|Home | About | Journals | Submit | Contact Us | Français|
(See the editorial commentary by Mbulaiteye an Goedert, on pages 575–7.)
Background.Human herpesvirus 8 (HHV-8) infection is endemic in sub-Saharan Africa. We examined sociodemographic, behavioral, and biological factors associated with HHV-8 infection in children and adults to determine HHV-8 seroprevalence and potential routes of transmission.
Methods.Participants were 1383 children and 1477 adults from a population-based sample in a rural community in Uganda. Serum samples were tested for HHV-8 antibodies with use of an enzyme immunoassay against K8.1.
Results.HHV-8 seroprevalence increased from 16% among children aged 1.5–2 years to 32% among children aged 10–13 years (P <.001) and from 37% among participants aged 14–19 years to 49% among adults aged ≥50 years (P <.05). HHV-8 seropositivity in children was independently associated with residing with a seropositive parent (P < .001) and residing with ≥1 other seropositive child aged <14 years (P < .001). History of sharing food and/or sauce plates was marginally associated with HHV-8 infection in children (P = .05). Among 1404 participants aged ≥15 years , there was no association between correlates of sexual behavior (eg, number of lifetime sex partners and HIV infection) and HHV-8 seropositivity (P > .10).
Conclusions.Our data suggest that HHV-8 is acquired primarily through horizontal transmission in childhood from intrafamilial contacts and that transmission continues into adulthood potentially through nonsexual routes.
Human herpesvirus 8 (HHV-8) is the etiologic agent that causes Kaposi sarcoma (KS)  and is associated with primary effusion lymphoma (body cavity–based lymphoma)  and multicentric Castelman's disease . In sub-Saharan Africa, estimates of HHV-8 seroprevalence in adult populations range from 14% to 83% and are highest in East and Central equatorial Africa (“the KS belt”) [4, 5]. Where population-based cancer registries are available, KS is the most common adult cancer in many regions of sub-Saharan Africa, representing up to 40% of all adult malignancies and 10% of all childhood cancers [6, 7]. KS incidence in children and adults has increased dramatically in many sub-Saharan African countries, concurrent with the HIV infection epidemic [6–8]. In Uganda, the incidence of pediatric KS has increased by >40 times, compared with the incidence observed before the HIV infection epidemic .
In regions of Africa where KS is endemic, most HHV-8 infections occur after age 1 year but before puberty, suggesting that HHV-8 is spread horizontally through nonsexual routes [10–12]. HHV-8 is frequently found in saliva, and salivary transmission is a recognized conduit for transmission of the virus . However, little is known about the specific acts by which African children are exposed to saliva. Although some have hypothesized that HHV-8 is spread through acts such as premastication of food (ie, the act of chewing food to make it soft and/or cool to feed child) [12, 14], it is not known whether premastication or other acts by which saliva is passed to children are responsible for transmission of HHV-8. Similarly, little is known about HHV-8 spread in adult populations in Africa. In regions where the seroprevalence of HHV-8 infection is low in the general population (eg, North America and Europe), there is evidence that HHV-8 is transmitted through a sexual route among men who have sex with men [15–21]. However, evidence for sexual transmission of HHV-8 in Africa is mixed. Although some studies have shown associations between HHV-8 infection and HIV infection, sexually transmitted infections, and/or sexual behavior [22–25], others have not [26–28]
We report data from a population-based study conducted in a rural community in Uganda to assess the prevalence of HHV-8 infection among children and adults and to identify sociodemographic, behavioral, and biologic risk factors associated with prevalent HHV-8 infection in children and adults.
The study population comprised all residents of Buziika B Parish, a rural area located in Mukono District, Uganda, ~90 kilometers east of the capital city, Kampala. In May 2002, a complete Global Positioning System–assisted enumeration of all households in Buziika B Parish was conducted. From June through August 2002, study personnel visited each household and invited all household members (defined as a person who had resided in the household for at least the prior 6 months and ate food prepared from a common source) to participate. After the provision of informed consent by the head of household, all other household members were offered enrollment. For household members not present at the time of the initial visit, appointments were given at the study office or study personnel returned to the home at a later date for a minimum of 3 attempts. Informed consent was obtained from all participants ≥18 years of age, and parental/guardian consent was obtained for the participation of children and adolescents <18 years of age. This study was approved by the Institutional Review Boards at the Uganda Virus Research Institute in Entebbe, Uganda; the US Centers for Disease Control and Prevention in Atlanta, Georgia; and the University of California, Berkeley in Berkeley, California.
Trained interviewers administered a structured questionnaire that requested information on demographic variables, socioeconomic status, and environmental exposures. In households where a child or children resided, 1 adult caregiver was asked to answer questions regarding specific behavioral practices through which resident children could have been exposed to the saliva of other household members. Specifically, with respect to each child in the household, respondents were asked whether the child had ever eaten food premasticated by the mother or had ever eaten food or sauce from a plate shared with other household members. Participants who were ≥15 years of age were asked about their number of sexual partners over the prior 12 months and history of genital ulcers or discharge over the prior 12 months. After the interview, blood samples were collected in EDTA tubes for serologic testing from all available consenting household members.
All serum samples were screened in duplicate for antibody to HHV-8 with use of an in-house enzyme immunoassay (EIA) using a synthetic peptide to the viral open reading frame (ORF) K8.1 as the antigen substrate, as previously described by Spira et al , shown to have 87% sensitivity and 100% specificity similar to K8.1-based assays developed by other laboratories . The cutoff value for positivity was the mean corrected optical density (OD) of 10 negative control specimens plus .125 OD units. Specimens with an OD value .001–.025 units above or below the cutoff were defined as equivocal and excluded from analysis.
In addition to HHV-8, serum samples from all participants were tested for the presence of antibody to Epstein-Barr virus (EBV) antibody (IgG EIA assay; Trinity Biotech), antibody to cytomegalovirus (CMV) antibody (Capita IgG EIA assay; Trinity Biotech), hepatitis B virus (HBV) core antibody (HBcAb; Abbot Murex EIA; Murex Biotech Limited), and antibody to herpes simplex virus type 1 (HSV-1; IgG ELISA; Kalon Biological) in accordance with their respective manufacturers’ instructions. For those who provided consent specifically for HIV testing, serum samples were tested for HIV antibody with use of EIAs (Murex Biotech and Recombigen, Cambridge Diagnostics); discrepant HIV results were resolved with Western blot (Genelabs; Genelabs Diagnostics). All serologic testing was performed at the Centers for Disease Control and Prevention–Uganda and Uganda Virus Research Institute laboratories in Entebbe, Uganda.
All analyses excluded children <18 months of age because of the possible presence of passively acquired maternal HHV-8 antibody. Overall and subgroup-specific HHV-8 antibody prevalences were estimated, and differences between subgroups were assessed using χ2 tests adjusted for clustering of multiple respondents in households with use of SAS Proc Surveylogistic (SAS Institute). Odds ratios (ORs) and 95% confidence intervals (CIs) for associations between sociodemographic and family characteristics, and HHV-8 seropositivity for children and adults were calculated by logistic regression with robust standard errors, using PROC Surveylogistic , to account for potential correlation of outcomes measured in the same household. To adjust for possible confounding, all potential explanatory variables were retained in these models.
For participants between 18 months and <14 years of age, we estimated ORs and 95% CIs for associations between presence of antibody to HBcAb, EBV, CMV, and HSV-1 and antibody to HHV-8. Furthermore, we estimated ORs and 95% CIs for associations between acts in which children may be exposed to saliva of others and presence of antibody to HHV-8, HBcAb, EBV, CMV, and HSV-1. Because it was assumed that household members who are infected with these viruses are the predominant source of horizontal transmission of these viruses to children in the household, these analyses were limited to children living in a home where ≥1 other household member was infected with the virus examined. For a subgroup of children living with at least one of their parents, we fit unadjusted and multivariate models to estimate ORs and 95% CIs for associations between sociodemographic variables, HHV-8 status of the mother, father, other children living in the household and other nonparental household members and of antibody to HHV-8 in the child.
In analyses limited to participants ≥15 years of age, we estimated ORs and 95% CIs for associations between indicators of sexual behavior, evidence of antibody presence to HBcAb and HIV, and presence of antibody to HHV-8. We included multiplicative interaction terms in the adult models to assess whether sex modified the association between participant characteristics, sexual behavior/correlates of sexual behavior, or antibody to HBcAb or HIV and HHV-8. We considered evidence for statistical interaction if these multiplicative interaction terms had adjusted P values <.05.
All statistical analyses were performed using SAS, version 9.1 (SAS Institute).
A total of 3335 individuals were approached (91% of the Census population), of whom 3133 (94%) were ≥18 months of age and, thus, eligible for inclusion in the study. Of those, 3129 (99.9%) provided consent and were enrolled in the study. HHV-8 serologic results were available for 3035 (97%) of these individuals; 175 (5.8%) were classified as having equivocal HHV-8 antibody results and were not considered in subsequent analyses. Characteristics of the 2860 participants from a total of 731 households are shown in Table 1.
Among 1383 children (18 months to ≤13 years of age), HHV-8 seroprevalence was 15% (30 of 194) among children 1.5–2 years of age and increased to 32% (112 of 350) among children 10–13 years of age (P <.001) (Table 1). Adjustment for sex, ethnicity, household density, drinking water source, and number of HHV-8–seropositive household members did not change the relationship between age and HHV-8 serostatus. In multivariable analysis, the odds of HHV-8 seropositivity were significantly higher among children living in households with ≥2 persons who were HHV-8 seropositive, compared with those living in a home with no other person who was HHV-8 seropositive, but was inversely associated with household density. We evaluated the associations between the HHV-8 serostatus of participants 18 months to ≤13 years of age and the HHV-8 serostatus of their parents and other household members (Table 2). Of the 1383 children in the study, 1042 lived in a home with at least 1 parent who had an evaluable HHV-8 result (350 with mother only, 112 with father only, and 580 with mother and father). In multivariate analyses adjusted for child's age, parental HHV-8 serostatus and number of HHV-8–seropositive household members ≤13 years of age, the odds of HHV-8 seropositivity in a child were significantly increased for those with at least 1 HHV-8–seropositive parent (P <.001) and for children residing in a home with ≥1 other seropositive household member aged <13 years (P < .001). Among the 580 children who lived with both parents, we assessed whether children's HHV-8 status was associated with that of their mother and/or father (Table 3). There was no evidence for an increased risk of HHV-8 infection in children due to HHV-8 in either the mother or the father alone (P = .31 and P = .51, respectively).
At least 90% of children had evidence of infection with HSV-1, EBV, and/or CMV (Table 4). Approximately 15% (202 of 1370) were HBcAb positive (Table 4) and <1% (4 of 1352) were HIV infected. Similar to findings for HHV-8, an increase with age was observed for HBcAb (P <.001), HSV-1 (P <.001), CMV (P = .01), and EBV (P = .02) positivity (Figure 1). After adjusting for age, there was an association between HHV-8 infection and the presence of antibody to HSV-1 but not the presence of antibody to EBV, CMV, or HBV core (Table 4).
We examined a possible association between exposure to food premasticated by the mother and seropositivity for HHV-8, HSV-1, EBV, CMV, and HBV in the child. Of the 923 children who lived with their mother, 81 (8.8%) were reported to have ever been exposed to food premasticated by the mother. In separate analyses restricted to children living with their mothers who were seropositive for the viruses examined (ie, HHV-8, HSV-1, EBV, CMV, or HBcAb) and adjusted for child's age only, there was no evidence for an association between having ever been exposed to food that had been premasticated by the mother and HHV-8 (P = .31), HSV-1 (P = .84), EBV (P = .36), CMV(P = .11), or HBcAb seropositivity in the child (P = .65) (Table 5).
Similarly, we examined the association between the act of sharing a food or sauce plate with other household members and seropositivity to HHV-8, HSV-1, EBV, CMV, and HBcAb in the child. In all, 1258 (91%) of 1377 children were reported to have ever shared a food and/or sauce plate with other household members. In separate analyses restricted to children living in a household with ≥1 other person seropositive for the viruses examined and adjusted for child's age only, there was no evidence for an association between CMV, EBV, and HSV-1 and having ever shared a food or sauce plate with other household members (Table 5). However, children who engaged in this practice had significantly greater odds of being HHV-8 and/or HBcAb seropositive (Table 5), compared with children who did not engage in this practice, although the effect was marginally significant for HHV-8 (for HHV-8, adjusted OR, 2.1 [95% CI, .99–4.3], P = .05; for HBcAb, adjusted OR, 3.0 [95% CI, 1.2–7.5], P = .02).
Of the 1477 participants ≥14 years of age, HHV-8 seroprevalence increases from 37% (118 of 321) among those 14–19 years of age to 49% (147 of 298) among persons ≥50 years of age (P =.01) (Table 1). Adjustment for sex, ethnicity, household density, drinking water source, and number of HHV-8–seropositive household members did not change the relationship between age and HHV-8 serostatus. In multivariable analyses, more men than women were HHV-8 seropositive (43% vs 38%; P = .04). HHV-8 seropositivity in adults was associated residing in a household with >2 HHV-8–seropositive persons but was inversely associated with household density.
Among 1404 participants ≥15 years of age, there was no association between correlates of sexual behavior (eg, number of lifetime sex partners, genital ulcers, discharge, or HIV infection) and HHV-8 seropositivity. There were not important differences according to sex (P values for interaction >.20), and hence, all partner types were analyzed together (Table 6). However, in analyses adjusted for age, the odds of HHV-8 seropositivity in women were significantly increased among those who were HBcAb-seropositive, compared with those who were HBcAb seronegative (Table 6); for men, there was no association between HBcAb seropositivity and HHV-8 seropositivity.
More than 90% of adults had evidence of infection with HSV-1, EBV, or CMV. Approximately 735 (50%) of 1469 adults were HBcAb positive, and 81 (5.7%) of 1434 were HIV infected. In contrast to HHV-8, HIV and HBcAb serostatus were associated with sexual activity. In age- and sex-adjusted analyses, the odds of HIV infection increased with number of lifetime sex partners (P <.001). The odds of HBcAb seropositivity were increased for adults with ≥4 lifetime sex partners, compared with those who had never had sex (adjusted OR, 1.8; 95% CI, 1.1–2.7; P = .01). Furthermore, adults who reported having a history of genital ulcers or ever having vaginal or penile discharge had greater odds of being HIV seropositive (adjusted OR, 2.7; 95% CI, 1.7–4.2; P < .001; adjusted OR, 2.4; 95% CI, 1.4–4.2; P = .001) or HBcAb seropositive (adjusted OR, 1.4; 95% CI, 1.1–1.7; P = .002; adjusted OR, 1.6; 95% CI, 1.1–2.2; P = .008).
In rural Uganda, HHV-8 infection was prevalent in early childhood and increased with age throughout adulthood, with the majority of infections occurring before age 14 years. This age-dependent pattern is consistent with other studies in countries where KS is endemic [4, 11, 32] including Uganda [5, 10–12]. The observed pattern of increasing seroprevalence of HHV-8 infection with increasing age was similar to that seen for HSV-1, HBV, EBV, and CMV infection, all of which are viruses known to be transmitted horizontally during childhood [33–39]. This pattern suggests that HHV-8 is mainly spread through nonsexual, horizontal routes during childhood and that transmission is ongoing throughout the adult years.
Common to all of the viruses examined—HHV-8, HSV-1, HBV, EBV, and CMV—is their presence in saliva [13, 36, 40–42], a recognized vehicle for transmission for these viruses and a possible explanation for a pattern of horizontal nonsexual spread among African children. To our knowledge, this is the first study to examine whether exposure to saliva through specific acts practiced with young children contributes to transmission of HHV-8 to children in sub-Saharan Africa. In our study, children who frequently shared food or sauce plates with other household members, among whom ≥1 were HHV-8 seropositive, had a marginally higher odds of being HHV-8 seropositive than those children who never shared a common plate. It is plausible that children who come into close contact with food or sauce plates shared with others are exposed to saliva of other household members, especially other children, who are actively shedding HHV-8 virus. It is also possible that this act is correlated with behaviors not examined here. Indeed, there are possibly a number of other acts by which children are exposed to saliva in this community, as have been revealed by more recent work in other sub-Saharan African communities [34, 43, 44]. We did not detect an association between reported exposure to food premasticated by the mother and seropositivity to any of the viruses examined. This could possibly be because mothers who were seropositive owing to distant childhood infection could have been shedding much lower numbers of viral particles in their saliva when the act occurred, compared with, for instance, younger siblings who were more recently infected. Furthermore, our inability to detect associations between risk factors examined and infection with HHV-8 or other viruses could be attributable to type II error, especially in the younger age groups, chance, or failure to measure risk behavior in caregivers other than the mother.
The finding of a 2–4-fold increase in seroprevalence of HHV-8 infection among children and adults living in households with ≥2 HHV-8–seropositive household members, compared with children and adults living in households with no other HHV-8–seropositive person provides evidence for intra-household HHV-8 transmission in children and adults. However, our observation of an age-related increase in seroprevalence of HHV-8 infection among children regardless of the presence of other HHV-8–seropositive household members suggests that transmission from HHV-8–infected persons outside the household may also be important. These findings, consistent with data from prior other studies [12, 14], imply that household members and persons residing outside the household may play an important role in the transmission of HHV-8 to children. Furthermore, we found that parental HHV-8 serostatus was independently associated with that of their child, although we found no association between child's infection status and that of her/his mother or father specifically.
Among children, seroprevalence of HHV-8 infection did not differ significantly by sex, but among adults, HHV-8 seroprevalence was significantly higher among men than among women, a finding that is consistent with at least 1 other report from this region . Consistent with other studies of adults in sub-Saharan Africa [27, 28, 45, 46], we found no evidence for an association between HHV-8 seropositivity and number of lifetime sex partners, history of genital ulcers, history of vaginal/penile discharge, or HIV infection. Furthermore, although there was an overall increase in HHV-8 seroprevalence with age in adults, there was little increase in HHV-8 seroprevalence in both women and men aged 14–34 years, the years of peak sexual activity with different partners. These findings are in marked contrast to our findings for HIV and HBV infection, both of which increased sharply after age 15 years and were significantly associated with all indicators of sexual activity. Although a statistically significant association between HHV-8 and HBcAb was observed for women, the association could also be explained by nonsexual horizontal transmission of HHV-8. Although we cannot rule out the possibility of some HHV-8 spread through sexual activity, the lack of association between HHV-8 serostatus and indicators of sexual activity suggests that it does not play a substantial role in transmission in our population.
A limitation of our work is the self-reported nature of acts in which saliva may be passed to children and, among adults, sexual behavior. However, because public health messages pertaining to either hygiene-related behavior or sexual behavior generally do not mention saliva, we do not believe that participants have underreported practices to provide socially desirable responses. Neither interviewers nor participants knew the participant's HHV-8, CMV, HSV-1, EBV, or HBV serostatus, thereby minimizing selective reporting. Finally, the cross-sectional study design precludes our ability to determine when infection occurred, and thus, it is not possible to establish causal associations between the behaviors examined and infection with HHV-8 or the other viruses examined.
In conclusion, we found seroprevalence of HHV-8 infection to be high among young children and to increase with age among children and adults in rural East Africa. HHV-8 transmission in this population appears to be mainly attributable to horizontal transmission from members in and outside households. Furthermore, our data suggest that transmission is ongoing in adulthood, most likely by nonsexual routes. Although the importance of horizontal transmission in childhood is evident, the specific routes of horizontal transmission remain unclear, and they need to be further investigated before effective prevention messages can be given. It is only through prospective longitudinal observation of uninfected newborns in their early childhood years in conjunction with detailed examination of their close contacts (ie, parents, siblings, other household members, and other persons outside of the home who have close contact with children) that we can definitively understand the types of interpersonal contact that transmit HHV-8, the biological factors in the infected and at-risk persons that mediate infectiousness and susceptibility to infection and the behavioral aspects of these contacts that mediate transmission.
This work was supported by the Global AIDS Program, CDC, and the National Institute of Child Health and Human Development (K01 HD052020 to LMB).
We thank members of the Buziika community and to the Buziika study staff, for their participation and work, and Drs. Arthur Reingold, Alan Hubbard, and Gertrude Buehring.