Because Bacon and colleagues (1) explored the utility of genetic testing, it is by definition circular to use the genetic test to discredit previously accepted diagnostic criteria for hemochromatosis in the absence of an independent gold standard. They found that some patients with liver disease who did not meet previously established criteria for hemochromatosis were homozygous for C282Y. The authors assumed that genotype should take precedence over the prior criteria, even though the predictive value of the various HFE genotypes has not yet been established. Indeed, given that C282Y homozygosity has been identified in patients with limited iron overload and no clinical manifestations (2), C282Y homozygosity in itself cannot be considered to be diagnostic. Furthermore, iron overload from hemochromatosis is known to exist in the absence of identifiable HFE mutations (3).

Nonetheless, we believe Bacon and colleagues may well be justified in arguing that the criteria for diagnosing hemochromatosis should be updated (for example, requiring a lower level of hepatic iron index or eliminating hepatic iron index as a criterion). This decision should be based on biological grounds, on better ways of quantifying iron overload, or on efficacy of treatment for patients who meet new criteria.

In reply to Dr. Powell, we agree that the use of tests for diagnosis of persons with illness differs greatly from the use of tests for screening healthy individuals. However, because genotype is not a perfect predictor of hemochromatosis, or its absence, we question use of genotype even in the routine evaluation of persons with iron overload. For example, in the circumstance of a person who otherwise meets criteria for hemochromatosis, a negative genetic test result is likely to prove more confusing than illuminating. Similarly, for patients with other identifiable causes of iron overload, genotyping is unlikely to be helpful unless the presentation is unusual.

Given the incomplete understanding of the genetic underpinnings of hemochromatosis, we believe it premature to use genotyping in routine clinical situations. Gathering information about the relation between phenotype and genotype, as Bacon and colleagues have done, is laudable and is essential to advancing knowledge so that genetic testing may someday become fully integrated into clinical practice.