Geography of AIDS-associated Kaposi's sarcoma in Europe

Background: Classical Kaposi's sarcoma (KS) is about four times more common in southern Europeans than in northern Europeans.

Objective: To describe the epidemiology of AIDS-associated KS (AIDS-KS) in Europe and to determine whether it occurs with increased frequency in southern Europe.

Methods: Analysis of the 'European non-aggregate AIDS data set', as of September 1995. Countries with a cumulative total of ≥ 50 KS cases as the presenting manifestation of AIDS were included. Homosexual men were excluded from south versus non-south comparisons because of possible confounding effects due to their route of HIV transmission.

Results: KS was the presenting manifestation of AIDS for 13.3% (16 367 out of 122 679) of men and 2% (491 out of 24 826) of women. In all countries, the risk for KS was higher in individuals who acquired HIV infection via sexual rather than parenteral transmission. Among AIDS patients, there is little difference by sex in the risk of KS in injecting drug users (IDU) or transfusion recipients. The percentage with KS increased with age among homosexual and bisexual men, from 10% in the age group 15-19 years to 23% in the age group 30-39 years. In all countries, the percentage with KS declined over time. The risk of KS was not significantly higher in southern Europe. The percentage with KS in southern Europe was slightly lower than in northern Europe (P > 0.1) in male IDU (1.8% versus 2.1%), and only slightly higher (P > 0.1) in female IDU (1.5% versus 1.1%), in male transfusion recipients (3.5% versus 3.0%), in female transfusion recipients (2.4% versus 2.3%), and in both heterosexual men (7.5% versus 6.2%) and women (2.0% versus 1.6%) excluding those originating from countries where heterosexual HIV transmission is frequent.

Conclusions: The strong geographic predilection described for classical KS in southern Europe was not seen for AIDS-KS. If KS is caused by a viral infection in an immunodeficient host, our findings suggest the geographical variations in classical KS are not due to variation in prevalence of the causative virus but may be due to geographical variations in the prevalence of a form of mild immunodeficiency.

Introduction

Kaposi's sarcoma (KS) was first described in Europe, but remained an uncommon tumor there until the emergence of the AIDS epidemic. The association of KS with AIDS and other causes of immunosuppression [^1-8] suggests that development of KS lesions requires an immunocompromised host. Epidemiologic studies of AIDS-associated KS (AIDS-KS) suggested that KS was caused by a sexually transmitted infectious agent [^9-11]. Thus KS appears to be caused by an infection in an immunocompromised host. Among many tumorigenic viruses studied, human herpesvirus 8 (HHV-8) has been proposed to be the etiologic cofactor for KS [^12,13]. HHV-8 has been detected in nearly all biopsy specimens from patients with both AIDS-KS and other forms of KS [^12-15]. However, results of studies of the prevalence of HHV-8 in the general population are inconsistent, with some studies finding a low prevalence [^16-20] and others a very high prevalence of infection [^21,22]. Therefore, the exact role that HHV-8 actually plays in the development of KS lesions remains unclear, perhaps because the current experimental HHV-8 tests are insufficiently sensitive or specific [^23,24].

Other pieces of the KS puzzle are missing. Before AIDS, classical KS was reported with fourfold higher frequency in the Mediterranean and southern European countries than in other non-African countries, and among those who lived in other countries but who were of southern European or Mediterranean origin [^25-32]. Over time, explanations for the geographic variations, including genetic, environmental, and hormonal factors were considered and discounted [^25-27]. If the development of KS requires a virus and an immunocompromised host, it may be argued that geographical variations in KS might be due to variations in prevalence of either the virus or the immunodeficiency in a given population. The AIDS epidemic provides an opportunity to shed light on this key epidemiologic question because all persons with AIDS are immuno-compromised. Thus if the geographic distribution of classical KS was due to differences in the prevalence of the virus, then AIDS-KS would be expected to have similar geographic distribution.

The geographic distribution of AIDS-KS is potentially influenced by other factors. AIDS-KS is 10-20-fold more common among homosexual men than in other HIV-transmission groups [^9-11]. KS in homosexual men with AIDS has been associated with their sexual practices [^33,35] and whether or not their sex partners were from places where AIDS-KS was common [³⁴]. This suggests that the agent causing KS is prevalent among homosexual men. Among other persons with AIDS, there are few factors associated with KS except having sex with bisexual men or originating from a country where KS was prevalent before AIDS [^9,11]. Therefore, to generate geographic patterns of KS in AIDS patients that reflect the pattern of the infectious agent for KS in Europe prior to the AIDS epidemic, the most appropriate group to study would be AIDS patients who are not homosexual and who do not originate from countries where KS was prevalent before AIDS.

In this paper, we describe the epidemiology of AIDS-KS in Europe and assess whether AIDS-KS is more common in the south as was reported for classical KS.

Methods

We analyzed the September 1995 public version of the 'European non-aggregate AIDS data set'. The European Centre for Epidemiological Monitoring of AIDS gathers data on reported AIDS cases from 44 of the 50 countries in the World Health Organization European region. The public-use version of the 'European non-aggregate AIDS data set' contains demographic data, HIV transmission risk group, and initial AIDS-indicator disease for AIDS cases (meeting the 1993 European AIDS surveillance case definition) [³⁶] recorded by national surveillance systems of 34 European countries in accordance with a standard core of epidemiological information. For the heterosexual risk group, but not for other transmission groups, cases are further classified into seven mutually exclusive subcategories, including whether the person originated from a country where heterosexual transmission of HIV is frequent.

In our analysis we included cases reported from the 13 countries that each reported at least 50 patients for whom KS was the initial AIDS indicator disease. We determined risk factors for KS. Stratifying by risk group, we looked at geographic variation in the percentage of AIDS patients with KS. Time trend analysis was done after grouping cases diagnosed before 1985 because of small numbers in the early years of the AIDS epidemic. Similarly, cases diagnosed in persons aged less than 15 years also were grouped together for analyses stratified by age.

We examined the percentage of AIDS patients with KS by country for all HIV transmission risk groups except for homosexual/bisexual men and persons who originated from countries where heterosexual HIV transmission is frequent. To determine the percentage of AIDS patients with KS by geographic region, the 13 countries were divided into two groups: south (Greece, France, Italy, Portugal, Spain) and non-south (Austria, Belgium, Denmark, Germany, The Netherlands, Sweden, Switzerland, UK). We then calculated percentages of AIDS patients with KS in these regions after stratifying by sex and risk groups. Heterosexuals not originating from countries where heterosexual HIV transmission is frequent were analyzed separately, except for Portugal and Spain, where data on country of origin was not available.

Results

A total of 154 307 AIDS cases were reported from 34 countries to the European Centre for Epidemiological Monitoring of AIDS, of whom 17 104 (9.0%) had a diagnosis of KS. After the exclusion of data from 21 countries with totals of fewer than 50 patients with KS, there were 147 516 patients with AIDS, and KS was the presenting manifestation of AIDS for 13.3% (16 367 out of 122 679) of men, for 2.0% (491 out of 24 826) of women, and for 0.3% (eight out of 2585) of children aged less than 15 years. Most of the AIDS cases were reported from the five southern European countries (men, 71.3%; women, 80.3%).

Risk factors

As in other reports KS was more common in homosexual/bisexual men (25.6%) than among persons with AIDS in other transmission categories. Other risk groups in which the percentage of KS is high include homosexual/bisexual men who are also injecting drug users (IDU; 12.6%) and persons who acquired HIV through heterosexual contact (5.6%) (Table 1).

Table 1

Among those who acquired HIV through nonsexual routes, the proportions with KS were similar among men and women. Among transfusion recipients, the percentage with KS was similar for men (3.4%, 56 out of 1636) and women (2.4%, 34 out of 1431) (P = 0.1). Among IDUs the percentage with KS was also similar for men (1.8%, 799 out of 44 705) and women (1.4%, 176 out of 12 497), although the difference was statistically significant (P = 0.004).

Among all heterosexuals the percentage with KS differed significantly between men (7.5%) and women (3.1%) which was true for all transmission categories except for those persons originating from a country in which heterosexual HIV transmission is frequent and in whom the risk of KS was similar in men (8.1%) and women (6.8%) (Table 2). Among heterosexual patients the highest risk of KS for both men and women was also among those persons originating from a country where heterosexual HIV transmission is frequent. Among all other heterosexuals the risk of KS was 7.2% (548 out of 7599) for men, and 1.9% (120 out of 6217) for women. Among heterosexual women the risk for KS was high also in those who had sex with bisexual men, 6.3% (24 out of 382). Among women the risk of KS varied greatly among heterosexual risk categories (range 1.3-6.8%) whereas among men there was not much difference in the risk of KS among heterosexual risk categories (range 5.4-8.15%) (Table 2).

Table 2

Age

Age-specific changes in the percentage of AIDS patients with KS was observed mainly in homosexual/ bisexual men (Fig. 1). The percentage of patients with KS among homosexual/bisexual men increased from 10.3% in those aged 15-19 years to 27.0% in those aged 35-39 years and declined to 20.1% among those aged 60 years or over.

Fig. 1

Trends over time

The percentage of AIDS patients diagnosed with KS declined every year in persons who acquired HIV infection by sexual contact rather than via parenteral transmission (Fig. 2). Because of small numbers we could not assess trends in AIDS-KS in all non-sexual risk groups. The absolute number of patients with KS and patients with AIDS increased consistently over time in all risk groups, except in homosexual/bisexual men in which the absolute number of patients with KS and patients with AIDS declined from 1992 onwards.

Fig. 2

In homosexual/bisexual men the percentage with KS decreased with time, from 37.3% in 1985 to 21.5% in 1994. This decline was observed in all countries. The percentage with KS also decreased between 1985 and 1994 for heterosexuals (from 17.0% to 6.0%), women (from 7.6% to 1.8%), and IDUs (from 6.8% to 1.3%). Country-specific yearly rates were not calculated for IDUs, heterosexuals, or women because of the small number of cases.

Geographic distribution

The percentage of AIDS patients with KS was not significantly higher in the south than in the non-south. In men the percentage of AIDS patients with KS was higher in the non-south (3.9%) than in the south (3.1%). However, in women (excluding heterosexuals from countries where heterosexual transmission of HIV is frequent) the percentage with KS was slightly higher in the south (1.6%) than in the non-south (1.3%). When stratified by risk and sex, there were no statistically significant differences between the south and the non-south in the percentages of AIDS patients with KS (Table 3). In five of the six categories, the percentage with KS was slightly higher in the south than in the non-south (P > 0.1): in female IDUs [relative risk (RR), 1.4], in male transfusion recipients (RR, 1.3), and in both heterosexual men (RR, 1.2) and women (RR, 1.3) after excluding those originating from countries where heterosexual HIV transmission is frequent. The percentage with KS was slightly lower in the south than in the non-south (P > 0.1) in male IDUs and female transfusion recipients. The number of patients with KS was too small to make a meaningful interpretation of the distribution of KS by country and risk (data not shown).

Table 3

Discussion

Our analysis of 16 859 AIDS-associated Kaposi's sarcoma (AIDS-KS) cases from 13 countries did not find the north-south difference in prevalence of Kaposi's sarcoma (KS) in magnitudes that was reported for classical KS. In some subgroups of persons with AIDS, southern Europeans were slightly more likely to have KS (up to 1.4 times) than non-southern Europeans, but this was not consistent and not statistically significant.

Studies demonstrating Mediterranean and southern European greater prevalence of classical KS have been hampered by the rarity of this neoplasm. Though most of the earlier evidence supporting this hypothesis has been anecdotal [^25-27], recent re-examination of cancer registries to obtain population-based incidence rates for classical KS consistently report a geographical bias. An Italian study based on 53 cases [²⁸] showed the incidence of classical KS to be two-to threefold higher than in the USA [²⁹] or Sweden [⁴] and more than tenfold higher than in England and Wales [³⁰]. In England and Wales (68 cases), an approximately ninefold increased incidence rate of classical KS was found in immigrants from Mediterranean Europe [³⁰]. Similarly, in Australia (26 cases) the highest incidence of classical KS was in those born in the Middle East or in southern or eastern Europe [³¹]. Such trends were also observed in the USA. An analysis of 244 cases from the Los Angeles Cancer Surveillance Program showed a threefold higher risk of classical KS in men born in southern Europe compared with those born in the USA [³²].

There is overwhelming evidence that KS is expressed during periods of immunodeficiency [^1-8]. The development of both KS and non-Hodgkin's lymphoma in persons with defective cell-mediated immunity is well documented [^1,2,7]. Transplant recipients are at high risk for KS because of the immunosuppressive drugs that are used to decrease graft rejection [^1-3]. Progression of KS lesions seems to be modulated by the degree of immunodeficiency whether iatrogenic or acquired. For example, patients with AIDS-KS [^5,6], and immunosuppressed transplant recipients with KS [⁷] often show regression of their lesions when immunosuppressive treatment is decreased.

Since KS lesions appear to require both immunodeficiency and a causative infectious agent, we considered two possible mechanisms that may give rise to a higher frequency of KS in a population: variations in the level of immunodeficiency and variations in the prevalence of the infectious agent. If geographical variation in the prevalence of the causative agent was responsible for geographic variation in classical KS, then the percentage of AIDS patients with KS might be highest in countries with high prevalence of classical KS. Because the infectious agent for KS is believed to be sexually transmitted, KS in persons who acquired HIV through non-sexual modes of transmission should best reflect the prevalence of the infectious agent for KS in the general population. This assertion is supported by the lack of age-related changes in the percentage of IDUs with KS. In contrast, age-related changes in the percentage of homosexual/bisexual men with KS (Fig. 1) suggest a recent epidemic of infection among men who have sex with men. We excluded homosexual men from our analysis because variations in the risk of KS from country to country may be due to variations in sexual practices [³³], region of origin of sexual partners [³⁴], and varying levels of changes in the sexual practices in response to the AIDS epidemic [³⁵].

The fact that the geographic pattern for classical KS is not seen in similar magnitude for AIDS-KS, suggests that the prevalence of the infectious agent may be relatively uniform in European countries. The geographic variation in classical KS may not be due to a higher prevalence of the infectious agent in southern Europe. This suggests there may be geographic variability in a mild form of immunodeficiency that facilitates the development of KS lesions. If this is true one might expect to find geographic variation in certain types of lymphoma prior to the AIDS epidemic. Current HHV-8 serological data is inconclusive to provide the geographic distribution of HHV-8. Large serosurveys of the general population in Europe have not been done. Studies to date have shown wide variations in HHV-8 seroprevalence in healthy adults with some suggesting that the infection is very rare [^16,17], and others suggesting that it is quite common [^21,22]. This variability is partly due to differences in sensitivity and specificity of these experimental tests [²³].

The risk factors for KS in this study are broadly consistent with earlier reports on AIDS-KS [^9-11]. Higher prevalence of AIDS-KS among persons who acquire HIV infection by sexual contact, and the increased risk of AIDS-KS in persons aged 20-50 years, which is most significant among homosexual/bisexual men and heterosexual men (Fig. 1) adds to the evidence suggesting sexual contact as the major mechanism of transmission of the infectious agent. In heterosexuals, the highest risk of KS was among those who originate from countries where heterosexual transmission of HIV is frequent, and among women who have sex with bisexual men. The differences in the risk of KS among heterosexual risk categories between men and women (Table 2) may be explained by the fact that some heterosexual men may have had other risk factors including sex with other men that may not have been reported and that risk categories in this data are not mutually exclusive. There is very little difference by sex in the percentage of KS in IDUs or transfusion recipients and among heterosexuals originating from countries where heterosexual HIV transmission is frequent (Tables 1 and 2), or by age among IDUs (Fig. 1). This implies that, in situations when the transmission of KS-causing agents and the degree of immunodeficiency is equal, the risk of KS also tends to be equal. In all countries and in persons who acquire HIV by sexual contact, the percentage of KS declined consistently over time despite an increase in the total number of AIDS cases (Fig. 2). The revised European AIDS surveillance case definition [³⁷] which was progressively implemented in 1993-1994 is not likely to have affected the decline observed in our analysis. The early presentation of KS in HIV infection and the documentation of a similar decline after accounting for the broadening of AIDS definition criteria [⁹] suggests that the decline seen among persons with sexual risk factors is real.

Our study has some limitations. It is possible that the validity of transmission category classification may vary slightly from country to country based on the degree of social acceptability of the risk category. If so, this could obscure differences in the risk of KS. Similarly the hierarchical method of risk categorisation, which also is used in this data set, is intended to classify a patient by the most probable route of transmission, as patients belonging to more than one transmission category are counted only once. Lack of information on country of origin of the patients or of the sex partners of the patients, except for heterosexuals, is a major limitation in country-specific analysis of KS. In countries with colonial links with Africa or the Caribbean, more patients may be of foreign origin. Although the differences in diagnostic or reporting practices might be an explanation for some variation between countries, it is not likely that such substantial differences would be accounted for by biases of this nature. Finally, studies based on AIDS surveillance data tend to underestimate the true proportion of AIDS patients with KS. In fact, KS diagnosed after notification of a case of AIDS is generally not recorded in AIDS registries, including the data set used here. The importance of such bias, however, is limited by the observation that KS is generally one of the earliest clinical manifestations of AIDS. Finally, it is possible that the geographic distribution of the KS infectious agent has changed through migration or other means since the papers on classical KS were written. The changes in the percentage of AIDS patients with KS by age (Fig. 1) suggests an increase in KS among those who acquired HIV through sexual contact, but not among others.

In conclusion, in this study population, which included only persons with immunodeficiency, we did not find the difference in geographic distribution for AIDS-KS in the magnitude that was reported for classical KS. This suggests that geographic difference in classical KS in Europe may not be due to geographic variation in the prevalence of the infectious agent for KS. There may have been variations in some requisite mild immunodeficiency that allows expression of KS lesions.

Acknowledgements

The European non-aggregate AIDS data set (ENAADS), release AIDSyymm. DAT^*, was prepared by the European Center for the Epidemiological Monitoring of AIDS (Saint Maurice, France). Compilation of this data file was made possible by the continuing participation of clinical doctors in mandatory and voluntary national AIDS reporting systems.

We thank W. Killean for assistance with the data analysis.