Fig. 11.1
Classification of gastric varices. Gastroesophageal varices: lesser curve GOV1 and fundal GOV2. Isolated gastric varices; fundal IGV1 and other IGV2. Reproduced with permission from John Wiley and Sons from [4]
Isolated fundal gastric varices, without oesophageal varices, typically occur in patients with splenic vein thrombosis [6]. The condition is termed sinistral or left-sided portal hypertension and can easily be missed by the unwary endoscopist. It usually occurs in patients with pancreatitis or pancreatic carcinoma. It is the one cause of portal hypertension and variceal bleeding which can be completely cured by surgery, i.e. splenectomy. In comparison with oesophageal varices, standard endoscopic techniques such as injection sclerotherapy and/or banding are less effective for gastric varices . More effective treatments include TIPS shunt, glue injection (cyanoacrylate) and balloon-occluded retrograde transvenous occlusion of varices (BRTO) . Cyanoacrylate glue injection and TIPS are mainstays of treatment in Western countries, whereas BRTO is widely used in the Far East.
Glue Injection
The advantage of using cyanoacrylate is that it does not require any special equipment and it can be done at the time of the initial diagnostic endoscopy. The main disadvantages are the risks of non-target embolism and glue damage to the endoscope. The risk of embolism is reasonably low. One large Chinese series included 635 patients with gastric varices [7]. Ectopic embolism was reported in five cases (0.8 %). These included three splenic infarcts, one small pulmonary infarct and a cerebral embolism causing transient paralysis which resolved after 5 days. Cyanoacrylate may damage the endoscope but this risk can be minimised by careful technique. The technique of cyanoacrylate injection is not standardised [8, 9]. Most centres use a mixture of cyanoacrylate and lipiodol. Typically, the injection needle and catheter are flushed with lipiodol or aterile water beforehand to prevent glue occlusion in the catheter lumen. Cyanoacrylate and lipiodol are mixed before injection. In the literature, various ratios range from 0.5 to 1.5 mls lipiodol per 0.5 ml cyanoacrylate. The volume injected per varix also varies from 1 to 2 mls. Some centres have extensive experience with this technique and use a form of cyanoacrylate which does not require mixing with lipiodol. Good visualisation is important when injecting glue. In active gastric variceal bleeding, the fundus may be obscured by blood or clot. In cases of active bleeding, insertion of a Sengstaken–Blakemore or Linton–Nachlas tube may stabilise the situation, allowing subsequent targeted injection with a clear visual field. To avoid damage to endoscopic equipment, some endoscopists now use recombinant human thrombin instead of cyanoacrylate and report good results [10]. Endoscopic ultrasound guided coil embolization of gastric varices has recently been described. This technique is less likely to result in non-target embolization but requires significant technical expertise [11].
Transjugular Intrahepatic Portosytemic Shunt
TIPS is a well-established treatment option in patients with portal hypertension. It is particularly useful as a salvage therapy for active bleeding not controlled by endoscopic therapy [12]. It is also useful for patients who rebleed despite endoscopic or pharmacological therapy. Portal vein thrombosis is a relative contra-indication. The major disadvantages are the risks of worsening liver failure in patients with high model for end-stage liver disease (MELD) scores (> 24) and the longer term risks of hepatic encephalopathy. Nevertheless, early TIPS has been shown to improve prognosis in patients with oesophageal variceal bleeding and Pugh’s scores ≤ 13 [13]. It should probably be considered early if glue injection fails.
Balloon Occluded Retrograde Transvenous Occlusion of Varices [14] (Fig. 11.2)
Fig. 11.2
Balloon occluded retrograde transvenous occlusion of varices (BRTO). A balloon occlusion catheter is passed from the inferior vena cava and the renal vein into a spleno-renal collateral vessel and sclerosant material injected. GV gastric varices. (Reproduced with permission from Elsevier from [15])
BRTO is an alternative to TIPS shunt. The initial descriptions from Japan described injection of ethanolamine oleate via gastrorenal collaterals. Ethanolamine may cause significant haemolysis and Japanese physicians use haptoglobin infusions to deal with this complication. Unfortunately, haptoglobin is not available in the West, which may explain why the technique was not widely adopted in Europe and America. Interventional radiologists in the USA now use a frothy concoction of one part lipiodol, five parts 3 % Sotradecol and two parts air/CO2. The collaterals can be accessed via the spleno–renal route or percutaneously via the portal and splenic veins (as with a TIPS shunt) [15]. The transportal route can be used as an adjunct to TIPS insertion, if required. The main advantage of BRTO is that it can be used in patients with high MELD scores and poor liver function who are not suitable for TIPS. In addition, it can be used in patients with bleeding gastric varices and a history of hepatic encephalopathy, which is a relative contra-indication to TIPS. It can also be used to occlude large portosystemic shunts in patients with disabling hepatic encephalopathy who are unsuitable for liver transplantation. The main disadvantages of the technique are the risk of non-target embolization and an increase in portal venous pressure. This may result in exacerbation of varices elsewhere or precipitate ascites formation. In a large study including 183 patients, technical success was achieved in 97 % with procedure related complications in 4.4 % [16]. These included five cases of pulmonary thromboembolism, one renal infarction, one ruptured gastro-renal shunt and one case of transient mental changes. In patients without oesophageal varices, new oesophageal varices appeared in 21/36 (58 %).
Comparative Studies
Compared to oesophageal varices, there are relatively few randomised controlled trials in patients with gastric varices and the available trials are relatively small. Lo et al. randomised 60 patients with bleeding gastric varices to treatment with either band ligation or cyanoacrylate. The cyanoacrylate group required less blood transfusion (4.2 vs. 2.6 units p < 0.01) and had fewer rebleeding episodes (54 % vs. 31 % p < 0.01) [17]. Mishra et al. randomised 67 patients presenting with gastric variceal bleeding to secondary prophylaxis with either nonselective β-blockade or cyanoacrylate injection [18]. Patients with active bleeding at the index endoscopy underwent one cyanoacrylate injection prior to randomisation. Patients randomised to cyanoacrylate had endoscopic injection 6 days after the index bleed. All visible gastric varices were injected and successful obliteration confirmed by palpating the varix with the needle hub. Just over half the patients required a repeat session 7 days later to confirm obliteration. Both rebleeding rates (15 % vs. 55 %) and mortality (3 % vs. 25 %) were significantly lower in the cyanoacrylate group. Mishra et al. also performed a controlled trial of primary prophylaxis controlled trial comparing cyanoacrylate injection, propranolol and no treatment in 89 patients [19]. Interestingly, the hepatic venous pressure gradient increased in both the injection and no treatment groups but fell in the propranolol group. Over a median follow-up of 24 months, bleeding was significantly less common in the cyanoacrylate group compared to either propranolol or no treatment (13, 28 and 45 %, respectively). Survival was significantly higher in the cyanoacrylate group compared to the no treatment group (90 % vs. 72 %; p = 0.48).
In terms of preventing rebleeding, TIPS is probably more effective than glue injection but is more invasive and expensive. In patients who had bled from gastric varices, Lo et al. randomised 35 to TIPS and 37 to cyanoacrylate injection [9]. TIPS insertion was successful in all patients. Rebleeding from gastric varices occurred in 4 patients in the TIPS group and 14 patients in the cyanoacrylate group (p < 0.05). Survival rates were similar. Sabri et al. reported a retrospective analysis on 50 patients treated with either TIPS or BRTO for bleeding gastric varices [20]. Technical success rates were 100 % for TIPS and 91 % for BRTO with 12-month rebleeding rates of 11 and 0 %, respectively.
Ectopic Varices
Prevalence of Ectopic Varices
Ectopic varices are varices which can appear anywhere in the gastrointestinal tract outside of the usual sites, i.e. gastro-oesophageal and ano-rectal [21]. Bleeding from ectopic varices is relatively uncommon representing 2–5 % of upper gastrointestinal haemorrhages [22]. Ectopic varices usually occur at or near sites of anatomical disruption or following venous impairment or thrombosis . Typically, they occur near sites of surgery, around stomas or areas of previous inflammation, e.g. pancreatitis. There is a paucity of data on the true prevalence of ectopic varices. The Japanese Society for Portal Hypertension performed a survey of their members for the years 2001 to 2005 [23]. Thirty-three institutions replied reporting a total of 173 cases. There were 77 rectal, 57 duodenal, 11 small intestinal, 10 anastomotic, 7 colonic and 8 biliary tract varices and 1 diaphragmatic varix. Eighty percent had cirrhosis and 58 % had received previous treatment for oesophageal varices. Haemorrhage from ectopic varices occurred in 78/173 (45 %) with the most common sites being rectal (30 cases) and duodenal (27 cases).
In terms of treatment, it is important to establish whether portal venous drainage of the site is intact [21]. If drainage is intact TIPS is an option. One study described 24 patients with bleeding from ectopic varices treated with TIPS [24]. Sites of bleeding included stomal [8], ileocolic [6], duodenum [5], anorectal [3], umbilical [1] and peritoneal [1]. Alcoholic cirrhosis was the commonest etiology [13], 12 had Pugh’s class B liver disease and 7 Pugh’s class C. No embolization was performed at the initial procedure. Cumulative variceal rebleeding rate was 23 % with overall survival of 80 % at 1 year. BRTO may be an option even in patients with portal vein thrombosis providing it is possible to access the appropriate collateral vessels. Local therapies, e.g. injection or banding are frequently used. Surgery may be appropriate in some cases.
Rectal Varices
Bleeding from rectal varices can be challenging, both diagnostically and therapeutically. Poorly targeted sclerotherapy or banding in the ano-rectum can cause troublesome ulceration and recurrent bleeding. Endoscopic ultrasound may be helpful in delineating rectal varices, which are not always easily apparent at endoscopy [25]. Some experts suggest that endoscopic bands should be placed at the highest point of inflow through perforating veins. This is analogous to the situation with oesophageal varices where banding or sclerotherapy are most effective when applied in the distal oesophagus where the perforating veins occur. TIPS shunt is another option although TIPS may be less effective for bleeding remote from the central portal venous system. TIPS combined with transvenous embolization may be useful. In one study of 12 patients with bleeding rectal varices [22], TIPS was successfully inserted in 11/12. One patient had uncontrolled bleeding despite TIPS; another rebled despite a patent TIPS shunt. Successful treatment with endoscopic sclerotherapy (5 % ethanolamine), BRTO and endoscopic ultrasound guided cyanoacrylate injection with coiling have all been described [26–28]. Surgery may also have a role. Kaul and Skaife describe a surgical rectal stapling technique used successfully in nine patients [29]. The technique is similar to stapling haemorrhoidectomy although the purse-string suture has to run beneath all the visible varices. After the device is fired, individual bleeding points are identified and sutured. This procedure should probably only be done by surgeons experienced in the technique.
Duodenal Varices (Fig. 11.3)
Fig. 11.3
Duodenal varix: recent bleed
Bleeding from duodenal varices is uncommon and may be difficult to diagnose. The varix may be collapsed at the time of diagnostic endoscopy and/or be obscured by bleeding. In a Japanese experience of 57 cases, 2 were in the bulb, 47 in the descending part and 8 in the third part [23]. Duodenal varices may occur in the absence of cirrhosis or other oesophago-gastric varices. Treatment options include banding, injection of cyano-acrylate, TIPS, BRTO and surgery. A review of the literature revealed 19 cases treated with banding [30]. Rebleeding occurred in 3/19 after banding. Two patients died of liver failure within 7 days. Two patients required surgery: one for recurrent bleeding and one for a duodenal perforation secondary to injection sclerotherapy for recurrent bleeding. There is a reported case of occlusion of the ampulla of Vater by banding, underlining the importance of identifying landmarks in this area [31]. Injection of cyanoacrylate can also be effective. Liu et al. described four patients with duodenal varices treated with cyanoacrylate and Mora-Soler described a further five [32, 33]. In the Spanish series, two patients rebled and three died during the initial hospital admission (one from active bleeding and two from liver failure/sepsis). In a Chinese series, there was no rebleeding. Two of the four patients died at 7 and 24 months of liver failure and sepsis, respectively. Cyanoacrylate injection can also cause biliary obstruction [34]. Kochar et al. described four patients treated with TIPS for bleeding duodenal varices [22]. Bleeding was controlled in 3/4. Tanaka et al. reviewed 12 cases of BRTO for bleeding duodenal varices [35]. Bleeding was controlled in all cases. There were two cases of new oesophageal varices but no reports of rebleeding.