Response: Screening for treatable disease is better than just screening for hemochromatosis

We proposed that the most efficient way of detecting the uncommon individuals homozygous for the C282Y mutation at risk for clinical disease was to determine the ferritin concentration of the serum.¹  The basis of our approach is the finding, by 3 independent groups, that cirrhosis of the liver, the most common morbid consequence of hemochromatosis, occurs almost exclusively in patients with serum ferritin levels greater than 1000 μg/L.^2-5  Åsberg et al state that in their large Norwegian study 4 homozygous subjects with serum ferritin levels below 1000 μg/L, ranging from 311 to 629 μg/L, had liver fibrosis and/or cirrhosis.⁶  It seems unlikely that the epidemiology of hemochromatosis in Norwegians differs from that of all other Europeans studied. The occurrence of cirrhosis in these patients is in contrast to the consistent findings among the other studies cited, in which only 1 of a total of 309 homozygotes with serum ferritin less than 1000 μg/L had cirrhosis.^2-5  Obviously, homozygotes for the C282Y mutation can have cirrhosis for a variety of reasons other than iron overload, and such factors probably played a role in some of the Norwegian patients.

If, then, we accept the fact that cirrhosis is rare in patients homozygous for the C282Y mutation who do not have greatly elevated serum ferritin levels, the remainder of the arguments presented by Åsberg et al are difficult for us to understand.

First, they point out that elevated transferrin saturation would identify 100% of male homozygotes in their population. Although this percentage is higher than that reported in all other large studies, ranging from 60% to 86%,⁷  it merely lends further support to our argument against use of transferrin saturation as a screen, that is, that it would detect largely homozygotes who would never develop clinical disease.

Åsberg et al also assert that the cost of measuring the transferrin saturation is less than that for serum ferritin, and suggest transferrin saturation measurement be done first. However, the cost of the 2 tests in the United States are roughly the same, with some laboratories even charging less for serum ferritin. Furthermore, transferrin saturation testing requires fasting and 2-stage testing; 3% to 4% of patients (1157/29 699 [3.9%] in our study) with elevated transferrin saturations would be recalled for subsequent serum ferritin determination. Because the goal of screening is to detect C282Y homozygotes at risk for cirrhosis, prescreening with transferrin saturation would only increase the cost of screening by introducing an additional 2 steps, with the only accomplishment being that patients with elevated serum ferritin levels for reasons other than HFE mutations would escape detection. Yet, we have clearly shown that the number of such individuals is relatively small and their detection potentially very valuable by identifying treatable diseases other than hemochromatosis.

Åsberg et al imply that screening for ferritin may increase, not decrease, the amount of anxiety produced in patients. The aim of the physician should not be to reduce all anxiety, only unnecessary anxiety, as is produced in the majority of patients with genetic hemochromatosis who do not have any clinical phenotype. Anxiety produced in the patient without hemochromatosis who has an elevated serum ferritin level is well placed, and may be rewarded, upon proper investigation, by an improved outcome.

As Åsberg et al point out, measuring serum ferritin would be the right thing to do if the primary goal is to find patients with treatable disease. That is our goal. To us it seems clearly a medically better goal than merely screening for hemochromatosis.