I. Pathophysiology

  1. Chronic increased pressure in portal vein can result in formation of varices in esophageal veins, and eventual bleeding
  2. Also produces arteriovenous anastomoses--caput medusa, esophageal vv., hemorrhoidal vv., short gastric v-to-splenic v.
  3. Causes:
    1. Prehepatic (most common in kids, best prognosis): portal vein thrombosis (e.g. in postpartum umbilical sepsis), "cavernous malformations"
    2. Hepatic: cirrhosis, schistosomiasis
    3. Post-hepatic: Budd-Chiari Sd

II. Assessing risk of bleeding from esophageal varices

  1. Independent predictors of esophageal varices in pts with cirrhosis: Splenomegaly, platelets < 88,000/mm3 (Am. J. Gastroent. 94:3285 and 94:3292, 1999--JW)
  2. Size of varices on esophagoscopy is correlated with risk
  3. Hepatic vein pressure gradient (hepatic vein wedge pressure - free hepatic vein pressure)
    1. Requires catheterization of hepatic vein through femoral or jugular entry
    2. > 5mm Hg = portal hypertension (> 10mm Hg = "clinically significant" portal hypertension)

III. Management of esophageal varices

  1. Primary prevention of varix formation
    1. Beta-blockers
      1. In a study in 213 pts with cirrhosis and portal hypertension (hepatic venous pressure gradient of 6mm Hg or more) but no esophageal varices on endoscopy randomized to Timolol vs. placebo; over median 55mo f/u, incidence of (development of esophageal varices or variceal bleeding) was not sig. diff. in the two groups (NEJM 353:2254, 2005--AFP)
  2. Primary prevention of variceal bleeding
    1. Beta-blockers
      1. In a randomized trial of nadolol vs. placebo in 161 pts with cirrhosis and small esophageal varices and no prior h/o variceal bleeding, with serial endoscopy, incidence of development of large varices over mean 36mo f/u was sig. lower in the nadolol group (20% vs. 51%), as was incidence of variceal bleeding (Gastroent. 127:476, 2004--abst)
    2. Organic nitrates--Not as effective as beta-blockers
      1. 118 pts with cirrhosis & esophageal varices but no previous bleeding randomized to propanolol (median 60mg/d) vs. isosorbide dinitrate (20mg TID). Over avg. 4y f/u, propranolol group had sig. less risk of initial bleeding (28% vs. 42%) and nonsig. less mortality in the propranolol group (Gastroent.113:1632, 1997)
    3. Surgical shunting
      1. Mesocaval
      2. Portocaval
      3. Transjugular intrahepatic portosystemic shunt (TIPS)-usually reserved for pts who fail secondary prevention for variceal bleeding with other treatments, e.g. oral beta-blockers.
    4. Endoscopic obliteration of varices
      1. Band ligation
        1. In a study in 140 pts with cirrhosis and esophageal varices and high-risk signs on endoscopy (e.g. "cherry-red spots") but no history of variceal hemorrhage, all of whom received nadolol (adjusted based on heart rate) and were randomized to endoscopic band ligation vs. no ligation, over median 26mo f/u, there was no sig. diff. in incidence of first variceal hemorrhage, upper GI bleeding, and death, though the band ligation recipients had sig. higher incidence of adverse events (Hepatol. 52:230, 2010-JW)
      2. Sclerotherapy (may be less effective than ligation)
    5. Comparisons of endoscopic tx vs. beta-blockers
      1. In a randomized trial in 172 pts with esophageal varices and no prior h/o bleeding, endoscopic ligation vs. propranolol titrated to 160mg/d vs. isosorbide-5-mononitrate titrated to 80mg/d, risk for bleeding was 6% w/ligation, 19% w/propranolol, and 28% w/isosorbide (all nonsig. differences); no sig. differences in overall mortality (Gastroent. 123:735, 2002--JW)
      2. In a meta-analysis of 9 studies involving 734 pts with esophageal varices and portal hypertension randomized to endoscopic band ligation vs. beta-blocker (propranolol or nadolol) therapy for primary prevention of variceal hemorrhage, EBL was associated with significantly lower incidence of first variceal hemorrhage (RR 0.63; NNT = 13); no sig. diff. in bleeding-related mortality or all-cause mortality observed.  However, EBL was associated with fatal banding-related bleeding in two patients (Eur. J. Gastroent. Hepatol. 19:835, 2007--JW)
  1. Secondary prevention of variceal bleeding
    1. Same treatment modalities used as for primary prevention (see above)
    2. Comparisons of endoscopic vs. medical tx
      1. In a randomized trial in 144 pts with cirrhosis hospitalized with bleeding esophageal varices randomized to nadolol + isosorbide dinitrate vs. endoscopic ligation, the former was ass'd with sig. lower risk of recurrent bleeding over median f/u 21mos (33% vs. 49%) and sig. lower risk of severe complications of tx (3% vs. 12%) (NEJM 345:647, 2001--JW)
      2. In a randomized trial of 121 pts with h/o variceal bleeding, endoscopic ligation vs. nadolol 40mg QD + isosorbide 20mg QD (doses titrated higher if tolerated) followed x median 25mos, rebleeding was sig. less likely in ligation group (20% vs. 42%), but ligation group had nonsig. higher total mortality (25% vs. 13%) (Gastroent. 123:728, 2002--JW)
      3. In a trial of 102 pts with cirrhosis and esophageal variceal bleeding (which had been brought under control) randomized to endoscopic band ligation (Q2wks until varices obliterated) vs. (propranolol 40mmg BID + isosorbide mononitrate after 8-12wks if needed to reduce hepatic vein pressure gradient), there were no sig. diffs between the groups in 1y incidence death or of rebleeding (Gastroent. 123:1013, 2002--JW)
    3. Comparisons of surgical vs. medical tx
      1. In a study of 91 pts with Child-Pugh class B or C cirrhosis s/p first variceal bleed randomized to TIPS vs. (propranolol + isosorbide mononitrate), TIPS ass'd with sig less rebleeding over avg. 15mo f/u (13% vs. 39%) but a sig. higher incidence of encephalopathy (38% vs. 14%); no diff. in survival or quality-of-life scores (Hepatology 35:386, 2002--JW)
      2. In a study in 63 pts with cirrhosis and initial episode of variceal hemorrhage, all on vasoactive drugs and prophylactic antibiotics, randomized to either (drug therapy with a nonselective beta-blocker + endoscopic band ligation) vs. TIPS, over 16mo f/u, TIPS recipients had sig. lower incidence of (rebleeding or failure to control bleeding) (1 pt vs. 14) (NEJM 362:2370, 2010-JW)
    4. Antibiotics for secondary prevention of variceal bleeding
      1. 120 pts with cirrhosis and prior variceal bleeding episodes controlled with endoscopic therapy randomized to ofloxacin 200mg IV Q12h x 2 then PO Q12h x 5d after endoscopic treatment vs. "on-demand" (only if clinically indicated); incidence of rebleeding over study period (not specified in summary of the study) was sig. lower in abx group, as was incidence of clinical bacterial infections (Hepatology 39:746, 2004--JW)
    5. Balloon-Occluded Retrograde Transvenous Obliteration
      1. A technique in which a radiographically-guided balloonb catheter is introduced into a gastric varix and a sclerosing agent is injected.
  1. Treatment of actively bleeding varices
  1. Endoscopic treatments
    1. Sclerotherapy
    2. Band ligation
    3. Local injection of fibrin glue (Lancet 350:692, 1997)
  2. Vasopressin IV (works as well as intra-arterial angio-guided admin.)
  3. Somatostatin
    1. Reduces pressure within varices
    2. Found, when used in combination with endoscopic tx (e.g. sclerotherapy), to be associated with sig. lower risk of rebleeding @ 5d (23% vs. 42% w/placebo) w/o sig. increased risk of major adverse events in a meta-analysis of 8 randomized trial (Hepatol. 35:609, 2002--JW)
    3. 5d IV infusion reduced % of pts found to have bleeding varices on EGD and improved sclerotherapy success rates but didn't change 6wk mortality in 206 pts with known cirrhosis or portal HTN presenting with UGI bleed (Lancet 350:1495, 1997--JW)
    4. BID SQ octreotide (a somatostatin analogue) c/w placebo x 6mos after sclerotherapy for variceal bleed in 32 pts with bx-proven cirrhosis was associated with sig. less rebleeding but no sig. decrease in mortality at 12mo f/u (BMJ 315:1338, 1997--JW)
  4. Vasopressin analogues
    1. Terlipressin-Not yet available in U.S. as of 2009.
  5. Combination of endoscopic therapy + pharmacotherapy
    1. In general, superior to either treatment alone
    2. In a study in 324 pts with cirrhosis and variceal hemorrhage randomized to terlipressin vs. octreotide (all underwent endoscopic band ligation and received fluid resuscitation and prophylactic antibiotics), in intent-to-treat analysis, achievement of initial control of hemorrhage was not sig. diff. between the two groups (Am. J. Gastroent. 104:617, 2009-JW)
  6. Baloon tamponade ("Sengstaken-Blakemore Tube")--for pre-op management
  7. Transjugular intrahepatic portosystemic shunts (TIPS) to prevent further bleeding (inconsistent results in trials comparing outcomes with TIPS vs. sclerotherapy to prevent rebleeding)
  8. Prophylaxis against bacterial infections
    1. Bacterial infections occur in 25-65% of pts with hepatic cirrhosis and GI bleeding
    2. Typically treated with a quinolone or a cephalosporin

"Porto-pulmonary HTN"

  1. Some pts with advanced liver disease and portal HTN will also develop pulmonary HTN.
  2. In contrast to PPH, tend to have high cardiac output and low SVR
  3. Etiology is unclear