Portal hypertension (PHTN) can occur in cirrhotic and noncirrhotic patients and can be classified as presinusoidal or prehepatic (extrahepatic or intrahepatic), sinusoidal or hepatic, or post-sinusoidal or post-hepatic (Fig. 61-1). Portal pressure can be measured directly, or more commonly indirectly, by calculating the hepatic vein pressure gradient (HVPG) by subtracting the measured free hepatic vein pressure (FHVP) from the wedged hepatic vein pressure (WHVP). Portal pressure is normally <6 mm Hg and clinically significant PHTN is defined as an HVPG greater than 10 to 12 mm Hg (Table 61-1).

The most common complications of PHTN include gastroesophageal varices, portal hypertensive gastropathy, splenomegaly and hypersplenism, ascites, hepatic hydrothorax, hepatic encephalopathy, hepatorenal syndrome, hepatopulmonary syndrome, portopulmonary hypertension, and cirrhotic cardiomyopathy (Table 61-2). The management of PHTN has changed dramatically over the past two decades and has been the subject of several clinical practice guideline publications.¹ Medical, endoscopic, and radiologic management strategies have largely replaced many surgical procedures such as selective and nonselective shunts, devascularization procedures, and peritoneovenous shunts. This chapter emphasizes the role of current therapies in the management of patients with PHTN and its complications.

Although descriptions of PHTN and its complications go back over four centuries, surgical therapy was pioneered by Nicolai Eck, who first performed an end-to-side portacaval shunt in an animal model in 1883.² Pavlov described hepatic encephalopathy, referred to as meat intoxication at the time, as a consequence of diverting portal flow, which he believed was due to nitrogenous compounds that were not being cleared by the liver. The first portosystemic shunt in a human was performed by Vidal in 1903, but Whipple and colleagues pioneered the era of surgical decompression of portal hypertension in the 1940s. In the 1960s and 1970s, Drapanas developed the mesocaval shunt, Warren and Inokuchi developed selective variceal bed decompression with the distal splenorenal and coronary vein-caval shunts, respectively, Sarfeh studied and popularized small diameter H-grafts, and Sugiura pioneered gastroesophageal devascularization and splenectomy. In the 1980s, endoscopic sclerotherapy and band ligation were introduced for control of variceal hemorrhage. The subsequent development of pharmacologic therapy, transjugular intrahepatic portosystemic shunts (TIPS), and the pioneering work of Starzl and Calne in the 1960s and 1970s, making liver transplantation a viable alternative for patients with end-stage liver disease, have all radically transformed the care of these patients.

Portal hypertension results from increased resistance to portal flow in association with increased portal collateral flow. The increased resistance occurs most commonly within the liver due to cirrhosis, but it can occur prehepatic as in portal vein thrombosis (PVT) or post-hepatic due to obstruction of hepatic venous flow (Budd–Chiari syndrome or veno-occlusive disease) (Fig. 61-1). Rarely, a hepatic arterial-portal venous fistula can cause PHTN.

As Bleibel et al. have noted,³ resistance to portal flow is the result of structural as well as physiologic derangements. Under normal physiologic conditions there is little resistance to portal venous flow and there is little intrinsic regulation of portal flow. In cirrhosis, however, collagen deposition and fibrosis, along with the contractile properties of stellate cells and myofibroblasts that surround the hepatic sinusoids and reside in fibrous septa along with vascular smooth muscle cells, lead to an increased resistance to portal flow. Initially, the splanchnic vascular bed response is vasoconstriction due to the release of thromboxane A2, norepinephrine, endothelins, and angiotensin-II, along with decreased nitric oxide–mediated vasodilatation. With progression of portal hypertension, the release of splanchnic vasodilators such as nitric oxide and vascular endothelial growth factor predominates, resulting in increased splanchnic inflow.

These changes result in the development of collaterals between the portal and systemic circulations (Fig. 61-2), plasma volume expansion, increased cardiac output, systemic vasodilatation, and hypotension. The development of a systemic hyperdynamic circulation results in systemic blood pressures of 100 to 110 mm Hg, cardiac outputs ranging from 10 to 15 L/min, and low calculated systemic vascular resistance of 250 to 500 dynes/cm⁵ that can impact fluid resuscitation and patient management.

In developed countries, 90% of patients with PHTN have cirrhosis most often caused by chronic viral hepatitis (hepatitis B, C), alcoholic liver disease, hemochromatosis, and nonalcoholic steatohepatitis. Less common causes include autoimmune hepatitis, primary and secondary biliary cirrhosis, primary sclerosing cholangitis, medications (eg, methotrexate), Wilson disease, α-1 antitrypsin deficiency, celiac disease, idiopathic adulthood ductopenia, granulomatous liver disease, idiopathic portal fibrosis, polycystic liver disease, right-sided heart failure, Budd–Chiari syndrome, and veno-occlusive disease (Table 61-3). A smaller percentage of patients will have noncirrhotic PHTN usually caused by PVT or hepatic fibrosis. In other parts of the world, noncirrhotic PHTN is much more common and is most often caused hepatic schistosomiasis or PVT.

Patients with cirrhosis are often asymptomatic even though 80% have an elevated HVPG and nearly 50% will have esophageal varices. Patients with PHTN and no varices will develop varices at a rate of about 8% per year.⁴ Patients come to the attention of their physician as the complications noted in the introduction develop (Table 61-2). Briefly, the clinical manifestations of bleeding gastroesophageal varices are usually hematemesis and/or melena, but can present with shock and vascular collapse from exsanguinating hemorrhage. Portal hypertensive gastropathy is manifest as diffuse mucosal oozing and may present with anemia from chronic blood loss. Splenomegaly and hypersplenism may present with leukopenia, thrombocytopenia, and sometimes anemia. Ascites, the buildup of fluid in the peritoneal cavity, may present with abdominal distention, weight gain, and shortness of breath from the increased fluid and intra-abdominal pressure. Patients can also develop hydrothorax (pleural effusion) from movement of fluid from the abdominal cavity into the pleural space, usually on the right side. They can also present with spontaneous bacterial peritonitis with fever, pain, and tenderness. Patients often first present with confusion or hepatic encephalopathy that can be precipitated by many factors including bleeding, infection, renal failure, and other manifestations of liver failure. Hepatorenal syndrome is the development of renal insufficiency in patients with cirrhosis. Patients with hepatopulmonary syndrome may be asymptomatic in its early stages, but may present with oxygen desaturation, shortness of breath, and dyspnea on exertion caused by intrapulmonary shunting. Patients with portopulmonary hypertension may have similar symptoms related to elevated pulmonary artery pressures.

The severity of liver decompensation can be characterized by the Pugh-modified Child-Turcotte (Child-Turcotte-Pugh [CTP]) classification scheme (Table 61-4)⁵ or the Model for End-Stage Liver Disease (MELD) score (Table 61-5) (discussed later). The CTP score, which uses clinical assessment and laboratory values, has been used to determine the functional status of the liver and estimate liver reserve as well as predict morbidity and mortality after shunt surgery and other general surgical procedures in cirrhotic patients.^6,7 Patients with Child A cirrhosis have adequate liver reserve and tolerate shunt and general surgery with survival rates similar to noncirrhotic patients. On the other hand, Child C patients have a high mortality, often exceeding 50%, and in general are not candidates for shunt or general surgical procedures, but are candidates for liver transplantation. The CTP has been largely replaced with the MELD score that was originally developed to predict the mortality for cirrhotic patients undergoing TIPS and has been adapted to predict the mortality of patients on the liver transplant waiting list and to prioritize liver allocation to the sickest patients.⁸

The goals of diagnostic studies in initially evaluating patents with PHTN and its complications are to determine the presence of hepatic disease, the level of obstruction to flow, the presence and extent of intra-abdominal portosystemic collaterals, the direction of blood flow in the portal vein (PV) (hepatopetal, toward the liver; hepatofugal, away from the liver), as well as the presence of thrombosis. The imaging is typically multimodality, including duplex ultrasound (US) and multiphase computed tomography (CT) as well as multiphase magnetic resonance imaging (MRI), and conventional angiography. CT arteriography (CTA), CT venography (CTV), MR angiography (MRA), and MR venography (MRV) can also be useful in certain circumstances.

Duplex Ultrasound

Duplex US, which includes both gray scale and color Doppler sonography, is a noninvasive, portable, and inexpensive technique that does not use ionizing radiation and is frequently used as the first-line examination in the diagnosis and follow-up of patients with liver disease and PHTN.

Grayscale interrogation of the liver is used to evaluate overall morphology as well as to locate focal lesions suggestive of hepatocellular carcinoma. Color Doppler US is useful in evaluation of the portal, splenic, hepatic, and superior mesenteric veins, the vena cava, and the hepatic artery, including flow direction, flow velocity, and large vessel thrombosis. Portosystemic collaterals are readily identified, and findings such as paraumbilical veins, spontaneous splenorenal circulation, dilated left and short gastric veins, and hepatofugal flow within in the portal system are 100% specific US signs of clinically significant PHTN.⁹ Duplex US is also useful in identifying causes of PHTN other than cirrhosis, such as portal or hepatic vein thrombosis.¹⁰ In a recent prospective study, a PV average maximum velocity of <15 cm/s was the only variable independently associated with a high risk of nonmalignant PVT.¹¹ Splenomegaly (diameter >12 cm and/or area >45 cm²) is the most commonly associated and sensitive sign of the presence of PHTN, is an independent predictor of esophageal varices, and is associated with clinically significant PHTN in compensated cirrhotic patients.¹²

Other US signs of clinically significant PHTN include dilatation of the PV (diameter >13 mm), lack of or reduced respiratory variations of splenic and superior mesenteric vein (SMV) diameter,¹³ lack of respiratory caliber variations in one of the major portal tributaries (splenic vein [SV] and/or SMV),¹⁴ reduced PV velocity (maximal and mean velocity of PV flow, respectively <16 cm/s and <10 to 12 cm/s),¹⁵ increased congestion index of PV (ratio between the cross-sectional area and blood flow velocity),¹⁶ altered hepatic venous Doppler pattern,¹⁷ increased intraparenchymal hepatic and renal artery impedance,¹⁸ and reduced mesenteric artery impedance.¹⁹

Finally, Duplex US is useful in the surveillance of patients who have been decompressed by TIPS. Doppler US evaluation can diagnose TIPS dysfunction by demonstrating in-stent velocities >250 cm/s or <50 cm/s with greater than 90% sensitivity and specificity.²⁰

Multiple-Detector Computed Tomography, CT arteriography, CT venography

Multiple-detector computed tomography (MDCT) uses ionizing radiation to acquire a volumetric data set that allows three-dimensional multiplanar reconstructions. Noncontrast images are acquired, intravenous contrast administered, and dynamic images are acquired in the hepatic arterial phase (12 sec following injection), portal venous phase (55 sec following injection), and finally in the late venous phase (120 sec following injection).²¹

MDCT can identify morphologic changes including hepatomegaly in the early stages of cirrhosis. In the later stages, there is frequent atrophy of the right lobe and medial segment of the left lobe with hypertrophy of the left lateral segment. MDCT can also characterize regenerative nodules, dysplastic nodules, and hepatocellular carcinoma (HCC), determine the patency of the venous and arterial systems, and measure liver volumes, which can be useful in preoperative planning for liver transplantation.

Portosystemic collaterals (varices) develop as portal pressures rise above 10 mm Hg, are a hallmark of PHTN, and appear as enhancing, well-defined tubular or serpentine structures that follow the enhancement characteristics of the PV. The left gastric (coronary) vein (LGV) is the most commonly seen portosystemic collateral and is dilated in up to 80% of cirrhotic patients. It normally drains the anterior and posterior surfaces of the stomach and ascends the lesser curvature to the gastroesophageal junction, where it receives the esophageal vein and supplies esophageal and paraesophageal varices that drain into the azygous/hemizygous venous systems.²² The LGV drains into the PV at the superior border of the duodenum. Dilated left gastric veins are visible between the anterior wall of the stomach and the posterior surface of the left hepatic lobe. A left gastric vein size larger than 5 to 6 mm or multiple veins 4 to 6 mm in diameter are indicative of PHTN. A left gastric vein greater than 7 mm has been shown to correspond to a HVPG of 10 mm Hg.²³ Approximately 30% to 70% of patients with PHTN develop varices and 9% to 36% are considered “high risk.”²⁴

Paraumbilical venous collaterals are found in 43% of patients with PHTN. They appear as tubular enhancing structures in the falciform ligament and are supplied by the left PV via a recanalized umbilical vein. They connect with the superior epigastric vein and/or internal thoracic veins which drain into the superior vena cava or connect with the inferior epigastric vein that drains into the external iliac vein. These abdominal wall varices form the caput medusae that are seen on physical exam as dilated subcutaneous veins at the umbilicus.

Gastric varices form in the face of PHTN and decreased drainage through the left gastric, posterior gastric, and short gastric veins. In this setting, portosystemic shunting can be through the left inferior phrenic or left adrenal veins forming a gastrorenal shunt. The incidence of gastric varices in patients with PHTN is approximately 30%.²⁰ Retroperitoneal varices are commonly seen collaterals and form between intestinal or retroperitoneal tributaries of the SMV or IMV and the systemic circulation.

PVT may be present in 1% of patients with early cirrhosis, in 30% of patients with advanced cirrhosis who are candidates for liver transplantation, and in 10% to 40% of patients with HCC,²⁵ as well in patients with no cirrhosis but who may have a hypercoagulable state. PVT is demonstrated by an intraluminal filling defect on US (Fig. 61-3A) and on CTA after the administration of IV contrast (Fig. 61-3B, C), and can be associated with bowel ischemia (Fig. 61-3D). After PVT, there is rapid development of numerous enhancing venous collaterals in the porta hepatitis that bypass the obstruction, referred to as cavernous transformation.²⁵

Magnetic Resonance Imaging, MR Angiography, MR Venography

MRI uses a high field-strength magnet in combination with radiofrequency energy to create three-dimensional images that can be viewed in multiple planes. Gadolinium chelates are used as contrast agents to improve tissue contrast and to perform angiography. Newer, liver-specific contrast agents such Eovist have been developed, with up to 50% of the injected dose of these liver-specific agents taken up by functioning hepatocytes and excreted in the bile. As with MDCT, MR imaging of patients with diffuse liver disease and PHTN is a dynamic process; T1 in-phase and out-of-phase, T2, and diffusion-weighted sequences are performed. Post-contrast studies are acquired in the arterial phase (20-35 sec), portal phase (70 sec), and equilibrium phase (3 min). A fourth phase is added with liver-specific agents and a delayed hepatobiliary phase at 20 minutes.²⁶

MRA can be performed using multiple techniques both with and without the administration of gadolinium contrast agents, and has an added benefit over MDCT in that it can provide information on both flow direction and flow velocity. MRI/MRV has proven useful in the evaluation of the portosystemic collaterals and the vena cava and evaluation of PVT, and compares equally to MDCT. MRI performs equally as well as MDCT in the diagnosis of HCC, but has increased accuracy in detecting smaller lesions.²⁶

Endoscopy

Esophagogastroduodenoscopy (EGD) is the gold standard for the diagnosis of esophageal and gastric varices and variceal hemorrhage⁴ (Fig. 61-4A-D). It is recommended that esophageal varices be classified as small (<5 mm in diameter) or large (>5 mm in diameter), with the large varices including medium-sized varices when three grades are used (small, medium, and large). Care should be taken to document the presence or absence of red signs (red wale marks or red spots) on varices that identify high-risk varices. Because therapy with β-blockers prevents bleeding in more than half of patients with medium or large varices, it is recommended that newly diagnosed patients with cirrhosis undergo screening EGD for varices. The prevalence of medium/large varices is about 15% to 25%, so most patients will have a negative EGD or have varices that do not warrant prophylactic treatment. Other noninvasive markers of varices such as platelet count, FibroTest, spleen size, PV diameter, and transient elastography have so far been inaccurate and cannot substitute for screening EGD. It is recommended that patients with no varices and compensated cirrhosis should undergo repeat EGD in 2 to 3 years. Patients with small varies should undergo repeat EGD in 1 to 2 years. Patients with decompensated cirrhosis should have yearly EGD. Esophageal capsule endoscopy may play a role in the future in screening for esophageal varices, although it is still not as sensitive as EGD.²⁷

Angiography and Measurement of Hepatic Venous Pressure Gradient

Historically, angiography played a larger role in the evaluation of patients with PHTN. The portal venous system was indirectly evaluated in the venous phase of celiac or superior mesenteric artery angiography or directly imaged through splenoportography or through transhepatic or transjugular portal venography (Fig. 61-5). However, with the advent of cross-sectional imaging, these diagnostic techniques are now rarely used.

Angiography still plays an important role in measurements of the HVPG that is an indirect measurement of portal venous pressure. The procedure is usually performed from a jugular vein access. The inferior vena cava (IVC) is cannulated with a diagnostic catheter and the right hepatic vein is subselected. If using a straight, end hole catheter, it is advanced peripherally into the hepatic vein until wedged, rendering the WHVP. Pressure measurements are obtained with the transducer at the right atrial level (mid-axillary line). A subsequent injection of contrast confirms the location by visualization of a characteristic sinusoidal pattern. The catheter is then withdrawn into the hepatic vein, and an FHVP is obtained. Alternatively, a balloon-tipped catheter is advanced into the middle third of the right hepatic vein and inflated to occlude the vein, which allows measurement of the WHVP. Again, injection of contrast with the balloon inflated confirms balloon occlusion of the hepatic vein. The FHVP is obtained after deflating the balloon. A prospective trial compared these two techniques against direct portal pressure measurements and found that that the balloon occlusion method is more reproducible and more accurately reflects the direct portal pressure.²⁸

The HVPG is calculated by subtracting the FHVP from the WHVP, which eliminates the changes in the pressure measurements caused by changes in intra-abdominal pressure. At the time of pressure measurements, transjugular liver biopsy can also be safely performed and is particularly advantageous in patients who do not meet criteria for percutaneous or open liver biopsy because of uncorrectable coagulopathy or ascites.²⁹

A normal HVPG is less than 6 mm Hg, and clinically significant PHTN develops at pressures of 10 to 12 mm Hg with the development of varices. Significant complications such as variceal bleeding and ascites usually arise at pressures greater than 12 mm Hg.³⁰

Ascites

Approximately 85% of patients in the United States with ascites have cirrhosis, and 15% have nonhepatic etiologies including cancer, heart failure, tuberculosis, or nephrotic syndrome.³¹ Ascites is usually suspected on the basis of the history and physical examination and confirmed by ultrasound and by paracentesis. Ascites from PHTN can easily be differentiated from other causes by straightforward fluid analysis. The diagnosis of spontaneous bacterial peritonitis (SBP) can be made by determining the absolute number of PMNs in the fluid (>250 PMNs per mm³).

MEDICAL MANAGEMENT

Recommendations for medical management of ascites include abstinence from alcohol if this is the cause of liver disease. Additional recommendations include sodium restriction and diuretics (spironolactone and furosemide), fluid restriction if the serum sodium is less than 120 to 125 mmol/L, and an initial therapeutic paracentesis in patients with tense ascites (Fig. 61-6). Patients who respond to diuretics are preferentially treated with sodium restriction and diuretics rather than serial paracentesis.³¹ Patients with cirrhosis and ascites should be considered for liver transplantation.

Patients have refractory ascites if they have fluid overload that is unresponsive to sodium restriction and high-dose diuretics or recurs quickly after undergoing therapeutic paracentesis. Treatment options include serial therapeutic paracentesis, TIPS, peritoneovenous shunt, or liver transplantation.

Controlled trials of serial large volume (5-10 liters) therapeutic paracentesis demonstrating its effectiveness and safety have been published³² and may be performed as often as biweekly or weekly. Although the administration of albumin after large-volume paracentesis is of uncertain value, it is recommended that an albumin infusion of 6 to 8 g/L of fluid be considered, particularly for volumes greater than 5 liters.³¹

SURGICAL MANAGEMENT

Patients with refractory ascites should be referred for liver transplantation unless there are other surgical or medical contraindications, because 21% die within 6 months.³³ Peritoneovenous shunts (LeVeen or Denver shunts) currently have few indications because of their poor long-term patency due to a high incidence of thrombotic, infectious, and technical complications. Furthermore, controlled trials have demonstrated no survival advantage compared to medical management.³⁴ Peritoneovenous shunts are now reserved for patients who are not candidates for serial paracentesis (eg, distance from a physician able to perform the procedure) and for patients who are not candidates for liver transplantation or TIPS.

RADIOLOGIC MANAGEMENT

Ultrasound

Although paracentesis in patients with massive ascites can be safely and easily performed without US guidance, in patients who are obese or otherwise have fluid that is difficult to localize by physical examination, paracentesis can be aided by performance under US guidance.

Transjugular Intrahepatic Portosystemic Shunt

The development of refractory ascites, which occurs in 5% to 10% of patients with ascites, is associated with a 1-year mortality of 50% to 90% and is a common indication for liver transplantation.^32,33 By reducing portal pressure, TIPS (discussed in more detail in the section on varices) has been shown to be effective in managing patients with refractory ascites.

A meta-analysis of five randomized control trials by D’Amico et al.³⁵ demonstrated a 7.1-fold reduction in the risk of recurrence of tense ascites after TIPS. Rates of improvement ranged from 38% to 84% after TIPS compared to 0% to 43% after large volume paracentesis. There was a trend toward a reduction in mortality in the TIPS group. Rates of hepatic encephalopathy were 2.2-fold higher in TIPS patients compared to repeated large-volume paracentesis. A meta-analysis by Salerno et al.³⁶ and review by Eesa et al.³⁷ also demonstrated TIPS to be superior to repeated large-volume paracentesis in controlling ascites and was associated with a significantly better transplant-free survival at 12 and 24 months.

The threshold portosystemic gradient (PSPG) (see section on TIPS for treatment of variceal hemorrhage) in the treatment of patients with refractory ascites has been a subject of some debate. The Society of Interventional Radiology (SIR) and American Association for the Study of Liver Disease (AASLD) guidelines recommended reducing the PSPG to less than 8 mm Hg.³⁸

Hepatic Hydrothorax

Hepatic hydrothorax develops in patients with cirrhosis and ascites when there is direct communication between the abdominal and thoracic cavities. It may develop in patients without clinically evident ascites. In most patients the defect is over the dome of the liver.³⁸ The use of TIPS in the treatment of hepatic hydrothorax is supported by several retrospective case series describing the outcomes of more than 150 patients.³⁹ At least partial improvement in clinical symptoms (dyspnea and decrease in frequency of thoracentesis) has been reported in 68% to 82% of patients, whereas complete resolution of hydrothorax was observed in 57% to 71% of patients.

Gastric and Esophageal Varices

Gastroesophageal varices occur in patients with cirrhosis and PHTN with an HVPG of at least 10 to 12 mm Hg. They are present in 50% of patients with cirrhosis, but their incidence is associated with the severity of the liver disease, ranging from 40% in patients with compensated cirrhosis to 85% in patients with decompensated cirrhosis. Varices develop at a rate of 8% per year in patients without varices, and the HVPG is the strongest predictor for their development.⁴

Variceal hemorrhage is the most life-threatening complication of portal hypertension and occurs at a yearly rate of 5% to 15% and at an overall rate of 25% to 70%⁴ (Fig. 61-7). Large varices on EGD and the presence of cherry red spots and red wale marks—linear, dilated venules that endoscopically look like whip marks on the variceal surface—are associated with a higher risk of hemorrhage (Fig. 61-4B, C), as is the degree of elevation of the HVPG. In spite of improvements in care, the mortality of esophageal variceal hemorrhage is still about 20%, but exceeds 60% in patients with HVPG >20 mm Hg.^40,41 This is in part related to the fact that patients with an HVPG >20 mm Hg are at increased risk for early rebleeding or inability to control bleeding. Overall for survivors of a first bleed, 30% will rebleed within 6 weeks, and by 1 to 2 years nearly 60% will have rebled.^42,43 Gastric varices are less common (5%-33%) than esophageal varices in patients with PHTN and about 25% will bleed over 2 years.

MEDICAL MANAGEMENT

Prophylaxis

Preventive management of variceal hemorrhage includes pharmacologic and endoscopic therapy. Nonselective β-blockade involves β₁ and β₂ adrenergic blockades, which are known to decrease cardiac output (β₁) and increase splanchnic arteriolar vasoconstriction (β₂), reducing portal flow (eg, propranolol and nadolol) (Table 61-6). Vasopressin also has splanchnic vasoconstrictive properties and decreases portal venous collateral flow and portal pressure. Somatostatin decreases splanchnic flow, but does so indirectly by reducing glucagon, substance P, and vasoactive intestinal peptide.