How I treat rare venous thromboses

Although venous thrombosis can involve any section of the venous system, the most common manifestation is deep vein thrombosis (DVT) of the lower extremities, with or without pulmonary embolism. Thrombosis can also involve other venous sites, such as the upper limbs, cerebral sinuses and splanchnic veins.¹  Although rare, these thrombotic events are often severe, accounting for important morbidity and mortality.²  Beside the therapeutic aspects of thrombosis in rare venous sites, this article will also address the pathogenic mechanisms and risk factors. Catheter-related upper-extremity deep vein thrombosis and retinal vein thrombosis will not be featured, as the former occurs in up to 30% of patients with central venous lines and the latter is the most common vascular disease of the eye, so that they cannot be considered rare

Cerebral vein thrombosis (CVT) is characterized by a variety of symptoms spanning from headache, lethargy, focal neurologic signs and seizures to more severe clinical manifestations such as stroke and coma.^3,4  Although in the past CVT was considered a severe disease characterized by significant short- and long-term morbidity and high mortality, its prognosis is now much more favorable. The introduction of noninvasive and sensitive diagnostic techniques (such as magnetic resonance imagine [MRI] and MRI angiography) has increased the frequency with which CVT is objectively diagnosed. A review of 19 studies reported a mortality rate of 5.6% (range, 0%-15.2%) during hospitalization and 9.4% (range, 0%-39%) at the end of the follow-up period, both rates being much lower than those of 30% to 50% reported in older studies.⁵  Furthermore, the majority of patients recover completely or with minor deficits (88%), and recurrence is uncommon (2.8%).⁵  These improvements are imputed not only to the more sensitive diagnostic techniques that detect earlier cases but also to the use of anticoagulant therapy.⁶ 

After the first observations of a relationship between CVT, inherited thrombophilia, and oral contraceptive use,^7,8  several recent studies have emphasized the prominent role of such risk factors. Two large studies showed that factor V Leiden increases the risk of CVT up to 5-fold, G20210A prothrombin to as much as 10-fold, mild to moderate hyperhomocysteinemia up to 4-fold.^8,9  These risk estimates became much higher in women taking oral contraceptives, indicating a synergic effect between this acquired risk factor and thrombophilia.⁹  Similar results, obtained in smaller studies (Table 1),^8-18  were summarized in a meta-analysis,¹⁹  that confirmed the strong association between CVT and G20210A prothrombin (OR 9.27; 95% confidence interval [CI], 5.85-13.67), factor V Leiden (OR 3.38; 95% CI, 2.27-5.05), hyperhomocysteinemia (OR 4.07; 95% CI, 2.54-6.52) and oral contraceptive use (OR 5.59; 95% CI, 3.95-7.91). On the whole, the mechanistic roles of gain-of-function mutations of coagulation factors prothrombin and V and of oral contraceptive intake sort out very clearly, apparently with a stronger role for G20210A prothrombin.¹⁹ 

Heparins followed by oral anticoagulant therapy (OAT) with vitamin K antagonists are commonly used in clinical practice, because these drugs do not increase the risk of intracranial hemorrhage.^20,21  Only 2 randomized control trials evaluated anticoagulation in patients with CVT.⁶  One compared unfractionated heparin with placebo and was stopped after the enrollment of 20 of the planned total of 60 patients because of the efficacy of treatment.²²  The other included 60 patients who received placebo or low-molecular-weight heparin (LMWH) for 3 weeks, followed by OAT for 3 months (target International Normalized Ratio of the prothrombin time between 2.5 and 3.5). Patients on anticoagulant treatment had more favorable, although not statistically significant, outcomes than controls.²³  Anticoagulation in the acute phase of CVT in patients with large cerebral hemorrhage at the time of diagnosis is matter of debate among experts.²⁴  On one hand, early anticoagulation may reduce the risk of thrombus extension, but on the other hand it may worsen hemorrhage. In the absence of clear data on efficacy and safety of early anticoagulation in these patients, we are among those who think that a delayed initiation of anticoagulation, until a stabilization or reduction of hemorrhage is objectively documented, is a sensible approach (Figure 1). Recently published consensus-based guidelines on the treatment of CVT recommend dose-adjusted intravenous heparin or body weight-adjusted LMWH followed by OAT of duration similar to that for lower limb DVT: 3 months if CVT is secondary to transient risk factors, 6-12 months if unprovoked or in the presence of mild hereditary thrombophilia (protein C and S deficiency, heterozygosity for factor V Leiden or G20210A prothrombin), or lifelong in the presence of recurrence or markers of severe thrombophilia (antiphospholipid antibodies and particularly the lupus anticoagulant, antithrombin deficiency, homozygosity for factor V Leiden, combined thrombophilia).²⁵  A comprehensive clinical evaluation of the presentation of each patient (idiopathic or secondary event, severity in terms of number of sinuses involved and sequels, type of thrombophilia) is recommended for optimal assessment of the risk/benefit profile and hence the duration of anticoagulation.

The most clinically relevant splanchnic vein thromboses (SVT) are extrahepatic portal vein thrombosis, Budd-Chiari syndrome (or hepatic vein thrombosis), and mesenteric vein thrombosis.²⁶  Imaging techniques (Doppler ultrasonography, computer tomography [CT], and MRI) facilitate early diagnosis, which is often difficult because the most frequent symptom (severe upper abdominal pain) is nonspecific. The geographic epidemiology of SVT is peculiar, portal vein thrombosis being more prevalent in Western countries and Budd-Chiari syndrome in Asia.²⁷ 

Among the predisposing conditions, there are such diseases as chronic myeloproliferative disorders (CMD), chronic inflammatory disorders, paroxysmal nocturnal hemoglobinuria, and Behçet disease, and also acquired thrombophilic conditions such as the antiphospholipid syndrome, pregnancy and puerperium, and the intake of oral contraceptives. Liver cirrhosis and abdominal inflammatory states (pancreatitis, abscess, inflammatory bowel disease, diverticulitis), surgical injury to the portal venous system, and abdominal cancer also play a mechanistic role. In Western countries, Philadelphia chromosome–negative CMDs are the leading cause of SVT not associated with cirrhosis, accounting for approximately 50% of Budd-Chiari syndrome and 25% of extrahepatic portal vein thrombosis.²⁸  However, the diagnosis of CMD associated with SVT is muddled by the frequent concomitant presence of portal hypertension and hypersplenism, which may mask any increase in blood cell counts. A significant advance in the diagnostic accuracy of CMD was thus provided by the discovery of acquired gain-of-function mutations in the JAK2 gene (V617F).^29-32  The JAK2V617F mutation was detected in 40% to 60% of patients with Budd-Chiari syndrome and in 30% to 40% of those with portal vein thrombosis, also in cases not fulfilling the diagnostic criteria for CMD.^28-31  A mutation in exon 12 of JAK2 was recently described in a much smaller proportion of patients with SVT and polycythemia vera.³²  In all, these results suggest that the search for JAK2 mutations is useful in the diagnostic workup of SVT, because it may help to unravel hidden myeloproliferation.

Table 2 summarizes the studies on the role of inherited thrombophilia in SVT.^33-37  The available studies are not homogeneous, because some but not others excluded patients with cirrhosis- or CMD-associated SVT. This limitation notwithstanding, mutations of coagulation factors prothrombin and V, alone or in synergy with oral contraceptives, appear to be prominent risk factors. The role of deficiencies of the natural anticoagulant proteins antithrombin, protein C, and protein S is less certain, not only because these proteins were rarely investigated but also because they are synthesized by the hepatocyte, so that it is difficult to establish whether deficiencies are causes of SVT or consequences of the resulting impairment of liver function.

Due to the rarity of SVT, its long-term prognosis, morbidity, and mortality are largely unknown, and only a few studies have addressed these issues and that of treatment. In general, Budd-Chiari syndrome has a more aggressive clinical presentation than thrombosis of the portal system. Whereas the former often leads to liver transplantation when medical therapies and transjugular portosystemic shunting are not effective, the latter may be diagnosed incidentally. Acute treatment with catheter-directed or systemic thrombolysis has been attempted in severe cases, particularly in patients with the Budd-Chiari syndrome, but only in single cases or very small series, with contrasting results in terms of efficacy and safety.³⁸  Amitrano et al followed up 121 patients with SVT for 41 months, observing that anticoagulation was associated with recanalization in 45% of cases and did apparently protect from recurrence, which occurred in 0 of 41 anticoagulated and in 10 of 54 (18.5%) nonanticoagulated patients.³⁹  The overall mortality rate was 10.7%. On the basis of these results, Amitrano et al recommend lifelong anticoagulation for patients with SVT who fail to achieve recanalization and for those with inherited thrombophilia. Similarly, 2 recently published consensus statements recommend lifelong anticoagulation for carriers of prothrombotic factors.^40,41 

In SVT, the issue of anticoagulant therapy is even more challenging than for other rare venous thromboses, for several reasons. Portal vein thrombosis often results in such complex and severe sequels as portal hypertension, cavernoma formation, and hypersplenism. The presence of portal hypertension increases the risk of bleeding from esophageal varices, and in Budd-Chiari syndrome the impairment of liver function often results in the reduced synthesis of coagulation factors and a prolonged prothrombin time. Frequent splenomegaly, causing a low circulating platelet count, increases the risk of bleeding associated with anticoagulant therapy. For these reasons, physicians are often concerned about harm to patients if this therapy is prescribed. Because the current recommendation of lifelong anticoagulant therapy is not justified by the aforementioned small studies of insufficient size, the decision has to be made on an individual basis, taking carefully into account the risk of bleeding. This risk is determined not only by the sequelae of SVT (ie, portal hypertension with esophageal varices and hypersplenism with thrombocytopenia), but also by the causes of SVT. For example, the risk of bleeding in patients whose SVT occurred as a complication of liver cirrhosis is higher than that in patients with SVT from other causes. If the risk of bleeding is due to the presence of esophageal varices, beta-blockers and endoscopic bending are suggested, whereas if the risk is due to the presence of thrombocytopenia secondary to hypersplenism, we prefer to avoid anticoagulant therapy when platelet count is below 50 000/mm³ (Figure 2). In patients with SVT associated with CMD unequivocally documented by a finding of JAK2 mutations, hydroxyurea is suggested even if myeloproliferation and the ensuing increase in peripheral cell count is hidden by hypersplenism.

Upper-extremity deep vein thrombosis (UEDVT) can involve the subclavian, axillary, or brachial veins and account for up to 4% of all DVTs. Three-fourths of the cases are related to direct vascular injuries due to central venous catheters, pacemakers, or cancer.⁴²  The remaining cases of UEDVT, called primary, are idiopathic or due to pinching of the axillary-subclavian vein at the thoracic outlet by muscular and osteotendineous structures induced by abduction and extension of the arm (the so-called Paget-Schroetter or effort syndrome). Although the compression of the axillary-subclavian vein is physiologic, it may be aggravated to the point of triggering thrombosis by anatomical abnormalities or by repetitive compressions, as may occur in athletes.^42-47 A peculiarly high frequency of UEDVT is observed in women undergoing artificial reproductive technology who develop the ovarian hyperstimulation syndrome.⁴⁸  It is thought that this specific localization is due to high levels of estrogen in the peritoneal fluid that is drained by the lymphatic system into the subclavian and jugular veins, leading to heightened local activation of coagulation and thrombus formation.⁴⁹ 

UEDVT is considered less severe than lower extremity DVT, because of lower rates of pulmonary embolism (3% vs 12%, respectively) and recurrence (1% vs 21%, respectively).^50,51  The highest rates of complications are observed in patients with thrombophilia.⁵²  Also the postthrombotic syndrome is rarer in UEDVT, being present in 15% to 25% of patients.

Inherited thrombophilia is associated with an increased risk of UEDVT. Table 3 summarizes the role of gain-of-function mutations and their interaction with oral contraceptives.^45-49,52-55  The largest study found a 5-fold increased risk for both G20210A prothrombin and deficiency of natural anticoagulant proteins taken together, and a 6-fold increased risk for factor V Leiden.⁴⁹  A higher recurrence rate was also observed in patients with thrombophilia (4.4% patient-year compared with 1.6% patient-years in those without thrombophilia), with a synergistic effect between oral contraceptive intake and inherited thrombophilia. For example, in carriers of G20210A prothrombin, oral contraceptives increase the risk of the disease up to 13-fold, suggesting a multiplicative interaction between these risk factors.⁴⁴ 

Treatment of UEDVT is similar to that of lower extremity DVT, although the rates of such complications as pulmonary embolism, recurrence, and postthrombotic syndrome are lower for the former.^50,51  Because neither the initial treatment nor the optimal duration of OAT has been established by randomized controlled trials, it is not known whether heparin is preferable to systemic or catheter-directed thrombolysis. In analogy with the more common and investigated lower extremity DVT, thrombolysis is not recommended for most patients with UEDVT, whereas heparins for initial treatment and OAT for secondary prophylaxis are the options of choice. Patients with a first UEDVT that occurred in the presence of a transient risk factor usually receive anticoagulant therapy for at least 3 months and those with one or more inherited causes of thrombophilia for up to 6 or 12 months. A longer period of anticoagulation is not warranted due to the relatively low recurrence rate and the rarity of postthrombotic syndrome.

The progress in diagnostic imaging has greatly contributed to the improved diagnosis of thrombosis in rare venous sites, leading to the earlier implementation of treatment essential to control morbidity and mortality. The progress in thrombophilia research has broadened our understanding of the mechanisms of rare venous thrombosis, emphasizing the role of gain-of-function mutations of coagulation factors prothrombin and V, in synergy with oral contraceptive intake in general, and with CMD in the context of SVT. However, because of the rarity of CVT, SVT, and UEDVT, there are no randomized trials of adequate sample size addressing either a possible association with weak or rare thrombophilic abnormalities or the optimal duration of anticoagulant therapy. Hence, treatment of thromboses at rare sites is based on opinions of experts. The majority of experts believe that rare thromboses should not be treated differently from lower extremity DVT, with the exception of SVT, where cytoreduction is useful to control myeloproliferation. Anticoagulant treatment with heparins followed by coumarins should be given for 3 months when the cause of thrombosis is identified, transient, and removable (eg, surgery, immobilization or oral contraceptive intake), and for 6 months in the presence of abnormalities that confer a moderately increased risk of thrombosis (eg, heterozygous factor V Leiden or G20210A prothrombin, deficiencies of protein C and protein S). Mimicking the strategy most often recommended in patients with lower extremity DVT, carriers of more severe thrombophilia traits are treated for 12 months or even longer if thrombosis was unprovoked and the underlying hypercoagulability particularly severe (antithrombin deficiency, lupus anticoagulant, combined abnormalities). Lifelong therapy is recommended in case of recurrent thrombosis.

SVT represents a particularly delicate setting, as these patients are often at increased risk of bleeding because of portal hypertension and/or the impairment of liver function. Many physicians recommend lifelong treatment in patients with rare thrombosis even after a single episode, in consideration of the critical sites in which these thromboses occur (particularly for CVT and SVT, less so for UEDVT). We believe that this behavior is unjustified, because there is little evidence that the recurrence rate is higher in rare thromboses than in lower extremity DVT. On the other hand, the rarity of these patients makes it difficult to predict that studies large enough to answer these questions will ever be conducted, so that large observational studies or international registries are warranted to circumvent this problem and to produce more evidence-based recommendations. At the moment, the choice of the duration of treatment is left to the judgment of caregivers, who must take into account the benefits but also the risks associated with lifelong anticoagulation.

Contribution: All of the authors contributed to the same extent to the design and writing of the final manuscript.

Conflict-of-interest: The authors declare no competing financial interests.

Correspondence: Pier Mannuccio Mannucci, Via Pace, 9, 20122 Milan, Italy; e-mail: piermannuccio.mannucci@unimi.it.