Thrombosis Risk Assessment in Polycythemia Vera (TRAP): A 4,636-Patient-Year Analysis of Arterial and Venous Thrombosis in Low-Risk Patients

Background

Conventional thrombosis risk stratification in polycythemia vera (PV) considers age > 60 years and thrombosis history for assignment to low (no risk factor) or high (at least 1 risk factor) risk category (J Clin Oncol.2011;29(6):761-70). The current study exclusively focused on low-risk PV with the following objectives: i) estimate the incidence of arterial and venous thrombosis; ii) examine the prognostic relevance of clinical and genetic risk factors for thrombosis; and iii) assess the impact of treatment on thrombosis risk.

Methods

Patients were retrospectively recruited from the Mayo Clinic, USA, and University of Florence, Italy, PV databases. Diagnosis was according to the International Consensus Classification criteria (Blood 2022;140:1200). Only first major arterial (AT) and unprovoked venous thrombosis (VT) were considered. Conventional statistical methods were applied (JMP Pro 17.0.0, SAS Institute, Cary, NC, USA).

Results

Mayo Clinic cohort

A total of 346 patients with low-risk PV were considered: median age 49 years; males 54%; median hemoglobin 17.7 g/dl; median leukocyte count 10.9 x 10⁹/L; leukocyte count >11 x 10⁹/L 49%; median platelet count 471 x 10⁹/L; cardiovascular (CV) risk factors 47%; hypertension 29%; hyperlipidemia 20%; smoking history 12%; diabetes mellitus 5%. Most frequent mutations, other than JAK2, were ASXL1 (13%) and TET2 (13%). Median follow-up was 13.4 years (1-52) (4,636 patient-years) during which 25 (7%) AT and 20 (6%) VT were documented. The corresponding incidence rates were 5.4 and 4.3 per 1,000 patient-years.

In multivariable logistic regression analysis, baseline hemoglobin >17.5 g/dl in women or >20 g/dl in men (15% vs 4%; OR 5.1; p<0.01) and absence of leukocytosis >11 x 10⁹/L (10% vs 4%; OR 2.9; p=0.02) were associated with increased risk of AT; in addition, diabetes mellitus (19% vs 7%; OR 3.9; p=0.06), showed borderline significance; overall incidence of AT was 20% in the presence of ≥2 risk factors (n=61), 7% in the presence of one risk factor (n=177), and 1% in the absence of all three risk factors (n=108; p<0.01). Multivariable analysis for arterial thrombosis-free survival (TFS) confirmed the predictive performance of these risk factors: hemoglobin thresholds (HR 4.4; p<0.01); absence of leukocytosis >11 x 10⁹/L (HR 2.7; p=0.03); and diabetes (HR 2.8; p=0.09); 20-year cumulative incidence of AT was 32% in the presence of ≥2 risk factors and 7% otherwise (p<0.01).

In multivariable analysis, smoking history was found to be surprisingly protective against VT (0% vs. 7%; p=0.02) while female gender was associated with a borderline higher risk (8% vs 4%; OR 2.2; p=0.09); VT incidence rates were 9% in the presence of both risk factors (n=144), 4% with one risk factor (n=172), and 0% in the absence of both risk factors (n=40; p=0.02); 20-year cumulative incidence of VT was 10% in the presence of both risk factors and 3% otherwise (p=0.03). ASXL1 and TET2 mutations did not appear to modify the risk of either VT (p=0.59) or AT (p=0.66). Treatment at diagnosis included aspirin and cytoreductive therapy in 63% and 24% of patients, respectively. Aspirin but not cytoreductive therapy was associated with a lower incidence of AT (5% vs 11%; 2.3 vs 3.0 per 1,000 patient-years; p=0.04) and VT (4% vs 10%; 1.9 vs 2.8 per 1,000 patient-years; p=0.03). Thrombosis-free survival was not influenced by documented treatment at diagnosis or at the time of event (p>0.1).

University of Florence cohort

A total of 284 low-risk PV patients were followed for a median of 9.1 years (1-38) with 19 (7%) AT and 24 (8%) VT events recorded; the corresponding incidence rates were 7.4 and 9.3 per 1,000 patient-years. In this Florence cohort, risk factors for AT were diabetes mellitus (HR 3.9), hyperlipidemia (HR 3.8) and advanced age, with only age remaining significant during multivariable analysis (p=0.03). Predictors of VT included hyperlipidemia (HR 4.2) and hypertension (HR 2.1). Treatment with aspirin or cytoreductive therapy did not impact thrombosis-free survival (p>0.1).

Conclusion

The current study confirms the infrequent occurrence of AT and VT in low-risk PV but also exposes the challenges in identifying risk factors that are reproducible in different but otherwise well-characterized patient cohorts. Discrepancies were also apparent in retrospective comparison of treatment effects on thrombosis, which reinforces the need for prospective studies.