Prostate cancer risk among men with diabetes mellitus (Spain)
A. Gonza ´ lez-Pe ´ rez,* & L.A. Garcı´a Rodrı´guez
Centro Espan˜ol de Investigacio´n Farmacoepidemiolo´gica (CEIFE), C/Almirante 28 2?, 28004, Madrid, Spain
Received 25 November 2004; accepted in revised form 29 March 2005
Key words: diabetes mellitus, epidemiologic studies, prostate cancer.
Objective: Observational studies have associated diabetes with a decreased risk of prostate cancer. We aimed to
evaluate this association using the General Practitioner Research Database in the UK.
Methods: Population based case–control study nested in a cohort.
Results: We identified 2,183 incident cases of prostate cancer between January 1995 and December 2001. We found
that diabetic patients had a decreased risk of prostate cancer (OR = 0.72; 95% CI: 0.59–0.87). This association was
observed among treated diabetics (OR = 0.63; 95% CI: 0.50–0.80) but not among untreated diabetics (OR = 1.01;
95% CI: 0.73–1.40). Our results suggest that the observed reduced risk could be restricted to users of insulin or
Conclusion: Patients with diabetes have a decreased risk of prostate cancer. The role of antidiabetic treatment in
this association warrants further research.
Several epidemiologic studies have assessed the rela-
tionship between history of diabetes and prostate can-
cer. A recent meta-analysis summarized the evidence
from 14 published studies concluding that patients with
diabetes presented a small decreased risk of prostate
The biological mechanisms for such association are
not fully understood. At least two different pathways
have been proposed. One involves alterations in sex
hormone levels in diabetic patients, and the other in-
volves the insulin-like growth factor (IGF)-1. Testos-
terone controls cell division in the prostate gland  and
high levels of this hormone have been associated with
increased risk of prostate cancer . Furthermore, cas-
trated men rarely develop prostate cancer . Since
diabetic patients have been found to have decreased
androgen levels , this could explain in part the
observed reduced risk of prostate cancer in this popu-
lation. IGF-1 has been shown to stimulate proliferation
and inhibit apoptosis in prostate cancer cells . This
factor has been associated with increased prostate can-
cer risk in some, but not all, epidemiologic studies .
Insulin down-regulates levels of IGF binding protein-1
(IGFBP-1) . Thus, in diabetics with reduced insulin
secretion, circulating IGF-1 levels are indirectly reduced
, which may suppress progression of prostate cancer
to clinical stage.
This study examines the strength of the association
between diabetes, antidiabetic agents and prostate can-
cer using the General Practice Research Database in the
We used data from the General Practice Research
Database (GPRD). This database contains computer-
ized information entered by general practitioners in the
UK . Data on over 2 million patients are systemat-
ically recorded and sent anonymously to the Medicines
and Health products Regulatory Agency (MHRA), that
collects and organizes this information in order to be
* Address for correspondence: Gonza ´ lez-Pe ´ rez A, Centro Espa-
n ˜ ol de Investigacio ´ n Farmacoepidemiolo ´ gica (CEIFE), C/Almirante
28 2?, 28004 Madrid, Spain. Ph. +34 91 524 0237; Fax: +34 91 524
2871 E-mail: firstname.lastname@example.org
Cancer Causes and Control (2005) 16:1055–1058
? Springer 2005
used for research projects. The computerized informa-
tion includes demographics, details from general prac-
titioner’s visits, diagnoses from specialist’s referrals and
hospital admissions, results of laboratory tests and a
free text section. Prescriptions issued by the general
practitioner are directly generated from the computer.
Several studies with the GPRD have documented the
validity and completeness of this database .
We identified all males 50–79 years old between January
1995 and December 2001. Patients with a code for any
cancer before starting date were excluded. The study
cohort comprised a total of 339,462 patients. A more
detailed description of the study cohort can be found
Case ascertainment and validation
We identified 2364 patients with a code of prostate
cancer during the study period, and manually reviewed
their computerized patient profiles. After the review of
the computerized information, 2185 patients were con-
sidered incident cases of prostate cancer. We validated a
random sample of 100 patients requesting the general
practitioner to confirm the diagnosis of prostate cancer.
Since 98% of cases for whom valid information was
received were confirmed as cases we decided not to re-
quest records for the remaining cases ascertained with
the computerized review. At the end, 2183 patients were
considered cases of prostate cancer.
Cohort and nested case–control analysis
Incidence rates of prostate cancer in our study popula-
tion and risk factors were described in detail in a pre-
vious report . A nested case–control analysis was
performed to assess the association between diabetes
and prostate cancer. All cases of prostate cancer
(n = 2183) identified in the study cohort were used in
the case–control analysis and we considered the date of
initial diagnosis as index date. A date during the study
period was generated at random for every member of
the study cohort. If the random date of a study member
was included in his eligible person-time, we used his
random date as the index date and marked that person
as an eligible control. This selection mechanism (i.e.,
incidence density sampling) allows that the likelihood of
being selected as a control is proportional to the person-
time at risk. Ten thousand controls were frequency-
matched by age (interval of 1 year) and calendar year
from the list of all eligible controls.
We ascertained patients with antecedents of prosta-
tism, defined as any mention of benign prostatic
hyperplasia (BPH) and/or prostatism (with or without
surgical intervention) recorded at least more than 1 year
before the index date. We also elicited subjects’ use of
health services (visits to the general practitioner, spe-
cialist referrals, and hospital admissions) in the 2 years
prior to the index date and their body mass index.
Among diabetic patients, we studied use of antidiabetic
drugs: sulphonylureas, biguanides (metformin is the
only available biguanide in the UK), and insulin. We
defined three time windows of exposure for each class of
drugs: current use, past use and never use. Current use
(treated) was categorized as use that lasted until the
index date or ended in the year prior to the index date
based on the supply of drug therapy as prescribed by the
general practitioner. Past use was use that ended more
than one year before the index date. The time window of
never use was defined as non-use of each respective drug
group at any time before the index date. Current users
were subdivided according to treatment duration into
less than 1 year, between 1 and 3 years, and more than
Estimate of odds ratio (OR), assumed to be a valid
estimate of the relative risk, and 95% confidence
interval (CI) associated with prior history of diabetes
compared to no prior history were computed using
unconditional logistic regression. We present the re-
sults from three different multivariate models. In
model 1 estimates are adjusted for age, calendar year,
and NSAID use. Model 2 includes all variables in
model 1 plus body mass index and history of pros-
tatism. The final model – model 3 – includes all
variables in model 2 along with health services utili-
zation. For the analyses presented in Table 1 we cre-
ated a single variable in which we categorized treated
diabetics into users of sulphonylureas only, biguanides
only, oral combination (sulphonylureas and bigua-
nides), insulin only, and insulin oral (insulin plus any
The median age was the same among cases and controls
(72 years old).Atotal of153cases and756 controlshada
1056 A. Gonza´lez-Pe´rez and L.A. Garcı´a Rodrı´guez
history of diabetes, representing prevalence rates of 7.0
and 7.6%, respectively. The percentage of untreated
diabetics was 35% among cases and 24% among con-
We found that diabetes patients had a reduced risk
of prostate cancer compared to non-diabetics. The
age-adjusted relative risk was 0.92 (95% CI: 0.77–
1.10). Adjusting for potential confounders resulted in
a lower relative risk (OR = 0.72, 95% CI: 0.59–
0.87). The observed association between diabetes and
prostate cancer was modified by antidiabetic treat-
ment. Thus, whereas those patients treated with an-
tidiabetic drugs experienced a 37% reduced risk of
prostate cancer, diabetic patients who had never used
these drugs had a similar risk than non-diabetics
(Table 1). Also, we found that the risk of prostate
cancer differed depending on the type of antidiabetic
Overall, current sulphonylureas and insulin use was
associated with an OR of 0.60 (95% CI: 0.42–0.85) and
0.49 (95% CI: 0.26–0.90), respectively. The reduced
risk was present since the beginning of either sulpho-
nylureas or insulin use and remained constant all over
the treatment duration period (data not shown). Cur-
rent use of biguanides (metformin) was not associated
with a decreased risk of prostate cancer (OR = 1.16,
95% CI: 0.63–2.14) after adjusting for other antidia-
betic treatment. We still did not observe a risk reduc-
tion in users of 3 years or more of metformin
(OR = 1.10, 95% CI: 0.60–2.01). The same analysis
was performed among diabetic patients only yielding
virtually identical results.
In our study, diabetes was associated with a reduced risk
of prostate cancer. The magnitude of this inverse asso-
ciation was 30% after adjustment for potential con-
founding factors. This estimate of risk is consistent with
results from three recently published studies [14–16].
Bonovas et al. examined the association between dia-
betes and prostate cancer by conducting a meta-analysis
of the relevant studies published up to 2003. They
analyzed a total of 14 studies and concluded that dia-
betes is associated with a moderate decreased risk of
prostate cancer (RR = 0.91, 95% CI: 0.86–0.96) .
The prevalence of diabetes in our study population
(7.6%) is in line with prior reports among men above
50 years of age in the UK (6.7%) . The proportion
of diabetic patients not treated with hypoglycemic
agents in our study is also consistent with a recently
published study .
To the best of our knowledge there is only one small
published study analyzing the effect of antidiabetic
treatment . The RR estimates for untreated diabet-
ics, as well as for users of sulphonylureas and insulin in
this study (0.99, 0.85, and 0.44, respectively) were simi-
lar to the ones presented here. However sparse data
resulted in very wide confidence intervals. Unfortunately
metformin was not approved in the US during the study
period and no estimate was reported for this drug.
Sulphonylureas and biguanides are the two more
common classes of oral antidiabetics used in our study
population. They have completely different modes of
action. While sulphonylureas act as insulin secreta-
Table 1. Risk of prostate cancer associated with diabetes
OR (95% CI)OR (95% CI)OR (95% CI)
Insulin and oral
aEstimates are adjusted for age, calendar year, NSAID use, and all the variables included in the table using unconditional logistic regression.
bEstimates are adjusted for age, calendar year, NSAID use, prior history of prostatism, body mass index and all the variables included in the
table using unconditional logistic regression.
cEstimates are adjusted for age, calendar year, NSAID use, prior history of prostatism, body mass index, health care utilization (GP visits,
referrals, hospitalizations), and all the variables included in the table using unconditional logistic regression.
Diabetes and prostate cancer1057
gogues, biguanides do not increase insulin circulating Download full-text
levels but reduce hepatic gluconeogenesis and stimulate
glycolysis in tissues . Also biguanides neither in-
crease weight nor provoke hypoglycemia, which makes
them suitable to treat overweight and obese type II
diabetic patients .
According to our results the reduced risk of prostate
cancer was only apparent among users of insulin and
sulphonylureas. These antidiabetic agents elevate insulin
blood levels. Whether the observed association is due to
the effect of these drugs or they act as mere markers for a
group of diabetic patients with impaired insulin secretion
is unclear. A similar rationale could explain the lack of
association among users of biguanides, which are most
effective in treating patients with insulin resistance
syndrome . However, caution must be taken in
interpreting these results, that warrant further research
before any mechanistic conclusion can be made.
We could not differentiate between type I and type II
diabetes. This would tend to dilute the association be-
tween a specific type of diabetes and prostate cancer.
However, given the age range (50–79 years) we expect
most of these patients to have type II diabetes. Also, we
did not have data on date of first diagnosis, and there-
fore we could not assess the effect of time since first
Patients with diabetes tend to have frequent visits to
their GPs as part of their routine care, and consequently
these patients are more likely to undergo prostate cancer
screening tests that could spuriously inflate the incidence
of diagnosed prostate cancer in this population. There-
fore any reduced risk observed among patients with
diabetes could be an underestimation of the true asso-
ciation. In order to overcome this potential bias, we
controlled for health care utilization (number of visits to
the GP, referrals and hospitalizations in the 2 years
prior to the index date) in the multivariate logistic
regression. Indeed, we observed a lower estimate in this
multivariate model compared to other models that did
not adjust for health care utilization.
This study adds further evidence to the suggestion that
diabetic patients present a decreased risk of prostate
cancer. The reduced risk appears to be restricted to those
patients treated either with oral sulphonylureas or insu-
lin, an association that warrants further investigation.
We thank the staff at GPRD, and the participating
general practitioners for their collaboration. We also
thank the Boston Collaborative Drug Surveillance
Program (BCDSP) for providing access to the database.
This study was partly supported by a research grant
1. Bonovas S, Filioussi K, Tsantes A (2004) Diabetes mellitus and
risk of prostate cancer: a meta-analysis. Diabetologia 47: 1071–8.
2. Coffey DS (1979) Physiological control of prostatic growth: an
overview. In: Prostate Cancer. International Union Against Cancer.
UICC Technical Report Series, Vol 48. Geneva, pp. 4–23.
3. Gann PH, Hennekens CH, Ma J, Longcope C, Stampfer MJ
(1996) Prospective study of sex hormone levels and risk of prostate
cancer. J Natl Cancer Inst 88: 1118–1126.
4. Hovenanian MS, Deming CL (1948) The heterologous growth of
cancer of the human prostate. Surg Gynecol Obstet 86: 29–35.
5. Ando S, Rubens R, Rottiers R (1984) Androgen plasma levels in
male diabetes. J Endocrinol Invest 7: 21–24.
6. LeRoith D, Baserga R, Helman L, Roberts CT (1995) Insulin-like
growth factors and cancer. Ann Intern Med 122: 54–59.
7. Stattin P,Rinaldi S,Biessy C,
Kaaks R (2004) High levels of circulating insulin-like growth
factor-I increase prostate cancer risk: a prospective study in a
population-based nonscreened cohort. J Clin Oncol 22: 3104–3112.
8. Suikkari AM,Koivisto VA,
Karonen SL, Seppala M (1988) Insulin regulates the serum levels
of low molecular weight insulin-like growth factor binding protein.
J Clin Endocrinol Metab 66: 266–272.
9. Orskov H (1996) Somatostatin, growth hormone, insulin-like
growth factor-1, and diabetes: friends or foes? Metabolism 45:
10. Garcı´a Rodrı´guez LA, Pe ´ rez Gutthann S (1998) Use of the U.K.
General Practice Research Database for pharmacoepidemiology.
Br J Clin Pharmacol 45: 419–426.
11. Jick H, Jick SS, Derby LE (1991) Validation of information
recorded on general practitioner based computerised data resource
in the United Kingdom. BMJ 302: 766–768.
12. Garcı´a Rodrı´guez LA, Gonza ´ lez-Pe ´ rez A (2004) Inverse associa-
tion between nonsteroidal anti-inflammatory drugs and prostate
cancer. Cancer Epidemiol Biomar Prev 13: 649–653.
13. Ronquist G, Garcı´a Rodrı´guez LA, Ruigo ´ mez A, et al. (2004)
Association between captopril, other antihypertensive drugs and
risk of prostate cancer. Prostate 58: 50–56.
14. Zhu K, Lee IM, Sesso HD, Buring JE, Levine RS, Gaziano JM
(2004) History of diabetes mellitus and risk of prostate cancer in
physicians. Am J Epidemiol 159: 978–982.
15. Coker AL, Sanderson M, Zheng W, Fadden MK (2004) Diabetes
mellitus and prostate cancer risk among older men: population-
based case-control study. Br J Cancer 90: 2171–2175.
16. Coughlin SS, Calle EE, Teras LR, Petrelli J, Thun MJ (2004)
Diabetes mellitus as a predictor of cancer mortality in a large
cohort of US adults. Am J Epidemiol 159: 1160–1167.
17. Joint Health Surveys Unit (1999) Health Survey for England 1998.
London: The Stationery Office.
18. Hippisley-Cox J, Pringle M (2004) Prevalence, care, and outcomes
for patients with diet-controlled diabetes in general practice: cross
sectional survey. Lancet 364: 423–428.
19. Rosenberg DJ, Neugut AI, Ahsan H, Shea S (2002) Diabetes
mellitus and the risk of prostate cancer. Cancer Invest 20: 157–165.
20. DeFronzo RA (1999) Pharmacologic therapy for type 2 diabetes
mellitus. Ann Intern Med 131: 281–303.
21. Scheen AJ, Lefebvre PJ (1998) Oral antidiabetic agents. A guide to
selection. Drugs 55: 225–236.
Stenman UH,Hallmans G,
Rutanen EM,Yki-Jarvinen H,
1058A. Gonza´lez-Pe´rez and L.A. Garcı´a Rodrı´guez