fax: (39) 02-50320723
Liver Failure and Liver Disease
Risk factors for thrombophilia in extrahepatic portal vein obstruction†
Article first published online: 22 FEB 2005
Copyright © 2005 American Association for the Study of Liver Diseases
Volume 41, Issue 3, pages 603–608, March 2005
How to Cite
Primignani, M., Martinelli, I., Bucciarelli, P., Battaglioli, T., Reati, R., Fabris, F., Dell'Era, A., Pappalardo, E. and Mannucci, P. M. (2005), Risk factors for thrombophilia in extrahepatic portal vein obstruction. Hepatology, 41: 603–608. doi: 10.1002/hep.20591
Conflict of interest: Nothing to report.
- Issue published online: 22 FEB 2005
- Article first published online: 22 FEB 2005
- Manuscript Accepted: 9 DEC 2004
- Manuscript Received: 10 AUG 2004
Scant information exists on the role of thrombophilia in extrahepatic portal vein obstruction (EHPVO). We studied 65 patients with EHPVO, 500 with deep vein thrombosis (DVT) of the lower limbs, and 700 healthy controls referred for thrombophilia screening, including the search for gain-of-function mutations in genes encoding coagulation factor V (factor V Leiden) and prothrombin (prothrombin G20210A); antithrombin, protein C, and protein S deficiency; and hyperhomocysteinemia. At least one abnormality in the thrombophilia screening was found in 40% of patients with either EHPVO or lower limb DVT and in 13% of controls, for odds ratios of 4.0 (95% CI, 2.3–7.0) and 4.4 (95% CI, 3.3–5.9), respectively. Statistically significant associations with EHPVO were observed for the prothrombin G20210A mutation (odds ratio, 8.1; 95% CI, 3.8–17.5) and the deficiencies of antithrombin, protein C, or protein S taken together (odds ratio, 4.5; 95% CI, 1.1–18.0). The odds ratio for the prothrombin G20210A was approximately twice that for lower limb DVT. Patients with factor V Leiden had an odds ratio for EHPVO of 0.8 (95% CI, 0.1–6.4) and for lower limb DVT of 7.5 (95% CI, 4.4–13.0). The odds ratio for EHPVO in patients with hyperhomocysteinemia was 2.0 (95% CI, 0.9–4.9). At variance with lower limb DVT, oral contraceptive use was not associated with an increased risk of EHPVO. Myeloproliferative disorders were diagnosed in 35% of patients with EHPVO. In conclusion, the risk for EHPVO is increased in the presence of thrombophilia resulting from the prothrombin G20210A mutation and from the deficiencies of the naturally occurring anticoagulant proteins, but not from factor V Leiden. (HEPATOLOGY 2005;41:603–608.)
In Western countries, extrahepatic portal vein obstruction (EHPVO) occurring in the absence of liver cirrhosis and portal invasion or constriction by a malignant tumor accounts for a small but not negligible proportion of cases of portal hypertension in adults.1 It has been known for many decades that myeloproliferative disorders, paroxysmal nocturnal hemoglobinuria, and abdominal inflammatory conditions2–4 are among the leading causes of EHPVO. More recently, inherited thrombophilia was associated with the occurrence of EHPVO, alone or in combination with the aforementioned risk factors. Among the causes of thrombophilia are the gain-of-function mutations in the genes encoding coagulation factor V (factor V Leiden G1691A) and prothrombin (prothrombin G20210A), the inherited deficiencies of anticoagulant proteins (antithrombin, protein C, and protein S), the antiphospholipid antibody syndrome, and the metabolic abnormality hyperhomocysteinemia.5 All these thrombophilic conditions consistently are associated with the most common clinical manifestation of venous thromboembolism; that is, deep vein thrombosis (DVT) of the lower limbs.5 In contrast, only a few studies, mainly small ones, have examined the role of thrombophilia in patients with EHPVO.4, 6–11 We performed a large case-control study of patients with EHPVO to investigate the spectrum of risk factors for the disease. We also investigated a parallel group of patients with DVT of the lower limbs to compare for the first time the magnitude of the risks in these two different localizations of venous thromboembolic disease.
Patients and Methods
We included in the study 65 patients with EHPVO that had occurred in the absence of overt neoplastic diseases or liver cirrhosis, selected on the basis of their referral for thrombophilia screening to the Thrombosis Center of the Ospedale Maggiore Policlinico of Milan between January 1994 and May 2003. Patients were asked to bring to the Center the diagnostic documentation of their thrombotic episodes. Diagnosis of EHPVO was made in 41 cases by Doppler ultrasound examination, in 15 by computed tomography, in 4 by magnetic resonance imaging, in 4 by arteriography, and in 1 during abdominal surgery. EHPVO involved only the portal vein in 19 patients and involved the mesenteric and/or splenic veins in 46 patients. Information on the risk factors for EHPVO present at the time of diagnosis was recorded. We considered as risk factors myeloproliferative disorders, inflammatory or infectious abdominal diseases (including diverticulitis, appendicitis, pancreatitis, duodenal ulcer, cholecystitis, ulcerative colitis, and Crohn's disease), surgery in the 3 months preceding the onset of symptoms, the intake of oral contraceptives, and pregnancy and puerperium (defined as the 3-month period after childbirth). These factors were considered to be associated with EHPVO only when their acute stage was concomitant to the event. In patients diagnosed at the late stage of portal cavernoma, the previous occurrence of these risk factors was, conservatively, not related to the event.
Because in patients with portal hypertension the criteria usually adopted to suspect chronic myeloproliferative disorders may be altered because of the presence of hypersplenism and increased plasma volume, we arbitrarily chose to suspect myeloproliferative disorders in the presence of at least one of the following in association with an enlarged spleen (longitudinal diameter more than 15 cm): platelet count higher than 300,000/mm2, hemoglobin higher than 15 g/dL in men and higher than 13 g/dL in women, and white blood cell count higher than 10,000/mm2 in the absence of infection. A bone marrow biopsy was carried out in 40 patients.
We also included in the study 559 patients with a first episode of lower limb DVT with or without pulmonary embolism. They were referred to the Thrombosis Center in the same study period as patients with EHPVO, and the objective documentation of their thrombotic episodes was reviewed by the same physicians. None of them had overt neoplastic or liver diseases. Fifty-nine patients were excluded because of uncertain diagnosis (n = 41) or incomplete thrombophilia screening (n = 18). Therefore, 500 patients with lower limb DVT were included in the final analysis. Diagnosis was made by compression ultrasonography or contrast venography. Symptomatic pulmonary embolism as complication of lower limb DVT occurred in 82 patients (16%) and was diagnosed by ventilation/perfusion lung scan or computed tomography. The presence of circumstantial risk factors for thrombosis, such as surgery in the 3 months preceding the event, trauma, prolonged immobilization, the use of oral contraceptives, and pregnancy and puerperium, were recorded.
Controls included 700 healthy individuals who were partners or friends of the patients and who agreed to be investigated with them at the Thrombosis Center. Previous thrombosis was excluded with a structured questionnaire validated for the retrospective diagnosis of thrombosis.12 At the time of blood sampling, controls were asked about the same risk factors for thrombosis recorded in patients. None of them had abnormal liver or renal function or overt neoplastic diseases. All patients and controls were Caucasian and gave their informed consent to the study, which was approved by the Institutional Review Board of the University of Milan.
DNA analysis for the 1691 guanine-to-adenine substitution in coagulation factor V gene (factor V Leiden) and for the 20210 guanine-to-adenine substitution in the 3′-untranslated region of the prothrombin gene was carried out as previously described.13, 14 Functional and/or antigenic assays for antithrombin, protein C, and protein S were carried out on plasma samples obtained from blood anticoagulated with sodium citrate (3.8% [wt/vol]), as previously described.15 Antiphospholipid antibodies (lupus anticoagulant and anticardiolipin antibodies) were diagnosed according to previously described methods16 that were carried out only in patients, because these tests are not part of the laboratory workup of our control population. Total plasma homocysteine was measured in blood samples anticoagulated with ethylenediaminetetraacetic acid after overnight fasting and 4 hours after an oral methionine load (3.8 g/m2 body surface area), according to previously described methods.17 Hyperhomocysteinemia was diagnosed when plasma levels of fasting total homocysteine or their postmethionine load increments above fasting levels exceeded the 95th percentile of the values obtained in the control group (19.25 nm/mL and 24.04 nm/mL in men and 14.88 nm/mL and 24.5 nm/mL in women, respectively).
Blood cell counts and liver function tests (serum aspartate aminotransferase, alanine aminotransferase, gamma-glutamyl transpeptidase, alkaline phosphatase, cholinesterase, albumin, and bilirubin) also were tested.
All analyses were performed with SPSS software (version 11, SPSS Inc., Chicago, IL). Continuous variables were expressed as median and range. Comparisons between groups were made by Mann-Whitney U test for continuous variables and the chi-square test for categorical variables. Odds ratios and 95% CIs were used as an approximate index of the relative risks. The odds ratios indicate the risk of EHPVO or lower limb DVT in the presence of a risk factor relative to the absence of that risk factor. Unconditional logistic regression analysis was used to adjust the odds ratios for possible confounders. A P value of less than .05 was the cutoff point for statistical significance.
Table 1 shows general characteristics, results of thrombophilia screening, and presence of risk factors in the study population. The two groups of patients, with EHPVO and with lower limb DVT, did not differ from controls in terms of age (at thrombosis for patients and at the visit for controls) and sex (63%, 55%, and 55% females, respectively). The median time elapsed from the event and the visit at the Thrombosis Center was 1 year (range, 3 months to 15 years) for patients with EHPVO and 8 months (range, 3 months to 12 years) for those with lower limb DVT. At the time of the visit, 30 patients (46%) with EHPVO had a portal cavernoma, whereas the remaining 35 did not.
|Patients With EHPVO||Patients With DVT With or Without PE||Controls|
|Median age, yr (range)||39 (15–66)||41 (8–80)||44 (12–84)|
|Body mass index, kg/m2 (mean ± SD)||23.1 ± 4.6||25.4 ± 4.9||24.2 ± 4.7|
|Thrombophilia abnormalities, n (%)|
|Any of the following*||26 (40)||199 (40)||91 (13)|
|Factor V Leiden||2 (3)||79 (16)||17 (2)|
|G20210A prothrombin||14 (22)||55 (11)||23 (3)|
|Antithrombin deficiency||2 (3)||8 (2)||4 (1)|
|Protein C deficiency†||0||7 (2)||0|
|Protein S deficiency†||2 (2)||5 (1)||2 (0.2)|
|Antiphospholipid antibodies||4 (6)||18 (4)||Not determined|
|Hyperhomocysteinemia||8 (12)||78 (16)||49 (7)|
|Other risk factors, n (%)|
|Myeloproliferative disorders||23 (35)||0||0|
|Inflammatory/infectious diseases||5 (8)||5 (1)||0|
|Immunomediated diseases||5 (8)||4 (1)||0|
|Surgery||6 (9)||61 (12)||0|
|Trauma or immobilization||0||84 (17)||0|
|Oral contraceptive use‡||7 (26)||115 (55)||73 (28)|
|Pregnancy/puerperium‡||1 (4)||40 (19)||0§|
Overall, at least one abnormality in the thrombophilia screening was found with the same frequency in patients with EHPVO or lower limb DVT (40%), contrasting with a lower frequency in controls (13%). Table 2 shows that compared with controls, the age- and sex-adjusted odds ratio for EHPVO in patients with any type of thrombophilia compared with those without was 4.0 (95% CI, 2.3–7.0) and that for those with lower limb DVT was 4.4 (95% CI, 3.3–5.9), both statistically significant. In 6 patients with EHPVO (9%), in 51 with lower limb DVT (10%), and in 4 controls (0.5%), more than one thrombophilic abnormality was identified. No homozygous carrier of factor V Leiden or G20210A prothrombin mutation was found in patients with EHPVO and controls, whereas among those with lower limb DVT homozygosity for either gain-of-function mutation was found in seven and two patients, respectively.
|EHPVO||DVT With or Without PE|
|Antithrombin, protein C, or protein S deficiency||7.7 (2.1–28.1)||4.5 (1.1–18.0)||4.8 (1.9–12.0)||4.8 (1.9–12.3)|
|Factor V Leiden||1.3 (0.3–5.6)||0.8 (0.1–6.4)||7.5 (4.4–12.9)||7.5 (4.4–13.0)|
|G20210A prothrombin||8.1 (3.9–16.6)||8.1 (3.8–17.5)||3.6 (2.2–6.0)||3.6 (2.1–6.1)|
|Hyperhomocysteinemia||1.9 (0.8–4.1)||2.1 (0.9–4.9)||2.5 (1.7–3.6)||2.6 (1.8–3.9)|
Table 2 also shows the odds ratios for EHPVO and lower limb DVT associated with the presence of each thrombophilia abnormality (inherited deficiencies of the anticoagulant proteins taken together, factor V Leiden, prothrombin mutation, and hyperhomocysteinemia). The hereditary nature of antithrombin, protein C, and protein S deficiencies was demonstrated with a family study (at least one first-degree relative had to have the same abnormality) carried out in all the deficient patients. Of the 18 patients with EHPVO and deficiency of one or more anticoagulant protein, only four had a hereditary deficiency, whereas the remaining 14, who had a nonhereditary deficiency presumably the result of the impairment of the liver synthetic function, were not considered to have thrombophilia. Each of the aforementioned thrombophilia abnormalities was associated with an increased risk of lower limb DVT, whereas only the inherited deficiencies of the anticoagulant proteins taken together and the prothrombin mutation increased the risk of EHPVO (Table 2). In particular, the odds ratio in the presence of the prothrombin mutation, taken as a measure of the relative risk for EHPVO, was nearly twice that for lower limb DVT. Hyperhomocysteinemia was associated with a two-fold increased risk of EHPVO, but the result did not reach statistical significance. In the eight patients with EHPVO, hyperhomocysteinemia was diagnosed on finding high fasting plasma levels of homocysteine, whereas in those with lower limb DVT, hyperhomocysteinemia was diagnosed on finding high fasting levels in 40 patients (51%) or high postmethionine load increments higher than fasting levels in 38 patients (49%).
Among the other risk factors (Table 1), the most common in patients with EHPVO was the presence of a myeloproliferative disorder, found in 23 patients (35%). Fourteen patients had essential thrombocythemia, three patients had polycythemia vera, two patients had idiopathic myelofibrosis, and in four patients, the myeloproliferative disorder affected more than one cellular line and was unclassified. No patient had signs of hemolysis, hemoglobinuria, or changes in blood cell counts suggestive of paroxysmal nocturnal hemoglobinuria. The use of oral contraceptives, present in a large proportion of women with lower limb DVT (55%), was recorded in a similar proportion of those with EHPVO and controls (26% and 28%, respectively), with an odds ratio of 3.3 (95% CI, 2.2–4.8) for lower limb DVT and 0.8 (95% CI, 0.3–2.0) for EHPVO. The distribution of the other risk factors (abdominal inflammatory or infectious diseases, surgery, and immunomediated diseases) was similar in patients with and without thrombophilia in the EHPVO (19% and 21%, respectively) and the lower limb DVT (30% and 25%, respectively) groups. The presence of abdominal inflammatory or infectious diseases and/or surgery was found in 10 (29%) of 35 patients diagnosed at the acute EHPVO stage and in 15% of those with lower limb DVT. The prevalence of thrombophilia was 39% and 41% in EHPVO patients with or without myeloproliferative disorders, abdominal inflammatory or infectious diseases, and surgery. In addition, five patients with EHPVO and four patients with lower limb DVT had immunomediated disorders (systemic lupus erythematosus, scleroderma, nodular vasculitis, Evan's syndrome, and celiac disease in the former, systemic lupus erythematosus in two patients and connectivitis in two additional patients in the latter). No risk factor for EHPVO was recognized in 15 patients (23%), whereas the presence of one risk factor only was recognized in 26 patients (40%), the presence of two risk factors was seen in 19 patients (29%), the presence of three risk factors was seen in 4 patients (6%), and the presence of four risk factors was seen in one patient (2%). Among EHPVO patients with more than one risk factor, the combined presence of thrombophilia and other risk factors was found in 12 patients (18%). In particular, 8 (35%) of the 23 patients with EHPVO and myeloproliferative disorders also had thrombophilia. The prevalence of patients with at least one risk factor was similar in EHPVO patients with (77%) or without (83%) portal cavernoma.
This study of a large number of patients with EHPVO and lower limb DVT investigated underlying risk factors, with a focus on thrombophilia. In addition to the inherited deficiencies of the anticoagulant proteins antithrombin, protein C, and protein S, and the gain-of-function mutations in coagulation factor V (factor V Leiden) and prothrombin, the metabolic abnormality hyperhomocysteinemia—evaluated only once before in patients with EHPVO11—also was investigated. The strongest thrombophilic risk factor for EHPVO was the G20210A prothrombin mutation, found in 22% of patients for an approximately eight-fold increased relative risk, which was approximately twice that found for lower limb DVT (four-fold). The latter estimate is consistent with that found in previous case-control studies of patients with lower limb DVT.5 In contrast, there was no association between EHPVO and factor V Leiden, which was associated with a seven-fold increased risk of lower limb DVT consistently in previous studies.5 Deficiencies of the anticoagulant proteins taken together and hyperhomocysteinemia also increased the relative risk of EHPVO, by factors of approximately five and two, respectively. Although the risks associated with the presence of antiphospholipid antibodies could not be estimated because the corresponding data were not obtained in controls, this thrombophilia marker was present in a similar proportion of patients with EHPVO (6%) and lower limb DVT (4%). As to the other risk factors, the most common was the presence of myeloproliferative disorders in 35% of patients with EHPVO. The relative risks associated with myeloproliferative disorders could not be calculated because changes in blood counts compatible with this diagnosis were not observed in the control population.
To date, thrombophilia was investigated in patients with EHPVO in six previous studies,4, 6–10 all of which have limitations because of their small sample size or lack of a control group. The only exception is the large case-control study of Janssen et al.8 that showed a weak association between EHPVO and factor V Leiden (odds ratio, 2.7; 95% CI, 1.1–6.9), but not with the prothrombin mutation (odds ratio, 1.4; 95% CI, 0.4–5.2). One of the reasons for the discrepancy on the role of each gain-of-function mutation between that and this study may be that in the latter, a selected population of patients without such important causes of EHPVO as cancer or liver cirrhosis was investigated. A low prevalence of factor V Leiden in patients with EHPVO found in this study is at variance with the findings of Janssen et al.8 but is consistent with other studies,4, 6, 10 in which the highest estimate of prevalence was 6%, approximately half of that reported in patients with lower limb DVT.5 This study shows that the metabolic abnormality hyperhomocysteinemia, which is usually more frequently acquired than inherited,18 is a mild risk factor for EHPVO, as it is for lower limb DVT. The measurement of fasting homocysteine seems to be sufficient to diagnose hyperhomocysteinemia in EHPVO patients, whereas in lower limb DVT patients, oral methionine load provides additional diagnostic information.19 At variance with lower limb DVT,5 the use of oral contraceptives was not associated with an increased risk for EHPVO. These findings strengthen the concept of specific risk factors for different thrombotic sites.20 For example, oral contraceptives alone increase the risk of lower limb DVT5 and cerebral vein thrombosis,15 but not that of upper limb DVT21 and of EHPVO. To our knowledge, the portal vein is the only venous thrombotic site in which the two gain-of-function mutations factor V Leiden and the prothrombin G20210A, both leading to blood hypercoagulability, seem to play a different role. However, the same discrepancy between the two mutations has been observed in the setting of arterial thrombosis. Postmenopausal women receiving hormone replacement therapy have a significantly higher risk of experiencing a first nonfatal myocardial infarction if they carry the prothrombin but not the factor V Leiden mutation.22 Similarly, an association has been observed between peripheral arterial disease and prothrombin G20210A, but not factor V Leiden.23
Among the limitations of this study is the fact that our population of patients was selected, because they were referred to a tertiary care center to be investigated for thrombophilia. Regarding the prevalence of circumstantial risk factors, such as abdominal inflammatory or infectious diseases, it may be that this was underestimated because these risk factors were not considered to be associated with the event in EHPVO patients diagnosed at the late stage of portal cavernoma (approximately half of this series), in whom the time of thrombosis could not be established. Moreover, whether hyperhomocysteinemia had a causal role in determining thrombosis cannot be established in a retrospective study like this, not only because the two-fold increase in relative risk did not reach statistical significance, but also because it cannot be ruled out that hyperhomocysteinemia was acquired as a consequence of an impairment of liver function associated with EHPVO.24 However, the possibility that, for the same reason, the deficiencies of the anticoagulant proteins antithrombin, protein C, and protein S may be acquired is likely to have been ruled out with the strict criteria adopted in this study of including only patients in whom at least one relative had the same deficiency.
In conclusion, we suggest that thrombophilia screening in patients with EHPVO should include DNA testing for prothrombin mutation, in addition to the measurements for antithrombin, protein C, and protein S. Whether the search for hyperhomocysteinemia and factor V Leiden is recommended in the diagnostic workup of patients with EHPVO requires further investigations, because the previous study of Janssen et al.8 found that there is an association between this gain-of-function mutation and EHPVO. Whether the knowledge of a thrombophilia status will help to optimize therapy in these patients remains to be assessed by specifically tailored studies.
- 1Portal vein thrombosis: a concise review. Am J Gastroenterol 2002; 97: 535–541., .Direct Link: