Risk profiling for cirrhosis and hepatocellular carcinoma in HFE hemochromatosis using mobilizable iron stores and alcohol consumption

Abstract

HFE hemochromatosis (HH) may cause cirrhosis and hepatocellular carcinoma (HCC). Progression to these endpoints depends on the severity of iron overload and cofactors, such as alcohol. We evaluated alcohol and iron-related risk factors in relation to cirrhosis at diagnosis and future development of HCC in a retrospective analysis of 197 HH subjects. The proportion of subjects either with cirrhosis or who developed HCC during follow-up were 29/197 (14.7%) or 10/197 (5.1%), respectively. The median (IQR) follow-up time after diagnosis was 15.2 (4.6 to 22.1) years. The median mobilizable iron stores and daily alcohol consumption (IQR) were 6.0 (3.8-11.0) g and 20 (0-40) g, respectively. An optimal logistic regression model for the odds of cirrhosis was developed by adding candidate liver insult variables (mobilizable iron, alcohol consumption, and age as a surrogate for duration of exposure) in a forward stepwise strategy using area under the receiver operating characteristic curve (AUROC) analysis and the corrected Akaike information criterion. This model demonstrated an AUROC (95% CI) of 0.966 (0.935-0.996), with sensitivity 76 (58-88)% and specificity 97 (93-99) % for prediction of cirrhosis and had a negative predictive value of 99.4 (95% CI 96.7-99.97) % for development of HCC. Thus, future risk of HCC can be assessed from mobilizable iron stores and alcohol consumption of HH subjects.

Keywords: HFE hemochromatosis (HH); Hepatic iron concentration (HIC); Hepatic iron index (HII); Hepatocellular carcinoma (HCC); Mobilizable iron (MobFe).

Fig. 1

Flow chart of subjects included in the study. 204 subjects diagnosed with HH between 1974 and 2010, of those 197 had completed clinical information and they were followed up for a median of 15.2 years. 29 patients had liver biopsy proven cirrhosis, of those 10 developed HCC during the follow-up period. 168 did not have liver biopsy proven cirrhosis, and none of these patients developed HCC during the follow-up period.

Fig. 2

Scatterplot of alcohol consumed per day against mobilized iron. The blue line indicates the loci of points where the predicted odds ratio for cirrhosis is 1.0. Note that two of the HCC datapoints overlap. All patients with HCC had cirrhosis.

Fig. 3

Probabilities of cirrhosis estimated by Model 3 for patients diagnosed without (N = 168) and with (N = 29) cirrhosis. The solid red lines represent the median predicted probabilities. Red datapoints indicate patients diagnosed with HCC.