The Uganda Kaposi's Sarcoma Study Group includes named authors and V. Sembajwe, M. Kalinaki, R. Byansi, C. Rwatooro, S. Nambooze, B. Tushimiere, N. Byabazaire (deceased), E. Bitamazire, E. Katabira, J. Mugerwa (deceased), D. Tindyebwa, C. Ateenyi-Agaba, L. Marum, J. Whitworth, B. Richardson and K. de Cock.
Windeyer Institute, University College London, London, United Kingdom
Kaposi's sarcoma-associated herpes virus (KSHV; also known as human herpesvirus-8 [HHV-8]) is considered to be a necessary cause of Kaposi's sarcoma.1, 2 The virus is most prevalent in groups or populations at highest risk of developing Kaposi's sarcoma, such as human immunodeficiency virus-1 (HIV-1)-infected homosexual men in Europe and the United States and in heterosexual men and women in parts of Africa, such as Uganda, where Kaposi's sarcoma was endemic before the spread of HIV.3, 4, 5, 6, 7 Marked increases in the incidence of Kaposi's sarcoma have been noted in several African centres, following the spread of the HIV epidemic.4, 8, 9
The modes of transmission of KSHV are not fully understood. In Europe and the United States, sex between men is the main behavioural risk factor for HIV-associated Kaposi's sarcoma and also appears to be a route of transmission of KSHV infection.10, 11, 12, 13 In contrast, there is some evidence for transmission of KSHV both heterosexually (i.e., the prevalence of the virus is higher among those with more sexual partners) and from a mother to her child in the South African black population.14, 15, 16 Other modes of transmission may also occur in Africa.17, 18, 19 The aim of our study was to investigate sociodemographic and lifestyle factors associated with KSHV infection. Uganda was thought to be an ideal setting in which to do this because Kaposi's sarcoma was relatively common there, in the years before HIV, representing up to 9% of all cancers in men.7, 20
MATERIAL AND METHODS
From 1994 to 1998, we systematically recruited adults 15 years or older with a provisional new diagnosis of cancer from the wards and outpatient clinics of the 4 main hospitals in Kampala, Uganda: Mulago (including the Uganda Cancer Institute), Nsambya, Mengo and Rubaga. After cancer patients were identified and gave informed consent, trained interviewers administered a standard questionnaire that requested information on demographic variables, socioeconomic status, environmental exposures, tobacco and alcohol habits and sexual and reproductive behaviour before the onset of illness. Further details of study methods can be found in Ziegler et al. on pages 233–240, which describes a case-control study of Kaposi's sarcoma.21
Having received pretest counseling, blood (or saliva) was taken from consenting patients both for an HIV test and for storage of serum. Results were reported back to the patient at the next clinic visit, along with posttest counseling. The HIV test employed the Cambridge Bioscience Recombigen ELISA (Cambridge, MA), and when serum was unavailable, we tested saliva using the GACELISA method (Murex, Dartford, UK). Our study was approved by the Committee on Human Research (VA Medical Centre and University of California San Francisco) and by the Uganda National Council for Science and Technology.
Any remaining sera were stored at −80°C and were later shipped on dry ice to University College London for KSHV testing. All assays were performed by a single investigator (D.B.), who was unaware of each patient's personal characteristics and diagnosis. A body-cavity-related B-cell lymphoma cell, BCP-1, which is positive for KSHV and negative for Epstein Barr virus, was used for an indirect immunofluorescence assay to detect IgG antibodies against KSHV antigen.5, 22 Latently infected BCP-1 cells were fixed in 4% paraformaldehyde and were made permeable with 0.2% Triton X-100. Cells were resuspended in PBS and fixed on glass slides. The samples were diluted 1:100 in PBS with 3% FCS. The diluted serum was added to fixed BCP-1 cells and incubated at 22°C for 45 min.
After the slides were washed in PBS with 3% FCS, rabbit antihuman IgG labeled with fluorescein isothiocyanate (Dako, High Wycombe, UK), diluted 1:40 in PBS with 3% FCS, was added and the slides were incubated at 22°C for 20 min. The slides were then washed in PBS without FCS and screened by ultraviolet microscopy for the nuclear stippling pattern characteristic of antibodies against the latent nuclear antigen of KSHV encoded by orf73.5
Serum samples that were positive for antibodies against KSHV by the immunofluorescence assay were then scored as low, medium or high, according to the intensity of the fluorescent signal. One hundred positive samples were retested in a blinded fashion at doubling dilutions, starting at 1:100 for antibodies against KSHV.23 The score for the fluorescent signal intensity was strongly related to antibody titre—the median titres were 1:800 for low-signal intensity (1+), 1:25,600 for medium-signal intensity (2+) and 1:51,200 for high-signal intensity (3+; p < 0.001). Further details of the scoring mechanism for signal intensity of anti-KSHV antibodies are provided elsewhere.15, 24
The people included in our study are restricted to 607 HIV-seronegative patients without Kaposi's sarcoma, for whom a stored blood sample was available for KSHV testing. The subjects comprised men (201) and women (406) with cancers of the oral cavity (21), oesophagus (36), stomach (19), liver (41), skin (21), breast (58), cervix (140), ovary (18), prostate (11), penis (10), eye (18) and non-Hodgkin's lymphoma (28), Hodgkin's disease (19) and other cancer sites or types (104). In addition, 63 patients with a provisional diagnosis of cancer who subsequently were diagnosed with a nonmalignant condition were included as a separate group. These people comprised part of the comparison group in a case-control study of Kaposi's sarcoma described in Ziegler et al.21
Anti-KSHV antibody status was also tested for in 432 patients who were seropositive for HIV-1 infection. Although HIV-1 infection was not associated with the presence of anti-KSHV antibodies (p = 0.2), those who were infected with HIV-1 were not included in the data presented here. Results on the anti-KSHV antibodies in HIV-1-infected individuals showed no clear association with the factors examined (data not shown).
Data were computerised by trained clerks using EPI-INFO software, and statistical analyses were conducted using STATA.25, 26 To examine the relation between cancer site or type and KSHV serostatus, an unconditional logistic regression model was used to estimate the log of the odds of KSHV seropositivity for each type of cancer, with adjustment for age group (<30, 30–44, 45+) and sex. The resulting estimates were used to obtain prevalence rates of KSHV for each cancer type (Fig. 1).15 An unconditional logistic regression model was used to estimate the odds ratio (OR) associated with KSHV seropositivity for each variable, in the whole HIV-1-seronegative study population, irrespective of cancer site, adjusted for age (<30, 30–44, 45+) and sex. All p-values are 2-sided. To exclude the possibility that people with apparently low titres of anti-KSHV antibodies (1+ fluorescent signal intensity; see above) were misclassified as being KSHV seropositive, the analyses were repeated, comparing people who had medium or high titres (2+ or 3+) with KSHV seronegatives. The data are not shown because the results did not materially change.
To investigate the determinants of high-titre infection among those with antibodies against KSHV infection, a further analysis was conducted. An unconditional logistic regression model was used to estimate, for each variable, the odds ratio (OR) associated with a medium- and high-signal intensity of anti-KSHV antibodies (as defined above) for each variable, as compared to those with a low-signal intensity response (i.e., low titre) to the immunofluorescent assay.
Of the 607 study participants, 302 (50%) had anti-KSHV antibodies and the prevalence did not vary significantly by cancer site or type (χ2 heterogeneity [14 degrees of freedom] = 15.4, p = 0.4; Fig. 1). The prevalence of anti-KSHV antibodies did not differ significantly by sex (p = 0.2), being 53% in men (107/201) and 48% in women (195/406), but increased linearly with age from 35% in those under 25 years (22/63), 45% for age 25–34 (39/86), 47% for ages 35–44 (69/146), to 55% in those over 45 years (172/312; p-trend < 0.001). Antibodies against KSHV were more common in tribal groups other than the Baganda tribe (54% and 45%, respectively; odds ratio [OR] = 1.5, 95% confidence intervals [CI] 1.1–2.0; p = 0.02), but there was no significant variation in seroprevalence by district of birth (p = 0.9), region of residence prior to becoming ill (p = 0.3) or nationality (p = 0.2; Table I).
Table I. Risk Factors for KSHV (in Those Without Kaposi's Sarcoma) Among HIV-Seronegative People in Uganda
Results for social and demographic factors are displayed in Table II. The prevalence of anti-KSHV antibodies declined from 55% in those who never went to school to 44% in those who left school at 16 or more years of age, although the results were of borderline statistical significance (OR = 0.7, 95% CI 0.4–1.0; p = 0.06). In addition, the prevalence of antibodies against KSHV decreased with increasing birth order (i.e., with increasing number of older siblings; p = 0.05). In the first or second born, the seroprevalence of KSHV was 53%, declining to 44% in those born fifth or later (OR = 0.7, 95% CI 0.5–1.0).
Table II. Social and Demographic Factors in Relation to KSHV Sero-Status
Odds ratios adjusted for age group and sex. df, degrees of freedom.
Age left home (years)
16+ (or never)
χ2 heterogeneity (1 df) = 0.01; p = 0.9
Years at current address
χ2 trend (1 df) = 0.01; p = 0.9
χ2 trend (1 df) = 0.09; p = 0.8
Travel time to market
χ2 heterogeneity (1 df) = 0.5; p = 0.5
Age left school
χ2 trend (1 df) = 3.5; p = 0.06
χ2 trend (1 df) = 1.1; p = 0.3
Number in house
χ2 heterogeneity (1 df) = 2.6; p = 0.1
Number of siblings
χ2 heterogeneity (1 df) = 2.0; p = 0.2
Birth order (no. older siblings)
χ2 trend (1 df) = 3.9; p = 0.05
Number sharing toilet
χ2 trend (1 df) = 0.01; p = 0.9
Table III displays results for variables characterising wealth in Uganda. None of the factors under consideration was significantly associated with antibodies against KSHV, although ownership of pigs or goats was borderline (OR = 1.4, 95% CI 1.0–1.9; p = 0.06). Similarly, in Table IV, the association of other exposures with anti-KSHV antibodies is shown—none were significantly linked to KSHV antibody seropositivity. Table V shows the results for sexual and reproductive behaviour in relation to anti-KSHV antibodies. The risk of being KSHV-seropositive was higher among people who reported ever having been married as compared to people who never married (OR = 2.1, 95% CI 1.1–4.1; p = 0.03). There was no significant association between any of the other factors studied and the presence of antibodies, with the exception of a lower risk of anti-KSHV antibodies among women who had undergone labial (genital) elongation (OR = 0.5, 95% CI 0.3–0.8; p = 0.01). Seventy-one percent of women had undergone labial elongation, which was particularly a feature of members of the Baganda tribe (94% vs. 49% in other tribal groups). After further adjustment for tribe, the result was no longer statistically significant, although the odds ratio had changed little (OR = 0.6, 95% CI 0.4–1.0; p = 0.06).
Table III. Some Factors Characterising Wealth in Relation to KSHV Sero-Status
Of the 302 individuals with anti-KSHV antibodies, 167 had a response that was graded as low-signal intensity on the immunofluorescent assay, and 135 (45%) had a medium- or high-signal intensity. Using those with low-intensity infection (corresponding to a low titre of anti-KSHV antibodies) as a comparison group, only 2 variables were statistically significantly associated with a medium- or high-signal intensity response (i.e., a high titre of anti-KSHV antibodies). The proportion of people with a medium- or high-signal intensity of anti-KSHV antibodies increased from 38% (23/61) in those under the age of 35 years to 54% (56/103) in those 55 years or older (χ2 trend = 4.9, p = 0.03). Those who reported having had a blood transfusion in the past were more likely than those who did not to have a medium- or high-signal intensity of anti-KSHV antibodies (OR = 2.0, 95% CI 1.1–3.6; p = 0.03). In people with anti-KSHV antibodies, no other variable investigated was significantly associated with medium- or high-signal intensity response to the immunofluorescent assay.
Little is known about the sero-epidemiology of KSHV in Africa, despite the fact that the virus is prevalent there and that Kaposi's sarcoma was endemic in some areas in the years before HIV. Even though nearly 50 potential risk factors were examined in our study, including sociodemographic, sexual and reproductive variables, few significant associations were found.
The seroprevalence of anti-KSHV antibodies in Uganda is one of the highest in the world. In our study, about 50% of adults were seropositive. Prevalence did not differ significantly by sex but increased with increasing age. These results are in accord with those from other studies in Africa.15, 17, 27 Antibodies against KSHV were less prevalent in members of the Baganda tribe compared to other tribal groups, and the reasons for this are not clear, particularly given the lack of any variation in the prevalence of antibodies by nationality, region of birth or region of residence.
The prevalence of anti-KSHV antibodies was slightly lower, but not significantly so, in those with more education. Although the results presented here were of borderline statistical significance, similar findings have been reported from South Africa.15 The prevalence of anti-KSHV antibodies also appeared to decrease with increasing birth order, but this result could also be a chance finding, since almost 50 significance tests were performed.
This investigation has not shed much light on the modes of transmission of KSHV in Ugandan adults. In South Africa, where Kaposi's sarcoma is less common than in Uganda, there was some evidence of an increasing seroprevalence of KSHV with increasing number of lifetime sexual partners.15 Those results were not replicated here, although people who had been or were married were about twice as likely to have anti-KSHV antibodies as compared to people who had never been married. The increasing prevalence of antibodies with age, however, suggests that new infections continue to be acquired throughout adult life. Studies from the United States indicate that saliva may be an important vehicle for transmission of KSHV infection but results presented here cannot shed further light on this hypothesis.28
There is evidence that high anti-KSHV antibody titres are an important determinant of the risk of Kaposi's sarcoma and perhaps also of transmission of KSHV from person to person.15, 16, 21, 24, 29 Presumably, the titre of antibodies against KSHV reflects the viral load, and there is currently some evidence to support this.30, 31 Nothing is known of the determinants of high antibody titres in individuals with anti-KSHV antibodies. In our study, among the 302 people with antibodies against KSHV, the proportion with high titres increased with increasing age and with a past history of blood transfusion. Injecting drug use has been found to be a risk factor for anti-KSHV antibodies in the United States, suggesting the possibility of parenteral transmission of KSHV, although the effect on antibody titre is unclear.33
In conclusion, this systematic investigation of factors associated with KSHV infection, in a population where the virus is relatively common, found that the prevalence of anti-KSHV antibodies increased with age but there is little evidence of variation in prevalence according to nearly 50 factors studied. This is in marked contrast to Kaposi's sarcoma (see Ziegler et al. on pages 233–240),21 which is associated with a number of factors other than infection with KSHV.
The authors would like to acknowledge the support of Mulago Hospital and Makerere Medical School, Kampala, Uganda.