Esophageal Varices Video (52.2 mb)

Portal hypertension is defined as increased pressure in the portal vein. With the right atrial pressure as a zero reference, normal portal venous pressure is approximately 4-8 mm Hg. The portal vein is formed by the confluence of the splenic and superior mesenteric veins. Its flow rate normally averages about 1-1.2 L/min. The simple phenomenon of increased pressure in this venous circulation unleashes a wide array of hemodynamic and metabolic consequences, including some of the most lethal and distressing complications of chronic liver disease.

The causes of portal hypertension are diverse (Table 16). Since portal pressure is the product of portal blood flow and intrahepatic resistance, any condition causing an increase in flow or resistance will increase portal pressure. An example of a "pure" flow increase is postsurgical or traumatic splenic arteriovenous fistula. The marked increase in splenic and thus portal venous flow leads to the development of portal hypertension. Almost all other causes of portal hypertension are mediated predominantly by increasing resistance, although evidence indicates that most high-resistance syndromes are also accompanied by increases in portal venous flow. In many conditions, the cause of the increased resistance is evident: inflammation and fibrosis lead to vascular distortion, architectural disturbance and impingement of the intravascular spaces. Other less evident factors are predominant in other conditions. For example, in acute alcoholic hepatitis, hepatocyte cell swelling and collagen deposition in the space of Disse lead to narrowing and distortion of sinusoidal spaces. The reasons for the increased mesenteric (and thus portal venous) blood flow in high-resistance states remain unclear. One theory postulates that a circulating vasodilatory humoral factor that would normally be inactivated by the liver escapes into the systemic circulation through shunts or hepatocellular insufficiency.

TABLE 16.   Causes of portal hypertension

There are two separate and sometimes overlapping classification systems for the causes of portal hypertension, using either the liver or the hepatic sinusoid as the reference point. The former classifies conditions into pre-hepatic, intrahepatic and posthepatic causes, while the latter divides conditions into presinusoidal, sinusoidal and postsinusoidal causes (Table 16). However, the exact site of increased resistance in many intrahepatic causes of portal hypertension has recently been questioned, and it is likely that the predominant resistance sites could change according to the stage of some disease processes. For example, early primary biliary cirrhosis is thought to produce mainly presinusoidal hypertension, but as dense cirrhosis supervenes, sinusoidal hypertension becomes more important. Similarly, an early lesion of alcoholic liver disease, the central or terminal hyaline sclerosis, characterized by zone 3 fibrosis, would cause postsinuoidal hypertension, with sinusoidal hypertension predominating as cirrhosis becomes established. In practical terms, there are reasons for trying to correctly classify resistance sites. One is for predicting responses to surgical shunting procedures: presinusoidal conditions generally have well-preserved hepatocellular function and thus respond well to diversion of portal blood, whereas sinusoidal and postsinusoidal conditions tend to be associated with varying degrees of hepatic insufficiency. Another is that ascites generally occurs only with sinusoidal and postsinusoidal hypertension.

Portal hypertension leads to many clinical complications. Ascites is directly related to the development of sinusoidal or postsinusoidal hypertension. Portal-systemic collateral vessels form in an attempt to decompress the portal hypertension (Table 17). The most troublesome site of collateral formation is around the proximal stomach and distal esophagus (gastroesophageal varices). Bleeding from such varices (or the gastric mucosa) and hepatocellular failure are the two commonest causes of death in cirrhosis. Indeed the mortality rates for variceal bleeding range from 15-50% depending on the degree of hepatic function: Child-Pugh class A, B and C patients have, respectively, 15%, 20-30% and 40-50% mortality rates when their varices bleed.

The risk of bleeding from gastroesophageal varices is related to several factors. First, a threshold minimum level of portal pressure of approximately 12 mm Hg appears necessary for varices to form. However, above this level it is unclear whether absolute height of portal pressure affects the bleeding risk. Factors such as intrathoracic pressure gradients induced by coughing, straining or sneezing, and damage to the variceal wall by acid reflux into the esophagus appear not to play a role. The two factors most important in determining bleeding risk are variceal size and local variceal wall characteristics. Several studies have shown that small varices almost never bleed, while the bleeding risk of medium-sized varices is approximately 10-15% over two years, and that of large varices, approximately 20-30% over the same period. During the past decade, it has become clear that certain varix wall characteristics that are visible through an endoscope are also predictive of high bleeding risk. These are the red and blue color signs. Small localized wall defects such as thin-walled blebs or sacs in the wall look like red spots or streaks and have variously been termed "red wale markings," "cherry-red spots" or "red streaks," while a diffuse pronounced blue color indicates a large varix (vein) with a stretched mucosa covering it.

Approximately 30-50% of upper GI bleeding episodes in patients with portal hypertension originate from nonvariceal sources. Cirrhotic patients have an increased incidence of acid-peptic disease, mostly erosive gastritis. This is probably due to the alcohol abuse that is common in this population. However, it has recently become clear that the majority of nonvariceal upper GI bleeding in cirrhosis is due to a peculiar form of gastropathy seen in the stomach in portal hypertension. Several features distinguish this portal hypertensive gastropathy from the erosive or inflammatory gastritis seen in nonhypertensive patients (Table 18). The major symptom of portal hypertensive gastropathy is bleeding. Pain or dyspepsia is uncommon as a presenting feature of this type of gastropathy. The appropriate treatment for this condition is still unclear, but it probably responds to measures to decrease portal pressure, although a possible role for cytoprotective agents has also been suggested.

Diagnosing portal hypertension is usually easy. The patient often has concomitant ascites and splenomegaly, along with the stigmata of chronic liver disease. However, it should be remembered that all the prehepatic and many of the presinusoidal conditions have well-preserved liver function and no ascites. Abdominal wall collaterals radiate outward from the umbilicus; when they are very prominent, it is easy to see why this condition is termed "caput medusae," after the fearsome creature in Greek mythology with the serpentine hairdo. Dilated abdominal wall veins, especially in the upper abdomen, are common, but caput medusae is rare. Another diagnostic clue may be the presence of anorectal varices masquerading as hemorrhoids. Gastroesophageal variceal bleeding produces large-volume, brisk bleeding with hematemesis and, later, melena or hematochezia. Portal hypertensive gastropathy may also produce brisk bleeding, but can occasionally cause low-volume oozing manifested only by melena.

Managing the acute bleeding episode consists of the general resuscitative measures such as volume and blood replacement, and specific measures to stop the bleeding. Various pharmacological, mechanical and surgical modes of arresting hemorrhage are used, usually in that order. Vasoconstrictive drugs to stop bleeding include vasopressin and somatostatin or their longer-acting analogues such as glypressin and octreotide, respectively. Vasopressin infusions induce generalized arteriolar and venous constriction, with resultant decreased portal venous flow and thus pressure, and at least temporary cessation of bleeding in 50-80% of cases. However, the generalized vasoconstriction also may result in peripheral vascular ischemia, myocardial ischemia or infarction and renal tubular damage. Concurrent nitrate administration has been suggested to attenuate some of these side effects, but whether it actually does so is still unproven. A safer alternative may be somatostatin or octreotide. Their mechanism of action is still unclear but probably relates to a suppressive effect on the release of vasodilatory hormones such as glucagon, leading to a net vasoconstrictive effect. Side effects are minimal. Whatever drug is used, it is generally inadvisable to continue drug therapy for more than one to two days.

Mechanical modes of therapy include inflatable balloons for direct tamponade. The Sengstaken-Blakemore tube has both an esophageal and a small gastric balloon; the Linton-Nachlas tube, with only a large gastric balloon, is attached to a small weight to stanch the cephalad flow of blood in the varices. Both tubes carry significant complication rates (15%), especially in inexperienced hands. The most common complications of esophageal balloon therapy for varices include aspiration, esophageal perforation and ischemic (pressure) necrosis of the mucosa.

The most common and probably the most effective nonsurgical therapies are endoscopic variceal sclerotherapy and ligation. Highly irritant solutions such as ethanolamine, polidocanol or even absolute ethanol are injected through endoscopic direct vision into and around the bleeding varix. The subsequent inflammation leads to eventual thrombosis and fibrosis of the varix lumen. Possible complications include chest pain, dysphagia, and esophageal ulceration and stricturing. The injection of irritant solutions that eventually lodge in the pulmonary circulation can result in lung function abnormalities, although these tend to be subclinical. A newer and probably safer method of endoscopic therapy is ligation or banding, similar to the rubber band ligations used to fibrose anorectal hemorrhoids. Initial studies suggest that its efficacy is similar to sclerotherapy, with fewer esophageal complications. The combination of endoscopic therapy and either balloon tamponade or drug therapy to control actively bleeding varices is successful in 80-95% of cases.

When all the above measures fail, emergency surgery may be tried. Emergency portacaval shunt surgery has been abandoned because of a 30-50% operative mortality rate. The simplest and probably best choice in the emergency situation is esophageal transection, in which a mechanical device transects and removes a ring of esophageal tissue, and then staples the ends together. Another type of "surgery" is the transjugular intrahepatic portal-systemic shunt (TIPS). In this procedure, an intrahepatic shunt between branches of the hepatic and portal veins is made by balloon dilation of liver tissue, and then an expandable metal stent of approximately 1 cm diameter is lodged into the fistula. The procedure can be done by a radiologist using fluoroscopy- guided catheterization, and requires only light sedation and local anesthesia.

Once the acute bleeding episode has been treated, how do we reduce the risk of future rebleeding? Before considering any other therapy, some obvious common-sense measures should be taken. For example, patients with cirrhosis caused by alcohol (the cause of approximately 50-60% of cirrhosis in Canada) absolutely must stop drinking; the rebleeding and mortality rates in patients who continue their alcohol use are much higher than in those who remain abstinent.

Prophylactic therapy to prevent bleeding may be divided into primary (to prevent the first bleed in a patient with varices who has never bled) and secondary prophylaxis (to prevent rebleeds). There is still much conflicting literature on these two topics, but for now, the following preliminary recommendations can be made. First, patients with large varices that have never bled should be started on beta blocker therapy at doses sufficient to reduce the resting heart rate by 20-25%. Beta-adrenergic antagonists are thought to produce arteriolar and venous constriction and significantly reduce blood flow through portal-systemic collaterals while modestly reducing portal pressure. Endoscopic sclerotherapy/banding, TIPS and surgery are ineffective and too risky for primary prophylaxis.

The appropriate secondary prophylaxis regimes remain controversial. There is probably a minority subgroup who respond favorably to beta blocker therapy, but they cannot be easily identified. One approach is to perform enough endoscopic sclerotherapy/banding sessions (usually 3-6) to obliterate varices or reduce them to small size. Treatment failures on this regime (e.g., those with recurrent bleeding) could be considered either for TIPS or surgery. TIPS should not be done in patients with a history of, or active, encephalopathy.

Prehepatic causes of portal hypertension such as portal vein thrombosis generally respond well to some type of portal-mesenteric diversion procedure such as mesocaval or portacaval shunting. In these cases, normal liver function protects against the development of encephalopathy or hepatic insufficiency when portal blood is diverted away from the liver.

Of course the definitive treatment for most of the complications of end- stage liver disease, including recurrent GI bleeding due to severe portal hypertension, is orthotopic liver transplantation. Since the presence of a surgical portacaval or mesocaval shunt greatly complicates the transplantation procedure, we have generally abandoned these types of shunting operations in patients with cirrhosis.