Correspondence

Disease-Related Conditions in Relatives of Patients with Hemochromatosis

N Engl J Med 2001; 344:1477-1478May 10, 2001

Article

To the Editor:

A decision on whether to recommend population screening for hereditary hemochromatosis requires information on disease penetrance — that is, the fraction of persons with a hemochromatosis genotype who will develop clinical symptoms of the disease if they are not treated. Bulaj et al. (Nov. 23 issue)1 derive a “minimal estimate” of the penetrance of hereditary hemochromatosis by studying the frequency of symptoms among relatives of patients with hemochromatosis identified on the basis of elevated transferrin-saturation values, rather than clinical symptoms. They calculate the minimal estimate of the incidence of disease-related conditions among homozygous relatives of these patients to be 29 percent for men over 40 years of age and 11 percent for women over 50 years of age.

Bulaj et al. report neither confidence intervals nor the numbers of observations on which these estimates are based. By deriving the numbers from their paper, we found that both estimates appear to be based on small samples, with 5 of 17 men and 1 of 9 women among the relatives exhibiting symptoms of hemochromatosis. The corresponding confidence intervals for these estimates of penetrance are 12 to 54 percent for men and 1 to 42 percent for women.2

The reported estimates of the penetrance of hereditary hemochromatosis are incomplete because they are cross-sectional. If larger numbers of observations were unavailable, life-table or survival analysis could be used to calculate the lifetime incidence.3,4 Reliable estimates of disease penetrance, which are needed for the evaluation of hemochromatosis screening, await the results of larger, collaborative studies with longitudinal follow-up.

Scott D. Grosse, Ph.D.
Jill M. Morris, Ph.D.
Muin J. Khoury, M.D., Ph.D.
Centers for Disease Control and Prevention, Atlanta, GA 30341

4 References

1. 1

   Bulaj ZJ, Ajioka RS, Phillips JD, et al. Disease-related conditions in relatives of patients with hemochromatosis. N Engl J Med 2000;343:1529-1535
   Full Text | Web of Science | Medline

2. 2

   Blyth CR, Still HA. Binomial confidence intervals. J Am Stat Assoc 1983;78:108-116
   CrossRef | Web of Science

3. 3

   Chase GA, Folstein MF, Breigtner JC, Beaty TH, Self SG. The use of life tables and survival analysis in testing genetic hypotheses, with an application to Alzheimer's disease. Am J Epidemiol 1983;117:590-597
   Web of Science | Medline

4. 4

   Struewing JP, Hartge P, Wacholder S, et al. The risk of cancer associated with specific mutations of BRCA1 and BRCA2 among Ashkenazi Jews. N Engl J Med 1997;336:1401-1408
   Full Text | Web of Science | Medline

Author/Editor Response

Dr. Kushner replies:

To the Editor: In our study, the incidence of disease-related conditions in relatives of probands with hemochromatosis was directly measured in 113 men and 101 women (Table 2 of the article). Fifty-two men were over 40 years of age, and 27 of these 52 (52 percent; 95 percent confidence interval, 38 to 66 percent) were clinically affected. Forty-three of the women were over 50, and 7 of these 43 (16 percent; 95 percent confidence interval, 7 to 71 percent) were clinically affected. Our data suggested that relatives of clinically affected probands were more likely to have disease-related conditions than relatives of probands identified through screening programs. We based our minimal estimates of penetrance on the values in the latter group to avoid any ascertainment bias related to familial factors predisposing to penetrance. There were 24 men over 40 years of age in the latter group, and 7 of these 24 (29 percent; 95 percent confidence interval, 13 to 51 percent) were clinically affected. There were only 9 women over 50 in the latter group, and 1 of these 9 women (11 percent; 95 percent confidence interval, 0.3 to 48 percent) was clinically affected. The true disease penetrance probably lies between the values obtained for the entire group and the values obtained for relatives of probands identified through screening programs.

We agree that larger, longitudinal studies would yield the most accurate estimates of penetrance, but this would require following genotyped homozygotes without providing phlebotomy therapy, until they developed a disease-related illness or died of other causes. Such studies are unlikely to be undertaken.

James P. Kushner, M.D.
University of Utah School of Medicine, Salt Lake City, UT 84132