Portal Hypertension and Cirrhosis Complications: Portal Hypertension: Varices and Bleeding

0:00Welcome to Board Pearls. This is episode one of three of the Portal Hypertension and Cirrhosis Complications chapter, in the Liver Disease module. This episode is portal hypertension and the bleeding it causes. A man with compensated cirrhosis who asks why he needs a pill shows how portal hypertension develops, how it's measured and how it's estimated without a catheter, and why a beta-blocker starts before any varix appears. A young man with varices and a nearly normal liver shows why the level of the obstruction matters, and a man who fails several organs at once shows acute-on-chronic liver failure. A woman who's never been scoped shows who can skip endoscopy, how often to look, and what to do with each size of varix, and a man with refractory ascites shows when to hold the beta-blocker. A man vomiting blood shows the bleeding bundle, the preemptive transjugular intrahepatic portosystemic shunt, TIPS, and secondary prophylaxis. And a woman with a bleeding fundal varix shows why gastric varices need different tools, followed by splenic vein thrombosis, portal hypertensive gastropathy, and gastric antral vascular ectasia, GAVE.

1:06Here's the first patient. A fifty-eight-year-old man has compensated cirrhosis. His hepatitis C was cured three years ago, and he also has fatty liver disease. He's never had ascites, bleeding, or encephalopathy. His FibroScan stiffness is twenty-eight kilopascals, and his platelets are ninety-five thousand. He asks why he'd need a pill when he feels fine and has never bled.

1:30His stiffness and platelets point to clinically significant portal hypertension, and that, not how he feels, is why treatment starts now. Portal hypertension is what turns compensated cirrhosis into a dangerous disease. About half of patients with cirrhosis develop varices, and about a third of those bleed, and ascites, spontaneous bacterial peritonitis, hepatic encephalopathy, hepatorenal syndrome, and the pulmonary vascular complications almost all flow from portal hypertension and the splanchnic vasodilation it produces. Mortality at six weeks after a variceal bleed is about fifteen to twenty percent, spontaneous bacterial peritonitis carries a hospital mortality of twenty to forty percent, and hepatorenal syndrome with acute kidney injury carries fifty to eighty percent without transplant. Preventing his first decompensation is worth a great deal.

2:20His portal hypertension developed in two steps. It began as increased resistance inside the liver, from the distorted sinusoidal architecture of cirrhosis of any cause. Then the splanchnic arteries dilated, raising portal inflow and pressure further, and the rest of the circulation was left underfilled. That underfilling activates the renin-angiotensin-aldosterone system and antidiuretic hormone, which retain sodium and water, and the result is a hyperdynamic circulation, the substrate for ascites, hepatorenal syndrome, and the lung complications. So treatment targets two problems, the resistance inside the liver and the inflow into the portal system.

2:59Portal pressure can be measured directly. The gold standard is the hepatic venous pressure gradient, the wedged minus the free hepatic venous pressure, measured at transjugular catheterization, and chapter seventeen covers the technique. A gradient over five millimeters of mercury defines portal hypertension. Ten or more defines clinically significant portal hypertension, CSPH, which identifies the patients who will develop varices and decompensate. And over twelve signals a risk of variceal bleeding. A response to a beta-blocker is a fall of more than twenty percent from baseline or to below twelve, and responders bleed and decompensate substantially less.

3:38He doesn't need a catheter, because the Baveno seven criteria estimate CSPH from stiffness and platelets. For compensated advanced chronic liver disease, a stiffness under ten kilopascals rules it out and ten to fifteen suggests it. For CSPH, a stiffness of twenty-five or more rules it in, and fifteen or less with platelets of one hundred fifty thousand or more rules it out. Between those, CSPH is probable with a stiffness of twenty to twenty-five and platelets under one hundred fifty thousand, or fifteen to twenty with platelets under one hundred ten thousand, and other combinations need more information. His stiffness of twenty-eight rules in CSPH on its own.

4:17That answers his question about the pill. A randomized trial took patients with compensated cirrhosis, CSPH, and no high-risk varices and stratified them by their acute gradient response to intravenous propranolol. Responders got propranolol or placebo, non-responders got carvedilol or placebo, and the beta-blocker arm significantly reduced first decompensation, especially ascites. That trial moved Baveno seven from starting a beta-blocker when varices are found to starting one for CSPH. The indication is preventing decompensation, which is separate from preventing varices: beta-blockers don't stop varices from forming in patients without them.

4:59Carvedilol is the beta-blocker of choice. Nonselective beta-blockers lower portal pressure through beta-one blockade, which lowers cardiac output, and beta-two blockade, which constricts the splanchnic vessels and cuts portal inflow. Carvedilol adds alpha-one blockade, which reduces resistance inside the liver, so it lowers the gradient more and is preferred. He starts carvedilol, six point two five milligrams daily, increasing to twice daily as tolerated. He feels fine because he's compensated, and carvedilol is meant to keep him that way.

5:32The second patient shows that portal hypertension doesn't always mean cirrhosis. A thirty-two-year-old man who grew up in Egypt has varices and an enlarged spleen, with normal liver function and only a mildly raised stiffness. The level of the obstruction varies, and portal hypertension can arise before or after the sinusoids without any cirrhotic architecture. Presinusoidal causes include schistosomiasis, the most common worldwide, in which eggs lodge in the portal venules and cause periportal fibrosis; porto-sinusoidal vascular disease, a group of obliterative portal venopathies; and portal vein thrombosis without cirrhosis. He most likely has schistosomiasis, and calling him cirrhotic would mislead his care.

6:15Postsinusoidal causes sit on the other side of the liver. They include Budd-Chiari syndrome, right heart failure, which transmits a high central venous pressure backward, constrictive pericarditis, and sinusoidal obstruction syndrome. Each level points to its own treatment: anticoagulation and angioplasty for Budd-Chiari syndrome, defibrotide for sinusoidal obstruction syndrome, treatment of the inflammation or thrombotic driver behind a portal vein thrombosis, and the usual cirrhosis care for sinusoidal disease. Episode three covers the vascular diseases.

6:47The third patient shows decompensation arriving all at once. A sixty-four-year-old man with alcohol-related cirrhosis develops pneumonia, then kidney failure and confusion within days. He has acute-on-chronic liver failure, ACLF, which is acute decompensation, from new ascites, encephalopathy, gastrointestinal, GI, bleeding, or bacterial infection, with organ failure on top of chronic liver disease. Sepsis is the most common trigger in the United States, and severe alcohol-associated hepatitis is another. Prognosis tracks the number of failed systems, among the liver, kidney, brain, coagulation, circulation, and lungs, and three or more, grade three ACLF, carries a short-term mortality above seventy-five percent. He has two. Unlike acute liver failure, ACLF doesn't cause significant cerebral edema.

7:42Once portal hypertension is established, the next question is whether varices are there, and the fourth patient shows who has to be scoped. A sixty-one-year-old woman has compensated alcohol-related cirrhosis and has been abstinent for two years. Her stiffness is twenty-two kilopascals and her platelets are one hundred twenty thousand. She's never been scoped. A friend with cirrhosis was told she didn't need an endoscopy.

8:05Her friend met the exception and she doesn't. Under the Baveno exception, a patient with a stiffness under twenty kilopascals and platelets over one hundred fifty thousand can defer endoscopy, with yearly elastography and platelet counts instead, because the yield of clinically significant varices is too low to change management. Both numbers are required, so a stiffness of fourteen with platelets of one hundred thirty thousand still means a scope. She misses both and needs endoscopy.

8:32The size of what she finds decides what happens next. Varices are present in about half of patients with cirrhosis, more often with a higher Child-Pugh class, and they grow about seven to eight percent a year. Large varices carry about a thirty percent chance of a first bleed within two years, about four times the risk with small ones, which is why prophylaxis starts at medium or large size. Her endoscopy shows large varices without red wale signs.

8:59Her large varices need primary prophylaxis, with either a nonselective beta-blocker or band ligation every two to four weeks until they're obliterated. Carvedilol is preferred under Baveno seven, because its alpha-one blockade relaxes the stellate cells through nitric oxide and lowers resistance inside the liver. It starts at six point two five milligrams daily and is titrated as tolerated to six point two five twice daily, a fixed target watched by tolerance and blood pressure. Propranolol or nadolol, if they're used, are titrated to a resting heart rate of fifty-five to sixty or a twenty-five percent fall, keeping the systolic pressure over ninety. A beta-blocker treats the pressure behind every varix, including those too small to band, while banding needs repeated sessions. With a normal blood pressure, sodium, and kidney function, she starts carvedilol, with banding in reserve. Once she's on a beta-blocker, she needs no surveillance endoscopy, because the result wouldn't change her treatment.

9:57Smaller findings follow different rules. Small varices get a beta-blocker when they have red wale signs, which mean a thin wall, because by Laplace's law wall tension equals pressure times radius divided by wall thickness. They also get one in Child-Pugh C cirrhosis, where portal pressure is high whatever the size. Otherwise they're rechecked in one to two years. After a negative endoscopy, the repeat is in two years with ongoing injury, such as active drinking, obesity, or untreated viral hepatitis, in three years when the injury is quiescent, after abstinence or a cured hepatitis C, and yearly in decompensated cirrhosis. A negative scope in her, abstinent for two years, would have meant a repeat in three.

10:42The fifth patient shows when the beta-blocker has to be held. A sixty-seven-year-old man with large varices on carvedilol has developed refractory ascites, with a systolic pressure of eighty-six and a sodium of one hundred twenty-eight. Beta-blockers aren't absolutely contraindicated in refractory ascites or after spontaneous bacterial peritonitis, but high doses are avoided, because lowering blood pressure in a patient who's already vasodilated and underfilled can worsen kidney perfusion and raise acute kidney injury and death. Baveno seven holds the beta-blocker for a systolic pressure under ninety, a sodium under one hundred thirty, or acute kidney injury, and resumes it after improvement. He meets two of those, so his carvedilol is held. His large varices still need prophylaxis, so he switches to banding every two to four weeks until they're obliterated.

11:35The sixth patient shows what happens when a varix bleeds. A forty-nine-year-old man with alcohol-related cirrhosis vomits a large amount of blood. His heart rate is one hundred twenty and his blood pressure ninety over fifty-five. His hemoglobin is six point eight. He has ascites and mild confusion. His bilirubin is three point two, his albumin two point six, his international normalized ratio, INR, one point nine, and his Child-Pugh score twelve, class C. The resident wants to transfuse him to a hemoglobin of ten.

12:10He has a presumed acute variceal bleed in Child-Pugh C cirrhosis, and the resident's target is wrong, because resuscitation in cirrhosis follows its own rules. Any GI bleeding in cirrhosis is variceal until proven otherwise, and chapter nine builds the full bundle. Transfusion targets a hemoglobin of about seven, because overtransfusion expands volume, raises portal pressure, and worsens rebleeding, so he gets two units to seven point four and stops there. The INR isn't corrected with plasma, because clotting in cirrhosis is rebalanced and the extra volume would raise portal pressure.

12:45Two drugs start before the scope. Ceftriaxone, one gram intravenously daily for up to seven days, is mandatory for any GI bleeding in cirrhosis, because hemorrhage disrupts the gut barrier and bacteria translocate into the ascites and blood. A meta-analysis showed that prophylaxis markedly reduced infection and mortality, and ceftriaxone is preferred because fluoroquinolone resistance is common in cirrhosis. Octreotide, a fifty-microgram bolus and then fifty micrograms an hour for two to five days, starts before endoscopy, because it lowers portal inflow and does no harm if the source turns out not to be variceal. And because he's confused and vomiting blood, he's intubated electively before the scope to protect his airway.

13:30He's banded within twelve hours for a spurting esophageal varix, and the bleeding stops. Banding achieves initial hemostasis in eighty to ninety percent. Early rebleeding after banding runs about thirty percent in the first seventy-two hours, so his octreotide and ceftriaxone continue. If banding had failed, balloon tamponade could have bridged him for no more than twenty-four hours, with his airway protected, or a covered esophageal stent for up to seven days, and salvage TIPS would have been indicated whatever his Child-Pugh class.

14:01His Child-Pugh C score sets up the most important cirrhosis-specific decision of the admission, preemptive TIPS. In a randomized trial in Child-Pugh C with scores of ten to thirteen, and in Child-Pugh B with active bleeding at the first endoscopy, a TIPS within seventy-two hours improved one-year survival substantially over standard therapy. It converts a failing, pressure-driven circulation into a controlled shunt before rebleeding drives multiorgan failure. Patients are selected during the first admission, not after a second bleed, because by then the benefit narrows, and a covered polytetrafluoroethylene-lined stent is the default because it stays open better. His score of twelve qualifies him. A Child-Pugh B patient qualifies only with active bleeding at endoscopy.

14:47The contraindications come from what the shunt does. Severe right heart failure or pulmonary hypertension rules it out, because the shunt raises right heart preload. Advanced cirrhosis with prior overt encephalopathy rules it out, because portal blood bypasses the liver's ammonia clearance. Active sepsis and biliary obstruction rule it out too. His echocardiogram is normal, he has no sepsis or obstruction, and his confusion belongs to this bleed rather than a prior episode, so he gets a covered TIPS on day two. Severe pulmonary hypertension would have meant secondary prophylaxis instead, accepting the higher risk of rebleeding.

15:25The seventh patient doesn't need TIPS but still needs protection from the next bleed. A fifty-six-year-old woman with Child-Pugh A cirrhosis bled from esophageal varices, was banded within twelve hours, and stopped bleeding. She's outside both high-risk groups, so she starts secondary prophylaxis before discharge: a nonselective beta-blocker, carvedilol, propranolol, or nadolol, plus banding until the varices are obliterated. The combination outperforms either alone, because the drug lowers the pressure behind every varix while banding removes the column that bled.

15:59Her banding schedule is tighter than in primary prophylaxis. Sessions are every seven to fourteen days, the time a post-banding ulcer takes to heal, rather than every two to four weeks. A proton pump inhibitor, PPI, for two weeks after each session reduces bleeding from the banding ulcer, which is sensitive to acid. After the varices are eradicated, surveillance endoscopy is repeated at three to six months and then every six to twelve months.

16:27Varices in the stomach behave differently, and the eighth patient shows why. A fifty-five-year-old woman with cirrhosis from primary biliary cholangitis has melena and a hemoglobin of seven. Endoscopy shows a large isolated fundal varix with a red spot and no esophageal varices. Computed tomography, CT, shows a large gastrorenal shunt. She has no ascites. The endoscopist reaches for the band ligator.

16:55She has a bleeding gastric varix, and the band ligator is the wrong tool, for reasons that start with its type. In the Sarin classification, gastroesophageal varix type one, GOV1, runs along the lesser curve as an extension of esophageal varices. It makes up about seventy-five percent of gastric varices, behaves like an esophageal varix, and is often banded. GOV2 extends from the esophagus into the fundus. Isolated gastric varix type one, IGV1, is a fundal varix, often with a gastrorenal shunt, and IGV2 is an isolated varix elsewhere in the stomach. The Saad-Caldwell classification adds whether a gastrorenal shunt is present, which directs the choice between TIPS and balloon-occluded retrograde transvenous obliteration, BRTO, so any fundal varix needs cross-sectional imaging as well as endoscopy. She has IGV1 with a gastrorenal shunt.

17:50Banding fails at her varix. The gastric submucosa is thicker, the varices are larger, and their position prevents a band from reliably capturing the whole varix. Cyanoacrylate glue injection, off label in the United States but widely used, polymerizes on contact with blood and forms a cast that the body extrudes over weeks, and it's used for IGV1 and GOV2. Coil embolization guided by endoscopic ultrasound is available at expert centers. Because endoscopic therapy is unreliable for fundal varices, the threshold for TIPS or BRTO is low. The Linton-Nachlas tube, with its single large gastric balloon, bridges uncontrolled gastric variceal bleeding, with the airway protected.

18:33Her shunt decides between BRTO and TIPS. BRTO passes a catheter through the gastrorenal shunt from the systemic side, occludes it with a balloon, and injects sodium tetradecyl sulfate or ethanolamine oleate backward to obliterate the variceal complex. It closes the shunt without lowering portal pressure, while TIPS lowers pressure throughout the portal system. So BRTO is preferred with a gastrorenal shunt and no other portal hypertensive complications, and TIPS is preferred with refractory ascites or refractory esophageal varices, which it also treats. She has a shunt and nothing else to fix, so she has BRTO, and her bleeding stops. She stays in cirrhosis care with esophageal variceal screening, because closing one shunt doesn't treat the portal hypertension behind it.

19:19The ninth patient shows that fundal varices can appear with a normal liver. A forty-seven-year-old man with prior alcohol-related pancreatitis has bleeding isolated fundal varices, no esophageal varices, normal liver function, and an enlarged spleen. Isolated fundal varices without esophageal varices prompt evaluation for splenic vein thrombosis, called sinistral or left-sided portal hypertension. The splenic vein runs behind the pancreas, so about half of splenic vein thromboses come from pancreatitis and about a fifth from pancreatic cancer, and the fundal varices become the spleen's collateral drainage.

19:59Splenectomy or splenic artery embolization cures it when the thrombosis is the only driver, because it removes the high-flow inlet to the blocked segment. His CT shows an occluded splenic vein with a patent portal vein, and he has splenic artery embolization. A TIPS would have treated a portal pressure that isn't the problem. Chapters twenty-five and twenty-six cover pancreatic disease.

20:22Two other gastric lesions bleed slowly, and the distinction between them decides whether portal pressure treatment helps. Portal hypertensive gastropathy shows a mosaic pattern when mild and a mosaic with red spots when severe, mostly in the proximal stomach, and it causes chronic low-grade blood loss. Its severity parallels portal pressure. So first-line treatment is iron with a nonselective beta-blocker, argon plasma coagulation helps localized bleeding, octreotide is for the rare acute bleed, and TIPS is for refractory disease that depends on transfusion.

20:56The last patient has a lesion that looks similar but isn't. A sixty-eight-year-old woman with systemic sclerosis and chronic kidney disease has iron deficiency anemia, and endoscopy shows red stripes radiating across the antrum without a mosaic background. She has gastric antral vascular ectasia, called watermelon stomach when it's linear and diffuse when it's scattered. It has red spots without a mosaic, it's antral rather than proximal, and it can occur without portal hypertension, with associations including systemic sclerosis, autoimmune disease, and chronic kidney disease. It's treated with argon plasma coagulation, radiofrequency ablation, cryotherapy, or band ligation. TIPS and beta-blockers don't help, because the lesion doesn't depend on portal pressure.

21:43Here are the traps the boards set in these patients. Clinical significance starts at a gradient of ten, not five. A stiffness of twenty-five or more establishes CSPH without a catheter. A stiffness in the gray zone with low platelets doesn't rule CSPH out. A beta-blocker for CSPH doesn't wait for varices, and it prevents decompensation, not varices. Carvedilol is preferred over propranolol. Not every patient with varices has cirrhosis. And a patient failing several organs at once has ACLF.

22:14The screening traps come next. A patient who meets both exception criteria doesn't need a scope, and one criterion alone isn't enough. Small varices with red wale signs or Child-Pugh C disease need treatment. A patient on a beta-blocker doesn't need surveillance endoscopy. The beta-blocker is held for hypotension, hyponatremia, or kidney injury, and banding replaces it while it's held.

22:39The bleeding traps follow. Transfusing to ten worsens rebleeding, and correcting the INR with plasma adds nothing. Antibiotics are given to every patient with cirrhosis and a GI bleed, with or without fever. Octreotide starts before the scope, and a confused, vomiting patient has the airway protected first. Preemptive TIPS in Child-Pugh C doesn't wait for a second bleed, and Child-Pugh B with active bleeding qualifies too. TIPS isn't placed with severe right heart failure, pulmonary hypertension, or prior overt encephalopathy. And banding after a bleed is every one to two weeks, with a PPI after each session.

23:18The gastric traps close the list. Fundal varices aren't banded. BRTO fits a gastrorenal shunt with nothing else to fix, and TIPS fits when ascites or esophageal varices coexist. Isolated fundal varices with a normal liver mean splenic vein thrombosis. Proximal gastropathy with a mosaic isn't antral GAVE, and GAVE doesn't respond to TIPS or a beta-blocker.

23:42So, start to finish. Portal hypertension begins with resistance inside the liver, and splanchnic vasodilation then drives the hyperdynamic circulation behind most complications. A gradient over five is portal hypertension, ten or more is clinically significant, and over twelve allows bleeding. Estimate CSPH by Baveno seven, and start carvedilol for compensated cirrhosis with CSPH to prevent decompensation. Identify presinusoidal and postsinusoidal disease, and recognize ACLF by its organ failures.

24:16Skip screening endoscopy only with a stiffness under twenty and platelets over one hundred fifty thousand. Treat medium or large varices, or small ones with red wale signs or Child-Pugh C disease, with carvedilol or banding to obliteration. Hold the beta-blocker for a systolic under ninety, a sodium under one hundred thirty, or kidney injury, and switch to banding.

24:39In acute bleeding, transfuse to about seven, give ceftriaxone and octreotide, protect the airway, and band within twelve hours. Place a preemptive covered TIPS within seventy-two hours for Child-Pugh C ten to thirteen or Child-Pugh B with active bleeding, after checking the contraindications. Everyone else gets a beta-blocker plus banding every one to two weeks, with a PPI after each session.

25:04GOV1 is banded like an esophageal varix. Fundal varices need glue, coils, BRTO with a gastrorenal shunt and nothing else to fix, or TIPS when ascites or esophageal varices coexist. Isolated fundal varices with a normal liver mean splenic vein thrombosis, cured by splenectomy or embolization. Gastropathy tracks portal pressure and responds to beta-blockers and TIPS, while GAVE doesn't and needs ablation.

25:34Episode two is ascites and its complications: the paracentesis, diuretics, refractory ascites and TIPS, hyponatremia, spontaneous bacterial peritonitis, hepatorenal syndrome, and hepatic encephalopathy.

25:50For the full chapter, the practice vignettes, and the topic-tagged question bank, head to board pearls dot com. You'll find the rest of the series on Apple Podcasts, Spotify, or wherever you listen to podcasts. That brings us to the end of episode one of three of chapter twenty-three, and I'll see you in the next one.