A Framework for the Cost-Effectiveness Analysis of Novel Biomarker Testing in Cardiovascular Disease

Objectives: Individuals are often prioritised for preventive cardiovascular disease (CVD) therapy based on 10-year risk of experiencing a primary CVD event. In Scotland, this risk is estimated with the ASSIGN risk score. Recent research has focused on identifying novel biomarkers to improve CVD risk diagnosis. The objective of this study was to develop a framework for the cost-effectiveness analysis (CEA) of novel biomarker testing given the inherent sparsity of data related to novel biomarkers. The framework was applied in the CEA of the urinary proteomic biomarker HF1. Methods: Gompertz regression was performed on data from the FLEMish Study on Environment, Genes, and Health Outcomes to establish hazard ratios associated with HF1 and key CVD outcomes. The results from this analysis were used to update the Scottish CVD Policy Model (SCVDPM), a previously-published decision-analytic model, and the ASSIGN score. The SCVDPM was employed to estimate Scottish population-level health and cost outcomes associated with prioritising individuals for preventive CVD therapy using the traditional and updated ASSIGN risk scores. Sensitivity analyses established a price at which HF1 testing would become cost-effective. Results: A framework was developed for the CEA of novel CVD biomarkers. HF1 was positively associated with risk of non-fatal coronary heart disease (adjusted hazard ratio, 1.69; 1.11-2.57) and combined CVD (adjusted hazard ratio 1.48; 1.03-2.11) events. Implementing HF1 testing for intermediate risk individuals was estimated to produce 437 QALYs at an incremental cost of £90,600,000 (ICER, £207,000/QALY). HF1 testing would become costeffective at a price of £40/person, down 89% from £350/person. Conclusions: External datasets can be used to update existing decision-analytic models and CVD risk scores, enabling the CEA of novel biomarkers. Prioritising individuals for preventive CVD therapy with an updated ASSIGN score which includes HF1 as a covariate would not be cost-effective compared to current practice without a large reduction in the price of testing