Diagnostic Value and Cost-Effectiveness of Next Generation Sequencing–Based Testing for Treatment of Patients with Advanced/Metastatic Non-Squamous Non-Small Cell Lung Cancer in the United States

The study evaluated the diagnostic value and cost-effectiveness of next generation sequencing (NGS)-based testing versus various combinations of single-gene tests (SGTs) for selection of first-line treatment for patients with advanced/metastatic non-squamous non-small cell lung cancer in the United States. A dynamic decision analysis model was developed comparing NGS versus SGT from a payer perspective. Inputs were obtained from published sources and included diagnostic performance, biomarker-positive disease rates, biomarker-directed recommendations for treatment, and survival outcomes. Costs were reported in 2020 US dollars. In the base case, NGS improved the detection of actionable biomarkers by 74.4%, increased the proportion of patients receiving biomarker-driven therapy by 11.9%, and decreased the proportion of patients with biomarker-positive disease receiving non–biomarker-driven first-line treatment by 40.5%. The incremental cost-effectiveness ratio per life-year gained of NGS testing versus SGT was $7224 (excluding post-diagnostic costs); the incremental cost-effectiveness ratio for NGS-directed therapy was $148,786 versus SGT-directed therapy. Sensitivity analyses confirmed the robustness of these findings; survival outcomes and targeted therapy costs had the greatest impact on results. Testing strategies with NGS are more comprehensive in the detection of actionable biomarkers and can improve the proportion of patients receiving biomarker-driven therapies. NGS testing may provide a cost-effective strategy for advanced/metastatic non-squamous non-small cell lung cancer; however, the value of NGS-directed therapy varies by the willingness-to-pay threshold of the decision-maker.

Lung cancer is one of the most common cancers and the leading cause of cancer-related mortality both globally and in the United States (Centers for Disease Control and Prevention, https://www.cdc.gov/cancer/uscs/index.htm, last accessed July 16, 2020). Non-small cell lung cancers (NSCLC) account for 80% to 85% of lung cancers; 57% of patients are diagnosed with metastatic disease, which has a 5-year survival rate of 6.9% (National Cancer Institute, https://seer.cancer.gov/archive/csr/1975_2017, last accessed December, 2021). Personalized medicine is rapidly becoming the primary treatment strategy for patients diagnosed with advanced or metastatic NSCLC.Atherly A.J. Camidge D.R. The cost-effectiveness of screening lung cancer patients for targeted drug sensitivity markers. In the last decade, there has been significant progress in the identification of molecular alterations that contribute to the pathogenesis of NSCLC.Parker D. Belaud-Rotureau M.A. Micro-cost analysis of ALK rearrangement testing by FISH to determine eligibility for crizotinib therapy in NSCLC: implications for cost effectiveness of testing and treatment. The use of biomarker testing to detect the presence of actionable genomic and/or nongenomic biomarkers is now considered the standard of care among patients diagnosed with advanced or metastatic NSCLCAtherly A.J. Camidge D.R. The cost-effectiveness of screening lung cancer patients for targeted drug sensitivity markers.,Hinrichs J.W. van Blokland W.T. Moons M.J. Radersma R.D. Radersma-van Loon J.H. de Voijs C.M. Rappel S.B. Koudijs M.J. Besselink N.J. Willems S.M. de Weger R.A. Comparison of next-generation sequencing and mutation-specific platforms in clinical practice. (National Comprehensive Cancer Network, https://www.nccn.org/professionals/physician_gls/default.aspx, last accessed June, 2020). The results of biomarker testing are subsequently used to select appropriate systemic therapy in these patients.Next generation sequencing (NGS)-based comprehensive genomic testing or an approach utilizing various combinations of single-gene tests (SGTs) could be used to detect genomic biomarkers. NGS-based testing offers the ability to evaluate the presence of numerous genomic biomarkers simultaneously whereas commonly used SGT strategies include use of multiple individual biomarker tests conducted sequentially or in parallel and can include multigene panels. Recent advancements in the identification of actionable biomarkers and the growing availability of targeted therapies in NSCLC, however, have drawn attention to limitations associated with using an SGT approach including limited tissue samples and the need for rebiopsies.Daniels M. Goh F. Wright C.M. Sriram K.B. Relan V. Clarke B.E. Duhig E.E. Bowman R.V. Yang I.A. Fong K.M. Whole genome sequencing for lung cancer.,Popper H.H. Tímár J. Ryska A. Olszewski W. Minimal requirements for the molecular testing of lung cancer.NGS-based comprehensive genomic profiling (CGP) with both large and small panels can detect four classes of genomic alterations, including base substitutions, short insertions and deletions (indels), amplifications, and complex fusions/rearrangements in hundreds to thousands of gene targets simultaneously, including potential resistant mutations to rule out ineffective treatment options.Signorovitch J. Zhou Z. Ryan J. Anhorn R. Chawla A. Budget impact analysis of comprehensive genomic profiling in patients with advanced non-small cell lung cancer.,McKenzie A.J. Dilks H.H. Jones S.F. Burris 3rd, H. Should next-generation sequencing tests be performed on all cancer patients?. Given the limitations of SGTs and the recognition that NGS-directed treatment can lead to improved patient outcomes,John A. Shah R.A. Wong W.B. Schneider C.E. Alexander M. Value of precision medicine in advanced non-small cell lung cancer: real-world outcomes associated with the use of companion diagnostics.,Zhu Y. Han Y. Bhandari N.R. Hess L.M. Genomic biomarker testing, treatments, and survival outcomes among patients with advanced or metastatic NSCLC in the US: a retrospective cohort study. National Comprehensive Cancer Network (NCCN) Clinical Practice Guidelines recommend a broad, panel-based approach, typically performed with NGS (National Comprehensive Cancer Network, https://www.nccn.org/professionals/physician_gls/default.aspx, last accessed June, 2020).Despite the data showing the clinical benefits of NGS-based testing in accordance with NCCN guidelines,Zhu Y. Han Y. Bhandari N.R. Hess L.M. Genomic biomarker testing, treatments, and survival outcomes among patients with advanced or metastatic NSCLC in the US: a retrospective cohort study.,Singal G. Miller P.G. Agarwala V. Li G. Kaushik G. Backenroth D. Gossai A. Frampton G.M. Torres A.Z. Lehnert E.M. Bourque D. O'Connell C. Bowser B. Caron T. Baydur E. Seidl-Rathkopf K. Ivanov I. Alpha-Cobb G. Guria A. He J. Frank S. Nunnally A.C. Bailey M. Jaskiw A. Feuchtbaum D. Nussbaum N. Abernethy A.P. Miller V.A. Association of patient characteristics and tumor genomics with clinical outcomes among patients with non–small cell lung cancer using a clinicogenomic database [Erratum appeared in: JAMA. 2020;323:480]. approximately 10%Robert N.J. Nwokeji E.D. Espirito J.L. Chen L. Karhade M. Evangelist M.C. Spira A.I. Neubauer M.A. Bullock S.A. Coleman R.L. on behalf of MYLUNG Consortium Collaborators: The U.S. Oncology Network and sponsors
Biomarker tissue journey among patients (pts) with untreated metastatic non-small cell lung cancer (mNSCLC) in the U.S. Oncology Network community practices. to 25%Bruno D.S. Hess L.M. Li X. Wen Su E. Zhu Y. Patel M. Racial disparities in biomarker testing and clinical trial enrollment in non-small cell lung cancer (NSCLC). of patients with advanced or metastatic NSCLC may not receive any biomarker testing due to a variety of factors.

A key knowledge gap exists regarding the cost-effectiveness of NGS testing and the value of NGS-directed therapy. Ideally, such an analysis would tabulate all the costs (eg, the change in cost associated with using one NGS panel versus multiple SGTs; the changes in treatment-related costs stemming from increased rates of biomarker detection) and benefits (eg, potential decrease in need for rebiopsy, improved survival associated with increased utilization of targeted therapies) associated with NGS compared with existing SGT practice.

Limited prior studies have reported cost-effectiveness of testing for specific genomic biomarkers [eg, epidermal growth factor receptor (EGFR), anaplastic lymphoma kinase (ALK)] in patients with NSCLC in the United StatesAtherly A.J. Camidge D.R. The cost-effectiveness of screening lung cancer patients for targeted drug sensitivity markers.,Romanus D. Cardarella S. Cutler D. Landrum M.B. Lindeman N.I. Gazelle G.S. Cost-effectiveness of multiplexed predictive biomarker screening in non-small-cell lung cancer.,Steuten L. Goulart B. Meropol N.J. Pritchard D. Ramsey S.D. Cost effectiveness of multigene panel sequencing for patients with advanced non-small-cell lung cancer.; however, no studies evaluating cost-effectiveness of NGS versus SGT in the US settings were identified. One study by Steuten et alSteuten L. Goulart B. Meropol N.J. Pritchard D. Ramsey S.D. Cost effectiveness of multigene panel sequencing for patients with advanced non-small-cell lung cancer. in 2019 compared multigene panel sequencing with SGT but did not include NGS specifically. Several studies have assessed the budget impact of NGS-based testing for patients diagnosed with NSCLC from the US healthcare payer perspective.Signorovitch J. Zhou Z. Ryan J. Anhorn R. Chawla A. Budget impact analysis of comprehensive genomic profiling in patients with advanced non-small cell lung cancer.,Pennell N.A. Mutebi A. Zhou Z.-Y. Ricculli M.L. Tang W. Wang H. Guerin A. Arnhart T. Dalal A. Sasane M. Wu K.Y. Culver K.W. Otterson G.A. Economic impact of next-generation sequencing versus single-gene testing to detect genomic alterations in metastatic non–small-cell lung cancer using a decision analytic model.,Yu T.M. Morrison C. Gold E.J. Tradonsky A. Arnold R.J.G. Budget impact of next-generation sequencing for molecular assessment of advanced non-small cell lung cancer. Nevertheless, these studies did not incorporate key factors such as sensitivity and specificity of the various testing options,Signorovitch J. Zhou Z. Ryan J. Anhorn R. Chawla A. Budget impact analysis of comprehensive genomic profiling in patients with advanced non-small cell lung cancer.,Pennell N.A. Mutebi A. Zhou Z.-Y. Ricculli M.L. Tang W. Wang H. Guerin A. Arnhart T. Dalal A. Sasane M. Wu K.Y. Culver K.W. Otterson G.A. Economic impact of next-generation sequencing versus single-gene testing to detect genomic alterations in metastatic non–small-cell lung cancer using a decision analytic model.,Yu T.M. Morrison C. Gold E.J. Tradonsky A. Arnold R.J.G. Budget impact of next-generation sequencing for molecular assessment of advanced non-small cell lung cancer. or failed to include long-term treatment benefitsPennell N.A. Mutebi A. Zhou Z.-Y. Ricculli M.L. Tang W. Wang H. Guerin A. Arnhart T. Dalal A. Sasane M. Wu K.Y. Culver K.W. Otterson G.A. Economic impact of next-generation sequencing versus single-gene testing to detect genomic alterations in metastatic non–small-cell lung cancer using a decision analytic model. that are critical to accurately estimate diagnostic and cost-effectiveness outcomes. Only one study accounted for sufficient tissue sample for testing and sequencing of multiple tests, and attempted to capture variability in SGT strategies.Pennell N.A. Mutebi A. Zhou Z.-Y. Ricculli M.L. Tang W. Wang H. Guerin A. Arnhart T. Dalal A. Sasane M. Wu K.Y. Culver K.W. Otterson G.A. Economic impact of next-generation sequencing versus single-gene testing to detect genomic alterations in metastatic non–small-cell lung cancer using a decision analytic model. As additional novel biomarkers are identified and newer treatments become available, existing economic models are quickly becoming obsolete.

The objective of this study was to develop an interactive model that would allow a user to evaluate the diagnostic value and cost-effectiveness of NGS-based testing versus various combinations of SGTs as well as to incorporate emerging biomarkers and the value of NGS-directed selection of first-line treatment for patients with advanced or metastatic non-squamous NSCLC in the United States.

Materials and MethodsModel StructureA dynamic economic model was developed in Microsoft Excel (Microsoft, Redmond, WA) using a decision analysis approach to compare biomarker testing strategies utilizing NGS-based methods versus SGT approaches (Figure 1). NGS strategies can include CGPs and small NGS panels, whereas SGT strategies can include multiple SGTs conducted in parallel and multigene panels to replicate current practice. Additionally, the model considered the health and cost consequences of treatment with targeted therapies among patients with biomarker-positive disease. The model employed a custom-designed Excel function implemented in Visual Basic for Applications to generate all possible outcomes from testing strategies in a heterogenous population and to calculate the probabilities of those outcomes. The model is dynamic in that a user may add emerging biomarkers, modify the distribution of testing strategies, and tailor utilization patterns and costs for site-specific scenarios in real time. The model was designed to consider costs from the perspective of a US third-party payer such as insurance companies, health maintenance organizations (ie, HMOs), and employers. The model used a lifetime time horizon, capturing all costs and benefits accrued until the entire simulated cohort of patients had died. The interactive model is available from Eli Lilly and Company.Figure thumbnail gr1

Figure 1Model Structure. Note: The calculation follows a decision tree structure where the likelihood of possible outcomes is represented by a node, and each potential outcome is represented by a branch. Example calculation 1: If the test strategy includes testing for EGFR and ALK, but the patient does not have either of these genomic alterations, the probability that the test returns positive for EGFR is (1 − specificityEGFR) × (specificityALK). Example calculation 2: If the test strategy includes testing for EGFR and ALK, but the patient does not have either of these genomic alterations, the probability that the test returns negative for any biomarker is (specificityEGFR) × (specificityALK). FP, false-positive; IO, immuno-oncology; NGS, next generation sequencing; NSCLC, non-small cell lung cancer; PFS, progression-free survival; SGT, single-gene test; TT, targeted therapy.

Study Population

The population included patients with non-squamous NSCLC who were diagnosed with advanced or metastatic disease. Patients were assumed to be treatment-naive for advanced disease. Therefore, the model is limited to the first-line treatment setting, and only one treatment regimen was assigned per patient when evaluating testing-directed therapies.

Patients entering the model were modeled as a cohort. Prevalence rates for the following biomarkers were obtained from the published literature based on those that were recommended in NCCN guidelines at the time of analysis (v6.2020) (National Comprehensive Cancer Network, https://www.nccn.org/professionals/physician_gls/default.aspx, last accessed June, 2020), including EGFR, ALK, ROS1, BRAF, KRAS, NTRK, MET, RET, and PD-L1. It was assumed that patients could only have one genomic alteration at a time. The rate of programed death-ligand 1 (PD-L1) expression ≥1% varied by presence or absence of each biomarker (Table 1).

Table 1Study Population: Prevalence of Actionable Biomarkers in Patients with NSCLC (Base-Case Scenario)

ALK, anaplastic lymphoma kinase; BRAF, B-Raf proto-oncogene; EGFR, epidermal growth factor receptor; KRAS, Kirsten rat sarcoma viral oncogene homologue; MET, mesenchymal to epithelial transition; NCCN, National Comprehensive Cancer Network; NSCLC, non-small cell lung cancer; NTRK, neurotrophic tropomyosin-related kinase; PD-L1, programed death-ligand 1; RET, rearranged during transfection; ROS-1, repressor of Silencing-1.

ComparatorsThe model structure allows for various combinations of SGT- and NGS-based strategies to be compared. The comparators included in the model are summarized in Table 2. In the base case, it was assumed that 100% of patients underwent biomarker testing in both groups (SGT and NGS); the distribution of testing strategies was informed by real-world utilization of various testing approaches and included common SGTs with PD-L1 in parallel or alone.Hess L.M. Zhu Y.E. Han Y. Van Hook E.C. Biomarker testing for patients with advanced/metastatic non-small cell lung cancer (NSCLC) in academic and community-based practices in the United States (US). NGS-based strategies were assumed to be equally distributed between CGP and small NGS panels, both of which tested for the same biomarkers in the base case. Different testing methods are available for certain biomarkers (eg, fluorescence in situ hybridization or PCR; DNA or RNA assay). The model included only one method for each biomarker in the base case; however, it is modifiable to include additional biomarker tests by the user.

Table 2Comparators: Sensitivity and Specificity of Biomarker Tests

ALK, anaplastic lymphoma kinase; BRAF, B-Raf proto-oncogene; CGP, comprehensive genomic profiling; EGFR, epidermal growth factor receptor; FISH, fluorescence in situ hybridization; KRAS, Kirsten rat sarcoma viral oncogene homologue; MET, mesenchymal to epithelial transition; NGS, next generation sequencing; NSCLC, non-small cell lung cancer; PCR, polymerase chain reaction; PD-L1, programed death-ligand 1; RET, rearranged during transfection; SGT, single-gene test.

The accuracy of commercially available tests and laboratory-developed tests in identifying a true-positive/negative biomarker varies. For the base-case analysis, the sensitivity and specificity of tests were based on commercially available tests. However, the sensitivity and specificity of some less commonly used tests could not be found, so information on known tests that use the same technology was applied. Due to lack of data, it was assumed that the sensitivity and specificity of each genomic alteration included in NGS testing were equivalent to those of the corresponding SGT.

The probability of correctly diagnosing patients with actionable biomarkers was based on biomarker prevalence and test sensitivity/specificity. Test results determined the assignment of first-line treatment a patient would receive under the assumption of complete concordance with NCCN guidelines, such that a patient with an identified biomarker would always be assigned to the targeted therapy for that biomarker. For patients with both a genomic alteration and high PD-L1 expression, the genomic alteration took precedence in guiding treatment selection, as recommended by NCCN guidelines. Untested patients who did not undergo biomarker testing or those without positive biomarker findings were assumed to receive platinum-based chemotherapy. Therapy provided to patients with false-negative or false-positive biomarker test results was considered to be suboptimal.

Cost InputsThe model included costs in three key categories: i) biomarker testing, ii) first-line treatment, and iii) routine management of advanced or metastatic NSCLC. Diagnostic testing costs were based on reimbursement costs (expected to cover the range of costs such as material and labor) from the Current Procedural Terminology coding system (Table 3). Unit costs for drugs (ie, the cost per box of vials/pills as applicable) used in first-line setting were obtained from IBM Watson Redbook.IBM Watson Redbooks
IBM Micromedex RED BOOK [Internet]. These unit costs were combined with drug dosing and schedules based on published product prescribing information and from clinical trial data to derive the cost of the entire treatment regimen (Table 4). The model also considered other, nonpharmacological healthcare costs incurred during routine management of disease in both pre- and post-progression phases, including consultations with a general physician or specialist, nurse time, routine scans, and hospital stays. Based on published literature (Healthcare Cost and Utilization Project, http://hcupnet.ahrq.gov, last accessed April 23, 2020; National Institute for Health and Care Excellence, https://www.nice.org.uk/guidance/ta520, last accessed July 16, 2020; Optum, Eden Prairie, MN), the monthly cost of routine management of NSCLC was estimated to be $1150 before disease progression and $1114 after disease progression.

Table 3Costs: Diagnostic Testing

ALK, anaplastic lymphoma kinase; BRAF, B-Raf proto-oncogene; CGP, comprehensive genomic profiling; DNA, deoxyribonucleic acid; EGFR, epidermal growth factor receptor; FISH, fluorescence in situ hybridization; IHC, immunohistochemistry; KRAS, Kirsten rat sarcoma viral oncogene homologue; MET, mesenchymal to epithelial transition; MSI, microsatellite instability; NGS, next generation sequencing; PCR, polymerase chain reaction; PD-L1, programed death-ligand 1; ROS-1, repressor of Silencing-1; SGT, single-gene test; TMB, tumor mutational burden.

Table 4Clinical Inputs (in Months) and Costs of First-Line Treatments (in 2020 US Dollars)

ALK, anaplastic lymphoma kinase; BRAF, B-Raf proto-oncogene; EGFR, epidermal growth factor receptor; KRAS, Kirsten rat sarcoma viral oncogene homologue; MET, mesenchymal to epithelial transition; NDC, National Drug Code; NTRK, neurotrophic tropomyosin-related kinase; OS, overall survival; PD-L1, programed death-ligand 1; PFS, progression-free survival; Q3W, every 3 weeks; RET, rearranged during transfection; ROS-1, repressor of Silencing-1.

Utility EstimatesHealthcare utilities measure the impact of cancer and its treatment on patient quality of life. These values are expressed as a value between 1 and 0, where 1 represents perfect health, and 0 represents the worst health state possible (York Health Economics Consortium, York, UK). Mathematically, utilities act as a multiplier that can be used to derive quality-adjusted outcomes (eg, a patient living 5 years with a utility value of 0.6 has lived 3 quality-adjusted years). In this model, utilities were linked to disease status (pre- or post-progression). Utility value inputs were obtained from the values provided in the National Institute for Health and Care Excellence technology appraisal of nivolumab for advanced non-squamous NSCLC (National Institute for Health and Care Excellence, https://www.nice.org.uk/guidance/ta713, last accessed August, 2021) and applied consistently across comparators. Patients whose disease had not yet progressed (during first-line treatment, pre-progression) were assigned a utility of 0.713 (95% CI, 0.642-0.784) and patients were assigned a utility of 0.569 (95% CI, 0.512-0.626) after disease progression (post-first line until death).Clinical InputsTreatment-specific clinical inputs included progression-free survival and overall survival (OS). Expected mean estimates of progression-free survival and OS were implemented by converting the published median survival from a clinical trial with the application of exponential distribution (ie, by assuming a constant rate of progression/death) for NCCN-preferred agents for each actionable biomarker, to reflect optimal outcomes of treatment (Table 4). Naive outcomes were included; no further adjustment was made to account for the heterogeneity in patient populations across studies and the variability between biomarker-specific patient groups assigned to individual treatments.Model Outcomes

NGS- and SGT-based diagnostic outcomes included the proportion of patients correctly identified as having or not having an actionable biomarker and the proportion of patients receiving guideline-recommended or suboptimal therapy. Survival outcomes were reported as life-years (LYs) and quality-adjusted life-years (QALYs; where the years of survival are multiplied by the utility value, as described above).

All costs were adjusted to 2020 US dollars and, in accordance with best practices for economic modeling, were discounted at an annual rate of 3%.Caro J.J. Briggs A.H. Siebert U. Kuntz K.M. ISPOR-SMDM Modeling Good Research Practices Task Force
Modeling good research practices--overview: a report of the ISPOR-SMDM Modeling Good Research Practices Task Force-1. Discounting of costs and health outcomes was based on the observation that, given the choice between experiencing the good health now or later, people prefer the former option.Hunink M. Glasziou P. Siegel J. Weeks J. Pliskin J. Elstein A. Weinstein M. Decision Making in Health and Medicine: Integrating Evidence and Values.

The key outcome of the model used to assess cost-effectiveness was the incremental cost-effectiveness ratio (ICER). The ICER represents the additional cost of an intervention per unit of improvement in health outcomes gained [in this case, per correctly identified biomarker, per LY, or per QALY gained (York Health Economics Consortium)].

Therefore, ICERs per LY gained and per QALY gained for NGS versus SGT were generated for the scenarios where post-diagnosis costs were included or excluded. Additionally, results were summarized as cost per additional patient receiving guideline-recommended therapies, and cost per additional patient with a correctly identified biomarker.

Scenario AnalysesGiven the variability of clinical practice, alternative strategies were explored for illustrative purposes; base-case assumptions were retained in these scenarios unless specified otherwise. In scenario 1, NGS (patients received comprehensive testing plus PD-L1) was compared with an alternate SGT strategy (patients receive SGT for ALK, EGFR, and ROS-1 plus PD-L1). In scenarios 2 and 3, testing 100% of the population with NGS-based strategies was compared with varied combinations of SGT strategies assuming a proportion of patients not tested, reflecting the lack of complete biomarker testing rates in current practice.Hess L.M. Zhu Y.E. Han Y. Van Hook E.C. Biomarker testing for patients with advanced/metastatic non-small cell lung cancer (NSCLC) in academic and community-based practices in the United States (US). A summary of the base case and scenarios 1 to 3 are presented in Table 5.

Table 5Comparators: Base-Case and Scenario Analyses

ALK, anaplastic lymphoma kinase; BRAF, B-Raf proto-oncogene; EGFR, epidermal growth factor receptor; KRAS, Kirsten rat sarcoma viral oncogene homologue; MET, mesenchymal to epithelial transition; NGS, next generation sequencing; NTRK, neurotrophic tropomyosin-related kinase; PD-L1, programed death-ligand 1; RET, rearranged during transfection; ROS-1, repressor of Silencing-1; SGT, single-gene test.

Sensitivity Analyses

A series of deterministic sensitivity analyses (DSAs) were conducted that evaluated each potential parameter that may affect the results and subsequent treatment outcomes. DSAs are conducted in accordance with best practice guidelines for economic modeling to determine which parameters have the greatest impact on the ICER. If appropriate data to calculate 95% CIs were not available, parameters were varied by ±5% for test sensitivities/specificities and ±20% for others. DSAs were conducted by systematically varying parameters from the base case one at a time while holding other parameters at their base-case values. All DSAs included the costs of biomarker testing, and all post-diagnostic costs and outcomes.

A probabilistic sensitivity analysis (PSA) was run to consider the impact of multiple parameters on the model result. A PSA simultaneously (York Health Economics Consortium) evaluates the uncertainty of each input parameter. The analysis included 1000 second-order Monte Carlo simulations (York Health Economics Consortium): for each simulation, all parameters included in the model (eg, utilities, costs, and clinical outcomes) were varied along a distribution representing a continuum of plausible outcomes given the known mean estimate and CIs for each model parameter. The distribution used for the PSA was informed by parameter characteristics: parameters were sampled from either beta distributions for parameters restricted to values between 0 and 1 such as sensitivities/specificities, or gamma distribution for parameters with values restricted by 0, such as costs and number of events. In cases where standard errors were not available or calculable, a standard error equal to 20% of the mean was assumed.

Model ValidationThis model and all its components were peer reviewed, including an evaluation of face validity of the structure, evidence, problem formulation, and results, in accordance with International Society of Pharmacoeconomic Outcomes Research Practice Guidelines.Caro J.J. Briggs A.H. Siebert U. Kuntz K.M. ISPOR-SMDM Modeling Good Research Practices Task Force
Modeling good research practices--overview: a report of the ISPOR-SMDM Modeling Good Research Practices Task Force-1. An oncologist evaluated the model for face and content validity specifically for its practical use, functional ease, and scientifically valid inputs for population-based healthcare decision-making. Lastly, a modeler independent of this study or any aspect of model development validated the logical structure of the model, the mathematical formulae and sequences of calculations, and the values of numbers supplied as model inputs.ResultsBase CaseNGS Testing

From a third-party payer perspective, the model predicted that the use of NGS-based diagnostic strategies versus SGT approaches would improve the detection of actionable biomarkers by a relative 74.4%, increase the proportion of patients who receive biomarker-driven therapy per guidelines by a relative 11.9%, and decrease the proportion of patients who initially receive suboptimal first-line treatment due to incorrect biomarker test results by a relative 40.5%.

The use of NGS-based testing incurred an additional diagnostic cost per patient of $2036 from the payer perspective. This translated to a cost per additional patient receiving biomarker-driven therapy of $22,091 and a cost of care per additional patient with one or more correctly identified biomarkers of $10,677 (Table 6).

Table 6Results: Diagnosis and Cost-Effectiveness Results of Base Case and Scenarios 1 to 3

LY, life-year; NGS, next generation sequencing; PFS, progression-free survival; QALY, quality-adjusted life-year; SGT, single-gene test.

NGS-Directed TherapyNGS-directed treatment strategies were associated with per-patient LY gains of 0.28 and QALY gains of 0.18 versus treatment informed by SGT strategies. NGS-directed treatment incurred increased total costs (including costs of diagnosis, first-line treatment, and routine medical resource use) by 16% versus SGT-directed strategies, driven by increased cost of first-line treatment (86% of total cost) (Table 7).

Table 7Results: Detailed Costs and Outcomes by Comparator Arm (Base Case)

留言 (0)

沒有登入
gif