[Analysis of risk factors for venous thromboembolism in patients with polycythemia vera and establishment of a prediction model]

Abstract

Objective: To investigate the risk factors of venous thrombosis in patients with polycythemia vera (PV) and establish a prediction model for venous thrombosis. Methods: PV patients with JAK2^V617F gene mutation positive in the Second Hospital of Tianjin Medical University from September 2017 to November 2023 were retrospectively included. The patients were divided into groups according to whether they had venous thrombosis. After matching age and gender factors with propensity scores, 102 patients were included in the venous thrombosis group [46 males, 56 females, with a median age M (Q₁, Q₃) of 52 (44, 60) years] and 204 cases were included in the group without venous thrombosis [92 males, 112 females, with a median age of 52 (44, 59) years]. The clinical and laboratory characteristics, disease progression and incidence of gene mutation were compared between the two groups. The follow-up cohort ended on November 20, 2023, with a median follow-up [M (Q₁, Q₃)] of 11 (1, 53) years. Multivariate Cox risk model was used to analyze the influencing factors of venous thrombosis in PV patients, and establish a scoring system for the venous thrombosis risk factor prediction model of PV patients. Receiver operating characteristic (ROC) curve was used to evaluate the predictive efficiency of the model. Results: Hemoglobin concentration, the ratio of hematopoietic volume≥55%, neutrophil to lymphocyte ratio≥5, hypertension, subcostal spleen≥5 cm and secondary myelofibrosis in venous thrombosis group were higher than those in non-venous thrombosis group (all P<0.05). In addition, the proportion of history of thromboembolism, V617F gene mutation load (V617F%)≥50%, diabetes mellitus, ASXL1 mutation and secondary reticular silver staining≥3 in the venous thrombosis group were higher than those in the non-venous thrombosis group (all P<0.05). The proportion of PV patients with 3 or more gene mutations was 44.1% (45/102) in venous thrombosis group, which was higher than that of PV patients without venous thrombosis 29.9% (61/204) (P=0.014). The proportion of ASXL1 gene mutation in venous thrombosis group was 17.6% (18/102), which was higher than the 4.9% (10/204) in non-venous thrombosis group (P<0.001). Multivariate Cox risk model analysis showed that previous thromboembolism history (HR=2.031, 95%CI: 1.297-3.179, P=0.002), V617F%≥50% (HR=2.141, 95%CI: 1.370-3.347, P=0.001), ASXL1 mutation (HR=4.632, 95%CI: 1.497-14.336, P=0.008), spleen subcostal≥5 cm (HR=1.771, 95%CI: 1.047-2.996, P=0.033) are the risk factors of venous thrombosis in PV patients. According to HR values, a score system for predicting risk of venous thrombosis in PV patients was established: previous history of thromboembolism, V617F%≥50% and spleen subcostoal≥5 cm were assigned 1 point respectively, and ASXL1 mutation was assigned 2 points. Low risk group: score 0, medium risk group: score 1-2, high risk group: score≥3. The ROC curve analysis of the model for predicting venous thrombosis in PV patients showed that the area under the curve (AUC) was 0.807 (95%CI: 0.755-0.860), with the sensitivity of 88.2% and the specificity of 59.8% when the Youden index was 0.48. Conclusions: Previous thromboembolism history, V617F%≥50%, ASXL1 mutation, spleen subcostoal≥5 cm are risk factors of venous thrombosis in PV patients. The established prediction model has good prediction efficiency.