Pierre Opolon

Liver fibrosis in overweight patients

BACKGROUND & AIMS A common clinical issue is whether overweight patients with abnormal liver function test results should undergo liver biopsy. Although serious liver injury can occur, its prevalence and risk factors are not well known. METHODS Ninety-three consecutive patients with abnormal liver function tests (but without overt liver disease), body mass index (BMI) > 25 kg/m(2), and no alcoholic, viral, autoimmune, drug-induced, or genetic liver disease were retrospectively studied. Clinical, biological, and histological variables were tested for association with septal fibrosis or cirrhosis. RESULTS Septal fibrosis was present in 28 patients (30%) including cirrhosis in 10 (11%). Age >/= 50 years (odds ratio [OR], 14.1), BMI >/= 28 kg/m(2) (OR, 5.7), triglycerides >/= 1.7 mmol/L (OR, 5), and alanine aminotransferase (ALT) >/= 2N (OR, 4.6) were independently associated with septal fibrosis. Among histological features, septal fibrosis was strongly associated with necroinflammatory activity (OR, 44). A score combining age, BMI, triglycerides, and ALT had 100% negative predictive value for septal fibrosis when scoring 0 or 1 (100% sensitivity for a specificity of 47%). CONCLUSIONS Septal fibrosis occurs frequently in overweight patients with abnormal liver function tests. A clinicobiological score combining BMI, age, ALT, and triglycerides could improve selection of patients for liver biopsy.

Extrahepatic manifestations of chronic hepatitis C

Objective To assess the prevalence of clinical and biologic extrahepatic manifestations of hepatitis C virus (HCV) infection and to identify associations between clinical and biologic manifestations. Methods To analyze the natural history of extrahepatic manifestations of HCV infection, we reviewed only the data recorded prospectively during the first visit of 1,614 patients with chronic HCV infection, coming from a single monocenter cohort. Exclusion criteria were positivity for hepatitis B surface antigen or human immunodeficiency virus. The prevalence of dermatologic, rheumatologic, neurologic, and nephrologic manifestations; diabetes; arterial hypertension; autoantibodies; and cryoglobulins were assessed. Then, using multivariate analysis, we identified demographic, biochemical, immunologic, virologic, and liver histologic factors associated with the presence of extrahepatic manifestations. Results At least 1 clinical extrahepatic manifestation was observed in each of 1,202 patients (74%). Five manifestations had a prevalence >10%: arthralgia (23%), paresthesia (17%), myalgia (15%), pruritus (15%), and sicca syndrome (11%). Four biologic abnormalities had a prevalence >5%: cryoglobulins (40%), antinuclear antibodies (10%), low thyroxine level (10%), and anti–smooth muscle antibodies (7%). Only vasculitis, arterial hypertension, purpura, lichen planus, arthralgia, and low thyroxine level were associated with cryoglobulin positivity. By univariate and multivariate analyses, the most frequent risk factors for the presence of clinical and biologic extrahepatic manifestations were age, female sex, and extensive liver fibrosis. Conclusion Extrahepatic clinical manifestations are frequently observed in HCV patients and involve primarily the joints, muscles, and skin. The most frequent immunologic abnormalities include mixed cryoglobulins, antinuclear antibodies, and anti–smooth muscle antibodies. The most frequent risk factors for the presence of clinical and biologic extrahepatic manifestations are advanced age, female sex, and extensive liver fibrosis.

Cryoglobulinemia in chronic liver diseases: Role of hepatitis C virus and liver damage

BACKGROUND/AIMS Mixed cryoglobulinemia is frequently associated with liver diseases. The respective role of hepatitis C virus (HCV) and liver damage in the pathogenesis of cryoglobulinemia is investigated in this study. METHODS The prevalence of cryoglobulinemia in 226 consecutive patients with chronic liver diseases (hepatitis C, 127; hepatitis B, 40; other diseases, 59) was studied, and the epidemiological, biological, histological, and virological features in these three groups were analyzed. Anti-HCV antibodies, HCV proteins, and HCV RNA were searched in the cryoprecipitates. RESULTS The prevalence of mixed cryoglobulinemia was high (41.5%) in patients with liver diseases and higher in patients with hepatitis C (54.3%) than in patients with hepatitis B (15%) or other causes of liver disease (32%). Patients with cryoglobulinemia had cirrhosis more frequently and had a longer history of hepatitis. In patients with hepatitis C, HCV RNA sequences and HCV proteins were detected in the cryoprecipitate. Cryoglobulins became undetectable in 21 of 43 patients treated with interferon. CONCLUSIONS These findings suggest that HCV is a major cause of cryoglobulinemia. Besides viral infection itself, multiple factors appear to be responsible for the production of cryoglobulins, including cirrhosis and duration of liver disease.

INFLUENCE OF DELTA INFECTION ON SEVERITY OF HEPATITIS B

The prevalence of serum markers of primary delta infection was determined in 532 patients with acute benign hepatitis B seen in Italy, and in 111 patients with fulminant hepatitis B seen in Italy, France and England. Patients with fulminant hepatitis had significantly higher prevalence of delta markers (43/111, 39%) than did those with benign hepatitis (101/532, 19%). In 25 of the 43 patients with delta-positive fulminant hepatitis, serum markers indicated a primary hepatitis B infection while in the remaining 18, IgM antibody to hepatitis B core antigen was absent, indicating that hepatitis B preceded superinfection with the delta agent. The increased morbidity of HBsAg hepatitis with delta infection may result from the cumulative simultaneous exposure to hepatitis B virus and delta, or from superinfection of HBsAg carriers with delta.

Prognostic significance of bacterial infection in bleeding cirrhotic patients: a prospective study.

BACKGROUND/AIMS In cirrhotic patients, bacterial infection is frequently associated with gastrointestinal bleeding and seems to increase mortality. The aims of this study were to determine the incidence of bacterial infections in bleeding cirrhotic patients and the influence of infections on the risk of rebleeding and death. METHODS Cirrhotic patients admitted for gastrointestinal bleeding who had not received antimicrobial chemotherapy in the previous 7 days were included. Blood, urine, and ascitic fluid cultures were systematically performed 1, 2, 4, and 7 days after admission. RESULTS Sixty-four patients were enrolled. Forty-two bacterial infections were documented in 23 patients (36%) within 7 days of admission. In patients with bacterial infection, mean Child-Pugh score and mean number of blood units transfused were significantly higher, early rebleeding was more frequent (43.5% vs. 9.8%; P < 0.01), and 4-week mortality was higher (47.8% vs. 14.6%; P < 0.01). Multivariate analysis only identified bacterial infections as predictive of early rebleeding (P < 0.02) and a high Child-Pugh score as predictive of death (P < 0.001). CONCLUSIONS In bleeding cirrhotic patients, bacterial infections only increase the risk of early rebleeding, and mortality is related to the severity of cirrhosis.

Effects of Lamivudine on Replication of Hepatitis B Virus in HIV-Infected Men

Chronic infection with hepatitis B virus (HBV) affects about 5% of the worlds population [1]. Of persons infected with the human immunodeficiency virus (HIV) as many as 10% are carriers of hepatitis B surface antigen (HBsAg) [2]. The high prevalence of HBV infection in HIV-infected patients occurs partly because the two viruses share routes of transmission. Carriers of HBsAg are usually asymptomatic, but they may develop cirrhosis, liver failure, and hepatocellular carcinoma. The 5-year survival rate of patients with established cirrhosis associated with chronic active hepatitis does not exceed 55% [3]. A high level of HBV replication or the presence of hepatitis B e antigen (HBeAg) predicts poor survival [4] and is common in persons infected with both HIV and HBV. In these patients, chronic HBV infection is more frequent than in the general population [5, 6] and has a peculiar course. Histologic and biological activities are lower and serum concentrations of HBV DNA are higher than those seen in HIV-negative patients [2, 7]. Interferon- is a cytokine that has antiviral and immunomodulatory properties [8] and is often proposed for the treatment of chronic HBV infection. In HIV-negative patients, the rate of response to interferon- (loss of serum HBV DNA and HBeAg) is 20% greater than the spontaneous seroconversion rate of 12% after 6 to 12 months of follow-up [9]. However, HIV infection is known to diminish the response to interferon-; these decreased response rates range from 0% [10] to 8% [11]. Therapies for HIV infection are already associated with increased survival [12]. Thus, the frequency of chronic HBV infection, which leads to cirrhosis and its complications, may increase in the population of HIV-infected patients who are coinfected with HBV. Therapies that are more effective against HBV than is interferon- alone are still needed in both HIV-positive and HIV-negative patients. Lamivudine, the enantiomer of 2-deoxy-3-thiacytidine, an inhibitor of HIV type 1 and HIV type 2 reverse transcriptase [13], has been shown to have antiretroviral activity in HIV-infected patients [14]. Lamivudine is also a potent selective inhibitor of HBV replication because this replication depends on reverse transcription of an intermediate RNA to a minus-stranded DNA, which then serves as a template for the synthesis of plus-stranded DNA [15]. Experimental studies have shown that lamivudine inhibits HBV DNA replication in transfected cell line 2.2.15 and in HBV-infected chimpanzees [15, 16]. Results of a phase II clinical trial [17] showed that lamivudine inhibited HBV replication in patients with chronic HBV infection. However, no information is available about the activity of lamivudine against HBV in HIV-infected patients. An open-label program of lamivudine therapy (NUCB 3004) has been in progress in France since December 1993. This program is for patients with progressive HIV disease who are refractory to or unable to tolerate therapies other than lamivudine. We evaluated the efficacy of lamivudine against HBV replication in 40 consecutive patients who were infected with both HIV and HBV and who entered this program. We report our results after 1 year of follow-up. Methods Patients From April to September 1994, 228 HIV-infected patients who were followed at the infectious diseases department at Groupe Hospitalier Pitie-Salpetriere and were eligible for the open-label lamivudine trial were prospectively tested for serologic markers of HBV infection (HBsAg, antibodies to HBsAg, HBeAg, and antibodies to HBeAg), IgG and IgM antibodies to hepatitis delta virus, and antibodies to hepatitis C virus (HCV). Forty of these patients were HBsAg carriers; these patients were consecutively included in our study and were followed for 1 year. Patients were seen at baseline and every 2 months. At each visit, physical examination was done and all clinical events were recorded. Blood samples were obtained for complete blood cell counts, measurement of CD4 lymphocyte counts, serum biochemistry tests, tests for serologic markers of HBV, and measurement of serum HBV DNA concentrations. Laboratory Assays Commercially available enzyme immunoassays were used to measure serologic markers of HBV (Abbott Diagnostics, Chicago, Illinois), IgG and IgM antibodies to hepatitis delta virus (Pasteur, Marne la Coquette, France), and antibodies to HCV (Ortho HCV ELISA, Ortho Diagnostic Systems, Inc., Raritan, New Jersey; RIBA HCV, Chiron Corp., Emeryville, California). Serum HBV DNA was assessed quantitatively and qualitatively. Serum concentrations of HBV DNA were measured by molecular hybridization through hybrid capture (Murex Diagnostics, Dartford, United Kingdom). Hybrid capture has a cut-off point of 5 pg/mL and allows titration to 2000 pg/mL. If the titer is greater than 2000 pg/mL, the test result is positive but exact titration cannot be done. We qualitatively assessed HBV DNA by using PCR at baseline and at months 2, 6, and 12 of lamivudine therapy only if serum concentrations of HBV DNA were less than 5 pg/mL as determined by molecular hybridization. We extracted DNA by heating 100 L of serum for 45 minutes. The two sets of primer used for the amplification were from the core region (HBV Core; Sorin-Biomedica, Saluggia, Italy). Polymerase chain reaction was done with the extracted DNA, and 35 cycles were run in a programmable thermoblock (Gen Amp PCR system 9600; Perkin Elmer Cetus Corp., Norwalk, Connecticut). To avoid contamination, each step of the PCR assay was done in a separate room and each series was done with at least three negative controls. We identified DNA by using DNA enzyme immunoassay (HBV Core Gene, ETI-K DEIA; Sorin-Biomedica). Single-stranded DNAs were hybridized with an oligonucleotide probe specific for the amplified region and were assayed by a monoclonal antibody that reacted only with double-stranded DNA. This assay can detect seven equivalent genomes per sample [18]. Antiviral Therapies In the French open-label treatment program, lamivudine was given to patients at a dosage of 300 mg twice daily from December 1993 to May 1995. Then, in May 1995, a protocol amendment recommended reducing the dose by 50% (to 150 mg twice daily). All study patients followed this protocol amendment. Thus, during the 12-month study period, 14 patients received 300 mg of lamivudine twice daily for 1 year and 26 patients received 300 mg of lamivudine twice daily for 9.5 1.2 months followed by 150 mg of lamivudine twice daily for 2.4 1.3 months. Seven patients with high HBV replication at baseline and 3 patients with low HBV replication at baseline received lamivudine in combination with zidovudine, 250 mg/d. The remaining 30 patients received lamivudine alone. Two patients with cytomegalovirus retinitis and no evidence of HBV replication (positivity for antibodies to HBeAg and an HBV DNA concentration < 5 pg/mL) started receiving lamivudine simultaneously with treatment for cytomegalovirus infection (1 patient received foscarnet and 1 received ganciclovir). No patients received interferon- during the course of the study. Other Concurrent Therapies At baseline, 38 patients (29 patients with high HBV replication at baseline and 9 patients with low HBV replication at baseline) were receiving sulfamethoxazole (800 mg/d) and trimethoprim (160 mg/d) as prophylaxis for Pneumocystis carinii pneumonia. One patient with high HBV replication and 1 patient with low HBV replication received sulfadiazine (2 g/d) and pyrimethamine (25 mg/d) as prophylaxis for recurrence of cerebral toxoplasmosis. Two patients with high replication and 1 patient with low replication were receiving bleomycin (5 mg/d for 3 days twice monthly) for cutaneous Kaposi sarcoma. Statistical Analysis Results are expressed as the mean SD. Data were compared using the Mann-Whitney test; a P value less than 0.05 was considered to be significant. Results The clinical and biological characteristics of the 40 study patients are shown in Table 1. No statistically significant differences in age, ratio of men to women, Centers for Disease Control and Prevention clinical stage of HIV infection [19], known duration of HIV infection, or CD4 lymphocyte counts were seen between the high-replication group and the low-replication group (data not shown). Table 1. Primary Clinical Findings at Baseline in Patients Infected with Both HIV and Hepatitis B Virus* Thirty-four patients had progressive HIV disease (a progressive decrease in CD4 lymphocyte count) despite receiving therapy with zidovudine and didanosine; 6 patients had progressive HIV disease despite receiving therapy with zidovudine and did not tolerate didanosine. All patients carried HBsAg for at least 1 year. In 27 patients, HBsAg and antibodies to HIV were found simultaneously in serum 5.9 2.8 years (range, 1 to 10 years) before lamivudine therapy was started. No patients had a history of initial acute hepatitis or decompensated liver disease or were infected with the hepatitis delta virus. One patient had detectable antibodies to HCV in serum. At baseline, two groups of patients were retrospectively identified according to serologic markers of HBV and serum concentrations of HBV DNA as measured by molecular hybridization. The first group comprised patients with high HBV replication (patients who were positive for HBeAg and had serum concentrations of HBV DNA > 5 pg/mL as determined by molecular hybridization) (n = 30). Liver biopsy was done in 6 of these patients 27 9 months (range, 12 to 36 months) before lamivudine therapy was started. These 6 patients had histologic findings consistent with chronic HBV infection, had a mean Knodell score [20] of 6 2.7 (range, 4 to 10), and did not lose serum HBV DNA while receiving interferon- therapy (5 million U/m2 body surface area three times a week for 6 months). Measurements of serum HBV DNA were available for 17 patients from 12 and 6 months before lamivudine therapy was started and for 4 patient

Effect of terlipressin (Glypressin) on hepatorenal syndrome in cirrhotic patients: results of a multicentre pilot study.

Objectives Hepatorenal syndrome (HRS) is a severe complication of cirrhosis, leading to death in more than 90% of cases in the absence of liver transplantation. Several treatments have been attempted as a bridge to liver transplantation. Among such treatments, terlipressin has been studied in several reports, two prospective pilot studies and a double-blind, short-term, controlled haemodynamic study. Promising results have been shown with this drug. The purpose of this multicentre retrospective study was to evaluate the effects of terlipressin on renal function and survival of patients with HRS. Patients and methods Eighteen patients recruited in three liver units with type 1 HRS in 16 cases and type 2 HRS in two cases were given 4 mg/day terlipressin (range 1.5–12) for 7 days (range 2–16). Electrolytes, renal function, mean urinary output, natriuresis, liver function tests, and tolerance of the treatment were monitored regularly. Results A total of 13/18 (72%) patients responded with a mean decline in serum creatinine ranging from 31 to 75% from day 0 to day 5. Eight of these 13 patients had a normal serum creatinine level at day 5. Liver function tests remained unaffected by terlipressin administration. Three local necrosis complications were noted in patients receiving terlipressin continuously via an infusion pump. Two responder patients survived: one of these underwent orthotopic liver transplantation with a follow-up of 24 months; the other is alive with a follow-up of more than 36 months. Patients who responded to terlipressin had lower baseline serum bilirubin and significantly higher serum sodium concentrations than patients who did not respond. Conclusion In this pilot study, improvement in renal function was noted in 72% of cases after administration of terlipressin, and was associated with long-term survival in two patients. Parameters associated with response to terlipressin and increased survival should be defined better in a large cohort of cirrhotic patients with HRS.

Truth survival in clinical research: An evidence-based requiem?

Science progresses through a series of paradigms that are held to be true until they are replaced by a better approximation of reality (1). Since the development of the steam engine in the late 18th century, economists have recognized 50-year cycles during which critical technological innovation is introduced (2). In 1997, Hall and Platell (3) estimated the half-life of dogma relating to the practice of surgery. From their analysis of 260 abstracts published from 1935 to 1994, they estimated that the half-life of truth for clinical conclusions in the surgical literature was 45 years. We hypothesized that some factors should be related to this truth survival. The first hypothesis was that conclusions derived from better methodology should have a longer half-life. If correct, this observation could be a validation of good methodology, often called evidence-based medicine (4). Therefore, we compared the survival of conclusions from meta-analyses with those from isolated, randomized trials or nonrandomized studies. For the conclusions from randomized trials (isolated trials or meta-analyses), we also compared the survival rate on the basis of high versus low methodologic scores. The second hypothesis was that survival of truth should be higher for negative conclusions than for positive conclusions. A negative conclusion has a better chance of survival because the only way it would not continue to be negative is if it were found to be false. A positive conclusion risks being found to be false or becoming obsolete. We also thought that publication of a negative conclusion in a reputable journal often indicated that a previous positive conclusion had been found to be false. We concluded, therefore, that this second-line analysis should be of higher quality. To reduce heterogeneity in sampling and evaluation, we chose a single medical disciplinecirrhosis and hepatitisand focused on two selective journals. We tried to categorize the conclusions into three groups: those that were true (referred to as true in this article), those that were not false but became obsolete (referred to as obsolete), and those that are now considered false (referred to as false). An example of an obsolete conclusion is the efficacy of immunoglobulins for preventing hepatitis A virus infection, since an effective vaccine is now available. An example of a false conclusion is the efficacy of corticosteroids for treating acute viral hepatitis. Methods We identified original articles about cirrhosis or hepatitis in adults from 1945 to 1999. The articles were divided into eleven 5-year periods. Nonoriginal studies and studies involving children were excluded. Selection of Nonrandomized Studies In each 5-year period, we selected 20 nonrandomized articles published in two journals10 from Lancet and 10 from Gastroenterology. We chose these journals because they have published clinical studies about hepatitis and cirrhosis since at least 1945, they are peer-reviewed and highly selective, and they have impact factors greater than 10. Articles from 1945 to 1985 were selected by a hand search. Because a true randomization was very difficult to organize, we selected by order of publication within each 5-year period. We selected the first article about cirrhosis or hepatitis that appeared within each 5-year period, then the last published article in the period, then the second article after the first, then the second-to-the-last article, and so on up to 20 articles. From 1985 to 1999, we used a PubMed electronic search and specified the following limits: cirrhosis or hepatitis, human, and Lancet or Gastroenterology. Abstracts were downloaded by using a similar selection method, stratified into 5-year periods. We selected the first abstract listed on the first electronic page, then the first on the last electronic page, then the last on the second electronic page, then the last on the next-to-the- last electronic page, and so on up to 20 articles. Selection of Randomized Trials In each 5-year period, we tried to select 20 randomized trials about cirrhosis or hepatitis, 10 from Lancet and 10 from Gastroenterology. This was possible from 1970 to 1999. From 1945 to 1969, we selected all randomized trials that could be identified in any journal (range, 4 trials [1945 to 1950] to 20 trials [1965 to 1969]). From 1945 to 1982, we used the hand searching method previously described (5). We completed the random selection by hand searching articles from 1982 to 1985 and using PubMed (as described for nonrandomized studies) to search for articles from 1985 to 1999. Selection of Meta-Analyses From 1945 to 1992, we used a hand-searching method, as described in a systematic review of meta-analyses (6). Thereafter, we used PubMed and specified the following limits: meta-analysis and cirrhosis or hepatitis. Because of the limited number of meta-analyses, we included all journals. All of the meta-analyses consisted solely of randomized trials. Database Development and Observer Review We obtained abstracts from all of the articles and selected the sentence from each abstract that seemed to best summarize the findings. These sentences were then copied to a database. Editing of these sentences was restricted to the rephrasing of outdated terminology and the elimination of redundant words. Six hepatologists, called observers, assessed a form that contained the selected conclusion sentences in a random order. Observers were full-time hepatologists with different clinical subspecialties (viral hepatitis, n = 2; HIV, n = 1; fibrosis, n = 1; alcoholic liver disease, n = 1; and transplantation, n = 1); worked in the same hospital; and were between 31 and 65 years of age. They had graduated from six different universities; three had worked in the United States, each in a different university. Observers were blinded to period, journal, authors, method (meta-analysis, randomized trial, or nonrandomized study), and the methodologic quality from which each conclusion was derived. They classified each conclusion into one of three categories: still true in 2000, obsolete but not false in 2000, or false in 2000. Quality Assessment of Methodology and Consideration of Prognostic Factors Independent of this study, the quality of the randomized trials was assessed by means of a scoring method (range, 2 to 14; mean, 12) that included 14 items (7, 8). Also independent of this study, the quality of the meta-analyses was assessed by means of a slightly modified version (6) of the scoring method established by Sacks and coworkers (9) and described in detail elsewhere (7). This scoring method (range, 0 to 54; mean, 27) included 27 items. We analyzed the meta-analyses that combined individual data as a separate category of research. In classic meta-analysis, the results of each trial are combined. In meta-analysis that combines individual data, the results for each patient are combined with the patients prognostic factors, thus permitting better adjustment of the treatment effect on prognostic factors. Articles were rated as high quality when the score was greater or equal to the mean (12 for randomized trials and 27 for meta-analyses) and as low quality when the score was less than the mean. The methodologic quality of nonrandomized studies was classified as low because no specific scoring method was available. In addition to methodologic quality, the following factors were considered: negative or positive conclusion, type of disease (hepatitis, portal hypertension, alcoholic liver disease, primary biliary cirrhosis, or miscellaneous), domain of clinical research (therapeutic, diagnostic, or cognitive study [cognitive studies were defined as explanatory studies that did not assess treatment or diagnostic tests]), journal of publication (Lancet, Gastroenterology, or other), and specialty (medicine or surgery). Statistical Analysis A conclusion was considered to be true, obsolete, or false when three or more observers stated it to be so. When there was a split decision (3 to 3) about whether conclusions were true or not true (9 of 474 articles [1.9%]), the conclusion was considered to be true. When there was a split decision (3 to 3) about whether conclusions were obsolete or not obsolete (26 of 474 articles [5.5%]), the conclusion was considered to be obsolete. When the article was not classified as either true or obsolete, it was considered to be false. Conclusions from older research are at greater risk for being refuted or becoming obsolete than are conclusions from more recent studies. Because the end points were highly time dependent, we used time-dependent analyses. The half-life was calculated according to the KaplanMeier method, using the censored time as the duration between the year of publication and the year 2000. The censored time is the time at risk for being refuted or found obsolete. For example, an article published in 1950 had a censored time of 50 years. First, we analyzed the truth survival: If the conclusion was assessed to be true, it was censored at 50 years. If the conclusion was assessed to be false or obsolete, it was considered a failure (death of truth). Second, we analyzed the nonfalse survival; true or obsolete conclusions were considered censored at 50 years. If the conclusion was false, it was considered a failure (death of nonfalse). The factors were compared by using the two-sided log-rank test and the multivariate proportional hazards regression analysis. Agreement among observers was analyzed by using statistics. Results Characteristics of the 474 identified articles are given in Table 1. All nonrandomized studies were published in Lancet (50%) or Gastroenterology (50%); randomized trials were published in Lancet (29%), Gastroenterology (41%), or other journals (30%); and 92% of meta-analyses were published in other journals. Compared with the total number of articles published every year about hepatitis or cirrhosis, this sample represents less than 0.1% of nonran

Blood neutrophil functions and cytokine release in severe alcoholic hepatitis : effect of corticosteroids

BACKGROUND/AIMS Several observations point to an important role of interactions between polymorphonuclear neutrophils and cytokines in severe alcoholic hepatitis. The polymorphonuclear neutrophil activation status and the local and systemic pro- and anti-inflammatory cytokine responses were quantified. The effect of corticosteroids, widely used in this setting, was evaluated using these parameters. METHODS We studied blood polymorphonuclear neutrophil functions in terms of L-selectin and beta2-integrin expression, H2O2 production and IL-8 and tumor necrosis factor alpha synthesis capacity. We also measured IL-8, tumor necrosis factor alpha and IL-10 plasma and liver tissue levels. Fifteen patients with alcoholic hepatitis were compared to 15 patients with alcoholic cirrhosis without alcoholic hepatitis, and to 10 healthy volunteers. The impact of a 28-day course of corticosteroids on blood neutrophils activation status and cytokine levels was evaluated in patients with alcoholic hepatitis. RESULTS Blood polymorphonuclear neutrophils were activated, as shown by increased H2O2 production (48+/-6 vs 29+/-6 MFI in healthy controls), and decreased L-selectin expression (300+/-61 vs 449+/-59 in healthy controls). Upon stimulation, polymorphonuclear neutrophils synthesized large amounts of IL-8 (21.7+/-9.2 ng/ml vs 8.8+/-10 ng/ml in healthy controls) and tumor necrosis factor alpha (524+/-132 pg/ml vs 79+/-144 pg/ml in healthy controls). Tumor necrosis factor alpha and IL-8 plasma and tissue levels were markedly increased as IL-10 was barely detectable in alcoholic hepatitis patients, compared to cirrhotic patients and healthy controls. During steroid therapy, plasma levels of the pro-inflammatory cytokine IL-8 fell as early as day 14, while levels of the anti-inflammatory cytokine IL-10 increased on day 21. Finally, polymorphonuclear neutrophil functions returned to normal after treatment. CONCLUSION Severe alcoholic hepatitis appears to be associated with polymorphonuclear neutrophil activation and an imbalance between pro- and anti-inflammatory cytokines; during steroid therapy a normalization of these parameters was observed.

Propranolol and sclerotherapy in the prevention of gastrointestinal rebleeding in patients with cirrhosis: a meta-analysis

BACKGROUND/AIMS A meta-analysis of nine selected randomized trials was performed to compare the effects of propranolol and sclerotherapy in the prevention of rebleeding and on survival in patients with cirrhosis. METHODS Five end points were assessed: rebleeding, esophageal rebleeding, death, death due to bleeding, and adverse events. Analyses were performed according to the intention-to-treat method. For each end point, heterogeneity and treatment efficacy were assessed by the Der Simonian and Peto methods. When a significant difference was observed, sensitivity analyses were performed by successive stratification according to treatment duration, type of publication, severity of cirrhosis, and methodological quality. RESULTS The mean percentage of patients free of rebleeding, the mean survival rate and the mean percentage of patients free of death from bleeding were not significantly different between patients treated with propranolol and those treated by sclerotherapy. The mean percentage of patients free of variceal rebleeding was 39% in propranolol group and 55% in sclerotherapy group (mean difference: 17%, 95% confidence interval: 9-25%, p < 0.001). The mean percentage of patients free of adverse events was significantly higher in the propranolol group than in the sclerotherapy group (mean difference: 22%, 95% confidence interval: 6-38%, p < 0.007). CONCLUSION In patients with cirrhosis and esophageal varices, endoscopic sclerotherapy is more effective than propranolol in preventing variceal rebleeding, but the incidence of adverse events is significantly higher with sclerotherapy. There was no difference in survival between the treatments. Propranolol should be considered as a first choice treatment for preventing rebleeding.