Correspondence to: Karima Chaabna, Section of Cancer Information, International Agency for Research on Cancer, 150 Cours Albert Thomas, 69372 Lyon, France, Tel.: +33-472-7381-72, Fax: +33-472-7386-96, E-mail: email@example.com
Trends in Kaposi sarcoma (KS) incidence over four decades were described for Zimbabwe and Uganda. KS data were retrieved from the population-based cancer registries of Bulawayo (1963–1971) and Harare (1990–2005), Zimbabwe and Kyadondo, Uganda (1960–1971 and 1991–2007). Joinpoint regression models were used to analyze time trends of KS incidence. Trends were compared to HIV/AIDS trends and were also described as rates versus birth cohort by age. In both countries, an increased incidence of KS accompanied the emergence of the HIV/AIDS epidemic (p-value < 0.0001). In Zimbabwe, KS incidence (both sexes, all ages) changed in parallel to that of HIV/AIDS prevalence, whereas in Uganda, despite an observed decrease in HIV/AIDS prevalence since 1992, we observed a decrease in KS incidence in men younger than 50 years (Annual Percent Change, APC after 1991 = −4.5 [−5.6; −3.4], p-value < 0.05) but not in men aged >50 years (APC after 1991 = 1.0 [−2.8; 5.0]) nor in women (APC = 1.0 [−0.6; 2.6]). In both populations, a period effect at older ages was observed, with initial increases in incidence in men followed subsequently by a downturn in rates of the same magnitude. The uniformly declining rates in younger men (aged less than 30 years) suggested that a recent cohort effect was also in operation with a reduced risk in generations born after the mid-1950s in Uganda and in the mid-1960s in Zimbabwe. The combined introduction of antiretroviral therapy and effective prevention programmes against HIV/AIDS appeared to be the key contributors to the KS decline observed in both Uganda and Zimbabwe.
human immunodeficiency virus/acquired immunodeciency syndrome
international agency for research on cancer
international classification of diseases for oncology
Kyadondo cancer registry
Kaposi sarcoma associated herpesvirus
standardized rate ratios
United State of America
Kaposi sarcoma (KS) has been a common form of cancer among indigenous populations in Sub-Saharan Africa for many decades. Prior to the HIV/AIDS epidemic (in the 1970s), KS accounted for around 9% of all histologically-confirmed malignant tumors in central African region and 3% in Zimbabwe. By 1990, the incidence rates of KS had increased 20-fold to 55 per 1000 in Uganda (Kyadondo County) and to 22 per 1000 in Zimbabwe (Harare). HIV infection and AIDS has been shown to greatly increase the risk of KS. A study in the USA showed that 81% of KS cases occurred in people living with HIV/AIDS. Despite the reduced risk of KS after the introduction of antiretroviral therapy (ART) in the mid-1990s in the USA, the rate of KS was still 140 times higher among men with HIV/AIDS as compared to those without HIV infection. The decline in HIV infection in Zimbabwe and Uganda has been attributed to HIV/AIDS prevention programmes including prevention of mother-to-child and sexual transmission, as well as to uptake of ART.[6, 7] A European study assessing the KS risk before and after implementation of ART for example, reported a 11-fold reduction in risk of KS. In Africa, primary prevention is a central strategy in HIV/AIDS control. Generally, the ART coverage in sub-Saharan Africa is lower compared with developed countries, but the recent uptake has been dramatic i.e., among the 4.7 million people in need of ART in Sub-Saharan Africa, 100,000 people received treatment at the end of 2003 (ART coverage = 2%) as compared with 310,000 at the end of 2004, and 810,000 at the end of 2005 (ART coverage = 17%).
This article aims to describe the changes in KS incidence before and subsequent to the HIV/AIDS epidemic in Zimbabwe and Uganda using the best available population-based cancer registry data available data from Zimbabwe and Uganda. Changes over time in the incidence of KS were analyzed alongside concomitant trends in HIV/AIDS prevalence. Additionally, sex-specific differences in KS incidence were examined and an assessment made of whether the reported male predominance of KS before the HIV/AIDS epidemic[3, 10, 11] prevails.
Material and Methods
HIV/AIDS prevalence data among adults aged 15–49 by calendar year (1990–2007) and country were obtained from United Nation AIDS. As for KS (ICDO = C46) incidence, data were taken from population-based cancer registries in two Sub-Saharan African countries: Kampala, 1960–1971 and 1991–2007 (Uganda), Bulawayo, 1963–1971 and Harare, 1990–2005 (Zimbabwe).
The Bulawayo Cancer Registry in Zimbabwe was founded in 1963 and functioned for 15 years until 1978. It covered the South-Western part of Zimbabwe with a population of around 160,000 circa 1969. The national Cancer Registry in Harare was established in 1985 to register cancers in Zimbabwe. It is located in the North-Eastern part of the country covering a population of 2.02 million in 2005 (17.4% of the total population). Kyadondo Cancer Registry (KCR) is a population-based cancer registry in Uganda covering a population of 1.6 million people in 2007 (6.6% of the total population). The area covered by KCR comprises of Kampala, the capital city, and part of Wakiso district. KCR was established in 1951 but stopped functioning between 1971 and 1990 because of instability in the country. It resumed activities in 1991.
Data from these registries appeared in successive Volumes of Cancer Incidence in Five Continents (CI5) published by IARC, for which specific quality criteria must be met, e.g. for the volume IX, a proportion of tumors which were morphologically verified (%MV) more than 60% but less than 99–100% and a percentage of cases identified as registered from death certificates only (%DCO) less than 20%.
Age-specific incidence rates (ASI) were estimated for the 4 age groups 0–19, 20–39, 40–59, and ≥60 in both genders from Zimbabwe (Bulawayo in 1963–72 and Harare in 1991–2002) and Uganda (Kyadondo in 1960–71 and 1991–2002). Furthermore, age standardized incidence rates (ASR) were estimated using direct standardization with the African standard population given its similar age structure to the populations studied and the high burden at younger ages. These rates were expressed as rates per 100,000 persons. Comparison between pre- and during AIDS epidemic (1991–2002) periods; and between genders and populations during AIDS epidemic (1991–2005) were performed using chi-square and Fisher exact's tests. Šidàk correction was used to counteract the problem of multiple comparisons and control the family-wise error rate because individual tests were assumed to be independent. Instead of using a p-value threshold of α = 0.05, a stricter threshold with β = 1 − (1 − α)1/n was used. Here, n represents the number of individual tests performed and β was equal to 0.004.
Rates were compared and expressed as standardized rate ratios (SRR). We calculated 95% confidence intervals (95% CI) for the SRR based on the Poisson distribution. SRRs were considered statistically significant when the 95% CI did not include unity.
KS annual ASR (overall and truncated in two age categories: <50 and ≥50-years-old) were computed by sex and population. A log-linear regression model with age-adjusted rates as the response was used to determine joinpoints in the trends of KS incidence. Standard errors of rates were included in the analysis to take into account the uncertainty stemming from different denominators for those rates. These are breakpoints in time for which significant changes in the linear trends are detected. They enable estimation of the magnitude of the trend–the annual percent change (APC) between each linear segment defined between two joinpoints. A maximum number of four joinpoints was specified. To quantify the trend for two fixed predetermined time periods, we estimated the average annual percent change (AAPC), based on a method that uses the underlying Joinpoint model to compute a summary measure over the prespecified intervals. It is computed as a weighted average of the APC's from the Joinpoint model, with the weights equal to the length of the APC interval. Joinpoint analyses were performed using the “Joinpoint” software from the Surveillance Research Program of the US National Cancer Institute.
To examine the KS incidence trends according to generation (cohort), 10-year synthetic birth cohorts were obtained by subtracting the midpoints of the 5-year age groups from the corresponding midpoints of 5-year calendar time. The trends are presented as rates versus birth cohort by age (in 5-year age groups).
Prior to the HIV/AIDS epidemic (during the 1960s), age-specific rates of KS ranged between close to 0 to 8/100,000 in Zimbabwe and Uganda (Table 1). During the emergence of HIV/AIDS, the highest rates of KS in males was observed for the 40–59 years age group with a 20- and 10-fold increase in risk of KS as compared with the pre-HIV/AIDS era in Zimbabwe and Uganda, respectively. A higher overall increase of KS among males in Zimbabwe (SRRpre/post = 29.2, p-value < 0.0001) resulted in a 30% higher rate of KS in Zimbabwe during the HIV/AIDS epidemic period as compared to Uganda (SRRZimbabwe/Uganda = 1.3, p-value < 0.0001).
Table 1. Age-specific and age standardized rates of Kaposi sarcoma incidence before (1960–1972) and during AIDS epidemic (1991–2002) according to sex in Uganda and Zimbabwe
There were 4,237 KS cases in men and 1,860 in women in Harare, Zimbabwe (1990–2005); and 2,750 in men and 2,020 in women in Kyadondo, Uganda (1991–2007). A male predominance was observed in both countries with a lower magnitude during AIDS-epidemic (e.g., SSRM/F = 17.1, p-value = 0.002 in Uganda in the pre-HIV/AIDS epidemic period versus 1.5, p-value < 0.0001 during the HIV/AIDS epidemic period) (Table 2). Reduced gender difference was partly a result of a higher overall increase in rates in females as compared to males (SSRpre/during = 72.4, p-value < 0.0001 in females in Zimbabwe and SSRpre/during = 126.5, p-value < 0.0001 in Uganda).
Table 2. Comparisons of age standardized rates of Kaposi sarcoma incidence according to sex and studied population in 1991–2005
Abbreviations: ASR: age standardized incidence (per 100,000 African standard population); SSR: standardized rate ratio; 95% CI: 95% confidence interval.
The time trends in KS rates in Zimbabwe followed almost exactly the pattern of HIV/AIDS prevalence (Fig. 1a). Between 1990 and 1998, the ASR for KS increased from 17 to 60 per 100,000 for males and from 8 to 31 per 100,000 for females. Annually among males a 41.2% (95% CI = 23.6–61.2) and 4% (95% CI: −0.3;9.5) increase in rate in 1990–1993 and in 1993–1998, respectively was observed. For females, the annual increase in rate between 1990 and 2001 was 8.9% (95% CI = 3.0;15.1). After 1998 in males (−8.5%, 95% CI = −12.2; −4.8 in 1998–2003; and −37.0%, 95% CI = −47.4;24.6 in 2003–2005) and after 2001 in females (−26.1%, 95% CI = −37.8; −12.3), rates steadily decreased in both sexes, and reached the rates observed in 1990 by 2005. The joinpoint analysis further supported this observation (Table 3). In Zimbabwe the trend of KS differed to that observed in Uganda in two ways. Firstly, the trend in KS followed that of HIV/AIDS prevalence for both sexes in Zimbabwe. On the other hand, only male KS rates followed the HIV/AIDS prevalence in Uganda (−3.2%, 95% CI = −4.4; −1.9] between 1990 and 2007), whereas the female rates for KS showed a relatively stable trend over the study period (Fig. 1b). Secondly, the timing of the change in incidence and prevalence differed: HIV/AIDS prevalence and KS incidence rates among males in Zimbabwe were decreasing only after 1998; in Uganda the decrease was seen since the early 90s. Age-specific trends in KS rates according to age groups (categorized into two groups below and over 50 years) and sex were also examined (results not shown). Incidence rates in both sexes in Zimbabwe decreased only during the early 2000s for both age groups. In Uganda, decrease in KS rate was observed during the whole period (between 1991 and 2007) in men younger than 50 years (AAPC = −4.5%, 95% CI = −5.6; −3.6]) but remained constant over time in men older than 50 years and in women.
Table 3. Trend of Kaposi sarcoma incidence in Zimbabwe (Harare, 1990–2005) and Uganda (Kyadondo, 1991–2007) according to sex
AAPC (95% CI)
AAPC (95% CI)
AAPC (95% CI)
Significantly different from zero (p < 0.05).
Abbreviations: AAPC: average annual percent change.
Figure 2 shows the KS incidence trends by 5-year age groups plotted against birth cohort in Zimbabwe (Fig. 2a) and Uganda (Fig. 2b). Both populations presented similar but shifted trends. A period effect was observed in older age groups with initial increases in KS incidence followed subsequently by a downturn of same magnitude in men aged 30 years and older in Zimbabwe and aged 40 years and older in Uganda. However, for men aged less than 29 years in Zimbabwe and aged less than 39 years in Uganda, rates uniformly decreased. This difference in trends between generations suggested a cohort effect was also in operation with a reduced risk for KS in generations born after the mid-1950s. The curves for older men appeared to turn downward near 1960s in Zimbabwe and near 1950s in Uganda.
In this study, we reported marked changes in KS incidence in two African populations during the last four decades. A strengthening of the decrease in KS rates was observed during the most recent time period. It is likely that the decline in the incidence of KS during the 1990s was related to the decrease in HIV/AIDS prevalence. The decline in KS incidence was mostly evident for males in both populations and females in Zimbabwe, whereas the rates for females were stable over the last two decades in Uganda.
KS was relatively common in sub-Saharan Africa before the HIV/AIDS epidemic and this “endemic” form of KS was recognized since the 1950s as a common neoplasm in native populations of Sub-Saharan Africa. During this period, we found KS incidence was higher in Uganda than in Zimbabwe. Furthermore, consistent with a previous report, we found a male predominance. After the emergence of the HIV/AIDS epidemic in the 1990s, we observed an increased incidence of KS in all age groups with an incidence peak at age 40–59 years that was consistent with the “epidemic” form of KS, which generally affects younger age groups.
The first HIV/AIDS epidemic was reported in Kinshasa (Democratic Republic of Congo) in the 1970s. It reached epidemic levels in the Eastern African region (including Uganda) in the early 1980s,[25, 26] and only afterwards in Southern Africa (including Zimbabwe). By the end of the 1980s, Zimbabwe and other southern African countries such as Malawi, Zambia, and Botswana overtook East Africa as the focus of the HIV/AIDS epidemic. We found a decrease in the KS incidence in Uganda occurring in the 1990s, followed subsequently by Zimbabwe (from around 2000). This marked decline in KS incidence was most likely caused by a combination of several changes surrounding the HIV/AIDS epidemic.
The decline in incidence trends observed in several birth cohorts among Ugandan and Zimbabwean men may be related to the introduction of ART in these populations irrespective of the age at diagnosis of HIV/AIDS. The decline in KS incidence began earlier in Uganda than in Zimbabwe, probably reflecting an earlier implementation of ART. The coverage of ART among adults increased considerably between 2006 and 2010; from 43 to 62% in Uganda and from 34 to 52% in Zimbabwe. The use of ART has substantially improved the clinical outcome of patients in Western countries, and in Europe, it has also been reported to reduce the risk of developing KS among patients with HIV/AIDS. Similar outcomes start to be observed in sub-Saharan Africa,[30, 31] and further increase of ART coverage is need in this region.
Our study reported a KS incidence decrease related to HIV/AIDS prevalence decline, in Uganda and Zimbabwe, and thus suggested an important role of other factors including primary prevention. The KS decline in rates in young men (aged under 30 years) is indicative of a reduced risk in generations born after the mid-1950s in Uganda and the mid-1960s in Zimbabwe, and behavioral changes partly due to effective multiple primary prevention programmes. Several reports,[33-35] concluded that the primary reason for Uganda's success with regard to the control of the HIV/AIDS epidemic is the decrease in casual/multiple sexual partner behavior. Others have suggested a prevention approach led by promotion of condom usage and HIV testing as well as the tackling of an array of broader societal factors, such as poverty, gender violence, and conflict.[36, 37] In Zimbabwe, the success of these prevention strategies has also been reported, and has been widely associated with the HIV/AIDS decline in the country.
KS incidence was 1.5–2.1 times higher in men, despite the lower prevalence of HIV infection among men in Uganda and Zimbabwe. A sex-specific difference in risk, with incidence rates elevated among men was observed prior to the HIV/AIDS epidemic in our study. This is probably because of the higher prevalence of Kaposi sarcoma associated herpes virus (KSHV) in males.[39-41] In 1994, KSHV was identified in AIDS-related KS cases and is now considered as the primary etiological agent for all forms of KS. The immunosuppression caused by HIV infection increases the likelihood of other opportunistic infection including KSHV. Other factors such as a genetic predisposition of men to KS have also been postulated to explain the male predominance of this tumor. In addition, the changing distribution of KSHV infection probably also partially explained the geographical and temporal differences in KS incidence in our study. For example, we observed a higher KS incidence among women in Uganda between 2001 and 2005 compared to Zimbabwe. Concurrently, several studies have showed that KSHV infection was more frequent during this period among women in Uganda than Zimbabwe.[40, 41]
Although we used data assessed as being among the best in Africa, our study had limitations related to data quality. The percentage of morphologically verified (%MV) KS was 65% and 75% between 2002–2007 in Harare and Kyadondo, respectively. The corresponding percentage of cases from death certificate only (%DCO) was 9% (between 2002 and 2007) in Harare. In Uganda, certification of death has been only carried out for legal reasons and has been known to be quite incomplete, so death certificates have not been used as a source of information by the registry. In Bulawayo, death certifications prior to burial were mandatory and the bodies of all the persons dying outside hospital in the African township of the city were brought to the mortuary for autopsy prior to certification. Between 1963 and 1967, autopsy rates were reported as 83% of all deaths in Bulawayo. In Harare (Zimbabwe) and Kyadondo County (Uganda), the low percentage of %MV (in comparison to European or American registry data quality) is a potential indicator of relative poor data validity in these registries. Some cases may be misclassified as KS. Moreover, the large %DCO observed in Harare indicated a substantial proportion of cancer cases had no information on basis of diagnosis. However, improvements in data quality were observed in both registries. In Harare, the %MV percentage increased from 60% to 65% between 1998–2002 and 2002–2007, while the %DCO decreased from 12 to 9% between the study periods.
The consistent declines in KS incidence in Zimbabwe and Uganda observed recently have been largely explained by changing patterns of HIV infection. In both countries, the combined introduction of ART and effective primary prevention programmes have been the main contributing factors to the decline of HIV/AIDS prevalence, and subsequently KS incidence. General differences in KS patterns by sex, time and geographical location during the HIV/AIDS epidemic has suggested the importance of additional risk factors, including differences in the prevalence of KSHV infection.