We agree with Drs. Knobler and Schattner that histologic examination of the liver in HCV-infected persons with and without type 2 diabetes mellitus contributes to our understanding of the relationship between these conditions. Previous studies have demonstrated that community-identified HCV infection is not usually associated with cirrhosis (1). Thus, we would not anticipate finding a high prevalence of cirrhosis among HCV-infected persons identified in a community setting, such as the participants in NHANES III. While this could not be confirmed in our analysis because liver biopsies were not performed in NHANES, the assumption is supported by a study by Knobler and coworkers, which demonstrated an association between HCV infection and type 2 diabetes mellitus in persons in whom cirrhosis was excluded by biopsy [2]). The biological mechanism of this association remains to be explained.

Several investigators, including Dr. Paris, have suggested iron overload as a possible explanation for the increased prevalence of type 2 diabetes mellitus among persons with HCV infection. Labropoulou-Karatza and colleagues (3) specifically examined serum ferritin levels among adults with ß-thalassemia and found that they had no effect on the relationship between HCV infection and diabetes. Similarly, in our study, serum ferritin levels did not affect the results of the multivariable analysis. Our results were comparable to those of Dr. Paris in that serum ferritin levels were elevated among persons with HCV infection and among those with type 2 diabetes mellitus. However, the inclusion of serum ferritin levels in the final multivariate model attenuated the odds ratio only slightly, from 3.77 (CI, 1.80 to 7.87) to 3.28 (CI, 1.80 to 5.99). Thus, to the extent that serum ferritin, an acute-phase reactant that may be affected by inflammation, reflects hepatic iron stores, iron overload does not seem to fully explain the increased occurrence of type 2 diabetes mellitus among HCV-infected persons in NHANES III.