- Email: [email protected]
Trans-Anastomotic Porto-Portal Varices in Patients with Gastrointestinal Haemorrhage
Clinical Radiology (2000) 55, 207–211 doi:10.1053/crad.1999.0347, available online at http://www.idealibrary.com on
Trans-Anastomotic Porto-Portal Varices in Patients with Gastrointestinal Haemorrhage A . W . M . MIT C HE L L* , J. E . J AC KSON The Departments of Imaging, Imperial College, Hammersmith Hospital, and *Charing Cross Hospital, Du Cane Road, London, W12 ONN, U.K. Received: 19 May 1999 Revised: 4 August 1999 Accepted: 15 August 1999 AIM: Porto-portal varices are commonly seen in patients with segmental extra-hepatic portal hypertension and develop to provide a collateral circulation around an area of portal venous obstruction. It is not well recognized that such communications may also develop across surgical anastomoses and be the source of gastrointestinal haemorrhage. The possible mode of development of such communications has not been previously discussed. MATERIALS AND METHODS: Over a 3-year period between 1995 and 1998, porto-portal varices were demonstrated across surgical anastomoses in four patients who were referred for the investigation of acute (two), acute-on-chronic (one) and chronic gastrointestinal bleeding (one). Their medical notes and the findings at angiography were reviewed. RESULTS: Three patients had segmental portal hypertension due to extra-hepatic portal vein (one) or superior mesenteric vein (two) stenosis/occlusion. One patient had mild portal hypertension due to hepatic fibrosis secondary to congenital biliary atresia. At angiography all patients were shown to have varices crossing previous surgical anastomoses. These varices were presumed to be the cause of bleeding in three of the four patients; the site of bleeding in the fourth individual was not determined. CONCLUSIONS: Trans-anastomotic porto-portal varices are rare. They develop in the presence of extra-hepatic portal hypertension and presumably arise within peri-anastomotic inflammatory tissue. Such varices may be difficult to manage and their prognosis is poor when bleeding occurs. Mitchell, A. W. M., Jackson, J. E. (2000). Clinical Radiology 55, 207–211. q 2000 The Royal College of Radiologists Key words: gastrointestinal tract, haemorrhage, angiography, portal vein, varices.
Gastrointestinal haemorrhage from portosystemic varices involving the lower oesophagus and the gastric fundus is common in patients with portal hypertension due to liver cirrhosis. Bleeding may also occur from so-called ‘ectopic varices’ at other well-described sites of portosystemic anastomosis such as the lower rectum and the retroperitoneum [1] or from neo-portosystemic anastomoses which develop at cutaneous enteral fistulae, e.g. ileostomies [2]. These varices all develop in order to decompress a high-pressure portal system into a lower pressure systemic system. Porto-portal varices, on the other hand, form a conduit through which an area of relatively high portal pressure drains into an area of lower portal pressure and these are commonly seen in patients with extra-hepatic portal hypertension. These varices are enlarged, pre-existing, collateral veins which bypass a main portal, splenic or superior mesenteric venous occlusion. It is not well recognized that portoportal varices may also develop across a surgical anastomosis. We report four patients undergoing visceral arteriography Author for correspondence and guarantor of study: Dr James Jackson. 0009-9260/00/030207+05 $35.00/0
during the investigation of gastrointestinal haemorrhage in whom trans-anastomotic porto-portal varices were present. These were felt to be the source of bleeding in three of the four patients in whom extensive investigations failed to demonstate any other significant abnormality.
MATERIALS AND METHODS
Between January 1995 and November 1998, 249 patients underwent visceral angiography to investigate gastrointestinal bleeding. From the reports of these angiograms, four patients were found in whom transanastomotic porto-portal varices were demonstrated at arteriography and the clinical notes and angiographic studies of these indivuals were reviewed. Followup was obtained from the clinical notes in three patients and by telephone in one. Three men and one woman with an age range of 11–70 years underwent visceral angiography for the investigation of acute (two), acute-on-chronic (one) and chronic (one) gastrointestinal haemorrhage. All individuals had a history of previous abdominal q 2000 The Royal College of Radiologists
208
CLINICAL RADIOLOGY
surgery which involved an ileo-colic anastomosis in two, an hepatico-jejunal anastomosis in one and a choledochocystjejunostomy in one.
Case 1 A 32-year-old man with a 21-year history of Crohn’s disease was referred for visceral angiography to investigate the cause of several acute episodes of gastrointestinal bleeding. He had no other illnesses and no features of liver disease. He had been managed medically for the first 17 years of his disease with intermittent courses of oral steroids and sulphasalazine or mesalazine. He then presented with a massive gastrointestinal haemorrhage requiring a laparotomy at which Crohn’s disease involving the entire small intestine was found. The caecum was inflamed and adherent to the posterior abdominal wall and a limited right hemicolectomy was performed together with resection of the distal 20 cm of ileum. Since this operation his disease had been difficult to control and for the 6 months prior to his referral for visceral angiography he had suffered six further episodes of severe gastrointestinal bleeding requiring transfusion of a total of 20 units of blood. Inferior mesenteric and coeliac axis arteriograms were normal. The superior mesenteric artery (SMA) angiogram demonstrated absence of the ileo-colic artery consistent with the history of a previous right hemicolectomy; the arterial phase was otherwise normal. During the venous phase, however, a tight stenosis of the superior mesenteric vein (SMV) was identified immediately below the first jejunal vein. Large jejunal varices were present in the upper small bowel bypassing this stenosis and, in addition, varices were seen at the site of the
ileo-colic anastomosis (Fig. 1). The latter drained via the right colic vein into the gastro-colic trunk which, in turn, drained into the upper portion of the SMV above the stenosis. No other source of gastrointestinal bleeding was identified and there were no porto-systemic varices. Treatment has been aimed at lowering portal venous pressure with propranolol. The patient remains well 18 months later with no further evidence of gastrointestinal blood loss.
Case 2 An 11-year-old male patient was referred for visceral angiography because of continued episodes of acute-on-chronic gastrointestinal blood loss. Within his first 4 weeks of life a Kasai procedure had been performed for non-correctable biliary atresia. Since the age of 2 years, the patient had been repeatedly admitted to his local hospital with increasingly severe episodes of haematemesis and melaena. His management had been complicated by the development of cirrhosis. Numerous investigations at the referring hospital, including angiography and a laparotomy, had failed to reveal the source of the haemorrhage. The investigations at the time of admission for angiography demonstrated mildly elevated bilirubin and liver enzymes, normal clotting screen, normal urea and electrolytes and albumen. An abdominal US showed a patent portal vein, splenic vein and SMV with no splenomegaly or ascites. The arterial phase of the SMA angiogram demonstrated no abnormality. The venous phase of the angiogram demonstrated patent superior mesenteric and main portal veins with retrograde filling of the gastric veins, indicative of portal hypertension, and of the jejunal vein within the bowel loop brought up to the liver hilum. Similar appearances were seen at a subsequent transhepatic portogram during which the portal pressure was measured at 20 mmHg (Fig. 2a). Selective catheterization of the jejunal arterial branch to the site of the hepatico-jejunostomy demonstrated normal arterial branches (Fig. 2b) but the venous phase of the study demonstrated a direct trans-anastomotic communication between the jejunal vein and the intrahepatic portal venous branches (Fig. 2c). Present treatment has been aimed at medical reduction of his portal hypertension but recurrent episodes of acute gastrointestinal haemorrhage remain a problem at 1 year follow-up.
Case 3
Fig. 1 – Patient 1. Venous phase image from a distal superior meseteric artery angiogram obtained using a digital subtraction technique. The main superior mesenteric vein is tightly stenosed (open curved arrow). Varices are seen at the ileo-colic anastomosis (short arrow). In addition large varices are present within the jejunal veins bypassing the SMV stenosis.
A 70-year-old man referred for investigation of acute severe gastrointestinal haemorrhage had undergone an open cholecystectomy for gallstones 25 years previously, from which he had made an uneventful recovery. Ten years following this he presented with an episode of cholangitis which resulted in further imaging and a diagnosis of choledochal cyst was made, for which he underwent a choledochocyst-duodenostomy. Five years later he developed further severe episodes of cholangitis and underwent a choledochocyst-jejunostomy (in addition to his choledochocyst-duodenostomy). He continued to suffer from repeated episodes of cholangitis following this and had been admitted on one occasion during the previous 5 years with anaemia.
209
PORTO-PORTAL VARICES IN GASTROINTESTINAL HAEMORRHAGE
(a)
(b)
(c)
Fig. 2 – (a) Patient 2. Image from a percutaneous, transhepatic portogram with the catheter placed within the superior mesenteric vein. There is retrograde filling of the left gastric vein (short arrow) and of a jejunal vein which courses up towards the liver hilum (curved arrow). (b) Arterial image from a selective jejunal arteriogram demonstrating the normal arterial branches supplying the jejunal loop which has been brought up to the liver hilum. (c) Venous phase from the arteriogram shown in Fig. 2(b) demonstrates that the jejunal vein within this loop of jejunum communicates across the hepatic-jejunal anastomosis with an intrahepatic portal venous branch.
He was admitted on this occasion to the referring hospital again because of severe anaemia (haemoglobin 5.2 gm/dl) and during the following 10 days he suffered from persistent gastrointestinal bleeding requiring a total of 32 units of blood. He underwent an angiogram which failed to demonstrate a source of bleeding and was subsequently referred to this institution for further investigation. An abdominal US showed main portal vein occlusion and an enlarged liver and spleen. The arterial phase of an SMA angiogram demonstrated a loop of jejunum pulled across the midline to the site of the choledochocyst-jejunostomy in the right upper quadrant of the abdomen. There was no evidence of active contrast medium extravasation. The venous phase of the SMA angiogram confirmed occlusion of the portal vein with associated portal varices and also demonstrated retrograde filling of a large jejunal vein to the site of the choledochocyst-jejunostomy at which point there were large varices
communicating with intrahepatic portal vein branches (Fig. 3a). The indirect spleno-portogram demonstrated a patent splenic vein and retrograde filling of the distal portion of the SMV with subsequent opacification of the same jejuno-portal varices. The patient continued to bleed despite maximum medical therapy and a distal splenorenal shunt (DSRS) was therefore performed. Early thrombosis of the DSRS occurred which necessitated the percutaneous insertion of a stent to restore shunt patency. The jejuno-portal varices were well demonstrated during stent insertion by direct catheterization of the superior mesenteric vein via the spleno-renal shunt (Fig. 3b). After stent insertion antegrade flow through the surgical shunt was restored, although variceal filling still persisted. Gastrointestinal bleeding continued although at a lower rate than previously and further massive haemorrhage did not occur. Unfortunately the patient subsequently died due to overwhelming sepsis and multi-organ failure.
210
CLINICAL RADIOLOGY
(a) Fig 4 – Patient 4. Venous phase from a distal superior mesenteric artery angiogram demonstrates distal ileal veins communicating across the ileo-transverse colic anastomosis (arrow) with the middle colic vein.
(b) Fig. 3 – (a) Patient 3. Venous phase from a superior mesenteric arteriogram demonstrates a markedly enlarged jejunal vein (arrow) coursing towards the liver hilum where there are numerous varices with faint filling of intrahepatic portal venous branches. (b). Image from a study in which the superior mesenteric vein has been catheterized from an internal jugular vein approach via a spleno-renal shunt. The large varices crossing the choledochocyst-jejunotomy between the jejunal vein and portal venous branches are well visualized.
Case 4 A 40-year-old woman was referred for assessment of an extensive carcinoma of the transverse colon invading the second and third parts of the duodenum. She had undergone a laparotomy at her local hospital at which an extended right hemicolectomy was performed. She was referred to this institution for further evaluation. An abdominal CT was performed
which demonstrated the involvement of the duodenum; liver metastases were not identified. An angiogram was performed as a ‘road-map’ and for an assessment of operability prior to a Whipple’s procedure. The coeliac axis angiogram was normal. The arterial phase of the SMA angiogram demonstrated changes consistent with a right hemicolectomy and was otherwise normal. The venous phase of the angiogram revealed involvement by tumour of several jejunal veins and the main trunk of the SMV which were narrowed. The main portal vein was patent. A Whipple’s procedure was subsequently performed as a debulking procedure and to prevent duodenal obstruction and from which the patient made an uneventful initial recovery. One month post-operatively, however, she became profoundly hypotensive and passed copious amounts of dark blood per rectum necessitating admission to the intensive care unit. Visceral angiography performed on the same day demonstrated no contrast medium extravasation; venous phase images from the SMA angiogram showed evidence of tumour progression with more extensive involvement of the main SMV. Venous return from the distal small bowel occurred via porto-portal varices which had developed across the ileo-colic anastomosis between the distal ileal veins and the middle colic vein which in turn drained above the SMV stenosis into the gastrocolic trunk (Fig. 4). Review of the pre-operative arteriogram showed that the ileo-colic varices had been present, although smaller, at this time. The patient died a few days later from continued gastrointestinal haemorrhage. DISCUSSION
Extrahepatic portal occlusion, whatever its cause, may result in the development of portosystemic varices in an attempt to
PORTO-PORTAL VARICES IN GASTROINTESTINAL HAEMORRHAGE
decompress the elevated portal venous pressure distal to the obstruction into the low-pressure systemic veins [1]. These varices represent enlarged but otherwise normal pre-existing anastomoses between the portal and the systemic venous circulations. There will be simultaneous enlargement of the normal veins in the portal venous system which bypass the occluded segment into other portal venous system [3] veins and these are usually the dominant, and only, varices in patients with extrahepatic portal venous occlusion who do not have accompanying hepatic portal hypertension. For example, in a patient with occlusion of the splenic vein due to a pancreatic neoplasm, the predominant venous return from the spleen will usually occur via the gastroepiploic and the epiploic arcades into the main portal vein and via the short gastric veins into the left and right gastric veins; porto-systemic (e.g. oesophageal) varices will often notably be absent. On the other hand, in a patient who has splenic venous occlusion due to chronic alcoholic pancreatitis, for example, who also has elevated main portal venous pressure due to alcoholic cirrhosis, the dominant varices are likely to be between the portal and the systemic circulation. Whilst portosystemic varices occur at sites of pre-existing normal portosystemic anastomoses, they may also occur at sites of previous surgery, though reports of these are relatively infrequent [2,4–7]. The most well-recognized of these are stomal varices which develop at the site of a colostomy or ileostomy in patients with portal hypertension between the colonic or ileal veins, respectively, and the anterior abdominal wall veins; such varices may be the cause of troublesome recurrent haemorrhage [8–10]. These varices are thought to develop within peri-anastomotic inflammatory tissue and must, therefore, involve neo-angiogenesis. These, and other extraoesophagogastric porto-systemic venous communications, have been termed ‘ectopic’ varices [6]. The development of porto-portal communications across surgical anastomoses, as seen in the four patients presented, is less well appreciated and has received very little attention in the literature [11,12]. Three of our patients (cases 1, 3 and 4) were similar in that they had extrahepatic portal venous obstruction without severe hepatic portal hypertension. Varices presumably developed in these individuals across their previous surgical anastomoses (ileo-colic in two and jejuno-portal in one) as these communications represented the most direct route from an area of high portal pressure to an area of normal portal pressure. Similar to the varices described above which develop across stomas, these communications presumably develop within peri-anastomotic scar tissue. The development, in case 2, of porto-portal varices across an hepatico-jejunal anastomosis is perhaps more difficult to explain. This patient had mild hepatic portal hypertension and no extrahepatic portal occlusion and yet still developed trans-anastomotic communications. The most likely explanation for this is that the intrahepatic portal venous branches with which the jejunal veins communicated were at lower pressure than the main portal vein due to the patchy nature of the patient’s liver cirrhosis. The mechanisms by which trans-anastomotic veno-venous
211
communications develop are not entirely clear. The development of varices across surgical anastomoses requires the formation of new vessels (angiogenesis) as there are no preexisting communications. It is well known that angiogenesis occurs during wound healing (as well as other processes). Although angiographically identifiable communications between the arterial and the venous circulations across surgical anastomoses are rarely (if ever) seen in patients without intra- or extrahepatic portal hypertension, it is likely that these develop in the majority of normal individuals as part of the normal healing processes [13]. If extrahepatic portal venous obstruction subsequently develops then these communications presumably enlarge to form trans-anastomotic varices. As with varices elsewhere, those crossing surgical enteric anastomoses are at risk of haemorrhage and their treatment may be very difficult as demonstrated in the four individuals presented. A lowering of portal pressure proximal to the extrahepatic portal occlusion (i.e. pre-occlusion) by surgical shunting may be expected to be useful and did temporarily improve matters in one of our patients (case 3). If the extrahepatic portal venous obstruction can be crossed with a catheter and guide wire, this might respond to angioplasty or stent insertion. This technique has been used by other workers in patients with small bowel varices secondary to SMV stenosis [3] and is a treatment option in one of our patients with a tight SMV stenosis (case 1) if bleeding recurs. REFERENCES 1 Sherlock S, Dooley J. Diseases of the Liver and Biliary System, 10th edn. Oxford: Blackwell Scientific Publications, 1997; chapter 10. 2 Ahn J, Cooper JM, Silberzweig JE, Mitty HA. Venographic appearance of portosystemic collateral pathways. Br J Radiol 1997;70:1302–1306. 3 Cohen GS, Ball DS, Flynn DE. Transjugular placement of a superior mesenteric vein stent for small bowel varices. J Vasc Intervent Radiol 1995;6:707–710. 4 Buhler L, Tamigneaux I, Giostra E, et al. Varices ectopiques, une case rare d’hemorrhage digestive. Schweiz-Med-Wochenschr-Suppl 1996; 79:70S–72S. 5 Cappell MS, Price JB. Characterization of the syndrome of small and large intestinal variceal bleeding. Dig Dis Sci 1987;32:422–427. 6 Lebrec D, Benahmou JP. Ectopic varices in portal hypertension. Clin Gastroenterol 1985;14:105–121. 7 Hamlyn AN, Lunzer MR, Morris JS, Puritz H, Dick R. Portal hypertension with varices at unusual sites. Lancet 1974;2:1531–1534. 8 Lashley DB, Saxon RR, Fuchs EF, Chin DH, Lowe BA. Bleeding ileal conduit stomal varices: diagnosis and management using trans jugular transhepatic angiography and embolization. Urology 1997;50:612– 614. 9 Wong RC, Berg CL. Portal hypertensive stomopathy: a newly described entity and its successful treatment by placement of a transjugular intrahepatic portosystemic shunt. Am J Gastroenterol 1997;92:1056– 1057. 10 Wolfson HC, Kozareck RA, Bredfeldt JE, Fenster LF, Brubacher LL. The role of endoscopic injection sclerotherapy in the management of peristomal varices. Gastrointest Endosc 1990;36:472–474. 11 Miller JT Jr, De Odorico I, Marx MV. Cholecystojejunostomy varices demonstrated by enteroclysis. Abdom Imag 1997;22:474–476. 12 Carpenter S, Brown KA. Chronic complications after cholecystojejunostomy. Am J Gastroenterol 1994;89:2073–2075. 13 Shikata J, Shida T. Experimental studies by the resin-casting method on the vascular structure of the colon following stapler anastomoses. Dis Colon Rectum 1985;28:341–346.
Trans-Anastomotic Porto-Portal Varices in Patients with Gastrointestinal Haemorrhage A . W . M . MIT C HE L L* , J. E . J AC KSON The Departments of Imaging, Imperial College, Hammersmith Hospital, and *Charing Cross Hospital, Du Cane Road, London, W12 ONN, U.K. Received: 19 May 1999 Revised: 4 August 1999 Accepted: 15 August 1999 AIM: Porto-portal varices are commonly seen in patients with segmental extra-hepatic portal hypertension and develop to provide a collateral circulation around an area of portal venous obstruction. It is not well recognized that such communications may also develop across surgical anastomoses and be the source of gastrointestinal haemorrhage. The possible mode of development of such communications has not been previously discussed. MATERIALS AND METHODS: Over a 3-year period between 1995 and 1998, porto-portal varices were demonstrated across surgical anastomoses in four patients who were referred for the investigation of acute (two), acute-on-chronic (one) and chronic gastrointestinal bleeding (one). Their medical notes and the findings at angiography were reviewed. RESULTS: Three patients had segmental portal hypertension due to extra-hepatic portal vein (one) or superior mesenteric vein (two) stenosis/occlusion. One patient had mild portal hypertension due to hepatic fibrosis secondary to congenital biliary atresia. At angiography all patients were shown to have varices crossing previous surgical anastomoses. These varices were presumed to be the cause of bleeding in three of the four patients; the site of bleeding in the fourth individual was not determined. CONCLUSIONS: Trans-anastomotic porto-portal varices are rare. They develop in the presence of extra-hepatic portal hypertension and presumably arise within peri-anastomotic inflammatory tissue. Such varices may be difficult to manage and their prognosis is poor when bleeding occurs. Mitchell, A. W. M., Jackson, J. E. (2000). Clinical Radiology 55, 207–211. q 2000 The Royal College of Radiologists Key words: gastrointestinal tract, haemorrhage, angiography, portal vein, varices.
Gastrointestinal haemorrhage from portosystemic varices involving the lower oesophagus and the gastric fundus is common in patients with portal hypertension due to liver cirrhosis. Bleeding may also occur from so-called ‘ectopic varices’ at other well-described sites of portosystemic anastomosis such as the lower rectum and the retroperitoneum [1] or from neo-portosystemic anastomoses which develop at cutaneous enteral fistulae, e.g. ileostomies [2]. These varices all develop in order to decompress a high-pressure portal system into a lower pressure systemic system. Porto-portal varices, on the other hand, form a conduit through which an area of relatively high portal pressure drains into an area of lower portal pressure and these are commonly seen in patients with extra-hepatic portal hypertension. These varices are enlarged, pre-existing, collateral veins which bypass a main portal, splenic or superior mesenteric venous occlusion. It is not well recognized that portoportal varices may also develop across a surgical anastomosis. We report four patients undergoing visceral arteriography Author for correspondence and guarantor of study: Dr James Jackson. 0009-9260/00/030207+05 $35.00/0
during the investigation of gastrointestinal haemorrhage in whom trans-anastomotic porto-portal varices were present. These were felt to be the source of bleeding in three of the four patients in whom extensive investigations failed to demonstate any other significant abnormality.
MATERIALS AND METHODS
Between January 1995 and November 1998, 249 patients underwent visceral angiography to investigate gastrointestinal bleeding. From the reports of these angiograms, four patients were found in whom transanastomotic porto-portal varices were demonstrated at arteriography and the clinical notes and angiographic studies of these indivuals were reviewed. Followup was obtained from the clinical notes in three patients and by telephone in one. Three men and one woman with an age range of 11–70 years underwent visceral angiography for the investigation of acute (two), acute-on-chronic (one) and chronic (one) gastrointestinal haemorrhage. All individuals had a history of previous abdominal q 2000 The Royal College of Radiologists
208
CLINICAL RADIOLOGY
surgery which involved an ileo-colic anastomosis in two, an hepatico-jejunal anastomosis in one and a choledochocystjejunostomy in one.
Case 1 A 32-year-old man with a 21-year history of Crohn’s disease was referred for visceral angiography to investigate the cause of several acute episodes of gastrointestinal bleeding. He had no other illnesses and no features of liver disease. He had been managed medically for the first 17 years of his disease with intermittent courses of oral steroids and sulphasalazine or mesalazine. He then presented with a massive gastrointestinal haemorrhage requiring a laparotomy at which Crohn’s disease involving the entire small intestine was found. The caecum was inflamed and adherent to the posterior abdominal wall and a limited right hemicolectomy was performed together with resection of the distal 20 cm of ileum. Since this operation his disease had been difficult to control and for the 6 months prior to his referral for visceral angiography he had suffered six further episodes of severe gastrointestinal bleeding requiring transfusion of a total of 20 units of blood. Inferior mesenteric and coeliac axis arteriograms were normal. The superior mesenteric artery (SMA) angiogram demonstrated absence of the ileo-colic artery consistent with the history of a previous right hemicolectomy; the arterial phase was otherwise normal. During the venous phase, however, a tight stenosis of the superior mesenteric vein (SMV) was identified immediately below the first jejunal vein. Large jejunal varices were present in the upper small bowel bypassing this stenosis and, in addition, varices were seen at the site of the
ileo-colic anastomosis (Fig. 1). The latter drained via the right colic vein into the gastro-colic trunk which, in turn, drained into the upper portion of the SMV above the stenosis. No other source of gastrointestinal bleeding was identified and there were no porto-systemic varices. Treatment has been aimed at lowering portal venous pressure with propranolol. The patient remains well 18 months later with no further evidence of gastrointestinal blood loss.
Case 2 An 11-year-old male patient was referred for visceral angiography because of continued episodes of acute-on-chronic gastrointestinal blood loss. Within his first 4 weeks of life a Kasai procedure had been performed for non-correctable biliary atresia. Since the age of 2 years, the patient had been repeatedly admitted to his local hospital with increasingly severe episodes of haematemesis and melaena. His management had been complicated by the development of cirrhosis. Numerous investigations at the referring hospital, including angiography and a laparotomy, had failed to reveal the source of the haemorrhage. The investigations at the time of admission for angiography demonstrated mildly elevated bilirubin and liver enzymes, normal clotting screen, normal urea and electrolytes and albumen. An abdominal US showed a patent portal vein, splenic vein and SMV with no splenomegaly or ascites. The arterial phase of the SMA angiogram demonstrated no abnormality. The venous phase of the angiogram demonstrated patent superior mesenteric and main portal veins with retrograde filling of the gastric veins, indicative of portal hypertension, and of the jejunal vein within the bowel loop brought up to the liver hilum. Similar appearances were seen at a subsequent transhepatic portogram during which the portal pressure was measured at 20 mmHg (Fig. 2a). Selective catheterization of the jejunal arterial branch to the site of the hepatico-jejunostomy demonstrated normal arterial branches (Fig. 2b) but the venous phase of the study demonstrated a direct trans-anastomotic communication between the jejunal vein and the intrahepatic portal venous branches (Fig. 2c). Present treatment has been aimed at medical reduction of his portal hypertension but recurrent episodes of acute gastrointestinal haemorrhage remain a problem at 1 year follow-up.
Case 3
Fig. 1 – Patient 1. Venous phase image from a distal superior meseteric artery angiogram obtained using a digital subtraction technique. The main superior mesenteric vein is tightly stenosed (open curved arrow). Varices are seen at the ileo-colic anastomosis (short arrow). In addition large varices are present within the jejunal veins bypassing the SMV stenosis.
A 70-year-old man referred for investigation of acute severe gastrointestinal haemorrhage had undergone an open cholecystectomy for gallstones 25 years previously, from which he had made an uneventful recovery. Ten years following this he presented with an episode of cholangitis which resulted in further imaging and a diagnosis of choledochal cyst was made, for which he underwent a choledochocyst-duodenostomy. Five years later he developed further severe episodes of cholangitis and underwent a choledochocyst-jejunostomy (in addition to his choledochocyst-duodenostomy). He continued to suffer from repeated episodes of cholangitis following this and had been admitted on one occasion during the previous 5 years with anaemia.
209
PORTO-PORTAL VARICES IN GASTROINTESTINAL HAEMORRHAGE
(a)
(b)
(c)
Fig. 2 – (a) Patient 2. Image from a percutaneous, transhepatic portogram with the catheter placed within the superior mesenteric vein. There is retrograde filling of the left gastric vein (short arrow) and of a jejunal vein which courses up towards the liver hilum (curved arrow). (b) Arterial image from a selective jejunal arteriogram demonstrating the normal arterial branches supplying the jejunal loop which has been brought up to the liver hilum. (c) Venous phase from the arteriogram shown in Fig. 2(b) demonstrates that the jejunal vein within this loop of jejunum communicates across the hepatic-jejunal anastomosis with an intrahepatic portal venous branch.
He was admitted on this occasion to the referring hospital again because of severe anaemia (haemoglobin 5.2 gm/dl) and during the following 10 days he suffered from persistent gastrointestinal bleeding requiring a total of 32 units of blood. He underwent an angiogram which failed to demonstrate a source of bleeding and was subsequently referred to this institution for further investigation. An abdominal US showed main portal vein occlusion and an enlarged liver and spleen. The arterial phase of an SMA angiogram demonstrated a loop of jejunum pulled across the midline to the site of the choledochocyst-jejunostomy in the right upper quadrant of the abdomen. There was no evidence of active contrast medium extravasation. The venous phase of the SMA angiogram confirmed occlusion of the portal vein with associated portal varices and also demonstrated retrograde filling of a large jejunal vein to the site of the choledochocyst-jejunostomy at which point there were large varices
communicating with intrahepatic portal vein branches (Fig. 3a). The indirect spleno-portogram demonstrated a patent splenic vein and retrograde filling of the distal portion of the SMV with subsequent opacification of the same jejuno-portal varices. The patient continued to bleed despite maximum medical therapy and a distal splenorenal shunt (DSRS) was therefore performed. Early thrombosis of the DSRS occurred which necessitated the percutaneous insertion of a stent to restore shunt patency. The jejuno-portal varices were well demonstrated during stent insertion by direct catheterization of the superior mesenteric vein via the spleno-renal shunt (Fig. 3b). After stent insertion antegrade flow through the surgical shunt was restored, although variceal filling still persisted. Gastrointestinal bleeding continued although at a lower rate than previously and further massive haemorrhage did not occur. Unfortunately the patient subsequently died due to overwhelming sepsis and multi-organ failure.
210
CLINICAL RADIOLOGY
(a) Fig 4 – Patient 4. Venous phase from a distal superior mesenteric artery angiogram demonstrates distal ileal veins communicating across the ileo-transverse colic anastomosis (arrow) with the middle colic vein.
(b) Fig. 3 – (a) Patient 3. Venous phase from a superior mesenteric arteriogram demonstrates a markedly enlarged jejunal vein (arrow) coursing towards the liver hilum where there are numerous varices with faint filling of intrahepatic portal venous branches. (b). Image from a study in which the superior mesenteric vein has been catheterized from an internal jugular vein approach via a spleno-renal shunt. The large varices crossing the choledochocyst-jejunotomy between the jejunal vein and portal venous branches are well visualized.
Case 4 A 40-year-old woman was referred for assessment of an extensive carcinoma of the transverse colon invading the second and third parts of the duodenum. She had undergone a laparotomy at her local hospital at which an extended right hemicolectomy was performed. She was referred to this institution for further evaluation. An abdominal CT was performed
which demonstrated the involvement of the duodenum; liver metastases were not identified. An angiogram was performed as a ‘road-map’ and for an assessment of operability prior to a Whipple’s procedure. The coeliac axis angiogram was normal. The arterial phase of the SMA angiogram demonstrated changes consistent with a right hemicolectomy and was otherwise normal. The venous phase of the angiogram revealed involvement by tumour of several jejunal veins and the main trunk of the SMV which were narrowed. The main portal vein was patent. A Whipple’s procedure was subsequently performed as a debulking procedure and to prevent duodenal obstruction and from which the patient made an uneventful initial recovery. One month post-operatively, however, she became profoundly hypotensive and passed copious amounts of dark blood per rectum necessitating admission to the intensive care unit. Visceral angiography performed on the same day demonstrated no contrast medium extravasation; venous phase images from the SMA angiogram showed evidence of tumour progression with more extensive involvement of the main SMV. Venous return from the distal small bowel occurred via porto-portal varices which had developed across the ileo-colic anastomosis between the distal ileal veins and the middle colic vein which in turn drained above the SMV stenosis into the gastrocolic trunk (Fig. 4). Review of the pre-operative arteriogram showed that the ileo-colic varices had been present, although smaller, at this time. The patient died a few days later from continued gastrointestinal haemorrhage. DISCUSSION
Extrahepatic portal occlusion, whatever its cause, may result in the development of portosystemic varices in an attempt to
PORTO-PORTAL VARICES IN GASTROINTESTINAL HAEMORRHAGE
decompress the elevated portal venous pressure distal to the obstruction into the low-pressure systemic veins [1]. These varices represent enlarged but otherwise normal pre-existing anastomoses between the portal and the systemic venous circulations. There will be simultaneous enlargement of the normal veins in the portal venous system which bypass the occluded segment into other portal venous system [3] veins and these are usually the dominant, and only, varices in patients with extrahepatic portal venous occlusion who do not have accompanying hepatic portal hypertension. For example, in a patient with occlusion of the splenic vein due to a pancreatic neoplasm, the predominant venous return from the spleen will usually occur via the gastroepiploic and the epiploic arcades into the main portal vein and via the short gastric veins into the left and right gastric veins; porto-systemic (e.g. oesophageal) varices will often notably be absent. On the other hand, in a patient who has splenic venous occlusion due to chronic alcoholic pancreatitis, for example, who also has elevated main portal venous pressure due to alcoholic cirrhosis, the dominant varices are likely to be between the portal and the systemic circulation. Whilst portosystemic varices occur at sites of pre-existing normal portosystemic anastomoses, they may also occur at sites of previous surgery, though reports of these are relatively infrequent [2,4–7]. The most well-recognized of these are stomal varices which develop at the site of a colostomy or ileostomy in patients with portal hypertension between the colonic or ileal veins, respectively, and the anterior abdominal wall veins; such varices may be the cause of troublesome recurrent haemorrhage [8–10]. These varices are thought to develop within peri-anastomotic inflammatory tissue and must, therefore, involve neo-angiogenesis. These, and other extraoesophagogastric porto-systemic venous communications, have been termed ‘ectopic’ varices [6]. The development of porto-portal communications across surgical anastomoses, as seen in the four patients presented, is less well appreciated and has received very little attention in the literature [11,12]. Three of our patients (cases 1, 3 and 4) were similar in that they had extrahepatic portal venous obstruction without severe hepatic portal hypertension. Varices presumably developed in these individuals across their previous surgical anastomoses (ileo-colic in two and jejuno-portal in one) as these communications represented the most direct route from an area of high portal pressure to an area of normal portal pressure. Similar to the varices described above which develop across stomas, these communications presumably develop within peri-anastomotic scar tissue. The development, in case 2, of porto-portal varices across an hepatico-jejunal anastomosis is perhaps more difficult to explain. This patient had mild hepatic portal hypertension and no extrahepatic portal occlusion and yet still developed trans-anastomotic communications. The most likely explanation for this is that the intrahepatic portal venous branches with which the jejunal veins communicated were at lower pressure than the main portal vein due to the patchy nature of the patient’s liver cirrhosis. The mechanisms by which trans-anastomotic veno-venous
211
communications develop are not entirely clear. The development of varices across surgical anastomoses requires the formation of new vessels (angiogenesis) as there are no preexisting communications. It is well known that angiogenesis occurs during wound healing (as well as other processes). Although angiographically identifiable communications between the arterial and the venous circulations across surgical anastomoses are rarely (if ever) seen in patients without intra- or extrahepatic portal hypertension, it is likely that these develop in the majority of normal individuals as part of the normal healing processes [13]. If extrahepatic portal venous obstruction subsequently develops then these communications presumably enlarge to form trans-anastomotic varices. As with varices elsewhere, those crossing surgical enteric anastomoses are at risk of haemorrhage and their treatment may be very difficult as demonstrated in the four individuals presented. A lowering of portal pressure proximal to the extrahepatic portal occlusion (i.e. pre-occlusion) by surgical shunting may be expected to be useful and did temporarily improve matters in one of our patients (case 3). If the extrahepatic portal venous obstruction can be crossed with a catheter and guide wire, this might respond to angioplasty or stent insertion. This technique has been used by other workers in patients with small bowel varices secondary to SMV stenosis [3] and is a treatment option in one of our patients with a tight SMV stenosis (case 1) if bleeding recurs. REFERENCES 1 Sherlock S, Dooley J. Diseases of the Liver and Biliary System, 10th edn. Oxford: Blackwell Scientific Publications, 1997; chapter 10. 2 Ahn J, Cooper JM, Silberzweig JE, Mitty HA. Venographic appearance of portosystemic collateral pathways. Br J Radiol 1997;70:1302–1306. 3 Cohen GS, Ball DS, Flynn DE. Transjugular placement of a superior mesenteric vein stent for small bowel varices. J Vasc Intervent Radiol 1995;6:707–710. 4 Buhler L, Tamigneaux I, Giostra E, et al. Varices ectopiques, une case rare d’hemorrhage digestive. Schweiz-Med-Wochenschr-Suppl 1996; 79:70S–72S. 5 Cappell MS, Price JB. Characterization of the syndrome of small and large intestinal variceal bleeding. Dig Dis Sci 1987;32:422–427. 6 Lebrec D, Benahmou JP. Ectopic varices in portal hypertension. Clin Gastroenterol 1985;14:105–121. 7 Hamlyn AN, Lunzer MR, Morris JS, Puritz H, Dick R. Portal hypertension with varices at unusual sites. Lancet 1974;2:1531–1534. 8 Lashley DB, Saxon RR, Fuchs EF, Chin DH, Lowe BA. Bleeding ileal conduit stomal varices: diagnosis and management using trans jugular transhepatic angiography and embolization. Urology 1997;50:612– 614. 9 Wong RC, Berg CL. Portal hypertensive stomopathy: a newly described entity and its successful treatment by placement of a transjugular intrahepatic portosystemic shunt. Am J Gastroenterol 1997;92:1056– 1057. 10 Wolfson HC, Kozareck RA, Bredfeldt JE, Fenster LF, Brubacher LL. The role of endoscopic injection sclerotherapy in the management of peristomal varices. Gastrointest Endosc 1990;36:472–474. 11 Miller JT Jr, De Odorico I, Marx MV. Cholecystojejunostomy varices demonstrated by enteroclysis. Abdom Imag 1997;22:474–476. 12 Carpenter S, Brown KA. Chronic complications after cholecystojejunostomy. Am J Gastroenterol 1994;89:2073–2075. 13 Shikata J, Shida T. Experimental studies by the resin-casting method on the vascular structure of the colon following stapler anastomoses. Dis Colon Rectum 1985;28:341–346.