Population-Level Patterns of Prostate Cancer Occurrence: Disparities in Virginia

Abstract

Prostate cancer is the most common cancer and the second leading cause of cancer-related deaths among men in the United States. In Virginia, which is a representative, ethnically diverse state of more than 8 million people that was established nearly 400 years ago, prostate cancer has the highest rate of new detection for any type of cancer. All men are at risk of developing prostate cancer regardless of demographics, but some men have an increased mortality risk due to cancer metastasis. Notably, one in five African American men will be diagnosed with prostate cancer in their lifetime and they have the highest prostate cancer mortality rate of any ethnic group in the United States, including Virginia. A person’s genetic profile and family history are important biological determinants of prostate cancer risk, but modifiable environmental factors (e.g., pollution) appear to be correlated with patterns of disease prevalence and risk. In this review, we examine current perspectives on population-level spatial patterns of prostate cancer in Virginia. For context, recent, publicly available data from the Centers for Disease Control and Prevention are highlighted and presented in spatial format. In addition, we explore possible co-morbidities of prostate cancer that may have demographic underpinnings highlighted in recent health disparity studies.

INTRODUCTION

The epidemiology of human prostate cancer (PCa) is undergoing rapid transformation. Improved access to screening sites, innovative therapeutic strategies and advances in tumor biology have led to worldwide reduction in cancer mortality and improved clinical care of PCa patients. Yet, disparities in PCa etiology persist as outcomes for African American (AA) men remain consistently worse than for other populations [1–3]. The disparity in occurrence and outcomes of PCa among the AA population has been linked to many factors including personal lifestyle, socio-economic status, biology, and environment. In particular, proximity to industrial sites that release carcinogens may also predispose populations to higher occurrence of PCa [4].

In Virginia, the PCa burden is concentrated in the southeastern region of the state, and the Elizabeth River is close to locations where the highest numbers of PCa diagnosis are made [5,6]. Considered a PCa hotspot, samples of soil and water obtained from locations adjacent the Elizabeth River contain both metallic and organic carcinogens [5]. Military and industrial installations may be contributing to the environmental contamination observed, although there are no definitive studies that directly link these facilities with elevated cancer rates. For example, it is conceivable that there could be a relatively stable resident population derived from a limited number of original settlers that passed on an unknown predisposing genetic condition. Nevertheless, a considerably more compelling hypothesis is that the heterogeneous distribution of carcinogenic contaminants highlighted above plays a key role in disease occurrence. This environmental condition represents a much more tangible and controllable risk factor that may have predisposed some Virginia residents to higher risks of PCa than in other parts of the state.

Important examples of pollution sources have already been designated as Federal Super Fund Sites in Virginia with steps taken to prevent human contact with the harmful contaminants found in these sites, which include metals and a range of organic compounds. The latter encompasses polychlorinated dibenzo-p-dioxins (PCDDs; dioxins), polychlorinated biphenyls (PCBs), and various pesticides. Importantly, these metals and organics represent endocrine disruptors that are known to affect both fetal and post-natal development of tissues in animals and humans. For example, high concentrations of metal and organic substances along with deformed biota (fish) exposed to river contaminants have been reported in the Hampton Roads area, a part of Virginia known for the one of the highest concentrations of PCa burden in the state [5,6].

Furthermore, mapping of Virginia PCa registry data [based on the Center for Disease Control (CDC) 2013–2017 data; QGIS] reveals a noticeable overlap between PCa hotpots (highest rates per 100,000 people), and locations with higher numbers of superfund sites (Figs. 1 and 2). Although this relationship remains to be formally tested by quantitative modelling of PCa distribution, the intersection of these two information sources is particularly intriguing because it suggests a connection between industrial pollution and PCa prevalence in Virginia. In this qualitative review, we examine current perspectives on population-level spatial patterns of PCa with a focus on environmental correlates. In addition, we discuss the use of geographic information science to inform demographic and spatial underpinnings in health disparity research.

Fig. 1.

Virginia GIS data illustrating the rate of prostate cancer per hundred thousand individuals (per CDC, 2017). Darker shades correspond to higher rates of prostate cancer occurrence and counties are indicated by name.

Fig. 2.

A heat map of the distribution of identified Federal Superfund Sites. Darker red shades indicate regions with greater numbers of known Federal Superfund Sites.

POPULATION LEVEL PATTERNS

Rates of new PCa diagnoses and PCa-related deaths are declining worldwide, including in Virginia [7], which is likely due to improvements in early detection and availability of more effective treatments. Despite global gains, declines in reported rates of PCa are not evenly distributed among all populations. Illustrating regional differences, a European Randomized Study of Screening for Prostate Cancer (ERSPC) showed significant reductions in PCa mortality, while an alternative study by the US Prostate Lung Colorectal and Ovarian (PLCO) trial did not observe reductions in PCa mortality in the US as observed in a global dataset [8]. Hence, PCa mortality appears to differ based on geographical location.

Innovations in treatment, imaging/diagnostics, and molecular characterization for advanced PCa have improved outcomes, but some aspects of disease management still require additional information to guide clinical practice [9]. Currently, patients experience delayed progression and longer overall survival compared to PCa patients diagnosed in previous years [10]. However, these improvements in PCa patient care do not necessarily translate into lower disease burden. When indolent cancers are treated with surgery or radiation, complications including impotence or incontinence may occur and if PCa is managed with active surveillance, up to 45% of men will eventually receive some form of treatment. These men will unnecessarily face frequent examinations, repeated invasive biopsies, and anxiety [11], thereby increasing PCa burden without favorably improving morbidity.

In the US, PCa disproportionately affects a specific subset of the population and this is particularly evident in Virginia: AA men have worse prognostic outcomes for PCa than Caucasian American (CA) men [12]. Based on clinical data obtained for other ethnicities combined or limited access to appropriate clinical resources, inappropriate treatment strategies may exacerbate observed PCa disparities. Although treatment options for PCa are more diversified and effective leading to higher survival rates overall, PCa survivors sometimes experience post-treatment effects such as incontinence, erectile dysfunction, hot flashes, sweating, and fatigue [13]. In addition, in about 40% of cases post-treatment PCa will recur [13]. If AA men are subject to suboptimal treatment regimen, then such post-treatment effects would further diminish success rates in an ethnically biased pattern. It is critically important to improve diagnosis and treatments for PCa patients and survivors [12]. However, it is equally urgent to address the causes of the health disparity that exists between AA men and other ethnicities, because eradicating these causes may yet yield the greatest gain in health improvements for PCa patients.

DISPARITY IN PROSTATE HEALTH

A person’s environment influences the likelihood of surviving PCa [14]. The context of location is important for health because it influences risk of exposure to disease, impacts ability to adopt a healthy lifestyle, and determines access to quality healthcare within a community (National Cancer Institute). Effective health programs integrate geography and demography because both contribute to a clear spatial assessment of the disease prevalence within a population. Recent GIS-based studies focusing on the PCa health disparity revealed a marked geographic pattern in the US that is significantly correlated with accessibility to screening and treatment health care facilities [13]. Significant racial and socioeconomic differences in cancer outcomes also correlate with patient residence location [2,14]. Rural-urban divides in cancer outcomes have been observed [13,15] including in Virginia, where the burden of PCa is concentrated in the southeastern part of the state around Hampton Roads (suburbs) and the Eastern Shore (rural) relative to more urbanized areas (Fig. 1).

Tracking spatial-temporal distributions in PCa can be challenging due to a complex multitude of factors (many of them correlated) involved in its etiology. In this regard, geographic information systems (GIS) data have been particularly useful for mapping health data in epidemiological studies. GIS data provide visual information to aid in identifying hotspots and enable resources to be deployed in areas where they are most needed [16,17]. Additionally, GIS data can be used to verify whether mapped quantities of multiple contributing factors coincide spatially and therefore locate places of special concern [18]. For example, Wang and colleagues [19] examined temporal and spatial trends in the diagnosis of highly aggressive PCa among men in Pennsylvania and found a statewide increase for both AA and CA men between 2004 and 2014. Interestingly, they also observed higher incidences of aggressive PCa in AA men, especially among AA men living in counties predominantly populated by CA. These findings suggest a bias of health care received among AA men in counties that are predominantly populated by CA.

Another variable revealed by GIS analyses in PCa prevalence is access to healthcare facilities. For PCa, this factor may contribute up to 30% in observed variation in PCa mortality and ~40% of disease recurrence following initial treatment [13]. The distance from a patient’s residence to healthcare facility impacts access to screening and follow-up [20]. Areas with limited access to screening and treatment showed higher levels of late stage PCa diagnosis, predisposing patients to fewer treatment options and increased mortality. Therefore, public health efforts should focus on areas with limited access to healthcare as a high-return strategy for narrowing the health disparities [13]. Insofar possible, planning the spatial location of a screening/health facility should specifically minimize travel times (e.g., bus, train, car) to the facility for an entire population, irrespective of spatial economic patterns. In principle, improved transit infrastructure should help solve the most basic problem of health care access disparities that are accentuated by long travel times.

Although valuable, spatial analyses of PCa can be complicated by limited data access. Public policy often requires protection of patient privacy and therefore limits the resolution of epidemiological studies to higher area scale units. This inadvertently introduces aggregation bias in spatial analyses of public data, smoothing out spatial heterogeneity and reducing the ability of researchers to detect relevant patterns [18]. The choice of area unit is therefore crucial in spatial studies. In epidemiological research, the most often used area unit is the census area unit – other commonly used population-based boundaries include zip codes, census tracts and census block-groups [18,21,22]. Results from GIS research are highly susceptible to scale and other methodological choices made by geographical researchers (e.g., proximity or coincidence; [23]). In fact, contradictions in GIS-based disparity studies are often related to the modifiable area unit problem (MAUP; [24]), whereby different boundary systems may be incorporated into an analysis. This problem limits the scalability of a study and hampers transferability of results among geographic locations. Methodological strategies must be repeated in each area of interest, which may not have matched data types available (e.g., zip codes in the US may function somewhat differently than in other countries). To overcome this problem, spatially sensitive strategies highlighting regions of greatest need for resolving factors related to PCa health disparities must be formulated. By extension, fine-scale approaches should converge on the ultimate goal of not only reducing the PCa health disparity, but also reducing the overall PCa burden in the US (including Virginia).

CANCER OCCURRENCE IN VIRGINIA AND POTENTIAL LINKS WITH ENVIRONMENTAL FACTORS

According to the CDC, the rate of new PCa has stabilized, PCa-related deaths are declining, and the five-year relative survival rates for CA, AA and other groups in the state of Virginia are 97.7%, 95.6% and 94.3%, respectively. Each year from 2013–2017, there were nearly 168 new cases of PCa per 100,000 people and 39 PCa-related deaths per 100,000 deaths. During the same period in Virginia, AA men were almost twice as likely (1.73 times) to get PCa than CA men, twice as likely as Hispanics and more than three times as likely to get PCa than Native American and Asian men. The rates of new cancer cases per 100,000 are presented for Virginia as follows: The Eastern Shore (Accomack 161 cases) and Northampton (67 cases) counties bear the highest burden for rate of new PCa cases. In addition, PCa burden is remarkable in Hampton Roads where Norfolk (597 cases), Suffolk (321 cases), and Chesapeake City (690 cases) all have higher than average cases for the state. Other areas of notable concern are Petersburg (184 cases), Lynchburg (190 cases), Warren county (102 cases), Westmoreland County (85 cases) and Brunswick county (95 cases) (Data from CDC; Fig. 1). Hence, the southeastern counties in Virginia appear to have a disproportionally high occurrence of new PCa cases.

Near Hampton Roads, the Elizabeth River is a natural feature, surrounded by industrial and military installations. This area holds a disproportionate number of PCa diagnoses in Virginia (Fig. 1). The river is designated a “toxic hot spot” with very high contaminant loads from metals and organic compounds in sediments including cadmium, chromium, copper, lead, and iron. Areas with high industrial and military activities, such as the ones adjacent to the river, have high concentrations of contaminants [5], including radioactive materials like radium-226. Additionally, metal forming installation, and repair of military equipment generate contaminants that are linked to PCa such as barium [25], cadmium [26–28], copper [29,30], lead [27,31], nickel [31], PCBs [32], volatile organic compounds [33], chromium [34], mercury [27], selenium [31], iron and silver [35], as well as zinc [26]. Stratifying the relative importance of each contaminant in the etiology of PCa is a challenging task, but carcinogen exposure has been linked to PCa risk, although correlations have yielded varying degrees of association [4]. It should be noted that at least some abundant industrial pollutants by themselves may not be carcinogenic in a strict sense. Yet, such contaminants could substantially enhance the tumorigenic effects of known carcinogens, including those present in low quantities that would preclude their detection in environmental samples.

The link between carcinogen exposure and the alteration of cell proliferation is well established, because samples taken from PCa patients often contain traces of such carcinogens. Notably, higher levels of bisphenol A (BPA) were reported in urine samples from PCa patients relative to controls [30]. Moderately chlorinated PCBs are strongly associated with PCa risk and pesticides such as simazine, dimethoate, have immunosuppressive and endocrine disrupting activities that render these compounds carcinogenic [36,37]. The status of Glyphosate (a heavily used herbicide in the US) as a probable carcinogen is currently being reviewed but will have important implications in PCa analysis because of the high number of people (including AA men) living in areas where this herbicide is used [38,39].

Heavy metals have also been significantly associated with PCa. In general, the mechanisms of heavy metal action toward carcinogenesis in cells include accumulation of reactive oxygen species, induction of migration/invasion, and autophagy deficiency, which increases the likelihood of tumorigenesis [28]. These general cancer-related mechanisms may therefore perhaps also influence PCa initiation and/or progression. High levels of Fe, Cu, and Mn in the hair and nails of patients were shown to be associated with an increase in PCa risk [40]. Further, exposure to Cd increases expression of survival proteins, while downregulating genes for proteins that control apoptosis and modulate carcinogenesis [41]. Interestingly, aluminum, which is detected in some populations at high levels, may be carcinogenic because it has endocrine disrupting properties [42]. Antimony (often associated with battery fabrication) was also found to be present in high concentrations in patients with PCa compared to PCa-free individuals [43], suggesting it may have a role in PCa etiology. Whether the roles of heavy metals in the development or progression of PCa have an ethnic bias is a topic that may also be explored via spatial analyses.

A high occurrence of PCa in Hampton Roads may be related to the presence of environmental carcinogens. Interestingly, several sites in Hampton Roads are already designated as Federal Superfund Sites where contaminants are intentionally sequestered. Despite containment efforts, these sites may still pose environmental risks to adjacent populations if carcinogens leach into underground and surface waters. A GIS-based study analyzed environmental contaminant data to determine whether proximity to hazardous substances predispose populations to higher rates of PCa [4]. The results were not yet fully conclusive, and hence additional studies will be required to iteratively identify direct correlations between key variables.

Additionally, research should focus on interactions among levels of pollution, PCa outcomes, and socio-demographic information within areas of Virginia that exhibit higher levels of PCa, to identify geographically linked causes of PCa that can be legislatively controlled [13]. Importantly, minority groups and socio-economically disadvantaged populations are at a higher risk of hazardous land uses in their local environment, in part due to the systematic marginalization of low-income neighborhoods. Legislation and policy to protect these groups are essential to reduce the burden of PCa among such populations, especially where housing and employment options have been historically constrained [23].

OTHER FACTORS IN PCa DISPARITY

AA men are disproportionately affected by number and rate of new PCa cases [12], and are twice as likely on average to have PCa as men from other ethnic groups in Virginia. Over the last four decades, studies have revealed a disproportionate effect of ethnicity and income, which may in part explain the higher prevalence of PCa within AA populations [23]. In fact, higher PCa occurrence in AA communities has been related to a multitude of factors such as behavior, diet and lifestyle [44–46]. There are cursory connections between diet and PCa risk/prevention [47]. For example, Vitamin D deficiency may be associated with increased PCa risk in AA men compared to CA men [44], but the multifactorial causes of PCa may reduce the relevance and phenotypic penetrance of these connections. For example, access to early PCa screening is arguably one of the most significant causes for disparity.

Assessing baseline prostate-specific antigen (PSA) levels during midlife is crucial for risk prediction of aggressive PCa in AA men [48]. However, AA men are often diagnosed with later-stage PCa than other groups and consequently experience worse outcomes [12]. Socio-demographic factors influence degree of medical mistrust among AA men [46]. A mistrust of the healthcare system (and lack of participation in clinical studies) can contribute to increases in risky behaviors that lead to poor outcomes for PCa among AA men. A systematic review of clinical studies evaluating the association between environmental exposure and male reproductive health in the U.S. revealed that > 80% of participants were CA and only 2–10% were AA [49]. A similar trend for inclusion of AA men in clinical trial participation has been observed [3]. Targeted outreach efforts are therefore needed to address disparate cancer screening and involvement of racial minorities, especially AA men that bear a disproportionately higher burden of PCa, and PCa-related deaths [45].

The development of innovative “risk alert” tools that can be widely accessed by the public at large has gained traction and is already making a positive contribution towards mitigating PCa disparity. For example, an online PCa risk assessment tool developed by the Prostate Biopsy Collaborative Group was designed for doctor-patient decision making and validation based on the analysis of many samples throughout Europe and the US. This tool makes use of multivariate logistic modeling approaches built by pooling data from several cohorts of PCa patients and can assign each patient into risk categories based on their ethnicity, age and family history [50]. As an example, higher PSA and age, abnormal digital rectal exam, African ancestry, and family history of PCa increases risk of high-grade PCa, while a history of prior negative prostate biopsy decreased the overall risk. This bioinformatics approach provides an interesting clinical perspective on the health disparity between AA and CA men in the US, as it recognizes an elevated risk based on African ancestry alone (it does not consider location of residence). With advances in artificial intelligence (AI) methodology and machine-learning approaches, in combination with other spatial information embodied in GIS data, it is likely that publicly available risk alert tools for PCa will further evolve and hopefully reduce barriers to PCa study participation, clinical detection and therapeutic treatment.

Also of potential relevance, obesity is associated with advanced PCa, as well as poor PCa prognosis [51]. Other considerations include genetic testing and genetic counselling [52], yet as with many multifactorial diseases, the treatment strategies should focus on controlling and limiting the co-morbidities most associated with disease initiation and progression. Similarly, a careful consideration of ethnicity should be included in the most appropriate plan to eliminate or reduce rates of PCa and PCa health disparities.

CONCLUSIONS

High global occurrence of PCa is an incentive to support cutting edge research and provide solutions to the problems of PCa faced by all men. Data sources for PCa including PCa registries and research institutions must maintain good quality data to support this goal (e.g., [47]). A new frontier in PCa data research should involve the use of spatial intelligence and GIS for determining PCa hotspots as focal points for the deployment of resources for PCa mitigation, and to help determine causes of health disparities. Effective solutions are especially needed for populations that are most impacted by PCa, such as AA men that are disproportionally affected by PCa in Virginia.

There is clearly a geographic component that compounds with related socioeconomic factors that predispose economically disadvantaged people and minorities (including AA men) to live in locations where there is higher likelihood for cancer. Our review of registry data reveals that in Virginia, these locations also overlap with areas where most Super Fund Sites are located in the state, suggesting that formal analyses involving environmental data deserve further exploration. We predict that combining geospatial and bioinformatics approaches for the identification of PCa risk should complement ongoing medical advancements to not only reduce the worldwide burden of PCa but also narrow the disparity between AA, CA and men from other racial groups in the US and Virginia.

Future studies should also consider the broader realm of cancer disparities. Because PCa is male specific, it will also be of interest to consider breast cancer (BCa) risk in the same geographic areas to assess whether environmental pollutants or other regional differences also affect BCa in AA women in a statistically and epidemiologically similar manner. While absence of a correlation between PCa and BCa would not be definitive, a positive correlation between the two could be particularly informative, because it would establish the same trends for two biologically distinct cancers in both genders. In addition, as cancer patient databases become more complete and refined, it will be of interest to expand the current concepts for PCa disparity towards other cancers that could be disproportionally affected by regional parameters and environmental factors like gastro-intestinal, pulmonary and hematopoietic cancers (e.g., leukemia and lymphoma).