High Prevalence of Screening-detected Prostate Cancer among
Afro-Caribbeans: The Tobago Prostate Cancer Survey
Clareann H. Bunker,2 Alan L. Patrick, Badrinath R. Konety, Rajiv Dhir, Adam M. Brufsky, Carlos A. Vivas, Michael J. Becich, Donald L. Trump, and Lewis H. Kuller
Departments of Epidemiology
M. J. B.], University of Pittsburgh, Pittsburgh, Pennsylvania 15261
Risk for prostate cancer is high among African Americans. We hypothesized that risk for prostate cancer is also high in other populations of African descent. Our objective was to determine the screening-detected prevalence of prostate cancer in the predominantly Afro- Caribbean population on the island of Tobago.
Male residents, ages 40–79 years, were invited to participate in a population-based screening for prostate cancer using serum prostate-specific antigen (PSA) and digital rectal exam (DRE). Men with elevated PSA (>4 ng/ml) or abnormal DRE were offered an ultrasound-guided sextant biopsy of the prostate gland. Men (2484), ages 40–79 years, underwent prostate cancer screening between September 1997 and June 2001. Mean age was 55.9, SD was 10.6 years, and median was 54 years. Mean serum PSA was 14.8 ng/ml, SD was 376 [excluding 4 values > 2 SD above the mean (1,112, 1,317, 1,818, and 18,330 ng/ml) mean PSA was 5.5 ng/ml and SD was 29.6], and median PSA was 1.2 ng/ml.
Elevated PSA and/or abnormal DRE were observed in 31% (759 of 2484) overall, and in age groups 40–49 (87 of 843, 10%), 50–59 (201 of 729, 28%), 60–69 (262 of 584, 45%), and 70–79 (209 of 328, 64%). Of 681 men biopsied, 259 (38%, or 10% of the 2484 screened) were diagnosed with prostate cancer. Age-specific rates of screening detected prostate cancer were: 1%, ages 40–79 years; 7%, ages 50–59 years; 18%, ages 60–69 years; and 28%, ages 70–79 years.
These screening results indicate a very high screening-detected prevalence of prostate cancer in this population of West African descent. These data support the hypothesis that populations of African descent share genetic and/or lifestyle factors that contribute to their elevated risk for prostate cancer.
Prostate cancer is a very serious personal and public health
problem affecting African Americans more frequently than
Caucasians. On the basis of 1990–1997 data from the U.S.
SEER4 of the National Cancer Institute (1), age-adjusted incidence
of prostate cancer is 225 of 100,000 among African
Americans compared with 149 of 100,000 among white non-
Hispanics. The mortality rate from prostate cancer was 2-fold
higher among persons of African descent (54 of 100,000)
compared with white non-Hispanics (23 of 100,000). Incidence
of prostate cancer in the United States increased dramatically in
both groups between the late 1980s and 1993, reflecting the
earlier diagnosis that occurred with the increasing use of serum
PSA screening (1). An encouraging downturn in prostate cancer
mortality rates was observed in both ethnic groups from 1993
to 1997 (1).
Established risk factors for prostate cancer include age,
ethnicity, family history of prostate cancer, and high-fat or meat
diet (2). Other factors suspected include hormone metabolism,
(3, 4) vitamin D metabolism, (5) and a few occupational exposures
(6). The relationships of a number of candidate genes
to prostate cancer are under investigation with most published
results limited to Caucasian populations (7, 8). The reasons for
the higher risk for prostate cancer among African Americans
Until recently, there has been little solid prevalence, incidence,
or mortality data for populations of African descent
outside the United States, although data published a few years
ago in an annual summary of worldwide data suggested high
rates of prostate cancer mortality in Martinique and Trinidad &
Tobago (9). Glover et al. (10) reported high rates of prostate
cancer incidence in the predominantly Afro-Caribbean population
of Jamaica. Data regarding screening parameters and prevalence
of prostate cancer in populations of African descent in
the United States are sparse (11, 12) and virtually absent in
other populations of African descent. However, a recent publication
has estimated prostate cancer prevalence for 1994
among African Americans and Caucasians using a model based
on incidence and survival functions calculated from the Connecticut
Tumor Registry, 1940–1993, and applied to the SEER
1973–1993 populations. The prevalence proportion ranged
from 7 of 100,000, ages 40–44 years, to 9,725 of 100,000, ages
Received 8/16/00; revised 4/1/02; accepted 4/22/02.
726 Vol. 11, 726–729, August 2002 Cancer Epidemiology, Biomarkers & Prevention
75–79 years, in Caucasians, compared with 14 of 100,000, ages
40–44 years, to 10,945 of 100,000, ages 75–79 years, in
African Americans (13).
We hypothesize that risk for prostate cancer is high among
populations of African descent living in diverse environments.
If so, this would lead us to hypothesize that populations of
African descent share genetic and/or lifestyle factors that increase
risk for prostate cancer.
On the island of Tobago, Trinidad & Tobago, we are
conducting a population-based, longitudinal study of prostate
cancer in the male population ages 40–79 years. In this report,
we present data from the initial cross-sectional screening using
serum PSA and DRE.
Materials and Methods
Population. The island of Tobago is about 7 26 miles in
size. According to the 1990 census (14) of Trinidad & Tobago,
the male population of Tobago, ages 40–79 years, numbered
5121. Ninety-two percent of Tobago residents reported that
they were of African descent. Most healthcare is provided by a
government-supported system through the Tobago Regional
Health Authority that manages the 19 neighborhood health
centers and one hospital. Some residents travel to Trinidad for
specialized care under the government system. Some care is
provided by private caregivers. PSA testing has been available
but generally limited to symptomatic men seeking care in the
Recruitment. The recruitment goal was 4000 men, 80% of the
male population of Tobago, ages 40–79 years. Currently,
3000 men are enrolled. Recruitment has been the result of
word of mouth, informing by healthcare workers at the hospital
and health centers, informing by private physicians, posters,
flyers, public service announcements, and public presentations
by oncologists and urologists from Trinidad and the United
Informed Consent. Consent was obtained using forms and
procedures approved by the Institutional Review Boards of
University of Pittsburgh Institutional Review Board and the
Tobago Ministry of Health.
Data Collection. Data were collected by locally resident study
staff at the study office located at the Tobago Regional Hospital.
Data collected included ethnicity, education, occupation,
smoking, medical history, personal and family cancer history,
vasectomy, prostate symptoms, health screening history, alcohol
intake, detailed occupational history, and height, weight,
waist, and hip measurements.
Biological Sample Collection. A 15-ml plain vacutainer of
peripheral blood was drawn from fasting subjects. Aliquots of
serum were frozen at 20°C for later measurement of PSA.
DRE. A systematic DRE was performed by a physician trained
according to the study protocol. This exam was scheduled after
the blood draw to avoid an artifactual increase in serum PSA
that may follow digital manipulation of the gland.
PSA Measurement. Serum PSA levels were measured at the
University of Pittsburgh Central Pathology Laboratory using
the automated Microparticle Enzyme Immunoassay, Abbot Ax-
SYM PSA assay (Abbott Laboratories, Abbott Park, IL).
Criteria for Referral for Prostate Biopsy. Subjects were
referred to the Tobago Regional Hospital for biopsy if the DRE
was abnormal (except for simple enlargement without palpably
abnormal areas) or if serum PSA was elevated (4.0 ng/ml).
Prostate Biopsy. Prostate biopsies were performed by urologists
or by surgeons trained by urologists from the University of
Pittsburgh Medical Center. Trans-rectal ultrasound guided biopsy
was performed using an 18 gauge, 21-cm spring-loaded
biopsy needle (Boston Scientific, Natick, MA). Sextant biopsies
were obtained according to a standard protocol.
Prostate Pathology. The formalin preserved specimens were
stored at room temperature and shipped to the University of
Pittsburgh for histopathological examination. The specimens
were examined for presence or absence of high-grade prostatic
intraepithelial neoplasia, presence or absence of cancer, Gleason
score of cancer, location of cancer, and perineural invasion.
Data Analysis. Age-specific prevalence rates (per 100
screened men) were calculated. Age-adjusted rates/100 and
SE/100 were calculated by direct standardization (15) based on
the age distribution (50–79 years) of the 1970 United States
standard million population (1). Positive predictive value of the
screening tests was calculated as number of men diagnosed
with prostate cancer divided by the number of men with abnormal
DRE and/or elevated PSA who underwent biopsy. All
statistical calculations were performed using SPSS 10.0 for
Windows (SPSS Inc., Chicago, IL).
PSA and/or DRE screening was completed for 2492 men, ages
40–79 years, 49% of the total male population in this age
group. Eight men who reported prior diagnosis of prostate
cancer are excluded from these analyses. Among the 2484 men
remaining, mean age was 55.9 years, SD was 10.6, and median
was 54 years. Ninety-two percent of men reported three or four
grandparents of African descent, whereas 5% reported one or
two of African descent. Twenty-four percent had completed
secondary school or higher education. Forty-two percent reported
ever smoking, whereas 14% were current smokers.
The serum PSA range was 0.1–18,330 ng/ml. Mean serum
PSA was 14.8 ng/ml and SD was 376. After excluding 4
values 2 SD above the mean [1,112, 1,317, 1,818, and 18,330
ng/ml], mean PSA was 5.5 ng/ml, SD was 29.6, median PSA
was 1.2 ng/ml, and range was 0.1–602 ng/ml.
Elevated serum PSA levels (4 ng/ml) were observed in
452 of 2437 men (19%), ranging from 2% of men, ages 40–49
years, to 53% of men, ages 70–79 years. DRE was abnormal in
514 of 2074 men (25%). Frequency of abnormal DRE increased
across age groups from 11 to 48%. Abnormal screening results
(PSA 4 ng/ml and/or abnormal DRE) are shown in Table 1.
PSA and/or DRE were abnormal in 759 of 2484 men (31%).
Thus, a high proportion of the screened men was referred for
biopsy: 10% of men, ages 40–49 years; 28%, ages 50–59
years, 45%, ages 60–69 years; and 64%, ages 70–79 years.
Of the 759 men referred for biopsy, 681 (90%) have
undergone prostate biopsy. Prostate cancer was diagnosed in
259 (38%) men, 2 (1%) with Gleason grade 5, 142 (55%) with
grade 6, 86 (33%) with grade 7, and 29 (11%) with grades 8, 9,
and 10. The prevalence of prostate cancer among screened men
was 10% (259 of 2484) among men ages 40–79 years, and 15%
(250 of 1585) among men ages 50–79 years. The age specific
results are shown in Table 1.
The high prevalence rate reported above reflects not only
the high rate of abnormal screening results but also a highpositive
predictive value for an abnormal screen: 12% of biopsied
men ages 40–49 years were diagnosed with prostate cancer;
27%, ages 50–59 years; 45%, ages 60–69 years; and 53%,
ages 70–79 years (Table 1).
Among 123 men reporting family history of prostate cancer,
117 reported one relative, 5 reported two relatives, and 1
reported three relatives with prostate cancer. The distribution of
727 Cancer Epidemiology, Biomarkers & Prevention
relatives included 78 fathers, 65 brothers, 3 half-brothers, 8
uncles, and 2 grandfathers. Thirteen (10.6%) of 123 men reporting
family history of prostate cancer were diagnosed with
prostate cancer, compared with 246 (10.4%) men diagnosed
with prostate cancer among 2361 men not reporting family
history of prostate cancer.
Eight percent of men reported that a physician had told
them they had benign prostatic hypertrophy. Waking to urinate
more than once/night was reported by 51% of men (62% of men
ages 50–79 years). The rate rose steadily from 33% among men
ages 40–49 years to 86% among men ages 70–79 years. Within
each age group, the rate was similar in cases and noncases.
The screening detected prevalence of prostate cancer in this
Afro-Caribbean population, ages 50–79 years, was about three
to four times higher than rates reported from screening studies
of predominantly Caucasian populations (16–18). These
United States studies, which reported results by age group, were
conducted between 1989 and 1992 when PSA screening was
just beginning to be widely used in the United States. After
direct age adjustment to the standard 1970 United States population,
ages 50–79 years (SEER), screening-detected prevalence
in Tobago men, ages 50–79, was 15.1 of 100, SE 0.9,
compared with 3.8 of 100, SE 0.2, in a population of 6501
United States men (92% Caucasian, 3% African American, and
5% other), ages 50–79 years, reported by Richie et al. (18).
Comparison of the age-specific screening-detected prevalence
rates of prostate cancer in these two populations are shown in
Factors that may influence the comparison of the screening
results from these two populations include the biopsy protocol,
the proportion of men with abnormal PSA/DRE who underwent
biopsy, prior level of screening in the populations, and recruitment
methods. In the study by Richie et al. (18), quadrant
ultrasound guided needle biopsies were performed, whereas
this study required sextant biopsies, which were likely to have
resulted in some increase in the probability of detecting cancer
if cancer was present (19). The biopsy rate among men with
abnormal screening results was 90% in the Tobago study,
compared with 69% in the study by Richie et al. (18). This
higher biopsy rate increased the opportunity to ascertain cases.
Thus, procedural differences may account for some of the
increased prevalence rate among the Tobago men.
The United States population in the previously mentioned
study (18) was screened in 1991–1992, a period during which
the incidence of prostate cancer was rising sharply (1), reflecting
increasing use of PSA testing. Thus, some prostate cancer
cases may have been removed from this United States population
by prior screening. However, 19% of the Tobago men, ages
50–79 years, reported a prior PSA test, suggesting that some
level of screening had also been available in this Tobago
Both the United States (18) and the Tobago populations
were self-referred. The Tobago population was recruited primarily
by word of mouth. The United States population was
recruited by advertisement. Among the United States population,
53% reported symptoms of prostatism. Sixty-two percent
of Tobago men, ages 50–79 years, reported waking to urinate
two or more times/night. If self-referral bias were to have a
strong effect, one would have expected higher prostate cancer
rates among men recruited earlier in the study compared with
later. As shown in Fig. 2, this was somewhat true, particularly
in the oldest age group among whom symptoms were more
likely to be related to prostate cancer. Compared with the
age-standardized rate of screening-detected prostate cancer
(15.1%) among men ages 50–79 years in the total screened
group, the rate based on the men recruited only in 2000–2001
was somewhat lower, 13.4%. Even after deflating the conservative
13.4% rate by the ratio of the biopsy rates in the two
populations, the even more conservative estimate of the agestandardized
rate of screening-detected prostate cancer of
10.3% is still almost 3-fold higher than in the United States
population, Richie et al. (18).
Similar age-specific prevalence rates from screening of
other populations of African descent have not been published.
Fig. 1. Screening detected prevalence of prostate cancer among 6501 predominantly
Caucasian United States men (18) and 2484 Afro-Tobagonian men.
Table 1 Screening and biopsy results in Tobago men by age group
and/or PSA n
(% of screened)
Biopsied n (%
Prostate cancer n
(% of biopsied)
Prostate cancer prevalence
(per 100 screened population)
40–49 843 87 (10) 77 (89) 9 (12) 9/843 (1)
50–59 729 201 (28) 188 (94) 50 (27) 50/729 (7)
60–69 584 262 (45) 240 (92) 107 (45) 107/584 (18)
70–79 328 209 (64) 176 (84) 93 (53) 93/328 (28)
Total age 2484 759 (31) 681 (90) 259 (38) 259/2484 (10)
Total age 1641 672 (41) 604 (90) 250 (41) 250/1641 (15)
728 High Prostate Cancer Prevalence in Afro-Caribbeans
Smith et al. (20) reported prostate cancer prevalence of 5.1%
among 804 African-American men screened between 1991 and
1995 using serum PSA and DRE. Application of the Tobago
prostate cancer rates to the reported age distribution in this
African-American population yielded an expected rate of
12.4%, compared with the observed prevalence of 5.1%, suggesting
that the Tobago rates are approximately twice those
observed in the African-American population.
One of the known risk factors for prostate cancer is ethnicity,
i.e., African descent, although we do not know how this
risk is mediated. One hypothesis is that genetic factors contribute
to the high risk for prostate cancer among populations of
African origin. If the Caucasian admixture rate in the Tobago
population is indeed low, then this population may carry a
higher burden of high-risk genes of African descent than the
more admixed populations in the United States.
In conclusion, compared with Caucasian and other populations,
the higher incidence of prostate cancer observed in
populations of African descent, African Americans (1) and
Jamaicans (10), and the very high screening-detected prevalence
of prostate cancer observed in this study among Tobagonians
support the hypothesis that these populations share ancestral
genetic factors that increase susceptibility to prostate
cancer. However, the variability in risk across these populations
of African descent suggests an important influence of unknown
environmental/lifestyle factors acting on prostate cancer risk in
these susceptible populations.