Portal Hypertension Related to Bleeding

Physiologic Principles of Portal Hypertension

The pressure within a vessel is determined by the flow and resistance within that vessel. This relationship is expressed by Ohm's law: Δ P = Q × R , where P is pressure, and Q and R are flow and resistance, respectively. Therefore either increase in flow or resistance can lead to an increase in pressure. In the case of portal hypertension, the increase in pressure occurs through both an increase in flow (Q) and an increase in resistance (R) . Resistance depends on a number of factors as defined by Poiseuille's law: R = 8 nL /π r , where n is the coefficient of viscosity, L the length of the vessel, and r the radius. Radius appears to be the most important factor, as small changes in radius are associated with large changes in resistance. These principles can be applied in relation to the development of portal hypertension and also for vascular regulation within each of the vascular beds in portal hypertension.

The vascular beds most salient for pathogenesis of portal hypertension include the intrahepatic circulation, the splanchnic circulation, and the portosystemic collateral circulation ( Fig. 16-1 ). Changes in the intrahepatic circulation are thought to be the primary impetus in the development of portal hypertension. These changes are characterized by an increase in intrahepatic resistance, the mechanisms of which are described in further detail below. Such changes not only initiate the development of portal hypertension but also result in secondary events that lead to a hyperdynamic circulatory state characterized by systemic and splanchnic vasodilation. Splanchnic vasodilation allows greater flow into the portal circulation, thereby further potentiating portal hypertension. At a critical level of portal pressure elevation, there is an expansion of the collateral circulation, which serves to decompress the portal circulation. As described further below, portal pressure reduction through collaterals comes at the expense of a number of complications of portal hypertension, including esophageal varices, and yet these collaterals fail to normalize portal pressure.

Increased Intrahepatic Resistance

Although it was previously thought that increased blood flow into the portal circulation may be the major driver of portal hypertension, current concepts indicate that an increase in vascular resistance is the major driver of most forms of portal hypertension. In early studies of cirrhotic livers, distortion and reduction of the hepatic microcirculation were observed. These vascular changes, in concert with compression of portal and sometimes hepatic veins by regenerative nodules, were believed to be a major cause for the increase in vascular resistance observed in cirrhotic livers. Portal hypertension may be found, however, without cirrhosis, and this observation has led investigators to question whether deposition of collagen alone is an important factor in increasing intrahepatic vascular resistance. Indeed, there also appears to be prominent changes in vascular structure and patency, which may precede actual fibrosis ( Fig. 16-2 ). For example, thrombosis of the portal and hepatic veins is thought to lead to ischemia, loss of parenchyma, and worsening of the fibrosis, supporting the idea that vascular change can precede and possibly promote cirrhosis. Some of these observations led to the concept that increased intrahepatic resistance in cirrhosis may have an irreversible component relating to fibrosis and a reversible component that was more dynamic in nature. Experimentally, the concept of a reversible and dynamic component of increased intrahepatic resistance was further crystallized by studies by Bhathal and colleagues which demonstrated that a vasodilator, sodium nitroprusside, could reduce perfusion pressure in the isolated perfused rat liver preparation. This observation was subsequently expanded in vivo and mechanistically and formed the basis for the concept that the nonfibrotic component of cirrhosis was an important aspect of portal hypertension and was amenable to therapeutic intervention. This is especially important because at the present time, the fibrotic component of portal hypertension is less amenable to pharmacologic modulation in humans.

The cell biology of hepatic vascular cells and their paracrine regulation of vascular structure and function have emerged as important drivers of the pathogenesis of increased intrahepatic resistance in portal hypertension, as well as a site for therapeutic intervention. Paradigmatically, this is represented by the interplay of two cells, the sinusoidal endothelial cell and the hepatic stellate cell. As previously mentioned, the sinusoidal endothelial cell is a unique endothelial cell in terms of its phenotype. It maintains fenestrae, which facilitate macromolecular transport across the space of Disse. Whereas it was initially thought to be a cell that exclusively regulates such transport functions, more recent models indicate that this cell also maintains traditional vasoregulatory functions within the hepatic circulation, including paracrine regulation of adjacent hepatic stellate cells via production of a number of vasoregulatory molecules including the canonic vasodilator and vasoconstrictor, nitric oxide (NO), and endothelin, respectively. Regulating mechanisms of NO are shown in Fig. 16-3 .

Hepatic stellate cells are interspersed within the hepatic sinusoids in the normal liver. However, in response to liver injury, these cells undergo a process termed activation , which is characterized by enhanced proliferation, migration, and collagen deposition capacity. These changes enhance the ability of sinusoidal endothelial cells to regulate the sinusoidal structure and function through hepatic stellate cells, which thereby reach the critical mass and level of activation necessary to act as an effector cell for sinusoidal endothelial cell-derived molecules (see Chapter 5 ). Although the hepatic stellate cell is the canonic cell type thought to mediate effector functions from endothelial cell–derived molecules, recent work suggests that the peribiliary fibroblast and other related mesenchymal cells can also activate to myofibroblasts and serve a similar function to the hepatic stellate cell. One theory proposes that epithelial cells may also transactivate into myofibroblasts through a process termed epithelial-mesenchymal transactivation (EMT). Thus the pool of contractile effector cells could be diverse in phenotype and may not be limited to hepatic stellate cells.

NO and endothelin are two prototypical vasoactive molecules that regulate interactions between sinusoidal endothelial cells and hepatic stellate cells and are discussed in more detail below. Many other vasoactive signaling pathways also exist and are relevant for portal pressure regulation but will not be discussed here.

In the intrahepatic system it is believed that an underproduction of NO contributes to increased intrahepatic vasoconstriction and increased intrahepatic resistance. This is in contradistinction to increases in NO generation in the splanchnic and systemic circulation that lead to the hyperdynamic circulation and increased portal inflow, which are discussed further below. The release of NO is reduced from the endothelial cells of cirrhotic animals, and the molecular mechanisms responsible for this hepatic endothelial dysfunction are an area of active investigation. Some experimental therapeutic approaches have determined that if this relative NO deficiency could be corrected then there would be a fall in intrahepatic resistance and portal pressure, suggesting that targeted increases in NO levels in the hepatic circulation could be used to lower portal pressure. For example, if drugs, such as nitrates, can be developed that increase NO levels in the portal system without causing further vasodilation in the systemic circulation, then new therapeutic opportunities could be possible. The most promising approaches at this time have included the statin class of medications, which appear to stimulate NO generation in the intrahepatic circulation without adverse NO generation in other vascular beds in both experimental models and in humans. This class of drugs requires further investigation as a potential therapy that directly targets portal hypertension.

Endothelins are a group of compounds that are potent vasoconstrictors. They bind to two different types of receptors termed ET _A and ET _B . Binding of endothelins to ET _A receptors on vascular smooth muscle cells leads to vasoconstriction, whereas binding to ET _B receptors on endothelial cells leads to release of NO and vasorelaxation. A number of different studies in combination indicate that endothelins may increase portal pressure in liver disease by binding to hepatic stellate cells, leading to their contraction and a rise in resistance within the liver microcirculation. In support of this idea are the findings that the acute administration of an ET _A/B antagonist leads to a fall in portal pressure in cirrhotic rats. Chronic administration of an ET _A/B antagonist in an animal model, however, failed to lower portal pressure. Studies in humans using these receptor antagonists are awaited. Concerns about this pathway as a therapeutic target in human portal hypertension include the need for evidence of ET _A/B antagonist safety profile in setting of liver dysfunction, better understanding of the effects of the various receptor subtypes, and the overall quantitatively small effect of modulation of this pathway that has been observed in animal and in vitro models.

Hepatic arterial flow also contributes to sinusoidal pressure because the arterial contribution to sinusoidal inflow is significant. Furthermore, with reduced portal blood flow into the cirrhotic liver there is a compensatory increase in hepatic arterial blood flow that aims to maintain total hepatic blood flow constant. Recent studies indicate that the hepatic artery participates in the generalized hyperdynamic arterial flow state that characterizes the systemic and splanchnic circulatory beds in portal hypertension, which is described in further detail below. This could theoretically increase intrahepatic resistance and exacerbate portal hypertension. Overall, the role of the hepatic artery in increasing portal pressure requires further investigation.

Recently, there has been greater emphasis on the vascular structural changes that occur in parallel with cirrhosis and portal hypertension. This includes remodeling of the hepatic sinusoidal vasculature and angiogenesis, which is the proliferation of existing endothelial cells. Close links between the processes of angiogenesis and cirrhosis have been identified with both processes going hand in hand. These have suggested that angiogenesis could also be a new target for portal hypertension treatment, although more work is needed in this regard. The sinusoidal remodeling changes in portal hypertension are also quite distinct. These include an increase in the mass of hepatic stellate cells that wrap around the endothelial cell tube. With regards to the endothelial cells themselves, they also undergo changes in phenotype characterized by dedifferentiation that includes loss of fenestrae, and development of basement membrane, termed capillarization . The pathologic significance of these changes is also anticipated to have therapeutic significance because reducing the contractile machinery and force of stellate cells should reduce intrahepatic resistance. Another prominent vascular structural change is the presence of “scar vessels” that transverse through dense cirrhotic scar. It has been postulated that these vessels may provide the metabolic and oxygen needs required for the scar to progress, akin to the role of angiogenesis required for tumors to continue to grow. Thus, the vascular changes of cirrhosis and portal hypertension may provide targets for therapies that not only target vasoregulation but also the related angiogenesis and sinusoidal structural changes that link intimately to the fibrotic process.

The advances in molecular mechanisms of portal hypertension are leading to a number of potential therapeutic agents that can eventually be tested in humans. In addition to the advances targeting specific complications of portal hypertension that are discussed in other chapters, there are also a number of agents that can directly target the elevated portal pressure and increased intrahepatic resistance. Presently, some of these agents include liver specific NO donors, statin class of medicines that stimulate intrahepatic NO generation, blockers of the endothelin pathway, blockers of the angiotensin pathway, and blockers of growth factors including platelet-derived growth factor (PDGFR)/TGF-β pathways, especially receptor tyrosine kinase inhibitors, such as sorafenib and imatinib. In total, most of these approaches are targeting hepatic stellate cell contractility, proliferation, migration, and activation, or alternatively are stimulating endothelial cell activation and its production of vasodilatory molecules such as NO.

Increased Splanchnic Blood Flow and the Hyperdynamic Circulatory State

Increased portal blood flow is an uncommon cause of portal hypertension in and of itself (aside from splanchnic arteriovenous shunts and splenomegaly), but rather is usually a propagator of portal hypertension triggered by increased intrahepatic resistance. This is because there is little outflow resistance from the liver, and so the increase in flow must be quite large to raise sinusoidal pressure purely on an inflow basis. However, in the setting of increased intrahepatic resistance, increased flow into the portal circulation is an important propagator of portal hypertension and in fact represents the most widely used site of pharmacologic therapeutic intervention such as octreotide, vasopressin, somatostatin, and β-blockers. Thus when the vascular resistance is increased, as is observed in cirrhosis, small increases in portal venous inflow may be associated with significant increases in portal vein pressure. Importantly, in cirrhosis, total flow entering the portal system does not equate to total flow entering the liver because the increase in pressure leads to portal-systemic collaterals that divert a significant component of the total flow entering the portal circulation, which are discussed in greater detail below.

With the development of portal hypertension, there is also the development of the hyperdynamic circulation , the pathogenesis of which is discussed in detail in Chapters 15 and 17 . In brief, the following sequence of events appears to occur in the patient with portal hypertension, leading to circulatory disturbances. Pressure in the portal circulation is increased either because of hepatic fibrosis or occlusion of the portal vein. Although portal blood flow into the liver declines, hepatic blood flow is partially maintained by an increase in hepatic arterial flow. In response to the rise in portal pressure, portal-systemic collaterals develop. Although this does not fully decrease portal hypertension, the shunting does cause vascular resistance to fall in the splanchnic bed, leading to the development of the hyperdynamic circulation. Splanchnic and portal venous inflow increases result in portal pressure to continue to be elevated despite the opening of the collateral circulation. As resistance in the liver continues to increase, there is a further increase in portal pressure, a fall in liver perfusion by the portal vein, and an increasing percentage of blood that is shunted through the collateral circulation. The liver is therefore deprived of portal blood, which may, over time, accelerate the progression of liver disease even if the underlying cause of cirrhosis has been reversed. Importantly, the hyperdynamic circulation contributes not only to the development of portal hypertension but also to the development of the hepatopulmonary syndrome (see Chapter 18 ), cirrhotic cardiomyopathy (see Chapter 18 ), and ascites and hepatorenal syndrome (see Chapters 15 and 17 ). Interestingly, the changes in the splanchnic circulation in portal hypertension may be viewed as a microcosm of the hyperdynamic systemic circulatory state observed in patients with cirrhosis, although there is controversy about this assertion because some vascular beds outside the splanchnic circulation may experience vasoconstriction rather than vasodilation.

From a vascular cell biology perspective, NO overproduction appears to contribute to the development of the hyperdynamic circulation ( Fig. 16-4 ). This conclusion is based on the findings in humans that in exhaled air, levels of NO are increased in cirrhotics before but not after liver transplantation. Additionally, patients with cirrhosis have increased plasma concentrations of NO. However, it is increasingly clear that whereas NO is a major driver of the hyperdynamic circulatory state, there are numerous other redundant and compensatory pathways, many of which are under active investigation.

The primary driver of increased NO generation remains unclear. Two mechanisms have received the greatest experimental attention: Vascular endothelial growth factor (VEGF) and mechanical stress induced by hemodynamic forces. For example, VEGF production is increased in cirrhosis and is a well-known stimulus of NO generation. Similarly, biomechanical forces, such as shear stress and stretch, are known to stimulate NO generation from endothelial cells and are increased in the splanchnic circulation of cirrhosis. However, a causal versus correlative relationship between these parameters continues to be actively debated.

Portosystemic Collateral Circulation in Portal Hypertension

Large portosystemic collaterals are seen in patients with portal hypertension and the higher the pressure the more extensive the collaterals. Indeed, these collaterals account for a significant component of morbidity and mortality attributable to patients with cirrhosis. For example, variceal formation and hemorrhage is directly attributable to gastroesophageal portosystemic collateral vessels that develop as a consequence of increased portal pressure. Although portosystemic collaterals that develop in response to the rise in portal pressure will tend to minimize the rate of rise in pressure, in the end the collateral circulation is insufficient to compensate for the factors that raise portal pressure.

Collateral vessels develop through several distinct but interrelated processes. These include vasodilation of existing collateral vessels, vascular remodeling of existing vessels, and angiogenesis; that is, the de novo formation of new vascular sprouts ( Fig. 16-5 ). Although vasodilation is a relatively rapid phenomenon, some of these other pathways require a longer timeframe of chronic portal hypertension that allows for the changes in vascular wall structure leading to remodeling of the vessel and sprouting of new endothelial sprouts. These interrelated processes are likely driven by both changes in growth factor levels in cirrhosis and mechanical factors related to increased pressure within the portal circulation that drives increased flow into the collateral bed. Importantly, these mechanisms have potential therapeutic implications. For example, VEGF is a growth factor that has been demonstrated to mediate the angiogenic collateral vessel response in portal hypertension, with inhibition of VEGF attenuating collateral vessel formation in animal models. Parallel strategies may be worthy of pursuit in humans with cirrhosis if safety of these drugs can be established.

Several factors eventually determine whether the collateral vessel will rupture as best exemplified by bleeding esophageal varices. These factors include the size of the varix, the thickness of the varix wall, and the pressure gradient between the variceal lumen and the esophageal lumen. The interplay of how these factors determine the risk of varix rupture is shown in Fig. 16-6 .

Portosystemic Collaterals

The portal venous system may decompress into the systemic venous system at several different sites. The most important site for this collateral circulation is within the mucosa of the proximal stomach and distal esophagus. When these collateral vessels dilate, gastric and esophageal varices develop. The normally obliterated umbilical vein, which lies in the ligamentum teres , is recanalized with increases in hepatic sinusoidal pressure and connects the left portal vein to systemic veins around the umbilicus. These veins then drain into the epigastric vessels and appear as caput medusae . Because the umbilical vein drains into the left portal vein, the presence of caput medusae rules out extrahepatic portal hypertension as the cause of portal hypertension. If the flow in the umbilical vein is high, an audible venous hum (Cruveilhier-Baumgarten murmur) may be heard over the course of the umbilical vein. When the inferior vena cava is occluded, as in Budd-Chiari syndrome, the veins in the flanks are more dilated and drain upward into the superior vena caval territory. These veins may be best appreciated by examining the back and the flank of the patient.

Large venous shunts may also form between the splenic vein and the left renal vein ( Fig. 16-7 ). These shunts are often large enough to decrease the risk of variceal bleeding, but may increase the risk of hepatic encephalopathy and portopulmonary hypertension. Collaterals may also develop between the portal venous system and the abdominal wall in relation to surgical scars or surgically created ostomies ( Fig. 16-8 ). Rectal varices ( Fig. 16-9 ) develop from collaterals between the superior hemorrhoidal vein, which continues as the inferior mesenteric vein, and the inferior rectal vein, which drains into the systemic pudendal vein. The prevalence of hemorrhoids in patients with portal hypertension may not be increased, although occasionally hemorrhoidal bleeding can be severe.

Assessment of Portal Venous System

The most common method of assessing for esophageal varices is upper gastrointestinal endoscopy. However, there are several other methods, including radiologic imaging and capsule endoscopy, to detect varices. An accurate comparison of the varying techniques to detect esophageal varices is difficult because the current gold standard for detecting varices, upper gastrointestinal endoscopy, may miss gastric varices, and considerable observer variation exists in assessing the size of esophageal varices.

Esophageal varices develop in response to an increase in portal pressure. The correlation between the size of esophageal varices and portal hypertension is variable because there are several other beds where collaterals may form, and the degree of decompression into these beds is variable. In addition, valves are present in the perforating veins of the esophagus that may prevent the flow of blood from periesophageal veins into vessels within the esophageal mucosa. The valves may become incompetent in some patients with portal hypertension, which may lead to an increase in variceal size.

Upper Gastrointestinal Endoscopy

The most common method of identifying and determining the size of gastroesophageal varices is by upper endoscopy. Various classifications are used to describe esophageal varices, namely size, form, and color. The simplest method of grading esophageal varices is based on size: small or large. Small varices are less than 5 mm in diameter ( Fig. 16-10 ), whereas large varices are greater than 5 mm in diameter. Large varices correspond to grade 2 and grade 3 varices in previous classifications. The size of the varices should be described in the lower third of the esophagus with the esophagus completely insufflated with air and examined on withdrawal of the instrument. Additional descriptors of the varices include red signs. A cherry red spot on a varix is 3 mm in diameter or less, and a hematocystic spot or blood blister on the varix is 4 mm in diameter or greater. The red wale sign describes a whiplike longitudinal mark on the varix ( Fig. 16-11 ). The red wale sign is indicative of a weakness in the varix wall and is a marker of increased risk of bleeding, although not as predictive of bleeding as is the size of the varix.

Gastroesophageal varices are classified as follows: Type 1 gastroesophageal varices are in continuity with esophageal varices and extend for a variable distance into the lesser curvature of the stomach; type 2 gastroesophageal varices are again in continuity with esophageal varices but extend into the cardia of the stomach ( Fig. 16-12 ). Isolated gastric varices can occur either in the fundus of the stomach (IGV type 1) or in the antrum of the stomach (IGV type 2). IGV type 2 varices are distinctly uncommon. Type 2 gastroesophageal varices are the most common site of bleeding from gastric varices, although the most severe bleeding is with IGV type 1 varices.

CT Scan

A computerized tomographic scan using multidetector arrays is another investigational modality to demonstrate esophageal varices ( Fig. 16-13 ). The advantage of CT scans is that varices may be detected with accuracy close to that of endoscopy. In addition, the portal venous anatomy, liver masses, and extrahepatic pathology may be visible. Patients greatly prefer CT scans over upper endoscopy. The risks of radiation with CT scans are minimal, but CT scans to screen for esophageal varices are probably best avoided in patients younger than 35 years in whom a minute risk of radiation exists. CT scan may be used to screen for varices in patients who do not wish to undergo endoscopy, especially if they have had a history of dysphagia.

Magnetic Resonance Imaging

Magnetic resonance imaging is another investigational modality for detection of varices. Magnetic resonance imaging has the same advantages over endoscopy as the CT scan without the risk of radiation. There have been limited studies that have compared upper endoscopy with magnetic resonance imaging in the detection of varices.

Magnetic resonance elastography (MRE) may determine the degree of hepatic fibrosis and possibly the degree of portal hypertension. The advantage of MRE over ultrasound transient elastography is that the stiffness across the whole liver can be determined; with transient elastography using ultrasound, stiffness is calculated across only a cylinder of liver approximately 1 cm in diameter and 2 to 4 cm in length. Spleen stiffness on MRE increases in patients with portal hypertension ( Fig. 16-14 ).

Endoscopic Ultrasound

Endoscopic ultrasonography uses radial or linear array echoendoscopes or ultrasound miniprobes passed through the biopsy channel of a conventional endoscope; it is still considered investigational in the evaluation of patients with portal hypertension. Endoscopic ultrasound allows measurement of the cross-sectional area of varices ( Fig. 16-15 ); the amount of blood flow in the varices and in the left gastric vein, azygos vein, portal vein, and splenic vein; and decrease in the size of varices following variceal ligation. Endosonography combined with variceal pressure measurement can potentially allow for measurement of variceal wall tension because both the radius of the varix and the transmural variceal pressure are measurable. Measurement of variceal wall tension is important because it is the major determinant of variceal bleeding. Despite all of its potential uses, perhaps the most practical use of endoscopic ultrasound is in determining whether a submucosal mass in the fundus of the stomach in a patient with cirrhosis is a tumor or a bunch of gastric varices.

Measurement of Portal Venous Pressure

Hepatic Vein Pressure Gradient Measurement

Catheterization of the hepatic vein is the usual technique used to measure HVPG. An end-hole catheter, or more commonly a balloon catheter, is used to determine pressure via either the femoral or the transjugular route. In the presence of tense ascites, catheterization of the hepatic veins may be more difficult using the femoral approach; it may be necessary to use a deflector to enter the hepatic vein. Not only is catheterization of the hepatic vein using the transjugular route somewhat easier, it also allows right-sided cardiac pressure to be measured. Contrast is injected into the hepatic vein to confirm the balloon is in a wedged position. Total occlusion of the hepatic vein by the inflated balloon results in a sinusoidal pattern being demonstrated on contrast injection without a collateral circulation to other hepatic veins. Once the balloon is deflated, the contrast washes out promptly. Additionally, there is a sharp increase in the pressure recorded on inflation of the balloon and a sharp drop when the balloon is deflated. When the balloon is correctly positioned and inflated, the pressure recorded is steady, with only respiratory variation. The HVPG should be measured at least three times to demonstrate reproducibility. Procedure-related complications with HVPG measurement are uncommon.

Wedging of the catheter in the hepatic vein or inflation of the balloon to occlude the hepatic vein creates a stagnant column of blood, which represents hepatic sinusoidal pressure in patients with cirrhosis. In the normal situation, this pressure is rapidly dissipated via the other sinusoids. In sinusoidal causes of portal hypertension, as in alcoholic cirrhosis and HCV-related cirrhosis, the pressure is not dissipated and there is a continuous column of blood extending from the hepatic vein to the portal vein ( Fig. 16-16 ). Thus wedged hepatic vein pressure (WHVP) represents portal pressure. Inferior vena caval pressure or the free hepatic vein pressure (FHVP) is measured at the junction of the hepatic vein and IVC, or with the balloon deflated in the hepatic vein. The FHVP is used as the reference standard and is recorded with the “zero” measured in the midaxillary line. The right atrial pressure should not be used as the reference standard.

The major drawback of using an end-hole catheter in the wedged position for measurement of the WHVP is that the pressure over only a small area of the liver is measured. Because there is regional variation in the degree of fibrosis, pressures may be higher in some veins than in the others. On the other hand, if a balloon-occluding catheter is used in the main right hepatic vein, the pressures are averaged over a wide area of the liver. Thus the WHVP using a balloon catheter is more accurate. It is important to note in portal hypertension secondary to portal vein thrombosis that the hepatic sinusoidal pressure is normal, and therefore the HVPG is normal. In primary biliary cholangitis, where presinusoidal resistance is also increased, the HVPG is somewhat lower than the true hepatic sinusoidal pressure.

At the current time, HVPG measurement is not routinely used to diagnose portal hypertension or to monitor therapy using pharmacologic agents. However, HVPG measurement may be used to determine whether the cause of portal hypertension is sinusoidal, presinusoidal, or postsinusoidal. HVPG measurement is particularly useful in combination with hepatic venography, transjugular liver biopsy, and right-sided heart pressure measurements to determine whether the cause of ascites is cardiac or hepatic in origin. HVPG has been used in patients with cirrhosis to assess the risk of hepatic resection, as an end point in trials using pharmacologic agents, and as a prognostic marker.

Measurement of Variceal Pressure and Flow

Measurement of variceal pressure may be particularly important in patients with large varices and portal vein thrombosis in whom the HVPG does not accurately measure portal pressure. Variceal pressure can be measured during upper endoscopy by inserting into the varix a needle connected to a fluid-filled catheter connected to a pressure transducer. Measurement of pressure is followed by sclerotherapy of the varix because of the small risk of bleeding associated with variceal puncture. Because injection sclerotherapy is seldom used nowadays, measurement of esophageal variceal pressure by needle puncture is not recommended.

Esophageal variceal pressure can also be measured using miniature pneumatic pressure-sensitive gauges. These methods require not only considerable procedural skill but also stable intraesophageal pressure. The principle of the pneumatic pressure gauge is that varices are thin walled and quite elastic. Because they are elastic structures, the pressure required to compress and collapse the varix equals the venous pressure within the varix. Patients with previous variceal bleeding have higher variceal pressures than patients who have never bled. A variceal pressure greater than 18 mm Hg in patients who have had a variceal bleed is associated with failure to control the bleed, as well as a predictor of early and subsequent rebleeding.

Measurement of Portal Venous and Hepatic Artery Blood Flow

Clearance studies measure only total hepatic blood flow; the relative contribution of the portal vein and hepatic artery to total blood flow cannot be determined accurately. The most commonly used agent to measure total hepatic blood flow is indocyanine green (ICG) because it is cleared by the liver and is nontoxic. Following injection or infusion of ICG, the disappearance of the compound in both the peripheral vein and hepatic vein is measured. Using the Fick principle, hepatic blood flow is calculated.

Determining the relative contributions of the hepatic artery and portal vein to hepatic blood flow is theoretically useful in selecting the patient who would benefit the most from a portosystemic shunt. It has been suggested that patients with normal portal venous blood flow might be poor candidates for portosystemic shunts because of rapid diversion of portal venous blood from the liver following creation of a shunt. However, the response of hepatic arterial blood flow to portal diversion may be more important. A poor outcome is likely following a portosystemic shunt if there is only a minimal increase in hepatic arterial blood flow in response to portal venous diversion. Additionally, decreased hepatic blood flow after liver transplantation may point to graft dysfunction.

Measurement of portal venous blood flow alone and hepatic artery blood flow is possible during surgery using flow meters. Measuring portal venous flow without surgery is cumbersome and requires catheterization of the superior mesenteric artery, hepatic veins, and umbilical vein. Following injection into the umbilical vein of 99-mTc pertechnetate, rapid images are collected from the heart, kidney, lungs, spleen, and liver, and the portal venous fraction calculated. In patients with cirrhosis, the portal venous fraction of total hepatic blood flow varies from the normal 66% to essentially no flow.

Because measurements of total hepatic, hepatic artery, and portal vein blood flow are cumbersome and invasive, they are seldom used in clinical practice. Doppler ultrasound has been used as an alternative modality. The velocity of flow of blood in the portal vein can be measured noninvasively using Doppler ultrasound. The velocity of flow multiplied by the cross-sectional area of the portal vein is then used to calculate the volume of blood flow in the portal vein. These results compare favorably with portal vein flow measured using flow meters. Technical problems associated with Doppler measurements, as well as body habitus, make it difficult to accurately estimate portal venous blood flow in some patients. Doppler is also used to measure hepatic arterial blood flow and resistance, and is a means of determining hepatic artery stenosis or thrombosis, especially following liver transplantation.

Portal Hypertension Secondary to Increased Portal Venous Blood Flow

Splanchnic Arteriovenous Fistula

The most common cause of increased portal venous blood flow is an arteriovenous fistula that may be intrahepatic or extrahepatic. A splenic artery–splenic vein fistula is an example of an extrahepatic splanchnic arteriovenous fistula ( Fig. 16-17 ). Fistulae may occur within the liver in hereditary hemorrhagic telangiectasia (HHT, Osler-Rendu-Weber syndrome), or may follow trauma such as a liver biopsy, or can result from a rupture of a hepatic artery aneurysm. Unlike patients with HHT with a hepatic artery–hepatic vein fistula who develop cardiac failure, patients with hepatic artery–portal vein fistula develop portal hypertension. Because the portal venous system is exposed to arterial blood pressure in the presence of a hepatic artery–portal vein fistula, the rate of delivery of blood to the liver exceeds the outflow, resulting in a rise in portal pressure. With time, the high flow results in sinusoidal fibrosis and an increase in intrahepatic resistance, which propagates the portal hypertension, a phenomenon confirmed in animal studies using hepatic artery–portal vein anastomosis. It is important to recognize the contribution of intrahepatic resistance because ligation of the fistula may not always result in resolution of the portal hypertension. Moreover, creation of a portosystemic shunt without occlusion of the fistula in an effort to ameliorate portal hypertension will actually result in the creation of a systemic arteriovenous fistula, which could lead to high-output cardiac failure.