Molecular and Cellular Endocrinology 197 (2002) 251
/255

www.elsevier.com/locate/mce

Canine prostate carcinoma: epidemiological evidence of an increased

risk in castrated dogs
E. Teske a,*, E.C. Naan a, E.M. van Dijk a, E. Van Garderen b, J.A. Schalken a,c
a
Department of Clinical Sciences of Companion Animals, Utrecht University, P.O. Box 80.154, 3508 TD Utrecht, The Netherlands
b
Department of Pathology, Utrecht University, Utrecht, The Netherlands
c
Department of Urology, University Medical Center Sint Radboud, Nijmegen, The Netherlands

Abstract

The present retrospective study investigated the frequency of prostate carcinoma (PCA) among prostate abnormalities in dogs
and determined whether castration influences the incidence of PCA in dogs. During the years 1993
/1998, 15363 male dogs were
admitted to the Utrecht University Clinic of Companion Animals, and of these dogs 225 were diagnosed with prostatic disease. In
addition, another 206 male dogs were diagnosed as having prostatic disease based on cytologic examination of aspiration biopsies
submitted by referring veterinarians. Benign prostatic hyperplasia was diagnosed in 246 dogs (57.1%), prostatitis in 83 dogs (19.3%),
and PCA in 56 dogs (13%). Dogs with PCA were significantly older (mean age /9.9 years) than dogs with other prostatic diseases
(mean age/8.4 years). The Bouvier des Flandres breed had an increased risk (odds ratio (OR) /8.44; 95% CI 4.38
/16.1) of having
PCA. Castration (26/56) increased the risk (OR /4.34; 95% CI 2.48
/7.62) of PCA. The mean age at diagnosis of PCA in castrated
dogs and in intact male dogs was not significantly different. The interval between castration and onset of prostatic problems was
highly variable, suggesting that castration does not initiate the development of PCA in the dog, but it does favour tumor
progression.
# 2002 Elsevier Science Ireland Ltd. All rights reserved.

Keywords: Dog; Prostate carcinoma; Risk factors; Androgen depletion

1. Introduction The dog is one of the few other species in which

Prostate cancer is one of the most important malig-
nancies in men in the Western world. In elderly men the
frequency of prostate cancer at autopsy has been
reported to exceed 40%. The probability of having a
prostate tumor diagnosed during lifetime or of dying of
prostate cancer is much lower, less than 9% (Landis et
al., 1999). Cancer confined to the prostate gland can be
cured by radical surgery or radiation therapy. However,
many patients present with locally advanced or meta-
static disease and are treated palliatively with androgen
deprivation (Chodak et al., 1994). Although the tumor is
initially hormone responsive, in most cases it develops
into an androgen-unresponsive neoplasm (McNeal et
al., 1986; Pilat et al., 1999).
* Corresponding author. Tel.: /31-30-2539411; fax: /31-30-
2518126
E-mail address: e.teske@vet.uu.nl (E. Teske).
0303-7207/02/$ - see front matter # 2002 Elsevier Science Ireland Ltd. All rights reserved.
PII: S 0 3 0 3 - 7 2 0 7 ( 0 2 ) 0 0 2 6 1 - 7
spontaneous prostate cancer occurs. Adenocarcinoma,
transitional cell carcinoma, and undifferentiated carci-
noma are the most common histological types (Leav and
King, 1968; Bell et al., 1991; Cornell et al., 2000).
Although the prevalence of prostate carcinoma (PCA) in
dogs is low, less than 0.6% in necropsy studies (Weaver,
1981), PCA in dogs shares several of the features of the
disease in humans (Maini et al., 1997). As in humans
prostate cancer occurs more frequently in older dogs
(Waters et al., 1996), tumor growth outside the prostate
is common, as are pulmonary and bone metastases (Bell
et al., 1991Leav and Ling, 1968). As in the late stage of
prostate cancer in humans, the canine tumor is unre-
sponsive to anti androgen therapy (Bell et al., 1991;
Johnston et al., 2000). Furthermore, the so-called high-
grade prostate intraepithelial neoplasia (HGPIN), which
is believed to be a precursor of carcinoma in humans
(Bostwick, 1995), is also found in the majority (55%) of
elderly sexually intact male dogs without clinical evi-
dence of prostate cancer (Waters and Bostwick, 1997).
252 E. Teske et al. / Molecular and Cellular Endocrinology 197 (2002) 251
/255
The prostate gland is an androgen-dependent organ.
Castration results in regression of prostatic tissue both
in normal dogs and in dogs with benign prostatic
hyperplasia (BPH), predominantly due to a decrease in
the number of luminal epithelium cells whereas the
number of basal cells remains unchanged (Mahapokai et
al., 2000). On the other hand androgen administration
to castrated dogs causes hypertrophy of the gland
(Walsh and Wilson, 1976). The reappearance of andro-
gen receptors in basal cells is considered a key factor in
stimulating basal cell proliferation and differentiation of
the regenerating prostate in the dog (Leav et al., 2001).
Therefore, it is surprising to find a high prevalence of
prostatic carcinoma in castrated dogs is surprising
(Obradovich et al., 1987; Bell et al., 1991).
The present retrospective study investigated the
frequency of prostate cancer among prostatic abnorm-
alities in dogs and evaluated whether castration influ-
ences the incidence of prostate cancer in dogs. Other
factors such as age at castration, age at diagnosis, and
breed were included in an attempt to improve insight
into the pathogenesis of prostate cancer in the dog.
2. Materials and methods
2.1. Animals
During the period January 1993
/December 1998,
15363 male dogs were referred to the Utrecht University
Clinic of Companion Animals (UUCCA). Of these
dogs, only animals with clinical signs of prostatic disease
underwent diagnostic work up of the prostate. In 225
dogs prostatic disease was diagnosed. Diagnoses were
made by means of cytological examination of fine-
needle aspiration biopsies of the prostate or urine
catheter aspiration biopsies taken under ultrasound
guidance. A modified Menghini needle, 18
/21 Gauge,
was used (Teske and Nickel, 1996) for the fine-needle
aspiration biopsy. In addition, another 206 male dogs
were diagnosed as having prostatic disease based on
cytological examination of specimens submitted by
referring veterinarians.
In all animals the following data were collected either
from the medical records or by means of telephone
inquiry: diagnosis of prostatic disease, age at diagnosis,
breed, information on castration, and interval between
castration and diagnosis of prostatic disease.
2.2. Statistical analysis
Statistical analysis of data was performed using the
SPSS windows 9.01 statistical package. Differences
among groups were assessed by the x2-test or Fisher
exact test for ordinal or ratio data, and the one-way
parametric analysis of variance for interval data. The
Mantel Haenszel test was used to calculate odds ratios
(OR) for the influence of castration and breed on the
incidence of prostatic disease and PCA. For the assess-
ment of the influence of castration, castration was
defined as removal of both testicles more than 100
days before the diagnosis of prostatic disease. At the
time of castration the dogs did not have to signs of
prostatic disease. The entire population of male dogs
admitted to the UUCCA during 1993
/1998 was used as
a control baseline population. The 95% confidence
intervals (CI) for proportions (P ) were calculated as
P9/(1.96 /S.E.). Differences were considered to be
significant at P B/0.05.
3. Results
BPH was diagnosed in 246 dogs (57.1%), prostatitis in
83 dogs (19.3%), and PCA in 56 dogs (13%). In 7 dogs
biopsies were inappropriate for assessment, and in the
other 39 dogs prostatic cysts (n/15), squamous meta-
plasia (n/7), sarcoma (n/2), malignant lymphoma
(n /2) and suspect carcinoma (n /13) were diagnosed
(Fig. 1). Of the 13 suspect carcinoma cases no definitive
diagnosis was made and therefore, the data of these
animals were not included in the calculations. In the few
cases in which BPH was combined with one of the other
diagnoses, the dog was listed under the other diagnosis.
There were no significant differences between dogs with
prostatic diseases seen at the UUCCA and the dogs of
which the biopsies were sent in by veterinary practi-
tioners, with regard to the incidence of the different
types of prostatic diseases or animal characteristics.
Therefore, all data were pooled in one group for the
analysis of data.
Twenty-three different breeds of animals were found
to have prostatic diseases. A significant increased risk
for any prostatic disease was found in the Scottish
terrier (OR /7.43, 95% CI 2.79
/18.6), the Bouvier des
Flandres (OR /5.51, 95% CI 4.22
/7.19), the Bernese
mountain dog (OR /2.41, 95% CI 1.59
/3.61), and the
German pointer (OR /2.19, 95% CI 1.16
/4.06). Of the
10 different breeds with PCA 14 dogs (25%) were
Bouvier des Flandres dogs. This breed had a significant
OR of 8.44 (95% CI 4.38
/16.1). No other breed
predisposition was found.
Dogs with PCA were significantly older (mean age/
9.9 years, range /5.8
/14.2) than dogs with other
prostatic diseases (mean age /8.4 years, range/1
/
15.4). In the UUCCA male dog population 20.4% of
the animals were castrated. Only 6.7% of the dogs with a
non-malignant prostatic disorder were castrated (OR /
0.28, 95% CI 0.18
/0.43). Twenty-six of the 56 dogs with
PCA had been castrated more than 100 days before the
first clinical signs of prostatic diseases, leading to an OR
of 4.34 (95% CI 2.48
/7.62). The mean age at diagnosis
 E. Teske et al. / Molecular and Cellular Endocrinology 197 (2002) 251
/255
Fig. 1. Frequency of different diagnoses in 431 dogs with prostatic disease either seen at the UUCCA (n/225) or at veterinary practices (biopsies
sent to UUCCA for evaluation, n/206).
of PCA in castrated dogs and in intact male dogs did not
differ significantly. The interval between castration and
onset of prostatic problems ranged from 1 to 10 years.
4. Discussion
The results of our study indicate that BPH is by far
the most common prostatic abnormality in dogs with
clinical signs of prostatic disease. The true percentage
may even be much higher than the 57% found in our
study, because veterinarians tend to treat the disease on
the basis of physical findings only. Although BPH is
present microscopically in the glands of all dogs older
than 6 years (Lowseth et al., 1990), most animals do not
exhibit signs and symptoms related to the hyperplastic
growth. In contrast to men, where the inward nodular
growth of the transition zone causes compression of the
urethra (McNeal, 1978), in dogs the gland expands
outward (Krawiec and Heflin, 1992; McDonnell, 1991).
BPH in dogs is hormone dependent and regresses after
castration. Moreover, the administration of either
dihydrotestosterone alone or in combination with estro-
gen to castrated dogs causes BPH (Coffey, 2001). In the
present study, only few non-malignant prostatic diseases
were found in castrated dogs, the majority being
prostatitis (n/8). We observed an increased risk of
benign prostatic disease in the Scottish terrier, the
Bouvier des Flandres, the Bernese mountain dog and
the German Pointer. To our knowledge, a breed
specificity for BPH, in Doberman Pinschers, has only
been reported once before (Krawiec and Heflin, 1992).
253
The low frequency of PCA, 56 of the 431 dogs with
prostatic diseases, is in accordance with past reports
(Krawiec and Heflin, 1992). Prostate cancer occurs less
frequently in dogs than in humans. The prevalence of
prostatic carcinoma reported from necropsy studies
ranges between 0.2 and 0.6% (Bell et al., 1991). The
median age of dogs with prostate cancer in the present
study was similar to the median age of 10 years reported
earlier (Bell et al., 1991; Cornell et al., 2000). When
converted to physiological age expressed in human
years, the mean age at prostatic cancer diagnosis is
almost similar in dogs and humans, 67 and 70 years,
respectively (Waters et al., 1996). There have been no
reports of an increased risk of PCA in a particular breed
of dog. The reason for the increased risk (OR /8.4) in
the Bouvier des Flandres breed is unknown. There is an
increasing support for the concept of a multistep
carcinogenesis in the progression of the normal prostate
epithelium to metastatic, androgen-independent PCA
(reviewed in Pilat et al., 1999; Bui and Reiter, 1999). An
inherited aberration may be the earliest step. In humans,
African American men are also at a higher risk of
developing prostatic carcinoma. These men have sig-
nificantly higher serum levels of total testosterone and
sex hormone-binding globulin (SHBG), whereas free
testosterone levels are normal (Wu et al., 1995; Winters
et al., 2001). It has been reported that SHBG increases
cAMP production in the prostate. cAMP-dependent
protein kinase A is a coactivator of the androgen
receptors, and Winters et al. (2001) postulated that
this may provide a mechanism by which circulating
androgens contribute to the increased risk of prostate
cancer. It remains to be clarified whether a similar
254 E. Teske et al. / Molecular and Cellular Endocrinology 197 (2002) 251
/255
mechanism contributes to the increased risk of PCA in
the Bouvier des Flandres.
The high prevalence of prostate cancer in castrated
male dogs is the most interesting finding of the present
study. This has been reported before (Obradovich et al.,
1987; Bell et al., 1991). However, the calculated OR
(4.34) in our study was even higher than the OR (2.38)
reported by Bell et al. (1991). The duration of androgen
exposure has been found to be significantly linked to the
histological type of the tumor (Cornell et al., 2000). In
the study of Cornell et al. (2000) adenocarcinoma was
the most common type of cancer found in sexually intact
male dogs, whereas other subtypes of PCA were found
more often in castrated dogs. Castration did not affect
the frequency or pattern of metastases. In the current
study, the mean age at diagnosis of PCA was similar in
castrated dogs and in intact male dogs. In contrast to
ferrets (Schoemaker et al., 2000), where castration leads
to adrenocortical tumors within a given time interval,
there was no such fixed interval between castration and
onset of prostatic disease in the dogs investigated. This
suggests that castration does not initiate the develop-
ment of prostatic carcinoma in the dog, but does favor
tumor progression.
The lack of markers for prostatic cancer in dogs (Bell
et al., 1995) makes early diagnosis difficult and therefore
the true incidence of prostate cancer is unknown, but it
may be higher than currently believed. In this respect, it
is interesting that HGPIN, which in humans is con-
sidered to be a precursor of prostatic carcinoma
(Bostwick, 1995), is also found in dogs (Waters and
Bostwick, 1997). HGPIN, which is morphologically
similar in dogs and humans, was seen in 66
/72% of
dogs with prostatic carcinoma. However, the prevalence
of HGPIN appears to vary widely: in pet dogs and
military working dogs without PCAs it is 55 and 3%,
respectively (Waters and Bostwick, 1997; Aquilina et al.,
1998). Aquilina et al. (1998) have postulated that genetic
and environmental factors, e.g. diet, may influence the
prevalence of HGPIN in dogs. The facts that castrated
dogs have a much lower frequency of HGPIN (8%) than
sexually intact dogs (55%) (Waters and Bostwick, 1997),
and that castrated dogs are at increased risk of devel-
oping prostatic carcinomas appear in conflict with the
concept that HGPIN is a precursor of canine PCA. As
in humans, it is also highly unlikely that BPH is a
precursor of PCA in the dog. BPH occurs in almost all
elderly male dogs, whereas the chance of PCA is less
than 0.6%. Castrated dogs have an increased risk of
PCA and do not spontaneously have BPH.
The main unanswered question is whether prostate
cancer in dogs develops from an androgen-independent
cell type or whether, as is believed to occur in humans, it
is initially androgen dependent but becomes indepen-
dent following hormonal depletion. Castration is
thought to promote the emergence of a hormone-
independent clone that continues to grow. In this
context, Bell et al. (1991) and Johnston et al. (2000)
have reported that androgen depletion by surgical
castration may result in the involution of non-neoplastic
tissue in the prostate, but does not appear to affect of
prostatic carcinoma in dogs. Selection for unresponsive
cells as a result of the loss of androgen receptors in the
tumor cells has been postulated as explanation for this
phenomenon. In support of this concept, Leav et al.
(2001) reported that only one of 19 canine prostatic
carcinomas expressed androgen receptors. Preliminary
unpublished results of our group (van Garderen) show,
however, that depending on the marker used AR (N-20;
sc-816) Santa Cruz Biotechnology Inc., Santa Cruz,
California. proportion of canine PCAs with androgen
receptor-positive tumor cells varies. Indeed, androgen
receptor-positive tumors were found in sexually intact
dogs and in castrated dogs. In this respect, these tumors
resemble the late stage of human PCA, where most
tumors are hormone refractory while continuing to
express androgen receptors (Craft and Sawyers, 1999;
Pilat et al., 1999). Although castration in dogs is
associated with a massive decrease in the number of
epithelial cells, indicating that these cells are androgen
dependent, an increased population of basal or stem
cells can be found which do not depend on androgens
for survival and which proliferate actively (Mahapokai
et al., 2000). It is possible that PCA arises from these
basal cells. Leav et al. (2001) have reported that most
canine prostate tumors originate from ductal structures.
They also demonstrated that the basal cells in the duct
areas and in the acinic areas are of different populations
and have a different sensitivity to factors that induce
neoplastic growth. The earlier findings of Cornell et al.
(2000) are consistent with this. They have reported that
well-differentiated adenocarcinomas are found more
frequently in sexually intact dogs while other histologi-
cal types with more anaplasia and even urethelial
differentiation are seen more often in at an early age
castrated dogs. Recently, it was shown by representa-
tional difference analysis, that androgen-independent
PCAs in humans are characterized by the overexpres-
sion of certain genes that are normally expressed in
prostatic basal cells (Bui and Reiter, 1999). These genes
play critical roles in the processes of self-renewal and
proliferation. Overexpression of these stem cell genes
leads to a transformed cell which survives and prolifer-
ates in the absence of androgen, obtaining its growth
advantage, and which may undergo new genetic
changes, leading to PCA (Bui and Reiter, 1999). These
genes might be responsible for the overexpression of
anti-apoptic factors, such as Bcl-2 and telomerase, and
peptide growth factors and their receptors, such as
EGFR, HGR/c-met. The overexpression of these genes,
not only in androgen-independent tumors, but also in
HGPIN, suggests that these genes are important in an
 E. Teske et al. / Molecular and Cellular Endocrinology 197 (2002) 251
/255
early stage of prostate cancer development (Bui and
Reiter, 1999). Future studies with dogs should investi-
gate the role of these factors in the development of
prostate cancer in this species.
In conclusion, the risk of PCA was increased in dogs
that had been castrated more than 100 days before
cancer diagnosis, independently of prostatic problems.
These PCAs developed at a variable time after castra-
tion, but at the same age in both castrated and intact
dogs, which suggests that castration is not an initiator of
cancer, but that it favors tumor progression.
References
Aquilina, J.W., McKinney, L., Pacelli, A., Richman, L.K., Waters,
D.J., Thompson, I., Burghardt, W.F., Jr., Botwick, D.G., 1998.
High grade prostatic intraepithelial neoplasia in military working
dogs with and without prostate cancer. Prostate 36, 189
/193.
Bell, F.W., Klausner, J.S., Hayden, D.W., Feeney, D.A., Johnston,
S.D., 1991. Clinical and pathological features of prostatic adeno-
carcinoma in sexually intact and castrated dogs: 31 cases (1970
/
1987). J. Am. Vet. Med. Assoc. 199, 1623
/1630.
Bell, F.W., Klausner, J.S., Hayden, D.W., Lund, E.M., Liebenstein,
B.B., Feeney, D.A., Johnston, S.D., Shivers, J.L., Ewing, C.M.,
Isaacs, W.B., 1995. Evaluation of serum and seminal plasma
markers of canine prostatic disorders. J. Vet. Int. Med. 9, 149
/153.
Bostwick, D.G., 1995. High grade prostatic intraepithelial neoplasia;
the most likely precursor of prostate cancer. Cancer 75 (Suppl.),
1823
/1836.
Bui, M., Reiter, R.E., 1999. Stem cell genes in androgen-independent
prostate cancer. Cancer Metastasis Rev. 17, 391
/399.
Chodak, G.W., Thisted, R.A., Gerber, G.S., Johansson, J.E., Adolfs-
son, J., Jones, G.W., Chisholm, G.D., Moskovitz, B., Livne, P.M.,
Warner, J., 1994. Results of conservative management of clinically
localised prostate cancer. N. Engl. J. Med. 330, 242
/248.
Coffey, D.S., 2001. Similarities of prostate and breast cancer:
evolution, diet, and estrogens. Urology 57 (Suppl. 4A), 31
/38.
Cornell, K.K., Bostwick, D.G., Cooley, D.M., Hall, G., Harvey, H.J.,
Hendrick, M.J., Pauli, B.U., Render, J.A., Stoica, G., Sweet, D.C.,
Waters, D.J., 2000. Clinical and pathological aspects of sponta-
neous canine prostate carcinoma: a retrospective analysis of 76
cases. Prostate 45, 173
/183.
Craft, N., Sawyers, C.L., 1999. Mechanistic concepts in androgen-
dependence of prostate cancer. Cancer Metastasis Rev. 17, 421
/
427.
Johnston, S.D., Kamolpatana, K., Root-Kustritz, M.V., Johnston,
G.R., 2000. Prostatic disorders in the dog. Anim. Reprod. Sci. 60
/
61, 405
/415.
Krawiec, D.R., Heflin, D., 1992. Study of prostatic disease in dogs: 177
cases (1981
/1986). J. Am. Vet. Med. Assoc. 200, 1119
/1122.
Landis, S.H., Murray, T., Bolden, S., Wingo, P.A., 1999. Cancer
statistics, 1999, CA. Cancer J. Clin. 49, 6
/29.
255
Leav, I., King, G.V., 1968. Adenocarcinoma of the canine prostate.
Cancer 22, 1329
/1345.
Leav, I., Schelling, K.H., Adams, J.Y., Merk, F.B., Alroy, J., 2001.
Role of canine basal cells in prostatic post natal development,
induction of hyperplasia, sex hormone-stimulated growth; and the
ductal origin of carcinoma. Prostate 47, 149
/163.
Lowseth, L.A., Gerlach, R.F., Gillet, N.A., Muggenburg, B.A., 1990.
Age-related changes in the prostate and testes of the beagle dog.
Vet. Pathol. 27, 347
/353.
Mahapokai, W., Xue, Y., van Garderen, E., van Sluijs, F.J., Mol, J.A.,
Schalken, J.A., 2000. Cell kinetics and differentiation after
hormonal-induced prostatic hyperplasia in the dog. Prostate 44,
40
/48.
Maini, A., Archer, C., Wang, C.Y., Haas, G.P., 1997. Comparative
pathology of benign prostatic hyperplasia and prostate cancer. In
Vivo 11, 293
/300.
McDonnell, J.D., 1991. The pathophysiology of benign prostatic
hyperplasia.
McNeal, J.E., 1978. Origin and evolution of benign prostatic enlarge-
ment. Invest. Urol. 15, 340
/345.
McNeal, J.E., Bostwick, D.G., Kindrachuk, R.A., Redwine, E.A.,
Freiha, F.S., Stamey, T.A., 1986. Patterns of progression in
prostate cancer. Lancet i, 60
/63.
Obradovich, J., Walshaw, R., Goullaud, E., 1987. The influence of
castration on the development of prostatic carcinoma in the dog. J.
Vet. Int. Med. 1, 183
/187.
Pilat, M.J.P., Kamradt, J.M., Pienta, K.J., 1999. Hormone resistance
in prostate cancer. Cancer Metastasis Rev. 17, 373
/381.
Schoemaker, N.J., Schuurmans, M., Moorman, H., Lumeij, J.T., 2000.
Correlation between age at neutering and age at onset of
hyperdrenocorticism in ferrets. J. Am. Vet. Assoc. 216, 195
/197.
Teske, E., Nickel, R.F., 1996. Zur Aussagekraft der Zytologie bei der
Diagnostik des Prostatakarzinoms beim Hund. Kleintierpraxis 41,
239
/247.
Walsh, P.C., Wilson, J.D., 1976. The induction of prostatic hyper-
trophy in the dog with androstabediol. J. Clin. Inves. 57, 1093
/
1097.
Waters, D.J., Patronek, G.J., Bostwick, D.G., Glickman, L.T., 1996.
Comparing the age at prostate cancer diagnosis in humans and
dogs. J. Natl. Cancer Inst. 88, 1686
/1687.
Waters, D.J., Bostwick, D.G., 1997. Prostatic intraepithelial neoplasia
occurs spontaneously in the canine prostate. J. Urol. 157, 713
/716.
Weaver, A.D., 1981. Fifteen cases of prostatic carcinoma in the dog.
Vet. Rec. 109, 71
/75.
Winters, S.J., Brufsky, A., Weissfeld, J., Trump, D.L., Dyky, M.A.,
Hadeed, V., 2001. Testosterone, sex hormone-binding globulin,
and body composition in young adult African American and
Caucasian men. Metabolism 50, 1242
/1247.
Wu, A.H., Whittemore, A.S., Kolonel, L.N., John, E.M., Gallagher,
R.P., West, D.W., Hanki, J., The, C.Z., Dreon, D.M., Paffenbar-
ger, R.S., 1995. Serum androgens and sex hormone-binding
globulins in relation to lifestyle factor in older African-American,
white, and Asian men in the United States and Canada. Cancer
Epidemiol. Biomarkers Prev. 4, 735
/741.