July 25, 1994
Kaposi's Sarcoma (KS) has an epidemiology unlike any other opportunistic infection (OI) or AIDS-related malignancy. As will be discussed later, the epidemiology of KS may provide some clues to the cause or one of the co-factors that leads to this disease. KS has a distribution that varies according to the method of HIV infection, sex, ethnicity, age, and residence of sexual partners.
Numerous epidemiological studies exist that provide additional pieces to this puzzle. However, while many conclusions can be strongly supported with additional studies, many issues remain controversial.
Kaposi's Sarcoma continues to be primarily a disease of homosexual/bisexual males. As HIV/AIDS has started affecting other populations, KS continues to be a greater risk among this population (Beral 1990; Selick 1987; Elford 1993; Rutherford 1990; Haverkos 1993). In the United States, KS is rare among those infected with HIV through heterosexual contact and even more rare among IVDUs, transfusion recipients, and hemophiliacs (Beral 1990; Rabkin 1990, 1992). Those who acquired HIV from heterosexual contact and had KS were more likely to be born in Haiti, other Caribbean countries, or Mexico/Central America (Beral 1990). One cohort study of hemophiliacs had only one case of KS, and that patient was also a homosexual living in New York City (Rabkin). Of extreme interest is the increase of KS among homosexual males who are not infected with HIV (Friedman-Kien 1990).
Geographic distribution of KS cases varies greatly in the US as well as Europe. The Multicenter AIDS Cohort Study (MACS) shows that patients living on or having sex partners from the West Coast, particularly in Los Angeles, a greater risk of KS (Armenian 1993). Out of 316 cases of KS, 52.9% were from Los Angeles, 19.9% from Chicago, 17.7% from Baltimore, and only 9.5 % from Pittsburgh. Rabkin has shown that the risk of KS is higher for those living in New York as opposed to Washington DC (Rabkin 1990). Beral et al. (Beral 1990) demonstrated that of the reported KS cases in the US, 3% were from Kansas, 6% from Iowa, 30% from California, and 31% from New York. This geographical distribution was found among all transmission groups with KS. Two Canadian studies support this hypothesis. A Canadian Surveillance study reported that KS was more common in cities that were considered primary epidemic centers such as Vancouver, Toronto, and Montreal (Schechter 1990). The Vancouver Lymphadenopathy- AIDS Group demonstrated that homosexual men were at a greater risk of KS if their sex partners were from Los Angeles, San Francisco, or New York (Archibald 1990). In the UK, Beral et al. have shown that patients were at greater risk of KS if they had sex partners from the US or Africa as opposed to the UK.
Women are at a much lower risk of KS than men. Pathogenesis studies suggest that this may be due to hormonal differences (see pathogenesis). One study has shown that women who were infected with HIV from bisexual males as opposed to being an IVDU were at a 4 times greater risk of KS (Beral 1990). An update report supports this finding (Haverkos 1993). Other studies have not shown this strong association (Elford 1993; Lassoued 1991; Benedetti 1991, Serraino 1992).
Age also plays a role. KS is extremely rare in children and is more of a risk for those older than 15 years (Beral 1990). Beral has shown that it is common for children with KS to be born to Haitian women. An Australian surveillance study reported that the youngest case of KS was a 17 year-old homosexual male. They reported no children with KS. It has been reported that KS is more common among homosexuals between 25-44 years (Beral 1990) . One Canadian study suggests that KS was more common in the 1945- 54 birth cohort of homosexual males (Schechter 1991).
Some studies have reported different KS risk among various racial groups. One surveillance study reports that the percentage of KS cases among whites and Hispanics was approximately twice that of blacks. However, this difference among racial groups was not consistent among the different HIV transmission categories (Beral 1990). In San Francisco, one study reports that KS was more common in White than Blacks or Hispanics (Reynolds 1993) while other studies of homosexual men report no racial differences (Lifson 1990; Haverkos 1993).
In Europe, epidemiological studies show similar distributions among people with HIV (Casabona 1991; Couturier 1991). KS appear to be more common in homosexual/bisexual males. However, one study showed that in Italy where most HIV infection is due to IVDUs, there is a greater proportion of heterosexuals with KS (Serraino 1992). One study showed geographic variances within Italy while another study showed no geographic differences within all of Europe. The Italian study mentions that it has a lower overall prevalence of KS relative to the United States. It was hypothesized that this difference may be due to a higher proportion of AIDS cases among IVDUs in Italy (70%) relative to the United States which has a higher proportion of homosexuals/bisexuals (Serraino 1992).
In summary, in the United States, KS is significantly more common among homosexual/bisexual males relative to other HIV transmission groups; is more prevalent in New York, Los Angeles, and San Francisco, areas considered the initial foci of the HIV/AIDS epidemic; rare among women unless they were infected with HIV from a bisexual male; extremely rare among children (less than 15 years) unless they were born to Haitian mothers; and conflicting data exists whether KS has a racial distribution. European and Australian studies show similar patterns in their respective countries. However, their studies have not collected enough data to demonstrate that women are at a greater risk of KS if they were infected with HIV from a bisexual male.
Many epidemiologists hypothesize that the reason for the unusual distribution of KS among people with HIV/AIDS is that it is caused by sexually-transmitted agent (not HIV). This theory received new support after several known cases of HIV-negative homosexual men were reported to have what is now considered epidemic KS (Kitchen 1990; Afrasiabi 1986, Friedman-Kien 1990; Lowdell 1989; Zucker-Franklin 1993; Castro 1992).
There are many other arguments in favor of KS being caused by an unknown sexually transmitted agent. Beral et al. (1990) have outlined these arguments. The high concentration of cases among homosexual men in the age group 25- 44 parallels the current HIV/AIDS epidemic. Homosexual males are at a greater risk of KS than bisexual males. KS is extremely rare among children. The majority of children with KS were born to mothers with classic KS is endemic. And the two most compelling arguments that KS may be sexually transmitted are the geographic distribution and increased risk for women infected with HIV from bisexual males. It has been shown that KS is more common among those who either live, travel to , and have sex with in someone who lives in Los Angeles, San Francisco, or New York (Archibald 1990; Beral 1990). An increased risk of KS among women infected from bisexual males is an even stronger argument. Why is KS more common in this group?
Studies of homosexual/bisexual males have been conducted to determine the etiology and possible mode of transmission of the "KS agent." No etiologic agent has been discovered nor has there been any conclusive evidence on the possible mode of transmission. However, some interesting data has been collected that may provide us with some clues to the cause of KS. Some studies have suggested that KS may be associated with fecal contact through rimming, fisting, and anal- insertive sex (Beral 1992, Jacobson 1990, Darrow 1992, Archibald 1992) Other studies have not found an association between these sexual activities and an increased risk of KS (Lifson 1990, Elford 1993). More recent studies have suggested a combination of factors that suggest a greater risk of KS may be associated with a previous history of sexually transmitted disease and a greater number of sexual partners (Armenian 1993; Abrams 1990). In addition, reports of HIV-negative homosexual males with epidemic KS report similar findings (Friedman-Kien 1990; Zucker-Franklin 1993; Lowdell 1989).
The use of amyl nitrate (poppers), once considered one of the possible causes of AIDS, has also been implicated as a possible cause of epidemic KS (Haverkos 1985; 1990; Archilbald 1992). This hypothesis has not been supported in other studies ( Lifson 1990; Beral 1992). After 57 of the original 87 men interviewed for the Haverkos study were reevaluated, it was shown that KS was more closely associated with rimming and that there was no significant association with the use of poppers (Darrow 1992).
When analyzing the data involved in these epidemiological studies, one needs to keep in mind that several variables are usually being analyzed through some form of regression analysis. It is impossible to analyze these data to get a pure picture of exactly what variable or variables are involved. While using various statistical tools to control for confounding among variables may help to provide a clearer picture, isolating the effects of each variable independent of others is difficult. The lesson we may have learned from the various studies looking at the use of poppers is that there probably is not one variable associated with KS. The use of poppers may lead to particular behaviors that may be more associated with KS. In addition, it has been suggested that poppers may provide some biological mechanism that may facilitate the transmission of the possible KS agent (Archibald 1992).
It may be that there is no ONE agent that is associated with KS but a variety of agents or interactions. More specifically, KS may not be caused by a sexually transmitted agent but by the hyperactivation of the immune system. While no one agent has yet to be associated with KS, one theme has been very strong; in homosexual men, KS is typically associated with a greater number of sexual partners, and a greater number of sexually transmitted diseases, which both lead to overactivation of an already overburdened immune system (Abrams 1990). KS may not be the direct result of one particular agent but the result of hyperactivation of the immune system (Robert C. Gallo, personal communication).
The two main issues that remain are:
Many researchers, oncologists, dermatologists, and primary care physicians, have reported that they are seeing much less KS than they did at the beginning of the AIDS epidemic. Several studies support the fact that the proportion of KS cases relative to other AIDS defining diseases has significantly decreased since 1981 in the United States as well as other countries (Selik 1987; Haverkos 1993; Beral 1990; Lifson 1990; Rutherford 1990; Elford 1993; Serraino 1992; Casabona 1991; Schechter 1991). In a cohort of 1,341 HIV-positive homosexual/bisexual men in San Francisco, Lifson et al. reported a decline of KS cases from 79% in 1981 to 25 % in 1989. Recent reports show that the percent of KS cases as an AIDS defining illness was as low as 9.9% in 1992 (Haverkos 1993). In most of these studies, it can be argued that the decline in cases of KS may be an artifact due to the method of data collection. Surveillance studies only collect initial AIDS defining diagnoses. As the AIDS definition has changed over time to allow for the reporting of other opportunistic infections and malignancies, the proportion of KS cases would naturally decrease over time.
In an attempt to compensate for the inability of the AIDS registries to collect all cases of KS, studies have been done to link AIDS registries with the cancer registries to determine the completeness of data collection within these registries. Reynolds et al. (Reynolds 1990) reported that only 72.3% of 1,330 reported KS cases appeared in both the San Francisco AIDS Registry and the California Tumor Registry between 1980-1986. Differences may be attributed to the fact that the AIDS registry required microscopic confirmation of KS while the CTR did not. Other differences noted were that private physicians' offices were more likely to report to the AIDS registry while hospitals and clinics were more likely to report to the CTR and the chances of only being reported to one or the other registry as opposed to both registries increased over time. This may be a sign that more private physicians or primary care providers were treating KS cases. Both registries showed a parallel increase in cases of KS during this time period. However, when the data from both registries were linked, there was an ever greater increase of KS cases. Reynolds et al. stated that these data may suggest an increase in the incidence of KS during this time period.
While not as sophisticated as the previous approach, the National Cancer Institute has also linked AIDS and cancer registries in San Francisco, the Bay Area exclusive of San Francisco, Los Angeles, San Diego, Orange County, Sacramento (which provided data on all other California counties), Florida, Metropolitan Atlanta, and New Jersey, to determine the completeness of KS case ascertainment. Cases were matched on social security number, name, and date of birth. When data collection was completed in November 1992, 10,350 cases of KS were reported: 6,987 were reported to both registries, 1,935 were reported only to the AIDS registry, and 1,428 were reported only to the cancer registry. Among those reported to both registries, 1,209 (17%) had KS reported only to the cancer registry and other AIDS-defining diseases reported to the AIDS registry. Cote et al. (Cote: unpublished, 1993) reported that AIDS and cancer registries were 87% and 81% complete, respectively, for KS case ascertainment among people with AIDS. AIDS registry completeness was higher for men than women, and for whites than for non-whites. While there is some potential for both under and overestimating registry completeness through their method, the authors conclude that this is a successful and inexpensive method of providing more complete AIDS/KS case collection. Reynolds and Cote acknowledge, however, that some cases are not been reported to either registry.
Because AIDS surveillance data mostly collects the primary AIDS defining illnesses, a secondary diagnosis of KS would not be reported. In order to correct for this artifact, prospective cohort studies have attempted to analyze the incidence of KS as a primary or subsequent diagnosis among fixed populations of HIV infected individuals (Jacobson 1990; Archibald 1990; Munoz 1993; Reynolds 1993). In an analysis of the MACS population between 9/84 and 9/88, Jacobson et al. reported no systematic trend in the incidence of KS. In a population-based study in San Francisco, Reynolds et al. reported little or no change in KS incidence between 1980-1987. In another analysis of the MACS population between 1985-1991, Munoz et al. found an increase in the incidence of KS, which they attributed to progressive immunosuppression in this population. After adjusting for CD4 count, however, this analysis indicated a downward trend in the incidence of KS. The Vancouver Lymphadenopathy-AIDS study also reports a decreasing incidence of KS. However, similar to Lifson et al., Archibald et al. used the proportion of KS cases relative to opportunistic infections to describe the decreasing incidence of KS in this cohort.
Several hypotheses attempt to explain the decrease in the proportion of KS cases and the possible decrease in incidence of KS among people with HIV/AIDS:
Lack of reporting (Beral 1990; Jacobson 1990). Reporting of an AIDS diagnosis only records the index disease . If a patient later gets KS, this may not be reported.
Changes in behavior which include safer sex techniques as well as a decrease in number of sexual partners among homosexual males (Beral 1990; Jacobson 1990; Elford 1993). If KS is associated with a particular agent, a combination of agents, or the hyperactivation of the immune system associated with chronic enteric infections due to particular sexual activities, then safer sex techniques and/or decreasing the number of sexual partners may have an impact on the risk of KS among people with HIV.
While the proportion of KS cases relative to other AIDS defining illnesses and the true incidence of KS among people with HIV are debatable issues, the bottom line is that the prevalence of KS (the absolute number of KS cases) continues to rise as the number of AIDS cases continues to rise and people with HIV/AIDS live longer through the use of improved OI prophylaxis and treatments. Between 1981 and 1983, when KS accounted for 32.7% of all AIDS diagnosis, the actual number of KS patients was only 957. In 1992 when KS accounted for only 9.9% of the reported AIDS cases, the absolute number of cases was 4, 659 (Haverkos 1993). Due to underestimation in case ascertainment, both of these numbers are likely to be less than the true number of cases.
In a retrospective analysis of patient records of primarily homosexual/bisexual males at a hospital in the United Kingdom, Peters et al. (Peters 1991) have shown that while the reporting of KS as an index diagnosis has decreased from 30% to 20% from 1984-1989, the prevalence of KS has remained constant at around 35%. In addition, Peters et al. reported that while the deaths due to PCP have decreased from 46% in 1986 to 3% in 1989, deaths attributed to KS have increased from 14% to 32% between 1984 and 1989. The authors stated that in 1989, KS was the most common cause of death in this cohort. In an analysis of the MACS population, between 1984 and 1992, Hoover et al. (Hoover 1993) supported this data by showing that overall, 37.4% of AIDS cases in this cohort were diagnosed with KS prior to death. The MACS population of homosexual/bisexual males is similar to that of the Peters study.
What we have seen is controversy over whether or not the incidence of KS is declining. The reality is, while the incidence may or may not be decreasing, the prevalence of KS continues to remain high if not rising (Peters 1991; Hoover 1993). In addition, as people with HIV/AIDS are living longer, their risk of ever getting KS increases. As people become more immunosuppressed their presentation and prognosis changes. Patients are now developing extensive visceral KS and dying because of systemic disease (Peters 1991).
Hoover et al. stated, "Many believe that Kaposi's Sarcoma is more localized and benign at higher CD4+ levels but becomes more invasive as T-cell function declines." This was demonstrated by the fact that in the MACS cohort between 1984 and 1992, the diagnosis of KS subsequent to an AIDS diagnosis almost doubled the hazard for death (Hoover 1993). Because 36% of all KS cases were reported as subsequent diagnoses, the author concluded that gay men presenting with an AIDS condition other than KS are at a significant risk for KS. It was also noted that a greater length of time from a non-KS AIDS-defining diagnosis to death was associated with a greater probability for later KS. While much more extensive than the Peters report, this data supports the increased risk and poorer prognosis of KS as people with HIV/AIDS live longer. Other researchers have seen similar patterns in their patient populations (Ronald Mitsuyasu, personal communication; Alexandra Levine, personal communication).
Autopsies performed on homosexual men with KS in the MACS study support previous reports that visceral KS may be going undetected (Ndimbie 1994). Of the 158 men autopsied, 46 had KS: Baltimore 12 (25.5%), Chicago 4 (28%), Pittsburgh 5 (11.4%), and Los Angeles with 24 (49%) . Of the 46 case of KS, 8 (17%) had visceral, 10 (22%) had cutaneous, and 28 (61%) has cutaneous and visceral. 36 of the 46 cases had a visceral component. 4 (40%) of the 10 cases with cutaneous KS were reported as having KS as their AIDS defining disease. 19 (68%) of the 28 cases of cutaneous and visceral KS were reported as having KS as their AIDS defining disease. Only 1 (12%) of the 8 cases of visceral KS has KS as their AIDS defining disease. Even more interesting is that 5 (68%) of the 8 visceral cases were not detected until autopsy! Of all KS cases, the mean time in days from KS diagnosis to death was shortest for the visceral group (22 days), followed by cutaneous (312 days), and cutaneous and visceral (488 days).
Payne et al. (Payne 1990) reported that survival for AIDS patients diagnosed with KS between 1981 and 1987 decreased over time. This was explained by the possibility of more aggressive tumors or the presentation of KS later in the course of HIV illness.
In a study of cutaneous neoplasms among HIV- positive patients in the military, the most common neoplasm was KS. Smith et al. reported that while the majority of patients in their cohort were in early stages of disease, the majority of patients with KS were in later stages of HIV disease. In addition, KS was associated with poor survival.
One small study of KS in women also showed an association with low CD4 cells (less than 100), more aggressive disease, and poor survival (Lassoued 1991).
One study by Miles et al. suggests that after adjusting for CD4 number, hematocrit, number of KS lesions, and body mass index, there was an increase in survival for patients with KS over the past six years (Miles: unpublished data).
While reviewing the distribution of KS it may be easy to be convinced that KS is associated with an unknown agent, possibly transmitted sexually. However, why have we not found this agent? Is it a combination of agents? Is it related to particular behaviors? Is KS the result of a hyperactive immune system? Why do HIV-negative homosexual males get KS? Why does epidemic KS have epidemiology patterns not seen in endemic KS or in KS as a result of immune suppression due to organ transplants? Only additional studies will provide additional clues to the true cause of epidemic KS.
Is the incidence of KS truly declining? While many explanations exist for a possible decrease of KS incidence, the absolute number of cases of KS remains significant. In addition, while KS was once considered an early stage disease easily treated, this pattern may also be changing to a more aggressive disease with decreased survival.
Epidemiology Recommendation: Provide a coordinated and uniform mechanism to collect data on the true incidence and prevalence of KS as a primary and subsequent AIDS diagnosis. Conduct prospective, longitudinal studies to better assess the natural history of AIDS-associated KS. Improve outreach to primary care physicians to emphasize the importance of reporting KS diagnoses to AIDS and cancer registries.