JAK2 V617F as a Marker for Long-Term Disease Progression and Mortality in Polycythemia Vera and its Role in Economic Modeling

Abstract

**Background:** In order to facilitate sound economic evaluations of novel treatments, health-economic models of polycythemia vera (PV) must combine effects on surrogate endpoints in trials with disease progression (DP) and mortality in long-term cohort data. **Objective:** We validate an economic model for PV that uses Janus Kinase 2 (JAK2) burden as a surrogate endpoint to predict DP (thrombosis, myelofibrosis, and acute leukemia) and overall survival (OS) based on progression-specific mortality. **Methods:** Long-term observational studies that include information about baseline JAK2 burden were identified via PubMed searches and used to validate the model. Kaplan-Meier (KM) OS curves were extracted using a digitizing software. External validity of the model was analyzed by visually comparing OS curves of the model with the KM curves of the included studies, as well as calculating differences in mean OS estimated as area under the curve (AUC). **Results:** The model’s predictions of cumulative DP were somewhat lower than the published studies. Over 20 years’ time, our base case model predicted a mean OS for a PV patient (15.0–16.5 years), which was in line with the published studies (15.8–17.5 years). Modeled mean OS was almost two years longer (1.6–1.9 years) for patients with JAK2 <50% than patients with JAK2 ≥50%. Only three long-term observational studies that satisfied the predefined criteria were found and could be used in the validation, but these studies did not capture JAK2 evolution over time. Improved model predictions of DP and mortality based on the longitudinal evolution of JAK2 could be derived from real-world data sources. Such data are currently scarce and future observational studies should be designed to capture the long-term impact of JAK2 on DP and mortality in PV. **Conclusions:** Our model, based on JAK2 burden as a marker for DP, generated OS estimations that are in line with results of published data.