multiway sensitivity analyses indicated that patient value was most sensitive to survival gains, value of a QALY, and discount rate, whereas manufacturer profits were most sensitive to production costs, number of patients eligible for axicabtagene ciloleucel, and discount rate (eAppendix Figures 5
Because the total social value of CAR T is determined by the survival gains and the production costs, it is expected that social value is most sensitive to the aforementioned parameters. Meanwhile, the price of CAR T, future reduction in its price, and patient income had no effect on total social value because the former 2 parameters affect only the patients’ and manufacturers’ shares of social value, whereas the latter affects only the amount of patient value attributable to productivity.
CAR T has provided the hope of a cure to patients who otherwise have limited treatment options and poor prognoses.29
Patients receiving CAR T are expected to experience meaningful improvements in life expectancy and QALYs, enabling them to contribute to overall productivity and generate social value. In both pALL and DLBCL, patients lost a substantial share of social value with treatment delays.
Various reasons have been reported for the treatment delays.11,14
One-time curative treatments such as tisagenlecleucel and axicabtagene ciloleucel present a challenge to existing payment systems because their costs accrue up front, whereas benefits accrue over a lifetime, in contrast with other cancer therapies that are administered over an extended time period. To address this challenge, novel financing mechanisms, such as an outcomes-based approach to reimbursement for tisagenlecleucel, are currently being discussed.16,30
One reimbursement approach under consideration would allow participating payers to pay for tisagenlecleucel only when patients respond within 1 month of treatment,31
allowing payers and manufacturers to share the financial risk. Additionally, aspects of the US healthcare system present challenges to outcomes-based contracts for curative therapies like CAR T. Because the average American changes health insurers every few years,32
the payers that pay the up-front costs of treatment with CAR T may not be the same ones that cover the cured individual years down the line. Thus, the payer may benefit from only a fraction of the savings, which reduces the incentive to invest in curative therapies.33
Creative solutions have been proposed to combat this “free-rider” problem.33,34
Although an outcomes-based contract developed for CAR T may help reduce payers’ risk of paying for nonresponse, the issue of up-front costs disincentivizing innovation remains.
Further, the development of formal policies to cover CAR T has been slow. Currently, reimbursement is frequently done on an individual basis,12
with hospitals facing high financial risk to treat patients with CAR T without a guarantee of payment from insurers.35
Although larger health plans are often better equipped than smaller regional plans to handle such requests, reviewing each case individually lengthens the authorization process.12
In some cases, waiting for CAR T reimbursement approval may take up to 90 days, which may be longer than a patient’s survival.11
Some payers, such as Medicare, have had success securing coverage of CAR T in the outpatient setting12
; however, challenges remain to provide sufficient reimbursement to hospitals to administer the treatment in an inpatient setting. Even with the recent approval of the new technology add-on payment of up to $186,500 per patient,36
intended to mitigate the additional costs of treatment, the reimbursement promised may fall short of the additional costs. When faced with high financial risk in the event that the costs of treating patients with CAR T exceed this payment cap, hospitals face disincentives for CAR T adoption. Thus, such policies may limit access for patients.
Additionally, CAR T is produced through a complex and individualized process37
that may be challenging to scale quickly. Efforts are currently under way to minimize delays caused by inefficiencies in production.38-40
Timely administration also necessitates that treatment centers be equipped with the proper equipment and human capital. Educating community oncologists is especially important in maximizing the efficacy and safety of CAR T, as patients are usually referred to their local oncologists for follow-up care after receiving treatment at the specified transplant centers.41
Professional organizations, such as the American Society of Hematology and the Foundation for the Accreditation of Cellular Therapy, are in the process of developing guidelines on CAR T.41,42
Our social value analysis indicates that facilitating timely patient access is a key consideration in determining an optimal financing approach. For patients with rapidly progressing cancer and high mortality rates,13,28
delaying treatment comes at a high cost. The case of CAR T provides a lesson to payers, policy makers, and innovators for incentivizing innovation and providing access to other curative therapies. In particular, therapies providing large QALY gains, such as curative therapies, bring large social value to society. Allowing innovators to share in that value incentivizes the development of future cures. However, stakeholders must work together to facilitate prompt patient access to such therapies. Efficient payment mechanisms, sufficient technological capabilities, adequate capital and human capital, and payment policy reform are required to minimize treatment delays for patients. Others have also argued that the price of CAR T should be lowered.43
These considerations are particularly important given other new or curative therapies in the pipeline, such as voretigene neparvovec-rzyl for mutation-associated retinal dystrophy,44
SPK-9001 for hemophilia,45
and LentiGlobin BB305 for sickle cell disease and beta-thalassemia.46
Our study is based upon the overall experience of patients with pALL and DLBCL and does not account for heterogeneity in patient experiences. We excluded caregiver burden from this analysis, but a reduction could be expected using CAR T-cell therapies, as they may offer patients a possible remission with fewer treatments and adverse events. Additionally, our study examined the impact of treatment delays of various lengths in only the first cohort of patients. It is uncertain how treatment delays may change in the future. Because of a lack of clinical data, we were also unable to account in our analysis for potential reductions in CAR T efficacy due to treatment delays. To the extent that delayed treatment reduces CAR T efficacy, our estimates of the social value lost because of treatment delays are conservative.
Moreover, the total cost of treatment with CAR T is not yet clear20
and may change over time. The average total costs of tisagenlecleucel used in our analysis ($736,265; obtained from the ICER report10
) included the average costs required by patients with pALL over the course of their treatment history (costs of CAR T, chemotherapy treatment, palliative chemotherapy, pretreatment, stem cell transplantation, adverse events, administration and monitoring, future healthcare, and end-of-life costs). This estimate substantially exceeded the average cost of treatment in the literature, which considered physician costs for leukapheresis and administration of lymphodepletion therapy, facilities, CAR T, drugs other than CAR T, facility fees for hospitalizations for cytokine release syndrome, and physician costs. These estimates ranged from $432,131 to $510,963 with the outcomes-based pricing arrangement.20
The average total cost of axicabtagene ciloleucel used in our analysis ($551,642) included costs accrued by patients over their treatment history (described above) and exceeded the $402,647 estimate reported by Hernandez et al in 2018.20
CAR T-cell therapies have the potential to provide significant benefit to patients with pALL and DLBCL and to society in the United States, particularly through gains in survival and productivity. However, the magnitude of benefit depends upon the ability of patients to access these treatments promptly.