- Email: [email protected]
The epidemiology of Budd–Chiari syndrome in France
G Model YDLD-3722; No. of Pages 7
ARTICLE IN PRESS Digestive and Liver Disease xxx (2018) xxx–xxx
Contents lists available at ScienceDirect
Digestive and Liver Disease journal homepage: www.elsevier.com/locate/dld
Liver, Pancreas and Biliary Tract
The epidemiology of Budd–Chiari syndrome in France Isabelle Ollivier-Hourmand a,∗,1 , Manon Allaire a,b,1 , Nathalie Goutte c , Rémy Morello d , Carine Chagneau-Derrode e , Odile Goria f , Jerôme Dumortier g , Jean Paul Cervoni h , Sébastien Dharancy i , Nathalie Ganne-Carrié j,k , Christophe Bureau l , Nicolas Carbonell m , Armand Abergel n , Jean Baptiste Nousbaum o , Rodolphe Anty p , Hélène Barraud q , Marie Pierre Ripault r , Victor De Ledinghen s , Anne Minello t , Frédéric Oberti u,v , Sylvie Radenne w , Noelle Bendersky x , Olivier Farges y , Isabelle Archambeaud z , Anne Guillygomarc’h A , Marie Ecochard B , Violaine Ozenne C , Marie Noelle Hilleret D , Eric Nguyen-Khac E , Barbara Dauvois F , Jean Marc Perarnau G , Pascale Lefilliatre H , Jean Jacques Raabe I , Michel Doffoel J , Jean Philippe Becquart K , Eric Saillard L , Dominique Valla x,M , Thong Dao a , Aurélie Plessier x,M , for the French Network for Vascular Disorders of the Liver a
Hepato Gastroenterology Nutrition Department, University Hospital Center Côte de Nacre, Caen, France Inserm Unit U1149, Research Center on Inflammation, Paris, France UMRS1193 and University Hospitals of Paris Sud, France d Public Health Unit, University Hospital Center Côte de Nacre, Caen, France e Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire, Poitiers, France f Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Charles Nicolle, Rouen, France g Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Edouard Herriot, Lyon, France h Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Jean Minjoz, Besanc¸on, France i Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Claude Huriez, Lille, France j Service d’hépatologie, Centre Hospitalo-Universitaire Hôpital Jean Verdier, APHP, Bondy, France k Inserm UNR-1162, « Génomique Fonctionnelle des Tumeurs Solides », Paris, France l Service d’Hépatologie, Centre Hospitalo-Universitaire Purpan, Toulouse, France m Service d’Hépatologie, Centre Hospitalo-Universitaire Saint Antoine, APHP, Paris, France n Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Estaing, Clermont Ferrand, France o Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire La Cavale Blanche, Brest, France p Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire, Nice, France q Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire de Nancy, Vandœuvre-lès-Nancy, France r Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Saint Eloi, Montpellier, France s Service d’Hépatologie, Hôpital Haut-Lévêque, Bordeaux, France t Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire de Dijon Bourgogne, Dijon, France u Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire, Angers, France v Laboratoire HIFIH, Université Bretagne Loire, Angers, France w Service d’Hépato-gastroentérologie, Hôpital de la Croix Rousse, Lyon, France x Service d’hépatologie, Hôpital Beaujon, APHP, Clichy-la-Garenne, France y Service de Chirurgie Hépato-Pancréato-Biliaire, Hôpital Beaujon, APHP, Clichy-la-Garenne, France z Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Hôtel-Dieu, Nantes, France A Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Pontchaillou, Rennes, France B Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire, Saint-Etienne, France C Service d’Hépatologie, Hôpital Lariboisière, APHP, Paris, France D Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Michallon, Grenoble, France E Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire, Amiens, France F Service d’Hépato-gastroentérologie, Centre Hospitalier Régional, Orléans, France G Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire, Tours, France H Service d’Hépato-gastroentérologie, Centre Hospitalier Général, Le Havre, France b c
∗ Corresponding author at: Service d’Hépato-Gastroentérologie et de Nutrition, CHU de Caen, Avenue Côte de Nacre, FR 14033 Caen, France. E-mail address: [email protected] (I. Ollivier-Hourmand). 1 Equal contributors. https://doi.org/10.1016/j.dld.2018.04.004 1590-8658/© 2018 Editrice Gastroenterologica Italiana S.r.l. Published by Elsevier Ltd. All rights reserved.
Please cite this article in press as: Ollivier-Hourmand I, et al. The epidemiology of Budd–Chiari syndrome in France. Dig Liver Dis (2018), https://doi.org/10.1016/j.dld.2018.04.004
G Model
ARTICLE IN PRESS
I
Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Metz-Thionville, Ars-Laquenexy, France Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire, Strasbourg, France Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Saint Pierre, La Réunion, France L Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Pointe à Pître, Guadeloupe, France M Centre de Recherche de l’Inflammation, Inserm UMR-1149, Université Paris Diderot, Paris, France J
K
a r t i c l e
i n f o
Article history: Received 30 November 2017 Accepted 3 April 2018 Available online xxx Keywords: Hepatic vein obstruction Incidence National database Questionnaire survey Risk factors
a b s t r a c t
Introduction: Epidemiological data is lacking on primary Budd–Chiari syndrome (BCS) in France. Methods: Two approaches were used: (1) A nationwide survey in specialized liver units for French adults. (2) A query of the French database of discharge diagnoses screening to identify incident cases in adults. BCS associated with cancer, alcoholic/viral cirrhosis, or occurring after liver transplantation were classified as secondary. Results: Approach (1) 178 primary BCS were identified (prevalence 4.04 per million inhabitants (pmi)), of which 30 were incident (incidence 0.68 pmi). Mean age was 40 ± 14 yrs. Risk factors included myeloproliferative neoplasms (MPN) (48%), oral contraceptives (35%) and factor V Leiden (16%). None were identified in 21% of patients, ≥2 risk factors in 25%. BMI was higher in the group without any risk factor (25.7 kg/m2 vs 23.7 kg/m2 , p < 0.001). Approach (2) 110 incident primary BCS were admitted to French hospitals (incidence 2.17 pmi). MPN was less common (30%) and inflammatory local factors predominated (39%). Conclusion: The entity of primary BCS as recorded in French liver units is 3 times less common than the entity recorded as nonmalignant hepatic vein obstruction in the hospital discharge database. The former entity is mostly related to MPN whereas the latter with abdominal inflammatory diseases. © 2018 Editrice Gastroenterologica Italiana S.r.l. Published by Elsevier Ltd. All rights reserved.
1. Introduction Budd–Chiari syndrome (BCS) is defined as an obstruction of the hepatic venous outflow tract located anywhere from the small hepatic venules to the entrance of the inferior vena cava into the right atrium, with the exclusion of cardiac and pericardial diseases, and sinusoidal obstruction syndrome/veno-occlusive disease [1]. Primary BCS is characterized by obstruction due to thrombosis or phlebitis, which usually excludes malignant invasion or prior cirrhosis. The available estimates of the incidence and prevalence of primary BCS in European countries are based on either small samples of individual cases collected from hospital liver units over a long period or from large queries of national databases of hospital discharge diagnoses [2–5]. Recent estimates differ widely depending on the studies, which are also markedly different. In France, an unpublished survey performed in academic liver units recorded 20 new cases of primary BCS in the adult population, corresponding to an estimated incidence of 0.36 per million inhabitants (pmi) in 1990 [5]. On the other hand, the estimated incidence in a recent population based study of hospital discharge diagnoses in Northern Italy was 6 times greater, with 2.1 pmi per year between 2000 and 2012 [4]. Because of these limited and conflicting data, the aim of the present study was to estimate the incidence, the prevalence and the main characteristics of primary BCS in French adult residents based on two different approaches: a nationwide questionnaire in specialized hospital liver units, and a query in the French database of hospital discharge diagnoses.
2. Materials and methods 2.1. Questionnaire Between January 1st and December 31st, 2010, we performed a declarative survey in all 32 French academic liver units including 14 groups in Ile de France (an administrative area that includes 18% of the French population), as well as in 16 large non-academic liver units throughout the country (Supplemental material 1). These liver units function as regional reference centers for vascular liver diseases since the French Ministry of Health created a nationwide network in 2005. We recorded incident BCS and the prevalence of primary BCS in inpatients or outpatients. Patients under 18 years of age who were not French residents, who developed BCS after liver transplantation, and those with malignancies except for myeloproliferative neoplasms (MPN) were excluded. The date of the first imaging test confirming the diagnosis of BCS was considered to be the date of the initial diagnosis. Characteristics at diagnosis were recorded using a standardized questionnaire, including the risk factors for venous thrombosis [1]. Deficiencies in protein C, protein S and anti-thrombin were not considered to be risk factors because of the difficulty of confirming them as primary defects in patients with liver diseases [1,6]. For similar reasons, antiphospholipid syndrome was only considered to be a risk factor if it was diagnosed before BCS or if systemic lupus was present. This observational survey received approval of the Commission nationale de l’informatique et des libertés (CNIL) and Comité de protection des personnes (CPP) – Nord-Ouest France. Incidences and prevalences were calculated using the 2010 population census performed by the Institut National de la Statistique et des Etudes Economiques (INSEE), for residents of France who were 18 years or older. Incident cases of BCS were defined as cases whose diagnosis was confirmed in 2010.
Please cite this article in press as: Ollivier-Hourmand I, et al. The epidemiology of Budd–Chiari syndrome in France. Dig Liver Dis (2018), https://doi.org/10.1016/j.dld.2018.04.004
G Model YDLD-3722; No. of Pages 7
ARTICLE IN PRESS I. Ollivier-Hourmand et al. / Digestive and Liver Disease xxx (2018) xxx–xxx
2.2. Search in the discharge diagnosis database for incident cases The French database of hospital discharge diagnoses (Programme de Medicalisation des Systèmes d’Information – PMSI) was queried to identify French residents 18 years or older with a new diagnosis of BCS between January 1st, 2007 and December 31st, 2012. In PMSI, each inpatient has a unique anonymous alphanumerical identification number that traces admissions in private and public hospitals throughout the country which excludes duplicates, and allows an exhaustive search. Discharge abstracts include the ICD-10 codes of main and associated diagnoses [7]. BCS was considered to be an incident case if the ICD-10 code I82.0 was recorded for the first time in 2012. 2012 was considered rather than 2010 to increase the washout period (2007–2012) without prior notification of code I82.0. In addition we performed a comparison of ages at diagnosis, genders, and regions between 2010 and 2012 in order to exclude duplicates. BCS cases were classified into the following 2 mutually exclusive categories: secondary BCS, associated with ICD-10 codes of cancer, alcoholic and viral cirrhosis or liver transplantation (to exclude secondary BCS due to transplant related venous outflow blockade), at any time between 2007 and 2012; and primary BCS from 2007 to 2012, when ICD-10 codes for secondary BCS were absent. The incidence was calculated using 2012 INSEE national census. This observational study was approved by CNIL and CPP – Ile de France. 2.3. Statistical analyses Quantitative variables were described using means and standard deviations (SD). Qualitative or categorical (QoC) variables were summarized using percentages. The Chi-square test was used to analyze QoC data and the Fischer’s exact test or the Fisher–Freeman–Halton test when the use conditions of the Chisquare test were not always verified. Means were compared with the Student-t test. Levene’s test was used to check the equality of variance. Most measures were given with their 95% confidence intervals. For the percentages, the Exact method was used, except where the ends of the confidence intervals are close to 0 or 1: in this case, the Score method provides more precise endpoints. All the tests were two tailed and the probability significance threshold ® ® was set at p < 0.05. IBM -SPSS 22.0 and R software (version 3.0.3; R Foundation for Statistic Computing, Vienna, Austria; http://www. r-project.org) for Windows were the statistical software used. 3. Results 3.1. Estimates of incidence and prevalence based on the 2010 survey Forty of the 48 selected centers completed the survey in 2010. A total of 186 cases of primary BCS were recorded. One incident case and 7 additional prevalent cases from 2 centers were excluded because of insufficient participation at the regional level. Thirty of the 178 primary BCS cases included in the cohort were in newly diagnosed symptomatic in-patients (Fig. 1). The resident population which was 18 years and older, corresponding to the 38 remaining centers in 2010, was 44.068.082 inhabitants. In this population, the incidence was 0.68 pmi per year and the prevalence was 4.04 pmi. Regional incidences varied from 0 to 2.12 pmi per year (Supplementary data 1 and 2). 3.2. Estimate of the incidence according to the discharge diagnosis database Between 2007 and 2012, 1666 in-patients were identified with the code I82.0. Two-hundred and eight of them were identified for
3
Table 1 Main characteristics of patients at the time of diagnosis in 2010 survey in hospital liver units (n = 178). Available data
Total
Age, yearsb Male gendera Body mass indexa Arterial hypertensiona Diabetesa Incident casesa
173 173 147 172 172 178
40.2 ± 13.9 53 (30.6) 24.1 ± 5.1 10 (5.8) 3 (1.7) 30 (16.9)
Location of thrombosis Isolated hepatic veina Combined hepatic vein and inferior vena cavaa Associated portal vein thrombosisa
158 158 158
121 (76.6) 37 (23.4) 5 (3.2)
Clinical manifestations Ascitesa Hepatomegalya Abdominal paina Oesophageal varicesa Splenomegalya Jaundicea Fevera Hydrothoraxa
164 164 156 135 159 133 154 154
122 (74.4) 115 (70.1) 113 (72.4) 74 (54.8) 78 (49.1) 27 (20.3) 31 (20.1) 20 (13.0)
Laboratory data Factor V 70–100%a 50–70%a <50%a
114 114 114
52 (45.6) 28 (24.6) 34 (29.8)
120 120
87 (72.5) 33 (27.5)
120 120
94 (78.3) 26 (21.7)
120 120
97 (80.8) 23 (19.2)
Total bilirubin <50 mol/La ≥50 mol/La ALT <5Na ≥5Na AST <5Na ≥5Na a b
Number (percentage) of patients. Median.
the first time in 2012, including 110 cases of primary BCS (Fig. 2). The census population over 18-years old who were residents in France in 2012 was 50.805.530 million inhabitants, and the estimated incidence in 2012 was 4.10 pmi for all cases, and 2.17 pmi for primary BCS. Regional incidences varied from 0 to 4.51 pmi per year (Supplementary data 3). 3.3. Characteristics and risks factors of BCS in hospital liver units in 2010 A total of 173/178 included cases of primary BCS had a complete set of data. The first cases were diagnosed in 1979, while 146 cases were diagnosed after 2000. The diagnosis was confirmed within the 7 days after the initial clinical manifestations in 9.9% of cases, and more than one month after the initial manifestations in 48.4% (Fig. 3). The main characteristics at diagnosis and risk factors are presented in Tables 1 and 2. Only one risk factor was identified in 94/173 patients with primary BCS, (54.3%), 2 risk factors in 31 (17.9%), ≥3 risk factors in 12 (7.0%), and none in 36 (20.8%). MPN was diagnosed in 72/151 patients (47.7%), based on the detection of the JAK 2 V617F mutation in 55 out of 139 tested, both bone marrow biopsy and endogenous erythroid colonies in 7, bone marrow biopsy only in 8, and endogenous erythroid colonies only in 2. Among BCS patients with MPN, 44 had polycythemia vera, 20 essential thrombocytemia, 7 latent MPN, and 1 myelofibrosis. MPN was diagnosed before BCS in 9 patients and none of them had received cytoreductive therapy. The mean body mass index (BMI) was 24.1 kg/m2 , higher in the group without than
Please cite this article in press as: Ollivier-Hourmand I, et al. The epidemiology of Budd–Chiari syndrome in France. Dig Liver Dis (2018), https://doi.org/10.1016/j.dld.2018.04.004
G Model YDLD-3722; No. of Pages 7 4
ARTICLE IN PRESS I. Ollivier-Hourmand et al. / Digestive and Liver Disease xxx (2018) xxx–xxx
Fig. 1. Flow chart for the inclusion and exclusion of cases into the survey performed in 2010 in hospital liver units.
Fig. 2. Flow chart for the inclusion and exclusion of cases into the survey 2012 from the queries addressed to the national hospital discharge diagnosis data base.
in that with identified risk factors (25.7 kg/m2 and 23.7 kg/m2 , respectively, p < 0.001), without difference in ascites frequency (70,6% (24/34) vs 75,4% (98/130), p = 0,66). Fifty percent (16/30) of patients without any risk factors had a BMI >25 kg/m2 compared to 26.9% (31/113) in those with at least one risk factor (p = 0.009).
3.4. Characteristics and risk factors for primary BCS in hospital discharge diagnosis database in 2012 The median age was 46.3 ± 18.4 years old in patients with primary BCS and 60.7 ± 15.1 years old in those with secondary BCS (p < 0.001). There were 68.1% of women with primary BCS, but only 25.5% with secondary BCS (p = 0.014).
Please cite this article in press as: Ollivier-Hourmand I, et al. The epidemiology of Budd–Chiari syndrome in France. Dig Liver Dis (2018), https://doi.org/10.1016/j.dld.2018.04.004
G Model
ARTICLE IN PRESS
YDLD-3722; No. of Pages 7
I. Ollivier-Hourmand et al. / Digestive and Liver Disease xxx (2018) xxx–xxx
5
Fig. 3. Delay to diagnosis in patients from hospital liver units.
Table 2 Risk factors patients in 2010 survey in hospital liver units (n = 178). Available data
Total
Acquired thrombophilia Myeloproliferative diseasea Paroxysmal nocturnal hemoglobinuriaa Behcet diseasea Hyperhomocysteinemiaa Systemic diseasea , c
151 112 101 118 173
72 (47.7) 10 (8.9) 6 (5.9) 9 (7.6) 14 (8.1)
Inherited thrombophilia Factor V Leidena Factor II G20210A mutationa MTHFR C677T mutationa Thalassemiaa
120 117 60 173
19 (15.8) 6 (5.1) 7 (11.7) 1 (0.6)
173
10 (5.8)
103 120
36 (35.0) 2 (1.7)
Local factora , b Hormonal risk factor Recent oral contraceptive usea Pregnancya a
Number (percentage) of patients. Including hepatic abscess (n = 3), hydatid cyst (n = 3), angiomatosis (n = 2), abdominal trauma (n = 1) and schistosomiasis (n = 1). c Including inflammatory bowel disease (n = 4), Takayasu disease (n = 1), sarcoidosis (n = 1), disseminated lupus erythematosus (n = 4), coeliac disease (n = 2), paludism (n = 2). b
Forty-seven (42.7%) of the 110 patients with primary BCS, had an associated ICD-10 code for conditions likely to increase the risk of hepatic vein thrombosis, including 14 (29.8%) MPN; 8 (17%) thrombophilia including lupus and other coagulation disorders; 5 (10.6%) rare diseases including Takayasu disease, sarcoidosis, Behcet disease, paroxysmal nocturnal haemoglobinuria (PNH), histiocytosis, familial Mediterranean fever, thalassemia, or sickle cell disease; 7 (14.9%) pregnancy; 1 (2.1%) abdominal trauma; and 18 (38.3%) local factors including inflammatory bowel disease, pancreatitis, cholecystitis, liver abscess, parasitosis and polycystic liver disease. In the 98 cases of secondary BCS, 9 (9.2%) had an associated ICD10 code for primary tumor of the liver excluding hepatocellular carcinoma (HCC), 18 (18.4%) HCC; 21 (21.4%) for a secondary tumor of the liver; 26 (26.5%) cancer without evidence of liver tumor; 19 (19.4%) alcoholic or viral cirrhosis without HCC; 10 (10.2%) liver graft complications. 6 (6.1%) MPN; 3 (3.1%) thrombophilia; 1 (1%) abdominal trauma; and 9 (9.1%) abdominal inflammatory disease associated with one of the previous codes. 4. Discussion This study based on a survey of hospital units participating in the nationwide network for vascular liver diseases shows that primary BCS is a rare condition in France because only 30 incident cases were identified in 2010, in an area with 44 million adult inhabi-
tants. This corresponds to an incidence rate of 0.68 pmi in 2010. However, in 2012, the nationwide discharge-diagnosis database recorded 110 new cases of primary BCS in a population of almost 51 million adult inhabitants, for an incidence rate of 2.17 pmi per year, which is 3 times higher than the previous estimate. Other findings in this survey suggest that these estimates differ because they apply to different entities, rather than different counts for the same disease. This hypothesis may be one of explanations to discrepancies between BCS incidence estimations around the world. The estimate obtained from the survey in the hospital liver units, 0.68 pmi in year 2010, applies to the entity presenting as a predominant liver disease that is often associated with general prothrombotic conditions. These 30 new patients identified in unselected but specialized liver units in the year 2010 must be compared to the 35/163 patients prospectively included in the multicenter European (En-Vie) cohort by French centers between October 2003 and October 2005 [8]. The pattern of obstructed veins and the associated prothrombotic conditions were as described in the EN-Vie cohort [8,9] and the predominance of MPN as a cause is fully and independently confirmed in the present survey. This survey also shows that local factors explaining why thrombosis develops at these unusual sites, are rare [9]. Thus, the present findings suggest that observations from the EN-Vie cohort study performed in tertiary referral units can be extrapolated nationally to patients who are seen in initial or secondary referral liver units. In other words, this survey demonstrates that results obtained in a survey among selected tertiary referal centers in Europe, are applicable to individual European countries. On the other hand, the queries of the national discharge diagnosis database identified a broader entity that was 3 times more frequent (2.17 incident cases pmi in 2012) suggesting that there was selection of BCS patients seen in hospital liver units compared to those admitted to hospitals for all causes. Spurious overestimation is unlikely because the recording system excludes duplicates and the ICD-10 code I82.0 is highly specific. Of note, an Italian survey of similar design, yielded very similar results [4]. It is interesting to note that local factors were identified in a greater percentage of cases than previously reported in series from hospital liver units, even after exclusion of cirrhosis of other origins and cancer. Indeed, a code for a concomitant disease that would probably facilitate the development of hepatic vein thrombosis through local mechanisms was recorded in 25% of these cases but in only 11% in the EN-Vie cohort [8,9]. The factor of abdominal inflammation played a major role, as reported in similar studies in smaller populations and longer periods of inception [4]. Finally it should be emphasized that 30.2% of all cases (57.2% of primary BCS) were not associated with a diagnostic code of a risk factor for BCS. This is a higher percentage than in the EN-Vie cohort [9] but is in between the findings
Please cite this article in press as: Ollivier-Hourmand I, et al. The epidemiology of Budd–Chiari syndrome in France. Dig Liver Dis (2018), https://doi.org/10.1016/j.dld.2018.04.004
G Model
ARTICLE IN PRESS
YDLD-3722; No. of Pages 7
I. Ollivier-Hourmand et al. / Digestive and Liver Disease xxx (2018) xxx–xxx
6
Table 3 Main characteristics of epidemiological surveys on Budd–Chiari syndrome.
Study period Age limit – years Malignancy excluded Population based Retrospective Prospective Declarative Inpatients Outpatients Incident cases – n Study population – 106 Incidence – pmi per year Patients (n) Prevalence pmi
Japan [10]
Denmark [2]
Franceb
Sweden [3]
China [11]
South Korea [12]
Franceb
Franceb
Italy [4]
1989 None Yes No Yes Yes Yes Yes No 21 123 0.17 160 2.4
1981–1985 None Yes Yes Yes No No Yes No 13 5 0.5 NAa NAa
2010 ≥18 Yes No Yes Yes Yes Yes Yes 30 44 0.68 178 4.04
1990–2001 None No Yes Yes No No Yes Yes 12 1.3 0.8 43 1.4
2000–2013 None No No Yes No No NA NA 20191 1300 0.88 NAa 7.69
2009–2013 None No Yes Yes No Yes Yes Yes 424 50 0.87 NAa 5.29
2012 ≥18 Yes Yes Yes No No Yes No 110 51 2.17 NAa NAa
2012 ≥18 No Yes Yes No No Yes No 208 51 4.1 NAa NAa
2000–2012 ≥20 No Yes Yes No No Yes No 287 13 2 NAa NAa
[n]
Reference of the paper. Not available. b Current survey. a
of similar searches of population-based databases in Northern Italy (61% of noncirrhotic nonmalignant BCS cases) [4] and Sweden (37% of noncirrhotic nonmalignant BCS) [3]. Because of the design of these studies, it is impossible to determine to what extent associated conditions were overlooked. Our findings and those of similar population-based studies using a hospital discharge diagnosis suggest that obstruction of hepatic venous outflow is a frequent but ancillary condition, that is, merely associated with a predominantly extrahepatic condition of the liver. The latter entity is likely not managed in specialized liver units, but probably in surgery units. As shown in Table 3, current and previously reported estimated incidences of BCS fall into 2 groups: one group with an incidence between 0.17 and 0.88 (present 2010 survey, [2,3,10,11,12]); and another with more than 2.0 (present 2012 survey, [4]). The two groups have different sample sizes and background populations of individual studies. Survey designs in the first group were based on a careful review of cases by liver specialists, except for the Danish study (performed when accurate noninvasive imaging was not available). In the lower incidence group, the estimate of our current survey in hospital liver units (0.68) is probably one of the most representative because of the large size of the nationwide population. Estimates from Sweden were based on only 15% of Swedish population [3]. Although these hospital liver unit surveys were performed at different dates, the estimates are relatively close, and similar to recent data suggesting a stable incidence of BCS since 2000. It is interesting to note that this group includes surveys in China and Korea, two Asian countries which were considered to have a high incidence but for which solid evidence was lacking [11,12]. The other group of studies includes 2 large population-based analyses of hospital discharge diagnoses with a similar design, performed at the same time in Italy and France, respectively. Identification of these cases was based on principal and secondary diagnoses. Thus, secondary diagnoses of BCS correspond to conditions in which obstructed hepatic veins were considered ancillary to an otherwise severe condition such as acute pancreatitis or inflammatory bowel disease. The Italian study did not individualize nonmalignant and noncirrhotic BCS, and covered only 22% of Italian population. If we apply similar inclusion criteria our overall estimated incidence of secondary and primary BCS was 4.1 pmi in 2012, twice the Italian estimate of 2.1 pmi per year, while our estimated incidence of primary BCS was 2.17. The third study using a hospital discharge diagnosis database was Danish and evaluated nonmalignant BCS. In that study the incidence of BCS (0.5 pmi per year) was probably underestimated because of a lack of sensitivity of liver vascular imaging between 1981 and 1985. Thus, even in the absence of cirrhosis or malignancy, hepatic vein obstruction
appears to be 2–4 times more common than the clinical entity of BCS managed in hospital liver units. Our survey of a recent, large population of consecutive patients enrolled in hospital liver units over a short and recent period in particular 2010, provides a realistic world picture of primary BCS in France, close to previous surveys [3,8,9,13–15,16,17]. The delay in the diagnosis of primary BCS patients was more than 6 months after the first clinical manifestations in 15.2% and more than one year in 5.9%. These findings suggest the presence of either unequal access to specialized care, or difficulties in diagnosis. To date the risk factors of primary BCS have mainly been analyzed in retrospective or multicenter studies making it difficult to assess selection biases [3,9,13–19]. The present study in 173 patients is representative of referrals to liver units nationwide. Our results confirm the multifactorial etiology of BCS. Indeed 24.9% of patients had at least 2 risk factors. Not surprisingly, MPN was present in 47.7%, factor V Leiden mutation in 15.8%, PNH in 8.9% and Behcet in 5.9%. Certain of the findings in this study are not in agreement with prior series. The proportion of oral contraceptive users in the survey (35.0% of women of child-bearing age) is less to or close to that in the general French population of women of the same age (47.5% in 2010 [20]). Despite the possibility of a lack of patient declaration, this suggests that current contraceptives are less of a risk factor of BCS. On the other hand, the present study shows for the first time that being overweight may be a new risk factor of primary BCS, since overweight was twice more common when no usual risk factor was identified without difference in ascites frequency. A recent study support this point in portal vein thrombosis with overweight and metabolic syndrome observed in respectively 82.0% and 47% of patients with idiopathic portal vein thrombosis vs 51% and 17% in the control group (p = 0.01). Thus, like in portal vein thrombosis with or without cirrhosis, overweight could impact on hepatic venous outflow thrombosis: one explanation could be an increase of plasminogen activator inhibitor 1 level [21,22]. In our study the result must be interpreted cautiously since the volume of ascites was not evaluable, and complete data concerning metabolic syndrome not available. Further studies are needed to confirm this hypothesis and analyze thoroughly metabolic syndrome in patients with and without a risk factor for thrombosis.
5. Conclusion In France, primary BCS is a rare entity when it is defined as a predominantly hepatic disease related to underlying prothrombotic conditions (incidence in liver units 0.68 pmi) whose main characteristics have been widely reported in large European multicenter
Please cite this article in press as: Ollivier-Hourmand I, et al. The epidemiology of Budd–Chiari syndrome in France. Dig Liver Dis (2018), https://doi.org/10.1016/j.dld.2018.04.004
G Model YDLD-3722; No. of Pages 7
ARTICLE IN PRESS I. Ollivier-Hourmand et al. / Digestive and Liver Disease xxx (2018) xxx–xxx
studies. In contrast, the entity recorded in the national database as nonmalignant noncirrhotic hepatic vein obstruction is 3 times more common and frequently accompanies local abdominal inflammatory diseases. The possible role of overweight as a risk factor for BCS merits further attention. Current oral contraception might be associated with a lesser risk of BCS. Conflict of interest None declared. Study concept and design Ollivier-Hourmand, Plessier, Valla. Acquisition of data and Critical revision of the manuscript for important intellectual content and final approval of the version to be published Ollivier-Hourmand, Allaire, Abergel, Anty, Archambeaud, Armand, Barraud, Becquart, Bureau, Carbonell, Chagneau-Derrode, Cervoni, Dharancy, Dauvois, De Ledinghen, Doffoel, Dumortier, Ecochard, Ganne-Carrié, Goria, Goutte, Guillygomarc’h, Hilleret, Lefilliatre, Nousbaum, Minello, Farges, Nguyen-Khac, Oberti, Ozenne, Perarneau, Raabe, Radenne, Ripault, Saillard. Analysis and interpretation of data and Drafting of the manuscript Ollivier-Hourmand, Allaire, Dao, Plessier, Valla. Statistical analysis Ollivier-Hourmand, Allaire, Morello, Goutte, Bendersky. Acknowledgments for technical assistance Delaval G, Zekrini K, Habersetzer F, Heurgues A, Cadranel JF, Sogni P, Rosa I, Grangé JD, Gornet JM, Parlier D, Del Tedesco E, Lombard F, Falissard B. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.dld.2018.04.004.
7
References [1] European Association for the Study of the Liver. EASL Clinical Practice Guidelines: vascular diseases of the liver. J Hepatol 2016;64(1):179–202. [2] Almdal TP, Sørensen TI. Incidence of parenchymal liver diseases in Denmark, 1981 to 1985: analysis of hospitalization registry data. The Danish Association for the Study of the Liver. Hepatology 1991;13:650–5. [3] Rajani R, Melin T, Björnsson E, Broomé U, Sangfelt P, Danielsson A, et al. BuddChiari syndrome in Sweden: epidemiology, clinical characteristics and survival – an 18-year experience. Liver Int 2009;29:253–9. [4] Ageno W, Dentali F, Pomero F, Fenoglio L, Squizzato A, Pagani G, et al. Incidence rate and case fatality rates of portal vein thrombosis and Budd-Chiari Syndrome. Thromb Haemost 2017;117:1–7. [5] Valla DC. Hepatic venous outflow tract obstruction etiopathogenesis: Asia versus the West. J Gastroenterol Hepatol 2004;19:S204–11. [6] Qi X, Han G, Guo X, De Stephano V, Xu K, Lu Z, et al. Review article: the aetiology of primary Budd-Chiari syndrome – differences between the West and China. Aliment Pharmacol Ther 2016;44(11–12):1152–67. [7] ICD-10 Version 2016. [Internet]. [Cited 2015 October 12] Available from: http:// apps.who.int/classifications/icd10/browse/2016/en. [8] Seijo S, Plessier A, Hoekstra J, Dell’era A, Mandair D, Rifai K, et al. Good long-term outcome of Budd-Chiari syndrome with a step-wise management. Hepatology 2013;57:1962–8. [9] Darwish Murad S, Plessier A, Hernandez-Guerra M, Fabris F, Eapen CE, Bahr MJ, et al. Etiology, management, and outcome of the Budd-Chiari syndrome. Ann Intern Med 2009;151:167–75. [10] Okuda H, Yamagata H, Obata H, Iwata H, Sasaki R, Imai F, et al. Epidemiological and clinical features of Budd-Chiari syndrome in Japan. J Hepatol 1995;22:1–9. [11] Zhang W, Qi X, Zhang X, Su H, Zhong H, Shi J, et al. Budd-Chiari Syndrome in China: A Systematic Analysis of Epidemiological Features Based on the Chinese Literature Survey. Gastroenterol Res Pract 2015;2015:738548. [12] Ki M, Choi HY, Kim KA, Kim BH, Jang ES, Jeong SH. Incidence, prevalence and complications of Budd-Chiari syndrome in South Korea: a nationwide, population-based study. Liver Int 2016;36(7):1067–73. [13] Garcia-Pagán JC, Heydtmann M, Raffa S, Plessier A, Murad S, Fabris F, et al. TIPS for Budd-Chiari syndrome: long term results and prognostics factors in 124 patients. Gastroenterology 2008;135:808–15. [14] Mentha G, Giostra E, Majno PE, Bechstein WO, Neuhaus P, O’Grady J, et al. Liver transplantation for Budd-Chiari syndrome: A European study on 248 patients from 51 centres. J Hepatol 2006;44:520–8. [15] Darwish Murad S, Valla DC, De Groen PC, Zeitoun G, Hopsman JA, Haagsma EB, et al. Determinants of survival and the effect of portosystemic shunting in patients with Budd-Chiari syndrome. Hepatology 2004;39:500–8. [16] Plessier A, Sibert A, Consigny Y, Hakime A, Zappa M, Denninger MH, et al. Aiming at minimal invasiveness as a therapeutic strategy for Budd-Chiari syndrome. Hepatology 2006;44:1308–16. [17] Moucari R, Rautou PE, Cazals-Hatem D, Geara A, Bureau C, Consigny Y, et al. Hepatocellular carcinoma in Budd-Chiari syndrome: characteristics and risk factors. Gut 2008;57:828–35. [18] Montano-Loza AJ, Tandon P, Kneteman N, Bailey R, Bain VG. Rotterdam score predicts early mortality in Budd-Chiari syndrome, and surgical shunting prolongs transplant-free survival. Aliment Pharmacol Ther 2009;30:1060–9. [19] Pavri TM, Herbst A, Reddy R, Forde KA. Budd-Chiari syndrome: a single-center experience. World J Gastroenterol 2014;20:16236–44. [20] Bajos N, Bohet A, Le Guen M, Moreau C. La contraception en France: nouveau contexte, nouvelles pratiques? Popul Soc 2012;(492):1–4. [21] Ayala R, Grande S, Bustelos R, Ribera C, Garcia-Sesma A, Jimenez C, et al. Obesity is an independent risk factor for pre-transplant portal vein thrombosis in liver recipients. BMC Gastroenterol 2012;12:114. [22] Bureau C, Laurent J, Robic MA, Christol C, Guillaume M, Ruidavets JB, et al. Central obesity is associated with non-cirrhotic portal vein thrombosis. J Hepatol 2016;64(2):427–32.
Please cite this article in press as: Ollivier-Hourmand I, et al. The epidemiology of Budd–Chiari syndrome in France. Dig Liver Dis (2018), https://doi.org/10.1016/j.dld.2018.04.004
ARTICLE IN PRESS Digestive and Liver Disease xxx (2018) xxx–xxx
Contents lists available at ScienceDirect
Digestive and Liver Disease journal homepage: www.elsevier.com/locate/dld
Liver, Pancreas and Biliary Tract
The epidemiology of Budd–Chiari syndrome in France Isabelle Ollivier-Hourmand a,∗,1 , Manon Allaire a,b,1 , Nathalie Goutte c , Rémy Morello d , Carine Chagneau-Derrode e , Odile Goria f , Jerôme Dumortier g , Jean Paul Cervoni h , Sébastien Dharancy i , Nathalie Ganne-Carrié j,k , Christophe Bureau l , Nicolas Carbonell m , Armand Abergel n , Jean Baptiste Nousbaum o , Rodolphe Anty p , Hélène Barraud q , Marie Pierre Ripault r , Victor De Ledinghen s , Anne Minello t , Frédéric Oberti u,v , Sylvie Radenne w , Noelle Bendersky x , Olivier Farges y , Isabelle Archambeaud z , Anne Guillygomarc’h A , Marie Ecochard B , Violaine Ozenne C , Marie Noelle Hilleret D , Eric Nguyen-Khac E , Barbara Dauvois F , Jean Marc Perarnau G , Pascale Lefilliatre H , Jean Jacques Raabe I , Michel Doffoel J , Jean Philippe Becquart K , Eric Saillard L , Dominique Valla x,M , Thong Dao a , Aurélie Plessier x,M , for the French Network for Vascular Disorders of the Liver a
Hepato Gastroenterology Nutrition Department, University Hospital Center Côte de Nacre, Caen, France Inserm Unit U1149, Research Center on Inflammation, Paris, France UMRS1193 and University Hospitals of Paris Sud, France d Public Health Unit, University Hospital Center Côte de Nacre, Caen, France e Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire, Poitiers, France f Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Charles Nicolle, Rouen, France g Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Edouard Herriot, Lyon, France h Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Jean Minjoz, Besanc¸on, France i Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Claude Huriez, Lille, France j Service d’hépatologie, Centre Hospitalo-Universitaire Hôpital Jean Verdier, APHP, Bondy, France k Inserm UNR-1162, « Génomique Fonctionnelle des Tumeurs Solides », Paris, France l Service d’Hépatologie, Centre Hospitalo-Universitaire Purpan, Toulouse, France m Service d’Hépatologie, Centre Hospitalo-Universitaire Saint Antoine, APHP, Paris, France n Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Estaing, Clermont Ferrand, France o Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire La Cavale Blanche, Brest, France p Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire, Nice, France q Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire de Nancy, Vandœuvre-lès-Nancy, France r Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Saint Eloi, Montpellier, France s Service d’Hépatologie, Hôpital Haut-Lévêque, Bordeaux, France t Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire de Dijon Bourgogne, Dijon, France u Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire, Angers, France v Laboratoire HIFIH, Université Bretagne Loire, Angers, France w Service d’Hépato-gastroentérologie, Hôpital de la Croix Rousse, Lyon, France x Service d’hépatologie, Hôpital Beaujon, APHP, Clichy-la-Garenne, France y Service de Chirurgie Hépato-Pancréato-Biliaire, Hôpital Beaujon, APHP, Clichy-la-Garenne, France z Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Hôtel-Dieu, Nantes, France A Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Pontchaillou, Rennes, France B Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire, Saint-Etienne, France C Service d’Hépatologie, Hôpital Lariboisière, APHP, Paris, France D Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Michallon, Grenoble, France E Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire, Amiens, France F Service d’Hépato-gastroentérologie, Centre Hospitalier Régional, Orléans, France G Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire, Tours, France H Service d’Hépato-gastroentérologie, Centre Hospitalier Général, Le Havre, France b c
∗ Corresponding author at: Service d’Hépato-Gastroentérologie et de Nutrition, CHU de Caen, Avenue Côte de Nacre, FR 14033 Caen, France. E-mail address: [email protected] (I. Ollivier-Hourmand). 1 Equal contributors. https://doi.org/10.1016/j.dld.2018.04.004 1590-8658/© 2018 Editrice Gastroenterologica Italiana S.r.l. Published by Elsevier Ltd. All rights reserved.
Please cite this article in press as: Ollivier-Hourmand I, et al. The epidemiology of Budd–Chiari syndrome in France. Dig Liver Dis (2018), https://doi.org/10.1016/j.dld.2018.04.004
G Model
ARTICLE IN PRESS
I
Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Metz-Thionville, Ars-Laquenexy, France Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire, Strasbourg, France Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Saint Pierre, La Réunion, France L Service d’Hépato-gastroentérologie, Centre Hospitalo-Universitaire Pointe à Pître, Guadeloupe, France M Centre de Recherche de l’Inflammation, Inserm UMR-1149, Université Paris Diderot, Paris, France J
K
a r t i c l e
i n f o
Article history: Received 30 November 2017 Accepted 3 April 2018 Available online xxx Keywords: Hepatic vein obstruction Incidence National database Questionnaire survey Risk factors
a b s t r a c t
Introduction: Epidemiological data is lacking on primary Budd–Chiari syndrome (BCS) in France. Methods: Two approaches were used: (1) A nationwide survey in specialized liver units for French adults. (2) A query of the French database of discharge diagnoses screening to identify incident cases in adults. BCS associated with cancer, alcoholic/viral cirrhosis, or occurring after liver transplantation were classified as secondary. Results: Approach (1) 178 primary BCS were identified (prevalence 4.04 per million inhabitants (pmi)), of which 30 were incident (incidence 0.68 pmi). Mean age was 40 ± 14 yrs. Risk factors included myeloproliferative neoplasms (MPN) (48%), oral contraceptives (35%) and factor V Leiden (16%). None were identified in 21% of patients, ≥2 risk factors in 25%. BMI was higher in the group without any risk factor (25.7 kg/m2 vs 23.7 kg/m2 , p < 0.001). Approach (2) 110 incident primary BCS were admitted to French hospitals (incidence 2.17 pmi). MPN was less common (30%) and inflammatory local factors predominated (39%). Conclusion: The entity of primary BCS as recorded in French liver units is 3 times less common than the entity recorded as nonmalignant hepatic vein obstruction in the hospital discharge database. The former entity is mostly related to MPN whereas the latter with abdominal inflammatory diseases. © 2018 Editrice Gastroenterologica Italiana S.r.l. Published by Elsevier Ltd. All rights reserved.
1. Introduction Budd–Chiari syndrome (BCS) is defined as an obstruction of the hepatic venous outflow tract located anywhere from the small hepatic venules to the entrance of the inferior vena cava into the right atrium, with the exclusion of cardiac and pericardial diseases, and sinusoidal obstruction syndrome/veno-occlusive disease [1]. Primary BCS is characterized by obstruction due to thrombosis or phlebitis, which usually excludes malignant invasion or prior cirrhosis. The available estimates of the incidence and prevalence of primary BCS in European countries are based on either small samples of individual cases collected from hospital liver units over a long period or from large queries of national databases of hospital discharge diagnoses [2–5]. Recent estimates differ widely depending on the studies, which are also markedly different. In France, an unpublished survey performed in academic liver units recorded 20 new cases of primary BCS in the adult population, corresponding to an estimated incidence of 0.36 per million inhabitants (pmi) in 1990 [5]. On the other hand, the estimated incidence in a recent population based study of hospital discharge diagnoses in Northern Italy was 6 times greater, with 2.1 pmi per year between 2000 and 2012 [4]. Because of these limited and conflicting data, the aim of the present study was to estimate the incidence, the prevalence and the main characteristics of primary BCS in French adult residents based on two different approaches: a nationwide questionnaire in specialized hospital liver units, and a query in the French database of hospital discharge diagnoses.
2. Materials and methods 2.1. Questionnaire Between January 1st and December 31st, 2010, we performed a declarative survey in all 32 French academic liver units including 14 groups in Ile de France (an administrative area that includes 18% of the French population), as well as in 16 large non-academic liver units throughout the country (Supplemental material 1). These liver units function as regional reference centers for vascular liver diseases since the French Ministry of Health created a nationwide network in 2005. We recorded incident BCS and the prevalence of primary BCS in inpatients or outpatients. Patients under 18 years of age who were not French residents, who developed BCS after liver transplantation, and those with malignancies except for myeloproliferative neoplasms (MPN) were excluded. The date of the first imaging test confirming the diagnosis of BCS was considered to be the date of the initial diagnosis. Characteristics at diagnosis were recorded using a standardized questionnaire, including the risk factors for venous thrombosis [1]. Deficiencies in protein C, protein S and anti-thrombin were not considered to be risk factors because of the difficulty of confirming them as primary defects in patients with liver diseases [1,6]. For similar reasons, antiphospholipid syndrome was only considered to be a risk factor if it was diagnosed before BCS or if systemic lupus was present. This observational survey received approval of the Commission nationale de l’informatique et des libertés (CNIL) and Comité de protection des personnes (CPP) – Nord-Ouest France. Incidences and prevalences were calculated using the 2010 population census performed by the Institut National de la Statistique et des Etudes Economiques (INSEE), for residents of France who were 18 years or older. Incident cases of BCS were defined as cases whose diagnosis was confirmed in 2010.
Please cite this article in press as: Ollivier-Hourmand I, et al. The epidemiology of Budd–Chiari syndrome in France. Dig Liver Dis (2018), https://doi.org/10.1016/j.dld.2018.04.004
G Model YDLD-3722; No. of Pages 7
ARTICLE IN PRESS I. Ollivier-Hourmand et al. / Digestive and Liver Disease xxx (2018) xxx–xxx
2.2. Search in the discharge diagnosis database for incident cases The French database of hospital discharge diagnoses (Programme de Medicalisation des Systèmes d’Information – PMSI) was queried to identify French residents 18 years or older with a new diagnosis of BCS between January 1st, 2007 and December 31st, 2012. In PMSI, each inpatient has a unique anonymous alphanumerical identification number that traces admissions in private and public hospitals throughout the country which excludes duplicates, and allows an exhaustive search. Discharge abstracts include the ICD-10 codes of main and associated diagnoses [7]. BCS was considered to be an incident case if the ICD-10 code I82.0 was recorded for the first time in 2012. 2012 was considered rather than 2010 to increase the washout period (2007–2012) without prior notification of code I82.0. In addition we performed a comparison of ages at diagnosis, genders, and regions between 2010 and 2012 in order to exclude duplicates. BCS cases were classified into the following 2 mutually exclusive categories: secondary BCS, associated with ICD-10 codes of cancer, alcoholic and viral cirrhosis or liver transplantation (to exclude secondary BCS due to transplant related venous outflow blockade), at any time between 2007 and 2012; and primary BCS from 2007 to 2012, when ICD-10 codes for secondary BCS were absent. The incidence was calculated using 2012 INSEE national census. This observational study was approved by CNIL and CPP – Ile de France. 2.3. Statistical analyses Quantitative variables were described using means and standard deviations (SD). Qualitative or categorical (QoC) variables were summarized using percentages. The Chi-square test was used to analyze QoC data and the Fischer’s exact test or the Fisher–Freeman–Halton test when the use conditions of the Chisquare test were not always verified. Means were compared with the Student-t test. Levene’s test was used to check the equality of variance. Most measures were given with their 95% confidence intervals. For the percentages, the Exact method was used, except where the ends of the confidence intervals are close to 0 or 1: in this case, the Score method provides more precise endpoints. All the tests were two tailed and the probability significance threshold ® ® was set at p < 0.05. IBM -SPSS 22.0 and R software (version 3.0.3; R Foundation for Statistic Computing, Vienna, Austria; http://www. r-project.org) for Windows were the statistical software used. 3. Results 3.1. Estimates of incidence and prevalence based on the 2010 survey Forty of the 48 selected centers completed the survey in 2010. A total of 186 cases of primary BCS were recorded. One incident case and 7 additional prevalent cases from 2 centers were excluded because of insufficient participation at the regional level. Thirty of the 178 primary BCS cases included in the cohort were in newly diagnosed symptomatic in-patients (Fig. 1). The resident population which was 18 years and older, corresponding to the 38 remaining centers in 2010, was 44.068.082 inhabitants. In this population, the incidence was 0.68 pmi per year and the prevalence was 4.04 pmi. Regional incidences varied from 0 to 2.12 pmi per year (Supplementary data 1 and 2). 3.2. Estimate of the incidence according to the discharge diagnosis database Between 2007 and 2012, 1666 in-patients were identified with the code I82.0. Two-hundred and eight of them were identified for
3
Table 1 Main characteristics of patients at the time of diagnosis in 2010 survey in hospital liver units (n = 178). Available data
Total
Age, yearsb Male gendera Body mass indexa Arterial hypertensiona Diabetesa Incident casesa
173 173 147 172 172 178
40.2 ± 13.9 53 (30.6) 24.1 ± 5.1 10 (5.8) 3 (1.7) 30 (16.9)
Location of thrombosis Isolated hepatic veina Combined hepatic vein and inferior vena cavaa Associated portal vein thrombosisa
158 158 158
121 (76.6) 37 (23.4) 5 (3.2)
Clinical manifestations Ascitesa Hepatomegalya Abdominal paina Oesophageal varicesa Splenomegalya Jaundicea Fevera Hydrothoraxa
164 164 156 135 159 133 154 154
122 (74.4) 115 (70.1) 113 (72.4) 74 (54.8) 78 (49.1) 27 (20.3) 31 (20.1) 20 (13.0)
Laboratory data Factor V 70–100%a 50–70%a <50%a
114 114 114
52 (45.6) 28 (24.6) 34 (29.8)
120 120
87 (72.5) 33 (27.5)
120 120
94 (78.3) 26 (21.7)
120 120
97 (80.8) 23 (19.2)
Total bilirubin <50 mol/La ≥50 mol/La ALT <5Na ≥5Na AST <5Na ≥5Na a b
Number (percentage) of patients. Median.
the first time in 2012, including 110 cases of primary BCS (Fig. 2). The census population over 18-years old who were residents in France in 2012 was 50.805.530 million inhabitants, and the estimated incidence in 2012 was 4.10 pmi for all cases, and 2.17 pmi for primary BCS. Regional incidences varied from 0 to 4.51 pmi per year (Supplementary data 3). 3.3. Characteristics and risks factors of BCS in hospital liver units in 2010 A total of 173/178 included cases of primary BCS had a complete set of data. The first cases were diagnosed in 1979, while 146 cases were diagnosed after 2000. The diagnosis was confirmed within the 7 days after the initial clinical manifestations in 9.9% of cases, and more than one month after the initial manifestations in 48.4% (Fig. 3). The main characteristics at diagnosis and risk factors are presented in Tables 1 and 2. Only one risk factor was identified in 94/173 patients with primary BCS, (54.3%), 2 risk factors in 31 (17.9%), ≥3 risk factors in 12 (7.0%), and none in 36 (20.8%). MPN was diagnosed in 72/151 patients (47.7%), based on the detection of the JAK 2 V617F mutation in 55 out of 139 tested, both bone marrow biopsy and endogenous erythroid colonies in 7, bone marrow biopsy only in 8, and endogenous erythroid colonies only in 2. Among BCS patients with MPN, 44 had polycythemia vera, 20 essential thrombocytemia, 7 latent MPN, and 1 myelofibrosis. MPN was diagnosed before BCS in 9 patients and none of them had received cytoreductive therapy. The mean body mass index (BMI) was 24.1 kg/m2 , higher in the group without than
Please cite this article in press as: Ollivier-Hourmand I, et al. The epidemiology of Budd–Chiari syndrome in France. Dig Liver Dis (2018), https://doi.org/10.1016/j.dld.2018.04.004
G Model YDLD-3722; No. of Pages 7 4
ARTICLE IN PRESS I. Ollivier-Hourmand et al. / Digestive and Liver Disease xxx (2018) xxx–xxx
Fig. 1. Flow chart for the inclusion and exclusion of cases into the survey performed in 2010 in hospital liver units.
Fig. 2. Flow chart for the inclusion and exclusion of cases into the survey 2012 from the queries addressed to the national hospital discharge diagnosis data base.
in that with identified risk factors (25.7 kg/m2 and 23.7 kg/m2 , respectively, p < 0.001), without difference in ascites frequency (70,6% (24/34) vs 75,4% (98/130), p = 0,66). Fifty percent (16/30) of patients without any risk factors had a BMI >25 kg/m2 compared to 26.9% (31/113) in those with at least one risk factor (p = 0.009).
3.4. Characteristics and risk factors for primary BCS in hospital discharge diagnosis database in 2012 The median age was 46.3 ± 18.4 years old in patients with primary BCS and 60.7 ± 15.1 years old in those with secondary BCS (p < 0.001). There were 68.1% of women with primary BCS, but only 25.5% with secondary BCS (p = 0.014).
Please cite this article in press as: Ollivier-Hourmand I, et al. The epidemiology of Budd–Chiari syndrome in France. Dig Liver Dis (2018), https://doi.org/10.1016/j.dld.2018.04.004
G Model
ARTICLE IN PRESS
YDLD-3722; No. of Pages 7
I. Ollivier-Hourmand et al. / Digestive and Liver Disease xxx (2018) xxx–xxx
5
Fig. 3. Delay to diagnosis in patients from hospital liver units.
Table 2 Risk factors patients in 2010 survey in hospital liver units (n = 178). Available data
Total
Acquired thrombophilia Myeloproliferative diseasea Paroxysmal nocturnal hemoglobinuriaa Behcet diseasea Hyperhomocysteinemiaa Systemic diseasea , c
151 112 101 118 173
72 (47.7) 10 (8.9) 6 (5.9) 9 (7.6) 14 (8.1)
Inherited thrombophilia Factor V Leidena Factor II G20210A mutationa MTHFR C677T mutationa Thalassemiaa
120 117 60 173
19 (15.8) 6 (5.1) 7 (11.7) 1 (0.6)
173
10 (5.8)
103 120
36 (35.0) 2 (1.7)
Local factora , b Hormonal risk factor Recent oral contraceptive usea Pregnancya a
Number (percentage) of patients. Including hepatic abscess (n = 3), hydatid cyst (n = 3), angiomatosis (n = 2), abdominal trauma (n = 1) and schistosomiasis (n = 1). c Including inflammatory bowel disease (n = 4), Takayasu disease (n = 1), sarcoidosis (n = 1), disseminated lupus erythematosus (n = 4), coeliac disease (n = 2), paludism (n = 2). b
Forty-seven (42.7%) of the 110 patients with primary BCS, had an associated ICD-10 code for conditions likely to increase the risk of hepatic vein thrombosis, including 14 (29.8%) MPN; 8 (17%) thrombophilia including lupus and other coagulation disorders; 5 (10.6%) rare diseases including Takayasu disease, sarcoidosis, Behcet disease, paroxysmal nocturnal haemoglobinuria (PNH), histiocytosis, familial Mediterranean fever, thalassemia, or sickle cell disease; 7 (14.9%) pregnancy; 1 (2.1%) abdominal trauma; and 18 (38.3%) local factors including inflammatory bowel disease, pancreatitis, cholecystitis, liver abscess, parasitosis and polycystic liver disease. In the 98 cases of secondary BCS, 9 (9.2%) had an associated ICD10 code for primary tumor of the liver excluding hepatocellular carcinoma (HCC), 18 (18.4%) HCC; 21 (21.4%) for a secondary tumor of the liver; 26 (26.5%) cancer without evidence of liver tumor; 19 (19.4%) alcoholic or viral cirrhosis without HCC; 10 (10.2%) liver graft complications. 6 (6.1%) MPN; 3 (3.1%) thrombophilia; 1 (1%) abdominal trauma; and 9 (9.1%) abdominal inflammatory disease associated with one of the previous codes. 4. Discussion This study based on a survey of hospital units participating in the nationwide network for vascular liver diseases shows that primary BCS is a rare condition in France because only 30 incident cases were identified in 2010, in an area with 44 million adult inhabi-
tants. This corresponds to an incidence rate of 0.68 pmi in 2010. However, in 2012, the nationwide discharge-diagnosis database recorded 110 new cases of primary BCS in a population of almost 51 million adult inhabitants, for an incidence rate of 2.17 pmi per year, which is 3 times higher than the previous estimate. Other findings in this survey suggest that these estimates differ because they apply to different entities, rather than different counts for the same disease. This hypothesis may be one of explanations to discrepancies between BCS incidence estimations around the world. The estimate obtained from the survey in the hospital liver units, 0.68 pmi in year 2010, applies to the entity presenting as a predominant liver disease that is often associated with general prothrombotic conditions. These 30 new patients identified in unselected but specialized liver units in the year 2010 must be compared to the 35/163 patients prospectively included in the multicenter European (En-Vie) cohort by French centers between October 2003 and October 2005 [8]. The pattern of obstructed veins and the associated prothrombotic conditions were as described in the EN-Vie cohort [8,9] and the predominance of MPN as a cause is fully and independently confirmed in the present survey. This survey also shows that local factors explaining why thrombosis develops at these unusual sites, are rare [9]. Thus, the present findings suggest that observations from the EN-Vie cohort study performed in tertiary referral units can be extrapolated nationally to patients who are seen in initial or secondary referral liver units. In other words, this survey demonstrates that results obtained in a survey among selected tertiary referal centers in Europe, are applicable to individual European countries. On the other hand, the queries of the national discharge diagnosis database identified a broader entity that was 3 times more frequent (2.17 incident cases pmi in 2012) suggesting that there was selection of BCS patients seen in hospital liver units compared to those admitted to hospitals for all causes. Spurious overestimation is unlikely because the recording system excludes duplicates and the ICD-10 code I82.0 is highly specific. Of note, an Italian survey of similar design, yielded very similar results [4]. It is interesting to note that local factors were identified in a greater percentage of cases than previously reported in series from hospital liver units, even after exclusion of cirrhosis of other origins and cancer. Indeed, a code for a concomitant disease that would probably facilitate the development of hepatic vein thrombosis through local mechanisms was recorded in 25% of these cases but in only 11% in the EN-Vie cohort [8,9]. The factor of abdominal inflammation played a major role, as reported in similar studies in smaller populations and longer periods of inception [4]. Finally it should be emphasized that 30.2% of all cases (57.2% of primary BCS) were not associated with a diagnostic code of a risk factor for BCS. This is a higher percentage than in the EN-Vie cohort [9] but is in between the findings
Please cite this article in press as: Ollivier-Hourmand I, et al. The epidemiology of Budd–Chiari syndrome in France. Dig Liver Dis (2018), https://doi.org/10.1016/j.dld.2018.04.004
G Model
ARTICLE IN PRESS
YDLD-3722; No. of Pages 7
I. Ollivier-Hourmand et al. / Digestive and Liver Disease xxx (2018) xxx–xxx
6
Table 3 Main characteristics of epidemiological surveys on Budd–Chiari syndrome.
Study period Age limit – years Malignancy excluded Population based Retrospective Prospective Declarative Inpatients Outpatients Incident cases – n Study population – 106 Incidence – pmi per year Patients (n) Prevalence pmi
Japan [10]
Denmark [2]
Franceb
Sweden [3]
China [11]
South Korea [12]
Franceb
Franceb
Italy [4]
1989 None Yes No Yes Yes Yes Yes No 21 123 0.17 160 2.4
1981–1985 None Yes Yes Yes No No Yes No 13 5 0.5 NAa NAa
2010 ≥18 Yes No Yes Yes Yes Yes Yes 30 44 0.68 178 4.04
1990–2001 None No Yes Yes No No Yes Yes 12 1.3 0.8 43 1.4
2000–2013 None No No Yes No No NA NA 20191 1300 0.88 NAa 7.69
2009–2013 None No Yes Yes No Yes Yes Yes 424 50 0.87 NAa 5.29
2012 ≥18 Yes Yes Yes No No Yes No 110 51 2.17 NAa NAa
2012 ≥18 No Yes Yes No No Yes No 208 51 4.1 NAa NAa
2000–2012 ≥20 No Yes Yes No No Yes No 287 13 2 NAa NAa
[n]
Reference of the paper. Not available. b Current survey. a
of similar searches of population-based databases in Northern Italy (61% of noncirrhotic nonmalignant BCS cases) [4] and Sweden (37% of noncirrhotic nonmalignant BCS) [3]. Because of the design of these studies, it is impossible to determine to what extent associated conditions were overlooked. Our findings and those of similar population-based studies using a hospital discharge diagnosis suggest that obstruction of hepatic venous outflow is a frequent but ancillary condition, that is, merely associated with a predominantly extrahepatic condition of the liver. The latter entity is likely not managed in specialized liver units, but probably in surgery units. As shown in Table 3, current and previously reported estimated incidences of BCS fall into 2 groups: one group with an incidence between 0.17 and 0.88 (present 2010 survey, [2,3,10,11,12]); and another with more than 2.0 (present 2012 survey, [4]). The two groups have different sample sizes and background populations of individual studies. Survey designs in the first group were based on a careful review of cases by liver specialists, except for the Danish study (performed when accurate noninvasive imaging was not available). In the lower incidence group, the estimate of our current survey in hospital liver units (0.68) is probably one of the most representative because of the large size of the nationwide population. Estimates from Sweden were based on only 15% of Swedish population [3]. Although these hospital liver unit surveys were performed at different dates, the estimates are relatively close, and similar to recent data suggesting a stable incidence of BCS since 2000. It is interesting to note that this group includes surveys in China and Korea, two Asian countries which were considered to have a high incidence but for which solid evidence was lacking [11,12]. The other group of studies includes 2 large population-based analyses of hospital discharge diagnoses with a similar design, performed at the same time in Italy and France, respectively. Identification of these cases was based on principal and secondary diagnoses. Thus, secondary diagnoses of BCS correspond to conditions in which obstructed hepatic veins were considered ancillary to an otherwise severe condition such as acute pancreatitis or inflammatory bowel disease. The Italian study did not individualize nonmalignant and noncirrhotic BCS, and covered only 22% of Italian population. If we apply similar inclusion criteria our overall estimated incidence of secondary and primary BCS was 4.1 pmi in 2012, twice the Italian estimate of 2.1 pmi per year, while our estimated incidence of primary BCS was 2.17. The third study using a hospital discharge diagnosis database was Danish and evaluated nonmalignant BCS. In that study the incidence of BCS (0.5 pmi per year) was probably underestimated because of a lack of sensitivity of liver vascular imaging between 1981 and 1985. Thus, even in the absence of cirrhosis or malignancy, hepatic vein obstruction
appears to be 2–4 times more common than the clinical entity of BCS managed in hospital liver units. Our survey of a recent, large population of consecutive patients enrolled in hospital liver units over a short and recent period in particular 2010, provides a realistic world picture of primary BCS in France, close to previous surveys [3,8,9,13–15,16,17]. The delay in the diagnosis of primary BCS patients was more than 6 months after the first clinical manifestations in 15.2% and more than one year in 5.9%. These findings suggest the presence of either unequal access to specialized care, or difficulties in diagnosis. To date the risk factors of primary BCS have mainly been analyzed in retrospective or multicenter studies making it difficult to assess selection biases [3,9,13–19]. The present study in 173 patients is representative of referrals to liver units nationwide. Our results confirm the multifactorial etiology of BCS. Indeed 24.9% of patients had at least 2 risk factors. Not surprisingly, MPN was present in 47.7%, factor V Leiden mutation in 15.8%, PNH in 8.9% and Behcet in 5.9%. Certain of the findings in this study are not in agreement with prior series. The proportion of oral contraceptive users in the survey (35.0% of women of child-bearing age) is less to or close to that in the general French population of women of the same age (47.5% in 2010 [20]). Despite the possibility of a lack of patient declaration, this suggests that current contraceptives are less of a risk factor of BCS. On the other hand, the present study shows for the first time that being overweight may be a new risk factor of primary BCS, since overweight was twice more common when no usual risk factor was identified without difference in ascites frequency. A recent study support this point in portal vein thrombosis with overweight and metabolic syndrome observed in respectively 82.0% and 47% of patients with idiopathic portal vein thrombosis vs 51% and 17% in the control group (p = 0.01). Thus, like in portal vein thrombosis with or without cirrhosis, overweight could impact on hepatic venous outflow thrombosis: one explanation could be an increase of plasminogen activator inhibitor 1 level [21,22]. In our study the result must be interpreted cautiously since the volume of ascites was not evaluable, and complete data concerning metabolic syndrome not available. Further studies are needed to confirm this hypothesis and analyze thoroughly metabolic syndrome in patients with and without a risk factor for thrombosis.
5. Conclusion In France, primary BCS is a rare entity when it is defined as a predominantly hepatic disease related to underlying prothrombotic conditions (incidence in liver units 0.68 pmi) whose main characteristics have been widely reported in large European multicenter
Please cite this article in press as: Ollivier-Hourmand I, et al. The epidemiology of Budd–Chiari syndrome in France. Dig Liver Dis (2018), https://doi.org/10.1016/j.dld.2018.04.004
G Model YDLD-3722; No. of Pages 7
ARTICLE IN PRESS I. Ollivier-Hourmand et al. / Digestive and Liver Disease xxx (2018) xxx–xxx
studies. In contrast, the entity recorded in the national database as nonmalignant noncirrhotic hepatic vein obstruction is 3 times more common and frequently accompanies local abdominal inflammatory diseases. The possible role of overweight as a risk factor for BCS merits further attention. Current oral contraception might be associated with a lesser risk of BCS. Conflict of interest None declared. Study concept and design Ollivier-Hourmand, Plessier, Valla. Acquisition of data and Critical revision of the manuscript for important intellectual content and final approval of the version to be published Ollivier-Hourmand, Allaire, Abergel, Anty, Archambeaud, Armand, Barraud, Becquart, Bureau, Carbonell, Chagneau-Derrode, Cervoni, Dharancy, Dauvois, De Ledinghen, Doffoel, Dumortier, Ecochard, Ganne-Carrié, Goria, Goutte, Guillygomarc’h, Hilleret, Lefilliatre, Nousbaum, Minello, Farges, Nguyen-Khac, Oberti, Ozenne, Perarneau, Raabe, Radenne, Ripault, Saillard. Analysis and interpretation of data and Drafting of the manuscript Ollivier-Hourmand, Allaire, Dao, Plessier, Valla. Statistical analysis Ollivier-Hourmand, Allaire, Morello, Goutte, Bendersky. Acknowledgments for technical assistance Delaval G, Zekrini K, Habersetzer F, Heurgues A, Cadranel JF, Sogni P, Rosa I, Grangé JD, Gornet JM, Parlier D, Del Tedesco E, Lombard F, Falissard B. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.dld.2018.04.004.
7
References [1] European Association for the Study of the Liver. EASL Clinical Practice Guidelines: vascular diseases of the liver. J Hepatol 2016;64(1):179–202. [2] Almdal TP, Sørensen TI. Incidence of parenchymal liver diseases in Denmark, 1981 to 1985: analysis of hospitalization registry data. The Danish Association for the Study of the Liver. Hepatology 1991;13:650–5. [3] Rajani R, Melin T, Björnsson E, Broomé U, Sangfelt P, Danielsson A, et al. BuddChiari syndrome in Sweden: epidemiology, clinical characteristics and survival – an 18-year experience. Liver Int 2009;29:253–9. [4] Ageno W, Dentali F, Pomero F, Fenoglio L, Squizzato A, Pagani G, et al. Incidence rate and case fatality rates of portal vein thrombosis and Budd-Chiari Syndrome. Thromb Haemost 2017;117:1–7. [5] Valla DC. Hepatic venous outflow tract obstruction etiopathogenesis: Asia versus the West. J Gastroenterol Hepatol 2004;19:S204–11. [6] Qi X, Han G, Guo X, De Stephano V, Xu K, Lu Z, et al. Review article: the aetiology of primary Budd-Chiari syndrome – differences between the West and China. Aliment Pharmacol Ther 2016;44(11–12):1152–67. [7] ICD-10 Version 2016. [Internet]. [Cited 2015 October 12] Available from: http:// apps.who.int/classifications/icd10/browse/2016/en. [8] Seijo S, Plessier A, Hoekstra J, Dell’era A, Mandair D, Rifai K, et al. Good long-term outcome of Budd-Chiari syndrome with a step-wise management. Hepatology 2013;57:1962–8. [9] Darwish Murad S, Plessier A, Hernandez-Guerra M, Fabris F, Eapen CE, Bahr MJ, et al. Etiology, management, and outcome of the Budd-Chiari syndrome. Ann Intern Med 2009;151:167–75. [10] Okuda H, Yamagata H, Obata H, Iwata H, Sasaki R, Imai F, et al. Epidemiological and clinical features of Budd-Chiari syndrome in Japan. J Hepatol 1995;22:1–9. [11] Zhang W, Qi X, Zhang X, Su H, Zhong H, Shi J, et al. Budd-Chiari Syndrome in China: A Systematic Analysis of Epidemiological Features Based on the Chinese Literature Survey. Gastroenterol Res Pract 2015;2015:738548. [12] Ki M, Choi HY, Kim KA, Kim BH, Jang ES, Jeong SH. Incidence, prevalence and complications of Budd-Chiari syndrome in South Korea: a nationwide, population-based study. Liver Int 2016;36(7):1067–73. [13] Garcia-Pagán JC, Heydtmann M, Raffa S, Plessier A, Murad S, Fabris F, et al. TIPS for Budd-Chiari syndrome: long term results and prognostics factors in 124 patients. Gastroenterology 2008;135:808–15. [14] Mentha G, Giostra E, Majno PE, Bechstein WO, Neuhaus P, O’Grady J, et al. Liver transplantation for Budd-Chiari syndrome: A European study on 248 patients from 51 centres. J Hepatol 2006;44:520–8. [15] Darwish Murad S, Valla DC, De Groen PC, Zeitoun G, Hopsman JA, Haagsma EB, et al. Determinants of survival and the effect of portosystemic shunting in patients with Budd-Chiari syndrome. Hepatology 2004;39:500–8. [16] Plessier A, Sibert A, Consigny Y, Hakime A, Zappa M, Denninger MH, et al. Aiming at minimal invasiveness as a therapeutic strategy for Budd-Chiari syndrome. Hepatology 2006;44:1308–16. [17] Moucari R, Rautou PE, Cazals-Hatem D, Geara A, Bureau C, Consigny Y, et al. Hepatocellular carcinoma in Budd-Chiari syndrome: characteristics and risk factors. Gut 2008;57:828–35. [18] Montano-Loza AJ, Tandon P, Kneteman N, Bailey R, Bain VG. Rotterdam score predicts early mortality in Budd-Chiari syndrome, and surgical shunting prolongs transplant-free survival. Aliment Pharmacol Ther 2009;30:1060–9. [19] Pavri TM, Herbst A, Reddy R, Forde KA. Budd-Chiari syndrome: a single-center experience. World J Gastroenterol 2014;20:16236–44. [20] Bajos N, Bohet A, Le Guen M, Moreau C. La contraception en France: nouveau contexte, nouvelles pratiques? Popul Soc 2012;(492):1–4. [21] Ayala R, Grande S, Bustelos R, Ribera C, Garcia-Sesma A, Jimenez C, et al. Obesity is an independent risk factor for pre-transplant portal vein thrombosis in liver recipients. BMC Gastroenterol 2012;12:114. [22] Bureau C, Laurent J, Robic MA, Christol C, Guillaume M, Ruidavets JB, et al. Central obesity is associated with non-cirrhotic portal vein thrombosis. J Hepatol 2016;64(2):427–32.
Please cite this article in press as: Ollivier-Hourmand I, et al. The epidemiology of Budd–Chiari syndrome in France. Dig Liver Dis (2018), https://doi.org/10.1016/j.dld.2018.04.004