Deciphering the individual contribution of absolute neutrophil and monocyte counts to thrombosis risk in polycythemia vera and essential thrombocythemia

Abstract

In addition to its adverse effect on survival and leukemic transformation, in both polycythemia vera (PV) and essential thrombocythemia (ET), leukocytosis has been implicated as a risk factor for thrombosis, in both of these myeloproliferative neoplasms (MPNs).1 Recent studies have underscored anatomical and biologic differences as well as common traits between arterial (AT) and venous (VTE) thrombosis. Consequently, different sets of risk factors for AT versus VTE in PV versus ET, have been promoted; in PV, these include prior AT event, hypertension, and hyperlipidemia, for AT, and prior venous event, older age, major hemorrhage, and leukocytosis, for VTE2, 3; in ET, the corresponding variables for AT include prior AT event, age > 60 years, cardiovascular risk factors, presence of JAK2V617F, and leukocytosis and, for VTE, male gender.4-6 In the current study, we sought to clarify the individual prognostic contribution of absolute neutrophil (ANC) and monocyte (AMC) counts for the association between leukocytosis and thrombosis in ET and PV. The current study included 487 patients with PV (n = 138) or ET (n = 349), recruited from the Mayo Clinic MPNs database, based on the availability of information on ANC and AMC, within 1 year of diagnosis. An external cohort of 424 patients with PV, from the University of Florence, Italy, was used to validate the observations from the Mayo Clinic PV cohort. Conventional criteria were used for diagnosis7 and definitions of major vascular events.6 Most patients were managed according to standard practice that included phlebotomy and aspirin therapy in low-risk patients and the addition of cytoreductive therapy, often in the form of hydroxyurea, in high-risk patients. Statistical analyses were performed using JMP Pro 14.0.0 software package, SAS Institute, Cary, NC. Multivariable analyses included previously established risk factors for AT or venous thrombosis. The Mayo Clinic PV patient cohort included 138 patients (median age 62 years, range 20–94; females 50%); presenting median (range) values were 17.9 g/dL (16.1–24) for hemoglobin, 11.8 × 10 (9)/L (2.7–65.8) for leukocyte count, and 434 × 10 (9)/L (44–1679) for platelet count; 57% of patients presented with leukocytosis >11 × 10 (9)/L; palpable splenomegaly was present in 37 (27%) patients; abnormal karyotype was documented in 17% of patients. Conventional risk category was high in 86 (62%) patients and low in 52 (38%). Median follow-up was 11 years (range 0.03–36.7). There were 22 (16%) documented venous events at diagnosis and 20 (15%) after diagnosis and 28 (20%) AT events at diagnosis and 15 (11%) after diagnosis. Neither ANC (p = .85) nor AMC (p = .14) correlated with AT at or prior to diagnosis, advanced age was the only other parameter with significant association in this regard (p = .01). Similarly, AT-free survival was not affected by either ANC (0.31) or AMC (0.27); advanced age was again the only other parameter tested that showed significance (p < .001) while prior AT events showed borderline significance (p = .05) that was no longer apparent during multivariable analysis. By contrast, VTE at or prior to diagnosis correlated with both ANC (p = .048) and AMC (0.06; borderline significance) while no other parameter tested, including age (0.7) or sex (p = .16), showed significant association. In univariate analysis, VTE-free survival was negatively affected by higher ANC (p = .001) and higher AMC (p = .036); ANC (p = .02) but not AMC (p = .71) retained significance during multivariable analysis that included age (p = .68), sex (p = .18) and prior venous events (p = .68). The adverse effect of higher ANC on VTE-free survival was most apparent in high (p = .004) versus low (p = .14) risk disease. The external validation cohort of PV patients was recruited from the University of Florence, Italy: median age 62 years (range 18–92); females 40%; presenting median (range) values of 17.8 g/dL (16.1–24) for hemoglobin, 9.8 × 10 (9)/L (4.5–26.9) for leukocyte count, and 461 × 10 (9)/L (154–1352) for platelet count. Median follow-up was 8 years (range 0.5–30.8); 46 (10.8%) venous events were documented at diagnosis and 33 (7.8%) after diagnosis; 58 (13.7%) AT events were documented at diagnosis and 39 (9.2%) after diagnosis. As was the case with the Mayo Clinic cohort, neither ANC nor AMC correlated with AT events at/prior to or after diagnosis. By contrast, VTE-free survival was negatively affected by higher ANC (p = .003) and prior venous events (p = .01), but not by AMC (p = .7), age (p = .4) or gender (p = .3); ANC (p = .01) and prior venous events (p = .04) retained significance during multivariable analysis. In the Florence cohort, the adverse effect of higher ANC on VTE-free survival was more prominent in low-risk diseases (p = .02), although a trend was also apparent in high-risk diseases (p = .07). On the other hand, unlike the case with the Mayo Clinic cohort, VTE at or prior to PV diagnosis did not correlate with either ANC (p = .4) or AMC (p = .2). The optimal ANC cutoff level for VTE-free survival risk stratification, determined by receiver operating characteristic (ROC) plot analysis, was 22 × 10 (9)/L for the Mayo Clinic cohort (Figure 1A) and 10 × 10 (9)/L for the Florence cohort (Figure 1B). The ET patient cohort included 349 patients (median age 57 years, range 18–89; females 61%): 46% JAK2, 34% CALR, 16% triple-negative, and 4% MPL mutated; international prognostic system for thrombosis in ET (IPSET) risk category was high 24%, intermediate 41%, and low 35%; presenting median (range) values were 13.8 g/dL (11.1–16.4) for hemoglobin, 8.2 × 10 (9)/L (3.2–29.8) for leukocyte count, and 859 × 10 (9)/L (451–3460) for platelet count; palpable splenomegaly was present in 48 (14%) patients and cardiovascular risk factors in 56%; median follow-up was 10 years (range 0–47). There were 38 (11%) documented venous events at diagnosis and 31 (9%) after diagnosis and 42 (12%) AT events at diagnosis and 64 (18%) after diagnosis. Neither ANC (p = .8) nor AMC (p = .38) correlated with AT at or prior to diagnosis; instead, parameters of significance included advanced age (p = .003), JAK2 mutation (p = .01), male sex (p = .03) and, of borderline significance, cardiovascular risk factors (p = .07). Similarly, AT-free survival was not affected by either ANC (0.1) or AMC (0.1); in multivariable analysis, significant variables included advanced age (p = .03), prior AT event (p = .04), and cardiovascular risk factors (p = .04). As was the case with PV, VTE at or prior to diagnosis correlated with both AMC (0.003) and ANC (p = .06), the latter with borderline significance; other parameters with borderline significance in this regard included JAK2 mutation (p = .06), advanced age (p = .08), and female sex (p = .08). During multivariable analysis, VTE-free survival was affected by advanced age only (p = .001). Observations from the current study flag neutrophils, and possibly monocytes, as potential contributors to venous thrombosis in PV, and possibly in ET as well. This would be consistent with emerging evidence on the role of neutrophils in venous thrombosis where the underlying mechanisms might also involve platelets, monocytes, the endothelium, procoagulant proteins, and neutrophil extracellular traps (NETs), with resulting interactive state of hypercoagulability and immune-mediate inflammation.8, 9 The persistent VTE risk in both high and low-risk PV patients presenting with increased ANC, as suggested by their inferior VTE-free survival, raises questions on the adequacy of current treatment in preventing venous thrombosis. A similar notation was made in a previously published controlled study of patients with ET where the risk of VTE was higher in patients receiving hydroxyurea versus anagrelide, both in combination with aspirin therapy.10 In two separate collaborative work involving patients with PV, we also showed adverse impact on VTE-free survival from higher JAK2V617F allele burden11 and neutrophil-lymphocyte ratio,12 which further supports the detrimental role of pro-thrombotic clonal burden associated with neutrophilia and associated inflammatory state, marked by lower lymphocyte count. Taken together, the three parallel studies,11, 12 including the current work, identify risk factors for VTE that are different than those for AT. Whether or not such information is actionable, in the context of current therapy, cannot be surmised at the present time, but signifies the need for venous thrombosis-focused analysis of future prospective and retrospective studies. The authors declare no potential conflicts of interest. Data available on request from the authors.

Topics & Concepts