Epidemiology of chronic hepatitis B infection in France: risk factors for significant fibrosis – results of a nationwide survey


Dr. J.-F. Cadranel, Service d’Hépato-Gastroentérologie et de Diabétologie, Centre Hospitalier Laennec, BP 72, 60109 Creil, France.
E-mail: jfrancois.cadranel@ch-creil.fr


Background  Epidemiological data concerning hepatitis B are scarce in France.

Aim  To describe epidemiological, clinical, virological and histological features of HBsAg-positive patients followed at non-academic hospitals in France.

Methods  Clinical, biological, virological and histological data of all HBsAg-positive consecutive patients observed from April 1, 2001 to May 31, 2002 in participating centres were recorded prospectively. Multivariate analyses of factors associated with significant fibrosis and cirrhosis were performed.

Results  Nearly 1166 HBsAg-positive patients were seen in the 58 centres: 671 males and 495 females from metropolitan France (32%) and from outside metropolitan France (68%); mean age 41 ± 15 years. Twenty-nine percent of patients were probable HBsAg inactive carriers, while 50% had chronic hepatitis; 43% of these were HBeAg-positive and 57% HBeAg-negative. Liver biopsy had been performed in 558 (51%) patients; 205 (17.6%) patients had cirrhosis. By multivariate analysis, factors associated with significant fibrosis were: age >40 years (P < 0.05), HBeAg-negative status (P < 0.02) and histological activity (P < 0.0001). Factors associated with cirrhosis: age (P < 0.0001), platelet count <150 000/mm3 (P < 0.0001) and viral co-infection (P < 0.03).

Conclusion  HBV infection represents a significant workload for hepatogastroenterologists at non-academic hospitals in France.


Chronic hepatitis B (HBV) infection affects 350 people individuals worldwide.1 It causes a spectrum of diseases ranging from an inactive carrier state to the development of severe cirrhosis-related complications and hepatocellular carcinoma.1, 2 Factors determining clinical outcome are largely unknown, and epidemiological studies of HBV are scarce in France. A recent epidemiological study conducted in France indicated that the prevalence of HBV infection was higher than previously thought.3 Moreover, HBeAg-negative chronic hepatitis B is now recognized as an important form of chronic hepatitis B. It is predominant in the Mediterranean area4–8 and the Far East9, 10. In 1994, the frequency of HBeAg-negative chronic hepatitis B, evaluated in a prospective study in academic hospitals, was 22.1% in France11. Recently, Zarski et al. in a large multicentre study performed at academic hospitals, reported an increased prevalence of HBeAg-negative chronic hepatitis which attained 72%.12 The aim of the present study was to describe current epidemiological and clinical features observed in general non-university hospitals, thereby avoiding a selection bias inherent to academic centres, in order to compare HBeAg-positive and HBeAg-negative patients with chronic hepatitis, and to elucidate factors associated with probable inactive carriage, on the one hand, and significant fibrosis and cirrhosis, on the other.

Patients and methods


Consecutively confirmed HBsAg-positive patients seen during the study period by hepatogastroenterologists working at general non-university hospitals were eligible for the study. The study was conducted prospectively from April 1, 2001 to May 31, 2002 in 58 centres, by members of the Association Nationale des Hépato-Gastroentérologues des Hôpitaux Généraux de France (ANGH). The population of areas from which studied patients originated was estimated at 15.620420 people. Among patients, 251 were under treatment at the time of study (data have been presented elsewhere).13


Data recorded

Data were anonymously recorded on a pre-established form for all HBsAg-positive consecutive patients seen during the study period at participating centres. Demographic data were the following: age (at the time of the study), gender, geographic origin, date and mode of discovery of HBsAg positivity, known duration of HBV infection when available, and known duration of HBsAg presence since discovery of HBsAg positivity. Laboratory data were recorded during the study period in the absence of any treatment. Alanine aminotransferase (ALT) levels were expressed as times the upper limit of normal range (ULNR). Recorded liver function tests included albumin level, total serum bilirubin concentration and prothrombin index. Platelet count and results of liver ultrasonography were noted. Presence of HBeAg or anti-HBe antibodies, and presence of serum HBV DNA through hybridization or sensitive quantitative methods were recorded in the absence of any antiviral treatment; presence or absence of delta antibodies and co-infection with hepatitis C or human immunodeficiency virus (HIV) viruses were also recorded. The results of liver biopsy performed before any antiviral treatments were recorded.

Serovirological methods

Testing for HBsAg (Biorad Monolisa Ag HBs Ultra, Marnes La Coquette, France), IgM anti-HBc antibodies (Abbott Axsym Core M), HBeAg (Abbott Axsym HBe) and HBe antibodies (Abbott Axsym anti-HBe) (Abbott Laboratories, Rungis, France) was carried out using a commercial enzyme immunoassay. HBV DNA was searched for using molecular hybridization (Hybrid Capture, Murex Diagnostic, France) with a cut-off of 1 pg/mL (i.e. 2.83.105 copies per mL = 5.45 log10 copies/mL)14 or, when available, by different sensitive quantitative methods according to each centre, including a signal amplification assay based on branched DNA (bDNA) technology. Although there was no strict correspondence between results of the different assays, we used the international standard for HBV DNA nucleic acid amplification techniques made available by the World Health Organization, and which probably leads to greater consistency in results of these assays.14 Results are expressed in log10 copies/mL.

Histological methods

At each centre, pathologists analysed biopsies in a blind fashion using the METAVIR classification15 and/or Knodell’s score.16 According to METAVIR, fibrosis was staged from F0 to F4: F0, no fibrosis; F1, portal fibrosis without septa; F2, portal fibrosis with few septa; F3, numerous septa without cirrhosis; and F4, cirrhosis. Necroinflammatory activity was graded from A0 to A3: A0, no activity; A1, mild activity; A2, moderate activity; and A3, severe activity. The term ‘significant fibrosis’ indicated a fibrosis stage equal to or above F2.15


Patients were classified according to the following pre-established criteria.

Probable HBsAg inactive carriers

Patients were considered probable HBsAg inactive carriers when they fulfilled all the following criteria in the absence of any antiviral treatment: (i) HBsAg positivity for at least 6 months; (ii) at least three repeated ALT levels below ULNR during the last 12 months; (iii) absence of HBeAg and presence of HBe antibodies; (iv) repeated HBV DNA serum negativity by hybridization and/or HBV DNA level below 105 copies/mL; (v) absence of co-infection with either delta, C or HIV virus; (vi) normal liver function tests and platelet counts; (vii) absence of other co-existing causes of liver disease (i.e. alcoholism, metabolic or autoimmune causes); and (viii) normal ultrasonography (except for gallbladder lithiasis and bright liver).

Acute hepatitis B

Patients were considered as harboring acute hepatitis B when (i) HBsAg positivity had been discovered within the past 6 months, (ii) they presented a positive test for IgM anti-HBc; and (iii) they had a marked increase in ALT activity (>10 upper limit of normal range) in the absence of signs of chronic liver disease.

Chronic hepatitis

HBeAg-positive chronic hepatitis.  HBsAg-positive patients were classified as having HBeAg-positive chronic hepatitis when they fulfilled the following criteria: (i) HBsAg positivity for at least 6 months; (ii) presence of HBeAg and absence of anti-HBc IgM antibodies; (iii) presence of HBV DNA by the hybridization technique and/or HBV DNA above 105 copies/mL; (iv) elevated ALT levels; and (v) liver biopsy showing chronic hepatitis (when available).

HBeAg-negative chronic hepatitis.  HBs-positive patients were classified as having HBe-negative chronic hepatitis when they fulfilled the following criteria: (i) HBsAg positivity for at least 6 months; (ii) absence of HBeAg and presence of HBe antibodies in the absence of any antiviral treatment, absence or presence at a low titer of anti-HBc IgM antibodies in the absence of epidemiological data for acute hepatitis; (iii) elevated ALT levels on at least one occasion; (iv) presence of HBV DNA by the hybridization technique and/or HBV DNA equal to or above 105 copies/mL; and (v) liver biopsy showing chronic hepatitis (when available).

Cirrhosis – hepatocellular carcinoma.  The diagnosis of cirrhosis relies either on histological criteria, or else a combination of other criteria including clinical (i.e. splenomegaly, ascites or any sign of liver decompensation), biological (low platelet count, prothrombin index below 75% in the absence of coagulation factor deficiency), ultrasonographic (i.e. hepatic dysmorphy or portal hypertension signs, ascites) and endoscopic (i.e. oesophageal varices) criteria. The term ‘decompensated cirrhosis’ refers to complications, such as jaundice, hepatic encephalopathy, ascites or variceal bleeding. The diagnosis of hepatocellular carcinoma relies on the usual criteria as established by the EASL Conference.17

Data analysis

Continuous variables were expressed as mean ± standard deviation (SD). Non-parametric data were expressed as percentage or median and range. Analyses were conducted using variance analysis, the chi-squared test, two-sided Fisher exact test, Mann–Whitney test, Wilcoxon test and two-sample Student’s t-test when appropriate. Categorical comparisons used logistic regression for multivariate analysis. Variables with a P-value <0.05 at univariate analysis were included in the final multivariate model. All statistical testing was two-tailed at the 5% level. The diagnostic value of non-invasive parameters of liver fibrosis evaluation was assessed by the area under the receiver-operating characteristics (AUROC) curve. The receiver-operating characteristics (ROC) curves were computed, and AUROC as well as 95% confidence intervals were calculated with the Mann–Whitney statistics as described by Hanley and McNeil. 18, 19 All analyses were performed using SPSS software (version 11.0; SPSS Inc., Chicago, IL, USA).



Nearly 1166 patients were consecutively included in this study during the 13-month survey. They included 671 males and 495 females with a mean age of 41 ± 15 years. They were followed at 58 centres; the number of patients followed at each centre ranged from 1 to 91 (median 12). Fourteen centres (24%) followed 30 patients or more, 16 centres (24%) 10–29 patients and 28 centres (48%) fewer than 10 patients. The mean follow-up of the total population was 6 ± 6 years (median 4 years; extremes: 1–32 years). Thirty-two percent of patients were from the French metropolitan area, whereas 68% were from elsewhere. The main characteristics of the population are shown in Table 1.

Table 1.   General characteristics of the population (n = 1166)
  1. * Mean ± standard deviation; † ULNR: upper limit of normal range.

Age (years) 41 ± 15
Female sex (%)43%
Geographic origin (%)
 Metropolitan France32%
 Outside France68%
 Sub-Saharan Africa24%
 Southern Europa12%
 Eastern Asia11%
Duration of HBsAg positivity (years)  5.7 ± 6
Liver tests
 Prothrombin index* (%) 86 ± 15
 Albumin level* (g/L) 40 ± 6
 Total bilirubin* (μmol/L) 23 ± 50
 ≤ ULNR (%)35%
 1–2 × ULNR (%)16%
 >2 × ULNR (%)49%
 Platelet count* (103/mm3)198 ± 78
Abnormal ultrasonography22%

Duration of HBV infection

The duration of HBV infection was known for 401 patients (34.4%). For those patients, the length of HBV infection was 35 ± 12 years.

Mode of discovery of HBsAg positivity

The mode of discovery of HBsAg was known for 1113 (95%) patients. HBsAg discovery took place in the absence of any overt clinical context in 922 (83%) patients: during screening of contact subjects of a known HBsAg carrier in 179 (16%) cases, during systematic screening in 127 cases (11%), in persons at high risk for HBV infection in 76 cases (7%), before blood donation or blood transfusion in 56 cases (5%) and during legal search during pregnancy in 198 cases (18%) cases. HBsAg discovery occurred during evaluation of persistent elevated ALT in 286 cases (26%). In 191 cases (17%) cases, a clinical context led to the discovery of HBsAg: acute hepatitis in 56 cases (5%), extrahepatic manifestations in 81 cases (7%), obvious cirrhosis in 43 cases (4%) and hepatocellular carcinoma in 11 (1%) patients.

Clinical presentation at the time of study

The clinical presentation of patients at the time of study was noted for 1145 (97%) patients in the cohort and is given in Table 2. According to pre-established definitions (see Patients and methods), 329 patients (29%) were classified as probable HBsAg inactive carriers, whereas 816 (71%) presented with acute or chronic liver disease. Anti-delta antibodies were observed in 3% of active carriers (32/1052), anti-HCV antibodies in 4% (46/1095) and anti-HIV antibodies in 4% (39/1070) of active carriers, respectively.

Table 2.   Clinical presentation at the time of the study (n = 1145)
Probable inactive carriers32929
Active carriers81672
Acute hepatitis283
Chronic hepatitis55369
Compensated cirrhosis14518
Decompensated cirrhosis354
Hepatocellular carcinoma253
Liver transplanted40.03

Probable HBS antigen inactive carriers (n = 329)

Comparative epidemiological and biological data of probable inactive and active carriers are given in Table 3. The length of follow-up of probable HBsAg inactive carriers was significantly shorter than that of active carriers (P < 0.0001). Probable inactive carriers were younger than active carriers (P < 0.0001), more often females (P < 0.0001) and more often born in sub-Saharan Africa than in France or eastern Asia. HBsAg positivity was more often discovered during pregnancy (P < 0.0001) or familial screening (P < 0.0001) in probable inactive carriers than in active carriers.

Table 3.   Comparison between probable inactive and active carriers of HBsAg
 Probable inactive carriers (N = 329)Active carriers (N = 816) P
  1. * Mean ± standard deviation; † ULNR: upper limit of normal mange.

Age (years) 35.3 ± 15.4 43 ± 12<0.0001
Female sex (%)213 (65)271 (33)<0.0001
Known duration of HBsAg positivity (years) mean ± S.E. [median]  4.6 ± 5.3 [3]  6.1 ± 6.5 [4]0.001
Geographic origin (%)
 Metropolitan France 23 39<0.0001
 Sub-Saharan Africa 40 19<0.0001
 Southern Europe  9 130.061
 Maghreb 12 100.213
 Eastern Asia  7 13<0.01
 Others  9  60.304
Prothrombin index* (%) 93 ± 9 84 ± 16<0.0001
Total bilirubin* (μmol/L) 12 ± 8.7 24.8 ± 54.3 0.014
Albumin level* (g/L) 40.3 ± 6.1 40.2 ± 6.3 0.876
Platelet count* (10.3/mm3)240 ± 72.5187.8 ± 75.9<0.0001
ALT (ULNR†) >2N (%)  0 61<0.0001
Active carriers (n = 816)
Comparison of active carriers with and without cirrhosis

Comparative data of these patients are given in Table 4. Nearly 205 patients had cirrhosis and 385 patients had histologically biopsy-proven chronic hepatitis without cirrhosis. The duration of known HBs positivity was similar in cirrhotic and non-cirrhotic patients. Patients with cirrhosis were more often males (P < 0.04), and were older than patients without cirrhosis (P < 0.0001). Cirrhotic patients were more often born in France (P < 0.02) and in the Maghreb region (P < 0.05), and less often in eastern Asia (P < 0.005) than patients without cirrhosis. Cirrhotic patients were more often HBeAg-negative (P < 0.05) and more frequently had viral co-infection (P < 0.02) than patients without cirrhosis. Activity grade was higher in cirrhotic patients (with liver histology; n = 145) than in non-cirrhotic patients (P < 0.003).

Table 4.   Comparison between active HBsAg carriers with biopsy-proven chronic hepatitis without cirrhosis (CH) and active carriers with cirrhosis
 CH patients*N = 385Cirrhotic patients N = 205P
  1. * CH: patients with chronic hepatitis and without cirrhosis on liver histology; † Mean ± standard deviation; ‡ Upper limit of normal range; § Metavir scoring system.

Age (years)39 ± 1453 ± 15<0.0001
Female sex (%)30220.033
Duration of HBsAg positivity† (years) [median]7 ± 7 [5]7 ± 7 [5]0.950
Geographic origin (%)
 Continental France36470.011
 Sub-Saharan Africa16130.275
 Southern Europe16130.396
 Eastern Asia1780.004
Prothrombin index† (%)89 ± 1171 ± 17<0.0001
Total bilirubin† (μmol/L)13 ± 930 ± 53 <0.0001
Albumin level† (g/L)40 ± 640 ± 70.659
Platelet† (10count/mm3)198 ± 58142 ± 78<0.0001
 <2 40480.057
 < 5 3732 
 ≥ 52320 
Viral co-infection (%)13240.013
HBeAg-positive (%)41330.048
Liver histology§
 Activity (%)
Viral load (log10 copies/mL)
 >7 70470.006
Comparison of patients with and without hepatocellular carcinoma

All 25 patients with hepatocellular carcinoma were cirrhotic. Patients with hepatocellular carcinoma were older than patients without the disease (58 ± 13 vs. 52 ± 15 years; (P = 0.06)) and were more often from eastern Asia (P < 0.0001) or of sub-Saharan origin (P < 0.08) than those without hepatocellular carcinoma.

Comparison of HBeAg-positive and -negative active carriers

Comparative data of these patients are given in Table 5. There were 257 (35%) HBeAg-positive patients and 468 (65%) HBeAg-negative patients. No difference in the female to male ratio was observed between HBeAg-positive and HBeAg-negative chronic hepatitis B patients. The known duration of HBsAg positivity was longer in HBeAg-negative than in HBeAg-positive patients (P < 0.009). HBeAg-negative patients were significantly older (P < 0.0001), more often born in southern Europe (P = 0.009) and the Maghreb (P < 0.013) and less often in eastern Asia (P < 0.0001) than HBeAg-positive patients. HBeAg-negative patients had lower ALT (P < 0.02) than HBeAg-positive patients. HBeAg-negative patients had lower HBV DNA (P < 0.0001) than HBeAg-positive patients. When considering only active carrier patients with liver biopsy, 225 (43%) were HBeAg-positive and 302 (57%) HBeAg-negative.

Table 5.   Epidemiological, clinical, biochemical, virological and histological characteristics of HBeAg-positive and HBeAg-negative active HBsAg carriers
  HBeAg-positive (N = 257)HBeAg-negative* (N = 468)P
  1. * HBe status not reported in 33 patients; † Mean ± standard deviation; ‡ Upper limit of normal range; § METAVIR scoring system.

Age† (years)39 ± 1744 ± 14<0.0001
Female sex (%)36350.77
Duration of HBsAg positivity† (years)5 ± 57 ± 70.009
Geographic origin (%)
 Continental France41360.212
 Sub-Saharan Africa16190.211
 Southern Europe9160.009
 Eastern Asia219<0.0001
Prothrombin index† (%) 85 ± 1584 ± 150.458
Total bilirubin† (μmol/L) 16 ± 1819 ± 370.375
Albumin level† (g/L) 40 ± 740 ± 60.572
Platelet count† (103/mm3) 189 ± 72181 ± 720.284
 ≥ 521300.015
Liver histology§
 Activity (%)
 Fibrosis (%)
 Median Knodell score8.58.50.679
Cirrhosis (%)
 Hepatocellular carcinoma33 
Transplanted (patients number)010.87
Viral load (log10 copies/mL)

Histological data

Nearly 558 patients (51%) underwent liver biopsy. Among these biopsies, 205 were classified according to the Knodell classification and 389 were classified according to the METAVIR scoring system. Sixteen HBeAg-negative patients had normal or subnormal liver histology, while 385 patients had chronic hepatitis and 145 had cirrhosis.

The median Knodell score was 91–16; the Knodell score ranged from 1 to 5 in 19% of cases, from 6 to 9 in 37% of cases, and above 9 in 44% of cases. The median Knodell score was similar in HBeAg-positive and HBeAg-negative patients. According to the METAVIR scoring system, median activity grade was 2. It was A0 in 4%, A1 in 43%, A2 in 40% and A3 in 13% of patients; 53% of patients thus presented with moderate to severe activity. The activity grade was similar in HBeAg-positive and HBeAg-negative patients. The median fibrosis stage was 2. Fibrosis stage was F0–F1 in 37%, F2 in 21%, F3 in 20% and F4 in 22%; 65% of patients presented with significant fibrosis, among whom 42% had severe fibrosis or cirrhosis. The fibrosis stage was higher in HBeAg-negative patients than in HBeAg-positive patients (F2-3-4 75% vs. 59%; P < 0.0001).

Factors associated with significant liver fibrosis (i.e. ≥ F2) and cirrhosis in multivariate analysis

Factors associated with significant fibrosis by multivariate analysis were the following: age >40 years (OR 1.8 [95% IC: 1.1; 3.3]; P < 0.05); HBe antigen-positive status (OR 0.5 [95% IC: 0.3; 0.9]; P < 0.02); platelet count <150 000/mm3 (OR 5.2 [95% IC: 1.7; 16.2]; P < 0.005); and liver activity ≥ A2 (OR 5.9 [95% IC: 3.3;10.5]; P < 0.0001). Factors associated with cirrhosis by multivariate analysis are given in Table 6. Age over 40 years (P < 0.0001), platelet count below 150 000/mm3 (P < 0.0001) and viral co-infection (P < 0.03) were significantly associated with cirrhosis.

Table 6.   Risk factors associated with compensated cirrhosis by multivariate analysis (N = 558)
 Tested variablesOR [95% CI]P
  1. *METAVIR scoring system.

Age (years)>404.2 [2; 8.5]<0.0001
SexFemale1.3 [0.6; 2.9]0.551
Geographic originMetropolitan France1.4 [0.7; 2.8]0.342
Platelet count (103/mm3)<1505 [2.8; 10.5]<0.0001
Histologic activity (%)*≥ A21.7 [0.8; 3.3]0.152
Viral co-infectionPresent2.3 [1.1; 5]0.032

Diagnostic performance of simple parameters for the diagnosis of significant fibrosis and cirrhosis

Sensitivity, specificity, positive and negative predictive values of prothrombin index, platelet count and serum bilirubin for the diagnosis of cirrhosis in patients with liver biopsy are given in Table 7. Prothrombin index and platelet count had the best diagnostic performances at the cutoff of 82% (sensitivity = 0.72 and specificity = 0.70) and 170 000/mm3 (sensitivity = 0.70; specificity = 0.70), respectively. For the diagnosis of significant fibrosis (F2–F3–F4) vs. no or mild fibrosis (F0–F1), the AUROC was 0.56 (CI: 95% 0.49–0.63) for serum bilirubin, 0.71 (CI: 95% 0.65–0.77) for platelet count and 0.71 (CI: 95%, 0.64–0.77) for prothrombin index with the following cut-off: 100/bilirubin 8 (Se 0.52; Sp 0.55), platelet count 185 000/mm3 (Se 0.66; Sp 0.65) and prothrombin index 82% (Se 0.70; Sp 0.64). For the diagnosis of cirrhosis vs. absence of cirrhosis, the AUROC was 0.59 (CI: 95% 0.51–0.66) for serum bilirubin, 0.73 (CI: 95% 0.66–0.81) for platelet count and 0.77 (CI: 95% 0.71–0.83) for prothrombin index with the following cutoff: 100/bilirubin 8 (Se 0.58; sp 0.56), platelet count 170 000 (Se 0.70; Sp 070) and prothrombin index 82% (Se 0.72; Sp 0.70).

Table 7.   Diagnostic values of biological tests for the diagnosis of biopsy-proven cirrhosis
Biological testsSensitivitySpecificityPositive predictive valueNegative predictive value
Prothrombin index (%) <80 and platelet count (103/mm3) <1500.480.950.840.76
Prothrombin index (%) <85 and platelet count (103/mm3) <1800.620.860.730.79
Prothrombin index (%) <80 0.660.810.680.80
Platelet count (103/mm3) <1500.620.820.670.78
Prothrombin index (%) <85 0.780.700.610.84
Platelet count (103/mm3) <1800.740.630.640.80


This large multicenter study was conducted in 58 non-academic hospitals throughout continental France over a 13-month period. Our series demonstrates that HBV infection accounts for a significant workload on the part of hepatogastroenterologists at these health care centres.

In 1992, an estimated HBsAg prevalence in France of 0.3% was suggested. 20 Since then, several bodies of evidence have strongly suggested an increasing prevalence of this infection in France. The Institut National de Veille Sanitaire recently estimated the prevalence of HBsAg carriers in France at 0.68% (IC: 95%; 0.44–1.05), with a prevalence of 1.19% in males (IC: 95%; 0.73–1.93) and 0.16% in females (IC: 95%; 0.07–0.37). An increasing prevalence was particularly noted in patients with low socio-economic status, and in immigrants from areas at high risk of hepatitis B.3 In a study recently performed in academic hospitals, a high proportion of patients born outside of continental France was also noted.12

The discovery of HBsAg during pregnancy accounted for 18% of patients in our cohort. Such results confirm the pertinence of the legal obligation to screen pregnant women. It should be noted that 68% of patients originated from outside continental France; they were from countries with a moderate or high prevalence of HBV infection.

This large-scale study enabled a comparison between HBeAg-positive and HBeAg-negative chronic hepatitis. Until recently, HBeAg-negative chronic hepatitis B was thought to be relatively rare and largely confined to Asian countries 9, 10 and the Mediterranean area4–7, 21, 22 rather than America or northern Europe.23, 24 A prevalence of 80–90% was reported in Italy 7, 21, Greece 8, 22 and Asia 9, 10. The first study comparing HBeAg-positive and HBeAg-negative patients with chronic hepatitis in France was performed in 1994 by Zarsky et al.11 The prevalence of HBeAg-negative patients was 22.1% in a population of 276 consecutive patients with chronic hepatitis seen in academic hospitals; most of the patients in that study were born in France or in northern European countries. In our study, among patients with biopsy-proven chronic hepatitis, 57% were HBeAg-negative and 43% HBeAg-positive. Several factors may account for the higher frequency of HBeAg-negative chronic hepatitis B that we observed. First, the influx of emigrants from countries with a high prevalence of this form of hepatitis B in recent years may have altered the epidemiology of HBV infection in our country, since HBeAg-negative chronic hepatitis was more often encountered in the present study in patients from southern Europe and the Maghreb. The increased prevalence of HBeAg-negative patients with chronic hepatitis is consistent with findings from a study conducted in academic hospitals12 and two uncontrolled trials, one performed in patients living in an area with low socio-economic level25 and another in a cohort of patients treated in France by lamivudine.26 Apart from their geographic origin, HBeAg-negative patients with chronic hepatitis differed from HBeAg-positive patients with chronic hepatitis in several ways. HBeAg-negative patients were older, had lower ALT levels, lower viral load and higher fibrosis scores than HBeAg-negative patients, with the prevalence of cirrhosis tending to be higher in HBeAg-negative patients. However, data concerning viral load that we observed in HBeAg-negative patients should be interpreted with caution, as insensitive methods were used in some patients. Taken together, however, these results are in agreement with the literature and with recent French studies.6–12, 26, 27

This large multicenter study gave us a unique opportunity to search for factors associated with probable inactive carriage, cirrhosis and hepatocellular carcinoma in France. Probable inactive carriers represented 29% of the entire cohort. Although the present study was designed prior to the conclusions of the EASL Conference,29 we feel that these patients were correctly classified. The median follow-up of these patients was as long as 3 years. Patients classified as probable inactive carriers were younger and more often female. They were more often born in sub-Saharan Africa and and less often in metropolitan France and eastern Asia than active HBeAg-negative patients. In addition, HBsAg positivity was more often discovered during pregnancy or screening in probable HBs Ag-inactive carriers than in active HBe Ag-negative patients.

Considering only patients with histologically biopsy-proven chronic hepatitis, cirrhotic patients were older, more often males and more often born in continental France or the Maghreb than in eastern Asia. Cirrhotic patients more often had viral co-infection than non-cirrhotic patients. Several studies have already shown the increased prevalence of significant fibrosis and cirrhosis in patients with B-delta or B-C co-infection,30, 31 as well as the deleterious effect of HIV co-infection32 on the course of HBV infection. All patients with hepatocellular carcinoma had cirrhosis; they were older and more often born in eastern Asia and sub-Saharan Africa than cirrhotic patients without hepatocellular carcinoma. These epidemiological characteristics are well-known factors in hepatocellular carcinoma.33

In our study, using multivariate analysis, significant fibrosis was associated with age above 40 years, HBeAg-negative status, a low platelet count and marked histological activity. In multivariate analysis, age above 40, a low platelet count and viral co-infection were significantly associated with cirrhosis. In the French study performed at academic hospitals, cirrhosis was associated with age over 50 years, male sex and low viral load.

Several differences in the populations studied may partly explain variations in the most important findings observed in these two large studies. First, the academic study included only patients with biopsy-proven chronic hepatitis scored by METAVIR classification; in contrast, in the present study, only 70% of patients classified as having chronic hepatitis had undergone liver biopsy. Second, our studies included all HBsAg-positive consecutive patients whatever their clinical status, whereas the academic study was restricted to outpatients and did not include patients with decompensated liver diseases, hepatocellular carcinoma or viral co-infection.12 Nevertheless, the number of cirrhotic patients was comparable in the two studies, i.e. 240 in the academic study12 and 205 in our study. Third, our population included a higher number of females (43%), with female gender being a well known factor associated with mild disease.

Factors associated with extensive fibrosis–cirrhosis have not been extensively studied in hepatitis B. In most studies,34–40 age and male sex were associated with cirrhosis; the absence of gender as a significant factor for bridging fibrosis–cirrhosis in our study may be in part explained by the fact that 18% of patients in the present study were detected during pregnancy, which combines female gender and youth. In a multicentre French study of patients treated with lamivudine,26 significant fibrosis was associated with age and male gender and only slightly with HBe status and non-Asian origin.

We had the opportunity of examining, on a large scale, diagnostic values for liver fibrosis of serum bilirubin, platelet count and prothrombin index. These simple parameters are routinely determined in patients with chronic liver diseases. We limited our evaluation to patients with liver biopsy and without hepatocellular carcinoma or decompensated cirrhosis, as these parameters are obviously disturbed in such settings. We found that diagnostic performances of platelet count and prothrombin index were more accurate for the determination of significant fibrosis and cirrhosis than that of serum bilirubin concentration.

Many non-invasive blood markers have been evaluated in chronic hepatitis C cohorts for evaluation of liver fibrosis, and accurate correlations with liver fibrosis according to the METAVIR score have been described. In contrast, few studies have evaluated the usefulness of non-invasive markers for diagnosis of significant fibrosis–cirrhosis in patients with chronic hepatitis B.41–46 The few indexes and or complex algorithms used rely on numerous blood parameters including bilirubin,41, 42 platelet count43, 45 and prothrombin time.44

Total serum bilirubin is part of the Fibrotest (Fibrosure in USA), a complex algorithm,41 and of discriminating function in the study by Hui et al.42 Fibrotest enabled distinguishing patients with significant fibrosis–cirrhosis from patients with mild or no fibrosis.41 However, total bilirubin alone showed a weak diagnostic performance.42 Indeed, the diagnostic performance of total bilirubin for diagnosis of significant fibrosis/cirrhosis in our study was poor. In addition, serum bilirubin is affected by haemolysis (frequent in African and Asian patients with haemoglobin diseases), and in Gilbert syndrome, which affects more than 5% of males in France.

The diagnostic performance of low platelet counts alone or combined with increased Asat (Asat/platelet ratio) for significant fibrosis is relatively accurate in patients with hepatitis C.43 However, the diagnostic performance of platelet count in patients with hepatitis B was poor in the study by Wai et al.45 with an AUROC of 0.63 and 0.73 only for diagnosis of significant fibrosis and cirrhosis, respectively. In our study, the AUROC of the platelet count was within the same range as in the study by Wai et al.45 Prothrombin index is an indirect marker of liver fibrosis46 and part of the complex algorithm of Cales et al.44 In our study, prothrombin index gave the best diagnostic performance, with an AUROC of 0.77 for cirrhosis.

In conclusion, this large-scale multicentre study contributes to the knowledge of epidemiological characteristics in France. It reveals that 65% of HB-virus-infected patients seen in non-academic hospitals had significant disease. HBeAg-negative chronic hepatitis is now predominant in France, as shown by two other recent French studies.12, 26 These epidemiological modifications should be taken into account in future evaluation of screening and treatment strategies Patients with hepatitis C infection and/or HIV infection should be screened for HBV infection and vaccinated against HBV in the absence of efficient immunization. A strong association between a moderate decrease in platelet count and prothrombin index and significant fibrosis/cirrhosis suggested that non-invasive simple methods should be developed to decrease the indications for liver biopsy, which remains the best method for correctly studying liver fibrosis. 47 The growing prevalence of chronic hepatitis B in France points to the urgent need for boosting the vaccination program against this virus at a nationwide level and for improving screening for this potentially severe disease.

ANGH participating centres and physicians

(1) Abbeville: J Butel; (2) Aix en Provence: C Wartelle, M Picon; (3) Aulnay sous Bois: G Bellaiche, G Tordjman, JL Slama; (4) Avignon: JP Arpurt, S Bellon; (5) Beauvais: P Stépani, C Bories; (6) Beziers: O Duhamel; (7) Blois: D Gargot, JP Ramain; (8) Bourg en Bresse: D Pillon; (9) Bourgoin Jallieu: T Fontanges; (10) Bry sur Marne: M Blazquez, B Condat; (11) Cahors: J Hadjadje; (12) Cambrai: T Thévenot; (13) Cayenne: D Louvel; (14) Châlons sur Saône: P Bernard, T Soupison; (15) Chartres: JP Cervoni, L Larvol; (16) Cherbourg: H Bertrand; (17) Compiègne: JC Barbare, JP Latrive; (18) Corbeil-Evry: J Denis, D Labayle, B Lambare; (19) Coulommiers: G Barjonet, C Njapoum; (20) Creil: JF Cadranel, A Cazier, P Dumouchel, K Hadj-Naccer, V Jouannaud; (21) Créteil: M Chousterman, I Rosa, V Auray-Cartier, H Hagège; (22) Douai: E Wohlschies; (23) Dreux: A Landau; (24) Eaubonne: M Howaizi; (25) Foix: JM Dramard; (26) Fontainebleau: JM Cayla; (27) Gonesse: A Pauwels, A Medini; (28) Hyères: C Renou; (29) Jolimont: J Henrion; (30) Le Mans: C Pilette, A Blanchi, B Bour; (31) Lens: T Davion; (32) Lourdes: JJ Meurisse; (33) Montauban: D Grasset, JL Payen; (34) Montélimar: B Nalet; (35) Montfermeil: B Lesgourgues, P Lahmek, L Traissac, S Nahon; (36) Moulins: R Combes, N Marcato; (37) Orléans: X Causse, N Si Ahmed; (38) Paris: Saint Joseph, L Mallet; (39) Pau: A Pariente; (40) Perpignan: AJ Rémy; (41) Oloron: JL Coquard; (42) Poissy: C Eugène, RL Vitte; (43) Pontivy: JA Seyrig; (44) Pontoise: O Danne, P Hervio; (45) Provins: L Jungman, E Gouffier; (46) Saint Brieuc: O Nouel; (47) Saint Denis: P Periac; (48) Saint-Nazaire: E Graf, T Martin; (49) Saint Quentin: P Levy, P Feydy; (50) Sarrebourg: H Jouin; (51) Senlis: C Halimi; (52) Soissons: O Ink; (53) Tarbes: T Morin; (54) Tourcoing: N Szostak-Talbodec; (55) Valence: H Moindrot; (56) Valenciennes: A Boruchowicz; (57) Vichy: P Cassan; (58) Villeneuve Saint Georges: L Bettan.


The authors thank Ms. Jerri Bram for English editing.

Declaration of personal and funding interests: None.