Risk factors for human Herpesvirus 8 infection and AIDS-associated Kaposi's sarcoma among men who have sex with men in a European multicentre study

Abstract

We aimed to identify risk factors for Kaposi's sarcoma (KS) among HIV-positive patients and behaviors associated with human Herpesvirus 8 (HHV-8) infection, as well as to assess KS incidence and mortality rates longitudinally. To fulfill the first objective, a European case-control study was designed in the early 1990s (each KS case was matched to 2 controls with another AIDS indicative disease). After the discovery of HHV-8, serology testing enabled us to assess risk factors for KS development among HHV-8 and HIV-1 coinfected men who have sex with men (MSM), as well as risk factors for HHV-8 infection. HHV-8 seroprevalence was determined using a latent immunofluorescence assay. Relevant information was obtained by means of a questionnaire and medical charts review. Assessment of risk factors for KS development and HHV-8 infection was performed using conditional and unconditional logistic regression models, respectively. A low CD4 count was the only significant risk factor for KS. HHV-8 infection was most strongly linked to the number of life-time sex partners, and multiple body fluids such as saliva and semen are quite likely involved in sexual transmission. Longitudinal follow up showed a significant protective role for highly-active antiretroviral therapy (HAART) both on KS development and mortality of KS patients. Although more conclusive data from cohort studies are needed to better define specific transmission mechanisms for HHV-8, our results contribute to explain why KS incidence is higher among MSM, and the decreasing KS incidence trend observed in countries with universal access to HAART. © 2006 Wiley-Liss, Inc.

Kaposi's sarcoma (KS), first described in 1872,1 is an angioproliferative disease occurring in 4 clinical-epidemiological forms. Epidemic or AIDS-KS occurs in HIV-infected individuals and is the most aggressive variant.2, 3 Human Herpesvirus 8 (HHV-8), which was identified in AIDS-KS tissue in 1994,4 is necessary for the development of all forms of KS.5 HHV-8 seroprevalence rates are low in Northern Europe and the US (0–5%), intermediate in Italy (5–35%), Greece (13–20%) and Spain (4.5–8%) and highest in regions of sub-Saharan Africa (30–60%).6, 7, 8

In the early 1980s, numerous cases of AIDS-KS among homosexual men arose in Europe and the US related to the HIV epidemic and the accompanying immunosuppression.9, 10, 11 AIDS-KS incidence decreased after that together with a decline in sexual promiscuity and HIV incidence,10, 12, 13 as well as after HAART introduction in 199614, 15, 16 probably due in part to enhanced immune reconstitution,17 and perhaps a direct effect of protease inhibitors as antiangiogenic factors.18 However, the proportion of AIDS diagnoses attributable to KS in Europe has not changed over time.16 KS is currently the 6th most common AIDS indicative disease and the first cause of neoplasm among AIDS patients in Western Europe.19

HIV infection has been suggested to contribute to the pathogenesis of KS through immunosuppression, perturbation in the expression of certain cytokines and the effect of HIV-1 Tat protein.20, 21 Other factors have also been suggested to influence KS development, such as degree of immunosuppression,11 temporal order of HHV-8 and HIV infection,11, 22 HIV or HHV-8 viral load23, 24, 25 and homosexual behaviour.26 However, only 30–50% of men who are coinfected with HHV-8 and HIV-1 develop KS within 5–10 years9, 11, 22, 26, 27 and, therefore, other risk factors might be involved.

Prevalence of HHV-8 infection and incidence of AIDS-KS are disproportionately higher among HIV positive men who have sex with men (MSM) when compared to other HIV-risk groups, which has long suggested that HHV-8 can be sexually transmitted.8, 28, 29, 30 However, identification of specific sexual practices among MSM as risk factors for HHV-8 infection has been so far controversial because of many possible sources of bias.31 Besides, detection of HHV-8 DNA in semen, saliva and rectal specimens has also been variable.32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43

The aims of this study were first, to identify potential risk factors for KS development among HIV-positive patients, especially among those coinfected with HIV-1 and HHV-8; and second, to describe specific sexual behaviors associated with HHV-8 infection. Finally, we followed up Spanish participants for a 10-year period to assess KS incidence and mortality rates.

AIDS, acquired immunodeficiency syndrome; CI, confidence interval; EURO-SHAKS, European study on HIV associated Kaposi's sarcoma; HAART, highly-active antiretroviral therapy; HHV-8, human Herpesvirus 8; HIV-1, human immunodeficiency virus type 1; IFA, immunofluorescence assay; KS, Kaposi's sarcoma; MSM, men who have sex with men; OR, odds ratio; RR, relative risk; STI, sexually transmitted infections; UK, United Kingdom.

Material and methods

Study design and inclusion criteria for patients

In 1993, a multicentre matched case-control study funded by the European Union (EURO-SHAKS: European Study on HIV Associated Kaposi's Sarcoma, DG XII)44 was designed to identify potential biological, behavioral and environmental risk factors for AIDS-KS. Participant countries were Spain, Belgium, Italy, the UK and Greece. Cases were defined as any AIDS patient with a diagnosis of KS (clinical stages I to IV), and controls as any AIDS patient with an indicative disease other than KS or with a CD4 cell count of less than 200,45 who had been diagnosed ±4 months from the date of the case diagnosis. Each case was matched with 2 controls by sex, age and transmission category. An informed consent was obtained from all study participants.

After the first identification of HHV-8, sera samples from both cases and controls were tested for the presence of antibodies against this virus. Spanish study controls were followed up yearly, by means of contacting the responsible physician and systematically reviewing medical records. Information regarding occurrence of KS, vital status and exposure to HAART were collected.

Clinical specimens and HHV-8 diagnostic methods

Among other specimens,44 peripheral blood was collected from all participants at study entry. Serum specimens were stored at −20°C and tested for the presence of antibodies against HHV-8 using an immunofluoresence assay (IFA) on latently infected cells, as previously described.7

Data collection

Demographic and behavioral data were collected from all participants by means of an extensive personal interview performed by trained persons. No personal identifier was used. Clinical data were collected by standardized review of medical records. CD4 values were obtained at the date of KS and AIDS diagnosis, respectively, for cases and controls. Among other questions, study participants were asked about their medical history during the 10 years preceding the AIDS diagnosis. Concerning sexual behavior, men who claimed having had sex with other men were asked about the number of male partners in their lifetime. Additionally, they were asked how often they practiced orogenital, oroanal or anogenital sex without condom. Answers were coded as 0 = “never”, 1 = “less than once a month”, 2 = “once a month or more”. All sexually transmitted infections (STI) were introduced as a single variable defined as the number of STI (NSTI, ranging from 0 to 6).

Statistical analysis

Conditional logistic regression models46 were used to study risk factors associated with KS development in the initial matched case-control design (analysis 1; Table I). Moreover, unconditional logistic regression models were performed for the analysis of determinants for KS among HHV-8 seropositive individuals (analysis 2; Table I), and risk factors for HHV-8 infection among controls (analysis 3; Table II) as well as among controls plus KS patients (analysis 4; Table II), since unmatched designs were considered. In all analyses, the following model-building strategy was used: variables with significant unadjusted odds ratio (OR) at level 0.15 were included in the initial multivariate model in order to take into account potential confounding factors. Variables with significant values of the Wald statistic at level 0.15 were retained in the most parsimonious final multivariate model.

Table I. Risk Factors for AIDS Associated Kaposi's Sarcoma (Univariate and Multivariate Analysis)

	Case–control design (N = 426)				Among HHV-8 positive (N = 247)
	Univariate		Multivariate		Univariate		Multivariate1
	OR (95% CI)	p	OR (95% CI)	P	OR (95% CI)	p	OR (95% CI)	p
KS, Kaposi's sarcoma; HHV-8, human Herpesvirus 8; NSTI, number of sexually transmitted infections; OR, odds ratio; CI, confidence interval. 1 Adjusted by age at diagnosis.
CD4 count
≥301	1	0.155	1	0.133	1	0.216	1	0.210
201–300	1.89 (0.78–4.57)	0.137	2.07 (0.80–5.35)	0.066	1.89 (0.69–5.21)	0.229	1.96 (0.69–5.62)	0.261
101–200	1.74 (0.84–3.63)	0.018	2.13 (0.95–4.81)	0.014	1.68 (0.72–3.93)	0.002	1.65 (0.69–3.97)	0.006
≤100	2.28 (1.15–4.50)		2.59 (1.21–5.57)		3.52 (1.56–7.97)		3.18 (1.39–7.29)
Age	0.99 (0.96–1.04)	0.834			0.98 (0.95–1.01)	0.162
NSTI	1.01 (0.91–1.12)	0.887			1.04 (0.84–1.29)	0.743
Sexual partners during life
≤50	1	0.100	1	0.188	1	0.204
50–300	1.72 (0.91–3.28)	0.211	1.64 (0.79–3.42)	0.032	0.59 (0.27–1.33)	0.462
301–600	1.61 (0.77–3.37)	0.006	2.48 (1.08–5.71)	0.001	0.70 (0.27–1.83)	0.497
≥601	2.48 (1.30–4.72)		3.63 (1.65–8.01)		0.75 (0.33–1.71)
Anogenital insertive sex
Never	1	0.038			1	0.339
Less than once a month	1.93 (1.04–4.33)	0.050			1.42 (0.69–2.90)	0.285
Once a month or more	1.75 (1.00–3.08)				1.40 (0.76–2.58)
Anogenital receptive sex
Never	1	0.071			1	0.569
Less than once a month	1.78 (0.95–3.31)	0.310			1.23 (0.60–2.52)	0.482
Once a month or more	1.30 (0.78–2.15)				1.25 (0.67–2.33)
Orogenital insertive sex
Never	1	0.485			1	0.599
Less than once a month	1.31 (0.51–2.88)	0.926			0.76 (0.27–2.13)	0.156
Once a month or more	1.03 (0.67–1.88)				0.56 (0.25–1.25)
Orogenital receptive sex
Never	1	0.005	1	0.030	1	0.122
Less than once a month	3.20 (1.42–7.22)	0.162	2.72 (1.11–6.67)	0.458	2.22 (0.81–6.13)	0.696
Once a month or more	1.56 (0.84–2.93)		1.32 (0.63–2.77)		1.02 (0.47–2.20)
Oroanal insertive sex
Never	1	0.417			1	0.023
Less than once a month	0.78 (0.42–1.43)	0.688			0.45 (0.22–0.90)	0.270
Once a month or more	0.89 (0.51–1.57)				0.68 (0.34–1.35)
Oroanal receptive sex
Never	1	0.530			1	0.376
Less than once a month	1.19 (0.69–2.04)	0.409			0.75 (0.39–1.42)	0.857
Once a month or more	1.26 (0.73–2.16)				0.94 (0.48–1.85)

Table II. Risk Factors for HHV-8 Infection among Controls only and among Controls Plus KS Patients (Univariate and Multivariate Analysis)

	162 HHV-8 seronegative vs 105 HHV-8 seropositive controls (N = 267)				162 HHV-8 seronegative vs 247 HHV-8 seropositive patients, including KS patients (N = 409)
	Univariate		Multivariate		Univariate		Multivariate
	OR (95% CI)	p	OR (95% CI)	p	OR (95% CI)	p	OR (95% CI)	p
KS, Kaposi's sarcoma; NSTI, number of sexually transmitted infections; OR, odds ratio; CI, confidence interval.
CD4 count
≥301	1	0.083	1	0.457	1	0.854
201–300	1.89 (0.92–3.86)	0.031	1.48 (0.53, 4.10)	0.519	0.95 (0.52–1.71)	0.117
101–200	2.66 (1.10–6.46)	<0.001	1.30 (0.59, 2.88)	0.223	1.80 (0.86–3.76)	0.113
≤100	2.38 (1.25–4.51)		0.63 (0.30, 1.33)		1.51 (0.91–2.52)
Age	1.04 (1.01–1.07)	0.019			1.02 (0.99–1.05)	0.071
NSTI	1.33 (1.06–1.68)	0.015			1.35 (1.12–1.63)	0.002	1.26 (1.03, 1.54)	0.027
Sexual partners during life
≤50	1	0.001	1	0.005	1	<0.001	1	0.004
50–300	3.68 (1.76–7.65)	0.010	3.03 (1.39, 6.58)	0.014	2.69 (1.58–4.57)	0.005	2.31 (1.31, 4.10)	0.018
301–600	3.30 (0.86–3.76)	<0.001	3.39 (1.29, 8.91)	<0.001	2.64 (1.34–5.20)	<0.001	2.47 (1.17, 5.21)	<0.001
≥601	6.85 (3.04–15.42)		6.44 (2.69, 15.43)		5.73 (3.01–10.64)		5.55 (2.79, 11.04)
Anogenital insertive sex
Never	1	0.365			1	0.071
Less than once a month	1.38 (0.69–2.76)	0.443			1.68 (0.96–2.94)	0.081
Once a month or more	1.25 (0.71–2.23)				1.51 (0.95–2.40)
Anogenital receptive sex
Never	1	0.169			1	0.038
Less than once a month	1.65 (0.81–3.38)	0.811			1.86 (1.04–3.348)	0.401
Once a month or more	1.07 (0.60–1.93)				1.22 (0.77–1.94)
Orogenital insertive sex
Never	1	0.183			1	0.181
Less than once a month	1.96 (0.73–5.30)	0.009			1.64 (0.80–3.37)	0.016
Once a month or more	2.78 (1.30–5.97)				1.94 (1.13–3.32)
Orogenital receptive sex
Never	1	0.472			1	0.018	1	0.058
Less than once a month	1.43 (0.54–3.78)	0.024			2.39 (1.16–4.90)	0.003	2.13 (0.97, 4.65)	0.248
Once a month or more	2.20 (1.11–4.34)				2.21 (1.30–3.77)		1.43 (0.78, 2.61)
Oroanal insertive sex
Never	1	0.007			1	0.090
Less than once a month	2.57 (1.30–5.10)	0.177			1.66 (0.93–2.99)	0.402
Once a month or more	1.58 (0.81–3.10)				1.26 (0.73–2.16)
Oroanal receptive sex
Never	1	0.012			1	0.016
Less than once a month	2.30 (1.20–4.41)	0.186			1.95 (1.13–3.37)	0.131
Once a month or more	1.57 (0.80–3.07)				1.51 (0.88–2.60)

Follow up of Spanish controls was administratively censored at May 1, 2004. Cumulative incidence rates of KS development, as well as cumulative mortality rates, were computed and expressed as rates/100 men. Relative risk (RR) of KS development (or death) and its 95% confidence interval (95% CI), as well as the RR adjusted through the Mantel-Haenszel estimator, were also computed.

Results

Study population

All 483 participants were men and the distribution by route of HIV acquisition reported by physicians was predominantly homosexual (91.3%), followed by intravenous drug use (4.3%), heterosexual (1.2%), transfusion recipients (1.2%) and unknown route (1.9%). For the current study, only those triplets (1 case and 2 matched controls) of men who reported having had sex with other men (MSM) in the questionnaire were included in the statistical analysis (n = 426, 142 cases and 284 controls). Mean age at the time of diagnosis was 38.0 years both for cases and controls, since age was used as a criterion for matching. Regarding country of enrolment, 104 (73%) triplets were recruited in Spain, 17 in Belgium, 9 in Italy, 6 in the UK, and another 6 in Greece. Among Spanish study participants, 227 subjects without KS at study entry (controls) were followed up to assess KS incidence and mortality rates.

Analysis of risk factors for KS

Among the 426 participants, serum specimens were not available for 23 (6 cases and 17 controls). Percentages of HHV-8 seropositivity among cases and controls were 72.06% (98 out of 136) and 39.33% (105 out of 267), respectively. An IFA positive result was highly associated with the presence of KS (OR: 4.82, 95% CI: 2.85–8.14). When risk factors were assessed using the matched case-control design (n = 426), significant associations were found between KS and some of the variables including certain sexual practices in the univariate analysis (Table I). Low CD4 counts, a high number of life-time sexual partners, and orogenital receptive unprotected sex were still significant in the multivariate analysis. In contrast, risk factors were analyzed among HHV-8 positive individuals including all 142 KS cases, who were assumed to be infected with HHV-8 regardless of HHV-8 serostatus, plus 105 controls that were HHV-8 seropositive. The only significant risk factor for disease development was having a low CD4 count, which was still significant when adjusted by age at AIDS diagnosis.

Analysis of risk factors for HHV-8 infection

Seventeen controls were excluded because serum was not available to perform IFA and, therefore, 267 controls without KS were included in subsequent analyses (162 HHV-8 negatives and 105 HHV-8 positives). A significant univariate association was found between HHV-8 infection and all unprotected sex practices but anogenital insertive sex, and risk for HHV-8 infection was significantly higher among men who had had more than 50 sexual life-time partners. However, in the multivariate analysis only the association with multiple sexual partners remained significant (Table II).

This analysis was also performed adding KS cases, who were by definition infected with HHV-8, to the HHV-8 positive group. Thus, the N increased to 409 subjects, including 162 HHV-8 negatives and 247 HHV-8 positives. In this analysis, the number of STI and the number of lifetime sexual partners were still statistically associated with HHV-8 infection in the multivariate model, and so was orogenital receptive sex.

Longitudinal follow-up

To assess KS incidence and the mortality rate, Spanish controls (227 HIV-infected MSM free of KS at study entry) were followed up twice a year until May 2004. Subjects that were lost over the first year after study entry (n = 7) were excluded from the analysis. The rest were followed up until their death or lost in follow up, for a median time of 7 years (range, from 0.4 months to 10.23 years).

Over this period, 16 (7.3%) subjects developed KS, including 9 (11.0%) individuals out of 82 HHV-8 seropositives and 7 (5.1%) out of 138 HHV-8 seronegatives at study entry. Cumulative KS incidence was thus 2.0 and 1.0 per 100 persons-year, respectively. Median time from study entry to KS development was 11.8 months (range, from 1 month to 4.5 years) and, according to the HHV-8 status at study entry, 10.0 and 23.1 months for HHV-8 seropositives and seronegatives, respectively. The RR for KS development among HHV-8 positives was higher than in patients who were seronegative at enrolment but did not reach statistical significance (RR = 2.0, 95% CI: 0.7–5.3). For 171 of the Spanish controls, information was available regarding whether and when they had started HAART. Once this variable was taken into account, among the 167 subjects who were free of KS before starting HAART, KS incidence was 0.8 per 100 persons-year in IFA-positive individuals who had started HAART and 8.0 per 100 persons-year in those IFA-positives who had not (RR = 10.5, 95% CI: 1.1–101.2).

Concerning mortality, 86 subjects (39.1%) died in a median time of 15.6 months (range, from 0.4 months to 8.2 years). The mortality rate and RR of death were higher among subjects who had developed KS than in those who had not (15.8 per 100 persons-year vs. 6.8 per 100 persons-year, RR = 2.3, 95% CI: 1.2–4.5). Moreover, the RR of death increased to 9.6 (95% CI: 3.2–28.1) when computed taking into account whether these patients had started HAART or not. Not having started HAART was a risk factor for death, both for subjects with KS (RR = 6.26; 95% CI: 1.7–23.3) and without (RR = 44.5; 95% CI: 23.1–85.5).

Discussion

The present European multicentre case-control study was designed to assess risk factors for KS among HIV-positive subjects before HHV-8 was first identified. Enrolment criteria for cases and controls were set on presence or absence of KS regardless of HHV-8 infection, but later serology testing revealed that 39% of the controls were actually seropositive. This fact enabled us to look at risk factors for KS development among HHV-8 and HIV-1 coinfected MSM. For this analysis, all KS patients were assumed to be HHV-8 infected since a gold standard for the identification of certainly infected individuals is not available for HHV-8, and viral DNA can be detected by PCR in lesions from virtually all patients.4, 47 We also looked at risk factors for HHV-8 infection. The observed sensitivity of the IFA was somewhat lower than that of other studies that detected HHV-8 antibodies in some 80% of HIV-positive patients with KS using similar serology techniques.7, 9, 28, 48 However, it is necessary to consider that the patients included in this study were in advanced stages of HIV-infection. AIDS patients might present poor antibody responses and even seroreversion,29 causing a low prognostic value of the used serology test in this population. The rather low IFA sensitivity might have resulted in an attenuation of associations.

Risk factors for KS

Study design is of utmost importance to assess risk factors for KS development. Early studies analyzed individuals with and without KS irrespective of HHV-8 serostatus to identify risk factors among specific homosexual activities.49, 50, 51 However, the KS-negative group might have included both HHV-8 positive (who were therefore at risk for KS) and negative subjects, as occurred in our study and, therefore, risk factors for KS development were probably confounded with those actually related to HHV-8 infection. Our study shows that sexual practices originally identified as risk factors for KS were no longer significant when data was reanalyzed taking HHV-8 infection into account. These results suggest that specific sexual behaviors, while related to HHV-8 infection, might not play a direct role in the development of KS.22 In our study, the only significant risk factor for KS was a low CD4 count, in agreement with previous work.11, 16, 26, 43, 52 Among the Spanish study participants, HLA-DRB1 showed a possible role in the development of KS, indicating that progression to disease may depend on host factors controlling the immunoresponse against HHV-8.53

When the original matching design was broken for the analysis of risk factors for HHV-8 infection and for KS development among HHV-8 positive subjects, some confounders may have been introduced. However, both cases and controls were in an advanced stage of HIV infection and also had a similar age. Therefore, we believe that there were no major confounders in the associations detected.

Risk factors for HHV-8 infection

The disproportionate prevalence of HHV-8 infection and KS incidence among homosexual and bisexual HIV-positive men compared to other HIV risk groups has long suggested that HHV-8 can be sexually transmitted.28, 29, 30 We found a strong association between HHV-8 seropositivity and the number of life-time sex partners among MSM with AIDS, which further confirms results obtained in previous studies.9, 10, 27

While HHV-8 infection has been linked to homosexual behaviour,10, 27, 54 association with specific sexual practices has been so far controversial regarding oroanal sex,9, 10, 49, 55, 56 anogenital sex10, 57, 58, 59 and orogenital sex.59,60 The assessment of transmission routes based on cross-sectional risk factor analyses should be interpreted with caution because the moment of infection is unknown and sexual behavior may change over time. While data obtained by studies including subjects with known HHV-8 seroconversion dates is more reliable, only a few such studies are available. Dukers et al.59 found no evidence that anogenital and oroanal insertive sex were related to HHV-8 seroconversion while orogenital receptive and insertive sex were significantly associated.

In our study, several sexual behaviors were significantly associated with HHV-8 infection in the univariate analysis. However, none of them but the number of life-time sexual partners remained significant in the multivariate model. Possible reasons that might have influenced these results are (i) the presence of statistically significant correlation among sexual practices, which causes multicollinearity and (ii) low statistical power limited the detection of associations. Since both presence of antibodies against HHV-8 and presence of KS can be considered surrogate markers of HHV-8 infection (HHV-8 DNA can be detected in lesions from virtually all KS patients), we reanalyzed the data grouping KS cases and HHV-8 serpositive patients to increase the statistical power. Despite the potential biases introduced with this analysis, the number of life-time sexual partners, number of STI as well as orogenital receptive unprotected sex remained significant in the multivariate model.

HHV-8 has been detected intermittently in semen samples from KS patients and homosexual men without KS,32, 33, 34 and in prostate specimens of AIDS-KS patients by PCR.36, 42 HHV-8 is shed in saliva,61, 62 and it has been frequently detected in oral fluids.35, 37, 40, 43 HHV-8 is transmitted though saliva among children in Africa63, 64 and viral shedding in the oral cavity is also quite likely to be involved in transmission among HHV-8 positive men.40

Longitudinal follow-up of Spanish cases

As reported in several studies, 30–50% of HHV-8 and HIV coinfected men develop KS within 5 or 10 years of dual infection.9, 11, 22, 26, 27, 65, 66 Although KS incidence tends to be higher among individuals in advanced stages of AIDS than in recent HIV seroconvertors,26 only 11% of our HHV-8 positive AIDS patients had developed KS at the end of follow up (up to 10 years). Interestingly, 80% of those who did not develop KS were on HAART. As expected, our results demonstrate a significant protective role for HAART both on KS development among HHV-8 seropositives, and on mortality of KS patients. These results are in agreement with studies showing that relatively long periods of HAART result in immunoreconstitution, undetectable HHV-8 DNA, KS regression and a better clinical outcome in patients with KS.67, 68

Follow up also revealed 7 (5%) incident KS cases among the 138 controls that were HHV-8 seronegative at enrolment. This implies that HHV-8 seroconversion as well as disease development could have occurred within the study period (median time from study entry to KS development for these patients was 23 months, range 6.4–54 months). This is in agreement with previous results showing that seroconversion may occur a few months before the onset of KS.29 In fact, KS develops more rapidly (i.e. 1–4 years) in HIV positive subjects who later seroconvert to HHV-8.11, 20, 22 Alternatively, these subjects could have had false negative HHV-8 serology results reflecting a low sensitivity of the IFA or a poor antibody response in immunocompromised patients (CD4 count ranged from 1 to 1341 cells/μl, median 63 cells/μl). This fact could have caused an underestimation of the RR for KS development among HHV-8 positive controls at study entry, which was only 2-fold more likely to develop KS during follow up than seronegatives. This result might have been also due to a confounding effect played by the progressive availability of HAART, decreasing KS incidence among HHV-8 positive patients.

Accurately determining the KS incubation period was not possible in our study because HHV-8 seroconversion dates were unknown. However, according to our results, this period may vary widely and be longer than 3–10 years for those AIDS patients that were HHV-8 seropositive at enrolment and had no sign of KS at the end of follow up. These results are in agreement with previously published reports. A study performed among HIV-positive MSM without AIDS estimated KS incubation time in 4.2 years (1.4–7.6 years) after HHV-8 seroconversion.26 In studies in which HHV-8 seroconversion date was unknown for some or all participants and assumed to be the enrolment date, this period varied from 1.3 to over 10 years, and the median time was around 5 years.9, 11, 27

In conclusion, the association of HHV-8 seropositivity with the number of sex partners found in our study confirms the relevance of sexual transmission among HIV-positive MSM. Accumulating evidence suggests that HHV-8 is intermittently shed in semen and saliva of infected individuals40, 42 and, therefore, this virus may be transmitted through multiple routes like other herpesviruses transmitted both by salivary and sexual contact. The main risk factors for KS development among HHV-8 seropositive MSM with AIDS were a low CD4 count and not being on antiretroviral therapy; our results demonstrate a significant protective role for HAART both on KS development among HHV-8 seropositives, and on mortality of KS patients. Finally, KS incubation time in MSM with AIDS may range from a few months to more than 9 years. Although more conclusive data from cohort studies are definitively needed to better define specific transmission mechanisms for HHV-8, our results contribute to explain why KS incidence is higher among MSM, as well as the decreasing KS incidence trend observed in countries with universal access to HAART.

Acknowledgements

We thank to all participating centres. We also acknowledge the support given by the CIRIT and the Health Department (Generalitat de Catalunya, Barcelona, Spain), the research networks RCESP and RIS (FIS), the UK Medical Research Council, and St. Stephen's Aids Trust (London, UK). We also thank all study participants.