Relationship Between Kaposi Sarcoma–Associated Herpesvirus and HIV

To the Editor: Dr Osmond and colleagues¹ found that the prevalence of Kaposi sarcoma–associated herpesvirus (KSHV) in 1978 and 1979 was 26.5% among homosexual men who later enrolled in the San Francisco City Clinic Cohort (SFCCC) study. This finding is important for understanding the subsequent epidemic of Kaposi sarcoma and its relationship with human immunodeficiency virus (HIV) in San Francisco. It was inappropriate, however, for Osmond et al to infer that the incidence of KSHV has not changed from 1978 through 1996 based on cross-sectional prevalence data from 3 heterogeneous studies, and to use these data in an ecological analysis of how behavior change might affect KSHV transmission.

Determining KSHV incidence from prevalence data is problematic² and requires assumptions that are not met in this analysis. One is that there was no net migration of KSHV-uninfected homosexual men into San Francisco during this period and no excess deaths among KSHV-infected men already in San Francisco, either of which would decrease KSHV prevalence. A second assumption is that the 3 cohorts represent unbiased samples of the same population. The SFCCC enrolled subjects in a public sexually transmitted diseases clinic from 1978 through 1980, while the San Francisco Men's Health Study (1984-1985) and the San Francisco Young Men's Health Study (1995-1996) had population-based designs. Subjects in the SFCCC likely had higher rates of sexually transmitted infections, including KSHV, than would have been found in a contemporaneous population-based study. The SFCCC had an even higher proportion of KSHV-infected subjects because it retrospectively included 97% of the 699 deceased subjects, but less than half of the remaining 6006 subjects.¹ Deceased subjects were much more frequently infected with HIV-1 at entry into SFCCC,³ and KSHV and HIV infections are strongly linked in this cohort.¹ Finally, there may be relevant temporal differences among SFCCC enrollees: subjects in the first phase of the hepatitis B virus (HBV) study on which SFCCC is based had a higher prevalence of HBV seromarkers than subsequent enrollees,³ which suggests that early enrollees had more sex partners. All of these biases could have obscured an increasing prevalence of KSHV among homosexual men in San Francisco during 1978 through 1985. Even so, KSHV prevalence in SFCCC actually increased by 34% between the January through June 1978 and the September 1979 through December 1980 periods.¹

Closed cohort studies are not limited by the above problems. Such studies demonstrate that KSHV incidence peaked during the early 1980s among homosexual men in New York, Copenhagen, and Washington, DC.^{4 - 5} Therefore, the conclusion that KSHV incidence was not changing during the onset of the HIV epidemic among homosexual men in San Francisco (and, by inference, in the rest of the United States) is doubtful.