Accepted Manuscript Cost-Effectiveness of Thoracic Radiotherapy for Extensive Stage Small Cell Lung Cancer Using Evidence from the CREST Trial

Introduction: The Chest Radiotherapy Extensive Stage Trial (CREST ) showed that adding thoracic radiotherapy (TRT) to the standard treatme nt (ST) paradigm of chemotherapy and prophylactic cranial irradiation improves overall ( OS) and progression-free survival (PFS), in extensive stage small cell lung cancer (ES-SCLC) pa tients. We evaluated the cost-effectiveness of adding TRT to ST in ES-SCLC patients. Methods: A cost-utility analysis was performed comparing TRT plus ST versus ST alone. The base case time horizon was 24 months; consistent wi th the maximum PFS reported in the CREST. OS was partitioned into two health states: P FS and post-progression survival. The proportion of patients in each health state over ti me was estimated by fitting parametric probability distributions to the CREST survival dat a. Costs were from a United States health care payer perspective and utilities were derived from t he literature. Incremental cost-effectiveness ratios (ICERs) were calculated per quality-adjusted lif year (QALY) using a 3% discount rate. Sensitivity analyses addressed uncertainty in key v ariables. M AN US CR IP T AC CE PT ED ACCEPTED MANUSCRIPT Cost-Effectiveness of TRT for ES-SCLC Results: In the base-case, adding TRT to ST was both cost-sa ving nd more effective thereby strongly dominating ST alone. At willingness-to-pay thresholds of $50,000, $100,000, and $200,000/QALY, TRT was preferred 68, 81, and 96 per cent of the time, respectively. In the lifetime scenario analysis, the TRT ICER increased to $194,726/QALY. Conclusions: Using the actual follow-up interval reported in th e CREST, adding TRT to ST strongly dominates a strategy of ST alone in ES-SCL C patients. Since the long-term survival benefit of TRT is small relative to ongoing costs o f progressive metastatic disease, we estimate less favorable ICERs for TRT over a lifetime horizo n. Introduction Small Cell Lung cancer (SCLC) accounts for approxi mately 15% of all bronchogenic carcinomas and about 70% of these patients present with extensive stage (ES) disease. 1 The initial treatment of ES-SCLC typically involves 4 t o 6 cycles of platinum-based chemotherapy. 2,3 Patients demonstrating a favorable response may be offer d prophylactic cranial irradiation (PCI), which lowers the risk of symptomatic brain m etastases and improves disease-free and overall survival (OS). 4,5 Despite these advances, treatment outcomes remain s uboptimal with a median OS of approximately 10 to 12 months. 6-8 In addition, thoracic control is inadequate with u p to 90% of patients progressing within the chest after treatme n with chemotherapy alone. 5,9 Administering thoracic radiotherapy (TRT) concurrently with chemo therapy improves OS for those with limited stage disease, but there is less compelling evidenc e that TRT is beneficial in the metastatic setting. 10-14 Although the results of a single-institution rando mized trial, a non-randomized phase 2 trial, and a few retrospective studies have sugge sted an advantage to TRT in this cohort, this M AN US CR IP T AC CE PT ED ACCEPTED MANUSCRIPT Cost-Effectiveness of TRT for ES-SCLC information has not changed clinical practice. 15-18 To more definitively evaluate TRT in ESSCLC, the multinational phase 3 randomized controll ed Chest Radiotherapy Extensive Stage Trial (CREST) assigned 498 patients who demonstrate d ny response to induction chemotherapy to receive TRT and PCI or PCI alone. 9 At a median follow up of 24 months, adding TRT to ST resulted in improved 2-year progression-free surviv al (PFS) and OS without additional adverse events. 9 In an environment of limited healthcare resources a nd with United States (US) national expenditures for lung cancer projected to reach bet we n $15 to 19 billion by 2020, healthcare decision-making authorities will increasingly requi re nformation on the cost-effectiveness of novel treatments such as the use of TRT for ES-SCLC to inform health policies. 19,20 The nontrivial differences in costs, efficacy and quali ty-of-life effects associated with the variety of treatment programs for metastatic lung cancer under score the importance of quantifying their relative cost-effectiveness so that funds can be op timally allocated to those interventions that are more cost-effective. The purpose of this study was to evaluate the cost-effectiveness of adding TRT to chemotherapy and PCI in patients with ES-SCL C treated according to the CREST. Materials and Methods Base-Case Population This analysis was based on the patient population of the CREST: median age 63 with ESSCLC (defined as disease beyond the hemithorax, hil ar, mediastinal or supraclavicular lymph nodes), World Health Organization (WHO) performance status 0-2, and any response to 4 to 6 cycles of platinum/etoposide chemotherapy. 9 Patients with brain, leptomeningeal or pleural M AN US CR IP T AC CE PT ED ACCEPTED MANUSCRIPT Cost-Effectiveness of TRT for ES-SCLC metastases, prior TRT, and those with bulky thoraci c disease precluding the safe delivery of TRT were excluded. 9 Decision Model Using Microsoft Excel (Microsoft Corporation, Redm ond, WA), a partitioned survival model was developed to estimate the direct medical costs from a US health care payer perspective, expected life years (LYs), and quality -adjusted life years (QALYs), associated with adding TRT to the standard treatment (ST) paradigm of chemotherapy and PCI compared to ST alone. As recommended by Levine et al, the base cas e time horizon was 24 months, consistent with the maximum PFS reported in the CREST, and not based on results projected into the future. However, the effect of a lifetime horizon was stud ied in a scenario analysis. 21 Based on well-established analytic methods for the economic evaluation of healthcare programs similar to Q-TWIST, overall survival was split into two health states: PFS and post-progression survival (PPS) in order to capture differences in costs and utilities within each health state. 22 The proportion of patients in each health state was bas ed on parametric survival functions with PPS assumed to be the difference between OS and PFS cal culated monthly. Patients entered the model in the PFS health state after completing indu ction chemotherapy. The TRT group received 30 Gy in 10 fractions using 3-D conformal technique s. It was assumed that PCI was administered after the completion of chemotherapy in both treatm ent groups and concurrently with TRT. As per the CREST and National Cancer Care Network (NCC ) guidelines, patients did not receive maintenance systemic therapy. 9,23 Those alive without progressive disease entered in to a surveillance program with physician visits and test ing performed according to NCCN guidelines. 23 Patients who progressed to the PPS health state un derwent a restaging workup and then received additional therapy or no active treat m nt based on patterns of care data from the M AN US CR IP T AC CE PT ED ACCEPTED MANUSCRIPT Cost-Effectiveness of TRT for ES-SCLC Surveillance, Epidemiology, and End Results (SEER) program. 24 We assumed no differences in adverse events between treatment strategies as was reported in the CREST. 9 Survival Data and Analytic Methods A single study estimate of effectiveness was used because the CREST is the largest, multicenter phase 3 randomized controlled trial to evaluate consolidative TRT in ES-SCLC patients who had all received ST according to curre nt national guidelines. 9,23 Using the method reported by Guyot et al, a close approximation of t he original individual patient time-to-event data (IPD) for PFS and OS was derived from the CRES T Kaplan-Meier survival functions. 25 Parametric probability distributions were independe ntly fit to the estimated IPD for OS and PFS for each treatment group in order to address uncert ainty associated with small patient numbers at the tails of the Kaplan-Meier survival curves. This al o provided a way to estimate costs and QALYs over a lifetime horizon. Curve fitting was pe rformed in R using the flexsurv package. 26 The best fitting distribution was selected based on fit statistics including Akaike and Bayesian information criterion, as well as by visual inspect ion. The log-normal and log-logistic distributions provided the best fit of the OS and P FS data, respectively, and were used in the baseline analyses (Figure 1). Direct Medical Costs: The cost of TRT was obtained from the 2016 Center for Medicare and Medicaid Services Physician Fee Schedule (CMSPFS) national payment am ount and was recognized at the start of month 1 (Table 1). 28 Post-treatment surveillance costs associated with the PFS health state were obtained from the 2016 CMSPFS and included a level 3 established patient office visit, chest/abdominal computed tomography scans, and lab work every 3 months during years 1 and 2, every 6 months during years 3 through 5, and ann u lly thereafter, consistent with NCCN M AN US CR IP T AC CE PT ED ACCEPTED MANUSCRIPT Cost-Effectiveness of TRT for ES-SCLC guidelines (Table 2). 23,28 At the time of progression an additional one-time cost was incurred for work-up and restaging of disease that was derived f rom the relapse patterns reported in the CREST and calculated using the 2016 CMSPFS (Table 2 ). Restaging consisted of a level 3 office visit, lab work, and imaging based on nation al standards. 23 The salvage treatment regimens used after progression were not specified in the CR EST. Therefore, costs of progressive disease were derived from an analysis of SEER-Medicare data that estimated the direct medical costs for continuing and terminal phases of care for ES-SCLC patients (Table 2). 24 We assumed that PPS costs were incurred through the second to last mont h f life and the terminal cost was assigned in the last month of life. When necessary, costs were inflated to 2016 US dollars using the medical care component of the US Chained Consumer Price Ind ex. Health State Utilities Patient preferences for the PFS and PP