Skip to main content
Springer logoLink to Springer
. 2025 Feb 21;43(5):583–594. doi: 10.1007/s40273-025-01474-3

Innovative Payment Models for Sickle-Cell Disease Gene Therapies in Medicaid: Leveraging Real-World Data and Insights from CMMI’s Gene Therapy Access Model

Antal Zemplenyi 1,2,, Jim Leonard 3, Garth C Wright 1, Michael J DiStefano 1, Kavita Nair 1, Kelly E Anderson 1, R Brett McQueen 1
PMCID: PMC12011968  PMID: 39982606

Abstract

Objective

This study aims to evaluate the financial implications of implementing various payment models, including outcome-based agreements (OBAs), volume-based rebates, and guaranteed rebates, for the newly approved gene therapies, exagamglogene autotemcel (exa-cel) and lovotibeglogene autotemcel (lovo-cel), in the treatment of sickle-cell disease (SCD) from the perspective of Colorado Medicaid. The analysis specifically examines the cost of standard of care (SoC) for severe SCD, the impact of different eligibility criteria based on vaso-occlusive events (VOEs), and the potential financial impacts associated with rebate structures.

Methods

Data from the Colorado Department of Health Care Policy & Financing (HCPF) database was used to estimate the annual costs for Medicaid-enrolled patients with severe SCD from 2018 to 2023. Patients were selected based on various eligibility criteria, including the number of VOEs, acute chest syndrome events, and stroke diagnoses. Three-state Markov models (SCD, stable, and dead) were constructed to compare the costs of SoC and gene therapies. The durability of gene therapy effectiveness and the financial impact of OBAs, volume-based rebates, and guaranteed rebates were evaluated over a 6-year contract period, with scenarios reflecting different VOE criteria and treatment durability.

Results

The average annual SoC cost for severe SCD patients (N = 138) was US$45,941 (SD US$59,653), with higher costs associated with more frequent VOEs. Gene therapies exa-cel and lovo-cel, with one-off list prices of US$2.2 million and US$3.1 million, respectively, exhibited high upfront costs, resulting in a negative cumulative balance averaging − US$2.11 million for exa-cel and − US$3.00 million for lovo-cel per patient over 6 years compared with SoC. Outcome-based rebates could potentially save Medicaid approximately US$260K (uncertainty interval 88K–772K) per patient on average for exa-cel and US$367K (uncertainty interval 122K–1111K) for lovo-cel after they pay the full up-front cost. Volume-based and guaranteed rebates also offered potential savings but varied in impact based on contract duration and effectiveness of gene therapy.

Conclusions

The study highlights critical considerations for Medicaid in negotiating OBAs for SCD gene therapies. Achieving budget neutrality over 6 years is unlikely due to low SoC costs. However, payment models can enhance value-based spending by linking high therapy costs and potential rebates to the health gains these treatments may offer. OBAs offer offsets contingent on therapy effectiveness durability and contract terms (such as length and price), while varying eligibility criteria impact budgets and outcomes. Medicaid real-world data is crucial for navigating complexities in defining eligible populations and structuring OBAs.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40273-025-01474-3.

Key Points for Decision Makers

The high initial cost of gene therapies for sickle-cell disease (SCD) is unlikely to be offset by eliminating standard-of-care costs, which vary from US$45K to US$128K annually, depending on eligibility criteria.
OBAs can significantly impact value-based financial outcomes in SCD, especially with longer contract time horizons that share financial risk due to the uncertain durability of gene therapy effectiveness.
Varying eligibility criteria tied to vaso-occlusive events can have a substantial budget impact. Volume-based or guaranteed rebates may encourage broader access by relaxing eligibility criteria.

Background

Exagamglogene autotemcel (exa-cel, tradename CASGEVY) and lovotibeglogene autotemcel (lovo-cel, tradename LYFGENIA), recently approved by the US Food and Drug Administration, are the first cell-based gene therapies for the treatment of sickle-cell disease (SCD) in patients aged 12 years and older [1]. SCD is a serious, debilitating and life-threatening blood disorder with substantial unmet need, in which red blood cells block blood vessels due to their deformation, impairing tissue oxygenation and triggering painful vaso-occlusive events (VOEs) or vaso-occlusive crises (VOCs) that can lead to organ damage and premature death [2, 3]. The associated health care costs are substantial, estimated at US$2.98 billion annually in the United States [4]. SCD is a rare disease that affects approximately 100,000 people in the US, although the exact prevalence remains uncertain [3].

These gene therapies hold promise for individuals profoundly affected by SCD. Current evidence suggests that these therapies can significantly reduce the incidence of painful VOEs and other serious complications compared with standard treatments, resulting in significant improvements for patients with this debilitating genetic disorder [5, 6]. However, these therapies have high upfront costs. The list price of lovo-cel is US$3.1 million, while exa-cel costs US$2.2 million [7]. Pivotal trials of these therapies have limitations, including single-arm design, small sample sizes and short follow-up periods, which contribute to uncertainty about long-term safety and sustained efficacy [6].

In general, the Centers for Medicare and Medicaid Services (CMS) mandate state Medicaid programs to provide coverage for all FDA-approved therapies, contingent upon the existence of a mandatory rebate agreement between the manufacturer and CMS [8]. A large portion of patients affected by conditions that gene therapies address, such as rare genetic disorders, rely on Medicaid for treatment access, considering that Medicaid covers approximately 28% of the US population [9]. Consequently, the considerable expenses linked with gene therapies represent a notable challenge for Medicaid's budget [10].

The long-term effectiveness of gene therapies (durability of effectiveness) influences the overall cost of disease management, as higher costs may be incurred when patients fail on a gene therapy and are required to switch back to the standard of care (SoC). The consequence of this uncertainty, combined with the substantial upfront costs, places the financial risk of funding a therapy that may not deliver sustained benefits entirely on the payer.

The Center for Medicare and Medicaid Innovation (CMMI) has developed a Cell and Gene Therapy Access Model [11], which proposes outcome-based agreements (OBAs), volume-based rebates, and guaranteed rebates. An OBA ties reimbursement to a treatment's real-world performance, sharing the financial risk between Medicaid and the manufacturer. If a patient does not respond to gene therapy or experiences early treatment failure, the manufacturer may reimburse part of the payment. Volume-based rebates are discounts provided based on the quantity of treatments purchased, while guaranteed rebates involve predetermined discounts or refunds based on specific performance metrics [12, 13]. However, establishing payment models involves key uncertainties, including challenges in accurately describing how eligibility criteria for gene therapy affect the magnitude and variability of SoC cost offsets by gene therapies, and the impact of the durability of the gene therapy's effectiveness on the financial outcomes of rebate agreements [14]. Quantifying these uncertainties can inform negotiations on OBA terms, including time horizon and payback amounts.

The CMMI access model requires participating manufacturers to offer either volume rebates, guaranteed rebates, or both, in addition to outcomes-based rebates [11]. However, no public estimations of the financial implications of different rebate scenarios for payers have been published. In this study, we first assess the actual SoC costs using various eligibility criteria defined by VOEs and different assumptions regarding treatment effectiveness durability. Then, we evaluate the cost of switching to gene therapies and the financial impacts of different payment models for gene therapies, including outcome-based rebates, volume-based rebates, and guaranteed rebates, compared with scenarios without these payment models, from the perspective of Colorado Medicaid.

Methods

Real-World Cost Estimation

Data Source

The analysis relied on data extracted from the Colorado Department of Health Care Policy & Financing (HCPF) database, specifically focusing on Medicaid-enrolled patients with diagnosed SCD based on ICD-10-CM codes (see Online Resource 1 in the electronic supplementary material [ESM]). The data encompassed the period from 2018 to 2023 and provided detailed information on healthcare utilization and associated costs. Key elements extracted from the dataset included demographic details (age, sex), diagnosis codes indicating clinical events with corresponding dates, treatments along with their dates, care settings (inpatient, outpatient, emergency department), and the total payment amounts made by Medicaid for each patient's healthcare services.

Patient Selection and Cost Analysis Method

To identify patients whose mean costs represent the SoC costs for the target group for gene therapies, an index date was established for each patient, and follow-up continued throughout the study period to estimate their costs. Multiple approaches were used to define severe SCD based on claims in the Medicaid database. In the base case, severe SCD was defined similarly to other recent claims analyses using specific criteria applied in the literature [1518]. These criteria included a diagnosis code for SCD of D57, excluding D57.3 for sickle-cell trait, and patients had to be at least 12 years of age. Additionally, the patient with SCD must have experienced at least four vaso-occlusive episodes (VOEs) within the past 2 years, or at least two acute chest syndrome diagnoses within the past 2 years, or a diagnosis of stroke (see Online Resource 1 in the ESM). These criteria were aligned with those used in clinical trials such as CLIMB SCD-121 (ClinicalTrials.gov identifier: NCT03745287) and HGB-206 (NCT02140554). Annual costs were calculated for each year of follow-up separately and then averaged to estimate the mean annual cost per patient.

The standard of care (SoC) cohort consisted of patients who received supportive care, hydroxyurea, blood transfusions, and newer therapies like voxelotor and crizanlizumab. However, the costs and potential impacts of these newer therapies were not analyzed separately. Instead, their contributions are included within the overall SoC estimates.

Beyond the base case, we examined the relationship between a higher number of VOEs (stricter eligibility criteria) and the number of patients meeting the criteria and their mean annual costs. The setting for treating VOEs, such as inpatient, outpatient, or emergency department (ED), might also be associated with disease severity [19]. Therefore, we developed scenarios based on where the VOE was treated.

Individuals with SCD undergoing regular red blood cell transfusions are more likely to avoid VOEs [20]. Due to the potential reduction in VOEs from transfusions, they might not meet eligibility criteria based on VOEs, even though they likely have a more severe form of SCD. As the potential benefits of gene therapy for these patients require more complex consideration, we excluded transfused patients who did not meet any of the other eligibility criteria (VOE, acute chest syndrome, stroke) from the overall cohort of potentially eligible patients and reported their costs separately.

VOEs were treated as distinct events if they occurred at least 3 days apart. Therefore, multiple visits within a span of < 3 days were considered a single VOE event. Overlapping inpatient, outpatient, and ED visits were also treated as a single event. Figure 1 illustrates the approach used to determine whether multiple healthcare visits are treated as separate or single VOEs. We excluded zero-cost claims for dual-eligible Medicare–Medicaid patients. Patients who met the criteria in any given year were retained throughout the observation period. The total cost was considered as the all-cause costs (beyond SCD) paid by Medicaid, which does not include any rebates. All costs were reported in 2023 US dollars.

Fig. 1.

Fig. 1

Visualization of the method of identifying discrete VOEs. OP outpatient, ED Emergency department, VOE vaso-occlusive events

A control group to separate SCD-specific from non-SCD-related healthcare costs could not be identified because of limitations in the claims dataset, including insufficient patient characteristics such as the absence of socioeconomic status or comorbidity profiles. An additional scenario analysis was conducted using the all-cause costs of a subgroup of patients with less severe SCD as a proxy to address this. This subgroup included newly diagnosed patients with minimal disease burden and no severe complications or transfusions. Non-SCD-specific costs from this subgroup were assigned to the no-SCD health states in both arms, and SCD-specific costs were calculated by subtracting these costs from overall all-cause costs. This approach provided an alternative perspective on cost estimates.

Comparison of All-Cause Costs with Standard-of-Care and Gene Therapies

Three-state Markov models (SCD, stable, and dead) were constructed to simulate the costs associated with SoC and gene therapies (exa-cel and lovo-cel), with results presented over a 6-year time frame. The SCD state refers to a health condition characterized by complications from the disease, like vaso-occlusive crises, requiring medical intervention. In contrast, the stable state describes a health condition where symptoms are controlled with no significant disease-related complications. Gene therapy treatment success, defined as freedom from severe VOEs, was estimated at 92% based on recent clinical data [21]. For the base case, the probability of reversion from the stable state to the SCD state was modeled using data from Frangoul et al. [22, 23], where one of 29 evaluable patients experienced a VOC. A 6-month cycle probability of 0.00688 was derived from these observed outcomes. This reversion rate was then applied both during the initial 30.5 months and extrapolated beyond that period due to limited long-term follow-up data. Sensitivity analyses tested the assumption of one out of 29 patients experiencing a VOC by applying this rate across different time horizons (e.g., 5 years, 10 years, 15 years, and 20 years).

Mortality was modeled using age-adjusted rates for SCD patients, with general population mortality rates multiplied by standardized mortality ratios (SMRs) based on age groups, following the approach applied by ICER’s SCD gene therapy review [5]: ages 12–18 years (SMR = 40.07), ages 19–35 years (SMR = 24.24), and ages 35+ years (SMR = 17.48). For patients achieving a stable state (no VOEs) after gene therapy, mortality hazard ratios were adjusted based on estimates from the ICER review [5].

It is acknowledged that the modeling simplifies the disease's complex and heterogeneous course. More detailed analyses focused on gene therapy value have already been developed [5, 2426]. Given the scope of this analysis and the 6-year time horizon, a generalized model was used to capture the broad cost trends associated with SCD. To address heterogeneity, the model integrates cost data from Medicaid claims for actual SCD patients across different levels of VOE, thereby capturing the variation in standard-of-care costs. Furthermore, the model explicitly considers uncertainty in two significant areas: 1) uncertainty around the ‘durability’ of the effectiveness of a cell and gene therapy at a population level; and 2) uncertainty around eligibility and payment approaches.

One author (RBM), who was not directly involved in the model's development, conducted a thorough review of the model. Key assumptions for the cost simulation model are presented in Table 1, while the inputs used in the model are listed in Table 2. The model estimated the cost of the gene therapies (1) without OBA and (2) with OBA, which includes paybacks in the event of gene therapy failure. Payback amounts (rebates) for a 6-year contract were estimated based on a proportionately declining percentage of the gene therapies' upfront cost over the contract duration. Payback rates for the gene therapies are presented in Online Resource 2 (see ESM).

Table 1.

Key model assumptions

Assumption Rationale
Hematopoietic stem cell transplantation (HSCT) is not included in SoC costs HSCT is prioritized over gene therapy when matched donors are available, and therefore, it is excluded from the SoC cost calculation
Treatment effectiveness is assessed by (1) moving to the stable state, and (2) improved survival in the stable state compared with the SCD state Effectiveness was modeled using clinical trial data, reporting sustained freedom from severe vaso-occlusive crises. Reducing VOEs was assumed to avoid related complications, thereby improving long-term survival outcomes
The durability of gene therapy effectiveness was modeled by assuming a gradual reversion to SoC, with a reversion rate informed by clinical trial data and extrapolated over time Reversion was modeled using clinical trial data (1 in 29 patients experienced a VOE), which was extrapolated into a cycle probability of 0.00688 over time
Chronic complications were assumed to be captured by the real-world cost estimates Colorado Medicaid cost data was used to represent the costs associated with different complications, reflecting the various stages and age groups of SCD patients
Mortality hazard ratios differ for adults and adolescents to account for organ damage in adults ICER’s approach, which distinguishes mortality hazard ratios between adults and adolescents based on organ damage, was applied in this study

HSCT hematopoietic stem cell transplantation, ICER Institute for Clinical and Economic Review, SCD sickle-cell disease, SoC standard of care, VOE vaso-occlusive events

Table 2.

Key model inputs

Description Deterministic Standard error Alpha Beta Distribution Source
Mortality
 Standardized mortality ratio, adolescent (12–18 years) 40.07 0.39 0.11 1.67 Lognormal ICER review Table E4 [5], Desai et al. 2020 [27]
 Standardized mortality ratio, adult (19–35 years) 24.24 0.39 0.05 1.50 Lognormal ICER review Table E4 [5], Desai et al. 2020 [27]
 Standardized mortality ratio, adult (35+ years) 17.48 0.39 0.06 1.30 Lognormal ICER review Table E4 [5], Desai et al. 2020 [27]
 Treatment effectiveness on mortality, 18 years and older 0.31 0.39 0.05 1.55 Lognormal ICER review Table E7 [5], Desai et al. 2020 [27]
 All-cause mortality US Life table for the total population, 2019 [28]
Treatment effectiveness
 Probability of achieving VOE-free state after gene therapy, both gene therapies 0.92 0.092 7.08 0.62 Beta Author’s estimation, Frangoul et al. 2024 [22, 23]
 Cycle (6 mo) probability of reversion to SCD, both gene therapies 0.00688 0.000688 99.31 14335.77 Beta Author’s estimation, Vertex press release [21]
Cost, US$
 Exa-cel list price, one-time 2,200,000 Red Book Online [29]
 Lovo-cel list price, one-time 3,100,000 Red Book Online [29]
 Gene therapy other upfront cost, one-time 114,227 28,557 16 7139 Gamma ICER review page E9 [5]
 Annual monitoring costs for the first 3 years, per 6 mo 8653 2163 16 541 Gamma ICER review page E9 [5]
 Annual cost of SoC 45,941 5078 82 561 Gamma Author’s analysis of HCPF claims data 2018–2023
 Unrelated annual healthcare costs, for scenario analysis only 7657 750 104 73 Gamma Author’s analysis of HCPF claims data 2018–2023
Number of VOEs per cycle without gene therapy 2.8 Author’s analysis of HCPF claims data 2018–2023
Starting age, years 21 Frangoul et al. 2024 [22]

HCPF Colorado Department of Health Care Policy & Financing, ICER Institute for Clinical and Economic Review, SCD sickle-cell disease, SoC standard of care, VOE vaso-occlusive events

To determine the uncertainty interval of the financial balance with gene therapies at 6 years, a Monte Carlo simulation with 1000 trials was run, employing the probabilistic value of the standardized mortality ratio, the treatment effectiveness on mortality, response rate, durability of effectiveness, gene therapy administration and monitoring costs, and SoC costs.

Assessing the Impacts of Payment Models

The financial implications of various payment models proposed by CMMI’s Cell and Gene Therapy Access Model were evaluated, focusing on OBAs, volume-based rebates, and guaranteed rebates. The analysis was conducted over a 6-year contract period, the maximum duration proposed by CMMI. It assessed the impact of gene therapy effectiveness duration, patient eligibility criteria, and rebate structures. For OBAs, financial impacts for Medicaid were estimated under different scenarios of response rate, gene therapy effectiveness durability, and across varying contract time horizons. Volume-based rebates were analyzed in relation to patient eligibility criteria, with hypothetical tiered rebates assigned based on varying thresholds for VOEs experienced in the 2 years prior to gene therapy qualification. These thresholds ranged from patients with ≥ 4 VOEs to those with ≥ 15 VOEs over the preceding 2 years. Guaranteed rebates were calculated based on hypothetical offsets of Medicaid costs over different time horizons, reflecting various SoC cost scenarios. These models aimed to explore potential financial savings and mitigate the budgetary impact on Medicaid while maintaining patient access to gene therapies.

Results

Actual Cost of Standard of Care

Base-Case Annual Costs per Patient Eligible for Gene Therapy

The study analyzed a cohort of 291 Colorado Medicaid patients with SCD between 2018 and 2023. Eligibility was assessed based on three criteria (see Table 3). The first group included 121 patients with four or more VOEs in two consecutive years. Expanding the criteria to include patients with two or more acute chest syndrome (ACS) events within 2 years, as well as those who had experienced a stroke, increased the total to 138 patients, representing 47% of the cohort. The mean annual cost for this group was US$45,941 (SD US$59,653).

Table 3.

Mean annual all-cause costs for severe SCD patients eligible for gene therapy

Criteria N Mean cost
$US
SD
$US
VOE: ≥ 4 in 2 years 121 44,499 57,199
ACS: ≥ 2 in 2 years and < 4 VOEs in 2 years < 30 61,254 113,641
Stroke: ≥ 1 in 1 year and < 4 VOEs and < 2 ACS in 2 years < 30 88,563 84,053
All eligible (including all the above) 138 45,941 59,653

Sample size < 30 and distribution by sex cannot be reported due to Medicaid's data protection regulation

ACS acute chest syndrome, SD standard deviation, VOE vaso-occlusive events; 2 years refers to 2 consecutive years during 2017–2023

Annual Cost with Different Eligibility Criteria Scenarios

The relationship between the number of VOEs over 2 years and the mean annual all-cause costs for severe SCD patients was assessed based on the setting where the VOE was treated. Table 4 shows that costs increase progressively with higher VOE thresholds, indicating that patients with more frequent VOEs incur significantly higher healthcare costs across all settings. Inpatient VOEs are linked to higher costs compared with those treated in outpatient or emergency department (OP/ED) settings. This suggests that inpatient VOEs may be more indicative of severe forms of SCD.

Table 4.

Number and mean annual costs for severe SCD patients applying VOE criteria for different settings

Number of VOEs in 2 years VOE in any setting (IP, OP or ED) VOE only in IP VOE only in OP or ED
N Mean cost (SD) $US N Mean cost (SD)
$US
N Mean cost (SD)
$US
≥ 4 121 44,912 (57,301) 51 68,647 (62,205) 98 43,000 (52,946)
≥ 6 91 49,838 (57,363) 33 84,541 (59,516) 74 45,708 (54,903)
≥ 8 71 52,563 (57,207) < 30 104,762 (61,084) 53 48,176 (56,617)
≥ 10 62 57,113 (58,560) < 30 122,451 (67,839) 46 52,348 (58,855)
≥ 15 43 61,577 (57,884) < 30 128,735 (77,210) 33 60,374 (62,584)

Sample size < 30 and distribution by sex cannot be reported due to Medicaid's data protection regulation. The counts exclude patients identified based on acute chest syndrome or stroke

ED emergency department, IP inpatient, OP outpatient, SD standard deviation, VOEs vaso-occlusive events

Annual Costs Per SCD Patient Receiving Transfusion

For patients without any transfusions, the mean annual cost was US$20,709 (SD 39,131). With one or more transfusions, the mean cost increased to US$54,155 (SD 65,882). A higher frequency of transfusions indicates increased all-cause medical costs, with patients with four or more transfusions in a year having a mean cost of US$63,568 (SD 71,463) (see Online Resource 3 in the ESM). In total, 101 patients received transfusions during the study period. About 52% of them met one of the criteria for severe SCD (as outlined in Sect. 3.1.1) and were therefore accounted for in the cost analysis. The remaining 48%, who did not have VOEs, had a lower mean annual cost of US$31,657 (SD 52,969).

Comparative Cost of Using Exa-cel

The results in this section demonstrate the impact for exa-cel; results for lovo-cel are presented in Online Resource 4 (see ESM). Figure 2 illustrates the projected costs over 6 years assuming a 92% response rate and an annual 3.4% rate of patients reverting from a VOE-free state back to SCD. The high upfront cost of gene therapy occurs in Year 1, with only a mild increase in cumulative costs over 6 years. In contrast, SoC costs gradually increase each year but remain low, offering minimal cost-offset potential. This results in a significantly high negative cumulative balance of −US$2.11M per patient on average. The Monte Carlo simulation shows the cumulative balance after 6 years within a 95% uncertainty interval, ranging from − US$2.18M to − US$2.01M. In the alternative scenario, a mean annual non-SCD-specific cost of US$7.6K (SD 8.8K) was assigned to the no-SCD health states in both arms. This approach adjusted the cumulative balance after 6 years to − US$2.07M.

Fig. 2.

Fig. 2

Comparison of per-patient costs between exa-cel gene therapy and SoC for SCD over 6 years. GT gene therapy, SCD sickle-cell disease, SoC standard of care

Payment Models

Outcome-Based Agreements

Over a 6-year contract, with a proportionally set payback amount and the base-case criteria outlined in Sect. 3.1.1, the estimated average balance is − US$1.85M per patient on average. Under these terms, OBAs could save Medicaid approximately US$260K per patient on average, with savings ranging from US$88K to US$772K based on the Monte Carlo simulation (see in more detail in Online Resource 5.1 in the ESM).

The financial impact of OBAs varies with response rates, with savings ranging from US$197K at a 95% response rate to US$507K at an 80% response rate per patient on average. Similarly, the financial impact depends on gene therapy effectiveness durability, with savings ranging from US$184K at 20 years to US$384K at 1 year per patient on average. OBAs can effectively reduce Medicaid’s financial burden, particularly in cases of early gene therapy failure. However, with better response rates and more durable therapy, savings may be lower, while providing improved health outcomes for patients (see Table 5).

Table 5.

Estimated average per-patient cost differences without and with OBA with various durability and response rate assumptions for exa-cel effectiveness

Balance without OBA at 6 years ($US) Balance with OBA at 6 years ($US) Financial impact of OBA at 6 years ($US)
Response rate
 80% − 2,130,000 − 1,623,000 507,000
 85% − 2,120,000 − 1,717,000 403,000
 90% − 2,111,000 − 1,810,000 301,000
 92% (base case) − 2,107,000 − 1,847,000 260,000
 95% − 2,101,000 − 1,904,000 197,000
Gene therapy effectiveness durability
 1 year − 2,119,000 − 1,735,000 384,000
 2.5 years (base case) − 2,107,000 − 1,847,000 260,000
 5 years − 2,103,000 − 1,886,000 217,000
 10 years − 2,101,000 − 1,906,000 195,000
 15 years − 2,100,000 − 1,913,000 187,000
 20 years − 2,100,000 − 1,916,000 184,000

‘Durability’ refers to the time horizon over which patients revert to SoC at the rate of 1/29; ‘Balance’ refers to the cost difference between exa-cel gene therapy and the SoC over 6 years; the ‘Financial impact of OBA’ refers to the difference in balances between the with- and without-OBA scenarios

OBA outcome-based agreement, SoC standard of care

The mean balance per patient between scenarios with and without OBAs can also vary significantly depending on the contract period. Unlike factors such as response rate or durability of the gene therapy effectiveness (as shown in Table 5), the contract horizon is a controllable parameter that Medicaid can determine. Extending the contract period increases the duration of risk-sharing and enhances the likelihood of achieving a payback (see Table 6).

Table 6.

Estimated per-patient cost differences without and with OBA over different contract time horizons

Contract time horizon Balance without OBA at 6 years ($US) Balance with OBA at 6 years ($US) Financial impact of OBA at 6 years ($US)
3 years − 2,107,000 − 1,882,000 225,000
4 years − 2,107,000 − 1,869,000 238,000
5 years − 2,107,000 − 1,857,000 250,000
6 years (base case) − 2,107,000 − 1,847,000 260,000

‘Balance’ refers to the cost difference between exa-cel gene therapy and the SoC over 6 years; the ‘Financial impact of OBA’ refers to the difference in balances between the with- and without-OBA scenarios

OBA outcome-based agreement, SoC standard of care

Volume-Based Rebates

The budget impact of gene therapy is heavily influenced by the number of eligible patients, which varies based on Medicaid’s eligibility criteria. With stricter criteria, fewer patients qualify for gene therapy. For example, the number of eligible patients drops from 121 at ≥ 4 VOEs to 43 at ≥ 15 VOEs. The balance per patient with OBA remains relatively consistent across different eligibility criteria, ranging from − US$1.85M to − US$1.77M. This indicates that the financial impact of OBAs per patient is relatively independent of the stringency of eligibility criteria. However, the overall budget impact is substantially impacted by the eligibility criteria, ranging from − US$157M with more lenient criteria (≥ 4 VOEs in 2 years) to − US$76M with stricter criteria (≥ 15 VOEs in 2 years). Implementing a volume-based rebate once a certain threshold is met can mitigate the budget impact while maintaining broader access through less strict criteria. Table 7 illustrates how hypothetical tier-based rebates of 30%, 20%, 10%, and 0%, combined with OBA, affect the total cost for Medicaid based on the eligibility criteria. The table shows that with less strict eligibility criteria (i.e., fewer VOEs required in 2 years), the number of eligible patients increases, leading to higher volume-based rebates. This approach helps balance broader access with cost control.

Table 7.

Impact of VOE-based eligibility criteria and hypothetical volume-based rebates on total costs

VOEs in 2 years N Balance with OBA per patient ($US) Cost for all eligible with OBA ($US) Volume-based rebate Total rebate ($US) Total cost ($US)
≥ 4 121 − 1,853,000 − 224,213,000 30% 67,263,900 − 156,949,100
≥ 6 91 − 1,828,000 − 166,348,000 20% 33,269,600 − 133,078,400
≥ 8 71 − 1,815,000 − 128,865,000 10% 12,886,500 − 115,978,500
≥ 10 62 − 1,792,000 − 111,104,000 0% 0 − 111,104,000
≥ 15 43 − 1,770,000 − 76,110,000 0% 0 − 76,110,000

We assumed that 100% of patients eligible based on VOEs would accept gene therapies

‘Balance’ refers to the cost difference between exa-cel gene therapy and the standard of care over 6 years

OBA outcome-based agreement, VOE vaso-occlusive event

Guaranteed Rebates

Manufacturers may offer guaranteed rebates from the gene therapy price. According to CMMI’s proposal, manufacturers must specify the proposed formula and components of the rebate. Here, we present a hypothetical approach to the rebate, where rebate amounts are set equivalent to the costs of SoC that Medicaid would otherwise incur over specific time horizons, rather than being tied to individual patients' costs, which may be variable from year to year. This approach is intended to inform the parties negotiating the contract on the magnitude of the rebate relative to Medicaid's average costs. As there is variation across state Medicaid programs, it may be valuable to estimate this separately for each state. Table 8 illustrates various rebate scenarios based on years of SoC costs and their impact on reducing the upfront financial burden on Medicaid.

Table 8.

Impact of guaranteed rebates on financial balance

Scenarios Rebate amount ($US) Rebate % Net price of exa-cel ($US) Balance with OBA ($US)
List price 0 0.0% 2,200,000 − 1,847,000
5 years of SoC costs 229,705 10.4% 1,970,295 − 1,645,000
7 years of SoC costs 321,587 14.6% 1,878,413 − 1,564,000
10 years of SoC costs 459,410 20.9% 1,740,590 − 1,442,000
15 years of SoC costs 689,115 31.3% 1,510,885 − 1,240,000

‘Balance’ refers to the cost difference between exa-cel gene therapy and the SoC over 6 years

OBA outcome-based agreement, SoC standard of care

Discussion

Our study highlights three key findings that can assist insurers and manufacturers in designing payment models for gene therapies for SCD. First, the costs of standard care for SCD are relatively low compared with those for other diseases treated with gene therapies, like hemophilia A and B [14]. As a result, gene therapies such as exa-cel or lovo-cel are unlikely to achieve budget neutrality for Medicaid. Second, OBAs can have a significant financial impact in SCD, depending on the average duration of gene therapy effectiveness. Longer contract time horizons may enable a greater sharing of financial risk due to the uncertainty surrounding the durability of gene therapy effectiveness. Third, varying eligibility criteria tied to the number of VOEs can greatly influence Medicaid’s overall budget impact. Volume-based or guaranteed rebates can incentivize payers to adopt more relaxed eligibility criteria, thereby expanding access to broader populations.

The Initial high cost of gene therapies for SCD will not be offset by eliminating SoC costs. The SoC costs range from US$45K to US$128K annually, depending on eligibility criteria, such as the number of VOEs used as proxies for disease severity. The highest annual cost is driven by inpatient VOEs, indicating more severe disease. For Medicaid-covered patients with severe complications, literature estimates are somewhat higher than our findings, with reported annual costs ranging from US$61,665 to US$81,333 [16, 18], possibly due to regional variations in healthcare utilizations and costs. Table 4 presents stricter eligibility criteria that yield costs that are more comparable to the literature, which were used in the scenario analysis in Table 7. Over a 6-year period, which is the maximum time period proposed by CMMI for payment models, it is highly unlikely that exa-cel or lovo-cel would reach the break-even point in Medicaid budgets. Coverage of these therapies will not be budget neutral, even with criteria identifying the most severe and costly patients, such as those with frequent VOEs, strokes, acute chest syndrome, or who require transfusions. Even at the high end of annual costs for SCD, the potential cost offsets from gene therapy are not as substantial as in other conditions, like hemophilia A and B, where gene therapies can save up to US$430–US$550K annually by replacing costly, recurring treatments [14]. However, while gene therapies for hemophilia largely sustain the good health already achieved with current care, therapies for SCD address significant unmet medical needs. This highlights the importance of considering the potential for substantial health gains alongside costs when designing payment models for SCD therapies [5, 16, 18, 26, 30]. OBAs can have a significant financial impact based on gene therapy effectiveness durability and contract duration. For 2.5-year durability, savings reach US$260K per patient, while longer durability (10–20 years) reduces rebates. However, when OBA’s impact is smaller, Medicaid receives more health benefits for its expenditures, resulting in more value-based spending. The small difference between the 3-year and 6-year results is due to most payback being linked to lack of initial response, as durable effects are expected for responders. For example, a 3-year contract yields $225K in savings, while a 6-year contract increases savings to US$265K. These agreements improve the alignment between costs and health benefits since payers can recoup costs for patients whose therapy is ineffective, while high costs are incurred only for those who achieve lasting health benefits. Therefore, collaboration among Medicaid organizations can increase leverage in negotiations. In the case of SCD, where more gene therapies are available and competition is higher, OBA terms can be key in selecting preferred therapies. This also offers manufacturers opportunities to differentiate their products from competitors.

The eligibility criteria linked with volume-based or guaranteed rebates proposed in CMMI’s Access Model could encourage payers to adopt less strict eligibility criteria, allowing broader access to gene therapies while managing budget impacts through substantial rebates. A hypothetical 30% volume-based rebate demonstrates this effect clearly. In the Colorado Medicaid example, such a rebate for exa-cel could save US$67M, reducing the total budget impact to US$157M and providing access to 121 patients. In contrast, using stricter eligibility criteria (≥ 10 VOEs in 2 years) without any volume-based rebate would cut the number of patients by 50% (N = 62) and lower the budget impact by only 25%, resulting in a US$117M cost. Guaranteed rebates could be linked to eligibility criteria in a similar way.

CMS is seeking applications to test whether its approach can achieve the three goals of the Access Model: improving Medicaid beneficiary access to innovative treatments, enhancing health outcomes, and reducing healthcare expenditures [11]. As state Medicaid programs introduce gene therapies for SCD into their coverage, they will need to navigate trade-offs between these goals.

In terms of accessibility, this study demonstrates that linking eligibility criteria with volume-based or guaranteed rebates could incentivize payers to adopt less strict criteria, thereby broadening access to gene therapies while managing budget impacts.

Regarding health outcomes, identifying SCD patients who would benefit most from gene therapies is complex, particularly when relying solely on claims data. For example, patients receiving transfusions may experience fewer VOEs but could develop complications like iron overload, leading to higher healthcare costs later on. Additionally, VOEs in inpatient and outpatient settings can vary significantly in severity, impacting both patient quality of life and healthcare expenditures. These complexities underscore the importance of tracking patient outcomes beyond merely counting VOEs, which could then also inform OBAs.

In terms of health expenditures, a major concern with gene therapies is how to share the financial risk between payers and manufacturers, especially when the high cost of new therapies is paid upfront but carries potential risks of failure. In cases where gene therapies, such as Valoctocogene roxaparvovec and Etranacogene dezaparvovec for hemophilia A and B, offer long-term savings, the focus is on when these therapies reach the break-even point, as they provide only modest health improvements since the established standard of care (SoC) already maintains relatively good patient health [31]. In contrast, SCD presents a substantial unmet medical need, with patients facing reduced life expectancy (57 years) and enduring painful VOEs that significantly diminish their quality of life [30, 32]. Thus, the value generated by gene therapies like exa-cel and lovo-cel should be central in negotiations. While it is unlikely that total health expenditures will decrease compared with SoC when using gene therapies, OBAs, along with volume-based and guaranteed rebates, not only ease budget constraints but also, through lower costs, contribute to more effective value-based spending at both the patient and population levels.

In summary, OBAs and the rebates proposed in CMMI’s Access Model are effective tools for managing the budgetary impact of gene therapies while expanding access. Leveraging these strategies can help balance financial burdens and improve accessibility and health outcomes for Medicaid patients.

Our study has several limitations. The simplification of SOC modeling is a limitation, as SCD is a highly heterogeneous condition. However, this approach was chosen to maintain focus on the study's key objectives within the available time frame and data. The absence of a control group to distinguish between SCD-specific and non-SCD-related healthcare costs is a key limitation. Due to limitations in the claims dataset, such as the absence of socioeconomic status or comorbidity profiles, we could not identify an appropriate non-SCD population for comparison. As a result, all-cause costs for SCD patients were modeled, which may overestimate costs attributable solely to SCD; however, an alternative scenario produced consistent results, supporting the robustness of our conclusions. The reliance on Medicaid data from a single state may limit the generalizability to other regions or payer systems. Assumptions about gene therapy effectiveness durability and cost projections are based on available literature and may change with new evidence. Longitudinal studies are needed to assess real-world effectiveness and durability. Additionally, the SoC costs were extrapolated based on the past, irrespective of how quickly patients developed complications, which could potentially increase all-cause annual costs. The study does not account for the potential impact of broader eligibility criteria on migration patterns. Specifically, there may be an incentive for individuals to shift to accessing gene therapy, which is not incorporated in the current modeling. The model assumed fixed SoC expenditures, not accounting for increased costs as SCD complications progress with age, potentially underestimating the burden of advanced disease. However, by including patients across all age groups, including those with advanced disease, the model captures a broad spectrum of patient costs, mitigating some of these limitations. Allogeneic transplant was not included as a competing therapy due to the complexity of transplant eligibility and the evolving use of mismatched donors, but this could be explored in future research. Additionally, the analysis did not account for loss to follow-up, so projected savings may not be fully realized by Medicaid. Patients could switch to other payers, such as Colorado's state healthcare exchange or employer-sponsored insurance, potentially reducing the realized savings and complicating the path to budget neutrality.

Conclusion

The study demonstrates critical considerations for payers regarding coverage and negotiation of terms for OBAs concerning gene therapies for SCD. The findings highlight the low probability, nearly impossible, of achieving budget neutrality for Medicaid over a 6-year period with exa-cel or lovo-cel, given the relatively low costs associated with the SoC for SCD. OBAs may offset the budget impact of gene therapy for SCD contingent upon the durability of gene therapy effectiveness and contract terms, such as time horizons and payback rates. Moreover, varying eligibility criteria significantly impact Medicaid's budget and health outcomes for enrollees. Importantly, negotiations must consider the substantial unmet medical need in SCD, which includes a wide range of complications that vary across patients. OBAs focusing on a single complication, such as VOEs, should not overlook others to ensure comprehensive care. A value-based pricing approach is essential to ensure therapy affordability and sustainability within the healthcare system. Despite the limitations, Medicaid real-world data can provide crucial insights into navigating the complexities of defining the population eligible for gene therapies and structuring contract terms for OBAs. This contribution is particularly significant in a disease landscape like SCD, where therapies are needed to address unmet medical needs, but the long-term effectiveness of treatments remains uncertain. Real-world data can contribute to designing payment models in the context of state Medicaid programs.

Supplementary Information

Below is the link to the electronic supplementary material.

Declarations

Conflict of interest

Dr Leonard is affiliated with the Colorado Department of Health Care Policy & Financing. The authors have disclosed no conflicts of interest concerning the research, authorship, or publication of this article. None of the authors are affiliated with or hold financial interests in the manufacturers of the gene therapies investigated in this study. The views expressed in this article represent the opinions of the authors and do not necessarily reflect those of their respective employers.

Funding

The University of Colorado is funded by the Department of Health Care Policy & Financing for the assessment of value-based contracts.

Ethics approval

The study was declared exempt by the Colorado Multiple Institutional Review Board.

Consent for publication

Not applicable.

Consent to participate

Not applicable.

Data availability

All data analyzed during this work are from the Colorado Department of Health Care Policy & Financing. The data are not publicly available. The model used during this work was developed by the University of Colorado under a contract with the Colorado Department of Health Care Policy & Financing.

Code availability

Not applicable.

Author contributions

Antal Zemplenyi, Brett McQueen, Jim Leonard, Michael DiStefano, Kelly Anderson, and Kavita Nair contributed to the study conception and design. Material preparation, data collection and analysis were performed by Garth Wright and Antal Zemplenyi. The first draft of the manuscript was written by Antal Zemplenyi and Brett McQueen, and all authors commented on previous versions of the manuscript. All authors read and approved the final version of the paper.

References

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Data Availability Statement

All data analyzed during this work are from the Colorado Department of Health Care Policy & Financing. The data are not publicly available. The model used during this work was developed by the University of Colorado under a contract with the Colorado Department of Health Care Policy & Financing.


Articles from Pharmacoeconomics are provided here courtesy of Springer

RESOURCES