| |
Liver Iron Levels NOT Directly Linked to Fibrosis
|
| |
| |
Liver iron is a surrogate marker of severe fibrosis in chronic hepatitis C
Journal of Hepatology April 2007
Dominique Guyader1, Anne-Sophie Thirouard2, Lars Erdtmann1, Nafissa Rakba1, Sylvie Jacquelinet1, Helene Danielou1, Michele Perrin1, Anne-Marie Jouanolle3, Pierre Brissot1, Yves Deugnier1
"....our study does not favor a direct link between iron and fibrosis progression and indicates that liver iron should be better considered as a surrogate marker of severe fibrosis in chronic hepatitis C than as a fibrogenic factor per se.... We analyzed the data using both the fibrosis index which allows to adjust the results to the duration of the infection and the Metavir fibrosis score which does not imply a linear evolution of fibrosis over time. In the multivariate analysis, we controlled the analysis for all potential confounding variables known to be associated with both iron metabolism disturbances and fibrosis progression. Using this approach, the results indicated that iron had no impact on fibrosis progression. The link which was evident in univariate analysis disappeared in multivariate analysis. Therefore, it is likely that the variables introduced for adjustment (age at infection or at the time of the biopsy and duration of the disease, histological activity, steatosis and excessive alcohol intake) were responsible for the link. Actually, all these variables are predictive of liver fibrosis but also responsible for disturbances of iron parameters...."
Introduction
Increased levels of serum iron, transferrin saturation and ferritinemia are encountered in 20-35% of patients with chronic hepatitis C and associated with mild to moderate increase of hepatic iron load, predominant in sinusoidal location, in 10-35% of cases [1], [2].
The frequency and the degree of iron overload increases with the stage of fibrosis. Experimental iron overload in HCV infected chimpanzees seems to enhance HCV pathogenicity [3]. Therefore, it was postulated that iron could enhance the rate of fibrosis progression, possibly through the activation of free radical generation [4]. However, from a clinical background, the impact of iron on fibrogenesis remains controversial. Most studies concluded on an association between the amount of liver iron and progression of fibrosis [5], [6], [7], [8], [9], [10] while more recent studies did not find such an association [11], [12]. The reasons for these discrepancies have been recently fully reviewed [2], [13]. Presumably, they rely on differences in study population and lack of control for confounding variables. Variables defined in studies on fibrosis progression in HCV infected patients were male sex, duration of infection, age at infection >40 years, daily consumption of alcohol >50/g/day degree of hepatocellular necrosis or inflammation, and features of the metabolic syndrome (such as high body mass index, diabetes, and liver steatosis) [14]. All these variables can also influence iron metabolism: serum ferritin levels increase with age and male sex [15] and iron metabolism disturbances are common in alcoholic liver diseases [16], insulin resistance iron overload syndrome [17] and non-alcoholic steatohepatitis (NASH) [18], [19]. Furthermore, it is noteworthy that, also in NASH, the impact of iron on fibrosis progression is still discussed [20]. Because all the factors responsible for rapid fibrosis progression can also be associated with an increase in ferritin and iron stores, the assertion of a link between iron and fibrogenesis can only rely on an appropriate analysis adjusted for all these confounding factors. Some studies were based on simple univariate comparison between patients with severe and those with moderate fibrosis and the results obtained were only adjusted with age, gender and duration of infection. Other reports adjusted the results according to alcohol intake or histological activity [10], [21], [22]. A recent Italian study [21] analyzed the link between steatosis and liver iron and found that hepatic steatosis was more common in patients with liver iron than in those without, but did not find a clear relationship between the amount of iron accumulation and the grade of steatosis. The age at infection which is a major determinant of fibrosis progression [23] has only been taken into account in a recent study which did not find any association between liver iron indices and fibrosis [11]. Finally, Ryder et al. did not find an association between liver iron and fibrosis progression between two paired liver biopsy performed at a median interval of 2.5 years in a particular population of untreated patients with mild disease [24].
Therefore, in order to better understand the relationship between iron and chronic hepatitis C and determine if iron has an independent influence on fibrosis progression, we analyzed a large number of consecutive patients, taking into account all variables known to influence iron metabolism and/or fibrosis progression.
ABSTRACT
Background/Aims
Patients with chronic hepatitis C have frequently mild to moderate liver iron overload which increases with fibrosis stage. Thus, it has been postulated that iron could enhance the progression of fibrosis. However, the real impact of iron is still controversial. The study was undertaken to determine the effect of confounding variables. All factors known to influence both iron overload and fibrosis were taken into account.
Methods
Five hundred and eighty-six patients, who had liver biopsy performed prior to antiviral treatment, were included. Serum ferritin and liver iron were correlated with clinical, biological and histological variables in univariate and multivariate analysis. The impact of iron on fibrosis was evaluated in multivariate analysis in the whole group and in the subgroup of 380 patients with available date of infection.
Results
Hyperferritinemia, encountered in 27%, was associated with liver iron deposits in only 46% of cases. Liver iron was elevated in 17%, and correlated with age, male sex, and alcohol intake. The univariate strong link which existed between liver iron and fibrosis disappeared after adjustment for confounding variables.
Conclusions
According to the results of this study, liver iron should be considered more as a surrogate marker for disease severity than as a fibrogenic factor per se.
Discussion
The issue of the relation between iron and fibrosis progression in hepatitis C is complex and debated. In this study, we investigated the impact of iron on fibrosis progression in HCV infected persons. The strength of the present analysis is related to the high number of patients included consecutively allowing for a high statistical power. This statistical power was validated by the fact that in the multivariate analysis, at the exception of gender which did not reach statistical significance, we individualized independent variables identified in pivotal studies (age at infection, duration of infection, alcohol intake, Metavir activity, steatosis) [23]. We analyzed the data using both the fibrosis index which allows to adjust the results to the duration of the infection and the Metavir fibrosis score which does not imply a linear evolution of fibrosis over time. In the multivariate analysis, we controlled the analysis for all potential confounding variables known to be associated with both iron metabolism disturbances and fibrosis progression. Using this approach, the results indicated that iron had no impact on fibrosis progression. The link which was evident in univariate analysis disappeared in multivariate analysis. Therefore, it is likely that the variables introduced for adjustment (age at infection or at the time of the biopsy and duration of the disease, histological activity, steatosis and excessive alcohol intake) were responsible for the link. Actually, all these variables are predictive of liver fibrosis but also responsible for disturbances of iron parameters. However, among these variables, Metavir activity has probably no influence because of the lack of correlation between TIS and Metavir activity in univariate analysis. The influence of steatosis is also debatable because of the absence of correlation with iron in multivariate analysis. It is likely that variables independently linked to liver iron in multivariate analysis (age and alcohol intake) have the greater influence. Therefore, according to our results, serum ferritin and liver iron should better be considered as surrogate markers for non-iron related variables associated with fibrosis progression than as a fibrogenic factor per se.
The retrospective collection of glycemia, serum cholesterol and triglycerides could have introduced a bias in the interpretation of those variables. However, the data were available in a large group of patients and, at the exception of a link between serum ferritin and serum glucose or triglycerides level, were not correlated with iron perturbations. Moreover, the hallmark of these metabolic features is represented by liver steatosis which was determined for all the patients.
Our results are thus in discordance with most data of the literature supporting a direct link between liver iron and fibrosis progression. We cannot exclude different genetic background among different population that might influence hepatic iron accumulation or fibrosis progression. However, the discrepancy can also be explained by several factors. To our knowledge, although some studies adjusted their results on different cofactors, no report took into account all the potential confounding factor in the same analysis. Particularly, age at infection was only reported in one recent study which did not find an association between liver iron and fibrosis [11]. Steatosis was considered in only one study in relation with liver iron content and did not show correlation [21]. Other potential bias explaining the different results of some previous studies can be the inclusion of patients irrespective of previous interferon treatment. Usually, patients treated by ribavirin were excluded in studies [21], [22] because it is well known that ribavirin could be responsible for an increase in liver iron content [28]. But patients previously treated by interferon should also be excluded because interferon treatment decreases fibrosis progression even in case of non-response [29] and also had an effect on liver iron load [30]. Also, some studies included HFE homozygous C82Y patients in the analysis and this is susceptible to overestimate the impact of iron on fibrogenesis [31]. Therefore, to avoid these bias, we included only patients who had not been previously treated and discarded C282Y homozygotes from analysis. In selection criteria, we excluded patients with previous iron depletion therapy. This could have introduced a selection bias if patients with most severe iron overload were excluded. Actually, no patient was discarded for this reason because the study was performed at the time of the first presentation before any therapeutic procedure.
Because of its multifactorial origin, hyperferritinemia was poorly correlated with liver iron content and stainable liver iron was not detected in almost 50% of patients with elevated serum ferritin. It is therefore inadequate to consider that the level of ferritin reflects the level of iron stores and the conclusions reached in some papers using serum ferritin should not be generalized in terms of iron stores [32]. The demonstration of iron overload should rely on liver biopsy, while magnetic resonance imaging could be a valuable non-invasive tool [33] if a liver biopsy is not indicated.
It is important to stress that we did not include patients with end-stage liver disease and therefore cannot extend our conclusions to this subgroup of patients. The cirrhotic patients included in the study had no hepatic insufficiency, nor decompensated liver disease, because they were selected only if a liver biopsy was performed. It has been demonstrated, both in non-biliary cirrhosis [27] and in hepatitis C cirrhosis [22], that the degree of liver iron increases with the severity of the liver disease, with 5-10% of patients having an hepatic iron index (liver iron divided by age)
>1.9, which was the classical diagnostic feature of genetic hemochromatosis before the discovery of the HFE gene. Also, this severity in end-stage liver disease contrasts with the minimal overload that we encountered in most of the cirrhotic patients included in the study. It is therefore likely that iron overload is a late and non-specific event in cirrhosis and presumably, at this stage, is secondary to different mechanisms related to the development of hepatic insufficiency or portal hypertension and shunting.
We did not find any influence of metabolic features (BMI, fasting glucose, triglycerides and cholesterol) or steatosis on liver iron staining whereas both factors were correlated with the serum level of ferritin. This was surprising because iron overload has been reported in insulin resistance syndrome [17]. It would incite to perform a similar study in the metabolic syndrome, in order to establish if there is a correlation between iron and the degree of metabolic abnormalities. One possible explanation could also be the difficulty to stain iron in case of massive steatosis which might lead to underestimation of TIS in that event. This could explain why TIS is higher in the group of patients with moderate steatosis by comparison of the group with massive steatosis (Fig. 3).
The absence of direct link between iron and fibrosis progression does not incite to propose iron depletion to patients with iron overload. However, venesections induce a significant and constant decrease in transaminase levels, without effect on HCV viral load, and could decrease liver inflammation [34]. Similar results have been obtained in NAFLD [35]. We must consider that those iron depletion trials were not designed to treat iron overload, and patients included were not selected on the basis of iron overload. Therefore, venesections do not treat iron overload but induce an iron deficient state which probably cannot be prolonged over time without significant side effects.
In conclusion, our study does not favor a direct link between iron and fibrosis progression and indicates that liver iron should be better considered as a surrogate marker of severe fibrosis in chronic hepatitis C than as a fibrogenic factor per se.
Patients and methods
Patients
Patients with chronic hepatitis C were consecutively included in the study if they had liver biopsy allowing for accurate assessment of liver fibrosis (length>20mm). Chronic hepatitis C was assessed by HCV RNA detection. Patients were excluded if the size of liver biopsy did not allow for accurate histological assessment of fibrosis, steatosis or liver iron, or in case of HbsAg positivity or HIV infection. Further criteria for exclusion were previous treatment with interferon and/or ribavirin, known C282Y homozygosity, previous iron depletion therapy, or coexisting affection which could influence interpretation of iron measurements or liver fibrosis (such as hematological disorders, chronic inflammatory diseases, porphyria).
Methods
The following clinical and biological data were recorded: route of exposure (transfusion, drug abuse, unknown), age at infection defined by the time of the first exposure to IV drug use or transfusion (available when the exposure route was identified), weight and height allowing for the determination of body mass index (BMI as kg/m2), serum iron (N<24μmol/L), transferrin saturation (N<0.45), serum ferritin (N<300 for men and 250μg/L for women), transaminases (ALAT and ASAT), expressed as multiple of the upper normal limit (xN), γ-glutamyl transpeptidase (GGTxN) and fasting serum glucose (μmol/L). Patients were considered as excessive drinkers if they had a daily consumption > 50g/day of alcohol during more than 2 years. Viral genotype, serum triglyceride, serum cholesterol were recorded retrospectively at the time of the liver biopsy. Viral load could not be studied because the data were missing for some patients and the commercial tests used throughout the study period were different. Blood for biological variables was drawn in a fasted state.
Liver biopsies were blindly reviewed. Activity and fibrosis were scored using the Metavir system [25]. Steatosis was recorded and classified as moderate (5-25%) or severe (>25%) steatosis according to the number of hepatocytes involved. Liver iron was assessed by the Deugnier's Total Iron Score (TIS), ranging from 0 to 60, which is the addition of the hepatocyte (HIS=0-36), sinusoidal (SIS=0-12), and mesenchymal (MIS=0-12) iron scores [26]. The use of this semi-quantitative score was preferred to the biochemical hepatic iron concentration because it gives information on the cellular and lobular distribution of iron and its correlation with hepatic iron concentration has been validated [26]. Moreover, the goal of the study was not to assess the level of iron overload, which has been evaluated in several well-designed studies, but to determine the factors associated with iron overload and the impact of iron on fibrosis. In these settings, the histopathological score has been proven to be a sensitive indicator of very mild iron deposition in non-hemochromatosis liver cirrhosis [27]. The fibrosis progression index was calculated by dividing the Metavir F score by the estimated duration of the infection at the time of the liver biopsy (UF/Year).
The protocol of the study was approved by the Ethic Committee of the university hospital of Rennes and patients gave their written informed consent to participate.
Statistical analysis
Data are presented as means±SD or median [interquartile range] according to variables distribution. Chi-square test, and non-parametric statistical tests (Spearman, Mann and Whitney U test, Kruskal-Wallis tests) were used. Multivariate analysis was done using logistic stepwise backward regression. Variables were included in multivariate model if the p value reached 0.10in univariate analysis. A p value of 0.05 was considered significant. Analyses were performed using SPSS 11.5 software (SPSS Inc. Chicago, IL).
Results
Patients
Five hundred and eighty-six out of 687 consecutive HCV positive patients were included in the study. The 101 patients who were not included were discarded because the size of the liver biopsy was insufficient for accurate assessment of fibrosis, iron, and steatosis or not available for blind review by the pathologist. 380/586 patients had estimable date of infection. No patient had previous depletion therapy. Main clinical, biological, histological features of the study population are given in Table 1.
Correlation between serum ferritin and histological iron
Serum iron was increased in 33% of cases (30μmol/L [7] in the group of patients with increased serum iron), transferrin saturation in 15% (0.54 [0.17]), serum ferritin in 27% (445μg/L [264]) and TIS in 21% (9.8±5.8; 8 [7]). The histological iron overload was hepatocytic in 34% of cases, sinusoidal in 12% and mixed in 54%. In the case of mixed iron overload, the median HIS/TIS ratio of 0.62 indicated a slight hepatocytic predominance. Although there was a statistical correlation between serum ferritin and TIS (Fig. 1: ρ=0.461; p<10-4), only 46% of patients with increased serum ferritin had stainable liver iron and no ferritin threshold value could definitely predict the presence of histological iron (except for ferritin values>1500μg/L reached only by a small number of patients). On the other hand, increase in TIS was detectable in 12% of patients with normal serum ferritin. Table 2 gives the main clinical, biological, and histological features according to serum ferritin level (see Fig. 2).
Liver iron: clinical, biological and histological correlations
TIS was significantly (p<10-4) higher in men (2.74±5.46) than in women (1.05±3.31). There was a significant correlation between TIS and age (ρ=0.219; p<10-4). There was no correlation with duration of infection (p=0.32). There was a significant correlation with alcohol intake (excessive drinkers 3.0±5.6 [0-25]; non-excessive drinkers 1.5±4.1 [0-24]; p=0.006). TIS was correlated with ASAT (ρ=0.111; p=0.007), GGT (ρ=0.175; p<10-4) and BMI (ρ=0.137; p=0.0023). TIS was not correlated with ALAT (p=0.14), cholesterol (p=0.18), triglycerides (p=0.32), glycemia (p=0.90) and genotype (p=0.52). Table 3 gives clinical, biological and histological features in patients who had no iron staining by comparison with patient who had positive iron staining. Fig. 3 shows the results of TIS in relation with histological parameters. There was no correlation with histological activity (p=0.24). There was a significant correlation with fibrosis (p<10-3) and steatosis (p=0.04). In multivariate logistic regression model (Table 5), the increase of TIS was independently associated with age, male sex, and excessive intake of alcohol. There was no influence of metabolic features (either steatosis p=0.58 or BMI p=0.66). Metavir Fibrosis score was not correlated with TIS in multivariate analysis (p=0.71).
Factors associated with fibrosis
Four hundred and seventy-four patients (81%) had Metavir F< 2 and 112 (19%) had Metavir F3 or F4 fibrosis. Table 4 gives univariate comparison between these two different groups. Table 5 gives the results of multivariate analysis performed in the whole group of patients and in the subgroup of patients with known duration of infection. TIS was not associated with fibrosis (p=0.45 in the whole group and p=0.21 in the group with known duration of infection). There was also a significant relationship (ρ=0.139; p=0.006) between TIS and fibrosis progression index in univariate analysis. The median fibrosis progression was 0.088 in the group of patients with stainable liver iron and 0.060 in the group of patients without stainable liver iron (p=0.007). However, this correlation between TIS and fibrosis progression index disappeared (p=0.40) after adjustment with age at infection, gender, alcohol intake, steatosis, and activity.
Table 5. Variables selected in multivariate analysis
In the multivariate analysis the usual suspects were linked to severe fibrosis: age at liver biopsy, excessive alcohol intake, steatosis, Metavir A, duration of infection.
*Odds ratio calculated for an increase of 10 years. The multivariate logistic regression analysis was performed in 584/586 patients and in 378/380 patients with available date of infection.
|
|
| |
| |
|
|
|
