We utilized data from the HIV/AIDS Cancer Match Study (HACM), a data linkage study of HIV and cancer in 14 regions in the US: Texas (2001–2015), New York (2001–2019), Colorado (2001–2015), Louisiana (2001–2015), Florida (2001–2019), Georgia (2004–2012), Puerto Rico (2003–2017), North Carolina (2001–2014), Washington D.C. (2007–2015), Maryland (2008–2018), New Jersey (2001–2012), Connecticut (2002–2016), Massachusetts (2007–2016), and Michigan (2001–2015). We identified individuals 20 years and older for whom the primary risk factor for HIV acquisition was recorded as MSM. We used the reported race and ethnicity variable from the HIV registry to categorize individuals as: non-Hispanic (NH) White, NH Black, or Hispanic/Latino. We excluded individuals of other racial identities due to the limited number of PWH and cancer diagnoses in these groups.

The HACM Study was approved by institutional review boards at participating registries and was exempt from review at the US National Institutes of Health. Consent of participants was not required for the use of data collected through public health surveillance.

We calculated incidence rates (IRs) for five cancers in MSMWH that are common and for which risk is elevated in people with HIV compared with the general population (i.e., Kaposi sarcoma [KS], non-Hodgkin lymphoma [NHL], anal cancer, Hodgkin lymphoma [HL], and liver cancer) by race and ethnicity, standardized to the age distribution of MSMWH in 2010. Follow-up started at the first of age 20 years, HIV report date or beginning of registry coverage, and ended at the earliest of age 85 years, end of registry coverage, or death. For analyses stratified by prior AIDS status, diagnosis of AIDS was treated as a time varying covariate. We include these estimates to emphasize the potential role of an AIDS diagnosis as a mediator in explaining observed racial and ethnic disparities in cancer risk for MSMWH.

To compare the differential impact of HIV on cancer risk within each race and ethnicity we estimated standardized incidence ratios (SIRs). SIRs were calculated to compare the observed counts of each cancer among MSMWH to the number of expected cancer cases based on rates from the general population of males in the US, stratified by race and ethnicity, age, region, and year. SIRs stratified by prior AIDS status were calculated to investigate how progression to AIDS affects cancer risk. For KS, expected rates were based on data collected by the Surveillance, Epidemiology, and End Results cancer registries before the HIV epidemic (1973–79) [12]. To further test for differences in SIRs between racial and ethnic groups after accounting for a prior diagnosis of AIDS we performed a global test for race and ethnicity in the adjusted model and report the corresponding global p-value [13, 14].

Incidence rate ratios (IRRs) were calculated using Poisson regression to compare risk of cancer among MSMWH between racial/ethnic groups, adjusted for age, region, and year. White MSMWH were used as the referent group because they contributed the most person-time allowing for greatest power to detect statistically significant differences in risk. Significance for IRRs were calculated using 95% CI. Goodness of fit tests were conducted on the residuals to assess appropriateness of the model and confirm that there was not concern for overdispersion.

SIRs were notably higher for KS among Hispanic/Latino and Black MSMWH (887 and 772, respectively) compared to White MSMWH (417) (Figure 1). A similar pattern was observed for NHL, with higher SIRs among Hispanic/Latino and Black MSMWH (5.5 and 4.8, respectively) than White MSMWH (3.1). White MSMWH had a higher SIR for anal cancer (38.4) compared to Hispanic/Latino (30.8) and Black MSMWH (24.1). HL risk was more elevated for Hispanic/Latino MSMWH compared to the general population (11.5) and elevated similarly for Black and White MSMWH (7.7 and 6.4, respectively). The highest SIR for liver cancer was among White MSMWH (2.27) followed by Black MSMWH (1.78) and Hispanic/Latino MSMWH (1.49).

Even after adjusting for a prior diagnosis of AIDS, there were statistically significant differences in SIRs between racial and ethnic groups based on the global p-value (Table 3). Within racial/ethnic groups, every SIR for those with a prior AIDS diagnosis was greater than for those who had not progressed to AIDS, except for liver cancer. For liver cancer, there were no significant differences in AIDS stratified SIRs among Hispanic/Latino or Black MSMWH, but the SIR for White MSMWH with a prior AIDS diagnosis was significantly higher than White MSMWH who had not progressed to AIDS (2.78 compared to 1.40).

Among Hispanic/Latino and Black MSMWH, KS was the most common cancer (115 and 110 per 100,000 person-years, respectively) followed by NHL (111 and 98.8, respectively) (Table 4). In contrast, NHL was the most common cancer among White MSMWH (106) followed by anal cancer (87.4), then KS (75.0).

Compared to White MSMWH, rates were higher among Hispanic/Latino MSMWH for KS (IRR 1.40; 95%CI=1.28–1.54), NHL (1.22; 95%CI=1.12–1.32), HL (1.79; 95%CI=1.52–2.11) and liver cancer (1.42; 95%CI=1.17–1.72). Risk of anal cancer was 32% lower for Hispanic/Latino compared to White MSMWH (0.68; 95%CI=0.60–0.76) (Table 4). Compared to White MSMWH, Black MSMWH had significantly higher rates of KS (IRR 1.24; 95%CI=1.13–1.36), HL (1.44; 95%CI=1.22–1.70), and liver cancer (1.54; 95%CI=1.29–1.84), but similar rates of both NHL and anal cancer. There was no evidence of differences in IRRs by immunosuppression status.

KS is caused by an infection of Kaposi sarcoma herpesvirus-8 (KSHV). Some cases of NHL and HL are caused by Epstein-Barr virus (EBV) among PWH. These viruses have a greater carcinogenicity in immunocompromised persons. KSHV prevalence is higher among MSM than in the general US population and, when adjusting for HIV status, there are no known differences in prevalence by race or ethnicity [15–18]. Studies of geographic and racial disparities over time in KS incidence among young men have reported annual improvements for White men and less significant improvements, even declines in the South US, among Black young men [19, 20]. Our data showed similar disparities in KS incidence with higher rates for Hispanic/Latino and Black MSMWH compared to White MSMWH. Hispanic/Latino and Black MSMWH had higher relative risks of KS compared to the general population, approximately twice that of White MSMWH. Further, relative risk of KS was greater among those with a prior AIDS diagnosis, consistent with an association with severe immunosuppression [15]. Prior research established that Black PWH have twice the mortality rate from HIV-associated KS compared to White and Hispanic/Latino PWH, even after controlling for covariates like HIV viral load and ART adherence, suggesting that our research does not capture the extent of disparities in outcomes like mortality from KS diagnoses among MSMWH [21].

Prior studies have established that MSMWH have a 30% higher risk of NHL compared to heterosexual men with HIV and this risk increases with greater severity of immunosuppression [22]. Additionally, risk for HL is 5 to 15 times higher among PWH compared to the general population [23]. Hispanic/Latino and Black MSMWH had higher SIRs than White MSMWH for both NHL and HL. EBV seroprevalence is higher among Black and Hispanic/Latino individuals, compared to White individuals in the US. [24, 25]. Racial disparities in prevalence of EBV among MSMWH have not been identified but likely reflect those observed in the general population. For each racial/ethnic group, MSMWH with a prior AIDS diagnosis had a higher relative risk of NHL than those who had not progressed to AIDS, emphasizing the importance of adherence to HIV treatment to maintaining viral suppression. Our research supports that preventing AIDS could reduce some racial disparities in risk of KS among MSMWH, but even after accounting for disparities in the risk of HIV progression, there are residual disparities in cancer risk. Access and adherence to cART due to systemic barriers could contribute to differences in morbidity and mortality from KS since those with HIV who receive cART had a much lower incidence of KS compared to those who do not [16].

In 2015, incidence of anal cancer among MSMWH was 9 times higher than in MSM without HIV, who still had a higher incidence than the general population [26]. Additionally, high risk anal HPV, which has a higher seroprevalence among Black MSM compared to White MSM, causes a higher proportion of anal cancer among Black and Hispanic/Latino PWH [27–29]. Our findings are consistent with these disparities since Black MSMWH had a lower SIR for anal cancer than White MSMWH even though the IRR was not statistically different. The contrast in these observed estimates could be driven by a higher number of expected anal cancers among Black men, which are estimated based on incidence rates in the general population. It is likely that the anal cancer SIR for Black MSMWH is an underestimation of the true relative risk, because 57% of anal cancer cases among Black men in the U.S. occur among those with HIV, thus the general population estimate is highly influenced by HIV, similar to what has previously been observed for KS [12].

Recent guidelines proposed for anal cancer screening identify PWH as a priority population that experience significant benefits from screening [30]. Because these screenings are not yet standard clinical practice, it is possible White MSM have had higher anal cancer screening rates, which have resulted in higher rates of early-stage, prevalent disease in comparison with other racial and ethnic groups [31]. One meta-analysis suggested White individuals in the US have the highest coverage of HPV vaccination, but non-White individuals are more likely to initiate the vaccination process with their provider, suggesting a closing in racial disparities in vaccination coverage [32, 33]. Although the HPV vaccine was too recently introduced to affect anal cancer rates today, HPV vaccination is highly effective in reducing the risk of anal cancer among MSMWH and should be better implemented in this group [34].

Rates of hepatocellular carcinoma (HCC), the most common histologic subtype of liver cancer, have largely plateaued in the general U.S. population, yet this decline has not been evenly distributed by race or ethnicity with White individuals having the lowest incidence rates between 2001–2019 [35, 36]. PWH have a higher burden and increased risk of HCC compared to those without HIV, due to higher prevalence of coinfection with hepatitis B virus (HBV), hepatitis C virus (HCV), behavioral risk factors, and immunosuppression [37]. Other work suggests that among MSMWH who acquired HCV sexually, higher viral load of HIV was associated with lower odds of spontaneous clearance of HCV [38]. Our findings are consistent with this suggestion that disparities in severe immunosuppression are associated with disparities in cancer risk.

Compared to White MSMWH, liver cancer incidence was significantly greater for Black MSMWH regardless of AIDS status, and for Hispanic/Latino MSMWH who had not progressed to an AIDS diagnosis. Prevention is paramount to reduce liver cancer risk. An effective HBV vaccine exists, yet vaccination rates among MSM are low, especially among older age groups and Black and Hispanic/Latino men [39–41]. With curative HCV treatment available, there are still disparities in access, especially among Medicaid enrollees who are disproportionately Black or Hispanic/Latino, resulting in disparities in HCC incidence [36, 42, 43]. Addressing disparities in HBV vaccination, access to HCV curative treatment, and increased access to earlier HIV treatment among Black and Hispanic/Latino MSMWH should be prioritized to reduce disparities in liver cancer among MSMWH.

We interpreted a prior diagnosis of AIDS as a proxy for greater immunosuppression, indicating several barriers to receiving adequate healthcare, including receiving a timely HIV diagnosis and reaching sustained viral suppression. Conversely, our comparator of people who had not progressed to AIDS is imperfect and may incorrectly assume one has sustained viral suppression when they have instead never received a recorded AIDS diagnosis. In theory, this could disproportionately impact those of minoritized groups and the differential misclassification of those without a diagnosis of AIDS. A diagnosis of AIDS is a single event which does not capture whether an individual later restores immune function. Immunosuppression is likely more sustained and severe for those who experience greater barriers to receiving care. More accurate determinations of immunosuppression could have been assessed if we had access to data on adherence to cART and longitudinal CD4 counts, but these data were unavailable. Future research should utilize these metrics. Also, we were unable to estimate cancer risk in racial and ethnic groups that provided too small samples or were not harmoniously ascertained between cancer registries and HIV surveillance. Thus, we were unable to include other racial/ethnic groups, but this information is vital in understanding disparities.

Cancers that are associated with a diagnosis of AIDS continue to contribute to morbidity and mortality in the aging population of people with HIV. However, improvements in HIV prevention and care can markedly reduce the incidence of certain cancers. Disparities in cancer risk may serve as an important proxy for disparities in HIV care. Elevated incidence of HIV among Black and Hispanic/Latino MSM contributes to racial inequities in cancer risk, especially for virus-associated cancers as examined in this study. Equitable access to cART will reduce racial and ethnic disparities for KS and NHL, which are strongly associated with greater immunosuppression. We further emphasize the importance of monitoring the trends in both the absolute and relative risk of these cancers within racial and ethnic groups as a potential proxy for assessing disparities in HIV care.