Liver histology is important for prognosis and treatment strategy in patients with hepatitis C. We report a 10-year experience of transjugular liver biopsy (TJLB) in patients with haemophilia and other congenital bleeding disorders (CBD) in terms of safety, efficiency and therapeutic consequences. TJLB was proposed to patients who were regularly followed for CBD, and were hepatitis C virus (HCV) positive by polymerase chain reaction. Patients with inhibitors or who were human immunodeficiency virus (HIV) positive with CD4 cells <0·2 × 109/l or with evidence of liver failure were excluded. TJLB was performed during a short hospitalization with factor replacement. Between 1992 and 2002, 88 TJLB were performed in 69 of 151 adult HCV patients (39% HIV positive). CBD was haemophilia A in 68% and haemophilia B in 24%. Few mild adverse events were recorded. Histology was assessable in 78 of 88 procedures (89%). Twenty-nine (37%) cases demonstrated minimal change (METAVIR A ≤ 1 and F ≤ 1). Extended fibrosis or cirrhosis was recorded in 23 procedures (26%), all in patients whose infection period was longer than 20 years. No relationship between liver histology, HIV status or HCV genotype was found. TJLB appears to be safe and useful in HCV patients with CBD. One-third of patients had minimal histological changes and could avoid systematic anti-HCV treatment.
The main objective of this study was to assess the safety and efficacy of TJLB in patients with CBD infected with HCV. The consequences of histological evaluation on therapeutic decision and the extra cost related to the correction of the underlying CBD were also measured.
Patients and methods
Since 1992, all patients with haemophilia or other CBD regularly followed in the haemophilia centre of Cochin hospital have been tested for anti-HCV with a second generation enzyme-linked immunosorbent assay (ELISA). We proposed TJLB to all adults with positive HCV antibodies and a positive HCV polymerase chain reaction (PCR) (since the availability of the test in 1995), who had abnormal alanine transaminase (ALT) levels if non-HIV co-infected (at least once during the year before) and in all HIV co-infected patients. Haemophiliacs with inhibitors or HIV-positive patients with a CD4 cell count below 0·2 × 109/l were excluded. Patients with clinical or biological evidence of liver failure (jaundice, ascites or prothrombin time below 70% of control) were also excluded, as were those with diagnosed hepatocellular carcinoma. Ultrasound examination of the liver, associated with Doppler, was performed before the procedure.
In patients who underwent TJLB, the date of HCV contamination was approximated according to their transfusion history. The retained date was that of the earliest exposition to blood before the screening of blood donors for HCV antibodies began in 1992 (Donahue et al, 1992) or to pool-derived concentrates before the availability of virus-inactivated clotting factors concentrates in 1987 (Garson et al, 1990; Mariani et al, 1993; Morfini et al, 1994).
Transjugular liver biopsy was performed during a 2–4-d hospitalization. Patients were infused with factor VIII concentrate (haemophilia A), factor IX concentrate (haemophilia B), von Willebrand factor concentrate (severe von Willebrand disease) or desmopressin (mild haemophilia A and mild von Willebrand disease) just before biopsy. Dosage was adjusted according to the severity of the disease (about 40 U/kg in patients with severe haemophilia A, 60 U/kg in patients with severe haemophilia B). Infusion was repeated every 8–12 h (20 U/kg in haemophilia A, 30 U/kg in haemophilia B) during a period that was progressively reduced: 48 h with additional infusions on days 6 and 8 for the first 19 biopsies, 48 h for the next 43 biopsies, 24 h for the last 26 biopsies.
Premedication was obtained with 10 mg of diazepam. All the procedures were performed by the same physician under local analgesia, with radio-fluoroscopic guidance. Briefly, a vessel dilator was introduced into the external jugular vein. A Quick-Core transvenous liver biopsy set was used (LABS-100; William Cook Europe, Bjaeverskov, Denmark).
In addition to liver sampling, transjugular access enabled measurement of the hepatic venous pressure gradient, calculated as the difference between wedged and free hepatic venous pressures (in normal subjects the gradient is ≤4 mmHg) according to previously described method (Lebrec et al, 1982). Briefly, correct wedging in a hepatic vein was assumed when the three following criteria were met: stable pressure curve, lack of reflux after injection of 2 ml of contrast, and a sharp fall in pressure after withdrawal of the catheter. Two consecutive similar values of wedged hepatic venous pressure were used. Free hepatic venous pressure was measured near the junction of the inferior vena cava and the hepatic vein. This gradient reflects postsinusoidal intra-hepatic portal hypertension.
Results are expressed as mean ± SE. Statistical analysis was performed by using analysis of variance (anova) or chi-square test with Yates correction as appropriate. A probability level of P < 0·05 was considered to be significant.
Of the 151 HCV-positive patients followed at our centre, 69 (three females, 66 males) underwent at least one liver biopsy during the study period. Among them, 27 had positive HIV serology and six had positive hepatitis B antigen. None had clinical or ultrasonographic evidence of cirrhosis. A total of 88 TJLB were performed between June 1992 and September 2002 (Fig 1). Fifty-three patients had one, 14 patients had two, one patient had three and one patient had four biopsies at various intervals. Repeated biopsy was performed either because of failure of the procedure (n = 10) or to assess evolution of the disease in treated or untreated patients (n = 9) at 2–8-year intervals.
Liver biopsy was not performed in the remaining 82 HCV-infected patients for the following reasons: patient refusal (n = 39), presence of factor VIII inhibitors (n = 16), advanced HIV disease (n = 9), other severe disease such as malignancy (n = 7, including two patients with hepatocellular carcinoma), age >75 years (n = 3), HCV-antibody positive but HCV-PCR negative (n = 4), persistent normal ALT in HIV-negative patients (n = 4).
The median age of the patients at the time of biopsy was 38·3 years (range 18·1–71·7 years). Biopsy was performed in patients with haemophilia A in 60 cases (severe in 34), haemophilia B in 21 (severe in 11), von Willebrand disease in five cases (four type 2 and one type 3), factor X deficiency in one case and factor VII deficiency in one case. Twenty-seven patients (corresponding to 36 biopsies) had positive HIV serology. HCV genotype was obtained in 52 patients (corresponding to 66 biopsies). Among these patients, 31 (60%) had genotype 1 and 21 (40%) had other genotypes (four genotype 2, nine genotype 3, four genotype 4 and four genotype 5). Six patients were positive for hepatitis B surface antigen (HBs-Ag).
None of the patients had clinical or biological evidence of cirrhosis or hepatic failure. In all the patients, the abdominal Doppler ultrasound examination was normal or showed non-specific steatosis.
The date of contamination for HCV infection could be determined in 54 patients (corresponding to 67 biopsies). The duration of contamination varied from 3 years (in a single patient infected in 1997 by intravenous drug injection) to 53 years (29·3 ± 1·5 years). The age at contamination varied from 0 to 49 years (9·9 ± 1·5 years).
Mild adverse events were recorded in seven cases (8%): two haematoma of the neck at the puncture point were reported. These haematoma were not compressive and did not modify the clotting factor replacement protocol. In another patient, transient palsy of the right arm was possibly due to a haematoma at the level of plexus brachial. Ultrasonographic examination was normal but factor replacement was prolonged for 10 d. Two cases of transient right hypochondrium pain were reported within 48 h after the procedure with normal physical and ultrasonographic examinations. One of the patients received an additional factor infusion. Right hypochondrium pain occurring in a patient a few days after the biopsy was diagnosed as hepatic colic with no signs of subcapsular haematoma or haemobilia. This patient had known biliary lithiasis on ultrasonographic examination before the biopsy. Finally, one patient presented a haematoma at the site of a central device used for clotting factor infusion. All these symptoms were transient and none were life-threatening.
Histological score could be evaluated in 78 of 88 procedures (89%). Mild hepatitis was recorded in 29 biopsies (33%), moderate hepatitis in 31 (35%) and there was evidence of extended fibrosis or cirrhosis in 18 (19%). In the remaining 10 cases, scoring was not possible because of catheterism failure or the sample was too small.
A significant difference was observed in liver histology according to the duration of HCV infection (Table I). The duration of infection was significantly longer in patients with extensive fibrosis or cirrhosis (35·9 ± 3·1 years) compared with mild or moderate hepatitis (27·3 ± 1·7 years). In particular, extensive fibrosis or cirrhosis was not observed in patients with a duration less than 20 years. Conversely, despite a trend to a lower frequency of mild hepatitis in HIV-positive patients, the difference was not statistically significant (Table II). The duration of HCV infection was not different in HIV-positive and -negative patients (30·4 ± 2·1 and 28·8 ± 1·9 years respectively). There is also no significant difference for the presence of genotype 1 or other genotype according to the result of liver histology (Table III), whereas the duration of HCV infection was similar in these two groups (31·2 ± 2·2 and 28·4 ± 2·8 years respectively).
Table I. Liver histology according to the duration of hepatitis C virus (HCV) infection in adult patients with congenital bleeding disorders infected with HCV.
Transjugular liver biopsies
Values are given as n (%).
Results are given for patients and for biopsies (some patients had more than one biopsy). Duration of HCV infection was obtained in 54 patients. Liver histology was not assessable for four patients. Liver histology was significantly worse in patients with HCV duration longer than 20 years (P = 0·04; chi-square test with Yates correction). Mild hepatitis corresponded to a METAVIR score of A ≤ 1 and F ≤ 1; moderate hepatitis to a METAVIR score of A > 1 and F = 0 or 1, or any A and F = 2; extended fibrosis or cirrhosis to a METAVIR score of any A and F = 3 or 4.
Extensive fibrosis or cirrhosis
Table II. Liver histology according to human immunodeficiency virus (HIV) serology in adult patients with congenital bleeding disorders infected with hepatitis C virus.
Transjugular liver biopsies
Values are given as n (%).
Results are given for patients and for biopsies (some patients had more than one biopsy). HIV serology was obtained in all 69 patients. Liver histology was not assessable for four patients. There was no significant difference in liver histology between patients with and without HIV co-infection (P = 0·19; chi-square test). Mild hepatitis corresponded to a METAVIR score of A ≤ 1 and F ≤ 1; moderate hepatitis to a METAVIR score of A > 1 and F = 0 or 1, or any A and F = 2; extended fibrosis or cirrhosis to a METAVIR score of any A and F = 3 or 4.
Extensive fibrosis or cirrhosis
Table III. Liver histology according to hepatitis C virus (HCV) genotype in adult patients with congenital bleeding disorders infected with HCV.
Transjugular liver biopsies
Genotype other than 1
Genotype other than 1
Values are given as n (%).
Results are given for patients and for biopsies (some patients had more than one biopsy). HCV genotypes were obtained in 52 patients. Liver histology was not assessable for three patients. There is no significant difference in liver histology between patients with infection genotype 1 and with infection genotype other than 1 (P = 0·93; chi-square test). Mild hepatitis corresponded to a METAVIR score of A ≤ 1 and F ≤ 1; moderate hepatitis to a METAVIR score of A > 1 and F = 0 or 1, or any A and F = 2; extended fibrosis or cirrhosis to a METAVIR score of any A and F = 3 or 4.
Extensive fibrosis or cirrhosis
In 17 cases (nine HIV positive), the platelet count was moderately decreased (median 130 × 109/l, range 77–142 × 109/l). The comparison between liver fibrosis and platelet count demonstrated a higher frequency of low platelet count in patients with extensive fibrosis or cirrhosis (80% vs. 11%; P < 0·01) corresponding to a sensitivity of 44% and a specificity of 90%.
Hepatic venous pressure gradient
The hepatic venous pressure gradient recorded during the 78 procedures was statistically different in mild hepatitis, moderate hepatitis and extensive fibrosis or cirrhosis (anova). When these three groups were analysed separately, patients with extensive fibrosis and cirrhosis had a significantly (P < 0·001) higher gradient (9·6 ± 2·1 mmHg) than those with mild hepatitis (2·8 ± 0·2 mmHg) or moderate hepatitis (3·3 ± 0·4 mmHg). In practice, the hepatic venous pressure gradient facilitated the evaluation of F3 and F4 fibrosis in the case of a small or fragmented biopsy sample.
Consequence of the TJLB on therapeutic counselling
In 49 of 78 assessable biopsies, there was evidence of moderate hepatitis (40%) or extended fibrosis or cirrhosis (23%). Consequently, relevant treatment was proposed to these patients (n = 39) and cirrhotic patients underwent a specific follow-up (upper digestive endoscopy to evaluate portal hypertension, and clinical evaluation every 6 months, Doppler ultrasound examination, blood liver function tests and alphafoetoprotein). Conversely, 29 biopsies (corresponding to 37% of assessable biopsies or 33% of all procedures) showed only minor or no histological change, no treatment was proposed at that time and the usual ALT and clinical follow-up was proposed with repeated TJLB at 3–5-year intervals.
Thirty-five of the 39 patients who had indication of treatment agreed to be treated. Two of them underwent three courses of treatment, eight of them two courses and the remainder 25 a single course. Treatment was interferon-alpha alone until 1998, interferon plus ribavirin in 1999 and 2000, and thereafter pegylated interferon plus ribavirin.
A sustained virological response, corresponding to a negative HCV-PCR 6 months after completion of treatment, was obtained in 13 of 30 assessable patients. Five patients are still under treatment. The rate of sustained response according to the type of treatment was 0/12 with interferon alone, 4/10 with interferon plus ribavirin and 9/20 with pegylated interferon plus ribavirin.
Extra cost related to bleeding disorder
The amount and duration of factor replacement was progressively reduced during the study, because of our increasing confidence in the safety of the TJLB. The initial mean cost for a patient (weighed 70 kg) with severe haemophilia was €12 000; this dropped to €4500 according to our on-going practice. This extra cost is lower in patients with minor haemophilia A or von Willebrand disease who can be biopsied under desmopressin (€120).
Indication of liver biopsy in patients with CBD and HCV infection is still a debatable issue. In the general population, the prevalence of severe complications, including death as a result of liver biopsy, is between 0·2% and 0·5% (Froehlich et al, 1993; Cadranel et al, 2000). The risk of this procedure is difficult to evaluate in particular populations, such as haemophiliacs, with regard to the small number reported, the different criteria for selection and the different biopsy techniques and clotting factor replacement requirement. Because of recurrent reports of high morbidity with this procedure and two former cases of death related to liver biopsy in haemophiliacs in the early 1980s (Aledort et al, 1985), a number of physicians are unwilling to consider this procedure in these patients (Lee, 1997; Telfer, 1997). However, other recent series of percutaneous liver biopsies (Ahmed et al, 1996; Hanley et al, 1996; McMahon et al, 2000; Venkataramani et al, 2000; Lethagen et al, 2001) reported low morbidity and indicated the decision-making interest of these procedures in large cohorts of patients. The transjugular route for liver biopsies was developed in liver disease units to allow pathological examination of the liver even in case of ascites or abnormal coagulation. Logically, TJLB associated with adequate factor replacement has been proposed for patients with CBD infected with HCV (Makris et al, 2001). However, only a few series of patients with CBD have been reported with this technique to date (Gupta et al, 1997; DiMichele et al, 2003). Gupta et al (1997) reported the results of TJLB in a cohort of six HIV-positive but HCV-negative patients with CBD. DiMichele et al (2003) reported their experience of 13 TJLB in adult CBD patients with HCV infection without significant morbidity.
Here, we reported our experience of 88 TJLB in 69 adult CBD patients with HCV infection. No major complication was observed in this study and only three minor events could be related directly to the procedure. In the four other cases, the relationship to the procedure was not clear.
A possible limitation of the TJLB was the size of the biopsy needed to permit accurate diagnosis. As reported previously in larger studies, the cut-off level of six portal tracts identified was used in this series as sufficient to grade and stage liver damage in hepatitis C (The French METAVIR Cooperative Study Group, 1994). In this study, the major limitation of the technique was the access and catheterization of the internal jugular vein or the hepatic vein. In case of procedural failure, a further procedure was proposed a few weeks later. Finally, the procedure was successfully performed in 89% (corresponding to 94% of patients), allowing an accurate histological examination.
Among factors reported to increase the development of fibrosis in hepatitis C, duration of infection, age at contamination (Poynard et al, 1997), alcohol intake (Pessione et al, 1998; Wiley et al, 1998) and HIV co-infection seem to play a major role (Lesens et al, 1999; Yee et al, 2000; Benhamou et al, 2001; Puoti et al, 2001; Ragni & Belle, 2001; Mohsen et al, 2003). The roles of age at contamination and of alcohol could not be evaluated in this study, because all patients, except one, were contaminated before 40 years of age (Poynard et al, 1997) and the alcohol daily intake was below 50 g/d in all but two. Regarding HIV co-infection, present in 39% of the patients, our results did not confirm the higher frequency of advanced liver disease in HIV and HCV co-infected patients; however, a trend to higher frequency of mild hepatitis in HIV-negative patients was observed. The detrimental effect of HIV co-infection was probably underestimated in this study, as patients with clinical or biological evidence of liver cirrhosis were excluded. Similarly, among HIV and HCV co-infected patients, it has been demonstrated that a deteriorated immunological status was associated with a more severe liver disease (Lesens et al, 1999; Benhamou et al, 2001; Puoti et al, 2001; Ragni & Belle, 2001; Mohsen et al, 2003). In this study, the negative effect of HIV was probably also underestimated, as patients with CD4 below 0·2 × 109/l were excluded.
In non-haemophiliacs and non-HIV co-infected patients with chronic hepatitis C, improvement of sustained virological response has been obtained by the combination of interferon-alpha and ribavirin compared with interferon-alpha alone (McHutchison et al, 1998; Poynard et al, 1998; Reichard et al, 1998). Using pegylated-interferon associated with ribavirin, a sustained virological response rate was reported to be as high as 55% in non-HIV co-infected patients (Manns et al, 2001; Fried et al, 2002). However, according to viral prognostic factors, the range for sustained virological response was large, from 42% for HCV genotype 1 to 82% for HCV genotype 2 or 3 (Manns et al, 2001) and from 42% for high viral load to 78% for low viral load (Manns et al, 2001). Haemophilia patients have virological characteristics that collectively tend to determine low responder group, namely high prevalence of HIV co-infection, of HCV genotype 1 and of high viral load (Makris et al, 2001). It has been reported that the sustained virological response rate obtained with interferon monotherapy was lower in haemophiliacs than in non-haemophiliacs (Makris et al, 2001). Early results with combination therapy suggested similar results in haemophiliacs and in non-haemophiliacs (Sauleda et al, 2000; Shields et al, 2000), even in prior non-responder to interferon monotherapy (Santagostino et al, 2002). However, these studies included a small number of highly selected patients and excluded HIV co-infected patients. Moreover, potential side effects of the combination of interferon with ribavirin have to be considered, particularly in HIV co-infected patients. Some authors have proposed that haemophiliacs that are HCV-antibodies positive, HCV-PCR positive and have abnormal ALT levels should be treated systematically, without liver biopsy (Rumi et al, 1997; Sauleda et al, 2000; Beurton et al, 2001; Santagostino et al, 2002). In our experience, 40% of HIV-negative CBD patients infected with HCV showed only minor changes at liver histology. In this situation, consensus conferences have concluded that patients may not need treatment and should be monitored periodically (Agence Nationale d'Accreditation et d'Evaluation en Sante, 2002; National Institutes of Health, 2002). Conversely, a majority of patients who had more advanced disease at histology agreed to be treated. A sustained virological response was recorded in more than one-third of patients (half of patients treated with the most recent schedule, pegylated interferon and ribavirin). Haemophiliacs with minor histological abnormalities, who were contaminated in the mid-1980s or earlier, could wait for the development of new more efficient drugs with less side effects, with the exception of patients with good prognosis factors (HCV genotype 2 or 3 and HIV-negative) who have a higher chance of sustained virological response. Conversely, in this study, cirrhosis was detected in 20% of cases without clinical, biological or ultrasonographic evidence and it is noteworthy that thrombocytopenia showed low sensitivity for identifying extended fibrosis or cirrhosis. The diagnosis of a clinically and biologically unsuspected cirrhosis leads not only to anti-viral treatment but also to ensuring an adequate follow-up in order to detect hepatocellular insufficiency, portal hypertension and hepatocellular carcinoma at the early stages.
At this time, liver biopsy remains the only procedure known to accurately assess liver damage. This procedure should be proposed to patients with CBD to the same extent as to other patients with chronic HCV according to recent consensus conferences (Agence Nationale d'Accreditation et d'Evaluation en Sante, 2002; National Institutes of Health, 2002). However, in patients with CBD and with clinical or biological evidence of cirrhosis as well as in haemophiliacs with inhibitors or uncontrolled HIV disease, liver biopsy is probably without benefit. With regard to the high rate of success with new anti-HCV treatment, patients with HCV genotype 2 or 3 and non-HIV co-infection could be treated without liver histology in certain circumstances. However, these patients represent a minority of patients with CBD. In our opinion, the transjugular route associated with factor replacement seems to be the method of choice to reduce the risks of liver biopsy in this population. The consequences of the information provided by biopsy in the management of HCV-infected patients justify the extra cost related to the underlying CBD. Alternative non-invasive methods for liver histology have to be developed, particularly in this population.