Successful recruitment of healthy African American men to genomic studies from high-volume community health fairs
Implications for future genomic research in minority populations
Study of genomic data obtained from patient biospecimens is frequent in research of subjects with prostate and other epithelial malignancies. Understanding of the characteristics of healthy men who participate in genomic research is limited.
Patients were identified through the Prostate Cancer Genetic Risk Evaluation of SNPs Study and the Indiana University Cancer Biomarker Study, 2 population-based biomarker and cohort studies. Between 2006 and 2010, healthy Caucasian (n = 774) and healthy African American (n = 381) men were recruited and enrolled at high-volume free community health fairs. Each participant completed a demographic questionnaire and provided a blood sample for genomic research investigations. Frequency differences between demographic features of healthy African American and Caucasian men were compared and analyzed by 2-sample t test and multivariate logistic regression after adjusting potential confounding variables with significance at the P < .05 level. Features examined included: age, body mass index (BMI), income, education, marital status, tobacco, alcohol, family history, prostate-specific antigen (PSA) level, and prior prostate cancer screening history.
Significant differences between healthy Caucasian and African American men participating in genomic research included: marital status (married, 69% Caucasian vs 46% African American, P< < .001), mean age (years, 58 Caucasian vs 54 African American, P < .001), mean BMI (kg/m2, 30.9 Caucasian vs 32.3 African American, P = .004), annual income (P = .038), education (P = .002), and mean PSA (ng/mL, 1.2 Caucasian vs 2.0 African American, P = .005).
Significant demographic differences exist between healthy Caucasian and African American men choosing to participate in genomic research. These differences may be important in designing genomic research study recruitment strategies. Cancer 2012;. © 2011 American Cancer Society.
With the completion of the human genome project, high-throughput genomic analysis platforms now permit identification of candidate cancer risk genes in large population studies.1-3 This technology may be particularly helpful in diseases such as prostate cancer, where no dominant singular oncogenes have emerged. A recent genome-wide association study identified 5 genetic variants in chromosomal regions (8q24, 3 regions; 17q12; 17q24.3) that independently or cumulatively confer susceptibility to prostate cancer.4 Although individuals with variants at all 5 loci possess a nearly 10-fold increased risk of developing prostate cancer, none of the loci demonstrated significant associations with aggressive prostate cancer. Thus, additional genomic studies in varied populations are likely required to develop an effective genomic screening mechanism to detect individuals at risk for lethal prostate cancer.
Given the complex contributions to carcinogenesis from both gene and environment, a critical need exists to recruit healthy individuals to prospective genomic cohort studies. At present, our understanding of the patient demographics of healthy individuals who choose to participate in genomic cancer research studies is limited. The current study was performed to characterize such patients and to test the experimental hypothesis that significant demographic differences exist between healthy African American and Caucasian men participating in genomic research studies.
MATERIALS AND METHODS
Prostate Cancer Genetic Risks Evaluation of SNPs Study and the Indiana University Cancer Biomarker Study
The Prostate Cancer Genetic Risks Evaluation of SNPs Study (PROGRESS) is a prospective biospecimen and clinical data collection protocol initiated at the Indiana University Melvin and Bren Simon Cancer Center in December 2006. The protocol is fully approved by the Indiana University Melvin and Bren Simon Cancer Center institutional review board (IRB), and all subjects provided written informed consent to participate in agreement with Declaration of Helsinki principles. All investigators completed the Indiana University Melvin and Bren Simon Cancer Center IRB Human Subject Protection Test before receiving consent from study subjects. The initial aims of the PROGRESS study are to identify biomarkers predictive of lethal prostate cancer risk in healthy men and markers predictive of response to therapy in men with prostate cancer. In addition, the data and samples collected in PROGRESS may be used to study other nonprostate diseases. All participants complete a clinical and demographic data questionnaire. In addition, peripheral blood samples (2 × 10 mL polypropylene tubes, 2 × 10 mL ethylenediaminetetraacetic acid tubes) for candidate biomarker identification were obtained from each patient. Multiple follow-up visits and sample collections were not performed for healthy subjects.
The Indiana University Cancer Biomarker Study (IU-CABS) is a more extensive prospective biospecimen and clinical data collection protocol built off of the PROGESS study experiences. All of the data fields and peripheral blood sample collections described above for PROGRESS were incorporated into the IU-CABS protocol. In addition, the IU-CABS protocol was expanded to collect samples and clinical data from healthy men and women or subjects from both sexes with any genitourinary malignancy. The IU-CABS protocol is fully IRB approved, required documented written informed consent from all study participants, and also complies with all research principles within the Declaration of Helsinki. The IU-CABS protocol opened to enrollment in July 2008. Since that date, all subsequent healthy subjects and genitourinary cancer patients have been enrolled through the IU-CABS program, whereas all prior prostate cancer patients enrolled through the PROGRESS study continued to be followed for outcomes data.
Subjects and Recruitment
Participants were identified and enrolled between December 2006 and July 2010 in multiple clinical settings located in Indianapolis, Indiana. Healthy subjects were enrolled at high-volume free community health fairs including the Indiana Black Expo and the Indiana State Fair. The Indiana Black Expo is a 2-week celebration of African American culture with an average of 250,000 attendees held in downtown Indianapolis with a 4-day free health fair open to all Indiana Black Expo attendees.5 The Indiana State Fair is a 10- to 14-day fair attended annually by >900,000 individuals.6-9 The Indiana State Fair focuses on agricultural and economic opportunities and innovations within Indiana and offers a free health fair conducted on all days. Healthy subjects were approached by research team members in the setting of an exhibit booth within the health fairs at both Indiana Black Expo and Indiana State Fair. Additional healthy subjects were recruited at smaller community health fair venues throughout the study period. Patients with prostate and other genitourinary cancers were recruited through the multidisciplinary genitourinary oncology clinics (oncology, urology, radiation oncology) within the Indiana University Melvin and Bren Simon Cancer Center system in Indianapolis, Indiana. Subjects received no monetary payment for participation. A small complementary tote bag was provided to each participant.
PROGRESS and IU-CABS healthy subject eligibility included age >18 years with no prior self-reported history of cancer and written informed consent. Individuals aged >18 years with a prior history of a genitourinary cancer who provided written informed consent were eligible to participate as members of the cancer cohorts. Only data on the healthy male subjects are analyzed in this report.
Each participant completed a detailed clinical and demographic questionnaire. Data collected from each participant included: age, height, weight, race/ethnicity, past and current tobacco and alcohol use, education level, household income, marital status, family history of prostate and other cancers, history of known genetic cancer risks (ie, BRCA1/BRCA2), current medications, prostate-specific antigen (PSA) and digital rectal examination (DRE) testing history, surgical history, and prostate biopsy history. National Institutes of Health recommended categories were used to collect data on race/ethnicity, education level, household income, and marital status.10 Tobacco use was categorized as current versus past and quantified as <1 pack per day, 1 to 2 packs per day, and >2 packs per day. Alcohol use was categorized as current versus past and quantified as drinks (1 beer equal to 1 mixed drink) per week. Family history of prostate cancer and known genetic cancer risks were categorized as yes versus no. Medications were grouped by class (eg, beta-blockers, antihyperlipidemic agents). PSA, DRE, surgical, and prostate biopsy history were identified as yes versus no, and results were recorded. In a subset of patients who had PSA testing performed by third-party providers participating in the health fairs, PSA values were available. Body mass indexes (BMIs) for each patient were calculated for each patient according to the formula: BMI = weight (kg)/(height [m]2).
Continuous variables were dichotomized according to the median values. Categorical variables were analyzed within each category and collectively. Univariate logistic regression testing was performed to compare demographic data (categorical variables) from African American and Caucasian men. A significance threshold of P = .05 was used in the univariate model to select variables for inclusion in the multivariate analyses. Multivariate logistic regression models were performed controlling for potential confounding covariates of interest, including age, socioeconomic status (income, education), and marital status. Odds ratios with 95% confidence intervals and P value from logistic regression were reported. Descriptive statistical methods were used to demonstrate distributions of continuous variables. Two sample t tests were done to test the significance of the relation between race and age, BMI, and PSA values (continuous variables). Statistical analyses were performed with SAS statistical software 9.1 (SAS Institute Inc., Cary, NC). Differences between demographic data from African American and Caucasian men were tested at a significance level of P = .05.
Between December 21, 2006 and July 18, 2010, 2349 subjects were screened at high-volume free community health fairs and Indiana University Melvin and Bren Simon Cancer Center clinics in Indianapolis, Indiana. At total of 2305 subjects provided consent and enrolled in the PROGRESS (1243 patients) or IU-CABS (1062 patients) studies. Patients excluded from analysis included: 301 healthy female participants, 603 patients with a confirmed cancer diagnosis, and 46 subjects with a self-reported race/ethnicity other than Caucasian or African American. The final analysis cohort was comprised of 1155 subjects, including 774 Caucasian and 381 African American healthy men. Full study details are shown in Figure 1. A comparison of the features of the Indiana Black Expo and Indiana State Fair health fairs is summarized in Table 1.
Table 1. Health Fair Features
|Dates||July (4-day weekend)||August (10-15 days)|
|Location||Indianapolis Convention Center||Indiana State Fairgrounds|
|Setting||Exhibit hall connected to convention center||Free-standing exhibit hall on fairground property|
|Cost||General admission, $8; health fair and testing, free||General admission, $8; health fair and testing, free|
|History||IBE, 25th year; health fair, 10th year||ISF, 154th year; health fair, 1st year|
Questionnaire Data Completion
Complete questionnaire data were available in 89% of Caucasian and 92% of African American subjects. In the remaining 19% of respondents, 1 to 3 questionnaire data fields were incomplete. The most common omitted variables included tobacco use (163 patients), age (54 patients), and DRE history (49 patients). PSA values were available in 294 (40%) Caucasian and 193 (51%) African American subjects.
Several demographic features differed significantly between healthy Caucasian and African American male participants. In general, African American men were younger, of lower socioeconomic status, and more likely unmarried compared with their Caucasian counterparts. The 2 cohorts were similar with respect to other lifestyle factors (alcohol, tobacco), family history of prostate cancer, and history of prostate cancer screening (PSA test and DRE). A full comparison of all demographic features is summarized in Table 2.
Table 2. Subject Demographic Comparisons
|Age, mean y||54.1 ± 1.4||58.1 ± 1.3||<.001|| || |
|BMI, mean kg/m2||32.3 ± 0.85||30.9 ± 0.8||.004|| || |
|PSA, mean ng/mL||2.0 ± 0.4||1.2 ± 0.4||.005|| || |
|Income, %|| || ||.038|| || |
| ≥$75,000||22.2||27.1|| ||Ref|| |
| < $75,000||11.9||19.3|| ||0.75||0.48-1.18|
| <$50,000||12.7||13.4|| ||1.22||0.75-1.96|
| <$35,000||12.1||10.4|| ||1.19||0.72-1.98|
| <$25,000||6.3||4.7|| ||1.35||0.71-2.55|
| <$20,000||5.0||2.0|| ||3.03||1.36-6.75|
| <$15,000||5.3||2.2|| ||2.09||0.93-4.72|
| <$10,000||5.5||2.6|| ||1.65||0.79-3.44|
| No answer||19.0||18.4|| ||1.10||0.70-1.74|
|Tobacco use, %|| || ||.366|| || |
| Yes||40.3||42.0|| ||1.15||0.85-1.56|
| No||59.7||58.0|| ||Ref|| |
|Alcohol consumption, %|| || ||.347|| || |
| Yes||45.4||46.9|| ||0.88||0.67-1.15|
| No||54.6||53.1|| ||Ref|| |
|Education, %|| || ||.002|| || |
| Grade 1-8||0.0||0.5|| || || |
| Grade 9-11||4.7||2.9|| ||1.80||0.85-3.80|
| Grade 12/GED||25.3||23.3|| ||1.40||0.98-2.01|
| College 1-3 years||37.4||28.4|| ||1.61||1.16-2.22|
| College ≥4 years||31.8||39.7|| ||Ref|| |
| No answer||0.8||5.2|| ||0.14||0.03-0.60|
|Marital status, %|| || ||<.001|| || |
| Married||45.8||69.2|| ||Ref|| |
| Divorced||17.6||11.1|| ||2.08||1.41-3.06|
| Separated||4.7||0.5|| ||9.15||3.0-28.24|
| Widowed||2.6||2.4|| ||2.06||0.85-4.98|
| Never married||24.0||9.4|| ||2.52||1.67-3.81|
| Member of unmarried couple||0.8||1.2|| ||0.72||0.19-2.76|
| No answer||4.5||6.3|| ||2.0||0.90-4.31|
|Family history, %|| || ||.410|| || |
| Yes||25.1||28.5|| ||0.84||0.63-1.11|
| No||71.1||67.2|| ||Ref|| |
| I do not know||3.7||4.3|| ||0.82||0.43-1.56|
|PSA test, %|| || ||.827|| || |
| Yes||58.4||70.2|| ||Ref|| |
| No||41.6||29.6|| ||0.90||0.63-1.27|
| Unknown||0.0||0.2|| || || |
|DRE, %|| || ||.402|| || |
| Yes||61.6||70.9|| ||Ref|| |
| No||38.4||29.2|| ||0.86||0.6-1.23|
With respect to specific demographic categories, African American men were younger than Caucasian participants (mean age 54 vs 58 years, P < .001). The mean BMI value was lower for Caucasian than African American subjects (30.8 vs 32.3, P = .004). In addition, African American men were characterized by lower education (P = .002) and income levels (P = .038). The differences in education were observed mainly in the 4-year and 1- to 3-year college level degree categories, with Caucasian men more likely to have a 4-year degree (39.7% vs 31.8%) compared with their African American counterparts. Income differences between cohorts were particularly apparent at annual income levels <$20,000 (16% African American vs 7.0% Caucasian, P = .016).
Furthermore, healthy African American men were significantly more likely to be in a nonmarried relationship than their Caucasian peers. Specifically, only 46% of healthy African American subjects were married compared with 69% of Caucasian men (P < .001). In contrast, 48% of African American men were divorced, separated, or never married, compared with only 23% of healthy Caucasian men (P < .001). Further stratifying the married groups, a smaller percentage of African American men reported being married than Caucasian men across all the age groups. However, a statistical significance was found only at age>60 years (married age 61-70 years: 57% African American vs 81% Caucasian, P < .001; married age >70 years: 67% African American vs 80% Caucasian, P < .001; data not shown in Table 2). A significant difference was observed between the 2 cohorts in actual PSA results (P = .005), with slightly higher values in African American as compared with Caucasian subjects.
Despite improvements in cancer therapeutics, disparities in cancer outcomes still persist, particularly among African Americans. Many factors including access to care, cost, education of prevention benefits, and cultural perceptions of cancer diagnosis have been identified.11-14 Although minority populations are frequently portrayed as less likely to participate in clinical therapeutic research studies, where opportunity exists, minority enrollment rates do not appear any lower than those of non-Hispanic white subjects.15-17
With the completion of the human genome project in 2000, characterizing mechanisms of disease risk and progression at the gene and molecular level is now possible.1 With these technological advances, an opportunity to study health disparities at the molecular level in conjunction with external environmental and societal influences exists for the first time. As such, it is paramount to increase minority representation not only in therapeutic trials, but also in correlative genomic research programs. This is particularly important in light of 2010 statistics demonstrating that African American men have a 14% higher incidence rate and a 34% higher death rate from all cancer sites compared with Caucasian men.18 Similarly, African American women have a 17% higher cancer mortality rate than their female Caucasian counterparts.18 Collectively, African Americans account for 10% of new cancer diagnoses and 11% of cancer deaths in the United States.19, 20 The Census Bureau currently estimates that approximately 40 million African Americans reside in the United States, comprising 13% of the total population.21 Of note, African Americans are a heterogeneous population, with variation in individual ancestries of African and Latin American origin.20 Given these disparities in clinical outcomes, the significant proportion of African Americans within the cumulative US cancer incidence and mortality figures, and the heterogeneous nature of the African American population, a critical need exist to recruit African Americans to genomic research studies to better characterize the biology of cancers within African Americans and thus improve prevention and therapeutic intervention strategies. An understanding of the demographic differences between African Americans and Caucasians who choose to participate in genomic research studies will be essential in optimizing future genomic recruitment efforts among African Americans.
Our findings are noteworthy because of their confirmation of prior observed socioeconomic differences between Caucasian and African American men and their presentation of new data in this previously under-reported patient population, namely healthy men participating in nontherapeutic correlative genomic research studies. Similar to prior investigators of therapeutic prostate cancer studies, African American men in our study were characterized by lower education and income levels than Caucasian men.22-24 In contrast to other reports, we observed no differences between smoking, alcohol, family history of prostate cancer, or prostate cancer screening patterns.25, 26
A striking difference in marital rates was observed between healthy African American and Caucasian men (46% African American vs 69% Caucasian). A reduced rate of marriage by African American men has been observed by other investigators; however, the difference observed in our study is larger than expected.27-31 This difference may have important implications on minority recruitment strategies for other population-based genomic studies at other centers in which large numbers of healthy male subjects are needed. In particular, the common practice of targeting faith-based organizations may not be as effective as previously believed for recruiting African American male subjects.
According to recent demographic data on historically African American protestant congregations, males comprise only 40% of worshippers.32 Compared with women, men place less emphasis on religion in their lives.33 In addition, men are less likely to attend religious services more than once per month compared with women (35.2% vs 46.0%).34-36 This gap in religious service attendance increases with age.37 Furthermore, a direct association between church attendance and marital status has been observed in African American men.38, 39 Similar associations between marital status and church attendance have also been observed in other ethnicities, including Caucasian and Latino men.39-42
Recruitment of healthy African American men from faith-based communities likely will remain a strategy for cancer epidemiologic studies. However, additional programs that engage community sites frequented by African American men in high volumes may also be needed to reach adequate sample sizes required for population-based men's health genomic studies. Our strategy of targeting large-volume free community health fairs is 1 potential alternative. Other examples include worksite health promotion programs (especially in vicinities of worksites including open-air garages, small concrete offices, and guard stations), housing projects, barber shops, athletic events, food pantries/food shelves, grocery stores, health clinics, a mass screening site, and minority community resource centers.43 Additional recruitment strategies for minorities successfully used in other cancer research studies include direct mail, Health Education Council, and media advertising (newspapers, employee newsletter, radio, and posters/flyers).44
Limitations in our study include the restricted population studied. Demographic differences between healthy women participating in similar genomic correlative research studies are unknown. Data collected on healthy women participating in genomic research through the IU-CABS study are too early for analysis. Our study did not analyze study subjects' views as to the importance of participating in genomic correlative research, thus limiting the ability to identify further opportunities for increased accrual. In addition, African American male recruiters were not exclusively used for recruitment of African American participants. Use of minority recruiters to enroll similar minority subjects has been suggested as a means to improve minority accrual, although prospective validation of this technique is not proven in many studies.13 Even when African American recruiters including recruitment coordinator were used in recruitment of African American subjects in some clinical trials, other factors including involvement, commitment, and interaction of recruiting staff, especially the principal investigator, with study participants and effective communication skills (verbal or nonverbal) play a more pivotal role in recruiting minorities in such trials.45 Although concerns about the use of specimens for genetic research and genetic discrimination issues have been raised by prior investigators as a potential barrier to genetic research efforts in minority populations, anecdotally we did not encounter such concerns with high frequency among our study population.46-49 It should be noted, however, that such data were not collected prospectively from study participants. The possibility of some selection bias for recruiting the patient population could not be completely ruled out, as our sample size was not collected from a single avenue (Indiana Black Expo consisted mainly of Urban African Americans, and Indiana State Fair consisted primarily of rural Caucasians). Lastly, our study benefitted tremendously from the presence of longstanding community events with health wellness opportunities. In other regions, immediate partnership with organizations such as Indiana Black Expo and Indiana State Fair targeting desired and under-represented minority populations may not be as quickly achievable.
In summary, our study presents initial demographic summary data from healthy African American and Caucasian men participating in a nontherapeutic genomic correlative study. Significantly lower rates of marriage in African American men may require alternatives to faith-based organizations as a recruitment site for male minority populations. The use of high-volume free community health fairs as a venue to recruit healthy patients for genomic correlative population studies presented in our study represents a novel recruitment approach.
Supported by grant funding from the Walther Cancer Research Institute and Clarian Health Partners, and institutional research support from the Indiana University School of Medicine.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.