Prostate cancer (PCa) is the most common cancer among men in the United States. In 2021, Black men are projected to experience 1.8 times the incidence and 2.1 times the mortality as non-Hispanic White (NHW) men (1). Prostate specific antigen (PSA), Gleason score (GS), and stage are prognostic factors of PCa, which together can be used to indicate PCa risk profile for clinical decision-making (2). Black men with PCa present with higher PSA levels on average compared to other racial groups, and for a given level of PSA, Black men have larger tumor volumes than NHW men (3-6). In California, we previously reported a 60% higher age-adjusted PCa mortality among Black men relative to NHW men, which was reduced to the null after adjustment for tumor, sociodemographic, institutional, and neighborhood characteristics (7). However, it remains unclear what factors contribute to the higher risk of advanced PCa at diagnosis among Black men.

Social determinants of disparities in cancer outcomes (e.g. clinicopathologic presentation, diagnosis, treatment, and survival (8)) are complex and intersecting. What we observe as racial disparities—adverse health consequences of racism for historically marginalized racial groups—may partially reflect overlapping inequities across other social factors associated with cancer outcomes (7,9-13). Individuals who reside in resource-poor settings as measured by low neighborhood socioeconomic status (nSES) are more likely to experience social isolation, stressors, have reduced access to medical and social services (14-22) and also experience disparate PCa outcomes (2,7,23). In addition, patient-level social factors including health insurance status type and marital status are associated with disease stage and mortality (7,24). Determining the relative contribution of intersecting social factors (at the individual- and contextual-level) to PCa risk is an area of ongoing investigation (2,7,23,25). To our knowledge, only one study has evaluated associations between these factors (in addition to race) and risk of advanced PCa at diagnosis (2). Elucidating these factors may provide insight to mitigate the Black-White racial disparity in PCa survival. To further examine intersecting social factors that may impact PCa risk profile, we conducted a population-based study using California Cancer Registry (CCR) data of Black men and NHW men in California with PCa from 2004 through 2017 with detailed information on PSA, GS, and stage, in addition to information on individual characteristics (age at and year of diagnosis, marital status, insurance status) and nSES.

In order to examine the extent to which social factors contribute to racial disparities in PCa risk profiles as defined by PSA, GS, clinical stage, and combined risk, we conducted a population-based study of all Black men and NHW men diagnosed with first primary invasive PCa in California from 2004 through 2017. NSES was the most influential factor contributing to age-adjusted racial disparities in PCa risk profile among Black men relative to NHW men for PSA, GS, stage, and combined risk; followed by marital status. Future studies are needed to elucidate the role of nSES and marital status in PCa risk profile at diagnosis. Specific areas of interest include a careful consideration of what nSES may be measuring, how that may be driving racial disparities in high PSA PCa in particular–including an increased understanding of the intersection of social factors of racism, social isolation, social stressors, and specific neighborhood factors–and whether such racial disparities in high PSA PCa are associated with worse survival outcomes after controlling for stage and GS.

Increased serum PSA, an androgen-regulated glycoprotein molecule involved in the liquefaction of seminal fluid, is a prognostic factor of PCa risk (34-36). Higher PSA among Black men relative to White men has been under investigation for decades (3,5). Some have speculated that higher PSA levels among Black men in comparison to NHW men at diagnosis of non-metastatic PCa may be due to higher tumor cell burden or screening detection later in the clinical course (i.e., differential disease courses) (37). Tumor volume, inflammation, number or cores positive and percentage of the higher versus lower GS components are reasons why patients could have the same stage and GS but a higher risk profile PCa. Other non-malignant clinical correlates of increased serum PSA include increasing age, larger prostate volume, infection or trauma to the prostate, and medical procedures that interfere with the prostate gland (38-42). It is therefore unclear whether high PSA among Black men compared to NHW men at diagnosis in our study is due to differences in underlying tumor biology, PSA expression or racial differences in PSA-based screening.

There are biological differences in high PSA at PCa diagnosis, which warrants further study since the majority of genetic research on PCa to date has been conducted among White men. Recent work has indicated that certain Kallikrein polymorphisms are associated with PSA levels in Black men but not NHW men (43). Additional studies are necessary to elucidate mechanisms for biological differences in PCa risk profiles observed between Black men and NHW men that are due to genetic ancestry versus social factors. Moreover, biological differences may be due underlying germline genetics or epigenetic expression in response to the embedding of racism. An example of one such study that considers the full breadth of factors that may contribute to racial disparities in PCa outcomes is our national, multicenter study; Research on Prostate Cancer in Men of African Ancestry: Defining the Roles of Genetics, Tumor Markers and Social Stress (RESPOND), currently in the field. In RESPOND, we are conducting the largest coordinated research effort to date to study multi-level determinants of enduring racial disparities in PCa among US Black men. We conceptualize increased risk of aggressive PCa and PCa mortality among Black men in the US as a combination of underlying germline genetics and the experience of racism through individual- and neighborhood-level social stressors across the lifecourse that “get under the skin” to cause biological vulnerability in somatic profiles, tumor inflammation, and other potential mechanisms.

Disparities in PSA-based screening across populations may also contribute to high PSA among Black men at PCa diagnosis. Although best available crude prevalence estimates for PSA screening are comparable between Black men and NHW men of age 40+ years in California (44,45), these estimates may not accurately reflect PSA screening prevalence for more recent years included in our study, or relevant PSA screening behavior. Multiple PSA tests need to occur to detect PCa early and it is unknown how many PSA tests Black men receive over time relative to NHW men. In our study, delayed diagnosis among Black men relative to NHW men resulting from differential longitudinal screening frequencies may have had a stronger impact on rates of high PSA at PCa diagnosis than on high GS at diagnosis since PCa among Black men tend to produce more PSA per tumor volume. Higher GS is generally considered to be less susceptible to early detection. We performed sensitivity analyses to examine whether Black-NHW disparities in high risk profile PCa were driven by changes in screening behavior following the 2012 United States Preventives Services Task Force (USPSTF) recommendation that clinicians should not screen men who do not express a preference for screening (46). We observed fewer men with PCa and proportionally more advanced PCa diagnosed in 2013-2017 than in preceding years for both racial groups, but adjustment for a period effect of year of diagnosis 2004-2012 vs. 2013-2017 did not result in changes in Black-NHW odds ratio estimates for PSA, GS, stage or PCa Risk of more than 1% (Supplemental Table 5). This suggests that our main findings were likely not impacted by changes in screening behavior following the 2012 USPSTF recommendations.

Our study is strengthened by the legal mandate in California for routine collection of tumor and patient characteristics on all persons with cancer in California. As part of the SEER registry program, component registries of the CCR meets stringent standards for quality, timeliness and completeness. Hence, our study is less prone to reporting and selection biases than studies within specific healthcare systems or patient populations. We utilized composite indices for measuring nSES that did not require patient report, which enabled us to provide evidence that nSES is a primary contributor to Black-NHW disparities in advanced PCa.

Our study was subject to limitations common to cancer registry-based analyses including lack of individual-level data on SES, family history of cancer, and lifestyle factors. We were unable to control for obesity, which is positively associated with GS (47), negatively associated with PSA (48), and disproportionately high in California for Black men relative to NHW men (e.g., 35.4% vs. 25.2%, respectively, in the California Health Interview Survey) (49). Another limitation of SEER data is inability to identify whether men were diagnosed with PCa following routine screening or based on a symptomatic indication. An additional limitation of cancer registry data is lack of information regarding prior residences or length at residence, which may be related to PCa risk profiles. We also lacked data on tumor volume, prostate size, and number and involvement of biopsy cores. Furthermore, our single PSA test result lacked potentially important information such as PSA kinetics (PSA velocity and doubling time) and free-to-total PSA ratio; potentially important predictors of PCa risk profile (35,36,50). Additionally, we observed some indication of non-random missingness in PCa risk profile outcome variables. The largest proportion of missing PCa risk profile data was observed among men with unknown or missing insurance status. However, men with missing or unknown insurance status comprised <5% of the overall sample. A recent study examining potential exclusion bias due to missing data when grouping prostate cancer cases using this D’Amico risk stratification in SEER data, found that tumor characteristics among men with missing PCa risk profile data were similar to those with complete data for risk profile (51). Our observation of <1% differences in the ORs between our models with and without multiple imputations provided evidence that our findings were likely not biased due to missing data.

Our findings suggest that racial disparities in high risk PCa among Black men relative to NHW men may be influenced in part by differences in nSES. These findings are consistent with previous findings of high PCa risk profiles for men in the lowest nSES quintile (2) and worse PCa survival among Black men than White men, which was attenuated by adjustment for nSES (23). Our findings that marital status contributes to racial disparities in high risk PCa among Black men relative to White men is consistent with findings that marital status is an independent and strong predictor of PCa survival, and a moderator of racial disparities therein (7,24). We interpret our findings for nSES and marital status in light of emerging evidence that PCa develops through complex interactions at the biological, individual, and social levels (10,23,52,53). Further work is necessary to elucidate potentially relevant adverse exposures among Black men residing in low nSES neighborhoods, such as perceived racism and social stress, how such factors may contribute to high risk PCa profiles. Furthermore, the association between high risk PCa at diagnosis and unmarried status warrants additional investigation, possibly to inform social support resources for health disparities populations.