Abstract
Importance
Among children and young adults with relapsed or refractory B-cell acute lymphoblastic leukemia, the rate of 5-year disease-free survival is 10% to 20%. Approval of tisagenlecleucel, a chimeric antigen receptor T-cell therapy, represents a new and potentially curative treatment option. However, tisagenlecleucel is expensive, with a current list price of $475 000 per one-time administration.
Objective
To estimate the long-term survival and value of tisagenlecleucel for children and young adults with B-cell acute lymphoblastic leukemia.
Design, Setting, and Participants
In this cost-effectiveness analysis, a decision analytic model was designed to extrapolate trial evidence to a patient lifetime horizon. The survival evidence for the model was extracted from 3 studies: B2202 (enrolled patients from April 8, 2015, to November 23, 2016), B2205J (enrolled patients from August 14, 2014, to February 1, 2016), and B2101J (enrolled patients from March 15, 2012, to November 30, 2015). Long-term survival and outcomes of patients younger than 25 years with B-cell acute lymphoblastic leukemia that is refractory or in second or later relapse were derived using flexible parametric modeling from the direct extrapolation of event-free survival and overall survival curves. The published Kaplan-Meier curves were digitized from November 1, 2017, to November 30, 2017, using an algorithm to impute patient-level time-to-event data. Sensitivity and scenario analyses assessed uncertainty in the evidence and model assumptions to further bound the range of cost-effectiveness. Data were analyzed from December 1, 2017, to March 31, 2018.
Interventions
The primary intervention of interest was tisagenlecleucel. The comparator of interest was the chemoimmunotherapeutic agent clofarabine.
Main Outcomes and Measures
Model outcomes included life-years gained, quality-adjusted life-years (QALYs) gained, and incremental costs per life-year and QALY gained.
Results
Forty percent of patients initiating treatment with tisagenlecleucel are expected to be long-term survivors, or alive and responding to treatment after 5 years. Tisagenlecleucel had a total discounted cost of $667 000, with discounted life-years gained of 10.34 years and 9.28 QALYs gained. The clofarabine comparator had a total discounted cost of approximately $337 000, with discounted life-years gained of 2.43 years and 2.10 QALYs gained. This difference resulted in an incremental cost-effectiveness ratio of approximately $42 000 per life-year gained and approximately $46 000 per QALY gained for tisagenlecleucel vs clofarabine. These results were robust to probabilistic sensitivity analyses. Across scenario analyses that included more conservative assumptions regarding long-term relapse and survival, the incremental cost-effectiveness ratio ranged from $37 000 to $78 000 per QALY gained.
Conclusions and Relevance
Tisagenlecleucel likely provides gains in survival and seems to be priced in alignment with these benefits. This study suggests that payers and innovators should develop novel payment models that reduce the risk and uncertainty around long-term value and provide safeguards to ensure high-value care.
This cost-effectiveness analysis uses a decision analytic model to extrapolate trial evidence to estimate the long-term survival and value of tisagenlecleucel for children and young adults with relapsed or refractory B-cell acute lymphoblastic leukemia.
Key Points
Question
What is the long-term survival and value of tisagenlecleucel for pediatric patients with relapsed or refractory leukemia?
Findings
In this decision-analytic modeling study using deidentified data, cost-effectiveness analysis generated an incremental cost-effectiveness ratio between $37 000 and $78 000 per quality-adjusted life-year gained over a patient lifetime horizon, with more than 40% of those initiating tisagenlecleucel treatment becoming long-term survivors.
Meaning
Tisagenlecleucel seems to be priced in alignment with its clinical benefits, although its value depends on assumptions around long-term survival.
Introduction
B-cell acute lymphoblastic leukemia (B-ALL) is the most common malignant neoplasm diagnosed in children, accounting for 20% to 30% of the overall incidence of childhood cancer.1,2 Owing to its presentation as a systemic disease, treatment for B-ALL has been based primarily on chemotherapy.3 However, among patients with relapsed or refractory B-ALL, rates of 5-year disease-free survival are only 10% to 20%.4,5,6 The recent approval of chimeric antigen receptor T-cell (CAR-T) therapies represents a new and potentially curative treatment option for B-ALL.7 Tisagenlecleucel, the first CAR-T therapy approved, is indicated for use in pediatric patients with relapsed or refractory B-ALL.7 Evidence for tisagenlecleucel demonstrated higher rates of response, event-free survival, and overall survival than other therapies used in pediatric patients with B-ALL who were heavily pretreated.8 However, follow-up for patients receiving tisagenlecleucel is limited, with a maximum duration of less than 4 years; therefore, uncertainty remains around its long-term benefit.
Tisagenlecleucel is expensive, currently listed at $475 000 per a one-time administration.9 Although the manufacturer has discussed an outcomes-based payment agreement, in which only patients who respond to treatment will be required to pay,10 the list price and payment agreement do not include additional costs, such as hospital markup or health care use for preparation, administration, management, or adverse events. The objective of this analysis was to estimate the long-term survival and value of tisagenlecleucel for pediatric patients with B-ALL.
Methods
Study Design
Because this study model uses already available evidence from studies, the Colorado Multiple Institutional Review Board waived approval for this study and for patient consents. A decision analytic model extrapolated trial evidence from 3 studies11 (B2202, enrolled patients from April 8, 2015, to November 23, 2016; B2205J, enrolled patients from August 14, 2014, to February 1, 2016; and B2101J, enrolled patients from March 15, 2012, to November 30, 2015) and aggregated all costs and outcomes expected from tisagenlecleucel. The analysis was from a payer perspective and estimated outcomes over a patient lifetime horizon. Data were analyzed from December 1, 2017, to March 31, 2018. The decision analytic model included a short-term decision tree and a long-term semi-Markov partitioned survival model (Figure 1). The decision tree tracked a patient from initiation of treatment through assessment of response. Initiation of treatment was to undergo leukapheresis in preparation for tisagenlecleucel infusion. At the first decision tree event node, patients had the following 3 options: continue with the tisagenlecleucel infusion after undergoing leukapheresis, discontinue before infusion because of adverse events (AEs) or manufacturing failure, or die before receiving the infusion. Those who discontinued the CAR-T arm owing to AEs were assumed to not be able to tolerate other active therapies and therefore transitioned to receive no further antileukemic therapy (ie, palliative care only). Those who discontinued the CAR-T arm owing to manufacturing failure were assumed to receive clofarabine. The second event node of the decision tree assessed a patient’s response to treatment, which had the 3 following options: alive and responding to treatment, alive and not responding to treatment, or dead before assessment of response. The model allowed for patients to receive or not receive stem cell transplantation (third event node of the decision tree) after assessment of response.
Figure 1. Model Framework.
The 2-part model included a short-term decision tree and a long-term Markov model. CAR-T indicates chimeric antigen receptor T-cell therapies; SCT, stem cell transplantation.
aPatients who discontinue owing to manufacturing failure will transition to the active comparator. Patients who discontinue owing to adverse event will receive no further antileukemic or antilymphomic therapy and will enter the semi-Markov partitioned survival model at M2.
bPatients who transition from M1 to M2 will receive palliative therapy.
After the decision tree, patients moved to the semi-Markov partitioned survival model, where they transitioned between the following 3 health states: alive and responding to treatment, alive and not responding to treatment, and dead. Patient movement through each of the health states was estimated from the direct extrapolation of event-free survival and overall survival curves. The published Kaplan-Meier curves were digitized from November 1, 2017, to November 30, 2017, using the algorithm by Guyot and colleagues12 to impute patient-level time-to-event data. The fitted model curves included the distributional forms of Weibull, exponential, log-normal, log-logistic, and Gompertz. The base-case distributional form was analyzed between December 1, 2017, and December 31, 2017, and was selected separately for each curve based on best model fit using Akaike information criterion values, visual comparison with Kaplan-Meier curves, and plausibility of long-term extrapolation. The eAppendix in the Supplement provides more details regarding the study design.
Flattening in the tail of the survival curves was observed for both tisagenlecleucel and clofarabine. Standard parametric models likely underestimate survival when flattening in the tail exists; therefore, we used a flexible parametric model to account for this flattening.13 We first fit a parametric curve function to the portion of the curve that was downward sloping. After the fitted curve intersected the flat portion (ie, slope of 0) of the Kaplan-Meier curve, a knot was introduced. After the knot, mortality occurred only owing to all-cause mortality. This flexible parametric approach extrapolated the published survival curves to 5 years, after which patients who were alive and responding to treatment were considered to be long-term survivors and effectively cured.14 Accordingly, patients who were alive and not responding to treatment died within 5 years of treatment completion. Mortality after 5 years was based on the age- and sex-adjusted all-cause risks of mortality.14 Modifications for excess disease-related mortality for long-term survivors was modeled by applying a standardized mortality ratio of 9.1 to the mortality risk.15 Model outcomes included patient survival, quality-adjusted survival, and health care costs. A full description of model assumptions, along with their rationale, can be found in eTable 1 in the Supplement.
Intervention and Comparator
The modeled cohort included patients younger than 25 years with B-ALL that was refractory or in second or later relapse. The intervention of interest was tisagenlecleucel. Evidence for tisagenlecleucel was abstracted from single-arm trials, which resulted in challenges selecting the most appropriate comparator therapy. Because a matching-adjusted indirect comparison was not possible owing to the lack of patient-level data and the available aggregated evidence being from single-arm trials,16 we chose the comparator with evidence from the most similar patient population. We engaged stakeholders and clinical experts in relapsed or refractory leukemia to inform our decision and considered various comparisons, including clofarabine monotherapy, clofarabine combination therapy, and blinatumomab. Clofarabine monotherapy was chosen as the comparator because the trial populations for both tisagenlecleucel and clofarabine monotherapy were most similar in relation to demographics and disease severity. eTable 2 in the Supplement describes the population characteristics for those receiving tisagenlecleucel and clofarabine. eTable 3 in the Supplement details the treatment regimen for the intervention and comparator. Furthermore, both clofarabine and tisagenlecleucel are indicated by the US Food and Drug Administration for relapsed or refractory acute lymphoblastic leukemia after at least 2 prior regimens. Blinatumomab was not selected as the comparator owing to the heterogeneity in the patient populations (eg, age, bone marrow blast level, and prior lines of therapy) in the evidence for tisagenlecleucel and blinatumomab.
Model Inputs
Clinical inputs included survival, AEs, and quality of life. The clinical inputs that informed the model are detailed in Table 1.8,14,17,18,19,20,21,22,23,24 Further detail on clinical model inputs can be found in eTables 4-11 in the Supplement. Economic inputs included the treatment acquisition cost, potential hospital markup, and health care use. Payment for tisagenlecleucel was assumed only for responders at 1 month, based on expected agreements between the manufacturer and payers. Both the tisagenlecleucel and clofarabine treatment arms assumed inpatient administration and were thus subject to potential hospital markup. Because the contracts between hospitals and payers related to hospital markup are still being developed for tisagenlecleucel and are confidential, we engaged stakeholders and CAR-T experts to inform the hospital markup for tisagenlecleucel. The assumed markup was capped at $100 000 based on comments from stakeholders and CAR-T experts that some facilities may not negotiate a markup while other facilities may charge a markup. Costs associated with health care use resulting from administration and monitoring were included in the model. The cost of a hospitalization for treatment administration was estimated using a fee-for-service approach, which included a per diem cost for hospital days and added costs of therapies administered during the hospitalization. Additional costs were included for AEs that were expected to prolong hospitalization (cytokine release syndrome) or extend beyond discharge (ie, B-cell aplasia). Future related and unrelated medical costs were included for long-term survivors. All costs were adjusted to 2017 US dollars. A full description of the economic inputs that informed the model can be found in eTables 12-14 in the Supplement.
Table 1. Clinical and Economic Inputs.
| Characteristic | Tisagenlecleucel | Clofarabine |
|---|---|---|
| Clinical Input | ||
| Response | ||
| Percentage of patients achieving response (complete or partial) | 84.48 | 30.017 |
| Percentage of patients dead before assessment of response | 7.48 | 25.017 |
| Percentage of patients achieving no response | 8.28 | 45.017 |
| Stem cell transplantation | ||
| Percentage of patients who receive stem cell transplantation | 10.5 | 14.8 |
| Adverse events | ||
| Cytokine release syndrome (grade III or higher) | 49 | 0 |
| Hypogammaglobulinemia (B-cell aplasia) | 43 | 0 |
| Utilities | ||
| Alive and not responding to treatment | 0.7518 | 0.7518 |
| Alive and responding to treatment | 0.9118 | 0.9118 |
| Long-term survivor—alive | 0.9118 | 0.9118 |
| Cytokine release syndrome (grade III or higher) | 014 for 8 d19 | 014 for 8 d19 |
| Disutilities | ||
| Chemotherapy | −0.42 for duration of treatment20 | –0.42 for duration of treatment20 |
| Stem cell transplantation | −0.57 for duration of treatment20 | −0.57 for duration of treatment20 |
| Economic Inputs, 2017 US $ | ||
| Treatment acquisition cost (with markup)$ | ||
| Tisagenlecleucel | 575 000 | 575 000 |
| Clofarabine | 264 per mg/mL21 | 264 per mg/mL21 |
| Methotrexate | 0.09 per mg/mL21 | 0.09 per mg/mL21 |
| Fludarabine | 3.70 per mg/mL21 | 3.70 per mg/mL21 |
| Cyclophosphamide | 0.74 per mg/mL21 | 0.74 per mg/mL21 |
| Cytarabine | 0.02 per mg/mL21 | 0.02 per mg/mL21 |
| Etoposide | 0.09 per mg/mL21 | 0.09 per mg/mL21 |
| Tocilizumab | 7.69 per mg/mL21 | 7.69 per mg/mL21 |
| Intravenous immunoglobulin | 0.14 per mg/mL21 | 0.14 per mg/mL21 |
| Healthcare use costs, $ | ||
| Cost per hospital day | 404922 | 404922 |
| Cost per day in intensive care unit | 404922 | 404922 |
| Office visit | 7423 | 7423 |
| Intravenous treatment administration (first hour) | 14023 | 14023 |
| Intravenous treatment administration (each additional hour) | 2923 | 2923 |
| Intravenous treatment administration (each additional sequence or drug) | 6623 | 6623 |
| Hematology panel | 1123 | 1123 |
| Liver function test | 823 | 823 |
Sensitivity Analyses
A probabilistic sensitivity analysis was conducted to assess variation in all model inputs simultaneously. Five thousand iterations were run and the percentage of all iterations that fell beneath common value thresholds of $50 000, $100 000, and $150 000 per quality-adjusted life-year (QALY) gained were computed.
Scenario Analyses
Previous literature has noted unique characteristics of decision analytic modeling for potentially curative therapies in the pediatric population, such as the discount rate, duration of benefit observed in trials, point of treatment initiation, and inclusion of future health care costs.25 For each of these characteristics, we conducted a scenario analysis. The resulting values from these scenario analyses served to bound the range of cost-effectiveness.
Discount Rate
In the United Kingdom, the National Institute for Health and Care Excellence recommends a discount rate of 1.5%, compared with the typical 3.5% discount rate, when an intervention is expected to result in substantial improvement in health over a long period of time.26 In the United States, there has been no recommendation on differential discounting for therapies with long-term and significant health improvement. In our base-case analysis, we used a discount rate of 3%, based on recommendations from the Second Panel on Cost-Effectiveness in Health and Medicine.27 However, in this scenario analysis, we used a discount rate of 1.5% for both costs and outcomes.
Duration of Trial Benefit
When tisagenlecleucel was approved by the US Food and Drug Administration, the median follow-up of Study B2101J, the trial with the longest duration, was only 18.6 months.11 Therefore, our decision analytic model required extrapolation of trial results. We tested the assumptions made through extrapolation in 2 scenario analyses. First, we estimated the number of years the benefit observed in the trial would need to continue to produce an incremental cost-effectiveness ratio beneath commonly cited value thresholds. Second, instead of using a flexible parametric model that accounted for the flattening observed in the Kaplan-Meier curves, we used standard parametric modeling to serve as a lower bound of survival and a conservative estimate of the long-term value.
Treatment Initiation
With tisagenlecleucel, patients initiate treatment by undergoing leukapheresis. However, patients who undergo leukapheresis may never receive the infusion owing to AEs, manufacturing failures, or death. In our base-case analysis, we started our model at leukapheresis; this approach is similar to an intent-to-treat analysis for a randomized clinical trial. In this scenario analysis, we started the model at infusion instead.
Future Health Care Costs
Criticism of models that include future unrelated health care costs is common, despite recommendations by the Second Panel on Cost-Effectiveness in Health and Medicine to include both future related and unrelated health care costs.27 Our base-case analysis included future related and unrelated health care costs for long-term survivors; however, in this scenario analysis, these future health care costs were excluded.
Results
Base Case
The total discounted costs over the patient lifetime horizon as well as the total undiscounted and discounted life-years (LYs) and QALYs are detailed in Table 2. Tisagenlecleucel had a total discounted cost of approximately $667 000, with discounted LYs gained of 10.34 years and 9.28 QALYs gained. A total of 42.6% of patients who underwent leukapheresis were considered to be long-term survivors. Clofarabine had a total discounted cost of approximately $337 000, with discounted life-years gained of 2.43 years and 2.10 QALYs gained. A total of 10.8% of patients receiving clofarabine treatment were considered to be long-term survivors. The incremental cost-effectiveness ratio between tisagenlecleucel and clofarabine was approximately $46 000 per QALY gained.
Table 2. Base-Case Results.
| Characteristic | Tisagenlecleucel | Clofarabine |
|---|---|---|
| Outcomes | ||
| Discounted life-years | 10.34 | 2.43 |
| Undiscounted life-years | 18.08 | 4.00 |
| Discounted QALYs | 9.28 | 2.10 |
| Undiscounted QALYs | 16.47 | 3.53 |
| Long-term survivors, % | 42.6 | 10.8 |
| Costs, 2017 US $ | ||
| Tisagenlecleucel | 405 490 | 0 |
| Clofarabine | 15 309 | 163 686 |
| Palliative chemotherapy | 2648 | 3973 |
| Pretreatment | 2979 | 0 |
| Stem cell transplantation | 47 744 | 64 648 |
| Adverse events | 33 534 | 0 |
| Administration and monitoring | 111 548 | 93 032 |
| Future health care costs | 45 901 | 9069 |
| End of life | 1602 | 2848 |
| Total | 666 754 | 337 256 |
| Incremental results | ||
| Incremental costs, $ | 329 498 | 329 498 |
| Incremental life-years | 7.91 | 7.91 |
| Incremental QALYs | 7.18 | 7.18 |
| ICER per life-year gained, $ | 41 642 | 41 642 |
| ICER per QALY gained, $ | 45 871 | 45 871 |
Abbreviations: ICER, incremental cost-effectiveness ratio; QALYs, quality-adjusted life-years.
Sensitivity Analysis
The mean incremental cost-effectiveness ratio from the probabilistic sensitivity analysis was $43 416 per QALY gained. All iterations were below a threshold of $100 000 per QALY gained; however, these sensitivity analyses did not account for uncertainty around survival curve parameters.
Scenario Analyses
Table 3 presents the results of each scenario analysis as compared with the base-case estimates. The cost-effectiveness estimates ranged from $37 000 to $77 500 per QALY gained across the scenario analyses.
Table 3. Scenario Analysis Results.
| Characteristic | Base-Case Assumption | Scenario Analysis Assumption | Incremental Costs, 2017 US $ | Incremental QALYs | Scenario Analysis ICER, 2017 US $ |
|---|---|---|---|---|---|
| Discount rate | Discount rate of 3% | Discount rate of 1.5% | 347 500 | 9.4 | 37 000 per QALY |
| Survival curve fitting | Flexible parametric modeling | Standard parametric modeling | 310 000 | 4.0 | 77 500 per QALY |
| Model start | Leukapheresis | Infusion | 454 900 | 9.1 | 50 000 per QALY |
| Future health care costs | Included | Not included | 292 700 | 7.2 | 40 700 per QALY |
Abbreviations: ICER, Incremental cost-effectiveness ratio; QALYs, quality-adjusted life-years.
Discount Rate
From the base-case analysis, the costs did not change, but the LYs and QALYs increased owing to the smaller discount rate. This change increased the incremental difference in health outcomes and resulted in a more favorable cost-effectiveness ratio of $37 000 per QALY gained (Table 3).
Duration of Trial Benefit
The trial-specific benefit would need to continue for at least 7 years after infusion to produce an incremental cost-effectiveness ratio less than $150 000 per QALY gained, for 12 years to be less than $100 000 per QALY gained, and for 36 years to be less than $50 000 per QALY gained. Figure 2 shows the cost-effectiveness estimate over all time horizons, from 1 year to lifetime. Furthermore, using standard parametric modeling, the LY for tisagenlecleucel decreased to 5.15 years and the QALYs for tisagenlecleucel decreased to 4.49 years. For clofarabine, the LYs decreased to 0.66 years and the QALYs decreased to 0.49 years. This change increased the incremental cost-effectiveness ratio to $77 500 per QALY gained (Table 3).
Figure 2. Incremental Cost-effectiveness Ratio (ICER) by Time Horizon.
The incremental cost-effectiveness ratio fell below $150 000 per QALY gained when the model time horizon was 7 years or longer. LY indicates life-year and QALY, quality-adjusted life-year. Cost-effectiveness was in 2017 US dollars.
Treatment Initiation
When the model started at treatment infusion, the incremental cost-effectiveness ratio slightly increased. Starting at leukapheresis, the incremental costs were $329 498 and incremental QALYs were 7.2 years. This change resulted in an incremental cost-effectiveness ratio of $45 871. When the model started at infusion, the incremental costs increased to $454 892 and incremental QALYs increased to 9.1 years. This change slightly increased the incremental cost-effectiveness ratio to $50 000 per QALY gained (Table 3).
Future Health Care Costs
In the final scenario analysis, we removed future related and unrelated medical costs. This resulted in a more favorable cost-effectiveness estimate of $40 700 per QALY gained (Table 3).
Discussion
Our base-case findings suggest that tisagenlecleucel in pediatric patients with B-ALL provides clinical benefits in quality-adjusted and overall survival compared with clofarabine. This result translates into cost-effectiveness estimates that meet commonly cited value thresholds under current model assumptions. The results were robust through sensitivity analyses, with no resulting incremental cost-effectiveness ratio extending beyond $100 000 per QALY gained. Owing to the unique characteristics of modeling potentially curative therapies in the pediatric population, multiple scenario analyses were conducted to test model assumptions. From these scenario analyses, the lowest cost-effectiveness estimate of approximately $37 000 resulted when the discount rate was reduced to 1.5%. The highest cost-effectiveness estimate of approximately $78 000 resulted when standard parametric modeling was used to estimate long-term survival from these therapies. We acknowledge that considerable uncertainty remains around the long-term benefit of tisagenlecleucel owing to limited available evidence; however, with current evidence and assumptions, tisagenlecleucel meets commonly cited value thresholds over a patient lifetime horizon, assuming payment for treatment acquisition for responders at 1 month.
With evidence for tisagenlecleucel coming from single-arm trials, comparator selection was challenging. We chose the comparator with the most similar baseline population characteristics, but also acknowledge that blinatumomab is frequently used to treat this patient population. We suspect that tisagenlecleucel would remain cost-effective compared with blinatumomab. A study conducted by other researchers found the incremental cost-effectiveness ratio of tisagenlecleucel vs blinatumomab was similar to the incremental cost-effectiveness ratio of tisagenlecleucel vs clofarabine (ie, $3000 more per QALY).28 Selection of the comparator is important in cost-effectiveness estimates; however, tisagenlecleucel for pediatric B-ALL likely remains below commonly cited value thresholds despite the chosen comparator.29
Limitations
This analysis is limited primarily by the lack of comparative evidence between the intervention and comparator; this lack complicates the calculation of the incremental difference in outcomes (denominator of the incremental cost-effectiveness ratio) between tisagenlecleucel and clofarabine. First, cross-trial differences in the population between tisagenlecleucel and clofarabine, and not necessarily the different treatments, may contribute to the differences in health outcomes observed between trials. This lack of comparative evidence also prohibits the ability to account for uncertainty about survival curve parameters in sensitivity analyses; therefore, the sensitivity analyses likely underestimate the uncertainty in this evidence. Second, follow-up on event-free survival and overall survival is limited for tisagenlecleucel, which required assumptions regarding long-term results and defining when a patient is effectively cured. A 5-year cure point was used based on evidence from clinical15 and economic14 literature. The cure point was established to determine at what point after trial follow-up patients who were alive but not responding to treatment would die. A later cure point (after 5 years) would result in slightly more favorable cost-effectiveness estimates. Similarly, an earlier cure point (shorter than 5 years) would result in slightly less favorable cost-effectiveness estimates. Uncertainty in long-term survival was partially accounted for in scenario analyses that evaluated different modeling assumptions and time horizons. Although scenario analysis results represent a reasonable lower bound on effectiveness based on current data, it is always possible that a worst-case scenario might present itself (eg, development of neutralizing antibodies with high rates of subsequent relapse). Value should therefore be reassessed once data with longer follow-up are available. Third, mechanisms for payment of tisagenlecleucel are still largely unknown (eg, bundled payment vs fee-for-service, amount of hospital markup, and outcomes-based pricing), necessitating assumptions around costs and payment of these therapies. Clinical experts and other key stakeholders were engaged to inform our assumptions, but value should also be reassessed once more information on payment for this novel therapy is available. Finally, because there is some expectation that tisagenlecleucel might be curative, we considered using a mixture cure model. However, without access to patient-level data, we elected to use a flexible parametric model with an assumed cure-point to approximate the aggregated data with fewer assumptions.
Conclusions
After extrapolating trial results over a patient lifetime horizon, we estimated that more than 40% of pediatric patients who undergo leukapheresis in preparation for tisagenlecleucel therapy will be considered long-term survivors, compared with approximately 10% of pediatric patients considered long-term survivors after receiving comparator therapy. With the evidence available at this time, tisagenlecleucel seems to be priced in alignment with benefits observed over a patient lifetime horizon. The findings are sensitive to the long-term benefit forecasting, but based on currently available evidence and extrapolation assumptions, the cost-effectiveness likely is between $37 000 and $78 000 over a patient lifetime horizon. As payers are negotiating their coverage and reimbursement of tisagenlecleucel, they should consider the uncertainties in the evidence and the assumptions made in this analysis. Novel payment models consistent with the present evidence may reduce the risk and uncertainty in long-term value and be more closely aligned with ensuring high-value care.30,31 Financing cures in the United States is challenging owing to the high up-front price, rapid uptake, and uncertainty in long-term outcomes31,32; however, innovative payment models are an opportunity to address some of these challenges and to promote patient access to novel and promising therapies.
eAppendix. Detailed Methods
eTable 1. Rationale for Model Assumptions
eTable 2. Population Characteristics for the Intervention and Comparator Group
eTable 3. Treatment Regimens
eTable 4. Response to Treatment
eTable 5. Source of Kaplan-Meier Curves to Calculate Transition Probabilities
eTable 6. Survival Curve Fit, Shape, and Scale Parameters for Final Model
eTable 7. Proportion of the Cohort in Each Health State
eTable 8. Receipt of Stem Cell Transplantation
eTable 9. Included Adverse Event Rates
eTable 10. Model Health State Utilities
eTable 11. Treatment-Related Disutilities
eTable 12. Treatment Acquisition Costs
eTable 13. Unit Costs for Healthcare Utilization
eTable 14. Administration and Monitoring
eReferences
References
- 1.Ward E, DeSantis C, Robbins A, Kohler B, Jemal A. Childhood and adolescent cancer statistics, 2014. CA Cancer J Clin. 2014;64(2):83-103. doi: 10.3322/caac.21219 [DOI] [PubMed] [Google Scholar]
- 2.Tai EW, Ward KC, Bonaventure A, Siegel DA, Coleman MP. Survival among children diagnosed with acute lymphoblastic leukemia in the United States, by race and age, 2001 to 2009: findings from the CONCORD-2 study. Cancer. 2017;123(suppl 24):5178-5189. doi: 10.1002/cncr.30899 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.American Cancer Society Treatment of children with acute lymphocytic leukemia (ALL). https://www.cancer.org/cancer/leukemia-in-children/treating/children-with-all.html. Accessed May 2, 2018.
- 4.Hamlin PA, Zelenetz AD, Kewalramani T, et al. . Age-adjusted International Prognostic Index predicts autologous stem cell transplantation outcome for patients with relapsed or primary refractory diffuse large B-cell lymphoma. Blood. 2003;102(6):1989-1996. doi: 10.1182/blood-2002-12-3837 [DOI] [PubMed] [Google Scholar]
- 5.Vose JM, Zhang MJ, Rowlings PA, et al. ; Autologous Blood and Marrow Transplant Registry Lymphoma Working Committee . Autologous transplantation for diffuse aggressive non-Hodgkin’s lymphoma in patients never achieving remission: a report from the Autologous Blood and Marrow Transplant Registry. J Clin Oncol. 2001;19(2):406-413. doi: 10.1200/JCO.2001.19.2.406 [DOI] [PubMed] [Google Scholar]
- 6.Mink SA, Armitage JO. High-dose therapy in lymphomas: a review of the current status of allogeneic and autologous stem cell transplantation in Hodgkin’s disease and non-Hodgkin’s lymphoma. Oncologist. 2001;6(3):247-256. doi: 10.1634/theoncologist.6-3-247 [DOI] [PubMed] [Google Scholar]
- 7.FDA.gov FDA approval brings first gene therapy to the United States. https://www.fda.gov/newsevents/newsroom/pressannouncements/ucm574058.htm. Published August 30, 2017. Accessed September 1, 2017.
- 8.Institute for Clinical and Economic Review CAR-T therapies: final evidence report. https://icer-review.org/material/car-t-final-report/. Accessed on March 23, 2018.
- 9.Hagen T. Novartis sets a price of $475,000 for CAR T-cell therapy. https://www.onclive.com/web-exclusives/novartis-sets-a-price-of-475000-for-car-tcell-therapy. Published August 30, 2017. Accessed September 1, 2017.
- 10.Novartis. Novartis receives first ever FDA approval for a CAR-T cell therapy, Kymriah(TM) (CTL019), for children and young adults with B-cell ALL that is refractory or has relapsed at least twice. https://www.novartis.com/news/media-releases/novartis-receives-first-ever-fda-approval-car-t-cell-therapy-kymriahtm-ctl019. Published August 30, 2017. Accessed August 16, 2018.
- 11.US Food and Drug Administration Tisagenlecleucel (CTL019) for the treatment of pediatric and young adult patients with relapsed/refractory b-bell acute lymphoblastic leukemia. FDA Advisory Committee Briefing Document. Washington, DC: US Food and Drug Administration; 2017.
- 12.Guyot P, Ades AE, Ouwens MJ, Welton NJ. Enhanced secondary analysis of survival data: reconstructing the data from published Kaplan-Meier survival curves. BMC Med Res Methodol. 2012;12:9. doi: 10.1186/1471-2288-12-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Huang M, Latimer N, Zhang Y, Mukhopadhyay P, Ouwens M, Briggs A Estimating the long-term outcomes associated with immuno-oncology therapies: challenges and approaches for overall survival extrapolations. Value & Outcomes Spotlight January/February 2018. https://www.ispor.org/publications/journals/value-outcomes-spotlight/abstract/january-february-2018/estimating-the-long-term-outcomes-associated-with-immuno-oncology-therapies-challenges-and-approaches-for-overall-survival-extrapolations. Accessed June 28, 2018.
- 14.Hettle R, Corbett M, Hinde S, et al. . The assessment and appraisal of regenerative medicines and cell therapy products: an exploration of methods for review, economic evaluation and appraisal. Health Technol Assess. 2017;21(7):1-204. doi: 10.3310/hta21070 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.MacArthur AC, Spinelli JJ, Rogers PC, Goddard KJ, Abanto ZU, McBride ML. Mortality among 5-year survivors of cancer diagnosed during childhood or adolescence in British Columbia, Canada. Pediatr Blood Cancer. 2007;48(4):460-467. doi: 10.1002/pbc.20922 [DOI] [PubMed] [Google Scholar]
- 16.Phillippo DM, Ades AE, Dias S, Palmer S, Abrams KR, Welton NJ. Methods for population-adjusted indirect comparisons in health technology appraisal. Med Decis Making. 2018;38(2):200-211. doi: 10.1177/0272989X17725740 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Jeha S, Gaynon PS, Razzouk BI, et al. . Phase II study of clofarabine in pediatric patients with refractory or relapsed acute lymphoblastic leukemia. J Clin Oncol. 2006;24(12):1917-1923. doi: 10.1200/JCO.2005.03.8554 [DOI] [PubMed] [Google Scholar]
- 18.Kelly MJ, Pauker SG, Parsons SK. Using nonrandomized studies to inform complex clinical decisions: the thorny issue of cranial radiation therapy for T-cell acute lymphoblastic leukemia. Pediatr Blood Cancer. 2015;62(5):790-797. doi: 10.1002/pbc.25451 [DOI] [PubMed] [Google Scholar]
- 19.KYMRIAH [package insert]. East Hanover, NJ: Novartis; 2017.
- 20.Sung L, Buckstein R, Doyle JJ, Crump M, Detsky AS. Treatment options for patients with acute myeloid leukemia with a matched sibling donor: a decision analysis. Cancer. 2003;97(3):592-600. doi: 10.1002/cncr.11098 [DOI] [PubMed] [Google Scholar]
- 21.IBM Micromedex. Redbook. 2018. [Google Scholar]
- 22.Price RA, Stranges E, Elixhauser E. Pediatric Cancer Hospitalizations, 2009. HCUP Statistical brief #132. Rockville, MD: Agency for Healthcare Research and Quality; 2012. [PubMed] [Google Scholar]
- 23.Centers for Medicare & Medicaid Services Physician Fee Schedule Search. https://www.cms.gov/apps/physician-fee-schedule/license-agreement.aspx. Accessed March 2, 2018.
- 24.Dasta JF, McLaughlin TP, Mody SH, Piech CT. Daily cost of an intensive care unit day: the contribution of mechanical ventilation. Crit Care Med. 2005;33(6):1266-1271. doi: 10.1097/01.CCM.0000164543.14619.00 [DOI] [PubMed] [Google Scholar]
- 25.Whittington MD, McQueen RB, Campbell JD. Considerations for cost-effectiveness analysis of curative pediatric therapies. JAMA Pediatr. 2018;172(5):409-410. doi: 10.1001/jamapediatrics.2018.0049 [DOI] [PubMed] [Google Scholar]
- 26.O’Mahony JF, Paulden M. NICE’s selective application of differential discounting: ambiguous, inconsistent, and unjustified. Value Health. 2014;17(5):493-496. doi: 10.1016/j.jval.2013.02.014 [DOI] [PubMed] [Google Scholar]
- 27.Sanders GD, Neumann PJ, Basu A, et al. . Recommendations for conduct, methodological practices, and reporting of cost-effectiveness analyses: Second Panel on Cost-Effectiveness in Health and Medicine. JAMA. 2016;316(10):1093-1103. doi: 10.1001/jama.2016.12195 [DOI] [PubMed] [Google Scholar]
- 28.Hao Y, Eldjerou L, Yang H, Qi CZ, Globe D. Cost-Effectiveness Analysis of Kymriah for the Treatment of Pediatric and Young Adult Patients with Relapsed or Refractory B-Cell Acute Lymphoblastic Leukemia in the United States. Atlanta, GA: American Society of Hematology; 2017. [Google Scholar]
- 29.Neumann PJ, Cohen JT, Weinstein MC. Updating cost-effectiveness—the curious resilience of the $50,000-per-QALY threshold. N Engl J Med. 2014;371(9):796-797. doi: 10.1056/NEJMp1405158 [DOI] [PubMed] [Google Scholar]
- 30.Danzon PM. Affordability challenges to value-based pricing: mass diseases, orphan diseases, and cures. Value Health. 2018;21(3):252-257. doi: 10.1016/j.jval.2017.12.018 [DOI] [PubMed] [Google Scholar]
- 31.Basu A. Financing cures in the United States. Expert Rev Pharmacoecon Outcomes Res. 2015;15(1):1-4. doi: 10.1586/14737167.2015.990887 [DOI] [PubMed] [Google Scholar]
- 32.Towse A, Mauskopf JA. Affordability of new technologies: the next frontier. Value Health. 2018;21(3):249-251. doi: 10.1016/j.jval.2018.01.011 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eAppendix. Detailed Methods
eTable 1. Rationale for Model Assumptions
eTable 2. Population Characteristics for the Intervention and Comparator Group
eTable 3. Treatment Regimens
eTable 4. Response to Treatment
eTable 5. Source of Kaplan-Meier Curves to Calculate Transition Probabilities
eTable 6. Survival Curve Fit, Shape, and Scale Parameters for Final Model
eTable 7. Proportion of the Cohort in Each Health State
eTable 8. Receipt of Stem Cell Transplantation
eTable 9. Included Adverse Event Rates
eTable 10. Model Health State Utilities
eTable 11. Treatment-Related Disutilities
eTable 12. Treatment Acquisition Costs
eTable 13. Unit Costs for Healthcare Utilization
eTable 14. Administration and Monitoring
eReferences