Correspondence

Testing for Antibodies in Kaposi's Sarcoma

N Engl J Med 1996; 335:819-820September 12, 1996

Article

To the Editor:

The report by Miller et al. (May 16 issue)1 on the development of a serologic test for Kaposi's sarcoma–associated herpeslike virus (KSHV) is an important initial step in determining the potential causative role of this new agent in Kaposi's sarcoma. Two aspects of their epidemiologic analysis, however, are questionable. First, the results in the convenience sample of 48 patients with human immunodeficiency virus (HIV) infection and Kaposi's sarcoma and 54 HIV-infected patients without Kaposi's sarcoma are stated in terms of predictive values. Presumably, the authors are considering the use of the serologic marker as a diagnostic test. In such an analysis, the positive and negative predictive values are highly dependent on the prevalence or pretest probability of disease (Kaposi's sarcoma) in the sample2 and are irrelevant when the subjects are chosen by the investigators on the basis of disease status. In this study, the prevalence was arbitrarily chosen to be 47 percent. If the prevalence had been, for example, 20 percent, as reported in the San Francisco Men's Health Study,3 the positive predictive value would have been only 56 percent, and the negative predictive value would have been 91 percent. If the pretest probability of disease had been 90 percent (as might be the case when testing patients with cutaneous lesions that are clinically suggestive of Kaposi's sarcoma), the positive predictive value would have been 98 percent, but the negative predictive value would have been only 23 percent. Furthermore, in a study of the serologic marker as a diagnostic test, the best control group would have been HIV-infected patients with cutaneous lesions confirmed by biopsy not to be Kaposi's sarcoma but otherwise difficult to distinguish from Kaposi's sarcoma on clinical grounds, instead of a convenience sample of patients without Kaposi's sarcoma.4

Second, the authors present unadjusted odds ratios for the association between a positive serologic response (presumably to KSHV) and clinical Kaposi's sarcoma as evidence of an etiologic role of KSHV. No data are provided on the numbers of sexual partners in the group of patients with Kaposi's sarcoma and the group without Kaposi's sarcoma. Failure to adjust for the number of sexual partners leaves open the alternative explanation that KSHV may be associated with sexual transmission and is simply a marker for the as-yet-undetected etiologic agent. Carefully controlled longitudinal studies may be able to resolve this issue.

Jeffrey N. Martin, M.D., M.P.H.
Julie L. Gerberding, M.D., M.P.H.
Dennis H. Osmond, Ph.D.
University of California, San Francisco, San Francisco, CA 94105

4 References

1. 1

   Miller G, Rigsby MO, Heston L, et al. Antibodies to butyrate-induci-ble antigens of Kaposi's sarcoma-associated herpesvirus in patients with HIV-1 infection. N Engl J Med 1996;334:1292-1297
   Full Text | Web of Science | Medline

2. 2

   Sackett DL, Haynes RB, Guyatt GH, Tugwell P. Clinical epidemiology: a basic science for clinical medicine. 2nd ed. Boston: Little, Brown, 1991.
   

3. 3

   Osmond D, Charlebois E, Lang W, Shiboski S, Moss A. Changes in AIDS survival time in two San Francisco cohorts of homosexual men, 1983 to 1993. JAMA 1994;271:1083-1087
   CrossRef | Web of Science | Medline

4. 4

   Jaeschke R, Guyatt G, Sackett DL. Users' guides to the medical literature. III. How to use an article about a diagnostic test. A. Are the results of the study valid? JAMA 1994;271:389-391
   CrossRef | Web of Science | Medline

Author/Editor Response

The authors reply:

To the Editor: Martin et al. correctly point out that the predictive value of a diagnostic test is dependent on the prevalence of the disease in a population. In our article we described the development of a serologic test for antibodies to KSHV; longitudinal, population-based surveys will be needed to establish its predictive value. We also agree that additional serologic studies in HIV-infected and HIV-uninfected populations will be needed to determine whether KSHV is the etiologic agent of Kaposi's sarcoma. Moreover, these serologic studies should be accompanied by information about sexual behavior and other risk factors in order to evaluate the importance of sexual transmission of KSHV. What our article provides is a description of first-generation serologic tools for addressing these important questions.

George Miller, M.D.
Michael Rigsby, M.D.
Yale University School of Medicine, New Haven, CT 06520-8064

Citing Articles (1)

Citing Articles

1. 1

   Martin, Jeffrey N., Ganem, Donald E., Osmond, Dennis H., Page-Shafer, Kimberly A., Macrae, Don, Kedes, Dean H., . (1998) Sexual Transmission and the Natural History of Human Herpesvirus 8 Infection. New England Journal of Medicine 338:14, 948-954
   Full Text