Kaposi Sarcoma Incidence, Burden, and Prevalence in United States People with HIV, 2000–2015

This article is featured in Highlights of This Issue, p. 1599

Kaposi sarcoma (KS) is a rare malignancy caused by KS-associated herpes virus (KSHV; ref. 1), which occurs at highly elevated rates among persons with HIV (PWH; refs. 2–4). KS is an AIDS-defining condition and is an indication of immune suppression (5). Though KS rates are elevated across subgroups of PWH, men who have sex with men (MSM) have the highest risk (6–10), likely due to a higher prevalence of KSHV infection (11).

The use of combination antiretroviral therapy (cART) to treat HIV began in 1996 in the United States and has become increasingly widespread, leading to significant reductions in KS rates among PWH (12–14). Relative to the general population, KS rates in PWH were elevated 2,800-fold in the pre-cART era (1991–1995; ref. 15) and 260-fold in 2009 to 2012 (14). Despite this dramatic decrease in KS rates, it remains unclear if rates have declined similarly across demographic and HIV transmission groups in the United States, or if KS rates are actually increasing in some groups, as two recent studies have suggested (16, 17). Identifying groups with increasing or even stable KS rates is critical, as KS may be an indicator of poor HIV control.

Recognizing these gaps in the epidemiology of KS, we utilized linked cancer and HIV registry data to identify a cohort of PWH to describe temporal trends in the incidence of KS over a recent 16-year period, both overall and within selected demographic and HIV transmission groups. In addition, we estimated the annual number of incident KS cases during 2008 to 2015 and the number of people diagnosed with KS in the last 5 years (5-year prevalence of KS) among PWH in the United States at the end of 2015.

Data were obtained from the HIV/AIDS Cancer Match (HACM) study, which is a population-based linkage study of 12 state and regional cancer and HIV registries across the United States (https://hivmatch.cancer.gov/). Specifically, data from Colorado (2000–2015), Connecticut (2005–2015), Georgia (2004–2012), Louisiana (2000–2015), Maryland (2008–2012), Michigan (2000–2015), North Carolina (2000–2014), New Jersey (2000–2012), New York (2001–2012), Puerto Rico (2003–2012), Texas (2000–2015), and Washington DC (2007–2015) were utilized. PWH were identified from the HIV registries, and follow-up for participants began at the latter of January 1 of the calendar year at the start of coverage for each registry, and the first of a person's HIV report date or AIDS diagnosis date. Participants 15 years or older were followed-up until death, the end of registry coverage or the end of our study period on December 31, 2015.

KS diagnosis, which was the outcome of interest, was identified in cancer registries using the ICD-O-3 histology code 9140 (18). Incident KS cases were defined as those with a KS diagnosis 4 or more months following HIV or AIDS diagnosis date. New KS cases recorded by the cancer registry 0 to 3 months after the earliest of HIV or AIDS diagnosis date were considered prevalent KS cases as the risk of KS is very high during this time period, likely due to concurrent diagnosis of HIV and cancer and cancer diagnosed during clinical work up related to new HIV diagnosis.

The state health departments of all states and Washington, DC, submit data on all HIV diagnoses to the Centers for Disease Control and Prevention's (CDC) National HIV Surveillance System. These national HIV surveillance data were adjusted for missing risk factor data (19) and were available for the full United States population during 2008 to 2015.

Incident KS rates relative to the general population were estimated with standardized incidence ratios (SIR) using data from the HACM Study. SIRs compared the observed incident KS cases among PWH to the expected number of cases based on general population rates, standardized by age, sex, and race (non-Hispanic White, non-Hispanic Black). As HIV-associated KS cases strongly influence general population rates, incidence rates recorded for the general population prior to the start of the HIV epidemic (before 1980) from Surveillance, Epidemiology and End Results data were used to calculate the expected case counts (20, 21). SIRs were calculated overall, and by age groups (<30, 30–39, 40–49, 50–59, and ≥60 years), sex, race/ethnicity (non-Hispanic White, non-Hispanic Black, and Hispanic), and HIV transmission groups [MSM, male and female persons who inject drugs (PWID), male heterosexual/other, female heterosexual/other]. The heterosexual/other category included those infected through mother to child transmission, receipt of clotting factor, transfusion/transplant, or an undetermined transmission route.

Adjusted incidence rate ratios (aIRR) were estimated for age groups, race/ethnicity, HIV transmission groups, and time since HIV diagnosis (0.33–2, 2.01–5, 5.01–10, 10.01–15, 15.01–20, >20 years), and were also adjusted for registry and calendar year using Poisson regression models. Annual changes in rates were estimated using Poisson regression with single calendar years as a linear term, adjusting for 5-year age groups, race/ethnicity, HIV transmission groups, and registry. Adjusted annual incidence rates for each calendar year were obtained as marginal estimates from these models (22, 23). Trends were assessed within strata defined by age groups (<30, 30–49, ≥50 years), race/ethnicity, and HIV transmission groups. To examine if our observed trends were unduly influenced by the four HACM registries that had data beyond 2012, we compared the temporal trends in incidence for the period 2000 to 2012 to that of our entire 16-year period. As no difference was observed, trends were evaluated on the basis of all available years. Joinpoint regression (24) was used to assess whether there were significant changes in the trajectory of incidence rates over time. In a sensitivity analysis, only those cases diagnosed 0 to 1 month after HIV report or AIDS diagnosis were excluded from incidence analysis.

To obtain annual estimates of the number of new KS cases (i.e., burden of incident KS cases) among PWH during 2008 to 2015, incidence rates were estimated from the Poisson models described above in strata defined by age group, race/ethnicity, HIV transmission group, and calendar years. Rate estimates included both incident KS cases and new cases diagnosed 0 to 3 months after the start of follow-up, considered to be prevalent in our study, to fully capture the annual number of new KS cases in the United States. These predicted incidence rates using HACM data were then multiplied by CDC estimates of the number of PWH in the United States in the same strata.

The 5-year limited-duration prevalence of KS in the United States at the end of 2015 was estimated using data on both incident and prevalent (diagnosed 0–3 months after the start of follow-up) KS cases at the end of each calendar year (2010–2015) restricted to people 15 years or older. A generalized estimation equation (GEE) logistic regression model was used to account for repeated observations of participants in each calendar year with the predicted 5-year KS prevalence as the outcome. This model was adjusted for age group, race/ethnicity, HIV transmission group, registry, and year. To quantify the number of PWH alive at the end of 2015 who had a KS diagnosis in the past 5 years, the 5-year limited duration KS prevalence was predicted on the basis of HACM data from the GEE model in strata defined by age group, HIV transmission group, and race/ethnicity and multiplied by the corresponding stratum specific total population of PWH in the United States in 2015, using CDC surveillance data. Except for the 5-year limited-duration prevalence where a parametric bootstrap was utilized, confidence intervals for incidence and burden estimates for the United States are based on nonparametric bootstrap resampling. All analyses were carried out in SAS (version 9.4).

From 2000 to 2015, 2,837 incident KS cases occurred during 4,488,522 person-years of follow-up in the HACM Study (Table 1) and the overall crude incidence rate during this period was 63.2 per 100,000 person-years (81.2 in men and 15.8 in women). In addition, a total of 1,492 KS cases occurred during the first 3 months after HIV or AIDS diagnosis. By the demographic and HIV transmission groups examined, 70.7% of incident KS cases occurred among PWH ages 30 to 49 years, 93.1% occurred among men, 25.8% among non-Hispanic White PWH, 48.7% among non-Hispanic Black PWH, 24.3% among Hispanic PWH, and 70.1% occurred among MSM. The overall SIR was 521 [95% confidence interval (CI), 502–540, Table 1] with the highest SIRs occurring among <30-year-olds (SIR: 358,000; 95% CI, 322,000–397,000, Table 1), females (SIR: 680; 95% CI, 588–783), and among MSM (SIR: 809; 95% CI, 774–845).

The aIRRs presented in Table 1 show that compared with 15- to 29-year-olds, KS incidence rates were significantly higher among 30- to 39-year-olds (1.31; 95% CI, 1.16–1.49, Table 1); however, in people 50 years or older rates were significantly lower (aIRR range: 0.47–0.66). Compared with non-Hispanic Whites, KS incidence was also significantly higher among non-Hispanic Blacks (1.50; 95% CI, 1.36–1.65) and Hispanics (1.27; 95% CI, 1.14–1.42). By HIV transmission group and time since HIV diagnosis, MSM and those with a recent HIV diagnosis, respectively had the highest KS incidence (Table 1). Results were similar when incident cases were defined as those occurring more than 1 month after HIV report date or AIDS diagnosis date.

During 2000 to 2015, KS rates declined at an average annual percent change (AAPC) of 6.1% per year (95% CI, −7.0% to −5.2% per year, Table 2) from an adjusted annual incidence rate of 108.9 per 100,000 person-years in 2000 to 46.6 per 100,000 person-years in 2015 (Fig. 1), although beginning in 2012 we noted an apparent stabilization of KS rates (P = 0.005). Substantial declines (P_trend < 0.005) in KS rates were observed across age groups (AAPC range: −8.4% to −3.5% per year, Table 2; Supplementary Fig. S1), race/ethnicity (AAPC range: −7.4% to −5.4% per year) and HIV transmission groups (AAPC range: −7.6% to −5.8% per year). KS rates also declined across all states included in our study (Supplementary Table S1).

From 2008 to 2015, an estimated 5,306 (95% CI, 5,018–5,616) incident KS cases occurred among PWH in the United States, declining from 719 incident KS cases in 2008 to 624 incident cases in 2015 (Fig. 2). An estimated 23 KS cases occurred among 15- to 19-year-olds, 968 in 20- to 29-year-olds, 3,258 in 30- to 49-year-olds, 798 in 50- to 59-year-olds, and 259 in ≥60-year-olds. Non-Hispanic Blacks (1,981; 37% of cases) and MSM (4,711; 89% of cases) respectively had the highest burden of incident KS cases (Fig. 2).

At the end of 2015, 0.20% (95% CI, 0.18%–0.21%) of PWH in the United States had received a diagnosis of KS within the prior 5 years, corresponding to 1,904 prevalent cases (Table 3). By age group, the proportion of 5-year prevalent KS cases recorded in people <30 years was 0.21% (n = 215 cases), 30 to 39 years 0.30% (n = 486), 40 to 49 years 0.22% (n = 539), 50 to 59 years 0.17% (n = 497), and those 60 years or older had the lowest 5-year prevalence of 0.11% (n = 166). By racial/ethnic group, non-Hispanic Whites and Hispanics had the highest KS prevalence of 0.24% (n = 712) and 0.22% (n = 458) respectively, and by HIV transmission group MSM had the highest prevalence [0.29% (n = 1,686), Table 3].

Our findings indicate that from 2000 to 2015 there was a >50% decrease in the incidence of KS among PWH in the United States and the slope of decline was similar across age, racial/ethnic, and HIV transmission groups. In 2015, in the United States, among PWH an estimated 624 new cases of KS were diagnosed and the 5-year prevalence of KS was 1,904. The substantial decrease in KS incidence may be attributed to sustained efforts in recent years to diagnose HIV earlier and initiate cART shortly after diagnosis, maintain sustained suppression of HIV RNA, increased availability of cART, tolerability of newer cART regimens, increased adherence to simple formulations with once-daily dosing, and increased linkage of PWH to care and retention in care (25). Continued efforts to expand optimal cART treatment to those less likely to engage in care are needed to sustain or accelerate further declines in KS rates.

Recent reports have shown increasing KS rates among non-Hispanic Blacks, particularly in specific age groups in some southern states (16, 17). Although we could not directly compare these published estimates to those in the current study due to differences in the states included in the analyses, we did not find evidence of increasing KS rates in any racial/ethnic group or state. Further, these prior studies utilized the general population as the denominator for rate estimation instead of the population of PWH, preventing the distinction between increasing KS occurrence among PWH and an increase in the at-risk population (i.e., increase in the number of PWH) over time. Subsequent analyses using cancer registry data and CDC HIV surveillance data have not shown increasing KS rates in these groups (26).

KS rates were highest among MSM compared with other HIV transmission groups, likely due to the elevated seroprevalence and the increased rate of sexual transmission of KSHV in MSM (11). This finding is also consistent with the 2-fold elevation in the seroprevalence of KSHV among PWID who are MSM compared with PWID who are male heterosexuals (27). KS incidence declined with increasing age. This is in contrast with the increased susceptibility to infections that occurs as people age and undergo immune senescence, and the increased risk of classic KS with age among immunocompetent individuals (28). In addition, prior studies in the general population have shown an age-associated increase in the seroprevalence of KSHV among people 30 years or older (29). However, consistent with our findings, the risk of KS among older PWH has been reported to decrease with increasing time since HIV diagnosis (30, 31) and was supported by the higher rate of cART adherence among older PWH (32, 33).

Although the 5-year limited duration prevalence of KS among PWH is low, the absolute number of prevalent cases in 2015, especially among MSM, is quite notable. The population of prevalent KS cases is likely to continue to grow further, as PWH continue to live longer after a KS diagnosis (34, 35) when controlled on cART (34). Therefore, the unique needs of PWH who are KS survivors should be considered, such as their potential increased risk for second cancers (36–38).

A major strength of our study is the use of population-based data from 12 regions across the United States in addition to CDC HIV surveillance data for the whole United States to describe the epidemiologic trends for KS in the current cART era. Second, our examination of temporal trends across key demographic and HIV transmission strata has shown consistent declines in KS incidence across major demographic and HIV transmission groups. In addition, our study includes all persons with diagnosed HIV in the United States and not just those continuously enrolled in care or adherent to cART. However, our findings are limited by the imperfect sensitivity of the HACM data match to identify HIV-infected cancer cases. As the sensitivity of the match has been reported to be 81%, it is likely that the burden of incident KS is underestimated (21). Another limitation of the study is the lack of information on cART and KSHV serostatus and incomplete information on CD4 cell counts and HIV RNA. Further, the 5-year limited duration prevalence at the end of 2015 is an underestimate of the true lifetime prevalence of KS among PWH in United States, as these 5-year limited duration prevalence estimates do not include KS diagnoses occurring prior to 2010.

In conclusion, this study provides a comprehensive description over a 16-year period of the current landscape of KS incidence, burden and prevalence across important population subgroups among PWH in the United States. Our findings indicate that though rates remain elevated relative to the general population there has been a consistent gradual decline in KS incidence among PWH in the United States during the current cART era, and this decrease is uniform across the demographic and HIV transmission groups examined. However, the number of PWH with a prior KS diagnosis at the end of 2015 is notable and the unique needs and risk profile of prevalent KS cases should be investigated further. As KS risk is strongly associated with immune suppression, increased HIV viral load and late initiation of cART, the continued strongly elevated risk of KS among PWH highlights the continued need for improvement in utilization of HIV diagnosis and treatment. Current efforts to control HIV in the United States through early initiation of cART, and linkage to and retention in care need to be expanded to sustain the recent declines in KS rates among PWH.

A. Monterosso reports grants from CDC during the conduct of the study. No disclosures were reported by the other authors.

The views expressed in this paper are those of the authors and should not be interpreted to reflect the views or official policies of the NCI, HIV/AIDS or cancer registries, the Centers for Disease Control and Prevention, or their contractors, nor does the mention of trade names, commercial practices, or organizations imply endorsement by the U.S. Government.

S. Peprah: Conceptualization, formal analysis, investigation, visualization, methodology, writing–original draft. E.A. Engels: Supervision, funding acquisition, methodology, writing–review and editing. M.-J. Horner: Data curation, methodology, writing–review and editing. A. Monterosso: Resources, data curation, writing–review and editing. H.I. Hall: Resources, data curation, writing–review and editing. A.S. Johnson: Resources, data curation, writing–review and editing. R.M. Pfeiffer: Conceptualization, supervision, investigation, methodology, writing–review and editing. M.S. Shiels: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, visualization, methodology, writing–review and editing.

The authors thank Timothy McNeel at Information Management Services for programming support. They also gratefully acknowledge the support and assistance provided by individuals at the following state HIV/AIDS and cancer registries: Colorado, Connecticut, District of Columbia, Georgia, Louisiana, Maryland, Michigan, New Jersey, New York, North Carolina, Puerto Rico, and Texas. This work was supported by the Intramural Research Program of the National Cancer Institute at the National Institutes of Health.

The following cancer registries were supported by the National Program of Cancer Registries of the Centers for Disease Control and Prevention: Colorado (NU58DP006347-01), Georgia (5U58DP003875-01), Louisiana (NU58DP006332-03-00), Maryland (5NU58DP003919-05-00), Michigan (17NU58DP006334), New Jersey (NU58/DP003931-05-00), New York (U58/DP003879), North Carolina (1NU58DP006281), and Texas (1NU58DP006308). District of Columbia was supported by the Centers for Disease Control and Prevention cooperative agreement DP006302.

The following cancer registries were supported by the SEER Program of the National Cancer Institute: Connecticut (HHSN261201300019I), Louisiana (HHSN261201800007I/HHSN26100002), and New Jersey (HHSN261201300021I, N01-PC-2013-00021). The New Jersey State Cancer Registry was also supported by the state of New Jersey, the Maryland Cancer Registry was supported by the State of Maryland and the Maryland Cigarette Restitution Fund, the Louisiana Tumor Registry was also supported by the state of Louisiana (0587200015), and the New York State Cancer Registry was also supported by the state of New York.

The following HIV registries were supported by HIV Incidence and Case Surveillance Branch of the Centers for Disease Control and Prevention, National HIV Surveillance Systems: Colorado (NU62PS003960), Connecticut (5U62PS001005-05), Louisiana (NU62PS924522-02-00), Michigan (U62PS004011-02), New Jersey (U62PS004001-2), New York (PS18-1802), and Texas (NU62PS924529): Integrated HIV Surveillance and Prevention Programs for Health Departments, National Center for HIV, Viral Hepatitis, STD, and TB Prevention (NCHHSTP).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.