Portal vein thrombosis – experience in a single centre

Authors


Portal vein thrombosis (PVT) is rare. Symptoms are non-specific and diagnosis is often incidental. Risk factors for PVT are multifactorial (Janssen et al, 2000; Sogaard et al, 2007), with cirrhosis and hepatobiliary malignancies being the most commonly identified local precipitating factors (Ogren et al, 2006). Systemic risks include both inherited and acquired thrombophilic conditions (Janssen et al, 2000; Bombeli et al, 2002) but detection in patients with PVT and liver disease is difficult due to the non-specific decrease in the plasma level of anticoagulant proteins. There is a well-described association between myeloproliferative disorders (MPD) and PVT (Kiladjian et al, 2008). Of recent renewed interest has been the association between PVT and the JAK2 mutation; the high specificity for MPD offers the opportunity to diagnose these conditions in patients with PVT (Kiladjian et al, 2008; Dentali et al, 2009). There are no randomized controlled trials of the optimum management of PVT. Anticoagulation is challenging, predominantly due to the bleeding risks associated with portal hypertension, varices, cytopenias and coagulopathy. Management is usually individualized, considering the underlying precipitating disorder and the bleeding against thrombotic risk.

We retrospectively analysed all cases of PVT diagnosed between January 2006 and July 2010 at University Hospitals of Leicester. Diagnostic inclusion criteria were of partial or complete PVT confirmed on imaging. Data was collected on demographics, presentation, local and systemic risk factors, anticoagulation management and recurrent thrombosis.

Data on 115 patients (67 male, 48 female) were available for analysis. Mean age was 60·1 years (range 1–93). Mean follow up was 4 years. Risk factors in this cohort are shown in Fig 1. Only 6·1% of patients had no identifiable risk factors for thrombosis with 65% of patients having two or more risks. Results of thrombophilia testing are shown in Table 1. 29·6% of the cohort was tested for thrombophilia and 55·9% of those tested were positive for at least one parameter.

Figure 1.

The percentage of patients with risk factors for thrombosis in this cohort. Recent surgery included any surgical intervention in the preceding year. Hormone-related included pregnancy, peurperium, oral contraceptive pill and hormone replacement therapy. ‘Other’ risks were inflammatory conditions (primary biliary cirrhosis, primary sclerosing cholangitis, ulcerative colitis, chronic pancreatits, sarcoidosis), chronic active infection (hepatitis B and C, human immunodeficiency virus), coeliac disease, perforated sigmoid colon, haemochromatosis, non-alcoholic steatohepatitis, nephrotic syndrome, α1 antitrypsin deficiency and congenital cardiac disease. VTE, Venous thromboembolism.

Table 1. Results of thrombophilia testing in this cohort
 Number testedNumber positivePercent (of tested) positive
LLAC23521·7
ACL2500·0
JAK2 V617F16318·8
PNH1000·0
F5 R506Q2015·0
F2 G20210A1700·0
 Number testedNumber reducedPercent (of tested) positive
  1. Parameters recorded were lupus anticoagulant (LLAC), anticardiolipin antibodies (ACL), presence of paroxysmal nocturnal haemoglobinuria clones (PNH), JAK2 V617F mutation, F5 R506Q, F2 G20210A and reduced levels of antithrombin (AT), protein C (PC) and protein S (PS). All testing was performed more than 4 weeks after the acute event as per local policy.

AT241250·0
PC17847·1
PS1715·9

38·3% were anticoagulated, 43·2% of those continuing in the long-term with warfarin. Documented reasons for not anticoagulating were cirrhosis, bleeding, varices, thrombocytopenia, chronic PVT and death before anticoagulation.

Death rate was high (55·7%) but only 7·8% of deaths were attributable to PVT. Death occurred at a median of 167 d after PVT (range 1–1675). In the 51 patients who were still alive, survival was a median of 3·5 years (range 0·9–13·2). Highest mortality was in the patients with malignancy and with malignancy and cirrhosis where 90·7% and 100% of patients had died, respectively; if these were excluded only 23·8% of patients were deceased. Death rate was lowest in the cohort of patients who had no risk factors (14·3%).

Five patients (4·3%) had recurrent thrombosis, three patients had a second recurrence. All had multiple risk factors for thrombosis including malignancy, MPD, central line, sepsis, cirrhosis, post-operative emergency splenectomy and heparin-induced thrombocytopenia. 16·7% of patients with reduced levels of anticoagulant proteins suffered a recurrent event.

Cirrhosis and malignancy were the most common risk factors for PVT, consistent with previous studies (Ogren et al, 2006). In the majority of cases, these diagnoses were made at the same time or subsequent to the PVT, prompting two suggestions; firstly, that patients presenting with PVT should be investigated for the presence of underlying cirrhosis and malignancy, and secondly, a higher index of suspicion for PVT is required in patients with known cirrhosis or malignancy presenting with abdominal symptoms.

The literature regarding prevalence of thrombophilia in PVT is variable (Denninger et al, 2000; Janssen et al, 2000; Bombeli et al, 2002; Sogaard et al, 2007). Few patients were investigated in our study, in accordance with recent guidelines (Baglin et al, 2010) but this approach can be questioned, as the prevalence in those tested was high and risk of recurrent thrombosis was threefold higher (16·7% vs. 4·3%) in patients with reduced AT compared to the whole cohort. The sample size was small and therefore statistical significance was not shown, however the difference was striking and this requires further study. Although antithrombin (AT), protein C (PC) and protein S (PS) levels are known to be reduced in patients with liver disease as well as following PVT (Denninger et al, 2000), only a quarter of our patients with reduced levels had evidence of cirrhosis. Whether the deficiency was inherited or acquired, recurrence rate was high in this group and may indicate a need for longer-term anticoagulation in these patients. Prevalence of F5 R506Q (Factor V Leiden) and F2 G20210A (prothrombin gene mutation) were no higher than in the general population, therefore did not appear to have significant risk for PVT.

The literature regarding prevalence of myeloproliferative disorders in PVT is also variable (Denninger et al, 2000; Janssen et al, 2000; De Stefano et al, 2007; Sogaard et al, 2007). 18·8% of those investigated were found to have mutation in JAK2; in one patient, this preceded any abnormality in the blood count. Four patients (3·5%) were diagnosed with an MPD, which probably under-represents the prevalence of this condition given the small number of patients tested. Whether routinely screening patients with VTE for this mutation is justified was addressed in a recent review (Dentali et al, 2009). The authors concluded that as JAK2 mutation is strongly associated with splanchnic vein thrombosis, screening is indicated in this patient group. Our findings would support this view.

Recurrent thrombosis was low, yet three of the patients with recurrence had an underlying MPD. This is surprisingly high given that an MPD was only found in four cases in total and reflects the strong thrombotic tendency in these patients. It also highlights the importance of making a diagnosis of MPD in patients with PVT, to enable appropriate follow up and management and adds further weight to JAK2 testing in this setting.

There is a general paucity of data regarding optimum management of patients with PVT. Most patients were not anticoagulated; a significant proportion due to increased bleeding risk. In the absence of randomized controlled trial evidence, management must be on an individual patient basis, taking into account risks of bleeding against risks of recurrent thrombosis. Current consensus opinion supports this approach (De Franchis, 2005).

In conclusion, we propose that patients with PVT should be investigated for cirrhosis and malignancy, which are associated with poor outcome and for AT, PC and PS deficiencies and JAK2 mutation, which are associated with high risk of recurrent thrombosis.

Author contributions

GS/SP designed the research study and analysed the data. GS performed the research and wrote the paper.

Ancillary