Primary prevention of esophageal variceal hemorrhage in cirrhosis — screening, risk stratification, and non-selective beta-blocker titration
Once a patient is diagnosed with compensated cirrhosis, the clinical question becomes whether esophageal or gastric varices have already formed as a consequence of portal hypertension. Traditionally this meant every cirrhotic patient underwent screening upper endoscopy (esophagogastroduodenoscopy, EGD). The Baveno VII consensus workshop (2021) reshaped this pathway by allowing many patients to skip invasive endoscopy entirely, using non-invasive markers to identify those at genuinely low probability of having varices needing treatment.
Cirrhosis distorts hepatic architecture with fibrous septa and regenerative nodules, raising resistance to portal venous flow. Once the hepatic venous pressure gradient (HVPG) — the difference between wedged and free hepatic venous pressure, a surrogate for portal pressure — exceeds approximately 10 mmHg, the threshold termed "clinically significant portal hypertension" (CSPH), portosystemic collateral vessels begin to form and dilate. The submucosal veins of the distal esophagus are a preferred collateral pathway because they connect the portal (splanchnic) circulation to the systemic azygos venous system with relatively little resistance.
As portal pressure rises further, these collaterals enlarge into visible, tortuous, saccular varices projecting into the esophageal lumen. Varices are essentially valveless, thin-walled veins under arterial-like pressure transmitted from the portal system; when wall tension exceeds the elastic limit of the overlying mucosa, rupture and life-threatening hemorrhage follow. Roughly 30–40% of patients have varices at the time cirrhosis is first diagnosed, and up to 90% will develop them over the natural history of the disease if they live long enough without effective portal-pressure-lowering therapy.
Traditional screening exposed large numbers of patients to endoscopy who did not, in fact, have varices requiring intervention — an invasive, resource-intensive, and imperfectly tolerated test performed essentially as a population screen. The Baveno VI and subsequently Baveno VII consensus criteria formalized a non-invasive alternative built on two widely available, inexpensive measurements:
• Liver stiffness measurement (LSM) by transient elastography (FibroScan), a validated surrogate for the degree of fibrosis and portal pressure • Platelet count, which falls as portal hypertension causes splenic sequestration (hypersplenism) from splenomegaly
Patients with LSM <20 kPa AND platelet count >150,000/µL have a probability of harboring varices needing treatment (VNT — meaning medium/large varices or small varices with red signs) below 5%. In this low-risk group, endoscopy can be safely deferred, with annual reassessment of stiffness and platelets to detect progression. Patients who do not meet both criteria proceed to endoscopic screening, where the yield of clinically important findings is substantially enriched.
The "Baveno VI criteria" (LSM <20 kPa, platelets >150,000/µL) were shown across multiple validation cohorts to spare roughly 20–25% of patients an unnecessary endoscopy while missing fewer than 5% of varices needing treatment — a favorable trade-off that Baveno VII reaffirmed as the standard non-invasive screening pathway in compensated advanced chronic liver disease (cACLD).
For patients who do not meet the low-risk elastography/platelet combination — or where transient elastography is unavailable, unreliable (e.g., ascites, obesity, narrow intercostal spaces), or the patient has already decompensated — upper endoscopy remains the diagnostic standard. During EGD, the endoscopist systematically documents:
• Presence, number, and location of esophageal varices (distal esophagus, extending proximally with severity) • Size, most simply dichotomized as small (<5mm, not obliterated by insufflation) versus large/medium (≥5mm) • Any red color signs on the variceal surface — red wale marks, cherry-red spots, hematocystic spots • Presence of gastric varices and portal hypertensive gastropathy • Child-Pugh class and other clinical decompensation markers, integrated with the endoscopic findings
This single examination anchors every subsequent decision about primary prophylaxis, and is repeated on a surveillance interval (typically 1–3 years, or annually if small varices or ongoing risk factors like active alcohol use are present) if prophylaxis is not yet indicated.
Not all varices carry the same rupture risk. Endoscopic grading combines three variables — size, the presence of red color signs, and the severity of underlying liver disease — to separate patients who need immediate prophylactic therapy from those who can be safely observed. This composite approach, formalized decades ago by the North Italian Endoscopic Club (NIEC), remains the conceptual backbone of contemporary risk stratification.
Variceal wall tension follows a Laplace-type relationship: for a given intravariceal pressure, wall tension rises with radius. A large, dilated varix therefore experiences greater circumferential wall stress than a small one at the same portal pressure, and is intrinsically closer to its rupture point. This is why size alone — dichotomized in most staging systems into small (<5mm) versus large/medium (≥5mm), sometimes further split into a three-tier small/medium/large scheme — is the single strongest endoscopic predictor of first hemorrhage.
Small varices without other risk features carry a modest bleeding risk and are frequently observed rather than treated, particularly since the incremental benefit of pharmacologic prophylaxis in this group is less firmly established than in large varices. Large or medium varices, by contrast, meet an unambiguous threshold for primary prophylaxis in essentially all published guidelines.
Independent of overall variceal caliber, the endoscopist looks for red-color signs on the variceal surface, which indicate a segment of variceal wall that is under focal excessive tension and mechanically closer to rupture:
• Red wale marks: longitudinal, whip-like red streaks running along the variceal surface — dilated intra-epithelial or subepithelial venules coursing over the varix • Cherry-red spots: discrete, flat red spots a few millimeters in diameter • Hematocystic spots: raised, blood-blister-like lesions, thought to represent the most imminent rupture risk of the three
Any of these red signs — even on a small varix — reclassifies the patient into a higher bleeding-risk category, functionally overriding the reassurance that small size alone would otherwise provide. Red signs are graded qualitatively (absent, mild, moderate, severe) but clinically any unambiguous red sign is treated as a trigger for consideration of prophylaxis.
The North Italian Endoscopic Club (NIEC) index, derived from a large prospective multicenter cohort in the 1980s, was among the first validated tools to formally combine three independent predictors into a single bleeding-risk estimate:
1. Variceal size (small / medium / large) 2. Severity of red wale marking (absent / mild / moderate / severe) 3. Child-Pugh class (A / B / C) — capturing the severity of the underlying synthetic liver dysfunction and, indirectly, the degree of portal hypertension
Child-Pugh class C carries elevated bleeding risk essentially independent of variceal size, because advanced hepatic decompensation correlates with higher portal pressure, coagulopathy, and reduced physiologic reserve to tolerate a bleeding event. A patient with small varices but Child-Pugh class C, or with red signs on any size varix, is therefore managed as high-risk — the modern practical rule distilled from the NIEC framework is: large/medium varices of any Child-Pugh class, OR small varices with red signs, OR small varices with Child-Pugh C, all warrant primary prophylaxis.
Key Insight — risk is multiplicative, not purely size-based: a small varix with high-grade red wale marking in a Child-Pugh C patient can carry a bleeding risk comparable to, or exceeding, that of a large varix with no red signs in a well-compensated Child-Pugh A patient. Endoscopic reporting must capture all three NIEC variables, not size alone, to correctly triage patients toward prophylaxis.
Once a patient is identified as high-risk — large/medium varices of any grade, or small varices with red signs or Child-Pugh C — two evidence-based options exist to prevent the first variceal bleed: pharmacologic reduction of portal pressure with a non-selective beta-blocker, or mechanical obliteration of the varices with endoscopic variceal ligation. Neither is universally superior; the choice is individualized.
Non-selective beta-blockade (NSBB) works physiologically, upstream of the varix: it lowers portal venous inflow and therefore portal pressure itself, reducing wall tension across every collateral vessel simultaneously — esophageal, gastric, and elsewhere. Endoscopic variceal ligation (EVL) works locally and mechanically: rubber bands are deployed around individual esophageal varices during serial endoscopy sessions, strangulating and eventually obliterating the treated columns. EVL does nothing to lower portal pressure and does not protect against bleeding from varices outside the reach of the endoscope (e.g., gastric or ectopic varices), whereas NSBB's systemic hemodynamic effect is agnostic to variceal location.
Multiple randomized trials and meta-analyses have found the two approaches broadly comparable in preventing first bleeding, with neither showing a clear, consistent mortality advantage over the other — which is why guideline panels frame the choice around patient-specific factors rather than a single preferred default.
Propranolol and nadolol are traditional non-selective beta-blockers, blocking both β1 and β2 adrenergic receptors. Carvedilol is also non-selective at the beta receptors but carries an additional, clinically important property: modest anti-alpha1-adrenergic activity, which produces direct vasodilation of the intrahepatic and portal-collateral vasculature. This combined beta-blockade-plus-alpha1-blockade mechanism produces a significantly greater reduction in HVPG than traditional NSBBs at equivalent tolerated doses in head-to-head hemodynamic studies.
Because of this superior portal-pressure-lowering effect, carvedilol has become an increasingly preferred first-line NSBB for primary prophylaxis in many contemporary pathways, particularly in patients without refractory ascites (where its more pronounced hypotensive tendency requires caution). Propranolol remains a well-validated, widely available, and appropriate first-line alternative, especially where carvedilol is unavailable, poorly tolerated, or the patient has significant ascites limiting its use.
EVL becomes the preferred initial strategy chiefly when NSBB therapy is contraindicated or not tolerated:
• Severe asthma or COPD, where non-selective beta-blockade risks clinically significant bronchospasm • High-grade atrioventricular block or severe bradyarrhythmia, worsened by beta-blockade • Baseline hypotension or a systolic blood pressure that cannot tolerate further pharmacologic lowering • Symptomatic intolerance during NSBB titration — fatigue, dyspnea, or hypotensive symptoms that persist despite dose adjustment • Patient preference, after being informed of the comparable efficacy, the need for repeated endoscopic sessions with EVL versus a daily oral medication with NSBB, and the small but real procedural risks of banding (post-banding ulceration, rare bleeding)
For small varices without red signs, the risk-benefit calculation shifts: evidence supporting NSBB to prevent variceal growth or first bleed in this lower-risk group is less robust than for large/medium varices, and EVL is not generally used at all for small varices. Careful observation with surveillance endoscopy remains a entirely reasonable strategy in this subgroup, with NSBB considered case-by-case.
Key Insight — modality choice is not purely a hemodynamic question. Guideline panels explicitly weigh patient adherence, comorbidities, ascites status, and access to endoscopic follow-up alongside pure efficacy data, because a therapy the patient cannot tolerate or attend for is not, in practice, an effective prophylactic strategy.
Selecting a non-selective beta-blocker is only the first step; the drug must then be titrated to a dose that meaningfully lowers portal pressure while remaining hemodynamically tolerable. Titration is guided pragmatically by resting heart rate and blood pressure, since routine repeat invasive HVPG measurement is not standard practice outside specialized centers or research protocols.
Non-selective beta-blockade lowers portal pressure through two simultaneous, complementary effects, both mediated by blocking receptors the drug does not distinguish between (hence "non-selective"):
• β1-receptor blockade (cardiac): reduces heart rate and myocardial contractility, lowering cardiac output. Because portal venous inflow is ultimately supplied by splanchnic arterial flow driven by cardiac output, a lower cardiac output modestly reduces total splanchnic blood flow.
• β2-receptor blockade (splanchnic vasculature): normally, β2 receptors on splanchnic arterioles mediate vasodilation. Blocking them leaves alpha-adrenergic vasoconstrictor tone unopposed, causing splanchnic arteriolar vasoconstriction. This is the dominant mechanism reducing portal venous inflow, and it is unique to non-selective agents — cardioselective (β1-only) beta-blockers lack this splanchnic effect and are not used for this indication.
The net result of reduced portal inflow (from both lower cardiac output and splanchnic vasoconstriction) is a fall in portal venous pressure and, correspondingly, in the hepatic venous pressure gradient (HVPG) — the pressure actually responsible for driving blood into, and distending, esophageal varices. Carvedilol adds a third mechanism: alpha1-adrenergic blockade directly vasodilates the intrahepatic and portal-collateral circulation, lowering intrahepatic resistance itself rather than only splanchnic inflow, which is the biologic basis for its larger average HVPG reduction.
Titration follows a "start low, go slow" pattern common to beta-blocker initiation in any indication, adjusted every few days as tolerated:
Propranolol: start 20 mg twice daily; increase in increments every 2–3 days toward a maximum tolerated dose, traditionally cited up to approximately 320 mg/day in patients without ascites, but capped lower — around 160 mg/day — in patients with ascites, who are more susceptible to hypotension and renal hypoperfusion from excessive beta-blockade-induced cardiac output reduction.
Carvedilol: start 6.25 mg once daily (sometimes twice daily); typically capped at a considerably lower ceiling, often cited around 12.5 mg/day, reflecting its more pronounced blood-pressure-lowering effect from added alpha1 blockade — pushing carvedilol doses higher than this risks clinically significant hypotension without proportionate additional HVPG benefit.
At each titration step, resting heart rate and blood pressure (and symptoms — fatigue, lightheadedness, dyspnea) are reassessed before further increases. Nadolol (once-daily, renally cleared) is an alternative traditional NSBB used similarly to propranolol in some practice settings.
Because routine repeat HVPG measurement is invasive, resource-intensive, and unavailable outside specialized hepatology/portal hypertension centers, dose titration is guided in everyday practice by two simple bedside parameters, used as a pragmatic surrogate for adequate splanchnic vasoconstriction and cardiac output reduction:
• Target resting heart rate of approximately 55–60 beats per minute, reflecting adequate β1-mediated chronotropic blockade • Systolic blood pressure maintained at or above approximately 90 mmHg, avoiding symptomatic hypotension
The dose is increased stepwise until either the heart rate target is reached, the maximum labeled/tolerated dose is achieved, or the patient develops limiting side effects (hypotension, bradycardia, fatigue, bronchospasm) — whichever comes first. This heart-rate-driven approach is an imperfect proxy for the true pharmacodynamic target (HVPG reduction), since heart-rate response and portal-pressure response do not always correlate perfectly across individual patients, but it remains the only practical, universally available titration tool.
Key Insight — heart rate is a surrogate, not the therapeutic target itself. A patient can reach a heart rate of 58 bpm with an inadequate fall in HVPG, or occasionally show a robust HVPG response at a heart rate somewhat above 60 bpm. Where available, invasive HVPG measurement before and after NSBB initiation remains the gold-standard way to confirm a true "hemodynamic response," but in routine practice, clinicians accept the heart-rate/blood-pressure endpoint as the pragmatic ceiling for titration.
Primary prophylaxis is not a one-time decision but an ongoing management relationship. Whether a patient is on NSBB therapy or undergoing serial endoscopic ligation, structured follow-up confirms that the chosen strategy is delivering its expected protective benefit and identifies patients who need escalation, a switch in modality, or closer surveillance.
For patients with high-risk varices left untreated, the risk of a first variceal hemorrhage is substantial: approximately 30% over roughly two years of follow-up in classic natural-history cohorts of large/medium or red-sign-bearing varices. Effective primary prophylaxis — whether NSBB or EVL — roughly halves this risk, bringing the 2-year first-bleeding probability down to approximately 15% in patients who receive and tolerate adequate therapy.
This is a clinically meaningful but incomplete protection: prophylaxis substantially lowers, but does not eliminate, bleeding risk, which is why ongoing monitoring, adherence support, and — where relevant — dose optimization or modality switching remain part of long-term care rather than a "set and forget" intervention.
Among patients treated with NSBB, the single best predictor of protection from bleeding (and, in some analyses, from other complications of portal hypertension such as ascites) is achieving a defined "hemodynamic response," measured by paired HVPG measurements before and after treatment initiation:
• Absolute HVPG falls to below 12 mmHg (the threshold below which variceal bleeding is rare, since 12 mmHg is the approximate pressure required to sustain variceal formation and rupture), OR • HVPG falls by at least 20% relative to the pre-treatment baseline, even if it remains above 12 mmHg
Patients meeting either criterion ("hemodynamic responders") have a markedly lower risk of first bleeding compared with non-responders on the same nominal drug and dose — underscoring that portal pressure reduction, not simply "being on a beta-blocker," is the operative protective mechanism.
Key Insight — because repeat HVPG catheterization is invasive and confined largely to specialized centers or clinical trials, most practicing clinicians never directly confirm hemodynamic response. Instead, achieving the heart-rate/blood-pressure titration endpoint described in Stage 4 is used as an imperfect but practical proxy, accepting that some nominal "responders" by heart-rate criteria are not true hemodynamic responders by HVPG criteria, and vice versa.
The two prophylactic strategies carry different surveillance requirements:
On NSBB: no routine repeat endoscopy is required purely to assess prophylaxis efficacy — the drug's effect is systemic and continuous as long as it is taken. Follow-up instead focuses on clinical tolerability (heart rate, blood pressure, symptoms) at intervals after initiation and dose changes, with reinforcement of adherence, since intermittent or discontinued NSBB use forfeits the protective effect and has been associated with rebound elevation of portal pressure.
On EVL: repeat banding sessions are performed approximately every 2–4 weeks until all esophageal varices are obliterated ("variceal eradication"), typically requiring 2–4 sessions in total. After eradication, surveillance endoscopy is performed periodically (commonly at 3–6 months, then annually) to detect and re-band any recurrent varices, since new collaterals can re-form over time as portal hypertension persists.
In either pathway, the underlying liver disease and portal hypertension are unchanged by local variceal treatment — ongoing management of the etiology of cirrhosis (e.g., alcohol cessation, antiviral therapy, weight loss in metabolic-associated liver disease) remains essential, since it is the only intervention capable of truly reducing portal pressure at its source over the long term, alongside NSBB therapy.