A great deal of research has been performed on cirrhotics to determine what interventions can be done to prevent the development of varices in those who have no varices (pre primary prophylaxis), prevent bleeding from varices in those who have varices that have never bled (primary prophylaxis), and to prevent re-bleeding in those who have a history of variceal bleeding (secondary prophylaxis).

Cirrhotics commonly develop portal hypertension as part of their disease and thus, are at risk for development of esophageal varices. If there was a safe and cost-effective intervention to significantly reduce the risk of esophageal variceal bleeding, it could be applied to all cirrhotics and endoscopy could be forgone, but such a treatment is not currently available. Endoscopy is a safe and effective method of screening cirrhotic patients for varices and can be done with moderate IV sedation unless the patient has significant cardiopulmonary disease, encephalopathy, or high tolerance to sedatives due to alcoholism or medications such as opioids and benzodiazpeines, in which case they should be sedated with monitored anesthesia. The patient is risk stratified based on the severity of cirrhosis and endoscopic exam to guide management of their varices.

In cirrhotics that do not have varices on endoscopy, the question arises as to how often they should be screened for development of varices, but there is no clear recommendation for screening interval. A prospective study 321 cirrhotics without history of bleeding were followed for a median of 2 years during which time 26.5 % had variceal bleeding . The three independent risk factors for bleeding were presence of red wales, large-sized varices, and advanced Child’s class. Of the etiologies of cirrhosis, alcohol was most associated with bleeding. Thus consensus guidelines recommend screening cirrhotics who have no or small varices every 2–3 years, but yearly if they have Child’s class C cirrhosis, alcoholic cirrhosis, or small varices with red wale sign [13, 16].

The class of medications most studied and found to be most useful in preventing variceal bleeding is nonselective beta-blockers (NNSB) , such as propranolol, timolol, and nadolol. The effect of NNSB on varices has been studied due to their effect of blocking the increased cardiac output effect of beta-1 adrenergic receptors and the splanchnic venous dilation effect of beta-2 adrenergic receptors thus reducing blood flow through varices [17]. When researching the effect of any medication or treatment for variceal bleeding prophylaxis, a physiologic marker used has been HVPG which is the accepted surrogate of portal pressure, and as stated earlier, effective treatment should result in a drop in HVPG to less than 12 mmHg or a reduction from baseline HVPG by at least 20 %. Due to HVPG being relatively invasive, some studies use the simple physiologic marker of a change in baseline heart rate to signify an adequate dose of NSBB based on an early study showing cirrhotics that were given propranolol sufficient to reduce baseline heart rate by 25 % resulted in a persistent reduction in portal , venous pressure [18]. Unfortunately, it has also been shown that 15 % of patients have at least relative contraindications to NSBB such as insulin dependent diabetes or emphysema, and 15 % may develop side effects that prevent reaching efficacious doses. Furthermore, there are recent studies showing a deleterious effect of NSBB on cirrhotics with refractory ascites in terms of mortality, perhaps due to the already compromised cardiac output of these patients, but large randomized controlled trials are lacking [17].

There has been interest in therapies to prevent the development of varices in cirrhotics who have no varices on screening endoscopy. This is based on the logic that if the formation of varices can be prevented, then variceal bleeding is not possible.

A study of 213 patients with minimal portal hypertension based on HVPG of 6 mmHg and no varices at baseline endoscopy were randomized to timolol or placebo to assess for development of varices. Over a median time of 4.5 years, approximately 40 % of patients developed varices in both groups but serious adverse events were more common in the timolol group (18–6 %) and this was statistically significant [19]. Similarly, another randomized double blind trial found no reduction in the rate of variceal formation when comparing cirrhotics on propranolol vs. placebo, although a third of patients were lost to follow-up [20]. Thus attempting to prevent the formation of varices using beta-blockers for patients with portal hypertension is not recommended. In contrast, a 12 year study of 218 compensated HCV cirrhotics found that none of the 34 patients who achieved sustained virologic response (SVR) developed esophageal varices, compared with 22/69 (31.8 %) untreated cirrhotics and 45/115 (39.1 %) non-SVR cirrhotics [21]. Therefore, the only recommended intervention to prevent development of varices in cirrhotics that do not have varices is to treat the underlying liver disease to whatever extent possible.

In contrast to pre-primary prophylaxis, medical intervention has had success in preventing bleeding in select patients with cirrhosis who have varices (primary prophylaxis). Many treatments have been studied, but the two most effective treatments have been found to be NNSB and endoscopic band ligation (Fig. 34.2).

A randomized controlled trial of 161 cirrhotics randomized to nadolol vs. placebo for a median period of 3 years, showed the rate of variceal growth from small to large was 37.2 % in the placebo group but only 10.8 % in the nadolol group , although there was no mortality difference and more adverse events in the nadolol group [22]. However, two more recent randomized controlled trials did not show positive results in cirrhotics with small varices [20, 23]. In one of the studies, 150 cirrhotics with small varices defined as <5 mm in diameter, were followed for 2 years and the growth to large varices was similar: 11 % in the propranolol group and 16 % in the placebo group (p = 0.79). This study also evaluated the role of HVPG measurements in these patients and found that a baseline HVPG could not reliably predict which patients would respond to NSBB, nor did a significant drop in HVPG due to NSBB therapy correlate with prevention of variceal growth [23]. A recent meta-analysis of six randomized controlled trials of cirrhotics without history of GI bleeding with either no or small varices that were treated with NSBB vs. placebo found no significant difference in the development of large varices, prevention of first variceal bleeding, or mortality, and even a subgroup analysis showed no differences between patients with no varices or small varices. However, there was a significant increase in adverse events in the NSBB group (OR 3.47; 95 % CI 0.08–3.70) [24].

The results are much more favorable for NSBB used in primary prophylaxis in cirrhotics with large varices. In such patients, nadolol was shown to significantly reduce the rate of first bleeding when compared to controls in a prospective, randomized trial of 79 patients followed for 2 years [25]. A multicenter, randomized trial comparing propranolol to placebo in preventing first bleeding from large varices included 174 patients, and in 3.5 years, 26 % in the propranolol group had bled while 41 % in the placebo group bled, although this was not statistically significant [26]. A larger, similar multicenter, randomized study of 230 Child’s C cirrhotics with large varices over 2 years showed that 26 % in the propranolol group bled, as in the previous study, but 61 % in the placebo group bled, and the survival difference was also statistically significant at 72 % in the propranolol group vs. 51 % in the placebo group [27]. A meta-analysis of 589 cirrhotics further supported the efficacy of both propranolol and nadolol in primary prophylaxis of variceal bleeding and improved mortality compared to placebo in high-risk patients [28]. Using a Markov model, propranolol for primary prophylaxis of variceal bleeding was found to be cost effective compared to placebo over a 5 year period [29].

There is emerging concern that although NSBB are effective in primary prophylaxis for variceal bleedin g, they may induce a poor outcome in the subset of cirrhotics with refractory ascites. It is thought that these patients have low systemic blood pressure that is more sensitive to the drop in cardiac output induced by beta blockade. A prospective nonrandomized study of 151 cirrhotics with refractory ascites compared the outcomes of those on propranolol versus those not on beta-blockers with a median follow-up of 8 months and they found a significant difference in 1 year morality of 19 % vs. 29 % in favor of those who did not receive propranolol [30]. There was some criticism regarding lack of HVPG analysis of patients and perhaps more severe liver disease in the propranolol group so more studies would be helpful [17].

Other treatments for primary prophylaxis have not been found to be effective, either on their own or in combination with NSBB. Based on the physiological finding of greater reduction in portal pressure with the combination of nitrates and beta-blockers compared to beta-blockers alone, it was theorized that this combination may also be effective in primary prophylaxis of variceal bleeding. However, a double blind randomized control trial of 349 cirrhotics that received a combination of propranolol and isosorbide mononitrate or propranolol with placebo showed nearly identical rates of variceal bleeding and mortality, while the nitrate group experienced significant more headaches [31].

Endoscopic sclerotherapy , where a sclerosant is directly injected into the varix to cause its obliteration by inducing fibrosis, has been found to be effective in controlling actively bleeding esophageal varices. However, in primary prophylaxis, sclerotherapy has not been effective in reducing the rate of variceal bleeding and its complication rate is as high as 50 % mainly due to chest pain [32]. Sclerotherapy is not recommended for prophylaxis of varices.

Endoscopic band ligation has been extensively studied for primary prophylaxis and is clearly effective in cirrhotics with high-risk varices (Fig. 34.3). An early randomized controlled trial of 68 cirrhotics with high-risk varices showed significant reduction in first bleed (8.6 % vs. 39.4 %) during the mean follow-up period of just over a year [33]. A meta-analysis of band ligation compared to placebo for primary prophylaxis consisting of 601 cirrhotics showed a relative risk reduction of first bleed of 64 % and risk reduction in mortality from upper GI bleeding of 80 %. The same study showed band ligation was more effective than propranolol for primary prophylaxis of variceal bleeding with a relative risk reduction of 52 % but there was no significant difference in bleeding related mortality or overall mortality between these two treatments [34].

A prospective randomized control study from India for primary prophylaxis of cirrhotics with medium or large (at least 3 mm diameter) esophageal varices was conducted comparing band ligation (41 patients) to propranolol (41 patients). They found the probability of variceal bleeding was 43 % in the propranolol group and 17 % in the ligation group over 18 months of follow-up, and this was statistically significant [35]. Another randomized control trial of 62 cirrhotics with high-risk varices were followed for a mean of 15 months and was terminated early due to an interim analysis showing a significant failure rate for propranolol with four deaths from variceal bleeding out of 31 patients compared to no deaths in the ligation group [36]. A meta-analysis of 16 randomized control trials of cirrhotics with high-risk varices, looked into this issue of primary prophylaxis, but focused on three trials with careful randomization to minimize selection bias, which the authors suspected were influencing the outcomes of many trials. In these three trials, they found no difference between band ligation and NSBB in mortality or rates of variceal bleeding. About half of the patients in the band ligation group developed superficial ulcers while 12 % of the NSBB group had to stop therapy and 20 % had to reduce dosage due to adverse effects [37].

The problem with the band ligation strategy is that it is costly, there is some risk of sedation and endoscopy, and the patient has to be compliant since it often requires multiple sessions to obliterate the varices. The banding itself results in sloughing of the mucosa leaving a shallow ulcer that tends to heal within 2 weeks; however the ulceration can potentially be deeper resulting in significant bleeding in 2–5 % of patients. A randomized, double blind, placebo-controlled trial of proton pump inhibitor therapy for 10 days after elective variceal banding followed by repeat endoscopy to assess for varices and ulcers showed that both groups had similar number of postbanding ulcers but the size of the ulcers in the placebo group was twice as large, and there were three patients with postband ligation bleeding in the placebo group but none in the pantoprazole group [38]. Therefore, PPI therapy after banding is recommended.

Since the two most effective treatments for primary prophylaxis of esophageal variceal bleeding have been found to be band ligation and NSBB, it was thought the combination of the two may be even more effective, and several trials have studied this possibility. A small single blind randomized controlled trial of 66 cirrhotics with medium to large varices were treated with band ligation alone or band ligation and propranolol until eradication of varices and then were followed up for a mean of 12 months to evaluate for variceal bleeding and recurrence of varices. All 66 patients achieved eradication of varices with no difference in time or number of endoscopic sessions needed to eradicate the varices. Only two patients experienced adverse effects from beta-blockers. There was no difference in bleeding with only a total of three patients who bled during the study of which only one was from variceal bleeding, and there was also no difference in mortality . Recurrence of varices, however, favored the combination group, 9–38 % with p = 0.003 [39]. A similarly designed randomized control trial of 144 patients was conducted requiring a mean of 3.3 endoscopic sessions over 2 months to achieve variceal eradication. Bleeding occurred evenly in both groups and all were from esophageal varices: 5 in combination group and 6 in the band ligation group with two bleeding deaths in the band ligation group. This study also found recurrence of varices favored the combination group, 19–33 % (p = 0.03). In the combination group, six patients discontinued propranolol due to side effects [40]. These studies show a modest benefit of combination therapy in select patients, but currently combination therapy of endoscopic band ligation and NSBB is not recommended for primary prophylaxis of esophageal varices.

More recently, an increased interest has developed in carvedilol, a NSBB that also has an alpha1-adrenergic receptor blocking effect that reduces vascular resistance causing both a reduction in both portal pressure and systemic pressure, and this effect seems to be stronger than propranolol [41]. A study of 104 patients was conducted to evaluate the efficacy of carvedilol in propranolol nonresponders. It included cirrhotics with esophageal varices without prior bleeding who were given propranolol as primary prophylaxis with a desired dose of 80–160 mg/day in order to achieve efficacy, which was evaluated after 4 weeks on the maximal tolerated dose by obtaining an HVPG measurement. If the HVPG was <12 mmHg or dropped by 20 %, the patient was deemed a propranolol responder and continued on that dose. If these parameters were not reached, the patient was switched to carvedilol, and the HVPG was again measured after 4 weeks on a stable dose, and continued if adequate drop in HVPG was reached. Of the 104 patients started on propranolol, only 37 (35.6 %) were propranolol responders with adequate HVPG response. Of the remaining 67 switched to carvedilol, 38 (56.7 %) showed adequate HVPG response, suggesting that carvedilol may be better tolerated and more efficacious than propranolol for primary prophylaxis [42]. Carvedilol has also been compared to band ligation for primary prophylaxis of esophageal variceal bleeding in cirrhotics. One such multicenter randomized control trial in Scotland of cirrhotics with medium or large varices followed for 2 years included 77 patients in the carvedilol group with a goal dose of 12.5 mg daily, and 75 patients in the band ligation group, who underwent band ligation every 2 weeks until eradication. Eleven patients (14.2 %) in the carvedilol group had to discontinue due to adverse reactions, while 23 patients (30.7 %) were unable to complete the endoscopy protocol in the band ligation group and 58 % reached variceal eradication. The study found no difference in mortality : 35 % in the carvedilol group and 37 % in the band ligation group. However, incidence of variceal bleeding during the follow-up period favored carvedilol 10–23 % (p = 0.04) [41].

In summary, cirrhotics without varices or small varices without high-risk stigmata are considered low risk for bleeding do not require primary prophylaxis for variceal bleeding, just periodic screening endoscopy. Cirrhotics with large varices or with high-risk stigmata can be treated with band ligation or NSBB. If tolerated, NSBB are a reasonable first option. Band ligation is appropriate in the 15 % of patients with contraindications to NSBB and in the 15–20 % who are unable to reach therapeutic doses due to side effects. Band ligation may also be a better choice in patients with refractory ascites [43]. Carvedilol may be the most effective NSBB. Combination therapy may have some benefit in reducing recurrence of varices but currently there is insufficient data showing a reduction in variceal bleeding or mortality compared to band ligation or NSBB alone.

Over time, the severity of liver disease progresses at variable rates, and this can result in the development and progression of esophageal varices. Studies have shown that in cirrhotics without varices, varices develop at a rate of 5–8 % per year. Furthermore, patients with small varices progress to large varices at a rate of 5–20 % per year. Because varices are not stable over time and do not cause symptoms until they rupture, periodic endoscopy is necessary to monitor for change. There is no strong data to determine the optimal frequency of surveillance but guidelines have been developed to assist in management depending on endoscopic findings:

No varices: every 2–3 years

Small varices, no red wale: every 1–2 years

Small varices with red wale or alcoholic cirrhosis: every 1 year

If large varices are present, then primary prophylaxis should be initiated. If NSBB is chosen for primary prophylaxis, then endoscopic surveillance is not necessary. If band ligation is chosen, then yearly endoscopy after variceal obliteration is reasonable [44].

Due to the proximal location of esophageal varices in the GI tract along with the potentially compromised mental state in cirrhotics from hepatic encephalopathy and/or volume loss, variceal bleeding poses a significant aspiration risk which has been shown to be 2.4–3.3 % during endoscopy [47]. A retrospective study was performed to analyze whether all cirrhotics with variceal bleeding should be intubated. This study of 69 hospital admissions with confirmed active variceal bleeding within 12 h of admission all had less than stage II hepatic encephalopathy, no signs of alcohol or drug intoxication, normal chest X-rays, and no signs of respiratory distress or aspiration prior to admission. Of these 69 hospital admissions, 47 underwent elective intubation for the endoscopy while 22 did not, and their characteristics were similar except that the non-intubation group had a significantly higher Child’s Pugh score of 9.1–8.0. Nonetheless, it was the intubation group which had the worse outcome with aspiration pneumonia in 19–0 % and nine deaths compared to just one in the non-intubation group. The reason for this worse outcome with intubation is unclear but it was not a prospective trial so there may have been factors leading to physicians selecting intubation that did not reveal themselves in the study [48]. Intubation of patients with suspected variceal bleeding for airway protection is reasonable in any cirrhotic with altered mental status, but is not necessary for all cirrhotics.

There is an understandable reaction on the part of clinicians to overt gastrointestinal hemorrhage, especially hematemesis, to aggressively resuscitate the patient with IV fluids and blood products to reduce the risk of circulatory shock and symptomatic anemia. This assumption has been studied objectively to determine the optimal resuscitation strategy. A randomized controlled trial of 889 patients admitted with gastrointestinal bleeding were randomized to liberal transfusion in which blood was transfused when the hemoglobin level dropped to 9 g/dl, or restrictive transfusion in which blood was transfused only when the hemoglobin level dropped to 7 g/dl. The study period was 45 days and included 277 cirrhotics of which 190 were found to have esophageal variceal bleeding. There was a significant mortality difference among cirrhotics favoring the restrictive strategy regardless of Child’s class by more than a 2:1 margin, 12–22 % [49]. Similarly, aggressive volume replacement with IV fluids may cause rebound portal hypertension which can exacerbate variceal bleeding, so it has been suggested that a patient should be resuscitated such that the heart rate be kept less than 100 bpm with a systolic blood pressure of 90–100 mmHg [50].

It has been found that cirrhotics admitted with GI bleeding are at significantly increased risk of infection during hospitalization, and this is associated with both variceal re-bleeding and increased mortality. Several studies have shown better outcomes in those treated empirically with antibiotics. A meta-analysis evaluated five randomized controlled trials comparing antibiotics to no treatment in 534 cirrhotics hospitalized with upper GI bleeding. Four of the studies used fluoroquinolones as the antibiotic which was given for a median of 7 days. The results favored the antibiotic group for several outcomes: remaining infection-free 86–55 %, free of spontaneous bacterial peritonitis (SBP) and bacteremia 92–73 %, and short-term survival up to 14 days 85–76 % [51]. Out of concern that there is emerging fluoroquinolone resistance, a randomized controlled trial of decompensated cirrhotics admitted with GI bleeding compared the outcome of 63 patients receiving oral norfloxacin to 61 patients receiving IV ceftriaxone. Over the course of 10 days, the norfloxacin group had a higher rate of infection, 33 % vs. 11 %, reaching statistical significance [52]. The available data strongly supports short-term antibiotic prophylaxis for all cirrhotics admitted with upper GI bleeding to reduce infection, re-bleeding, and improve survival. Fluoroquinolones are acceptable but if the patient is already on fluoroquinolones or there is concern for resistance, IV ceftriaxone is a reasonable choice.

Like SBP, variceal bleeding can precipitate hepatic encephalopathy in cirrhotics. Patients with hepatic encephalopathy should obviously continue therapy with lactulose if they develop upper GI bleeding. A study was conducted to determine if cirrhotics without known hepatic encephalopathy would also benefit from lactulose during hospitalization for acute variceal bleeding. It was a prospective, randomized trial of 35 patients each in the lactulose group and placebo group. Randomization was performed at time of endoscopy and the endoscopist and patient were not blinded but the clinician performing twice daily assessment of encephalopathy during hospitalization remained blinded. They found that 14 (40 %) patients in the control group but only 5 (14 %) patients in the treatment group developed clinical evidence of encephalopathy (p = 0.03) with a median encephalopathy grade of two, and a median time to development of encephalopathy of 2 days. This study supports a low threshold for starting lactulose in any cirrhotic admitted with variceal bleeding, but more data is needed to make it the standard of care [53].

Medications that acutely reduce portal pressure and hepatic blood flow in patients with portal hypertension have been given at the time of acute variceal bleeding in an attempt to reduce severity of bleeding in order to improve outcomes [54]. Vasoactive drugs include anti diuretic hormone analogues such as vasopressin and terlipressin, and somatostain analogues such as octreotide. Somatostatin tends to be well tolerated but vasopressin is known to cause many serious side effects due to intense systemic vasoconstriction including mesenteric and peripheral ischemia, myocardial infarction, and arrhythmias in over a third of patients, and this has limited its use [50]. An early randomized trial comparing vasopressin to octreotide as an adjunct to endoscopy for acute variceal bleeding was published in 1992. It included 48 cirrhotics with active variceal bleeding randomized to continuous infusion of octreotide or vasopressin after endoscopy. Complete bleeding control was achieved in 63 % of the octreotide group and 46 % in the vasopressin group, although there was no difference in mortality after 42 days. Serious side effects were more frequent in the vasopressin group, 40–12 % [55]. A meta-analysis comparing somatostatin analogues in addition to endoscopy compared to endoscopy alone for acute variceal bleeding evaluated eight randomized trials with a total of 939 patients and found that combination therapy was associated with a significant improvement in early hemostasis but no difference in mortality or adverse events [56]. Thus, a somatostatin analogue is the preferred vasoactive drug for any cirrhotic hospitalized with suspicion of variceal bleeding and should be given as an IV infusion for 3–5 days if variceal bleeding is confirmed on endoscopy [16]. Vasoactive drugs are not, however, adequate as a replacement to endoscopy.