Prediction of Thrombosis in Polycythemia Vera：Development and Validation of a Globally Applicable “ANA Model”

Background: Avoiding or reducing the occurrence of thrombus is still the focus of clinical management in patients with polycythemia vera. Although the conventional stratification system of thrombosis has improved clinical diagnosis and treatment, it still needs to be further improved. Objectives: The aim of this study is to develop a prediction model of thrombosis for polycythemia vera that is globally applicable with inexpensive and readily available predictors. Methods: We collected clinical information of 213 newly diagnosed patients with polycythemia vera from Qilu Hospital of Shandong University,then we randomly divided them into a training cohort and a validation cohort.Multivariable Cox regression analyses were conducted for the identification of thrombotic risk factors and model development. Results: We investigated data from 213 patients with polycythemia vera who were diagnosed according to the 2016 World Health Organization (WHO) criteria to identify covariates associated with thrombosis.Multivariable analyses indicated that age ≥60 years (hazard ratio [HR] 2.48, 95% CI 1.00-6.15, P=0.049),neutrophil count >8×10^9/L (HR 4.43, 95% CI 1.44-13.60, P=0.009),arterial thrombosis at or prior to diagnosis (HR 2.93, 95% CI 1.11-7.76, P =0.03) were independent risk factors of thrombosis. After assigning coefficient-weighted scores to those risk factor, a new thrombosis prediction model named “ANA model(age,neutrophil count,arterial thrombosis history-ANA)” was developed, classifying patients into low-risk and high-risk groups according to the final total score.Patients in the 2 groups had notably different thrombosis-free survival rates both in training cohort (P < .001) and validation cohort (P < .001). The ANA model outperformed the conventional model in discrimination power (C-statistic: 0.786 [95% CI 0.729-0.843] vs 0.759 [95% CI 0.718-0.800]).Moreover, the DCA curve of ANA model showed better net benefits compared to the conventional model especially when the threshold is between 12 .5%-50%, encouragingly, the incidence of thrombosis after polycythemia vera diagnosis is within this threshold range. Conclusion: This retrospective study confirmed the risk factors affecting thrombosis-free survival in PV.And the ANA model shows excellent accuracy and utility for thrombosis prediction in WHO-defined polycythemia vera.