Chronic portomesenteric (PM) and portosplenomesenteric

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2007;5:80 86 Chronic Mesenteric Venous Thrombosis: Evaluation and Determinants of Survival During Long-Term Follow-up DAVID W. ORR,* PHILLIP M. HARRISON,* JOHN DEVLIN,* JOHN B. KARANI, PAULINE A. KANE, NIGEL D. HEATON,* JOHN G. O GRADY,* and MICHAEL A. HENEGHAN* *Institute of Liver Studies, and Department of Radiology, King s College Hospital, Denmark Hill, London, England See CME exam on page 3. Background & Aims: The natural history of chronic portomesenteric (PM) and portosplenomesenteric (PSM) venous thrombosis is defined poorly. Therapeutic options are limited, and are directed at the prevention of variceal bleeding and the control of abdominal pain related to gastrointestinal hyperemia. Methods: Patients with extensive PM and PSM thrombosis were reviewed retrospectively to evaluate the efficacy of medical therapy and to determine which clinical variables had prognostic significance regarding long-term survival. Results: Sixty patients, with a median age at diagnosis of 44 years (range, y), were assessed. The median follow-up period was 3.5 years (range, y). The overall survival rate was 73.3%, with 1- and 5-year survival rates of 81.6%, and 78.3%, respectively. One- and 5-year survival rates, excluding patients who died from malignancy-related causes, were 85.7% and 82.1%, respectively. Factors associated with improved survival included treatment with -blockers (P.02; odds ratio [OR],.09; 95% confidence interval [CI], ) and anticoagulation (P.005; OR, 0.01; 95% CI, <0.01 to 0.26). Eighteen patients in total were anticoagulated, including 8 patients who had variceal bleeding, all of whom underwent endoscopic band ligation of esophageal varices before anticoagulation. By using Cox regression analysis, variables associated with reduced survival were the presence of ascites (P.001; OR, 42.6; 95% CI, ), and hyperbilirubinemia (P.01; OR, 13.8; 95% CI, ) at presentation. Six patients died of variceal hemorrhage. Conclusions: Patients with chronic PM and PSM venous thrombosis without underlying malignancy have an acceptable long-term survival. Treatment with -blockers and anticoagulation appears to improve outcome. Chronic portomesenteric (PM) and portosplenomesenteric (PSM) venous thrombosis are rare conditions that usually are related to thrombophilia, intra-abdominal inflammatory disease, and, less commonly, owing to cirrhosis and malignancy. The clinical presentation varies according to the temporal relationship to thrombosis. Patients with acute mesenteric and portal vein (PV) thrombosis present with sudden onset of abdominal pain as a result of bowel ischemia and infarction. For such patients, the mortality of acute thrombosis is high, with reported 30-day and 3-year mortality rates of 23% 27% and 64% 67%, respectively. 1,2 In contrast, the presenting symptoms and signs of chronic PM and PSM venous thrombosis are frequently nonspecific with abdominal symptoms or variceal bleeding. The natural history of chronic PM and PSM venous thrombosis is unknown, with current therapeutic measures directed at prevention of variceal bleeding and symptom control. Because conventional portosystemic surgical and radiologic shunts are not possible, medical management is the mainstay of treatment. Evidence from retrospective case series, of patients with cirrhotic and noncirrhotic extrahepatic portal vein thrombosis (EHPVT), suggests that death from variceal bleeding occurs infrequently, with mortality owing to the underlying disease process or bowel infarction. 3,4 Case series of patients with acute noncirrhotic PV and superior mesenteric vein (SMV) thrombosis managed with early initiation of anticoagulation have shown benefit by reducing complications of bowel necrosis owing to recanalization of thrombosed vessels. 3,5 7 However, none have shown improved survival with medical management. Because the risk-benefit analysis of anticoagulation in patients with chronic thrombosis is unknown, initiation of treatment is tempered by the theoretic increased risk of life-threatening hemorrhage. For that reason, we have undertaken a review of patients with PM and PSM thrombosis to examine the causative risk factors, clinical determinants of long-term survival, and the efficacy of anticoagulation and -blockers in this patient group. Patients and Methods All patients with chronic PM and PSM venous thrombosis diagnosed at King s College Hospital in London between January 1973 and June 2005 were reviewed. The diagnosis of chronic PM and PSM venous thrombosis was made on the basis of computerized tomography (CT), magnetic resonance imaging, and/or aortoportography. The chronicity of thrombosis was established by a follow-up period of more than 3 months, with evidence of persistent splanchnic venous occlusion on radiologic examination. Supportive evidence of chronic thrombosis was suggested by the presence of extensive intra-abdominal venous collaterals, cavernous transformation of the portal vein, and splenomegaly. Patient demographics including clinical presentation, laboratory data, hematologic investigations for procoagulant conditions, medical treatment, and endoscopic data were collated from the medical record. The assessment for hematologic procoagulant conditions was in accordance with medical practice of the day. Liver biopsy examinations were performed as deemed clinically appropriate. The presence of esophageal and gastric varices and variceal bleeding was assessed by esophagogastroduodenoscopy. Abbreviations used in this paper: CI, confidence interval; CT, computerized tomography; EHPVT, extrahepatic portal vein thrombosis; EV, esophageal varices; EVBL, endoscopic variceal band ligation; MPD, myeloproliferative; MRV, magnetic resonance venography; OR, odds ratio; PM, portomesenteric; PSM, portosplenomesenteric; PV, portal vein; SMV, superior mesenteric vein by the AGA Institute /07/$32.00 doi: /j.cgh

2 January 2007 CHRONIC PORTOMESENTERIC VENOUS THROMBOSIS 81 The primary outcome measure was survival. Secondary outcomes included prevention of primary or recurrent variceal bleeding. The management of portal hypertension at King s College Hospital is aimed at the prevention of first variceal bleeding episode with nonselective -blockers (propranolol). Treatment was considered adequate if there was a 25% reduction in the resting heart rate or reduction of the resting heart rate to 60 or less per minute. Secondary prophylaxis was with -blockade, as described previously, and endoscopic injection sclerotherapy or endoscopic variceal band ligation (EVBL) with eradication of esophageal varices. Gastric varices, which had bled or showed high-risk stigmata of bleeding, were injected with Histoacryl glue (Braun Medical Ltd, Melsungen, Germany). Coumadin was commenced in patients with a recognized procoagulant condition, in whom no contraindication to treatment existed, aiming for a therapeutic international normalized ratio of Anticoagulation was commenced only in patients with previous variceal bleeding after eradication of varices. Statistical Analysis For quantitative data, analysis was performed using the Mann Whitney and Kruskal Wallis tests for comparisons of 2 or more independent groups. The difference in proportions of categoric data was ascertained by the Fisher exact test when the number of subjects was less than 5, and by the 2 test for 2 2 tables when the number of subjects in each cell was 5 or more. Continuous data were categorized into normal and abnormal values for multivariate analysis. Multivariate Cox regression analysis was conducted to analyze clinical variables and treatment outcome. Survival plots were generated according to the Kaplan Meier method and were compared using the log-rank test. Data are expressed as means or median, range and interquartile range when appropriate. P values less than.05 were considered significant. All analysis was performed using SPSS 13.0 (SPSS Inc., Chicago, IL). Results The study group was composed of 60 patients (31 women, 29 men). The median age at the time of diagnosis of PM or PSM venous thrombosis was 44 years (range, y). Forty-eight of 60 (80%) patients were of European ethnicity, 8 (13.3%) from the Indian subcontinent, and 4 (6.7%) were of African ethnicity. The median follow-up period was 3.5 years (range, y). All patients underwent abdominal ultrasound scan with Doppler assessment of the portal vein, confirming extrahepatic PV occlusion. Occlusion of the PM or PSM veins was confirmed by a combination of CT, magnetic resonance venography (MRV), and/or aortoportography. Fifty patients underwent CT scan, 23 patients MRV, and 29 patients aortoportography. The pattern of splanchnic venous thrombosis was PSM thrombosis in 54 patients and PM thrombosis in 6 patients. These 6 patients had evidence of proximal splenic vein patency. The presenting features were variceal bleeding in 31 of 60 (51.7%) patients, abdominal pain in 18 of 60 (30.0%) patients, abdominal pain and ascites in 4 of 60 (6.7%) patients, ascites in 2 of 60 (3.3%) patients, and 5 of 60 (8.3%) cases were discovered incidentally during investigation of thrombocytopenia, anemia, splenomegaly, or increased liver enzyme levels. Table 1. Cause of Chronic PM and PSM Venous Thrombosis Risk factor (n) n (%) Acquired disorder 17/60 (28.3) Myeloproliferative disorder Polycythemia rubra vera 4/60 (6.7) Essential thrombocythemia 3/60 (5.0) Myelofibrosis 1/60 (1.7) Unclassified 3/60 (5.0) Antiphospholipid syndrome 5/51 (9.8) Monoclonal gammopathy 1/49 (2.0) Heritable disorder 9/60 (15.0) Factor V Leiden mutation 3/49 (6.1) Paroxysmal nocturnal hemoglobinuria 2/34 (5.9) Sickle cell disease 1/9 (11.1) Antithrombin III deficiency 1/49 (2.0) Protein S deficiency 1/49 (2.0) Plasminogen deficiency 1/36 (2.8) Cirrhosis (ALD 2, HBV 2, AIH 1, 6/60 (10.0) cryptogenic 1) Intra-abdominal infection (4 umbilical 5/60 (8.3) vein sepsis, 1 liver abscess) Cancer (HCC 3, TCC 1) 4/60 (6.7) Oral contraceptive pill 4/60 (6.7) Postoperative 3/60 (5.0) Pancreatitis 2/60 (3.3) Congenital venous anomaly 2/60 (3.3) Intestinal volvulus 1/60 (1.7) Idiopathic (unknown) 17/60 (28.3) NOTE. Seventy risk factors were identified in 60 patients. ALD, alcoholic liver disease; HBV, hepatitis B virus; AIH, autoimmune hepatitis; HCC, hepatocellular carcinoma; TCC, transitional cell carcinoma. Etiology of Portomesenteric and Portosplenomesenteric Thrombosis The causes of PM and PSM venous thrombosis are shown in Table 1. Ten (16.7%) patients had more than 1 procoagulant risk factor identified. Acquired procoagulant conditions, predominantly myeloproliferative disorders, accounted for thrombosis in 17 of 60 (28.3%) patients. Seventeen (28.3%) patients had no identifiable procoagulant condition. Six (10%) patients had cirrhosis, 2 of these patients had additional procoagulant risk factors of factor V Leiden mutation and hepatocellular carcinoma. Thirty patients (50%) underwent liver biopsy examination, including the aforementioned 6 patients with cirrhosis. Of the remaining patients, 19 of 30 (63.3%) had normal biopsy specimens or mild fibrosis, and 5 of 30 (16.7%) had nodular regenerative hyperplasia. Of those patients with malignancy, 3 patients with hepatocellular carcinoma died from metastatic cancer at a median of 4 months after the diagnosis of venous thrombosis. The fourth patient with transitional cell carcinoma presented with variceal bleeding and died 2 years later of metastatic disease. Three patients developed PSM thrombosis after abdominal surgery, presenting a median of 5 years after intervention. Outcome and Variables Associated With Outcome During the follow-up period, 16 patients died, with an overall survival rate of 73.3%. The causes of death are shown in Table 2. The overall survival at 1- and 5-years was 81.6%, and

3 82 ORR ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 1 Table 2. Cause of Death in 16 Patients With Chronic Mesenteric Venous Thrombosis Cause of death n 16 (%) Variceal bleeding 6 (37.5) Metastatic carcinoma 4 (25.0) Septicemia 3 (18.75) Decompensated liver disease 1 (6.25) Intracerebral bleed 1 (6.25) Ischemic gut 1 (6.25) 78.3%, respectively. The 1- and 5-year survival rates excluding patients diagnosed with cancer was 85.7% and 82.1%, respectively. Patient characteristics at the time of diagnosis and medical therapy were evaluated for their prognostic significance (Table 3). Patient age, sex, or mode of clinical presentation did not influence survival. Most patients (35 of 60 [58.3%]) were diagnosed after January The mortality rate was higher in patients diagnosed after this time on Cox-regression analysis (P.04; odds ratio [OR], 3.9; 95% confidence interval [CI], ). Patients diagnosed after January 2000 were more likely to present with abdominal pain (15 of 35 [42.9%] vs 4 of 25 [16.0%], respectively, P.05) and less likely to present with variceal bleeding (13 of 35 [37.1%] vs 19 of 25 [76.0%], respectively, P.004). There was no significant difference in the other clinical variables. Laboratory values including hematologic indices, international normalized ratio, and aspartate aminotransferase levels were not predictive of outcome (data not shown). The extent of thrombosis, classified according to either PM or PSM thrombosis, was not predictive of outcome. However, clinical variables at presentation associated with poor outcome included increased bilirubin level greater than 20 mol/l, albumin level less than 35g/L, and the presence of ascites on index ultrasound examination (P.05) Table 3. Patient Characteristics, Symptoms, and Radiologic Findings at Diagnosis in 60 Patients Classified According to Survival Outcome (% and interquartile range) Patient characteristics and treatment Survived (n 44) Died (n 16) P value Age, y (median) 43 (33 53) 46 (32 60).76 Sex Male 18 (62.1) 11 (37.9).06 Female 26 (83.9) 5 (16.1) Laboratory values Bilirubin level ( mol/l) Median (interquartile range) 13 (10 22) 21 (11 49) (52.2) 11 (47.8) (86.5) 5 (23.5) Albumin level (g/l) Median (interquartile range) 39 (36 41) 36 (33 38) 35 8 (53.3) 7 (46.7) (80.0) 9 (20.0) Variceal bleeding Present 23 (71.9) 9 (28.1).79 Absent 21 (75.0) 7 (25.0) Varices at endoscopy a Present 37 (74.0) 13 (26.0).99 Absent 5 (83.3) 1 (16.7) Abdominal pain Present 14 (73.7) 5 (26.3).97 Absent 30 (73.2) 11(26.8) Ascites (radiologic) Present 8 (50.0) 8 (50.0).01 Absent 36 (81.8) 8 (18.2) Cirrhosis Present 4 (66.7) 2 (33.3).70 Absent 40 (74.1) 14 (25.9) Cancer Present 0 (0) 4 (100).004 Absent 44 (78.6) 12 (21.4) Medical management -blockers Yes 29 (82.9) 6 (17.1).05 No 15 (60.0) 10 (40.0) Warfarin Yes 17 (94.4) 1 (5.6).02 No 27 (64.3) 15 (35.7) a 56 patients underwent esophagogastroduodenoscopy.

4 January 2007 CHRONIC PORTOMESENTERIC VENOUS THROMBOSIS 83 (Table 3). After exclusion of cirrhotic patients from the analysis, increased bilirubin level and low albumin and ascite levels still were associated with a worse outcome (P.05). Patient survival was significantly worse in patients with malignancy (P.01). Other risk factors for thrombosis, including the presence of cirrhosis, were not associated with a worse outcome. Complications of Thrombosis Fifty-six patients underwent esophagogastroduodenoscopy for variceal assessment. Fifty-one of 56 (91%) patients had esophageal varices (EVs), and 28 of 56 (50%) patients had gastric varices. Thirty-nine of 56 (76%) patients with varices had an episode of variceal bleeding. The sentinel variceal bleed in all patients was from EVs. After management of the primary EV bleed, 18 of 39 (46%) patients had recurrent variceal bleeding. The median time from treatment of the primary bleed to a secondary bleed was 48 months (range, mo). No clinical or laboratory variable except presentation with variceal bleeding was predictive of rebleeding. The secondary bleed was from EVs in 11 patients. Seven patients had recurrent bleeding from gastric varices, 7 patients from small-bowel varices, and 2 from colonic varices. In total, 6 patients died from variceal bleeding, 1 from EV, 3 from gastric varices, and 1 each from small-bowel varices and colonic varices at a median of 12 months (range, mo) after diagnosis. Biliary disease was clinically significant in 5 patients. All had clinical cholangitis with cholestatic liver function tests, as a result of extrinsic compression from cavernous transformation of the PV and choledochal varices in 2 patients and extrahepatic biliary stricturing in 3 patients. Endoscopic retrograde cholangiopancreatography with biliary stenting was performed in all patients. Four patients developed recurrent cholangitis (range, 2 5 clinical events), with a median of 3 endoscopic retrograde cholangiopancreatographies (range, 2 11 clinical events) required during the follow-up evaluation. Recurrent cholangitis resulted in septicemia and death in 1 patient. Emergency laparotomy, with resection of ischemic small bowel, was required in 3 patients. The time from diagnosis of PSM thrombosis to the development of transmural bowel infarction ranged from 2 to 34 months. One patient died as a result of small-bowel infarction. None of these patients were anticoagulated at the time of development of gut ischemia/infarction. During follow-up evaluation, no patient had evidence of extrasplanchnic thrombosis. Postprandial abdominal pain was the most common symptom reported in 40 of 60 (66.6%) patients during the follow-up evaluation. Five patients had persistent severe abdominal pain unresponsive to -blockade, simple analgesia, and neuropathic modulating drugs such as gabapentin. Abdominal radiograph and CT scanning were performed to exclude subacute bowel obstruction and to confirm edematous small bowel. These patients were commenced on subcutaneous octreotide (Sandostatin; Novartis Pharmaceuticals, East Hanover, NJ), mcg, 3 times/day, and of these 4 patients had self-reported improvement in symptoms. Two of 5 patients were converted to long-acting octreotide (Sandostatin LAR; Novartis Pharmaceuticals), mg/mo, and have required long-term octreotide to control abdominal pain. Medical Therapy Eighteen patients were maintained on long-term anticoagulation. Of these, 9 of 18 (50%) had variceal bleeding. Five patients had EV bleeding before commencing coumadin, all of whom had endoscopic injection sclerotherapy/evbl with eradication of EV before anticoagulation. The median time from variceal bleed to commencing coumadin was 18 months (range, 3 30 mo). None of these patients have had recurrent variceal bleeding. Four patients had a primary EV bleed at a median of 46 months (range, mo) after anticoagulation. Of these, 3 were re-established on coumadin after eradication of EVs, although 1 has recurrent small-bowel variceal bleeds, but is maintained on anticoagulation because of multiple previous thrombotic events. The remaining patient died from EV bleeding. In total, 2 of 9 (22.2%) anticoagulated patients have had recurrent variceal bleeding, compared with 16 of 30 (53.3%) patients not treated with anticoagulation (P NS). The 9 anticoagulated patients without variceal bleeding have had regular endoscopic screening. Of these, 4 had no varices and 5 had small varices. During radiologic surveillance, 3 patients treated with coumadin had evidence of partial recanalization of the PSM veins. Recanalization included the SMV, the portomesenteric veins, and the portal vein. Recanalization was not seen in any of the nonanticoagulated patients. A comparative assessment of anticoagulated with nonanticoagulated patients showed that there were no statistically significant differences between patient age, sex, duration of follow-up period, presence of ascites, hematologic indices, and liver function tests between groups. Patients were more likely to be anticoagulated if they were diagnosed with myeloproliferative (MPD) (P.01), or if they presented with abdominal pain (P.05). Patients were less likely to be treated with coumadin if they presented with variceal bleeding (P.01). However, the clinical presentations of abdominal pain, variceal bleeding, or underlying procoagulant disorder were not predictive of survival (Table 3). Patient outcome on univariate analysis and log-rank test showed improved survival in patients treated with long-term coumadin (94.4% vs 64.3%, P.02, Table 3, Figure 1A). Thirty-five patients (58.3%) were maintained on long-term nonselective -blockers. Of these patients, 24 of 35 (68.6%) were commenced on treatment as secondary prophylaxis of variceal bleeding. The remaining 11 patients were commenced on -blockers as primary prophylaxis of variceal bleeding caused by the presence of varices on esophagogastroduodenoscopy (n 10) or the presence of postprandial abdominal pain (n 1). When comparing patients treated with -blockers with those without, there was no statistically significant difference between groups in relation to age at presentation, sex, duration of follow-up evaluation, presence of ascites, hematologic indices, liver function tests, and cause of thrombosis. The incidence of variceal bleeding in patients who received primary prophylaxis with -blockers was compared with those patients with endoscopically apparent grade 1 2 EV who did not receive primary prophylaxis. Ten patients were treated, of whom 1 (10.0%) had a variceal bleed. In contrast, 5 of 11 (45.5%) patients who did not receive primary prophylaxis had a variceal bleed (P.15). Twenty-four patients were treated with -blockers as secondary prophylaxis. Twelve (50.0%) had recurrent variceal bleeding. In comparison, 6 of 15 (40%) patients who were not treated had recurrent bleeding (P.54). Although the median number of recurrent variceal bleeds in the treatment group was less than untreated patients, this was not significant. Patient outcome on 2 and log-rank analysis showed improved survival in patients treated

5 84 ORR ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 1 Figure 2. Survival curves (in months) for patients with chronic PM or PSM venous thrombosis who received -blocker monotherapy, coumadin monotherapy, or combination treatment vs no treatment (P.05, log-rank test)., Coumadin (n 8);, -blockers (n 25); -----, -blockers and coumadin (n 10;, no treatment (n 17). Clinical variables predictive of a poor prognosis included radiologically apparent ascites and increased bilirubin levels at presentation (P.001; OR, 42.6; 95% CI, and P.01; OR, 13.8; 95% CI, , respectively). Significantly, patients treated with coumadin and/or -blockers had an improved survival on multivariate analysis (P.005; OR, 0.01; 95% CI, 0.01 to 0.26 and P.02; OR, 0.09; 95% CI, , respectively). Figure 1. Survival curves (in months) for patients with chronic PM or PSM venous thrombosis who received: (A) long-term anticoagulation or no anticoagulation (P.05, log rank test) and (B) -blockade or no -blockade (P.04, log-rank test). (A) (solid line), Coumadin; (broken line), no coumadin; (B)(solid line), -blocker; (broken line), no -blocker. with nonselective -blockers (82.9% vs 60.0%, respectively; P.05, Table 3, Figure 1B). Ten patients were treated with the combination of -blockade and anticoagulation, 17 patients received neither treatment. Survival at 1 and 5 years was significantly better in the combination therapy group (100% and 75% vs 58.8% and 46.6%, respectively, P.03). There was no significant difference between outcome comparing monotherapy with coumadin or -blockade and combination therapy (Figure 2). The multivariate survival analysis by Cox regression is shown in Table 4. Previous variceal bleeding and MPD were included in the analysis because of increased -blocker use in variceal bleeding and anticoagulation in MPD. Patients with malignancy were excluded from analysis because all of these patients died as a consequence of their cancer rather than from complications of venous thrombosis. Discussion In this study, we have characterized the clinical features, risk factors, and outcomes of patients with chronic PM and PSM venous thrombosis. We also have shown that therapy with -blockade and anticoagulation were associated with improved survival in this patient cohort. The causes of PM and PSM thrombosis are diverse, with most patients in our series (71.7%) having an identifiable procoagulant condition. Acquired and inherited thrombophilic states were the most commonly identified risk factors in this report, which is consistent with 2 published series in noncirrhotic EHPVT, in which up to 75% of patients who underwent full thrombophilia Table 4. Clinical Variables and Medical Treatment Influencing Survival by Cox Regression Analysis Clinical variable a OR 95% CI P value Coumadin to blocker Ascites Bilirubin ( 20 mol/l) Albumin ( 35 g/l) Variceal bleed MPD NOTE. OR of death with 95% CIs are shown. a Patients with cancer (n 4) censored from multivariate analysis.

6 January 2007 CHRONIC PORTOMESENTERIC VENOUS THROMBOSIS 85 screening were identified with an acquired or hereditary procoagulant predisposition. 8,9 Inherited prothrombotic disorders were identified in 15% of our patients, with factor V Leiden mutation identified in 6.1% of all patients tested. This is similar to the reported prevalence of the mutation in European populations. 10,11 Of note, these 3 patients had additional risk factors to the factor V Leiden mutation and, significantly, 10 of 60 (17%) patients had more than 1 identifiable risk factor. The presence of concurrent procoagulant disorders should therefore be sought in all patients, and indeed analysis of JAK2 mutation, prothrombin gene mutation, and methylene-tetrahydrofolate reductase mutation 9,15 may be of diagnostic value in this cohort. In this series, the long-term outcome of patients without evidence of malignancy is acceptable, with 1- and 5-year survival rates of 85.7% and 82.1%, respectively. Survival, when compared with patients with noncirrhotic EHPVT, is worse (1- and 5-year survival, 95% and 89%, respectively), but significantly better than all-comers (unselected patients) with EHPVT (1- and 5-year survival, 70% and 61%, respectively). 4 Janssen et al 4 reported that hepatic disease, predominantly cirrhosis, and malignancy contributed to EHPVT in 71% of patients, in comparison with 16% reported in this series, with the majority of patients dying from either malignancy or liver failure. Most deaths in this series arose from variceal bleeding or from septic complications within the first year of diagnosis. The difference in long-term survival between patients with EHPVT and chronic PSM thrombosis may be related to the predominant underlying cause. Patients with chronic PSM thrombosis typically have a defect in coagulation or a diffuse inflammatory process rather than localized disease. To this end, medical management in chronic PSM thrombosis should be directed at reducing the risk of portal hypertensive and thrombotic complications. In this series, 76% of patients had an EV bleed, all of whom survived their index event. Death caused by variceal hemorrhage was from recurrent variceal bleeding, either from gastric or ectopic varices. In general, recommendations for managing portal hypertension in noncirrhotic patients are extrapolated from randomized controlled trials and meta-analyses of data derived from cirrhotic patients In this report, although there was a trend toward a reduced risk of primary variceal bleeding in patients treated with -blockers (11.1% vs 45.4%), this was not significant. However, in view of the numbers in this aspect of the study, the chance of a type II error is high. Recent evidence has shown that EVBL can be effective in the primary prevention of variceal bleeding Thus, EVBL should be considered when -blockers are contraindicated or when patients are intolerant. In this study, secondary prophylaxis of bleeding used a strategy of variceal eradication and -blockade. Re-bleeding was common, occurring in 46% of patients despite regular endoscopic follow-up evaluation. During follow-up evaluation, 61.1% of patients rebled from EVs, and 83.3% developed recurrent bleeding from either gastric or ectopic variceal sources. -blockers did not confer a reduced risk of secondary variceal bleeding. From this study it is apparent that there is an increased mortality from variceal bleeding in patients with chronic PSM thrombosis (10%) compared with noncirrhotic PVT, in whom mortality related to bleeding has been reported to be as low as 1.5%. 3,23 This difference is explained by an increased incidence of gastric and ectopic varices and variceal bleeding from more extensive thrombosis and collateralization. When analyzed using multivariate analysis, treatment with -blockers was associated with a significantly better survival rate (P.04; OR, 0.10; 95% CI, ). In patients with extensive venous thrombosis, the use of anticoagulation must balance the risk of thrombus extension and bowel infarction against the risk of variceal bleeding, both of which have a significant associated mortality. 1 3 There is increasing evidence that the risk-benefit ratio favors anticoagulation in noncirrhotic EHPVT. 3,5,6 However, no study has shown improved survival. A retrospective study of patients with early PV or SMV thrombosis showed that the majority (93%) of anticoagulated patients either partially or completely recanalized the occluded vessel. 7 A larger study from the same group showed that the risk of thrombotic complications in PVT appears to be as significant as the risk of variceal bleeding, with a reported increased risk of splanchnic venous infarction in patients who did not receive anticoagulation. 3 In this series, patients were less likely to be anticoagulated when a self-limiting cause for PSM thrombosis was identified, the cause of thrombosis was not identified, or they presented with variceal bleeding. However, 8 patients who had a variceal bleed were anticoagulated after eradication of EV. Although patients with variceal bleeding at presentation were less likely to be anticoagulated, the mode of presentation was not predictive of outcome. Furthermore, there was no evidence of an increased risk of variceal bleeding in anticoagulated patients. Indeed, one of the most pertinent findings from this study was the improved survival noted in patients who were anticoagulated. This improved survival can be postulated to result either from recanalization of vessels or in the prevention of thrombus propagation within the small-bowel anastomotic loops. We believe that anticoagulation therefore reduces the risk of transmural bowel infarction. Anticoagulation of patients with a self-limiting cause for the venous thrombosis needs to be defined further, and, in such patients, the benefit may be less because inherent fibrinolytic mechanisms may prevent further extension of the thrombosis. The presented data do not show that coumadin and -blockers reduce the risk of variceal bleeding or transmural bowel infarction. Although there is a trend toward reduced primary variceal bleeding in patients treated with -blockers and secondary variceal bleeding in anticoagulated patients, these findings are not sufficient to explain the improved survival. Aspects that have not been investigated because of small patient numbers and incomplete data are the severity of secondary variceal bleeding and the number of endoscopies with variceal eradication. Patients maintained on -blockers may have less severe secondary variceal hemorrhage and patients with known portal hypertension treated with anticoagulation may undergo more regular endoscopy and variceal eradication of high-risk varices, improving patient outcome. In this study, clinical determinants of poor outcome were the findings of an increased bilirubin level and the presence of ascites on ultrasound examination. Thus, relatively minor changes in synthetic function appear to impact on outcome. Biliary complications in chronic PSM thrombosis resulted in morbidity and mortality in 8.3% and 1.7% of patients, respectively, in this study. Five patients developed cholestatic liver enzymes and cholangitis, which resulted in the death of 1 patient. Endoscopy remains the mainstay of therapy in managing symptomatic choledochal varices. 24 Biliary surgery carries a significant morbidity and mortality, 24 and conventional surgical portosystemic shunting or transjugular intrahepatic portosystemic shunt is not possible in this patient group. Previous studies have shown that long-term biliary stenting is effective without causing morbidity, 25,26 however, in our experience, stent occlusion and recurrent cholangitis can result in significant morbidity and mortality.

7 86 ORR ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 1 The limitations of this study are that it represents a retrospective analysis of a heterogeneous patient cohort. Including all patients treated at our institution minimized selection bias in as much as this was possible. Furthermore, the long study period included patients who presented in the 1970s and this may have resulted in treatment bias. Undoubtedly, management of variceal bleeding has improved with time, and new endoscopic techniques and therapeutic agents have been introduced Despite improved therapeutic intervention, the apparent survival in patients diagnosed within the past 5 years is worse. Patients diagnosed before the availability of high-resolution, multislice CT and MRV are more likely to have been diagnosed with aortoportography. This diagnostic investigation is more likely to have been performed because of the chronicity of their condition and for consideration of possible shunt surgery, thereby improving the apparent survival in the earlier patient group. Treatment with -blockers was directed at patients with large varices or who had experienced a variceal bleed, and because most deaths were related to variceal hemorrhage, patients without varices were less likely to be treated. Thus the efficacy of -blockers may be underestimated. Indeed, the converse also may be true for patients treated with anticoagulation: patients presenting with variceal hemorrhage were less likely to be treated, such that the efficacy of anticoagulation may be overestimated. In conclusion, we have shown that patients with PM and PSM venous thrombosis, without malignancy, have an acceptable longterm survival. Clinical features predictive of poor outcome include mild hyperbilirubinemia and ascites at presentation. Variceal bleeding is the primary cause of morbidity and mortality, yet treatment with -blockade and anticoagulation appears to improve survival. -blockade should be considered in all patients and we recommend long-term anticoagulation in all patients in whom a prothrombotic disorder has been identified. References 1. Rhee RY, Gloviczki P, Mendonca CT, et al. Mesenteric venous thrombosis: still a lethal disease in the 1990s. J Vasc Surg 1994;20: Morasch M, Ebaugh J, Chiou A, et al. Mesenteric venous thrombosis: a changing clinical entity. J Vasc Surg 2001;34: Condat B, Pessione F, Hillaire S, et al. Current outcome of portal vein thrombosis in adults: risk and benefit of anticoagulant therapy. Gastroenterology 2001;120: Janssen HL, Wijnhoud A, Haagsma EB, et al. Extrahepatic portal vein thrombosis: aetiology and determinants of survival. Gut 2001;49: Joh JH, Kim DI. Mesenteric and portal vein thrombosis: treated with early initiation of anticoagulation. Eur J Vasc Endovasc Surg 2005;29: Kocher G, Himmelmann A. Portal vein thrombosis (PVT): a study of 20 non-cirrhotic cases. Swiss Med Wkly 2005;135: Condat B, Pessione F, Helene Denninger M, et al. Recent portal or mesenteric venous thrombosis: increased recognition and frequent recanalization on anticoagulant therapy. Hepatology 2000;32: Denninger M, Chait Y, Casadevall N, et al. Cause of portal or hepatic venous thrombosis in adults: the role of multiple concurrent factors. Hepatology 2000;31: Primignani M, Martinelli I, Bucciarelli P, et al. Risk factors for thrombophilia in extrahepatic portal vein obstruction. Hepatology 2005;41: Mahmoud A, Elias E, Beauchamp N, et al. Prevalence of the factor V Leiden mutation in hepatic and portal vein thrombosis. Gut 1997;40: Heit J, Sobell J, Li H, et al. The incidence of venous thromboembolism among Factor V Leiden carriers: a community-based cohort study. J Thromb Haemost 2005;3: Tefferi A, Lasho T, Schwager S, et al. The JAK2(V617F) tyrosine kinase mutation in myelofibrosis with myeloid metaplasia: lineage specificity and clinical correlates. Br J Haematol 2005;131: Kralovics R, Passamonti F, Buser A, et al. A gain-of-function mutation of JAK2 in myeloproliferative disorders. N Engl J Med 2005;352: Tefferi A, Gilliland D. JAK2 in myeloproliferative disorders is not just another kinase. Cell Cycle 2005;4: Amitrano L, Brancaccio V, Guardascione M, et al. Inherited coagulation disorders in cirrhotic patients with portal vein thrombosis. Hepatology 2000;31: Berrut G, Ghali A, Quere I, et al. A common mutation C677T in the 5,10-methyltetrahydrofolate reductase gene is associated to idiopathic deep venous thrombosis. Rev Med Interne 2003;24: D Amico G, Pagliaro L, Bosch J. The treatment of portal hypertension: a meta-analytic review. Hepatology 1995;22: Pagliaro L, D Amico G, Sorensen T, et al. Prevention of first bleeding in cirrhosis. A meta-analysis of randomized trials of nonsurgical treatment. Ann Intern Med 1992;117: Bernard B, Lebrec D, Mathurin P, et al. Beta-adrenergic antagonists in the prevention of gastrointestinal rebleeding in patients with cirrhosis: a meta-analysis. Hepatology 1997;25: Lui H, Stanley A, Forrest E, et al. Primary prophylaxis of variceal hemorrhage: a randomized controlled trial comparing band ligation, propranolol, and isosorbide mononitrate. Gastroenterology 2002;123: Sarin S, Lamba G, Kumar M, et al. Comparison of endoscopic ligation and propranolol for the primary prevention of variceal bleeding. N Engl J Med 1999;340: Schepke M, Kleber G, Nurnberg D, et al. Ligation versus propranolol for the primary prophylaxis of variceal bleeding in cirrhosis. Hepatology 2004;40: Vleggaar F, van Buuren H, Schalm S. Endoscopic sclerotherapy for bleeding oesophagogastric varices secondary to extrahepatic portal vein obstruction in an adult Caucasian population. Eur J Gastroenterol Hepatol 1998;10: Khare R, Sikora S, Srikanth G, et al. Extrahepatic portal venous obstruction and obstructive jaundice: approach to management. J Gastroenterol Hepatol 2005;20: Sezgin O, Oguz D, Altintas E, et al. Endoscopic management of biliary obstruction caused by cavernous transformation of the portal vein. Gastrointest Endosc 2003;58: Berkelhammer C, Kortan P, Haber G. Endoscopic biliary prostheses as treatment for benign postoperative bile duct strictures. Gastrointest Endosc 1989;35: Heneghan M, Rockey D. Is tips shunt effective for secondary prophylaxis of gastric variceal hemorrhage? Gastroenterology 2003;125: Villanueva C, Minana J, Ortiz J, et al. Endoscopic ligation compared with combined treatment with nadolol and isosorbide mononitrate to prevent recurrent variceal bleeding. N Engl J Med 2001;345: Carbonell N, Pauwels A, Serfaty L, et al. Improved survival after variceal bleeding in patients with cirrhosis over the past two decades. Hepatology 2004;40: Address requests for reprints to: Dr Michael A. Heneghan, Consultant Hepatologist, Institute of Liver Studies, King s College Hospital, London SE5 9RS, England. michael.heneghan@kingsch.nhs.uk; fax: (44)