Cost–effectiveness of adjuvant atezolizumab for patients with stage II–IIIA PD-L1+ non-small-cell lung cancer

Lung cancer is responsible for the most cancer-related deaths in the USA [1]. Non-small-cell lung cancer (NSCLC) accounts for the majority (88%) of all lung cancer cases, and approximately half of all patients are diagnosed with early-stage NSCLC (eNSCLC), which is considered to be stages I, II or resectable stage III disease [2,3]. The 5-year survival estimate is only 36% for patients with stage IIIA NSCLC [3,4]. Surgical resection with or without neoadjuvant or adjuvant therapy is the primary treatment option for operable patients with eNSCLC, and cisplatin-based chemotherapy has been the historical standard of care for adjuvant treatment of resectable stage IB–IIIA eNSCLC [5].

Improvements in overall survival have been modest for patients with eNSCLC receiving adjuvant chemotherapy [6], with high rates of recurrence, especially for patients with stage II-IIIA disease [7]. The programmed cell death 1 ligand 1 (PD-L1) immunotherapy (IO) atezolizumab demonstrated a significant disease-free survival (DFS) benefit compared with best supportive care (BSC) among resected patients who had received adjuvant chemotherapy based on the randomized, open-label, Phase III IMpower010 clinical trial (NCT02486718) [8,9]. In October 2021, atezolizumab was approved by US Food and Drug Administration as an adjuvant treatment following resection and platinum-based chemotherapy for adults with stage II-IIIA NSCLC (American Joint Committee on Cancer, 7th edition) and PD-L1 expression on ≥1% of tumor cells (PD-L1+).

Quantifying the cost–effectiveness of new therapies is essential to understanding their value and to informing payer decisions. When this study was conducted, atezolizumab was the first and only IO approved in the adjuvant setting for patients with eNSCLC in the USA, for which the previous standard of care was chemotherapy alone after resection. Therefore, this study evaluated the cost–effectiveness of atezolizumab versus BSC following adjuvant chemotherapy from a US payer perspective using a $150,000 per quality-adjusted life-year (QALY) willingness-to-pay (WTP) threshold.

Materials & methods

Model overview

A Markov model was developed to compare the clinical and economic outcomes of atezolizumab versus BSC as adjuvant treatment following resection and platinum-based chemotherapy for adult patients with stage II-IIIA NSCLC whose tumors have PD-L1 expression on ≥1% of tumor cells (Figure 1). The model was developed from a US commercial payer perspective using a lifetime time horizon. Moreover, it uses a monthly cycle length and, as such, uses monthly probabilities to inform transitions across health states.

Health states

All patients start and remain in the DFS health state while alive and disease free. Patients can transition from the DFS health state to the locoregional recurrence (LR) health state and may be treated with curative treatment (80%) or palliative treatment or no treatment (20%) [10–12]. For both curative and palliative treatment, the model assumes that patients receive stereotactic ablative radiotherapy (SABR; 25%), re-resection (25%), or systemic therapy (chemotherapy or IO) with or without radiotherapy (50%) for a maximum of 6 months. Patients with LR receiving treatment with curative intent can transition to first line (1L) metastatic recurrence (MR) or death. Those receiving palliative treatment or no treatment can only progress to death.

Patients transitioning to 1L MR from the DFS or LR health state may be treated (77%) or may not receive treatment (23%) [12,13]. Patients who receive treatment remain in this health state while they are alive and progression free and receive 1 of 4 treatment options for up to 12 months based on US clinical practice guidelines and clinical expert consultation (Supplementary Table 1). Those who receive treatment can progress to 2L MR or death, and those not receiving treatment can only transition to death.

Additionally, patients in the atezolizumab arm who experience LR or 1L MR are allowed to re-challenge with another IO depending on the timing of the recurrence. Specifically, patients are re-challenged with IO if recurrence occurs ≥17 months after atezolizumab initiation (or ~6 months after the end of 16 cycles of atezolizumab). If recurrence occurs <17 months after atezolizumab initiation, patients can only be re-treated with chemotherapy in the model. Patients transitioning to 2L MR may be treated (70%) or not treated (30%) after metastatic progression, based on consultation with US clinical experts. Four treatment options are available for treated patients (Supplementary Table 1).

Efficacy

DFS data beyond the IMpower010 32-month follow-up period were extrapolated using a log-logistic distribution, which provided the most appropriate estimate of the proportion of patients receiving atezolizumab who may be disease free at 10 years based on goodness-of-fit criteria, specifically the Akaike Information Criteria and Bayesian Information Criteria. The extrapolated DFS data were then adjusted based on input from clinical experts. The unadjusted and adjusted DFS curves are provided in Supplementary Figure 1. First, a cure adjustment was applied where a proportion of patients are assumed to no longer be at risk of a DFS event, such as recurrence or death. Based on the findings of Sonoda et al., the model assumes that a maximum of 91.5% of patients can be considered free of a DFS event (or ‘cured’) at 5 years post-surgery [14]. Second, the model does not allow for the probability of an uncured patient with eNSCLC dying to be smaller than that of an individual from the general population. Also, among patients with eNSCLC who are considered cured and no longer at risk of cancer-related death, the model allows for their probability of death to be adjusted with a standardized mortality ratio to account for excess mortality of 25% faced by these patients relative to the general population [15,16]. Finally, the model allows the treatment effect of atezolizumab to decrease over time, but there are insufficient data from IMpower010 and the literature to identify the time point where the treatment effect of atezolizumab ceases. Therefore, the model assumes that it ceases at year 5 based on clinical expert opinion.

During each 1-month cycle, the probabilities of death, LR, or MR were based on data from IMpower010. Transition probabilities among health states that were unavailable from IMpower010 were derived from the literature and other clinical trials (Table 1).

Costs

All costs from a US payer perspective are included and expressed in 2022 US dollars adjusted for inflation using data from the US Bureau of Labor Statistics data. The model includes costs associated with treatment of eNSCLC as well as for LR or MR, which include both drugs and procedures. In addition, the model considers administration costs and costs associated with the management of common adverse events (AEs).

Treatment costs for patients with eNSCLC in the intervention arm included atezolizumab 1200 mg every 3 weeks for a treatment duration based on time-to-off treatment from the IMpower010 clinical trial, followed by follow-up care for up to 5 years [8,9]. Patients in the BSC arm only receive follow-up care for up to 5 years. Commercial drug costs used weighted average costs from the Medi-Span Price Rx database (Price Rx.Medi-Span.com, accessed February 2022), and Medicare drug costs used the average selling price from the Medicare Fee Schedule 2022 (Supplementary Table 2). The indicated dosing for each product was based on the prescribing information, and the dosing for IO and chemotherapy regimens followed US treatment guidelines.

The specific regimens for the treatment of recurrence and their market shares are provided in Supplementary Table 1. LR costs include radiotherapy cost per fraction at $297 [19] and a one-time cost for biopsy of $2499 [20] during the first LR and/or 1L MR. Other costs associated with LR – such as radiotherapy, biopsy, SABR, or re-resection – are provided in Supplementary Table 3.

Drug administration costs from the commercial perspective were based on IBM MarketScan^® data analysis [21], and Medicare administration costs were from the Center for Medicare & Medicaid Services Physician Fee Schedule [22] (Supplementary Table 3). Follow-up care comprised computed tomography scans every 6 months for 3 years based on US treatment guidelines and clinical expert consultation (Supplementary Table 3). End-of-life costs include all hospital utilization and spending within the last 6 months of life for patients who do not receive hospice care but do include nursing home costs (Supplementary Table 3).

Finally, the model captures the costs associated with the management of treatment-related grade ≥3 AEs for atezolizumab patients in the DFS health state (Supplementary Table 4) based on the IMpower010 trial [23], with AE-related costs from the Healthcare Cost and Utilization Project [19]. Monthly AE management costs are provided in Supplementary Tables 4. The model assumes that patients on IO and chemotherapy realize these AE management costs associated with atezolizumab and chemotherapy regimens used in the trials, regardless of the specific drug that they receive for treatment.

Health state utilities

Since IMpower010 did not collect patient-reported outcomes, health state utilities were derived from EQ-5D scores published in the literature or from the IMpower150 trial (Table 2). The health state utility values in the model decrease with progressive disease. However, the utility values are not time-variant and may be greater than age-adjusted general population utility values in certain cycles. In those cases, the model switches to use of age-adjusted general population utility values from Ara and Brazier (2011) [24]. This adjustment is necessary to ensure that the model does not assume that patients with eNSCLC have a better quality of life than the general population.

Analyses

The primary outcome was the incremental cost–effectiveness ratio (ICER), calculated as the difference in costs divided by the difference in QALYs. Both costs and QALYs were discounted at a rate of 3% per year. A deterministic 1-way sensitivity analysis was performed using the 20th and 80th percentile values as lower and upper bounds for each parameter to identify influential parameters. A probabilistic sensitivity analysis was also performed to evaluate the impact of the joint uncertainty of all model parameters and to test the robustness of model results. Last, a scenario analysis was performed using Medicare-specific costs and assuming an average patient age of 65 years to reflect the Medicare population perspective.

Results

Base case cost–effectiveness results

In the base case analysis, atezolizumab generated 1.05 additional QALYs at an additional cost of $48,956. The base case ICER for the comparison of atezolizumab versus BSC was $46,859/QALY (Table 3), which is below the $150,000/QALY WTP threshold. The relatively higher drug treatment costs associated with atezolizumab during the DFS health state were partly offset by the lower treatment costs in later metastatic health states (Supplementary Table 5).

Sensitivity analysis results

The deterministic sensitivity analyses showed that no model parameters increased the ICER above the $150,000/QALY WTP threshold. The most influential variable was the time to rechallenge with IO in the 1L metastatic health state (Figure 2). Among the utility inputs, the model was most sensitive to the utility value in DFS, due to the time spent in DFS.

The results of the probabilistic sensitivity analyses were consistent with those of the base-case analysis, with an incremental cost of $50,549 and incremental QALY of 1.03, resulting in an ICER of $49,142/QALY. At a WTP threshold of $150,000/QALY, the probability of atezolizumab being cost-effective was ~91% (Figure 3).

Scenario analysis results

Results from the scenario analysis showed that atezolizumab is also cost-effective in the Medicare population, where atezolizumab generated 0.98 additional QALYs at an additional cost of $47,382 for an ICER of $48,512/QALY with atezolizumab versus BSC (Table 4).

Discussion

This study demonstrated the cost–effectiveness of atezolizumab relative to BSC, with an ICER of $46,859/QALY and a WTP threshold of $150,000/QALY from a US commercial payer perspective. This result was driven primarily by patients spending more time in the DFS health state before transitioning to LR or metastatic disease, thereby decreasing treatment costs in the metastatic settings, as approximately a third of the treatment costs for the atezolizumab arm is accrued during the DFS health state after initial resection. Overall findings were similar in the scenario analysis that considered the Medicare perspective, with an ICER of $46,859/QALY. The sensitivity analyses supported the robustness of the primary findings. The cost–effectiveness of new therapies is an important consideration for healthcare treatment decisions, especially given the increased focus on demonstrating value in the USA as evidenced by the recent development of value-based frameworks [29,30]. While drug pricing policies are continually evolving in the USA, in light of the recently passed Inflation Reduction Act, there will be increasing focus on the price of new therapeutics and the value they bring to patients and health systems.

To our knowledge this is the first demonstration of cost–effectiveness of cancer IO for eNSCLC adjuvant treatment in the US setting, while there has been one cost–effectiveness study from the perspective of China's healthcare system [31]. These results highlight how the incremental drug treatment costs with IO can offset the economic burden associated with future recurrences and death. The economic burden associated with recurrences and deaths in the eNSCLC have been previously reported [32]. Sharma and colleagues (2022) reported an estimated 2387 of 4400 patients with eNSCLC in the USA would experience a recurrence within 5 years of BSC and would be eligible for adjuvant atezolizumab treatment [32]. The authors estimated that 1030 recurrences and 369 deaths could be avoided with atezolizumab treatment, with subsequent reductions in recurrence-related costs of $785 million in direct costs and $32 million in terminal-care costs over 5 years [32]. Others have also reported total all-cause healthcare costs to be nearly fivefold higher (+$4577 per patient per month) for patients with stage IA to IIIB NSCLC with recurrence over an average follow-up period of 30 months ($5889 vs $1312, respectively; p < 0.001), driven largely by hospitalizations ($2739 vs $559 per patient per month, respectively) [33]. Thus, the longer time in the DFS health state observed for patients in the atezolizumab arm based on clinical trial data is likely to translate to meaningful healthcare savings for this patient population.

At the time of this publication, atezolizumab was the only IO approved in the eNSCLC adjuvant treatment setting. Nivolumab was also approved as neoadjuvant treatment in combination with platinum-doublet chemotherapy for patients with NSCLC, but was not included in this study scope, which focused on the adjuvant setting [34]. Additional clinical trials of IO across treatment strategies are ongoing, and updated cost–effectiveness analyses will be needed.

Interpretation of these findings should consider certain inherent limitations of cost–effectiveness modeling. Extrapolation of survival data from clinical trials leads to uncertainty around the incremental benefit of the intervention beyond the length of the trial follow-up period. However, the extrapolation used the most conservative distribution. Additionally, health utilities were not collected in IMpower010, and model inputs were instead derived from the literature or other clinical trials, which may have introduced bias related to underlying differences between study populations. Additional healthcare costs associated with LR and MR that are unrelated to treatment or follow-up were not included in the base-case analysis. Indirect costs, such as work productivity, were also not considered. However, including these costs would likely decrease the incremental cost of atezolizumab and, thus, further improve the observed ICER for atezolizumab versus BSC.

Conclusion

Atezolizumab provided 1.05 additional QALYs at an additional cost of $48,956 with an ICER of $46,859/QALY, which is well below the $150,000/QALY WTP threshold. Sensitivity analysis showed the probability of atezolizumab being cost-effective to be ≈91%. The relatively higher drug costs associated with atezolizumab during the DFS health state were partly offset by delaying progression and lowering treatment costs in later metastatic health states. Atezolizumab is also cost-effective in the Medicare population, with an ICER of $48,512/QALY versus BSC. In conclusion, atezolizumab is cost-effective versus BSC for the adjuvant treatment of resected patients with PD-L1+ stage II-IIIA NSCLC at a WTP threshold of $150,000/QALY. Adjuvant atezolizumab is likely to offer additional value to patients and payers as the new standard of care in this setting.