Home e-TOC, the free content delivery service
Home Help [Feedback] [For Subscribers] [Archive] [Search] [Search Result] || Sign In as Member || Contact Subscription Administrator at this Site || FAQ
Reprint (PDF) Version of this Article
PubMed Citation
Related Articles in PubMed
Download to Citation Manager
This article has been cited by other articles
Search Medline for articles by:
GUYADER, D. || DEUGNIER, Y.

GASTROENTEROLOGY 1998;115:929-936

Noninvasive Prediction of Fibrosis in C282Y Homozygous Hemochromatosis

DOMINIQUE GUYADER*, CHRISTIAN JACQUELINET, ROMAIN MOIRAND*, BRUNO TURLIN§, MICHEL H. MENDLER*, JACQUES CHAPERONparallel , VÉRONIQUE DAVID, PIERRE BRISSOT*, PAUL ADAMS#, and YVES DEUGNIER*

* Clinique des Maladies du Foie and INSERM Unité 49,  Etablissement Français des Greffes, § Anatomie Pathologique B, parallel  Département de Santé Publique,  Laboratoire de Génétique Moléculaire; Hôpital Universitaire Pontchaillou, Rennes, France; and # Department of Medicine, University of Western Ontario, London, Ontario, Canada

    Abstract
Top
Abstract
Introduction
Methods
Results
Discussion
References

Background & Aims: The diagnosis of hemochromatosis is now possible for C282Y homozygous patients using noninvasive molecular genetic tests. The aim of this study was to define noninvasive factors predictive of severe fibrosis (bridging fibrosis or cirrhosis) to avoid unnecessary liver biopsies in such patients. Methods: Clinical and biological data were recorded at the time of diagnosis in 197 French C282Y homozygous patients, 52 (26%) of whom had severe fibrosis. Variables significantly linked to severe fibrosis using univariate analysis were entered into a multivariate stepwise analysis. These variables were combined to obtain a simple index allowing for prediction of severe fibrosis. Results: Serum ferritin, hepatomegaly, and serum aspartate aminotransferase were selected using multivariate analysis. Their combination applied to the 96 patients with ferritin level of <= 1000 µg/L, normal aspartate aminotransferase values, and absence of hepatomegaly showed that no severe fibrosis was encountered in this subgroup of patients. The results were validated in 113 C282Y homozygous patients in Canada with a good reproducibility of negative prediction but a poor reproducibility of the positive prediction of severe fibrosis. Conclusions: In C282Y homozygous patients, the diagnosis of severe fibrosis relies on liver biopsy, but absence of severe fibrosis can be accurately predicted in most patients on the basis of simple clinical and biochemical variables.

    Introduction
Top
Abstract
Introduction
Methods
Results
Discussion
References

Genetic hemochromatosis is an autosomal recessive disorder characterized by an increased intestinal absorption of iron, leading to early abnormalities in serum iron test results (especially elevated transferrin saturation) and late clinical symptoms.1 When diagnosed before the onset of cirrhosis and treated adequately, hemochromatosis does not reduce life expectancy.2,3 On the contrary, when hemochromatosis is detected after cirrhosis has developed, there is a high risk of hepatocellular carcinoma2-4 that persists even in iron-depleted patients.5 Patients with cirrhosis have a 5.5 relative risk of death compared with noncirrhotic patients.6

Liver biopsy has previously been an important step in the diagnosis of hemochromatosis7 because (1) it shows liver iron overload and assesses the hepatocytic type of iron deposition with typical periportal and perilobular distribution8; (2) it allows for biochemical determination of hepatic iron concentration (HIC), and a hepatic iron index (HIC/age ratio) >1.9 is a helpful means of establishing a diagnosis of homozygous hemochromatosis9,10; and (3) it provides assessment of the degree of fibrosis, which is of major prognostic significance.

The discovery of the hemochromatosis gene will lead to new diagnostic strategies. Feder et al.11 have identified a C282Y mutation (change in cysteine to tyrosine at position 282) on a gene initially called HLA-H and now HFE,12 which encodes a major histocompatibility complex class I-like molecule. The HFE C282Y mutation is strongly associated with hemochromatosis because 60%-100% of patients with typical phenotypic hemochromatosis worldwide are homozygous for this mutation.11,13-19 The determination of the C282Y status represents a simple and efficient means of diagnosing hemochromatosis. Therefore, in the case of a homozygous C282Y mutation patient, liver biopsy is no longer indicated for the diagnosis, and the rationale for performing liver biopsy is now restricted to the need for assessment of fibrosis. Although controversial, there is a general agreement to perform a liver biopsy in most hemochromatosis patients, except for young (<30 years) asymptomatic patients identified by family studies or by screening with a genetic test, because it is unlikely that those individuals will have cirrhosis or a significant increase in fibrosis.7 However, until now, no data were available that could permit to select, among homozygous hemochromatotic patients, those at low risk of cirrhosis for whom liver biopsy is unnecessary. Liver biopsy is an invasive procedure, relatively safe, but still associated with a significant morbidity and a mortality rate estimated between 0.015% and 0.03%.20,21

The aim of the present study was to identify noninvasive variables that could predict or exclude severe fibrosis to avoid an unnecessary liver biopsy in such patients.

    Patients and Methods
Top
Abstract
Introduction
Methods
Results
Discussion
References

Patients

Patients were selected from a database of HFE-genotyped patients with various conditions of hepatic iron overload on the following criteria: (1) homozygosity for the C282Y mutation; (2) availability of clinical examination, biological data, and liver biopsy performed at the time of diagnosis allowing for accurate assessment of hepatic fibrosis; and (3) absence of hepatitis C virus antibodies, hepatitis B surface antigen, and/or histological picture of chronic active hepatitis. Using these criteria, 197 patients were eligible for analysis.

Methods

The following data were recorded at the time of diagnosis: age, gender, occurrence of hepatomegaly (assessed either clinically or by ultrasonography), prothrombin index, serum iron (N <=  25 µmol/L), transferrin saturation (N <=  0.45), serum ferritin (measured by an immunoradiometric assay using Ferritin Magic; Ciba Corning SA, Le Vésinet, France; N <=  400 µg/L), serum aspartate aminotransferase (AST), alanine aminotransferase (ALT), and gamma -glutamyltranspeptidase expressed as multiple of upper limit of normal value (ULN). In addition, history of alcohol consumption was carefully recorded, and patients were considered chronic excessive drinkers if they had drunk more than 80 g/day of alcohol for men and 60 g/day of alcohol for women during more than 5 years and/or in case of clinical, biological, or histological evidence of chronic alcoholism.

Needle liver biopsy was obtained at the time of diagnosis before treatment and routinely processed after fixation in 10% neutral formaldehyde and embedding in paraffin. Slides were stained using H&E-saffron, Gordon-Sweets, Sirius red, or Masson's trichrome for connective tissue assessment and Perl's for iron assessment. Fibrosis was graded in a 5-grade scale (stage 0, no fibrosis; stage 1, portal fibrosis; stage 2, extensive portal fibrosis; stage 3, bridging fibrosis; and stage 4, cirrhosis). Pathologists were blinded to clinical information, and the same criteria were used in France and in Canada. The term severe fibrosis referred to stage 3 or 4 fibrosis. HIC was measured using Barry and Sherlock's method22 (normal <=  36 µmol/g of dry liver weight).

Genetic Studies

Analysis of the C282Y mutation was performed by an amplification of exon 4 before restriction fragment length polymorphism analysis as previously described.17

Statistical Analysis

Results were expressed as median (25th-75th percentile). Values were considered significant with P < 0.05. Univariate analysis was performed using an analysis of variance for quantitative variables with normal distribution, the nonparametric Mann-Whitney U test for quantitative variables with nonnormal distribution, and chi 2 test for qualitative data. Three models were assessed to predict severe fibrosis. Two of them, in which variables significantly associated with severe fibrosis were entered in univariate analysis, relied on classical multivariate analysis tools (Statistical Package for Social Sciences; SPSS Inc., Chicago, IL). The first model was elaborated using stepwise logistic regression (backward and forward procedures, Wald criteria). The second model was constructed using discriminant analysis with a stepwise procedure for selection of variables. The third model attempted to be a simplified model for clinical use. This simplified clinical model was based on a decision tree combining variables selected in multivariate analysis. Optimization of thresholds was based on receiver operating characteristics (ROC) curve.

Validation of the Predictive Models

Finally, these three models were validated in a second population of 113 homozygous C282Y Canadian patients. Serum ferritin was measured using a commercial radioimmunoassay (Quantimmune ferritin IRMA; Bio-Rad, Hercules, CA; upper limit of normal value, 350 µg/L).

    Results
Top
Abstract
Introduction
Methods
Results
Discussion
References

Patients

Main clinical and biological data of the 197 French C282Y patients are given in Table 1. Among these, 98 (50%) had no fibrosis, 47 (24%) had moderate (stage 1 or 2) fibrosis, and 52 (26%) had severe (stage 3 or 4) fibrosis. Patients were identified by family screening in 51 cases (26%), by asymptomatic increase in serum iron test values (serum iron, transferrin saturation, and/or serum ferritin) in 79 other cases (40%), and by clinical symptoms suggestive of hepatic disease and/or iron overload in the 67 remaining cases (34%). The Canadian population consisted of 87 men and 26 women with a median age of 50 years (25th percentile, 40 years; 75th percentile, 62 years; range, 16-74 years). Twenty-eight of 113 patients (25%) had severe fibrosis, and 4 were alcoholics (4%).

 
View this table:
[in this window] [in a new window]
  		Table 1. Main Clinical and Biological Data of the Patients

Univariate Analysis

Mode of diagnosis (proband or family screening), age at diagnosis, hepatomegaly, chronic alcoholism, serum AST, ALT, and gamma -glutamyltranspeptidase levels, transferrin saturation, serum ferritin level, HIC, and HIC/age ratio were significantly associated with severe fibrosis (Table 1). There was a higher frequency of patients detected through familial screening when there was no severe fibrosis (43/145, 30%) compared with patients with severe fibrosis (8/52, 15%). There was a higher percentage of men in patients with severe fibrosis (40/52, 77%) than in patients without severe fibrosis (92/145, 63%), which was close to significant P values (0.07). There was a strong link between alcoholism and severe fibrosis. The frequency of alcoholism was 12 of 145 (8%) in patients without severe fibrosis vs. 28 of 52 (54%) in patients with severe fibrosis. Hepatomegaly was also highly suggestive of severe fibrosis. The frequency of hepatomegaly was 12 of 145 (8%) patients without severe fibrosis vs. 38 of 52 (73%) patients with severe fibrosis. The repartition of the different variables was studied to separate patients with and without severe fibrosis (Figure 1). No patient among the 39 individuals under the age of 35 years had severe fibrosis. The frequency of severe fibrosis was very low in patients with transferrin saturation <= 0.65 (1/25, 4%), serum ferritin <= 1500 µg/L (4/122, 3%), or serum ferritin <= 1000 µg/L (1/105, 1%). The frequency of severe fibrosis was high in patients with elevated AST values (21/26, 81%) compared with 31 of 171 patients (18%) with normal AST values (chi 2, P < 10-4). Although HIC and HIC/age were significantly correlated to severe fibrosis, there was a considerable overlap between groups: the frequency of severe fibrosis was only 36 of 56 (64%) in the case of HIC values >400 µmol/g and was 20 of 27 (74%) for HIC values >500 µmol/g.


View larger version (27K):
[in this window]
[in a new window]
  		Figure 1. Representation of the distribution of variables: 50% of values are comprised within the box (between the 25th and 75th percentile), and the horizontal bar gives the median; 80% of values are comprised between the extremities of vertical bars (between the 10th and 90th percentile), and the extreme values are represented as individual points. square , No severe fibrosis; , severe fibrosis.

Multivariate Analysis

Variables significantly linked to severe fibrosis in univariate analysis were entered into multivariate analysis. Using the logistic regression model, three independent variables were used to predict severe fibrosis: AST, serum ferritin, and presence of hepatomegaly (parameters estimation is given on Table 2). The mode of diagnosis was not selected as an independent predictive variable. With a threshold of 0.5 for Y predicted (which represents the probability of having severe fibrosis; patients were predicted "severe fibrosis" if Y predicted >0.50 and "no severe fibrosis" if Y predicted <0.50), 183 of 197 patients were correctly predicted by logistic regression analysis (diagnostic accuracy, 0.93) showing 140 of 145 patients without severe fibrosis and 43 of 52 patients with severe fibrosis (sensitivity, 0.83; specificity, 0.97; positive predictive value, 0.90; negative predictive value, 0.94). Using the discriminant analysis model, four independent variables were used to separate patients with and without severe fibrosis: AST, serum ferritin, hepatomegaly, and alcoholism; 186 of 197 patients were correctly classified by discriminant analysis (diagnostic accuracy, 0.94) showing 140 of 145 patients without severe fibrosis and 46 of 52 patients with severe fibrosis (sensitivity, 0.89; specificity, 0.97; positive predictive value, 0.90; negative predictive value, 0.96). When applying the logistic regression analysis separately for the 146 probands and the 51 relatives diagnosed through family screening, serum ferritin was selected in both groups and seemed to be the key variable for prediction. The other independent variables were: AST, hepatomegaly, and transferrin saturation for the proband group and age for the relative group.

 
View this table:
[in this window] [in a new window]
  		Table 2. Logistic Regression Analysis (Odds Ratio)

Clinical Predictive Model of Severe Fibrosis

Because the mathematical formulation of the two models based on multivariate analysis makes their clinical use difficult, the variables selected by multivariate analysis (serum AST, serum ferritin, and hepatomegaly) were then combined to get a simple clinical predictive model of severe fibrosis. Figure 2 shows the correlation between serum ferritin and AST values. Figure 3 gives the prevalence of severe fibrosis according to the different distributions of the three variables on a decision tree. ROC curves using different AST thresholds between 0.8 and 1.4 ULN were constructed to determine the best threshold for ferritin. Thresholds between 1690 µg/L and 1000 µg/L gave the best compromise in terms of sensitivity and specificity for prediction of severe fibrosis. A threshold of 1000 µg/L was chosen to maximize the negative prediction of severe fibrosis. Figure 3 shows that serum ferritin was the most powerful variable for negative prediction of severe fibrosis. Only 1 of 105 patients (1%) with serum ferritin <= 1000 µmol/L had severe fibrosis vs. 51 of 92 patients (55%) with serum ferritin >1000 µg/L. When patients were excluded with hepatomegaly, ferritin >1000 µg/L, or an elevated AST, none of the 94 remaining patients had severe fibrosis. In contrast, the positive prediction of severe fibrosis was more difficult; when selecting patients with either serum ferritin >1000 µg/L, hepatomegaly, or elevated serum AST, the prevalence of severe fibrosis was only 52 of 103 (50%). When considering patients with serum ferritin >1000 µg, it seemed that the prevalence of severe fibrosis was much higher in case of hepatomegaly (37 of 41 [90%] vs. 14 of 51 [27%]) or elevated serum AST (21 of 24 [87%] vs. 30 of 68 [44%]). For a maximum security in positive prediction of severe fibrosis, when selecting patients with serum ferritin >1000 µg/L, hepatomegaly, and elevated AST values, 17 of 18 patients (94%) had severe fibrosis.


View larger version (25K):
[in this window]
[in a new window]
  		Figure 2. Correlation between serum ferritin and AST values. Spearman test: rho  = 0.628; P < 10-4. The horizontal and vertical bars represent the chosen thresholds for serum ferritin = 1000 µg/L and AST = 1 ULN.


View larger version (18K):
[in this window]
[in a new window]
  		Figure 3. Decision tree using the three variables (serum ferritin, hepatomegaly, and serum AST) selected by multivariate analysis. The term fibrosis refers to severe (stage 3 or 4) fibrosis.

Validation of the Predictive Models

These three different models were applied to the Canadian population. Using the multiple logistic regression model, 95 of 113 patients were correctly classified (diagnostic accuracy, 0.84) showing 71 of 75 patients with a negative prediction of severe fibrosis and 24 of 38 patients with a positive prediction of severe fibrosis (sensitivity, 0.86; specificity, 0.84; positive predictive value, 0.63; negative predictive value, 0.95). Using the discriminant model, 93 of 113 patients were correctly classified (diagnostic accuracy, 0.82) showing 82 of 99 patients with a negative prediction of severe fibrosis and 11 of 14 with a positive prediction of cirrhosis (sensitivity, 0.32; specificity, 0.96; positive predictive value, 0.79; negative predictive value, 0.83). Using the simplified clinical model, 2 of 60 patients (3%) with serum ferritin <= 1000 µg/L had severe fibrosis vs. 26 of 53 patients (49%) with serum ferritin >1000 µg/L. No severe fibrosis was encountered for patients with serum ferritin <= 1000 µg/L, normal AST values, and absence of hepatomegaly; 14 of 18 (78%) patients with serum ferritin >1000 µg/L, elevated AST values, and hepatomegaly had severe fibrosis. Details of results are given in Figure 4.


View larger version (18K):
[in this window]
[in a new window]
  		Figure 4. Validation of the clinical predictive model in the Canadian population. The term fibrosis refers to severe (stage 3 or 4) fibrosis.

Finally, when serum ferritin, serum AST, hepatomegaly, and alcoholism were entered in stepwise multivariate analysis using the same methods as that in the French population, serum ferritin and hepatomegaly, but not AST or alcoholism, were selected in the model.

    Discussion
Top
Abstract
Introduction
Methods
Results
Discussion
References

This study based on multivariate analysis of variables significantly linked to severe fibrosis in C282Y homozygous patients showed that absence of severe fibrosis can be accurately predicted using simple clinical and biochemical variables. The most significant variable for negative prediction was serum ferritin, and the prevalence of severe fibrosis was extremely low when serum ferritin was <= 1000 µg/L (1%-3%). Therefore, on the basis of these results, it is not advisable to perform liver biopsy in this subgroup of patients, which represented about half of the total population and three-fourths of nonfibrotic patients. If patients with hepatomegaly and elevated AST values were excluded, there were no patients with severe fibrosis in both the French and the Canadian population. By contrast, the positive prediction of severe fibrosis was more difficult. Thus, the positive diagnosis of fibrosis still relies on liver biopsy.

The two methods of multivariate analysis gave similar results in French patients. The use of the multiple logistic regression model has the advantage of producing a probability of severe fibrosis, whereas discriminant analysis gives a score that has to be compared with a discriminant threshold score. In the multiple logistic regression model, the probability threshold for prediction (Y predicted) used in the study was 0.50. It is possible to choose different probability thresholds for clinical use to increase the positive (higher Y predicted) or negative (lower Y predicted) predictive value of severe fibrosis. The validation of these different models on a different population, selected in another center, is a very important step. It showed that multivariate models had a good negative predictive but a poor positive predictive value when applied to the Canadian group. Several explanations can be proposed: (1) ferritin measurements are not standardized, and comparison between values obtained using different kits of radioimmunologic measurement is difficult; (2) percutaneous liver biopsy that was selected as the gold standard in the study may give false-negative results, and false-negative rates of 24% are reported in some series of blind liver biopsies23; and (3) difference in fibrosis cofactors such as alcoholism or environmental cofactors may exist between different populations selected in different countries. This fact is highlighted by the results of multivariate analysis in the Canadian population, in which serum AST and alcoholism were not selected when using a method identical to those applied to the French population. This difference is probably linked to a different prevalence of excessive drinkers (22% in the French population vs. 4% in the Canadian) and explains why the results obtained by discriminant analysis (which took into account the variable "alcoholism") were less efficient when applied to the Canadian population. Therefore, due to this absence of reproducibility of positive prediction, we cannot recommend the use of multivariate models for this purpose unless each center establishes its own equation of regression on nonalcoholic patients using a standardized ferritin measurement. By contrast with positive prediction, absence of severe fibrosis was accurately predicted in both groups (negative predictive value, 0.94 in the French population and 0.95 in the Canadian population, using the multiple logistic regression model). Because of the need for mathematical calculation of the regression equation, which will probably hamper its clinical bedside use, it is easier to use the simplified clinical predictive model. The rate of severe fibrosis was very low when serum ferritin was <= 1000 µg/L, and no patient had severe fibrosis when serum ferritin was <= 1000 µg/L in the absence of hepatomegaly or increase in serum AST. This logical formulation was constructed to avoid false-negative prediction of cirrhosis and would have avoided liver biopsy in 70% of nonfibrotic patients. The choice of a higher serum ferritin threshold of 1500 µg/L would have resulted only in a slight decrease in negative prediction and would have avoided liver biopsy in a larger number of patients but, however, can result in misclassification of patients in the absence of standardized ferritin measurement. The use of multiple logistic regression is advisable for negative prediction if possible because, with an excellent negative prediction, it would have avoided liver biopsy in a larger number of nonfibrotic patients (140 of 145 [97%] in the French population and 71 of 85 [84%] in the Canadian population). Because the mode of diagnosis was not selected as an independent variable, it is not useful to establish different models for probands and patients selected through familial screening.

The absence of liver biopsy implies that no information will be available regarding the amount of hepatic iron and the existence of associated hepatic histological abnormalities such as iron-free foci.24 Actually, biochemical or histological assessment of liver iron content is no longer essential in C282Y homozygous patients because the diagnosis no longer relies on the determination of the hepatic iron index but rather on the genetic test. Moreover, the venesection record of the amount of blood removed before reaching iron depletion retrospectively will give the best evaluation of excess iron storage (the removal of 500 mL of whole blood is equivalent to approximately 250 mg of iron). The correlation of HIC with the amount of iron removed is not better than the correlation obtained with serum ferritin25 and does not accurately predict the duration of treatment. Finally, if pretherapeutic quantification of hepatic iron is judged necessary by the physician, noninvasive techniques such as magnetic resonance imaging provide excellent results.26,27 Iron-free foci, defined as clear-cut sublobular clusters of more than 20 hepatocytes devoid of iron in an otherwise iron-loaded liver, are found in the liver of 7.6% of patients with hemochromatosis and are considered as preneoplastic lesions. However, all patients with iron-free foci have heavy iron overload and severe fibrosis.24 Therefore, the detection of iron-free nodules does not modify clinical practice and does not justify the performance of liver biopsy in nonfibrotic patients.

Age and HIC, although significantly correlated with severe fibrosis in univariate analysis, were not selected as independent variables linked to severe fibrosis by regression analysis in the total population. Nevertheless, age was selected in the subgroup of patients diagnosed through family screening, and distribution of the values showed that no severe fibrosis was encountered under the age of 35. This validates the empiric attitude proposed by most physicians7 and includes the majority of patients detected in family screening. It did not provide any more information than the combined use of AST and serum ferritin because all patients younger than 35 years had normal AST values and serum ferritin level <= 1000 µg/L. We found that the majority of patients with HIC >500 µmol/g had severe fibrosis, which is in accordance with previous studies,10,28 but, clearly, a lot of patients with HIC lower than 500 µmol/g had severe fibrosis even in the absence of alcoholism. We were not able to confirm the findings of Basset et al.,10 who separated nonalcoholic patients with and without cirrhosis using an HIC threshold of 400 µmol/g. The absence of severe fibrosis in a significant proportion of patients despite massive hepatic iron overload indicates that iron itself is poorly fibrogenic and that the influence of cofactors is probably important29 and explains the poor performance of positive prediction of severe fibrosis that we encountered.

It is important to point out that our study included C282Y homozygous patients and that the prediction is therefore restricted to these patients. This genotype represents more than 95% of patients having a typical phenotype of hemochromatosis in western France,30 England,31 Australia,18 and Canada32 but could be less common in other countries such as southern France,13 Italy,14 or southern United States.15 The reasons for these differences are not yet clear, and these discrepancies may be related to differences in the phenotypic diagnostic criteria or to genetic heterogeneity of the disease. Moreover, iron storage disorders can be encountered in various non-C282Y-linked iron disorders such as inefficient erythropoiesis,33 end-stage cirrhosis,34-36 iron overload with polymetabolic disorders,37 or chronic viral hepatitis.38 In these different pathological conditions, in patients who are not homozygous for the C282Y mutation of the HFE gene, liver histology is essential for the evaluation of fibrosis and remains the key element of the diagnosis. Another point is that several recent reports,30,39,40 using C282Y genotyping for the purpose of family screening, found a significant proportion (3%-15%) of C282Y homozygous subjects, higher in females than in males,39 who did not express iron overload. Performing liver biopsy in patients who have normal serum iron, transferrin saturation, and serum ferritin values only to insure the absence of a slight liver iron overload seems excessive. The goal of treatment is to normalize transferrin saturation and ferritin,7 and, therefore, phlebotomies are not necessary when these tests are already normal. However, regular follow-up is crucial for those patients because iron overload may develop with time, and if serum iron tests tend to increase, phlebotomies are required to keep the tests within normal range. Finally, the use of these predictive models is proposed to improve rather than replace clinical judgment. There will continue to be individual patients with other risk factors in whom clinical judgment may justify a liver biopsy despite the predictive models.

On the basis of our results, we propose the use of the clinical model to decide whether or not to perform a biopsy on a C282Y homozygous patient. In case of serum ferritin <1000 µg/L, absence of hepatomegaly and normal serum AST level, it is not useful to perform a liver biopsy because there is no risk of significant liver fibrosis.

    Footnotes

   Received December 23, 1997. Accepted June 23, 1998.
   Address requests for reprints to: Dominique Guyader, M.D., Clinique des Maladies du Foie, CHU Pontchaillou. Rue H. Le Guilloux, 35033 Rennes, France. e-mail: Dominique.Guyader@univ-rennes1.fr; fax: (33) 299-28-41-12. 

    Abbreviations

Abbreviations used in this paper: HIC, hepatic iron concentration. ROC, receiver operating characteristics. ULN, upper limit of normal value.

    References
Top
Abstract
Introduction
Methods
Results
Discussion
References

1. Brissot P, Deugnier Y. Genetic hemochromatosis. In: McIntyre N, Benhamou J, Bircher J, Rizzetto M, Rodes J, eds. Oxford textbook of clinical hepatology. Oxford: Oxford University Press, 1998 (in press).

2. Niederau C, Fischer R, Purschel A, Stremmel W, Haussinger D, Strohmeyer G.  Long-term survival in patients with hereditary hemochromatosis.  Gastroenterology  1996;110:1107-1119[Abstract].

3. Niederau C, Fischer R, Sonnenberg A, Stremmel W, Trampisch HJ, Strohmeyer G.  Survival and causes of death in cirrhotic and in noncirrhotic patients with primary hemochromatosis.  N Engl J Med  1985;313:1256-1262[Medline].

4. Bradbear RA, Bain C, Siskind V, Schofield FD, Webb S, Axelsen EM, Halliday J, Bassett ML, Powell LW.  Cohort study of internal malignancy in genetic hemochromatosis and other chronic nonalcoholic liver diseases.  J Natl Cancer Inst  1985;75:81-84[Medline].

5. Deugnier Y, Guyader D, Crantock L, Lopez JM, Turlin B, Yaouanq J, Jouanolle H, Campion JP, Launois B, Halliday JW, Powell LW, Brissot P.  Primary liver cancer in genetic hemochromatosis: a clinical, pathological, and pathogenetic study of 54 cases.  Gastroenterology  1993;104:228-234[Abstract].

6. Adams PC, Speechley M, Kertesz AE.  Long-term survival analysis in hereditary hemochromatosis.  Gastroenterology  1991;101:368-372[Abstract].

7. Bacon BR.  Diagnosis and management of hemochromatosis.  Gastroenterology  1997;113:995-999[Medline].

8. Deugnier YM, Loréal O, Turlin B, Guyader D, Jouanolle H, Moirand R, Jacquelinet C, Brissot P.  Liver pathology in genetic hemochromatosis: a review of 135 homozygous cases and their bioclinical correlations.  Gastroenterology  1992;102:2050-2059[Abstract].

9. Kowdley KV, Trainer TD, Saltzman JR, Pedrosa M, Krawitt EL, Knox TA, Susskind K, Pratt D, Bonkovsky HL, Grace ND, Kaplan MM.  Utility of hepatic iron index in American patients with hereditary hemochromatosis: a multicenter study.  Gastroenterology  1997;113:1270-1277[Abstract].

10. Bassett ML, Halliday JW, Powell LW.  Value of hepatic iron measurements in early hemochromatosis and determination of the critical iron level associated with fibrosis.  Hepatology  1986;6:24-29[Medline].

11. Feder JN, Gnirke A, Thomas W, Tsuchihashi Z, Ruddy DA, Basava A, Dormishian F, Domingo R, Ellis MC, Fullan A, Hinton LM, Jones NL, Kimmel BE, Dronmal IGS, Lauer P, Lee VK, Loeb DB, Mapa FA, McClelland E, Meyer MC, Mintier GA, Moeller N, Moorz T, Morikang E, Prass CE, Quintana L, Starnes SM, Schatzman RC, Brunke KJ, Drayna DT, Risch NJ, Bacon BR, Wolff RK.  A novel MHC class I-like gene is mutated in patients with hereditary haemochromatosis.  Nat Genet  1996;13:399-409[Medline].

12. Bodmer JG, Parham P, Albert E, Marsh S.  Putting a hold on HLA-H.  Nat Genet  1997;15:234-235.

13. Borot N, Roth MP, Malfroy L, Demangel C, Vinel JP, Pascal JP, Coppin H.  Mutations in the MCH class I-like candidate gene for hemochomatosis in French patients.  Immunogenetics  1997;45:320-324[Medline].

14. Carella M, D'ambrosio L, Totaro A, Grifa A, Valentino MA, Piperno A, Girelli D, Roetto A, Franco B, Gasparini P, Camaschella C.  Mutation analysis of the HLA-H gene in Italian hemochromatosis patients.  Am J Hum Genet  1997;60:828-832[Medline].

15. Barton JC, Shih WWH, Sawada-Hirai R, Acton RT, Harmon L, Rivers C, Rothenberg BE.  Genetic and clinical description of hemochromatosis probands and heterozygotes: evidence that multiple genes linked to the major histocompatibility complex are responsible for hemochromatosis.  Blood Cells Mol Dis  1997;23:135-145[Medline].

16. Beutler E, Gelbart T, West C, Lee P, Adams M, Blackstone R, Pockros P, Kosty M, Venditti CP, Phatak PD, Seese NK, Chorney KA, Ten Elshof AE, Gerhard GS, Chorney M.  Mutation analysis in hereditary hemochromatosis.  Blood Cells Mol Dis  1996;22:187-194[Medline].

17. Jouanolle AM, Fergelot P, Gandon G, Yaouanq J, Le Gall JY, David V.  A candidate gene for hemochromatosis: frequency of the C282Y and H63D mutations.  Human Genet  1997;100:544-547[Medline].

18. Jazwinska EC, Cullen LM, Busfield F, Pyper WR, Webb SI, Powell LW, Morris CP, Walsh TP.  Hæmochromatosis and HLA-H.  Nat Genet  1996;14:249-251[Medline].

19. Jouanolle AM, Gandon G, Jezequel P, Blayau M, Campion ML, Yaouanq J, Mosser J, Fergelot P, Chauvel B, Bouric P, Carn G, Andrieux N, Gicquel I, Le Gall JY, David V.  Hæmochromatosis and HLA-H.  Nat Genet  1996;14:251-252[Medline].

20. Garcia-Tsao G, Boyer JL.  Outpatient liver biopsy: how safe is it?  Ann Intern Med  1993;118:150-153[Medline].

21. Piccinino F, Sagnelli E, Pasquale G, Giusti G.  Complications following percutaneous liver biopsy: a multicenter retrospective study on 68, 276 biopsies.  J Hepatol  1986;2:165-173[Medline].

22. Barry M, Sherlock S.  Measurement of liver-iron concentration in needle-biopsy specimens.  Lancet  1971;2:100-103[Medline].

23. Schalm SW.  The diagnosis of cirrhosis: clinical relevance and methodology.  J Hepatol  1997;27:1118-1119[Medline].

24. Deugnier YM, Charalambous P, Le Quilleuc D, Turlin B, Searle J, Brissot P, Powell LW, Halliday JW.  Preneoplastic significance of hepatic iron-free foci in genetic hemochromatosis: a study of 185 patients.  Hepatology  1993;18:1363-1369[Medline].

25. Baldus WP, Fairbanks VF, Moyer TP. Predicting phlebotomy requirements in genetic hemochromatosis (abstr). Proceedings of fifth conference of the International Association for the Study of Disorders of Iron. Metabolism 1995:67.

26. Bonkovsky HL, Slaker DP, Bills EB, Wolf DC.  Usefulness and limitations of laboratory and hepatic imaging studies in iron-storage disease.  Gastroenterology  1990;99:1079-1091[Abstract].

27. Gandon Y, Guyader D, Heautot JF, Reda MI, Yaouanq J, Buhe T, Brissot P, Carsin M, Deugnier Y.  Hemochromatosis: diagnosis and quantification of liver iron with gradient-echo MR imaging.  Radiology  1994;193:533-538[Medline].

28. Loréal O, Deugnier Y, Moirand R, Lauvin L, Guyader D, Jouanolle H, Turlin B, Lescoat G, Brissot P.  Liver fibrosis in genetic hemochromatosis: respective roles of iron and noniron related factors in 127 homozygous patients.  J Hepatol  1992;16:122-127[Medline].

29. Pietrangelo A.  Iron, oxidative stress and liver fibrosis.  J Hepatol  1998;28:8-13.

30. Brissot P, Moirand R, Jouanolle AM, Guyader D, Le Gall JY, Deugnier Y, David V.  A genotypic study of 207 unrelated probands diagnosed as "genetic hemochromatosis" on "classical" phenotypic criteria (abstr).  Hepatology  1997;26:370A.

31. Merryweather-Clarke AT, Pointon JJ, Shearman JD, Robson KJH.  Global prevalence of putative haemochromatosis mutations.  J Med Genet  1997;34:275-278[Medline].

32. Adams PC, Chakrabarti S.  Genotypic/phenotypic correlations in genetic hemochromatosis: evolution of diagnostic criteria.  Gastroenterology  1998;114:319-323[Abstract/Full Text].

33. Tavill AS, Sharma BK, Bacon BR.  Iron and the liver: genetic hemochromatosis and other hepatic overload disorders.  Prog Liver Dis  1990;9:281-305[Medline].

34. Ludwig J, Hashimoto E, Porayko MK, Moyer TP, Baldus WP.  Hemosiderosis in cirrhosis: a study of 447 native livers.  Gastroenterology  1997;112:882-888[Abstract].

35. Deugnier Y, Turlin B, Le Quilleuc D, Moirand R, Loréal O, Messner M, Meunier B, Brissot P, Launois B.  A reappraisal of hepatic siderosis in patients with end-stage cirrhosis: practical implications for the diagnosis of hemochromatosis.  Am J Surg Pathol  1997;21:669-675[Medline].

36. Villeneuve JP, Bilodeau M, Lepage R, Côté J, Lefebvre M.  Variability in hepatic iron concentration measurements from needle-biopsy specimens.  J Hepatol  1996;25:172-177[Medline].

37. Moirand R, Abdel Majid M, Loréal O, Paillard F, Brissot P, Deugnier Y.  A new syndrome of liver iron overload with normal transferrin saturation.  Lancet  1997;349:95-97[Medline].

38. Bonkovsky HL, Banner BF, Rothman AL.  Iron and chronic viral hepatitis.  Hepatology  1997;25:759-768[Medline].

39. Crawford DHG, Jazwinska EC, Cullen LM, Powell LW.  Expression of HLA-linked haemochromatosis in homozygous and heterozygous subjects diagnosed according to the C282Y mutation: evaluation of previous diagnostic criteria (abstr).  Hepatology  1997;26:198[Abstract]A.

40. Adams PC, Chakrabarti S.  Genotypic/phenotypic correlations in genetic hemochromatosis: evolution of diagnostic criteria.  Gastroenterology  1998;114:319-323[Abstract/Full Text].

GASTROENTEROLOGY 1998;115:929-936
© 1998 by The American Gastroenterological Association
0016-5085/98/$3.00
Reprint (PDF) Version of this Article
PubMed Citation
Related Articles in PubMed
Download to Citation Manager
Search Medline for articles by:
GUYADER, D. || DEUGNIER, Y.

This article has been cited by other articles:


Home Help [Feedback] [For Subscribers] [Archive] [Search] [Search Result] || Sign In as Member || Contact Subscription Administrator at this Site || FAQ