|Home | About | Journals | Submit | Contact Us | Français|
There is conflicting evidence about the contribution of heterosexual transmission to the spread of human herpesvirus 8 (HHV-8) in southern Africa. This study evaluated the hypothesis that HHV-8 infection is associated with risk factors for human immunodeficiency virus type 1 (HIV-1) and other sexually transmitted infections among Zimbabwean men.
HHV-8 seroprevalence was determined for 2750 participants in the Zimbabwe AIDS Prevention Project cohort of male factory workers in Harare, Zimbabwe. Potential associations of HHV-8 antibody detection with risk factors for HIV-1 infection were examined by univariate analysis. Variables with P < .1 in the univariate analysis were included in a multivariate logistic regression model. HHV-8 seroprevalence was also determined among 297 heterosexual couples.
Prevalence of HHV-8, HIV-1, and HHV-8 and HIV-1 coinfection was 28.5% (95% confidence interval [CI], 26.8%–30.2%), 19.5% (95% CI, 18.0%–20.9%), and 6.5% (95% CI, 5.6%–7.5%), respectively. Detection of HHV-8 antibodies was independently associated with older age and HIV-1 infection but not with number of recent sex partners, marital status, education, condom use, prior sexually transmitted infections, payment for sex, chronic hepatitis B infection, or incident HIV-1 infection. HHV-8 seroprevalence was 31.7% (95% CI, 26.3–37.0) among wives in the couples tested, but HHV-8 infection of wives was not associated with HHV-8 infection of husbands (odds ratio, 1.08; 95% CI, 0.62–1.88; P = .8).
HHV-8 and HIV-1 infection did not have common sexual risk factors among urban Zimbabwean men. Sexual transmission does not explain the high prevalence of HHV-8 in this population.
The risk of human herpesvirus 8 (HHV-8) infection in American men is directly related to the number of reported sexual contacts with other men . In the United States, HHV-8 infection is associated with sexually transmitted infection (STI) risk factors in men regardless of reported sexual orientation, but it is not associated with STI risk factors in women . Because underreporting of sex between men could not be ruled out, the latter study could not conclusively estimate the contribution of heterosexual transmission of HHV-8 among American men.
Investigations of HHV-8 epidemiology in Africa have yielded conflicting results about sexual transmission of HHV-8 in African populations. Although some studies have found associations between HHV-8 antibody detection and STIs or STI risk factors [3–6], other studies did not identify associations between HHV-8 infection and sexual behaviors [7, 8]. In a recent study of a South African population with a high prevalence of human immunodeficiency virus type 1 (HIV-1) infection and other STIs, HHV-8 infection was not associated with STIs or measures of sexual behavior .
Between 1987 and 1995, the incidence of Kaposi sarcoma increased more than 40-fold among Zimbabwean men . The high incidence of Kaposi sarcoma among Zimbabwean men in parallel with the emergence of the HIV-1 epidemic suggests that HIV-1 and HHV-8 coinfection is prevalent in this population. However, the characteristics of Zimbabweans who are infected with HHV-8 and the factors that place Zimbabweans at risk for HHV-8 infection are unknown. The Zimbabwe AIDS Prevention Project (ZAPP) established a cohort of male Zimbabwean factory workers that was assessed at regular intervals for sexual risk factors and was tested for HIV-1 infection and other STIs. We hypothesized that, if sexual transmission was important for the acquisition of HHV-8 infection in Zimbabwean men, HHV-8 infection would be correlated with high-risk sexual behaviors in the ZAPP cohort.
ZAPP was a longitudinal cohort study with participants recruited at 40 factories in Harare, Zimbabwe . From March 1993 through June 1997, 3383 men were enrolled. The wives of 373 participants were also enrolled. Informed consent was obtained from all participants, and the human experimentation guidelines of the US Department of Health and Human Services, the Medical Research Council of Zimbabwe, the Colorado Multiple Institutional Review Board, and the Stanford University Medical Center Institutional Review Board were followed. At study entry, participants were questioned about STI risk-related behaviors and were tested for HIV-1 antibodies and hepatitis B antigen. Participants were followed up at 6-month intervals through June 1997. In previous reports from the ZAPP cohort, HIV-1 prevalence at entry was 19.1%, and the incidence of HIV-1 infection and syndromic STIs during cohort follow-up were 2.93 cases per 100 person-years (95% confidence interval [CI], 2.18–3.86 cases per 100 person-years) and 10.19 cases per 100 person-years (95% CI, 8.96–11.6 cases per 100 person-years), respectively [12, 13].
Antibodies to HHV-8 latency-associated nuclear antigen were detected in a 1:10 dilution of serum (collected at study entry) by use of a latent BCP-1 cell immunofluorescent assay, as described elsewhere . Only wells that contained cells with punctate nuclear fluorescence were scored as a positive result for HHV-8 antibody. To ensure quality control, all assays included wells with diluted plasma from a person with Kaposi sarcoma and from a person with no clinical, serological, or polymerase chain reaction evidence of HHV-8 infection as positive and negative controls, respectively. All latency-associated nuclear antigen antibody assays were conducted and interpreted by a single investigator, who was blinded to the identity of the specimen and other demographic and serological data.
Statistical analyses were conducted using SAS, version 9.1 (SAS Institute). Potential associations of HHV-8 antibody with variables previously shown to be risk factors for HIV-1 infection in the ZAPP cohort were examined by univariate analysis. These variables were marital status, age, education level, condom use, payment for sex, number of reported sex partners, history of STI, and hepatitis B surface antigen detection. The results of HIV-1 antibody tests performed at study entry and periodically throughout the ZAPP follow-up period were included in analyses of factors associated with HHV-8 infection. Variables with P < .1 in the univariate analysis were included in a multivariate logistic regression model with use of the forward step-wise process.
Serum samples for HHV-8 testing at ZAPP entry were available for 3043 (90%) of 3383 men participating in the ZAPP cohort. One or more baseline data items were not available for an additional 293 participants; these participants were not included in further analyses. Thus, the final study group consisted of 2750 participants (81% of the ZAPP cohort). HHV-8 seroprevalence was 28.7% (95% CI, 27.1%–30.3%). All participants were African men. The median age was 28 years (range, 17–75 years). Most participants (64%) were married, and most (80%) had completed at least 8 years of school. The median income of participants was ~$40 per month. Approximately one-half of the study group had >1 lifetime sex partner, 43% did not use condoms, 24% admitted to having provided payment for sex, 18% had had a prior STI, and 1.4% had a reactive syphilis serologic test result.
Baseline HIV-1 prevalence in the study group was 19.5% (95% CI, 18.0–20.9; 535 of 2750 participants were infected). An additional 113 participants (4%) acquired HIV-1 infection during ZAPP follow-up. In univariate analyses, HIV-1 infection at study entry was significantly associated with marital status, age, education, condom use, payment for sex, number of sex partners, history of an STI, a reactive VDRL test, and detectable hepatitis B surface antigen (table 1; P ≤.03 for each variable). Incident HIV-1 infection was associated with marital status (P =.008), age (P =.002), income (P =.01), condom use (P =.04), number of lifetime sex partners (P < .001), history of an STI (P =.001), and detectable hepatitis B surface antigen (P =.009). In the multivariate model, men who were widowed (adjusted odds ratio [OR], 5.09; 95% CI, 1.73–15.0) or divorced (adjusted OR, 2.48; 95% CI, 1.32–4.67) were associated with the greatest prevalence of HIV-1 infection (figure 1A). Age, education, payment for sex, prior STIs, and chronic hepatitis B infection were also independently associated with increased prevalence of HIV-1 infection in the multivariate model.
HHV-8 seroprevalence was 28.5% (95% CI, 26.8–30.2; 784 of 2750 participants seropositive) in the study group at ZAPP entry. In univariate analyses, marital status, age, income, payment for sex, and HIV-1 infection at study entry were associated with an increased prevalence of HHV-8 infection (P ≤.1). Only patient age of ≥40 years (adjusted OR, 1.54; 95% CI, 1.14–2.07) and HIV-1 infection at study entry (adjusted OR, 1.28; 95% CI, 1.04–1.59) were independently associated with baseline HHV-8 antibody detection in the multivariate model (figure 1B). STI risk factors, including number of sex partners in the last year, a history of prior STIs, chronic hepatitis B infection, or a reactive syphilis serology (VDRL test) at ZAPP entry, and HIV-1 seroconversion during ZAPP follow-up were not associated with HHV-8 infection (table 1).
The prevalence of HIV-1 infection and HHV-8 infection, condom usage, syphilis serological status, and marital status did not differ between the final study group and the 633 participants that were excluded because of missing data (P > .3, for each comparison), but participants with missing data were younger, were less educated, had lower income, had more sex partners, had more self-reported STIs, were more likely to have paid for sex, and were more likely to have detectable hepatitis B surface antigen, compared with the study group (P < .02, for each comparison). When the participants with missing data were added to the study group data set, there was no change in the outcome of the univariate analyses of factors associated with HIV-1 or HHV-8 infection, except that a significant association between HHV-8 and hepatitis B surface antigen detection was no longer observed (P = .2).
Age was independently associated with both HIV-1 and HHV-8 infection, but the qualitative relationships between age and HIV-1 and HHV-8 prevalence differed (figure 2). HIV-1 prevalence was 5.1% (95% CI, 2.7%–7.6%) among participants aged 17–21 years, peaked at 30.6% (95% CI, 25.3%–36.9%) among participants aged 29–32 years, and was 14.8% (95% CI, 10.5%–19.2%) among participants aged ≥47 years. In contrast, HHV-8 prevalence was 23.3% among participants aged 17–21 years and gradually increased with age, with a maximum prevalence of 35.7% (95% CI, 29.9%–41.6%) among participants aged ≥47 years. The overall prevalence of HIV-1 and HHV-8 coinfection was 6.5% (95% CI, 5.6%–7.5%). The prevalence of coinfection was 1.28% (95% CI, 0.03%–2.53%) among participants aged 17–21 years; prevalence was highest among participants aged 26–28 years (9.62%; 95% CI, 6.57%–12.7) and was similar to this level in the older age groups.
Baseline serum samples from the wives of 297 ZAPP participants were available. Among the couples, 59 ZAPP participants (20%) and 69 spouses (23%) were HIV-1 seropositive. Overall, 247 couples (83%) were concordant for HIV-1 infection. The prevalence of HIV-1 infection among wives was 66% (95% CI, 53.7%–78.6%) if the husband was HIV-1 infected but was only 12.6% (95% CI, 8.4%–16.9%) if the husband was HIV-1 negative (figure 3). If a man was HIV-1 infected, the OR that his wife had HIV-1 infection was 13.5 (95% CI, 7.0–26.2; P < .001). Seventy-nine ZAPP participants (27%) and 95 spouses (32%) were HHV-8 seropositive. One hundred seventy-six couples (59%) were concordant for HHV-8 infection. The prevalence of HHV-8 among wives was 32.9% (95% CI, 22.3%–43.5%) if the husband was HHV-8 infected and was 31.2% (95% CI, 25.0%–37.4%) if the husband was HHV-8 negative. If a man was HHV-8 infected, the OR that his spouse was HHV-8 infected was 1.08 (95% CI, 0.62–1.88; P = .8).
This analysis of HHV-8 seroprevalence in the ZAPP cohort is, to our knowledge, the first systematic evaluation of HHV-8 epidemiology in Zimbabwe, where AIDS-associated Kaposi sarcoma is a major cause of morbidity and mortality among HIV-1–infected persons. Although HHV-8 infection was common among men in urban Zimbabwe, the demographic characteristics of the ZAPP study group are different from those of the Zimbabwean population at large; therefore, any extrapolation of our HHV-8 infection prevalence estimate to the general population should be performed cautiously. All participants in the study group were men, but the prevalence of HHV-8 infection was similar in a smaller group of women (wives of factory workers in the ZAPP cohort); thus, we found no evidence that HHV-8 prevalence differs among Zimbabwean women. Overall, prevalence of HHV-8 infection among ZAPP participants is similar to the prevalence in neighboring South Africa [15, 16], and our findings provide further evidence of the wide dissemination of this virus in southern Africa.
Available assays to detect HHV-8 antibodies make use of use latent or lytic viral proteins expressed in infected cells, whole virus particles, and recombinant proteins. Our study used the immunofluorescent assay format to detect antibodies to latency-associated nuclear antigen, because it has been used widely in studies of HHV-8 epidemiology, is inexpensive, and has superior sensitivity, compared with commercial HHV-8 serological assays. A recent comparison of various HHV-8 serological assays found that a latency-associated nuclear antigen immunofluorescent assay similar to the one used in our study had 72% sensitivity and 100% specificity, whereas a lytic antigen immunofluorescent assay had improved sensitivity (85%) but decreased specificity (92%) . Although the sensitivity for detection of HHV-8 antibody in our study could have been improved by including an assay for a second antigen, the high specificity of the latency-associated nuclear antigen immunofluorescent assay used in our study guarded against identification of potential false associations with HHV-8 infection.
Age is associated with HHV-8 infection in other areas of Africa [3, 5, 7, 8]; thus, it was not surprising to find that older age was the greatest risk factor for HHV-8 infection in the ZAPP cohort, in which HHV-8 prevalence was 12.4% higher in the oldest versus the youngest age group. There are several plausible explanations for the relationship between age and HHV-8 prevalence. First, new infections could be acquired throughout adult life. Second, because HHV-8 antibody detection is associated with socioeconomic factors in Africa [7, 18], younger Africans might not have been exposed to the same environmental or behavioral factors associated with HHV-8 infection as have their elders. Last, because HHV-8 antibody titers increase with age , the serological assays used to detect HHV-8 infection could have greater sensitivity in older individuals. Our study and the previous studies that have identified associations between HHV-8 antibody detection and age are limited by a cross-sectional design. Measurement of HHV-8 incidence in longitudinal studies is needed to determine whether HHV-8 infection is acquired during adulthood in Africa.
Despite the high prevalence of HIV-1 and HHV-8 infection among Zimbabwean men, coinfection with these 2 viruses was much less common. The pattern of increasing coinfection prevalence in young adulthood and stable coinfection prevalence after an age of 25 years (figure 3) mirrors the age-specific incidence rates of AIDS-associated Kaposi sarcoma among Zimbabwean men . Because persons infected with both HIV-1 and HHV-8 are at the highest risk for development of Kaposi sarcoma disease, efforts to identify coinfected persons and target them with interventions to prevent Kaposi sarcoma could potentially help to reduce morbidity and mortality due to this disease in African settings.
Our study measured HHV-8 prevalence in the ZAPP cohort because risk factors for STIs have been well characterized and both HIV-1 and herpes simplex virus type 2 are highly prevalent in this cohort [11, 21, 22]. As expected from previous analyses of the ZAPP cohort, age, education, payment for sex, prior STIs, and chronic hepatitis B infection were independently associated with the prevalence of HIV-1 infection in our study group [11, 12]. Of note, the risk of HIV-1 infection was decreased among men who reported occasional or no condom use, compared with men who reported regular condom use, and the risk of HIV-1 infection was higher among married men, compared with single men. The finding of an inverse relationship between self-reported condom use and HIV-1 prevalence in ZAPP was reported previously and is interpreted to mean that men who report regular use of condoms identify themselves as being at risk for HIV-1 infection; condom use itself is not interpreted to be a risk factor for HIV-1 infection . The paradoxical association between marital status and HIV-1 prevalence is thought to be attributable to a higher likelihood of high-risk sexual behavior among married men, compared with single men . The analyses of HIV-1 risk factors in the study group confirm that our data set and analytical methods reproduced the findings of previous studies and that the study group was sexually active and at risk for acquisition of infections that are transmitted through heterosexual intercourse.
A strength of our study was the parallel analyses of risk factors for HIV-1 and HHV-8 infection in the study group. This approach provided an internal control for the sensitivity of our methods to detect potential sexual risk factors for HHV-8 infection. The multivariate model was able to detect associations between HIV-1 infection and known sexual risk factors, such as payment for sex (adjusted OR, 1.75; 95% CI, 1.38–2.23) and self-reported history of an STI (adjusted OR, 2.28; 95% CI, 1.79–2.90), but HHV-8 antibody detection was not independently associated with any sexual risk factors other than prevalence of HIV-1 infection. The lack of an association between HHV-8 infection in spouses, despite strong concordance between spouses with regard to HIV-1 infection, provides further evidence that heterosexual transmission is not a major factor in the spread of HHV-8 infection in urban Zimbabwe.
Although ZAPP participants with HIV-1 infection were at increased risk of HHV-8 infection, the adjusted odds of HHV-8 antibody detection were increased only 40% among HIV-1–infected persons. A similar increase in the odds of HHV-8 infection were observed among HIV-1–infected antenatal Zambian women , and HHV-8 antibody detection is associated with HIV-1 infection status among South African women . Associations between HHV-8 infection and HIV-1 infection were not observed in other African populations [3, 4, 15]. Weak associations between HIV-1 infection and HHV-8 infection could be explained by higher HHV-8 antibody titers in HIV-1–infected persons [14–16, 19], which could differentially increase the sensitivity of HHV-8 antibody detection in HIV-1–infected persons, compared with HIV-1–negative persons. Weak and inconsistent associations between HHV-8 infection and HIV-1 infection across different studies could also be explained by a low level of heterosexual transmission of HHV-8 in African populations.
Some studies have supported a role for heterosexual transmission in the spread of HHV-8 in Africa. HHV-8 prevalence was associated with syphilis, not using condoms, and no circumcision in Kenya ; alcohol consumption and gonorrhea in Kenya ; genital warts and HIV-1 coinfection in Zambia ; and commercial sex work and STIs in Nigeria . However, these studies did not detect significant associations between HHV-8 infection and multiple other sexual risk factors, and risk factor detection was not consistent across studies. Other studies failed to identify any associations between HHV-8 prevalence and sexual risk factors. Among HIV-1–seronegative patients with cancer in Uganda, despite an overall HHV-8 prevalence of 50%, there was no association between HHV-8 antibody detection and sexual behavior, including marital status, number of sex partners, condom use, exchange of gifts for sex, or history of genital discharge . Likewise, a study of heterosexual persons in South Africa found no association between HHV-8 infection and multiple sexual risk factors, despite high prevalences of HIV-1 infection and herpes simplex virus 2 infection in the population [8, 9]. The inconsistent associations between HHV-8 infection and sexual risk factors in Africa have led some investigators to doubt whether heterosexual transmission of HHV-8 occurs . The lack of an association between HHV-8 infection and sexual risk factors among Zimbabwean men who are at high risk of STIs provides additional evidence to support this interpretation.
The findings in the ZAPP study group suggest that it is unlikely that sexual transmission is the dominant mode of HHV-8 transmission in Zimbabwe. Although detection of HHV-8 in female genital tract specimens is uncommon in the absence of Kaposi sarcoma, HHV-8 is often detected in the saliva of African women . Several studies provide evidence that HHV-8 can be transmitted through saliva from mother to child in Zambia and South Africa [25, 26], and it is hypothesized that the use of saliva in traditional medical, ritual, and feeding practices could contribute to transmission of HHV-8 in African settings . Although our study does not directly address the role of saliva in the transmission of HHV-8 in Africa, the lack of concordance of HHV-8 antibody detection between husbands and wives suggests that salivary transmission during sexual relations or other adult-to-adult contact is not common in urban Zimbabwe.
We thank the ZAPP for providing the HIV-1 prevalence, incidence, behavioral, and demographic data used in our analyses.
Financial support. This work was supported by grants from the National Institutes of Health Fogarty International Center (TW0123), the Colorado Center for AIDS Research (AI054907), and the Michael Gelfand Medical Research Foundation.
Presented in part: XVI International AIDS Conference, Toronto, Canada, August 2006 (abstract MOPE0409).
Potential conflicts of interest. All authors: no conflicts.