Recruitment strategies and comparison of prostate cancer-specific clinical data on African-American and Caucasian males with and without family history

Abstract

Prostate cancer is the most common cancer in men in the United States. This is a complex disease with high heterogeneity and the exact causes are unknown in population-specific samples. Family history is a primary risk factor irrespective of race. Identifying prostate cancer families with multiple affected cancer cases is challenging. Herein we document recruitment techniques and present prostate cancer clinical factors described in a cohort of African Americans and Caucasians with or without a strong family history. A total of 521 prostate cancer patients (241 African Americans and 280 Caucasians) were identified using a novel cooperative methodology involving a combination of treating physicians and tumor registries. Higher prostate-specific antigen (PSA, P=0.0269) was found in familial cases as compared to sporadic cases in African-American men. In addition, PSA values for familial cases were higher (P=0.0093) in African-American as compared to Caucasian men. No differences were detected in Gleason score values in either race, regardless of family history. These findings remained the same after adjustment was made for age at diagnosis. In conclusion, methodologies for cohort acquisition, and clinical characteristics, are described for men with and without a family history of prostate cancer using both Caucasian and African-American populations.

Introduction

Prostate cancer (PCa) is the most common malignancy in men in the United States and the causes of PCa are not clearly understood. The three major risk factors in PCa include age, race and family history. Incidence of this cancer increases around 15 times in men 65 years or older as compared to younger men.¹ Variation in the incidence rates of PCa is related to race. Epidemiological studies show that African-American men have one of the highest rates of PCa as compared to other ethnic subgroup of the world’s population. A high incidence of PCa has also been reported in the predominantly African-Caribbean population of Jamaica.² African-American men are also 2–3 times more likely to die from this disease.³ Biology of progression of PCa may or may not be responsible for the difference in survival between these two racial groups.⁴ African-American men develop PCa at an earlier age than Caucasians.⁵ Environmental, dietary, hormonal and genetic factors are under study to better understand these disparities.

In addition to the likelihood of increasing incidence of the disease with increasing age and an African ancestry, familial aggregation of PCa significantly increases the risk of the disease. A trend has been observed of an increasing risk of PCa being associated with an increasing number of affected relatives and an earlier age at onset in affected relatives.^6–9 Risk of PCa is increased 5 to 11-fold in men with two or more affected first-degree relatives.⁹ Specifically, familial aggregation is observed in 15–25% of the diagnosed cases with 5–10% of those cases estimated to be due to hereditary risk factors.^10–14

To date, very few studies have reported on PCa-affected individuals with African descent.^15–20 Interestingly, the recently described 8q24 risk allele is present in a higher percentage of the African-American population as compared to other ethnic groups; both risk and subsequent progression are potentially influenced by this genetic factor.^21,22 Role of family history in combination with racial information may provide key insight for explaining the variability in clinical expressions of affected individuals, which may aid in identifying effective screening strategies and lead to better understanding of the disease in different ethnic groups. Explanations for the differences observed in the clinical characteristics of PCa between African Americans and Caucasian remain controversial.²³

Recruitment of subjects for studies attempting to identify genetic components of disease has been a challenge in complex disease research, especially in the diseases with late age of onset. Researchers within the Department of Genetics at Louisiana State University Health Sciences Center (LSUHSC) in New Orleans initiated a genetic linkage study of PCa with recruitment beginning in March 2000. Recruitment was largely derived from patients residing in Louisiana. The objective of the present study was to report on the recruitment techniques and to present the clinicopathological characteristics of PCa in the African-American and Caucasian cases in relation to number of affected family members to further understand this complex and heterogeneous disease.

Materials and methods

Ascertainment of prostate cancer-affected individuals

We implemented several distinct strategies in building the cohorts. There is little published data concerning experience in recruiting families with multiple cases of PCa for linkage studies, especially in African-American patients. It can be both difficult and expensive to collect families with multiple affected members, and the initial setup of a network consisting hospitals and referring doctors is potentially quite time consuming.²⁴ Ascertainment of our study subjects was accomplished as described in Figure 1 followed by a detailed description.

Figure 1.

Study design to enroll familial (≥3) cases of prostate cancer.

Recruitment strategies

(1) Letters inviting participation were sent to area hospital administrators. After obtaining permission from the appropriate hospital authority, the tumor registrars in individual hospitals were contacted. The study was explained to the tumor registrar and sample letters requesting study participation were provided to the tumor registry personnel.

Cooperating tumor registries analyzed the hospital database for PCa diagnosis to identify study subjects. Letters were sent to the identified individuals from their physicians’ office inviting the patient to participate in our PCa research study with a prestamped postcard addressed to us. Potential participants were asked to return the response form indicating their preferred contact time/information in the stamped, addressed envelope provided. Alternatively, individuals could choose to phone our research office directly.

(2) Another strategy was for an LSU-based research associate to volunteer at participating hospitals and assist the tumor registry contact in the preparation of the above-mentioned paperwork that would be forwarded to the physicians. The agreement between the hospital and the researcher was that the researcher could not leave with any patient documentation, and that all patient information remains confidential.

(3) Acquisition of potential subjects was also achieved through our collaborating physicians’ direct referral of patients to our research office.

Interview to obtain family history and clinical information

All subjects expressing interest in the study underwent an interview with one of our research associates by telephone. At point of contact with our office, the individual received a short verbal introduction to the study and a request to complete a 2–3 min questionnaire. The questionnaire included questions on race, ethnicity, date of diagnosis, age of diagnosis, prostate-specific antigen (PSA) at diagnosis, Gleason score and family history. A letter and study pamphlets were sent to the participating individuals if their family information fulfilled the criteria of our study. If the individual wanted more study information before divulging this personal information, they received the same letter and pamphlet mentioned above. If the family fulfilled the criteria of having a positive family history (three or more PCa cases), the contact person (proband) was asked to discuss the study with the family members (who after our initial evaluation were considered informative for genetic analysis) and also the other affected family members (for their consent to obtain medical records verifying the primary cancer sites) to see if they would grant consent to be contacted by our office. A family pedigree was constituted from information provided by the proband and additional family members were contacted thereafter by the study coordinator for verification of their disease status.

Consent forms

Consent forms were read and signed by each participant and contained information about the study and risks associated with participation in the study, if there was any. Signed copy of the consent form was provided to each study participant for their record. All study-related materials were approved by LSU Health Sciences Center Institutional Review Board (IRB).

Review of medical records

Release of information forms was obtained from all affected individuals or their family representatives (in the event of a deceased individual) and medical records were obtained to confirm the diagnosis, date of birth and age of subject, and grading and staging of tumor, and PSA at the time of diagnosis, if available. In Louisiana, medical records are maintained for 10 years in the hospitals/clinics where the diagnosis is made. We documented all the pertinent medical information available, including, age of diagnosis, PSA levels, Gleason scores and clinical stages, where available. Medical records were checked carefully to document the diagnosis of primary PCa.

In the process of recruiting individuals with the goal of identifying a susceptible allele for PCa, we created an additional cohort that focused on clinical parameters in cases without a family history of PCa in the African-American and Caucasian populations. To make a comparison of clinical histories of sporadic and familial cases, additional medical records were reviewed from the database of a collaborative clinician after complying with required Health Insurance Portability and Accountability Act (HIPAA) and receiving IRB approval.

After receiving postcards from the physicians’ office, 521 individuals with PCa contacted our office expressing interest in participating in the study. Of these men, 241 were African Americans and 280 were Caucasians. Demographic information of the individuals recruited is provided in Table 1. For the purpose of our original linkage study, only individuals with three or more relatives affected with PCa were eligible for inclusion. In addition, at least two of the affected relatives must be still living so that blood samples could be obtained on those affected individuals. Medical documentation verifying the primary site as the prostate was collected from the probands and other affected members in the family, which produced a database of 167 individuals with clinical data. If medical documentation did not include an age for each individual, the record was not included in the analysis. At the time of conducting the screening interview to determine the eligibility to enroll in the study fulfilling our study criteria, if it was revealed that the proband did not have a family history of PCa, we briefly described our alternative study on the comparison of clinical parameters and if the proband consented, we requested authorization to obtain medical records. The medical charts of additional sporadic cases were reviewed from the patient pool of one of our collaborating physicians after confirming the family history of PCa, which produced 54 more complete medical records. From the medical charts obtained from our collaborator, we included only those where we were able to verify the number of affected members in the family upon contacting the patients directly. Therefore, a total of 221 medical charts were reviewed, including both familial (with three or more affected individuals in the family) and sporadic cases (single cases only). Two individuals with extraordinarily elevated PSA values (>100 ng/ml) were not included in the analysis due to the difficulty in achieving a useful normalizing transformation in this highly skewed data set. Individuals were listed as having family history only if the medical charts of affected family members confirming the prostate as a primary cancer site were accessed. We requested medical release forms from all living affected individuals in the family if the participating family had multiple cases of PCa. Medical charts release form for deceased PCa patients were signed by next of kin. Of the total of 221 medical charts reviewed, 77 were African Americans and 144 were Caucasians. PSA values were tabulated only at the time of diagnosis. No PSA values were included after the individual obtained medical treatment. Some familial cases had missing PSA values and the final database contained PSA on 46 African-American and 66 Caucasian familial cases. Demographic information and clinical characteristics are provided in Table 2.

Table 1.

Demographics of families with prostate cancer using our recruitment strategy

No. of prostate cancer cases in families of each individual	No. of African Americans, % (N = 241)	No. of Caucasians, % (N = 280)
Single cases	143 (59)	164 (59)
Two cases	58 (24)	75 (27)
Three or more cases	40 (17)	41 (15)

Table 2.

Demographics of individuals for which medical charts were obtained to verify prostate as the primary site of origin

Ethnicity	Family history	Clinical parameters	Median	Mean±s.d.	P-values
African-American (N = 77)	Sporadica	Age at onset (years)	65.50 (53–83)	65.1±6.8	—
	Familialb		60.00 (38–74)	59.5±8.6
	Sporadica	PSA values (ng/ml)	6.8	6.5±6.3
	Familialb		7.2	9.8±7.6	0.0269
	Sporadica	Gleason scores	6.0	6.1±0.99	—
	Familialb		6.0	6.5±1.20
Caucasian (N = 144)	Sporadicc	Age at onset (years)	68.50 (47–83)	68.5±7.4	—
	Familiald		66.50 (44–83)	65.9±7.8
	Sporadicc	PSA values (ng/ml)	7.4	8.1±9.2	0.3056
	Familiald		6.4	7.1±8.9
	Sporadicc	Gleason scores	6.8	6.5±0.93	—
	Familiald		6.0	6.4±1.20

Abbreviation: PSA, prostate-specific antigen.

N = Sample size with

^aN = 30

^bN = 47

^cN = 70

^dN = 74.

Data analysis

Statistical analysis was performed on the clinical data using version 9.2 of the Statistical Analysis System (SAS, SAS Institute, Cary, NC, USA). Due to the skewness of PSA values, we transformed PSA values by taking the base 10 logarithm (after adding 1 to each PSA value to avoid taking the log of 0). This transformation worked well to ‘normalize’ PSA values for statistical purposes. Subsequent analyses were based on the transformed PSA values. No transformation was required for the other clinical values studied. In addition to the log-transformed PSA values, Gleason scores were tabulated.

General linear models were fit using PSA as the dependent response and included dichotomous indicator variables for family history of PCa (familial or sporadic) and race (African American or Caucasian) along with a continuous covariate for age. Familial history was labeled familial if there were at least three members affected in the family.

Results

Serum PSA values were compared in the African-American and Caucasian men with and without a family history of the disease. When PSA was used as the response, an interaction was detected between race and familial case status (P=0.0183). Further analysis was thus required to uncover the nature of the relationships between the factors and their influence on PSA. Familial cases had higher levels of PSA (P=0.0269) when compared with sporadic cases in African-American men. In contrast, PSA was not found to differ with respect to familial status in Caucasians (P=0.3056).

African Americans were found to have higher PSA (P=0.0093) than their Caucasian counterparts in familial cases. However, no difference in PSA between races was observed in the sporadic cases group (P=0.3801). There was no indication of differences in Gleason scores in any of the factors (race, familial status and age) or their interaction.

Discussion

With respect to the recruitment effort in identifying PCa-affected individuals with family history, the strategy to recruit families with PCa aggregation has been successful using the techniques described above. It is often extraordinarily difficult to recruit families for complex diseases in genetic epidemiological studies.^25,26 Although it is difficult to accrue families at a rapid rate due to the stringent privacy controls, our procedures have met all IRB and ethical reviews and represent a methodology that can be adopted to recruit data for these types of studies. In our experience, referral by the physician can make a substantial difference in family ascertainment in PCa. Families are more receptive to participation if their own physician explains the importance of the study during an office visit or phone call. Assistance/collaboration of the tumor registrar/tumor registry contact person is a very important step. Various incentives for participation may be appropriate as volunteer efforts are typically problematic for tumor registries that typically operate on tight budgets and have little time for research studies.

Ensuring confidentiality and privacy improves participation. The families feel secure if maintenance of privacy and confidentiality is assured during the first phone conversation with the proband. Where possible, ability or the willingness of the study coordinator to visit families to draw blood samples enhanced the number of family members (informative for linkage analysis) willing to participate. In our experience, it is difficult to recruit qualified families fulfilling the criteria for a linkage study (with three or more affected with two living family members); yet our study identified a number of such families with the assistance of collaborating clinicians and a dedicated study coordinator.

In general, overall response rate for the Caucasian population was 18 vs 22% for the African-American population. Clinical collaborators in our institution were actively involved in the recruitment process of African-American participants once a patient was identified to have a positive family history. Without the direct referral of our clinical collaborators at LSUHSC, response rate for the Caucasians and the African Americans were 17 and 15%, respectively. These data indicate that African-American males were more receptive to contacting us if they were referred by their physicians directly. In total, we were able to enroll 20 African-American families and 24 Caucasian families in our linkage study. A number of families initially detected to have three or more cases of PCa did not eventually qualify due to not having enough living relatives affected in the family or having no communication or cooperation from affected family members. Also, in some situations families identified were turned out to be bilineal (affected relatives on both paternal and maternal sides) and thus were ineligible to qualify.

Earlier studies have documented no substantial clinical or pathological differences with respect to family history.^10,13,27 Keetch et al.²⁸ observed comparable values of median serum PSA except for the lower histological grades in hereditary PCa compared to sporadic carcinomas. But none of these studies involved different races in the study population, and in addition, many of those studies only included the organ-confined PCa cases. Comparison of clinical characteristics in these two races in sporadic and familial cases in the population presents new data in this context. Age at onset in African-American men in familial cases is lower than sporadic cases. Age at onset in Caucasian men also differs between familial and sporadic cases but the difference is not that dramatic as in the African-American men studied. Thus, positive family history exhibits an effect on the age of onset of PCa in both populations studied. Significant difference was observed in PSA values in African-American sporadic vs familial cases. Differences between the races in PSA for familial vs nonfamilial cases have not been previously described.

Data were also analyzed with all men together in any race and then were stratified in two age groups: one with the individuals ≤65 years and another group with the individuals older than 65 years (data not shown). We used this criterion since 65 years is the cut-off age limit for early onset of PCa in a number of genetic studies. Using age as a dichotomous variable, we did not find any other significant association. Also, age-adjusted analysis involving PSA values yielded approximately similar P-values.

In conclusion, we have recruited a substantial number of African Americans and Caucasians with a familial history of PCa using a series of techniques that can be broadly applied. The time interval between diagnosis and enrollment did not vary between races or presence/absence of family history due to the nature of the study, since only the patients interested participating in the study contacted us and based on the available family history information or feasibility to contact all the informative family members we decided to enroll patient.

Significant obstacles in recruiting the entire families for the linkage study were noted in the African-American cases, typically for not having contact information for other family members. Differences in familial and sporadic cases between African Americans and Caucasians were detected including younger age at onset and higher PSA values in the African-American familial cases. Because levels of PSA and age at diagnosis are dependent not only on biological variables, but also variables such as frequency of screening, we interpret these data with caution. We have no data on the PSA screening frequency of the participating individuals and the impact of this variable on our data is potentially profound. Thus, although these observations are novel, confirmation of these findings in larger studies that control for PSA screening frequency is needed before they can be accepted as biologically valid.

Accrual to the study is ongoing. Enrollment to the linkage study was affected by Hurricane Katrina, since for a number of families, while we were in the middle of documenting the medical record and collecting biological samples, some of those families could not be traced back due to nonworking original phone numbers or nonexisting addresses. Some of the medical charts located in hospitals out of service after the storm could not be retrieved. Data presented in the current manuscript list only those collected pre-Katrina.