Introduction

Since cancer patients are at high risk for both arterial and venous thrombosis compared to the general population, and appropriate prevention and treatment of cancer-associated thrombosis (CAT) is of critical prognostic impact, CAT has been the subject of intensive research, and various expert panels have published guidelines for CAT in recent years [1,2,3,4].

CAT is also a common complication among patients with hematologic malignancies. In a Danish health care registry-based population study, the adjusted standardized hazard ratios for the incidence of thrombosis by cancer type in the first six months after the cancer diagnosis were as follows: leukemia, 3.55 (95% confidence interval 2.82–4.49); non-Hodgkin lymphoma (HL) 7.44 (6.07–9.13); HL, 8.05 (5.98–10.83); and multiple myeloma (MM), 7.96 (6.30–10.04) [5].

The Khorana score is the most well-known risk assessment tool to guide the indication for anticoagulation in CAT, but the only blood cancer patients included in the original literature were those with lymphoma [6, 7]; leukemia, MM, and myeloproliferative neoplasm (MPN) patients were not enrolled in their study. The Khorana score, therefore, may not be applicable for stratifying the risk of CAT in hematologic malignancy, especially in acute leukemia, as the Khorana score is overestimated by anemia and leukocytosis [8, 9].

In recent years, a number of clinical trials have reported on the prevention [10, 11] and treatment [12,13,14] of CAT with direct oral anticoagulants (DOACs), and their efficacy is being established. However, most of these trials included only 10% or fewer patients with hematologic malignancies, and some studies excluded patients with acute leukemia (Table 1).

Table 1 Pivotal randomized-controlled trials of cancer-associated VTE prophylaxis or treatment in which hematologic malignancy patients were included
Full size table

This article reviews the specific considerations of CAT in hematologic malignancies, up-to-date perspectives, and unresolved clinical questions on a disease-by-disease basis (Fig. 1).

Fig. 1
figure 1

Factors to consider in optimizing prevention and treatment strategies for CAT in hematologic malignancies

Full size image

Antithrombotic therapy in thrombocytopenic patients

The International Initiative on Thrombosis and Cancer (ITAC) clinical practice guidelines recommend low-molecular-weight heparin (LMWH) for the prevention and treatment of thrombosis associated with solid tumors [2]. However, patients with hematologic malignancies, especially those with acute leukemia, often present with severe thrombocytopenia due to underlying disease or intensive chemotherapy, which makes the bleeding risk associated with heparinoid administration a major concern.

Several clinical studies have examined the efficacy and safety of thromboprophylaxis by comparing standard versus reduced doses of LMWH in CAT patients with platelet counts below 50,000–100,000/μL [15, 16]. Reduced-dose LMWH was associated with similar or better outcomes in these large studies. In contrast, in another single-center retrospective study of 78 patients with platelet counts below 50,000/μL who needed treatment for venous thromboembolism (VTE) during chemotherapy for acute leukemia, lymphoma, and plasma cell tumors, patients who prematurely discontinued anticoagulation experienced fewer bleeding events at the cost of more recurrent VTE [17].

Practically, the NCCN guidelines recommend the following LMWH doses for CAT with thrombocytopenia: full dose for patients with platelet counts > 50,000/μL, a dose reduction for patients with platelet counts 25,000–50,000/μL, and discontinuation for patients with platelet counts < 25,000/μL [4].

Acute promyelocytic leukemia (APL)

Sixty-to-seventy percent of newly diagnosed APL cases are complicated by disseminated intravascular coagulation (DIC). At the same time, not only DIC but also thrombosis complicates approximately 9% of cases upon diagnosis and during treatment of APL [18]. Various coagulation and platelet aggregation activators, including podoplanin, are expressed by APL cells, which may contribute to thrombosis [19]. The binding of plasminogen and plasminogen activator to this tetramer results in efficient plasmin production and activation of the fibrinolytic system. All-trans retinoic acid (ATRA), a key drug in APL treatment, exerts its antifibrinolytic effect by suppressing the expression of the Annexin II/S100A10 heterotetramer, which is present on the APL cell surface [20, 21]. Several retrospective studies suggest that prompt administration of ATRA may lead to early resolution of DIC and improved survival [22, 23].

In the pre-ATRA era, antifibrinolytic therapy with the plasmin inhibitor tranexamic acid (TXA) in combination with transfusion support failed to demonstrate improvement in remission rates or early mortality due to bleeding compared to transfusion alone [24]. TXA is no longer recommended to be administered in combination with ATRA, because it can increase the risk of thrombosis [25].

Acute myeloid leukemia (AML) other than APL

DIC is also a risk factor for thrombotic events in non-APL AML, which is present in approximately 30% of AML patients at the initial presentation [26, 27]. The incidence of VTE in AML is 5–15%, most of which occurs within 3 months of AML diagnosis [9, 28,29,30,31,32,33]. In a single-center retrospective analysis of 222 AML cases, AML patients who developed thrombosis had a significantly worse prognosis than those who did not [9].

Several attempts to develop acute leukemia-specific CAT risk assessments have been reported. Previously identified risk factors for thromboembolic complications in AML include advanced age, high D-dimer, high white blood cell (WBC) count, and tobacco smoking [9, 27, 28, 33]. It is noteworthy that a preserved platelet count (> 50,000–100,000/μL) at diagnosis is also a risk factor for thromboembolic complications in AML [31] (Table 2).

Table 2 Common and disease-specific risk factors for VTE previously reported in hematologic malignancies
Full size table

Acute lymphoblastic leukemia (ALL)

The incidence of DIC in ALL is less than that in AML, approximately 10%, while thrombosis occurs at approximately the same or a slightly higher frequency in ALL than in AML [29, 34]. A retrospective analysis of 2482 ALL patients enrolled in a US cancer registry showed a 2-year cumulative incidence of thromboembolic complications of 4.5%, and the overall survival (OS) of ALL patients who developed thrombosis was inferior to that of those who did not [28].

ALL patients remain at high risk of thrombosis even after remission induction therapy, in contrast to AML patients [35, 36]. This may be partly explained by the high-dose corticosteroids and L-asparaginase (L-Asp) used to treat ALL, both of which cause an imbalance in coagulation and fibrinolysis [35]. Other known risk factors for thrombosis in ALL include advanced age, comorbidities, and high D-dimer levels, as in AML [37] (Table 2). L-Asp also inhibits the synthesis of various factors involved in blood coagulation and fibrinolysis, most notably plasma antithrombin (AT) activity and fibrinogen [38].

There have been no randomized-controlled trials (RCTs) examining the thromboprophylactic effect of AT concentrates in ALL patients receiving L-Asp, and no preventive measures with a high level of evidence have been established [39]. However, based on several retrospective analyses, the International Society on Thrombosis and Hemostasis (ISTH) recommends measuring plasma AT activity once a week and administering AT concentrates to maintain plasma AT activity above 50–60% [40, 41]. It is also weakly recommended that prophylactic fresh frozen plasma (FFP) not be administered routinely because the asparagine contained in FFP can compromise the antileukemic activity of l-Asp [41].

Lymphoma

The Khorana score defines lymphoma as the highest-risk disease by cancer type [6], even though validation studies raise the question of whether it is adequate for high/intermediate risk stratification [42, 43]. A retrospective analysis of 18,018 patients with lymphoma enrolled in 29 clinical studies showed a thrombosis incidence of 6.4%, with incidences of 5.3% for deep vein thrombosis (DVT) and 1.2% for pulmonary embolism (PE) [44]. By histological type of lymphoma, the incidence of VTE was 8.3% in high-grade non-Hodgkin lymphoma (NHL), 6.3% in low-grade NHL, and 4.7% in HL.

The ITAC guidelines recommend that intermediate- and high-risk patients with a Khorana score of 2 or higher receive prophylaxis with LMWH, apixaban, or rivaroxaban [2]. However, lymphoma patients accounted for only 10% of those enrolled in the original Khorana score study. The Thrombosis Lymphoma (ThroLy) risk model was developed to specifically predict the development of VTE in patients with lymphoma [45]. The model is scored by assessing a history of thrombosis, body mass index > 30 kg/m2, mediastinal disease, extranodal involvement, performance status, neutropenia, and hemoglobin levels. Furthermore, an increased risk of developing VTE has been reported in patients with primary central nervous system lymphoma, primary mediastinal large B-cell lymphoma, use of anthracycline or methotrexate, and B symptoms [46,47,48,49] (Table 2).

A retrospective comparison of the efficacy and safety of LMWH and warfarin in 57 patients with lymphoma showed the superiority of LMWH [50], and the ITAC guideline recommends DOAC at the same level as LMWH for this indication (Table 1) [2].

Multiple myeloma (MM)

The use of immunomodulatory drugs such as thalidomide and lenalidomide (Len) in combination with dexamethasone (Dex) in patients with MM increases the risk of developing VTE compared with their use as single agents [51]. With regard to proteasome inhibitors, a retrospective analysis of the incidence of VTE in 672 patients with newly diagnosed MM showed that carfilzomib in combination with Len/Dex increased the incidence of VTE approximately twofold (21.1% vs. 9.6%) compared with bortezomib in combination with Len/Dex [52]. The anti-CD38 monoclonal antibody daratumumab does not confer an additional risk of VTE when used in combination with other drugs [53].

The IMPEDE VTE score was developed to predict VTE risk in MM patients based on an analysis of 4,446 patients registered at the Veterans Administration Central Cancer Registry (VACCR) [54]. IMPEDE VTE was validated in a large independent cohort of the Surveillance, Epidemiology, End Results (SEER)-program database. A similar SAVED score, conversely, was developed based on the SEER database and validated in patients registered in the VACCR [55]. Asian ethnicity is a characteristic factor associated with a reduction in the development of VTE common to IMPEDE VTE and SAVED (Table 2).

Low-dose aspirin and LMWH or DOAC are recommended for low- and high-risk MM patients, respectively, based on these risk prediction models [56, 57].

Chronic myeloid leukemia (CML)

Four meta-analyses have evaluated the impact of BCR::ABL1 tyrosine kinase inhibitors (TKIs) on cardiovascular events in CML patients. Compared to imatinib, second- and third-generation TKIs, excluding bosutinib, were shown to increase the incidence of arterial occlusive disease but not VTE [58,59,60,61]. The adverse effects are particularly considerable with ponatinib, which may be due in part to direct cytotoxicity to the vascular endothelium [62]. Prophylactic low-dose aspirin was recommended later in clinical trials of ponatinib [63].

Common cardiovascular risk factors, such as smoking, hypertension, dyslipidemia, and diabetes, are also known to predict cardiovascular events in CML patients on TKI treatment. It is important to select an appropriate TKI based on an individual patient's risk assessment. The European Society of Cardiology has proposed Systematic Coronary Risk Evaluation (SCORE) as a comprehensive risk assessment system for coronary artery disease in general [64]. In a retrospective clinical study analyzing the incidence of cardiovascular events in 192 CML patients with hypertension, the 5-year cumulative incidence of arterial occlusive disease was 33% in patients with a SCORE of high or very high, compared to 14% in the remaining patients [65].

Asciminib, a recently approved allosteric inhibitor that targets the ABL1 myristoyl pocket, has been suggested to be potentially less toxic to the cardiovascular system than conventional TKIs in a preclinical animal study [66]. The clinical safety of asciminib remains to be verified; 8.7% of the subjects experienced arterial occlusive events in the long-term follow-up of the asciminib phase 1 trial [67].

BCR::ABL1-negative MPN: polycythemia vera (PV) and essential thrombocythemia (ET)

Both PV and ET are hematopoietic neoplasms with a high rate of arterial and venous thrombosis, which is a major prognostic factor, although the pathophysiology and treatment of these conditions differ from one another [68,69,70]. The common feature is that the recurrent genetic abnormality JAK2 V617F is a significant risk factor, particularly for VTE [71,72,73,74]. In JAK2 V617F knock-in mouse models, increased megakaryocyte migration and platelet aggregation are observed, suggesting that not only the quantity but also the quality of blood cells may contribute to the development of thrombosis [75,76,77].

PV

Established risk factors for thrombosis in PV are advanced age (> 60–65 years) and a history of thrombosis [78, 79]. In addition to the common cardiovascular factors, such as dyslipidemia and diabetes, leukocytosis is also a risk factor for arterial thrombosis in PV [80,81,82,83] (Table 2).

Low-dose aspirin has been shown to reduce arterial thrombosis in PV patients in double-blind RCTs and is recommended in Cochrane reviews [84, 85]. If venous thrombosis develops while on low-dose aspirin, switching aspirin to warfarin or DOACs should be considered, although high-level evidence is lacking [86].

Controlling hematocrit (Ht) with phlebotomy and/or hydroxyurea (HU) also improves OS in PV [87]. PV patients harboring JAK2 V617F who were prospectively randomized to phlebotomy and/or HU with an intensive target Ht < 45% had fewer cardiovascular events and deaths from thrombosis than patients randomized to a control target Ht 45–50% [88]. In this trial, however, not only Ht control but also the suppressive effect of HU on leukocytes and platelets should be taken into consideration. In another study on HU-resistant PV, poor control of WBC and platelet count, not Ht, was most associated with the risk of thrombotic and bleeding events [89]. In a propensity score matching analysis of PV patients who received phlebotomy alone (342 cases) or HU alone (681 cases), OS and cumulative incidence of cardiovascular events were better in the HU group, despite the lower Ht and WBC counts in the phlebotomy group, and this association was particularly significant in high-risk PV patients [90,91,92].

The JAK inhibitor ruxolitinib can be chosen to alleviate splenomegaly and constitutional symptoms in PV patients resistant or intolerant to HU. JAK inhibitors have theoretically favorable pharmacological profiles to prevent thrombosis, suppressing the production of pro-thrombotic mediators such as P-selectin, von Willebrand factor, interleukin-6, and tissue factor in vitro [93]. Nevertheless, the thromboprophylactic effect of JAK inhibitors was not demonstrated until recently [94, 95]. The MAJIC-PV study showed the superiority of ruxolitinib over best available therapy in OS and event-free survival, including thrombosis as an endpoint, in HU-resistant/intolerant PV, regardless of splenomegaly [96].

ET

IPSET-thrombosis, stratified by age > 60 years, previous thrombosis, and JAK2 V617F, is a widely accepted risk score for arterial thrombosis in ET [73, 97]. Notably, platelet count is not a significant risk factor for thrombosis [98, 99] (Table 2). ET patients with JAK2 V617F, who are at high risk for thrombosis, tend to have paradoxically lower platelet counts than those with mutated CALR [72, 100].

A pivotal RCT showed improved thrombosis-free survival with HU by suppressing platelet counts to less than 600,000/μL in ET [101]. However, this study was conducted years before the detailed molecular pathogenesis and classification of ET were characterized, and the mechanisms of action of HU may not be attributable to simply the decreased platelet counts [99]. Especially in low-risk ET, defined as age younger than 60 years old, without cardiovascular risk, and with platelet counts < 1.5 × 106/μL, cytoreduction with potentially carcinogenic HU is not recommended over aspirin alone, since the addition of HU did not improve key outcomes, such as vascular events, myelofibrosis, and leukemic transformation [102]. There is inconsistency among trials as to whether anagrelide, an alternative cytoreductive agent, is sufficient to prevent thrombosis in HU-resistant/intolerant ET [103, 104].

In contrast to PV, RCTs demonstrating the benefit of aspirin in ET are lacking, although it is empirically used in many cases [85]. In a retrospective analysis, low-dose aspirin for patients with CALR-mutated ET did not prevent thrombosis but rather increased bleeding [105]. Individual patient factors such as JAK2 V617F, advanced age, and cardiovascular comorbidities should be considered to determine the indication of aspirin in ET [106, 107].

Future directions and perspectives

Many clinical questions are still unsolved in CAT, especially in hematologic malignancies, due to the limited patient population and unique pathophysiology compared to solid cancer. The safety and efficacy of DOACs, which are becoming widely used in CAT in solid cancers based on accumulating high-quality evidence, are expected to be evaluated for a range of hematologic malignancies in prospective clinical trials. Hematologists must carefully assess the patient's risk of CAT to offer appropriate prophylaxis and treatment of thromboembolic complications while minimizing the risk of bleeding and maximizing the efficacy of anticancer treatment, even in the midst of a rapidly changing landscape of clinical oncology practice.