Estimates of the prevalence of PVT have fallen into a relatively broad range (about 4–25 %), probably due to variations in the characteristics of the patients and the definition used to define PVT [1, 3, 4]. Overall, it appears that in patients with cirrhosis admitted to hospital but otherwise unselected, the prevalence of partial and occlusive PVT is in the order of 7–10 % and 2–4 %, respectively. The incidence of PVT has been reported 7.8 % over a mean follow-up period of 12 months in patients wait-listed for LT [5], 16 % over a mean follow-up period of 16 months in patients participating in an endoscopic sclerotherapy program after variceal bleeding [6], and 10.7 % by 5 years when assessed prospectively in patients initially with Child A cirrhosis and no hepatocellular carcinoma (HCC) [7].

The causal factors most commonly implicated in the development of PVT are listed in Table 20.2. Searching for possible causes in patients with cirrhosis has generated many data. However, the cross-sectional design of most studies makes it difficult to infer whether cause or consequence explains the observed associations with PVT. Cross-sectional studies have shown PVT to be associated with smaller liver weight, higher model for end-stage liver disease (MELD), or Child–Pugh scores, ascites, and encephalopathy [4, 5, 8, 9]. A recent prospective study in patients with compensated cirrhosis at baseline found that PVT developed more frequently in patients with features of initially more severe liver disease, but there was no evidence for a direct temporal relationship between progression of liver disease and the occurrence of PVT [7]. Therefore, it remains unclear if progression of liver disease causes the development of PVT.

In patients with cirrhosis, PVT has been associated with decreased levels of coagulation inhibitors [9–11]. The direction of this association is likewise difficult to interpret because advanced liver disease induces a decrease in plasma levels of coagulation inhibitors (particularly protein C, but also protein S and antithrombin). Molecular studies of Factor V Leiden and prothrombin gene mutation have given inconsistent results regarding any association with the development of PVT [10, 11].

Recent studies have shown that contrary to general belief, thrombin generation capacity is preserved in plasma from patients with cirrhosis (provided platelet counts are above 60,000/μL), which contrasts with the decreased levels of most coagulation factors [10]. This apparent paradox is actually explained by a simultaneous decrease in the plasma levels of both coagulation inhibitors and most coagulation factors. Furthermore, a degree of resistance to the activation of the protein C pathway system has been shown, corresponding to a procoagulant state. This procoagulant state could be related to the marked decrease in plasma protein C levels, together with the marked increase in plasma factor VIII levels. The magnitude of these changes parallels the severity of cirrhosis. The clinical relevance of these laboratory changes is suggested by epidemiological evidence for an increased risk of venous thromboembolism in patients with cirrhosis. However, the data linking procoagulant changes with an increased risk of venous thrombosis in general—and PVT in particular—are still lacking.

A prospective longitudinal study disclosed a strong association of reduced portal vein blood flow velocity at baseline with the subsequent (1-year) development of PVT, independent of baseline MELD score [12]. In another study, however, the decrease in portal blood flow velocity with time was not found to be an independent factor for the later development of PVT [7]. The limitations in assessing portal blood flow velocity by noninvasive means cannot be ignored. This area clearly deserves further study.

Several surveys found PVT to be associated with previous splenectomy, surgical portosystemic shunting, or endoscopic therapy for esophageal varices [3, 9, 13]. However, in the absence of randomized control trials, it is not possible to assess whether surgery directly caused PVT, or whether the need for surgery (i.e., severe portal hypertension) was a marker for a greater risk of developing PVT.

Alcoholic cirrhosis, diabetes, and obesity have been associated with the development of PVT [13, 14]. However, a comprehensive assessment, taking into account all the possible causal factors for cirrhosis and particularly the metabolic syndrome, remains to be performed.

Routine imaging for HCC screening is the most frequent situation in which PVT is currently recognized, followed by a recent complication of cirrhosis, including gastrointestinal bleeding; and much less commonly, features of intestinal ischemia [4, 9]. It is difficult to determine whether symptoms or complications, if any, are directly related to the development of PVT or whether they led to a fortuitous uncovering of PVT. PVT in patients with cirrhosis does not appear to induce clinical or laboratory features of hepatic ischemia. However, among patients with cirrhosis, and acute ischemic hepatitis related to bleeding, the prevalence of PVT was 29 % [15], which is about twice the prevalence expected among unselected patients with cirrhosis and acute bleeding (16 %) [16].

An accurate diagnosis can be obtained at Doppler ultrasound of the portal vein and its main branches [17]. Doppler assessment is needed to avoid a false-negative result at ultrasound where a void-appearing portal vein can actually be occupied by a fresh thrombus. Enhanced computerized tomography (CT) or magnetic resonance imaging (MRI) confirms the diagnosis of PVT. It may be easier to assess the degree (partial or occlusive) and the extent (venous segments involved) at CT scan or MRI than at ultrasound.

The main differential diagnosis for PVT in patients with cirrhosis is portal venous invasion by a malignant tumor (usually HCC; Table 20.3). This entity has been mistakenly referred to as “malignant PVT,” although the obstruction is not related to thrombosis but to tumor ingrowth. The main differential feature is enhancement of the endoluminal material at the arterial phase of a CT or MRI scan [18, 19]. Additional features favoring a diagnosis of tumor invasion include proximity to a typical HCC nodule, a markedly enlarged portal vein, and washout of the endoluminal material at the portal and late phase [18, 19]. It is almost impossible to differentiate pure tumor invasion from tumor invasion with superimposed thrombosis. The clinical relevance of the latter distinction is doubtful, whereas the differentiation of pure thrombosis from malignant invasion is critical. A marked elevation in serum α-fetoprotein level may be seen with malignant vascular invasion.

In some patients, particularly those with large extrahepatic portosystemic shunts, portal flow is reversed (hepatofugal) or stagnant. Rarely, in such patients, the portal vein may not even be visible at all.

A spontaneous decrease in size or resolution of PVT has been reported in up to 40 % of patients at subsequent 3–6-month imaging [7, 20–22]. However, extension has also been reported in up to 72 % of patients not given anticoagulation [23]. Data are missing to clarify whether resolution is influenced by the partial or occlusive nature of the thrombus and the length of its extent. Short-term recurrence after disappearance also appears to be common but not constant [24]. Development of a portal cavernoma seems to be extremely unusual in patients with a persistent thrombus [7, 21, 22]. Therefore, venous changes following acute PVT differ considerably when cirrhosis is present from when it is absent [25].

The impact of PVT on outcome remains difficult to ascertain. Table 20.4 lists features associated with the development of PVT. As noted above, the association of PVT with the severity of cirrhosis could be explained by PVT causing liver disease to worsen. Indeed, PVT could exacerbate portal hypertension by superimposing a prehepatic block to the intrahepatic block, precipitating gastrointestinal bleeding and ascites formation, increasing portosystemic shunting and encephalopathy. Furthermore, by decreasing portal perfusion, PVT could induce parenchymal atrophy and worsen hepatic dysfunction. Studies that address this issue are sparse. In a prospective study, the development of PVT at any time during the course of initially compensated cirrhosis was not associated with a subsequent progression of liver disease [7]. Similarly, retrospective but longitudinal surveys disclosed no association between the persistence or the resolution of PVT and the progression of liver disease [21, 22]. In a recent controlled trial, enoxaparin administration for 48 weeks prevented the progression of liver disease, much more so than the development of PVT [26]. Therefore, it is unlikely that the obstruction to portal flow, created by a thrombus, explains the totality of the association between PVT and progression of liver disease. Actually, three scenarios could explain the association of PVT with liver disease progression: (i) advanced liver disease could precipitate the development of PVT, (ii) PVT could induce a progression of liver disease, and (iii) a common determinant (e.g., disordered hepatic or intestinal circulation) could independently and simultaneously explain the progression of liver disease and the development of PVT, as illustrated in Fig. 20.1. These scenarios are not mutually exclusive. Scenario (iii) appears to be most compatible with the data discussed above. Clarifying which of these scenarios is correct would tip the balance for or against potential treatments targeting portal vein recanalization.