Portal Hypertension

Meet the patient

A 54-year-old man with known alcohol-related cirrhosis is brought to ED at 2am having vomited bright red blood onto the bedroom floor — three bowlfuls, his wife says. He is pale, sweaty, heart rate 112, blood pressure 88 over 52, and a spider naevus sits on his chest.^[1][2]

Two questions now decide the next two hours: is this a variceal bleed? and can I keep him alive until the endoscopist arrives? Everything below exists to answer those two questions at consultant depth. Hold them, and the four numbers, the five anastomoses, and the acute-bleed bundle all fall into place.^[1]

The HVPG ladder — five defines, ten is clinically significant, twelve marks a haemodynamic response

Portal hypertension is a pressure gradient, and the HVPG numbers decide diagnosis, risk, and treatment response. The gradient is the hepatic venous pressure gradient (HVPG): the wedged (balloon-occluded) hepatic venous pressure, which reads sinusoidal pressure, minus the free hepatic venous pressure, which reads intra-abdominal IVC pressure.^[17][2]

HVPG higher than 5 mmHg defines portal hypertension; Baveno VII specifies that values over 5 mmHg indicate sinusoidal portal hypertension. HVPG of 10 mmHg or more is clinically significant portal hypertension (CSPH) — the level that predisposes to decompensation. A fall to under 12 mmHg, or by more than 20 percent, on a non-selective beta-blocker is the haemodynamic response associated with protection from portal-hypertension-related events. After TIPS, reducing the portal pressure gradient to under 12 mmHg is associated with near-complete protection from portal hypertensive bleeding. HVPG of 16 mmHg or more predicts increased short-term mortality after non-hepatic abdominal surgery — not six-week death from variceal bleeding. The exam mantra "over 16 kills" is that surgical figure, misapplied.^[17][2]

The portal vein forms behind the neck of the pancreas from the superior mesenteric vein (draining the midgut) and the splenic vein (draining the spleen and foregut via the left gastric and short gastric veins). It is valveless and sits between two capillary beds — splanchnic and hepatic sinusoid — so any obstruction, before or within or after the liver, transmits pressure backwards and prises open the embryonic portosystemic anastomoses. The portal vein supplies most hepatic inflow; the hepatic artery supplies the remainder — a conventional anatomic split, not a trial-derived percentage.^[1]

Where the pressure escapes — the five portosystemic anastomoses (OPERA)

A raised portal pressure decompresses through five named sites, and the first one kills patients. These are the embryonic portosystemic channels that re-open when the portal vein cannot drain — and the oesophago-gastric junction is the lethal one.^[1]

The five sites, each a portocaval meeting point:^[1]

1. Oesophago-gastric junction — left gastric (coronary) vein (portal) meets the oesophageal veins draining to the azygos and hemi-azygos (systemic). These dilate into oesophageal and gastric varices — the commonest source of life-threatening haemorrhage.
2. Rectum and anal canal — superior rectal vein (portal, via inferior mesenteric) meets middle and inferior rectal veins (systemic, internal iliac), producing rectal varices.
3. Periumbilical region — paraumbilical veins in the falciform ligament (portal) meet superficial epigastric and thoraco-epigastric veins (systemic), producing caput medusae and the audible Cruveilhier-Baumgarten murmur.
4. Retroperitoneum — veins of the second part of the duodenum, pancreas, and bare areas dilate; typically silent but bleed at surgery.
5. Bare areas of the liver — diaphragmatic and retroperitoneal venous communications.^[1]

^[1]

The classic trap: rectal varices are not haemorrhoids. Varices lie above the dentate line, are portosystemic collaterals, and bleed back into the portal system. Treating them as piles misses the portal hypertension driving them.^[1]

Three sites of obstruction — pre, intra, post

The level of the block predicts whether liver synthetic function is preserved — and that single fact changes the whole management. Pre-hepatic and pre-sinusoidal obstruction leaves hepatocytes intact; sinusoidal cirrhosis and post-hepatic outflow obstruction do not.^[1]

^[1]

Enlarge

The discriminator line: a cirrhotic patient with a variceal bleed has abnormal liver function; a schistosomal or portal-vein-thrombosis patient with the same bleed often has perfectly normal albumin and INR. That single split tells you whether transplant is ever on the table — it is not, when the liver is structurally preserved.^[1]

Who gets it — cirrhosis in the West, schistosomiasis in the tropics

In most of the world portal hypertension is a complication of cirrhosis, and cirrhosis is a top-ten killer of the middle-aged. The global aetiological burden is shifting: alcohol dominates in Europe and the Americas; chronic hepatitis B drives disease in sub-Saharan Africa, East and South-East Asia; hepatitis C is declining where direct-acting antivirals reach patients but persists where they do not; and metabolic dysfunction-associated steatohepatitis (MASH, formerly NASH) tied to obesity and type 2 diabetes is the fastest-rising cause in the West.^[1]

At diagnosis of cirrhosis, varices are already present in about 30 percent of compensated and 60 percent of decompensated patients. The incidence of oesophageal variceal bleeding is about 5 percent per year, rising to as high as 30 percent with large varices. Untreated survivors of a variceal haemorrhage have about a 60 percent risk of rebleeding over 1 to 2 years.^[1][18]

^[1]

The two-hit mechanism — resistance up, inflow up

Portal pressure follows Ohm's law: pressure equals flow multiplied by resistance. Cirrhosis raises both. That dual hit is why a single drug rarely controls it and why beta-blockers — which cut inflow — are the backbone of prophylaxis.^[1]

Enlarge

Hit one — intrahepatic resistance rises. The structural component is obvious: fibrous septa, regenerative nodules, and sinusoidal capillarisation physically distort the vascular bed. But there is also a dynamic, reversible component — injured hepatocytes and sinusoidal endothelial cells activate hepatic stellate cells, which trans-differentiate into contractile myofibroblasts wrapped around the sinusoids. These constrict in response to endothelin-1, angiotensin II, and thromboxane, and relax to nitric oxide. In cirrhosis the intrahepatic NO balance falls, so the sinusoid constricts. This dynamic element is therapeutically important — it is why drugs that lower intrahepatic resistance (carvedilol, statins in trial) have a target.^[1]

Hit two — splanchnic inflow rises. As portal pressure climbs, the splanchnic circulation paradoxically vasodilates, driven by excess nitric oxide, glucagon, and prostacyclin from a hyperdynamic splanchnic endothelium. More blood arrives at a higher-resistance liver, amplifying the hypertension and producing the hyperdynamic circulation of cirrhosis — bounding pulse, warm peripheries, low systemic vascular resistance, high cardiac output — that ultimately decompensates into cardiomyopathy and hepatorenal failure. This is the rationale for non-selective beta-blockers: beta-2 blockade leaves unopposed alpha-mediated splanchnic vasoconstriction, cutting inflow.^[1]

Variceal rupture follows Laplace's law. Wall tension is proportional to transmural pressure multiplied by radius, divided by wall thickness. A varix therefore ruptures when portal pressure is high, the varix is large, and its wall is thin — the basis of endoscopic grading by size and the red wale signs and cherry-red spots that mark wall thinning. It is also why band ligation works: it obliterates the thin-walled varix and replaces it with a fibrotic nodule.^[1]

Four complications declare the disease

Portal hypertension declares itself through four complications, overlaid on the stigmata of chronic liver disease when the cause is cirrhosis. Learn the four as a cluster: varices, ascites, splenomegaly, encephalopathy.^[1]

^{[1] [1]}

When the cause is cirrhosis, the examination also reveals the stigmata of hepatocellular failure: palmar erythema, spider naevi (in the superior vena cava distribution), gynaecomastia and testicular atrophy from impaired oestrogen metabolism, parotid enlargement, Dupuytren contracture, jaundice, fetor hepaticus, and asterixis. Caput medusae — a crown of dilated periumbilical veins radiating from a recanalised umbilical vein — with a venous hum and thrill over the umbilicus is the Cruveilhier-Baumgarten sign, virtually pathognomonic of portal hypertension.^[1]

Examiners test the corners. Consider portal hypertension in a child with painless haematemesis (extrahepatic portal vein obstruction), in a young woman with sudden hepatomegaly and ascites (Budd-Chiari, often post-partum or in a myeloproliferative disorder), in a patient with a known pancreatic mass and isolated gastric varices (splenic vein thrombosis), and in chronic constrictive pericarditis presenting with ascites out of proportion to leg oedema. A patient who bleeds from varices with entirely normal liver synthetic function should prompt a search for a pre-sinusoidal (schistosomiasis) or pre-hepatic (portal vein thrombosis) cause.^[1]

The differential — what else bleeds, what else swells

The acute presentation is upper GI haemorrhage; the chronic is ascites with splenomegaly. Both generate wide differentials, and varices are one cause of each. The table below is the cornerstone of the SAQ.^[1]

| Condition | Key distinguishing feature |^[1] |---|---| | Oesophageal or gastric varices | Known chronic liver disease; massive painless haematemesis; signs of portal HTN; SAAG over 11 g per litre if ascites present | | Peptic ulcer (gastric or duodenal) | Epigastric pain, melaena more than haematemesis, normal liver function, NSAID or H. pylori history | | Mallory-Weiss tear | Post-emesis haematemesis after initial non-bloody vomit; normal liver usually | | Gastric or oesophageal malignancy | Weight loss, dysphagia, anaemia; mass on endoscopy | | Aorto-enteric fistula | Prior aortic graft, sentinel bleed then catastrophic haematemesis; urgent CT angiography | | Dieulafoy lesion or angiodysplasia | Painless bleed, normal liver, diagnosis at endoscopy | | Budd-Chiari syndrome | Acute painful hepatomegaly, ascites, tender liver; hepatic vein thrombosis on imaging | | Right heart failure or constrictive pericarditis | Raised JVP, hepatomegaly, ascites; echocardiography diagnostic | | Intra-abdominal malignancy with peritoneal deposits | SAAG under 11 g per litre (exudate), cytology positive, fixed pelvic mass |

^[1]

The discriminator line: a serum-to-ascites albumin gradient (SAAG) under 1.1 g per decilitre (11 g per litre) makes portal hypertension very unlikely — summary likelihood ratio 0.06 — so think malignancy, tuberculosis, or pancreatic ascites instead, and stop chasing portal pressure.^[6][14]

The bedside round — look for the four, smell the liver

Bedside examination in suspected portal hypertension rarely provides one diagnostic sign; its job is to map the four complications and the stigmata of the underlying cause. Begin with a focused general examination: pallor, jaundice, scratch marks, spider naevi, palmar erythema, gynaecomastia, testicular atrophy, parotid enlargement, Dupuytren contracture, clubbing. Smell for fetor hepaticus — sweet, musty, mercaptan breath. In any bleeder, assess hydration and the signs of shock: tachycardia, hypotension, cold peripheries, oliguria, confusion.^[1]

Abdominal examination looks for hepatosplenomegaly, ascites (shifting dullness, fluid thrill), caput medusae, and superficial abdominal wall collaterals. A venous hum over the umbilicus (Cruveilhier-Baumgarten) is pathognomonic. Test for asterixis (flapping tremor with wrists dorsiflexed) and constructional apraxia (Reitan trail, clock drawing) for encephalopathy. A digital rectal examination may reveal melaena, and anoscopy distinguishes true rectal varices (above the dentate line, blanching bluish) from haemorrhoids.^[1]

Bedside pearl: what does a platelet count of 70 with splenomegaly tell you? (answer)ShowHide

Baveno VII splits two non-invasive questions. CSPH is ruled out when liver stiffness by transient elastography is 15 kPa or less and the platelet count is 150 or more, and ruled in when stiffness is 25 kPa or more (virus- or alcohol-related, and non-obese NASH, cACLD). High-risk varices are still ruled out with the Baveno VI pair: stiffness under 20 kPa and platelets over 150 — not stiffness under 25 kPa. The "25 kPa plus platelets 150" mash-up is a common mix-up of the CSPH rule-in cut-off with the variceal screening rule-out.^[2]

Investigations — confirm, locate, grade

Investigation serves three purposes: confirm portal hypertension, define its cause and site, and grade severity for prognosis and transplant listing. Run them in that order and you will not miss a step.^[1]

Bloods. A full blood count shows thrombocytopaenia (hypersplenism), anaemia (blood loss, alcohol, folate deficiency, haemodilution), and leucopaenia. Liver function tests show a cholestatic or hepatocellular pattern depending on aetiology. Coagulation (INR) and albumin gauge synthetic function and feed the Child-Pugh and MELD scores. Renal function and electrolytes screen for hepatorenal syndrome and hyponatraemia. Send viral serology (HBsAg, anti-HCV), autoimmune and metabolic markers (ferritin, caeruloplasmin, alpha-1-antitrypsin, immunoglobulins, anti-mitochondrial antibody), and alpha-fetoprotein for hepatocellular carcinoma surveillance.^[1]

Endoscopy (OGD). The cornerstone for diagnosis and therapy. Varices are graded by size (small, medium, large) and inspected for red signs (red wale marks, cherry-red spots, haematocystic spots) that predict bleeding. Gastric varices are classified by Sarin (GOV1, GOV2, IGV1, IGV2). At the same session, band ligation or glue injection delivers therapy.^[1]

Ultrasound with Doppler is first-line imaging. It shows liver nodularity and echotexture, portal vein patency and flow direction (hepatofugal — away from the liver — in advanced disease), splenomegaly, ascites, and screens for hepatocellular carcinoma. Portal vein thrombosis appears as an echogenic clot or absence of flow with cavernous transformation when chronic. Splenic vein patency should be checked specifically when isolated gastric varices are found.^[1]

Cross-sectional imaging — CT and MR portography, contrast-enhanced ultrasound. These define vascular anatomy, confirm thrombosis, plan TIPSS or surgery, screen for hepatocellular carcinoma, and quantify collaterals. MR elastography and transient elastography (FibroScan) estimate liver stiffness, which — combined with platelet count — stratifies the probability of CSPH and high-risk varices non-invasively, per Baveno VII.^[2]

HVPG is the gold standard for diagnosing and grading portal hypertension. A balloon-tipped catheter is passed under fluoroscopy via the internal jugular vein to a hepatic vein; the wedged pressure reflects sinusoidal pressure and the free pressure the IVC. The difference is the HVPG. It is reproducible and prognostic and is the metric by which beta-blocker response is judged — but it is invasive and requires expertise, so it is reserved for diagnostic uncertainty, research, and assessing treatment response in specialised centres.^[1]

Severity scores — Child-Pugh and MELD

^[2]

Child-Pugh combines bilirubin, albumin, INR (or prothrombin time), encephalopathy, and ascites to give a 5 to 15 score (A, B, or C). MELD uses bilirubin, INR, and creatinine (plus sodium in MELD-Na) and, with Child-Pugh, is the score most often used to gauge severity in acute variceal bleeding. Median survival is more than 15 years in compensated cirrhosis, about 2 years once decompensated, and about 9 months with further decompensation — and every decompensated patient should be considered for liver transplant, rather than waiting on an unsourced MELD-15 cut-off.^[2][10][13]

The acute bleed bundle — ABC-TAB

Enlarge

An acute variceal bleed is a medical emergency with a fixed bundle, started before endoscopy and continued through it. The four pillars — airway, restrictive transfusion, vasoactive drug, and antibiotic — are delivered simultaneously, not sequentially, alongside plans for urgent endoscopy and a defined rescue pathway.^[1][2]

^{[1] [1]}

Two principles are examiner-favourite traps. First, transfusion is restrictive: a threshold of haemoglobin 7 g per decilitre beat a liberal 9 g per decilitre threshold for six-week survival (95 versus 91 percent; hazard ratio for death 0.55) and reduced further bleeding (10 versus 16 percent), with benefit driven largely by the cirrhotic subgroup. Second, antibiotic prophylaxis reduces bacterial infections after a variceal bleed — a risk difference of 15.2 percent favouring prophylaxis — but the mortality benefit is now uncertain: a 2025 Bayesian meta-analysis of 14 randomised trials found shorter (2 to 3 day) or no prophylaxis noninferior for all-cause mortality, so guideline courses of 5 to 7 days rest on low-quality evidence.^[7][16]

The vasoactive drug. The standard terlipressin bolus regimen is 2 mg intravenously every 4 hours; head-to-head data show that a low-dose continuous infusion (4 mg over 24 hours for 5 days) achieved a higher 24-hour HVPG response than boluses (85.4 versus 58.2 percent) at roughly half the daily dose, with fewer adverse events and less very-early rebleeding. Current expert guidance names octreotide as the vasoactive drug of choice on safety grounds, reserves terlipressin where its potency is needed, and contraindicates terlipressin in ongoing coronary, peripheral, or mesenteric ischaemia and in hypoxaemia. A systematic review and meta-analysis found no difference between terlipressin or vasopressin versus octreotide or somatostatin in mortality, bleeding control, rebleeding, transfusion, or length of stay — though the terlipressin group had more adverse events.^[11][10][4]

Refractory bleeding — the rescue ladder

If bleeding continues despite vasoactive drug and endoscopic therapy, escalate in this fixed order. Do not improvise — each rung has a defined role. Balloon tamponade and covered metal stents are bridges to PTFE-covered TIPSS, not destinations.^[2]

1. Balloon tamponade (Sengstaken-Blakemore or Minnesota tube, or the self-expanding Cook tube). The gastric balloon is inflated to tamponade the varix at the gastro-oesophageal junction. Use only as a bridge to definitive treatment (covered TIPSS). Baveno VII rates self-expanding covered metal stents as efficacious as balloon tamponade and safer. Risks of balloon tamponade include aspiration, oesophageal ulceration and perforation, and pressure necrosis. Endotracheal protection should be in place.
2. Self-expanding covered oesophageal metal stent — an alternative bridge that is easier to place and carries a lower perforation risk.
3. Transjugular intrahepatic portosystemic shunt (TIPSS) — the definitive rescue therapy, decompressing the portal system by stenting a tract between the portal and hepatic veins.^[1]

The classic trap: the Sengstaken-Blakemore tube is a bridge, not a treatment. It buys time to reach covered TIPSS or a self-expanding metal stent — nothing more. A covered metal stent is the safer bridge when available.^[2]

Pre-emptive TIPSS — the 72-hour fork for high-risk bleeders

For the highest-risk patients, TIPSS is not rescue — it is the plan from hour one. Baveno VII indicates pre-emptive PTFE-covered TIPSS within 72 hours (ideally under 24 hours) for bleeding from oesophageal or GOV1/GOV2 varices if Child-Pugh C under 14, or Child-Pugh B greater than 7 with active bleeding at initial endoscopy, or HVPG over 20 mmHg at the time of haemorrhage.^[2]

The landmark Early TIPS trial (Garcia-Pagan, NEJM 2010) randomised 63 high-risk patients (Child-Pugh C, or class B with persistent bleeding at endoscopy) within 24 hours of admission to covered-stent TIPSS within 72 hours after randomisation versus pharmacotherapy plus EVL. Early TIPSS gave a 97 percent versus 50 percent one-year probability of remaining free of the composite endpoint (failure to control bleeding or rebleeding) and improved one-year survival (86 percent versus 61 percent), without excess serious adverse events. Baveno VII later tightened the class-B criterion to B greater than 7 with active bleeding.^[3][2]

Consultant confession: the hardest part of this decision is not the evidence — it is recognising, at 3am, that the patient in front of you qualifies. Calculate Child-Pugh on admission, before the bleed distorts the labs, and flag the high-risk patient for pre-emptive TIPSS the moment endoscopy confirms active bleeding.^[3]

Prophylaxis — stop the first bleed, stop the next one

Primary prophylaxis — prevent the first bleed

All patients with cirrhosis should be screened for varices. The aim of primary prophylaxis is to prevent the first variceal haemorrhage in those found to have medium or large varices, or small varices with red signs or Child-Pugh B or C disease. Do not mash the cut-offs: CSPH is ruled out when liver stiffness is 15 kPa or less with platelets 150 or more, and ruled in at 25 kPa or more. Endoscopy for high-risk varices can be spared using Baveno VI — stiffness under 20 kPa and platelets over 150 — in patients who cannot take an NSBB. Carvedilol is the preferred NSBB in compensated cirrhosis.^[2]

Non-selective beta-blocker (NSBB). The role of NSBBs in primary prophylaxis has expanded across the Baveno editions, and variceal surveillance is now individualised by liver stiffness and platelet count rather than routine annual endoscopy; endoscopic band ligation is the preferred alternative for patients intolerant of NSBBs.^[1] Carvedilol is the practical first choice when a haemodynamic response matters: in an HVPG-guided dose-response cohort, carvedilol 6.25 mg daily produced an HVPG fall of at least 20 percent (or to 12 mmHg or below) in 39 percent of patients, rising to 53 percent after escalation to 12.5 mg daily — the same responder definition used to judge beta-blocker success.^[12]

Endoscopic variceal ligation (EVL) is first-line for large varices, or where NSBB are contraindicated or not tolerated. Bands are applied at two- to eight-week intervals until variceal eradication (typically two to four sessions).^[1]

Secondary prophylaxis — prevent rebleed

Every patient who survives a variceal bleed needs secondary prophylaxis — untreated 1- to 2-year rebleeding is about 60 percent. First-line is combination traditional NSBB or carvedilol plus EVL.^[18][2]

TIPSS, shunts, and transplant — the decompressive ladder

TIPSS places a covered stent between an intrahepatic portal vein branch and a hepatic vein under fluoroscopic guidance via the internal jugular vein, decompressing the portal system. It controls refractory bleeding, refractory ascites, hepatic hydrothorax, and Budd-Chiari syndrome, and is the bridge to transplant.^[1]

Early covered-stent TIPSS in high-risk bleeders achieved a 97 percent versus 50 percent one-year freedom from failure-to-control-bleeding-or-rebleeding. Versus repeated paracentesis for refractory ascites, TIPS more than doubles hepatic encephalopathy (odds ratio 2.24). Stenosis or thrombosis requires surveillance Doppler. Contraindications include severe heart failure, severe liver failure, polycystic liver disease, uncontrolled encephalopathy, and active sepsis.^[3][15]

Surgical shunts are now rarely used — supplanted by TIPSS, which is less invasive and carries lower peri-procedural mortality. They are reserved for failed TIPSS, TIPSS not technically feasible, or specialised centres with surgical expertise:^[1]

• Distal splenorenal (Warren) shunt — selectively decompresses gastro-oesophageal varices into the left renal vein while preserving mesenteric-portal perfusion; lowest encephalopathy rate of the surgical shunts.
• Mesocaval shunt — jugular or prosthetic graft between SMV and IVC.
• Portocaval shunt — side-to-side or end-to-side; the most effective decompression but the highest encephalopathy rate.^[1]

Devascularisation procedures (oesophageal transection, Sugiura-Futagawa) are historical salvage options now virtually obsolete in centres with TIPSS.^[1]

Liver transplant is the only definitive cure for portal hypertension arising from cirrhosis. Reserved for decompensated cirrhosis — Baveno VII says every decompensated patient should be considered for transplant, including after a decompensating event such as refractory ascites, recurrent encephalopathy, or hepatorenal syndrome. Allocation is by MELD or MELD-Na score. Portal hypertension from pre-hepatic or pre-sinusoidal causes (portal vein thrombosis, schistosomiasis) does not require transplant — the liver is structurally preserved.^[1]

Ascites and encephalopathy — the stepwise ladders

Ascites is managed by a stepwise ladder, from salt restriction to transplant.^[14]

1. Modest salt restriction plus diuretic therapy, beginning with spironolactone (or an equivalent aldosterone antagonist) in the first instance; diuretics are added stepwise while sodium restriction continues.^[14]
2. Gross or tense ascites is treated with therapeutic paracentesis followed by colloid volume expansion, and then diuretic therapy to prevent re-accumulation.^[14]
3. Large-volume paracentesis — give intravenous albumin at the time of a tap over 5 litres to prevent acute kidney injury and death; albumin is also the volume expander of choice in hospitalised patients with ascites and AKI, but is not indicated in uncomplicated ascites.^[10]
4. Refractory ascites — repeated large-volume paracentesis or TIPSS; TIPS reduces ascites re-accumulation at 3 and 12 months but more than doubles the odds of hepatic encephalopathy (odds ratio 2.24), with no significant difference in mortality, bleeding, infection, or acute renal failure.^[14][15]
5. Liver transplant for refractory ascites as a decompensating event.^[14]

Hepatic encephalopathy combines precipitant removal with ammonia-lowering therapy. Most episodes resolve when you find and fix the trigger — so hunt for it first.^[9]

1. Identify and treat the precipitant — infection (including SBP), gastrointestinal bleeding, constipation, dehydration, electrolyte disturbance, sedatives or opioids, TIPSS, and worsening liver failure. This alone resolves many episodes.
2. Lactulose remains the backbone of therapy — over 90 percent of patients in the pivotal rifaximin maintenance trial received concomitant lactulose.^[9]
3. Rifaximin 550 mg twice daily added to lactulose reduces breakthrough encephalopathy (22.1 versus 45.9 percent at six months; hazard ratio 0.42) and hospitalisation involving encephalopathy (13.6 versus 22.6 percent; hazard ratio 0.50).^[9]
4. Supportive care — airway protection in grade III to IV encephalopathy, nutrition with adequate protein, correction of electrolytes.^[1]

The classic trap: do not restrict protein in hepatic encephalopathy. A randomised trial found a normal-protein diet safe in episodic encephalopathy, with no benefit from restriction and higher protein breakdown on the low-protein diet. Treat the precipitant, lower the ammonia, feed the patient.^[19]

The complications that kill — SBP, HRS, encephalopathy

Variceal bleeding carries a six-week mortality of about 20 percent per episode — 21.1 percent in a large tertiary cohort — even with modern endoscopic and pharmacological therapy. Independent predictors of death are rebleeding, Child-Pugh class C, and haemodynamic instability at admission. Bleeding also precipitates encephalopathy, infection, and hepatorenal syndrome — so the post-bleed patient is never "stable", only "between complications".^[13]

Spontaneous bacterial peritonitis (SBP) is infection of ascitic fluid without an obvious source; a diagnostic tap is mandatory in every cirrhotic with ascites who deteriorates. Diagnostic tap: an ascitic polymorphonuclear count over 250 cells per microlitre raises the likelihood of SBP (summary likelihood ratio 6.4), and a count at or under 250 lowers it (likelihood ratio 0.2); fluid should be inoculated into blood culture bottles at the bedside, which improves microbiological yield from 77 to 100 percent. Treat with a third-generation cephalosporin — cefotaxime, dosed according to renal function — plus intravenous albumin 1.5 g per kg at diagnosis and 1 g per kg on day three: in the landmark trial this reduced renal impairment from 33 to 10 percent and in-hospital mortality from 29 to 10 percent.^[6][8]

Hepatorenal syndrome (HRS-AKI) is functional renal failure from intense splanchnic vasodilation reducing effective circulating volume and renal perfusion; the kidneys are histologically normal. Diagnosed by the ICA criteria (rise in creatinine, no response to albumin challenge, exclusion of other causes). HRS-AKI (formerly type 1) is rapid and fatal untreated; treated with terlipressin plus albumin (or noradrenaline in ICU). HRS with refractory ascites (formerly type 2) is indolent. Definitive treatment is liver transplant.^[1]

Other complications to name. Hepatopulmonary syndrome (intrapulmonary shunting, hypoxaemia, orthodeoxia) and portopulmonary hypertension (pulmonary arterial hypertension from portosystemic shunting) are pulmonary vascular complications. Hepatic hydrothorax (usually right-sided) and umbilical hernia with risk of rupture are mechanical complications of ascites. Gastric antral vascular ectasia and portal hypertensive gastropathy cause chronic occult bleeding.^[1]

The six classic pitfalls

• Over-transfusing a variceal bleed — a restrictive threshold (transfuse when haemoglobin falls under 7 g per decilitre) improved six-week survival and reduced further bleeding versus a liberal one.^[7]
• Forgetting the diagnostic tap in a deteriorating cirrhotic with ascites — an ascitic PMN count over 250 cells per microlitre diagnoses SBP; inoculate the culture bottles at the bedside.^[6]
• Misattributing low-gradient ascites to portal hypertension — a SAAG under 1.1 g per decilitre (11 g per litre) makes portal hypertension very unlikely; malignancy and tuberculosis are the non-cirrhotic causes to exclude.^[6][14]
• Delaying TIPSS in a high-risk bleeder — early covered-stent TIPSS placed within 72 hours of randomisation beat pharmacotherapy plus EVL for rebleeding and for survival.^[3]
• Missing post-TIPSS encephalopathy — TIPS more than doubles hepatic encephalopathy versus paracentesis (odds ratio 2.24); rifaximin 550 mg twice daily plus lactulose prevents breakthrough episodes.^[15][9]
• Stopping the vasoactive drug too early — start on suspicion, before endoscopy, and continue 2 to 5 days after haemostasis to prevent early rebleeding.^[10]

Special situations — the corners examiners test

Portal vein thrombosis (pre-hepatic). Baveno VII splits recent PVT without cirrhosis from PVT in cirrhosis. In recent PVT without cirrhosis, anticoagulation is given for at least 6 months; after that, long-term anticoagulation is recommended if there is a permanent prothrombotic state. Chronic extrahepatic obstruction produces cavernoma (periportal collaterals) and portal hypertension with preserved synthetic function. In cirrhosis with PVT, acute variceal bleeding is managed with the same bundle as without PVT when possible — do not transplant the non-cirrhotic 6-month rule onto every cirrhotic thrombus.^[2]

Budd-Chiari syndrome (post-hepatic). Baveno defines BCS as obstruction of hepatic venous outflow from the small hepatic veins to the IVC entrance into the right atrium. Presentation is extremely diverse — do not wait for a textbook triad. Management is stepwise at an experienced centre: anticoagulation, then angioplasty, stent, thrombectomy or thrombolysis, then TIPS, then liver transplant. Long-term anticoagulation is given to all primary BCS. TIPS is attempted when angioplasty is not feasible and the patient does not improve on medical therapy including anticoagulants; transplant is considered when that ladder fails or in acute liver failure.^[2]

Splenic vein thrombosis (sinistral, left-sided portal hypertension). Teaching association: isolated gastric (fundal) varices with preserved liver function after pancreatitis or pancreatic cancer. Splenectomy as "curative" is conventional surgical teaching and is not a Baveno VII statement.^[1]

Gastric varices. For acute bleeding, Baveno VII recommends tissue adhesives (N-butyl-cyanoacrylate or thrombin) for isolated gastric varices and for GOV2 that extends beyond the cardia; EVL or tissue adhesive can be used for GOV1. BRTO may be considered as an alternative to endoscopic treatment or TIPS for GOV2, IGV1, and ectopic varices when the shunt anatomy is suitable and local expertise exists.^[2]

Children (portal cavernoma / extrahepatic portal vein obstruction). Baveno VII: mesenteric-left portal vein bypass (Meso-Rex) should be considered in all children with complications of portal cavernoma, referred to centres experienced in the operation. That is the sourced indication — not an unsourced claim that extrahepatic obstruction is "the commonest paediatric cause."^[2]

Pregnancy. Portal hypertension in pregnancy is managed as a high-bleed-risk state (expanded blood volume and intra-abdominal pressure). Screen and eradicate varices before conception where possible. Terlipressin is generally avoided in pregnancy because of uterine vasospasm — that caution is conventional obstetric practice and is not a Baveno VII statement. Endoscopic therapy is the practical haemostatic option; TIPSS and surgery are deferred where possible. Vaginal delivery is acceptable with good pain control; caesarean is for obstetric indications.^[1]

The anticoagulated patient. Anticoagulation is not contraindicated in cirrhosis once varices are treated; portal vein thrombosis on anticoagulation should be managed jointly with hepatology. Reverse warfarin with vitamin K and prothrombin complex concentrate for an acute bleed; direct oral anticoagulants have specific reversal strategies.^[1]

The elderly and comorbid. Frail, elderly patients tolerate bleeds poorly; beta-blocker dosing should start low and titrate slowly; pre-emptive TIPSS decisions weigh encephalopathy risk against bleeding risk.^[1]

Prognosis and disposition

Variceal bleeding carries a six-week mortality of about 20 percent per episode in modern series (21.1 percent in a large tertiary cohort, predicted by rebleeding, Child-Pugh class C, and shock at presentation), lower with adherence to the full bundle — restrictive transfusion, vasoactive drug, antibiotic, EVL, and early TIPSS where indicated. Survivors need structured secondary prophylaxis and surveillance endoscopy.^[13]

Cirrhosis-stage median survival (not Child-Pugh class percentages): more than 15 years compensated, about 2 years decompensated, about 9 months with further decompensation. MELD and Child-Pugh remain the scores used to gauge severity; decompensation itself is the transplant-assessment trigger.^[10][2]

TIPSS in the Early TIPS trial gave 97 versus 50 percent one-year freedom from failure-to-control-bleeding-or-rebleeding and 86 versus 61 percent one-year survival in high-risk bleeders. For refractory ascites, TIPS reduces re-accumulation but more than doubles encephalopathy (odds ratio 2.24) without a proven mortality difference versus paracentesis.^[3][15]

Pre-hepatic and pre-sinusoidal causes (portal vein thrombosis, schistosomiasis, congenital hepatic fibrosis) carry a far better prognosis because hepatocellular function is preserved; patients tolerate bleeds well and live normal spans with variceal control.^[1]

Disposition. Every acute variceal bleed is admitted to a unit with 24-hour endoscopy and interventional radiology. High-risk patients (Child-Pugh C under 14, Child-Pugh B greater than 7 with active bleeding, or HVPG over 20 mmHg) are flagged early for pre-emptive TIPSS. Survivors are enrolled in a surveillance and secondary prophylaxis programme, with transplant referral at decompensation.^[2]

Evidence, guidelines, and regional deltas

Baveno VII consensus (2022) is the current international standard for portal hypertension — "Personalised Care for Portal Hypertension" — superseding Baveno VI. Key recommendations: CSPH is defined non-invasively by liver stiffness 25 kPa or more (rule-in) or 15 kPa or less with platelets 150 or more (rule-out); high-risk varices are ruled out with Baveno VI — stiffness under 20 kPa and platelets over 150; NSBBs (preferably carvedilol) are first-line for CSPH; EVL if NSBB-intolerant; the acute bleed bundle is vasoactive drug, antibiotic from admission, and EVL within 12 hours, with pre-emptive PTFE-covered TIPSS within 72 hours (ideally under 24) for Child-Pugh C under 14, Child-Pugh B greater than 7 with active bleeding, or HVPG over 20 mmHg; secondary prophylaxis combines NSBB and EVL.^[2]

The Early TIPS trial (Garcia-Pagan, NEJM 2010). Sixty-three high-risk patients (Child-Pugh C, or class B with persistent bleeding at endoscopy) were randomised within 24 hours of admission to covered-stent TIPSS within 72 hours after randomisation versus pharmacotherapy plus EVL. Early TIPSS gave 97 percent versus 50 percent one-year freedom from failure-to-control-bleeding-or-rebleeding and 86 percent versus 61 percent one-year survival, without excess serious adverse events. Baveno VII later specified C under 14 and B greater than 7 with active bleeding.^[3][2]

Vasoactive agent meta-analysis (Huaringa-Marcelo 2021). Twenty-one RCTs comparing terlipressin or vasopressin with octreotide or somatostatin found no difference in mortality, bleeding control, rebleeding, transfusion, or length of stay; the terlipressin group had more adverse events. The choice of vasoactive agent is therefore often driven by availability, cost, and contraindications rather than efficacy.^[4]

The mantra

Over 5 defines portal hypertension, over 10 is clinically significant, a fall to under 12 mmHg or by over 20 percent is an NSBB response, and 16 mmHg is a surgical-risk threshold — not a variceal-bleed mortality cut-off. The four numbers (correctly labelled), the five anastomoses (OPERA), the four complications (varices, ascites, splenomegaly, encephalopathy), and the acute-bleed bundle — restrictive transfusion, a vasoactive drug started before endoscopy, antibiotic prophylaxis from admission, and band ligation within 12 hours — are the spine of every portal-hypertension answer.^[17][2]

Ward-round test

Stem 1 — the man from the top of the topic (answer)ShowHide

The 54-year-old cirrhotic who vomited three bowlfuls of blood at 2am. Heart rate 112, blood pressure 88 over 52. What do you do in the next 30 minutes? Model: This is an acute variceal bleed until proven otherwise. Run the bundle simultaneously: restrictive transfusion — transfuse only when the haemoglobin falls under 7 g per decilitre, because a restrictive strategy improved six-week survival and cut further bleeding versus a liberal threshold; start the vasoactive drug now, before endoscopy — the standard terlipressin bolus regimen is 2 mg every 4 hours, continued 2 to 5 days after haemostasis; antibiotic prophylaxis, which reduces post-bleed infection though its mortality benefit is now questioned by a 2025 Bayesian meta-analysis; and endoscopic variceal ligation at the urgent endoscopy. Then decide whether he is a candidate for early TIPSS — the landmark trial randomised within 24 hours and placed a covered stent within 72 hours, with markedly better rebleeding-free and overall survival.^{[7][10][11][16][3]}

Stem 2 — the well-looking patient who bled from varices with normal liver function (answer)ShowHide

A 28-year-old man from a schistosomiasis-endemic region presents with painless haematemesis. Endoscopy shows oesophageal varices. His albumin, INR, and bilirubin are entirely normal. What is the mechanism, and what does this tell you about transplant? Model: This is pre-sinusoidal portal hypertension from schistosomiasis — eggs carried continuously through the portal circulation provoke granulomatous inflammation, amputation of the intrahepatic venules, and periportal pipe-stem (Symmers) fibrosis, producing an intrahepatic pre-sinusoidal block with a well-preserved liver parenchyma. Hepatocytes are intact, so advanced liver failure is not seen unless another liver disease coexists, and gastrointestinal bleeding is the most frequent cause of death. Treat the parasite with praziquantel (or oxamniquine) and manage the varices endoscopically or surgically — splenectomy with gastro-oesophageal devascularisation is a recognised option. Transplant is not the answer: the liver is structurally preserved.^[5]

Stem 3 — the ascites with SAAG under 11 (answer)ShowHide

A cirrhotic patient has tense ascites. The diagnostic tap returns fluid with a SAAG of 8 g per litre. The registrar attributes it to portal hypertension. What is the error? Model: A SAAG under 1.1 g per decilitre (11 g per litre) argues strongly against portal hypertension — summary likelihood ratio 0.06. Portal hypertensive ascites sits above that gradient. A low-gradient ascites means malignancy, tuberculosis, or pancreatic ascites until proven otherwise — the International Ascites Club stresses that these non-cirrhotic causes occur with increased frequency in patients with liver disease — so send cytology and culture, inoculating blood culture bottles at the bedside, and stop chasing portal pressure as the cause of the fluid.^[6][14]

Stem 4 — the post-TIPSS confusion (answer)ShowHide

Three days after a TIPSS for refractory ascites, a 60-year-old cirrhotic becomes drowsy and develops a flapping tremor. What happened, and what do you do? Model: This is post-TIPSS hepatic encephalopathy — a complication significantly more frequent after TIPS than after paracentesis (odds ratio 2.24 in meta-analysis), because the shunt diverts nitrogenous toxins past the liver. Hunt for a precipitant (infection, constipation, sedatives), continue lactulose, and add rifaximin 550 mg twice daily, which reduced breakthrough encephalopathy from 45.9 to 22.1 percent and encephalopathy hospitalisations from 22.6 to 13.6 percent at six months on top of lactulose. Most episodes resolve with medical therapy.^[15][9]

References19ShowHide

1. [1]Brzdęk M, Dobrowolska K, Janczura J, et al. Advances in portal hypertension management: Evolution of the Baveno guidelines World J Gastroenterol, 2025.PMID 41024757
2. [2]de Franchis R, Bosch J, Garcia-Tsao G, et al. Baveno VII - Renewing consensus in portal hypertension J Hepatol, 2022.PMID 35120736
3. [3]García-Pagán JC, Caca K, Bureau C, et al. Early use of TIPS in patients with cirrhosis and variceal bleeding N Engl J Med, 2010.PMID 20573925
4. [4]Huaringa-Marcelo J, Huaman MR, Brañez-Condorena A, et al. Vasoactive Agents for the Management of Acute Variceal Bleeding: A Systematic Review and Meta-analysis J Gastrointestin Liver Dis, 2021.PMID 33723542
5. [5]Da Silva LC, Carrilho FJ Hepatosplenic schistosomiasis. Pathophysiology and treatment Gastroenterol Clin North Am, 1992.PMID 1568771
6. [6]Wong CL, Holroyd-Leduc J, Thorpe KE, et al. Does this patient have bacterial peritonitis or portal hypertension? How do I perform a paracentesis and analyze the results? JAMA, 2008.PMID 18334692
7. [7]Villanueva C, Colomo A, Bosch A, et al. Transfusion strategies for acute upper gastrointestinal bleeding N Engl J Med, 2013.PMID 23281973
8. [8]Sort P, Navasa M, Arroyo V, et al. Effect of intravenous albumin on renal impairment and mortality in patients with cirrhosis and spontaneous bacterial peritonitis N Engl J Med, 1999.PMID 10432325
9. [9]Bass NM, Mullen KD, Sanyal A, et al. Rifaximin treatment in hepatic encephalopathy N Engl J Med, 2010.PMID 20335583
10. [10]Garcia-Tsao G, Abraldes JG, Rich NE, et al. AGA Clinical Practice Update on the Use of Vasoactive Drugs and Intravenous Albumin in Cirrhosis: Expert Review Gastroenterology, 2024.PMID 37978969
11. [11]Arora V, Choudhary SP, Maiwall R, et al. Low-dose continuous terlipressin infusion is effective and safer than intravenous bolus injections in reducing portal pressure and control of acute variceal bleeding Hepatol Int, 2023.PMID 36542261
12. [12]Schwarzer R, Kivaranovic D, Paternostro R, et al. Carvedilol for reducing portal pressure in primary prophylaxis of variceal bleeding: a dose-response study Aliment Pharmacol Ther, 2018.PMID 29492989
13. [13]Raza M, Ul Hassan M, Zaheer M, et al. Clinical Outcomes and Predictors of Mortality in Variceal Upper Gastrointestinal Bleeding: A Retrospective Study From a Tertiary Care Hospital Cureus, 2026.PMID 41952939
14. [14]Moore KP, Wong F, Gines P, et al. The management of ascites in cirrhosis: report on the consensus conference of the International Ascites Club Hepatology, 2003.PMID 12830009
15. [15]Saab S, Nieto JM, Lewis SK, et al. TIPS versus paracentesis for cirrhotic patients with refractory ascites Cochrane Database Syst Rev, 2006.PMID 17054221
16. [16]Prosty C, Noutsios D, Dubé LR, et al. Prophylactic Antibiotics for Upper Gastrointestinal Bleeding in Patients With Cirrhosis: A Systematic Review and Bayesian Meta-Analysis JAMA Intern Med, 2025.PMID 40788637
17. [17]Procopeţ B, Tantau M, Bureau C Are there any alternative methods to hepatic venous pressure gradient in portal hypertension assessment? J Gastrointestin Liver Dis, 2013.PMID 23539394
18. [18]de Franchis R, Primignani M Natural history of portal hypertension in patients with cirrhosis Clin Liver Dis, 2001.PMID 11565135
19. [19]Córdoba J, López-Hellín J, Planas M, et al. Normal protein diet for episodic hepatic encephalopathy: results of a randomized study J Hepatol, 2004.PMID 15246205