Diabetes and sex-specific colorectal cancer risks in Newfoundland and Labrador: a population-based retrospective cohort study.
Sikdar, Khokan C. ; Walsh, Stephanie J. ; Roche, Madonna 等
Type 2 diabetes and colorectal cancer (CRC) are major causes of
morbidity and mortality in Canada and the burden of these diseases is
rapidly growing. (1,2) In 2008/09, almost 2 million Canadians (about 1
in 15) aged [greater than or equal to] 1 year had been diagnosed with
diabetes and this number is predicted to reach 3.7 million in 2018/19.
Rates are consistently highest in Aboriginal communities and the
Atlantic Provinces. (1) Currently, Newfoundland and Labrador (NL) has
the highest age-standardized prevalence of diabetes in Canada. (1)
Likewise, the number of new CRC cases in Canada has risen by 35% in the
past decade, from an estimated 17,200 new cases in 2001 to an estimated
23,300 cases in 2012. (2,3) CRC is the second leading cause of cancer
mortality in Canada; an estimated 9,200 Canadians died of CRC in 2012.
(2) The province of NL has the highest incidence of CRC in the world.
(4)
Previous studies exploring the relationship between diabetes and
CRC have not been consistent. Some studies report a lack of association,
(5,6) some have found that diabetes increases the risk of CRC, (7-11)
while others have found that the relationship between diabetes and CRC
is subsite-specific, for example, a stronger association for colon
cancer than for rectal cancer. (12-14) The literature in relation to
sex-specific association between diabetes and CRC revealed inconsistent
results; a meta-analysis-based study found a strong relationship between
diabetes and increased risk of CRC in both males and females, (10) while
others have reported the association in women only. (12,15) Other work
has shown that different types of CRC have different etiologies, (16)
which are differentially affected by sex. Thus, it is plausible that
diabetes may affect subsite-specific CRCs differently for males compared
to females.
To further assess the relationship between diabetes and the risk of
CRC(overall and subsite-specific) for males and females, we conducted a
population-based study in NL.
METHODS
This was a population-based retrospective cohort study that used
health administrative databases in NL, a province with a population of
509,000 (in 2012). This study protocol was approved by the Health
Research Ethics Authority of NL.
Data sources
Individuals eligible to be included in the study were residents of
NL aged [greater than or equal to] 30 years at study entry. The study
population was obtained from the Cancer and Chronic Disease Research
Database (CCDRD), which was built for ongoing research related to the
association between cancer and chronic diseases. The patient population
in the CCDRD was assembled by linking two databases: the NL component of
the Canadian Chronic Disease Surveillance System (CCDSS) and the
provincial Oncology Patient Information System (OPIS). The CCDSS
compiles administrative health care data relating to several chronic
conditions, including diabetes. The information from which the CCDSS is
composed includes the provincial health insurance registry, hospital
discharge records, and fee-for-service physician claims. The OPIS is a
province-wide cancer surveillance system and contains information on new
histology-confirmed cancer cases in the province.
Formation of the Diabetes Inception Cohort
The diabetes cohort consisted of individuals with a new diabetes
diagnosis between October 1, 1995 and March 31, 2004. The diabetes cases
were previously extracted using a nationally validated case definition
for diabetes: one or more hospitalizations or two or more
fee-for-service physician claims with a diagnosis of diabetes within a
two-year period. (1) This case criterion was shown to have a high
sensitivity (79.5% to 91%), and specificity (~99%). (17) While we were
unable to distinguish between type 1 and type 2 diabetes, the vast
majority of individuals aged [greater than or equal to] 30 years are
expected to have type 2 diabetes. (10) In order to minimize the
inclusion of individuals having diabetes before the study entry date, a
six-month wash-out period from April 1, 1995 to October 1, 1995 was
used. The study entry date for those with diabetes was the date of their
initial diabetes diagnosis.
Selection of the Comparison Group
Individuals without any evidence of diabetes at baseline were
eligible to be included in the comparison group. Using frequency
matching by 5-year age groups and sex, four non-diabetes individuals
were selected for each diabetes case. Non-diabetes individuals were
assigned the same entry date as their matched diabetes counterparts.
Those who died prior to their assigned study entry date were excluded.
Outcome and follow-up
The outcome of interest was CRC incidence overall and by anatomic
subsite based on a new diagnosis of CRC, for which we adopted a
definition from previous research. (2,9,12) According to the
International Classification of Disease-Oncology, version three
(ICD-O-3) diagnosis codes, cases of CRC comprised codes C18-C21 and C26.
Colon cancers comprised ICD-O-3 codes C18.0-C18.9, with specification as
to whether the diagnosis was in proximal or distal colon (ICD-O-3 codes
C18.0-C18.5 and C18.6-C18.7, respectively). For identification of rectal
cancer, ICD-O-3 codes C19 to C21 were employed. For individuals with and
without diabetes, the follow-up period started one year after their
entry into the study and ended on the earliest of the following events:
incident CRC, death, or end of the study (March 31, 2007). Maximum
follow-up was 10.5 years (from October 1, 1996 to March 31, 2007).
Individuals with previous cancer history or those who developed cancer
within the first year of entry were excluded. Individuals identified as
having cancer or diabetes at baseline in the hospital discharge abstract
or physician billing records, but not in the OPIS or CCDSS, were also
excluded.
Covariates
The following covariates were included in the analysis: baseline
age, sex, and severity of illness. Severity of illness at baseline was
estimated using the Charlson Comorbidity Index (CCI). (18) The CCI
measure was used to control for the presence of a number of serious
health conditions for the 18 months prior to study entry that may have
altered cancer diagnosis rates. After identifying co-morbidities through
diagnosis codes in the fee-for-service physician claims and hospital
discharge abstracts databases, a co-morbidity score (representing
severity of illness) was assigned to each individual based on the
presence or absence of 13 specific conditions, identified using
ICD-9(all physician claims, and hospital records prior to April 1, 2001)
and ICD-10-CA codes (hospital records, as of April 1, 2001). Given that
the aim was to assess the risk of developing CRC for those with
diabetes, the two diabetes (diabetes and diabetes with complications)
and two cancer (cancer and metastatic cancer) co-morbidities were
removed from the CCI.
Statistical analysis
For individuals with and without diabetes, characteristics of the
study subjects were compared between males and females using
Student's t-tests for continuous variables and [chi square] tests
for categorical variables. Severity of illness was categorized as
"not severe"(CCI = 0) and "severe"(CCI = [greater
than or equal to] 1). Overall and sex-specific incidences of CRC were
calculated separately for individuals with and without diabetes by
dividing the number of incidence cases by category-specific person-years
and presented by age group and severity of illness. Sex-specific
incidence rates of CRC by anatomic subsite (i.e., colon, proximal colon,
distal colon and rectal cancers) were calculated following a similar
manner used in calculating CRC incidence. Ninety-five per cent
confidence intervals (95% CIs) for the CRC incidence rates were
calculated assuming a Poisson distribution. To examine unadjusted CRC
risk, we compared Kaplan-Meier estimates of survival probabilities
between diabetes and non-diabetes groups. Cox proportional hazards
regression, adjusted for age and severity of illness, was performed to
estimate hazard ratios (HR) with corresponding 95% CIs for overall and
subsite-specific CRC associated with diabetes. To investigate whether
any observed association of diabetes with CRC was due to a high
proportion of overweight and obese individuals in the diabetes group,
the relative risk (RR) of diabetes with CRC was calculated after
adjusting for overweight and obesity in the population (Appendix A). To
do this adjustment, the RR of overweight/obesity with CRC was obtained
from a published meta-analysis of 15 cohort studies, (19) and the
prevalence of overweight and obese in the NL population aged 30 years
and older was derived from the 2005 public use micro data file of the
Canadian Community Health Survey (CCHS) Cycle 3.1. These two estimates
were used to derive the obesity-attributable CRC cases in our study
cohort, which was subsequently subtracted from overall CRC cases to
adjust for obesity effect on CRC. All statistical analysis was performed
using SAS version 9.2 (SAS Institute Inc., Cary, NC) software.
RESULTS
The initial study cohort comprised 130,710 individuals (26,142 with
diabetes and 104,568 without diabetes). Following exclusions mentioned
above, the analysis cohort consisted of a total of 122,228 individuals
(25,304 with diabetes and 96, 924 without diabetes). Characteristics of
the study subjects are presented in Table 1. There were almost equal
numbers of males and females, 61,156 (50%) and 61,072(50%),
respectively. Females were older than males (mean age 58.4 versus 55.9
years, p<0.01). Females with diabetes were also older than males with
diabetes (58.7 versus 56.3 years, p<0.01) as were females without
diabetes compared to males without diabetes (58.3 versus 55.9 years,
p<0.01).
[FIGURE 1 OMITTED]
During 751,562 person-years of follow-up, a total of 1,663 newly
diagnosed CRC cases(414 for diabetes and 1,249 for non-diabetes) were
identified(Table 2). The incidence of CRC was higher for individuals
with diabetes compared to those without diabetes (28.7, 95% CI 25.9-31.4
versus 20.6, 95% CI 19.4-21.7 per 10,000 person-years) and increased
with age. The same pattern was observed when the analysis was stratified
by sex. As shown in Figure 1, diabetes negatively affected survival of
patients with CRC in both sexes.
Incidence rates of CRC by anatomic subsite and hazard ratios
stratified by sex are presented in Table 3. Among males, incidences of
specific CRC were higher for individuals with diabetes compared to those
without diabetes for colon (22.4, 95% CI 18.9-25.9 versus 15.5, 95% CI
14.1-16.9 per 10,000 person-years) and distal colon (9.2, 95% CI
6.9-11.4 versus 5.8, 95% CI 4.9-6.7 per 10,000 person-years). Among
females, a higher incidence of colon (17.4, 95% CI 14.4-20.5 versus
12.3, 95% CI 11.1-13.5 per 10,000 person-years) and proximal colon
cancer (11.5, 95% CI 9.1-14.0 versus 7.7, 95% CI 6.7-8.7 per 10,000
person-years) was found for those with diabetes compared to those
without diabetes, respectively. Overlapping confidence intervals
indicated that incidences of proximal colon cancer in males with and
without diabetes were similar, and incidences of distal colon cancer in
the two groups for females were similar. For both sexes, rectal cancer
incidence rates were not significantly different for those with and
without diabetes.
As shown in Table 3, after adjusting for age and severity of
illness, CRC risk was increased by 38% (HR=1.38, 95% CI 1.19-1.60) among
males with diabetes and by 52% (HR=1.52, 95% CI 1.27-1.80) among females
with diabetes, compared to males and females without diabetes.
Subsite-specific stratified analyses indicated that among males,
diabetes was positively associated with overall colon cancer risk
(HR=1.49, 95% CI 1.24-1.78), proximal colon cancer risk (HR=1.35, 95% CI
1.05-1.78) and distal colon cancer risk (HR=1.61, 95% CI 1.21-2.15).
Among females, diabetes was significantly associated with increased risk
of overall colon cancer (HR =1.47, 95% CI 1.20-1.80), proximal colon
cancer (HR=1.58, 95% CI 1.22-2.02) and rectal cancer (HR=1.56, 95% CI
1.10-2.22). No significant association was observed for diabetes and the
risk of rectal cancer in males, nor for diabetes and the risk of distal
colon cancer in females.
Obesity-adjusted RR of CRC on diabetes estimated separately for
males and females demonstrates that controlling for obesity did not
substantially alter the degree of CRC risk on diabetes. Following this
adjustment, the RR of diabetes with CRC was 1.30 (95% CI 1.28-1.32)
among males and 1.44 (95% CI 1.40-1.48) among females. These results did
not differ from the HR (1.38, 95% CI 1.19-1.60 for males and 1.52, 95%
CI 1.27-1.80 for females) when obesity-attributable CRC cases were
included in the analysis.
DISCUSSION
This is the first population-based cohort study that employed
population-based administrative data to examine the relationship between
diabetes and CRC incidence in Canada. We found that diabetes was
associated with a 38% increased risk of CRC among males and a 52%
increased risk of CRC among females. Adjusting for obesity effect did
not alter the diabetes-associated risk of CRC. Findings from this study
support those of other cohort studies that have demonstrated a
significant increase in CRC incidence among individuals with diabetes.
(8-11) The CRC incidence rates in this study are notably higher than
those of other studies. For example, one population-based cohort study
from Singapore reported 208.9 and 140.2 incidence per 100,000
person-years for those with diabetes and those without diabetes,
respectively, (20) which is considerably lower than the 287 and 206 per
100,000 person-years, respectively, found in the current study. The
higher rates in this study likely reflect the added influence of Western
lifestyle factors.
The findings of this study are consistent with the hypothesis that
the relationship between diabetes and CRC is subsite-specific (12,13)
and, furthermore, that the patterns of subsite-specific associations
differ for males and females. For males, diabetes significantly
increases the risk of overall CRC, as well as proximal, distal, and
overall colon cancers. Diabetes does not significantly increase the risk
of rectal cancer in males. For females, diabetes significantly increases
the risk of overall CRC, as well as proximal and overall colon and
rectal cancers, but not of distal colon cancer. The results suggest that
there is a stronger association between diabetes and CRC for females
than for males; however, this trend was not significant. This finding
may be due primarily to the contributions of proximal colon and rectal
cancers.
The findings of the current study also support those of studies
that have shown that diabetes increases the risk of proximal colon
cancer, but not distal colon cancer, in females. (12,14) There are
several plausible reasons for why the association between diabetes and
distal CRC was observed in males only. Previous studies have
demonstrated that smoking significantly increases insulin resistance,
(21) which is thought to increase CRC risk; (22,23) it is possible that
the distal colon may be especially sensitive to increased insulin and
insulin-like growth factors. (24) One study which reported a similar
result also found a significantly higher proportion of smokers among
males with diabetes than among females with diabetes. (8) In the current
study, smoking status was unknown. Another possible explanation for sex
differences may be the differential moderating effect of estrogen on
levels of insulin and IGFs, as estrogen has been linked to reduced serum
IGF levels. (24,25) There is also evidence to suggest that different
genetic pathways to CRC dominate in the proximal and distal colon, which
are influenced by different sex-related factors; (26) these pathways may
differentially interact with diabetes status.
The results of this study should be interpreted in the context of
the limitations of the available data. Primarily, the magnitude of the
association between type 2 diabetes and CRC risk may be underestimated
for several reasons. First, it was not possible to distinguish between
type 1 and type 2 diabetes. This is important as type 1 diabetes (which
accounts for about 5-10% of the total population living with diabetes)
(1) may not be related to CRC. (12) Second, this study included only
diabetes cases identified using a validated case definition applied to
administrative data. Those with diabetes or its precursors who did not
meet the criteria for the CCDSS definition (e.g., they had only one
physician claim with a diabetes diagnosis) may have been included in the
comparison group as an individual without diabetes. Third, in later
stages of diabetes, insulin levels might decline, which may result in
variable associations between diabetes and cancer risk, if the
hyperinsulinemia hypothesis holds true.(22) Previous work has shown that
the risk of CRC may decrease with increased follow-up time. (27) Future
work should include all diagnoses of pre-diabetes, hyperinsulinemia,
hyperglycemia, and other factors related to insulin resistance, as well
as a more in-depth analysis by follow-up time.
A further limitation of the current study is that, since type 2
diabetes and CRC share common risk factors such as smoking, physical
inactivity, Western diet, and obesity, the observed increased risk of
CRC associated with a history of diabetes may be confounded by these
factors. However, as we have shown, controlling for obesity did not
alter the association. Also, a meta-analysis found a positive
association between diabetes and CRC when the analysis was limited to
studies that controlled for activity level and BMI. (10) In future work,
controlling for lifestyle factors would reduce error variation and
provide a more precise estimate of the strength of the relationship
between diabetes and CRC. Similarly, data on insulin use and other
medications may be another contributing factor to be included in future
investigations, as insulin therapy has also been found to increase CRC
risk. (7,28,29)
While the data used in this study posed some minor problems in
terms of missing information and other discrepancies, using
population-based and large-size administrative datasets along with a
cohort design is a major strength which has the ability to provide the
reliable estimates for diabetes and cancer incidence available at this
time for NL and many other provinces. This design allowed us to identify
the cases of CRC and carry out successful record linkage of patients via
a unique health care number. The cohort design and record linkage of
population-based historical data enable us to alleviate issues related
to selection and recall bias. Also, exclusion of individuals who
developed CRC within the first year of diabetes diagnosis was a
thoughtful measure that allowed us to mitigate detection bias or
overestimation of the risk. Because of the large datasets and long
follow-up period, we were able to identify a relatively large number of
CRC cases and, thus, examine the association with diabetes by subsite
and stratify the analysis by sex.
Appendix A. Approach to the estimation of diabetes-attributable
risk for colorectal cancer after adjusting for overweight/obesity; where
colorectal cancer (CRC) is the outcome, diabetes is an exposure and
overweight/obesity (OB) is a known confounder.
The formulae for attributable risk percent (AR%) and population
attributable risk percent (PAR%) were the modified version of the
previously used formulae. (31,32)
Formula 1a
AR% was calculated using the overweight/obesity-associated relative
risk ([RR.sub.ob]) of colorectal cancer
AR% = [[RR.sub.ob]-1]/[RR.sub.ob]
where [RR.sub.ob] is the relative risk of CRC associated with
overweight/obesity, obtained from Dai et al.; (19) [RR.sub.ob]-1
indicates excess relative risk of CRC for overweight/obesity.
Formula 1b
PAR% was calculated separately for diabetes and non-diabetes group
according to the formula
PAR% = AR% * p,
where p is the proportion of population overweight/obese in
diabetes and non-diabetes groups, obtained from the Canadian Community
Health Survey Cycle 3.1.
Formula 2a
The overweight/obesity attributable CRC cases ([x.sub.ob]) were
estimated separately for diabetes and non-diabetes groups
[x.sub.ob] = x * PAR%,
where x is the number of CRC cases obtained from the study sample
in each of the diabetes and non-diabetes groups.
Then x and [x.sub.ob] were used to estimate the CRC cases
([x.sub.adj]) in diabetes and non-diabetes groups after adjusting for
overweight/obesity
[x.sub.adj] = x-[x.sub.ob]
Formula 2b
The final step is the calculation of diabetes associated relative
risk ([RR.sub.adj]) of colorectal cancer after adjusting for
overweight/obesity using the following formula
[RR.sub.adj] = [Incidence of CRC in diabetes group after adjusting
for overweight/obesity]/[Incidence of CRC in non- diabetes group after
adjusting for overweight/obesity]
[RR.sub.adj] = [[x.sub.adj] in diabetes group/Person-years in
diabetes group]/[[x.sub.adj] in non-diabetes group/Person-years in
non-diabetes group]
In summary, the results of this study have important clinical and
public health implications as an association between diabetes and
increased risk of CRC in both males and females was found. These
findings also provide indirect epidemiological evidence for the
hypothesis that either hyperinsulinemia or factors related to insulin
resistance may play a role in increasing the risk of CRC by promoting
growth of colon tumours, stimulating insulin-like growth factor
receptors and acting as cell mitogen. (30) Given the evidence of shared
etiologies, along with the increasing burden of both diabetes and CRC in
Canada, these findings have implications for screening protocols and
preventive initiatives. Also, preventive initiatives should directly
address the shared risk factors (smoking, Western diet, obesity, and
sedentary lifestyle). Future studies will be necessary to demonstrate
whether lessening the burden of hyperinsulinemia and factors related to
insulin resistance will be an effective strategy in the prevention of
both type 2 diabetes and CRC incidence.
Conflict of Interest: None to declare.
REFERENCES
(1.) Public Health Agency of Canada. Diabetes in Canada, 2011:
Facts and Figures From a Public Health Perspective. Ottawa, ON: Public
Health Agency of Canada, 2011. Available at:
http://www.phac-aspc.gc.ca/cd-mc/publications/diabetes-diabete/facts-figures-faits- chiffres-2011/pdf/facts-figures-faits- chiffres-eng.pdf
(Accessed March 14, 2013).
(2.) Canadian Cancer Society. Canadian Cancer Statistics 2012.
Ottawa: Canadian Cancer Society, 2010. Available at:
http://www.cancer.ca/~/media/CCS/
Canada%20wide/Files%20List/English%20files%20heading/PDF%20-%20Policy%20-%20Canadian%20Cancer%20Statistics%20- %20English/Canadian%20Can-
cer%20Statistics%202012%20-%20English.ashx (Accessed August 27, 2012).
(3.) Canadian Cancer Society. Canadian Cancer Statistics 2001.
Ottawa: Canadian Cancer Society, 2001. Available at:
http://publications.gc.ca/collections/
collection_2011/statcan/CS2-37-2001-eng.pdf (Accessed June 22, 2012).
(4.) Zhao J, Halfyard B, West R, Buehler S, Sun Z, Squires J, et
al. Tobacco smoking and colorectal cancer: A population-based
case-control study in Newfoundland and Labrador. Can J Public Health
2010; 101:281-89.
(5.) Kune GA, Kune S, Watson LF. Colorectal cancer risk, chronic
illnesses, operations, and medications: Case control results from the
Melbourne Colorectal Cancer Study. Cancer Res 1988; 48:4399-404.
(6.) O'Mara BA, Byers T, Schoenfeld E. Diabetes mellitus and
cancer risk: A multisite case control study. J Chron Dis 1985;
38:435-41.
(7.) Berster JM, Goke B. Type 2 diabetes mellitus as risk factor
for colorectal cancer. Arch Physiol Biochemistry 2008; 114:84-98.
(8.) Limburg PJ, Vierkant RA, Fredericksen ZS, Leibson CL, Rizza
RA, Gupta AK, et al. Clinically confirmed type 2 diabetes mellitus and
colorectal cancer risk: A population-based, retrospective cohort study.
Am J Gastroenterol 2006;101:1872-79.
(9.) Inoue M, Iwasaki M, Otani T, Sasazuki S, Noda M, Tsugane S.
Diabetes mellitus and the risk of cancer: Results from a large-scale
population-based cohort study in Japan. Arch Intern Med
2006;166:1871-77.
(10.) Larsson SC, Orsini N, Wolk A. Diabetes mellitus and risk of
colorectal cancer: A meta-analysis. J Natl Cancer Inst 2005;97:1679-87.
(11.) Coughlin SS, Calle EE, Teras LR, Petrelli J, Thun MJ.
Diabetes mellitus as a predictor of cancer mortality in a large cohort
of US adults. Am J Epidemiol 2004;159:1160-67.
(12.) Limburg PJ, Anderson KE, Johnson TW, Jacobs DR Jr, Lazovich
D, Hong CP, et al. Diabetes mellitus and subsite-specific colorectal
cancer risks in the Iowa Women's Health Study. Cancer Epidemiol
Biomarkers Prev 2005; 14:133-37.
(13.) Lim E, Gibbs P, McLaughlin S, Johns J. Subsite-specific
colorectal cancer in diabetic and nondiabetic patients. Cancer Epidemiol
Biomarkers Prev 2005;14:1579.
(14.) Hu FB, Mason JE, Liu S, Hunter D, Colditz GA, Michels KB, et
al. Prospective study of adult onset diabetes mellitus (type 2) and risk
of colorectal cancer in women. J Natl Cancer Inst 1999; 91:542-47.
(15.) Levi F, Pasche C, Lucchini F, La Vecchia C. Diabetes
mellitus, family history, and colorectal cancer. J Epidemiol Community
Health 2002;56:479-80.
(16.) Nilson TI, Vatten LJ. Prospective study of colorectal cancer
risk and physical activity, diabetes, blood glucose and BMI: Exploring
the hyperinsulinaemia hypothesis. Br J Cancer 2001;84:417-22.
(17.) Public Health Agency of Canada. National Diabetes
Surveillance System Methods Documentation, 2008. Ottawa: Public Health
Agency of Canada, 2009. Available at:
www.phac-aspc.gc.ca/ccdpc-cpcmc/ndss-snsd/english/diabetes_data/00-06/pdf/method_v208-eng.pdf (Accessed June 22, 2012).
(18.) Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of
classifying prognostic comorbidity in longitudinal studies: Development
and validation. J Chron Dis 1987; 40:373-83.
(19.) Dai Z, Xu Y, Niu L. Obesity and colorectal cancer risk: A
meta-analysis of cohort studies. World J Gastroenterol 2007;13:4199-206.
(20.) Seow A, Yuan JM, Koh WP, Lee HP, Yu MC. Diabetes mellitus and
risk of colorectal cancer in the Singapore Chinese health study. J Natl
Cancer Inst 2006;98:135-38.
(21.) Oh SW, Kim YH, Choi YS, Chang DK, Son HJ, Rhee PL, et al. The
comparison of the risk factors and clinical manifestations of proximal
and distal colorectal cancer. Dis Colon Rectum 2008;51:56-61.
(22.) Giovannucci E. Modifiable risk factors for colon cancer.
Gastroenterol Clin North Am 2003;31:925-43.
(23.) Giovannucci E, Michaud D. The role of obesity and related
metabolic disturbances in cancers of the colon, prostate, and pancreas.
Gastroenterology 2007;132:2208-25.
(24.) Tricoli JV, Rall LB, Karakousis CP, Herrera L, Petrelli NJ,
Bell GI, Shows TB. Enhanced levels of insulin-like growth factor
messenger RNA in human colon carcinomas and liposarcomas. Cancer Res
1986;46:6169-73.
(25.) Morimoto LM, Newcomb PA, White E, Bigler J, Potter JD.
Variation in plasma insulin-like growth factor-1 and insulin-like growth
factor binding protein-3: Personal and lifestyle factors (United
States). Cancer Causes Control 2005;16:917-27.
(26.) Breivik J, Lothe RA, Meling GI, Rognum TO, B0rresen-Dale AL,
Gaudernack G. Different genetic pathways to proximal and distal
colorectal cancer influenced by sex-related factors. Int J Cancer
1997;74:664-69.
(27.) Weiderpass E, Gridley G, Nyren O, Ekbom A, Persson I, Adami
HO. Diabetes mellitus and risk of large bowel cancer. J Natl Cancer Inst
1997;89:660-61.
(28.) Hemkens LG, Grouven U, Bender R, GUnster C, Gutschmidt S,
Selke GW, Sawicki PT. Risk of malignancies in patients with diabetes
treated with human insulin or insulin analogues: A cohort study.
Diabetologia 2009;52:1732-44.
(29.) Yang YX, Hennessy S, Lewis JD. Insulin therapy and colorectal
cancer risk among type 2 diabetes mellitus patients. Gastroenterology
2004;127:1044-50.
(30.) Will JC, Galuska DA, Vinicor F, Calle EE. Colorectal cancer:
Another complication of diabetes mellitus. Am J Epidemiol
1998;147:816-25.
(31.) Parkin DM, Boyd L. Cancer attributable to overweight and
obesity in the UK in 2010. Br J Cancer 2011;105:S34-S37.
(32.) Simons E, To T, Dell S. The population attributable fraction
of asthma among Canadian children. Can J Public Health 2011;
102(1):35-41.
Received: September 13, 2012
Accepted: December 8, 2012
Khokan C. Sikdar, PhD, [1,2] Stephanie J. Walsh, MSc, [3] Madonna
Roche, MSc, [2,3] Ying Jiang, MD, MSc, [4] Ania Syrowatka, MSc, [4]
Kayla D. Collins, MSc, PhD, [2,3]
Author Affiliations
[1.] School of Public Health, University of Alberta, Edmonton, AB
[2.] Faculty of Medicine, Memorial University of Newfoundland, St.
John's, NL
[3.] Research and Evaluation Department, Newfoundland and Labrador
Centre for Health Information, St. John's, NL
[4.] Centre for Chronic Disease Prevention and Control, Public
Health Agency of Canada, Ottawa, ON
Correspondence: Khokan C. Sikdar, Research and Evaluation,
Community Oncology, Alberta Health Services-Cancer Care, 1500 Sun Life
Place, 10123 99 St NW, Edmonton, AB T5J 3H1, Tel: 780-643-4345, E-mail:
[email protected] Acknowledgements: The authors
thank Jeffrey Dowden for assistance in data analysis, and gratefully
acknowledge support of the Newfoundland and Labrador Centre for Health
Information for facilitating this work and presenting at the Health Data
Users Conference held in Ottawa, ON, Canada, November 22-23, 2011. This
study was financially supported by the Public Health Agency of Canada.
The funding source played no role in the study design, the analysis or
the interpretation of the data.
Table 1. Characteristics of the Study
Sample by Diabetes Status and Sex
Diabetes
Males Females p-value
(n=12,667) (n=12,637)
Mean age * (yr) (SD) 56.3 (12.8) 58.7 (14.3) <0.01
Age at baseline
([dagger]) (yr) (%)
30-44 2644 (20.9) 2467 (19.5) <0.01
45-54 3494 (27.6) 2863 (22.7)
55-64 3137 (24.8) 2782 (22.0)
65-74 2315 (18.3) 2650 (21.0)
75+ 1077 (8.5) 1875 (14.8)
Co-morbidities at
baseline ([dagger])
([double dagger]) (%)
AMI 460 (3.6) 276 (2.2) <0.01
CHF 576 (4.6) 577 (4.6) 0.94
PVD 395 (3.1) 300 (2.4) <0.01
CVD 391 (3.1) 333 (2.6) <0.05
Dementia 99 (0.8) 170 (1.4) <0.01
PD 1187 (9.4) 1369 (10.8) <0.01
CTD 688 (5.4) 1060 (8.4) <0.01
Peptic ulcer 292 (2.3) 265 (2.1) 0.26
Liver disease 88 (0.7) 53 (0.4) <0.01
Paraplegia 52 (0.4) 44 (0.4) 0.42
Renal disease 202 (1.6) 191 (0.5) 0.59
SLD 14 (0.1) 4 (0.0) <0.05
Mean CCI score * (SD) 0.4 (0.7) 0.4 (0.7) 0.15
Severity of illness ([dagger])
([section])
Not severe (%) 9,331 (73.7) 9135 (72.3) <0.05
Severe (%) 3,336 (26.3) 3502 (27.7)
No Diabetes
Males Females p-value
(n=48,489) (n=48,435)
<0.01
Mean age * (yr) (SD) 55.9 (12.8) 58.3 (14.3)
Age at baseline
([dagger]) (yr) (%) <0.01
30-44 10,478 (21.6) 9734 (20.1)
45-54 13,596 (28.0) 11,128 (23.0)
55-64 11,940 (24.6) 10,614 (21.9)
65-74 8637 (17.8) 10,011 (20.7)
75+ 3838 (7.9) 6948 (14.3)
Co-morbidities at
baseline ([dagger])
([double dagger]) (%) <0.01
AMI 828 (1.7) 432 (0.9) 0.19
CHF 895 (1.9) 840 (1.7) <0.01
PVD 770 (1.6) 527 (1.1) <0.14
CVD 712 (1.5) 657 (1.4) <0.01
Dementia 234 (0.5) 497 (1.0) <0.60
PD 2874 (5.9) 2909 (6.0) <0.01
CTD 1924 (4.0) 2778 (5.7) <0.05
Peptic ulcer 824 (1.7) 727 (1.5) 0.15
Liver disease 97 (0.2) 78 (0.2) 0.27
Paraplegia 111 (0.2) 95 (0.2) <0.01
Renal disease 367 (0.8) 229 (0.5) 0.32
SLD 21 (0.0) 15 (0.0) 0.87
Mean CCI score * (SD) 0.2 (0.6) 0.2 (0.5)
Severity of illness
([dagger]) ([section]) 0.01
Not severe (%) 40,924 (84.4) 40,556 (83.7)
Severe (%) 7565 (15.6) 7879 (16.3)
AMI = acute myocardial infarction, CHF = congestive heart failure,
PVD = peripheral vascular disease, CVD = cerebral vascular disease,
PD = pulmonary disease,
CTD = connective tissue disorder, SLD = severe liver disease,
CRC = colorectal cancer, CCI = Charlson Comorbidity Index.
* Comparisons of demographic and clinical characteristics between
males and females were performed using Student's t-tests.
([dagger]) Comparisons of demographic and clinical characteristics
between males and females were performed using [chi square] tests.
([double dagger]) HIV (Human Immunodeficiency Virus) was excluded
from individual co-morbidities due to low numbers of incident cases.
([section]) Not severe: 0; severe: 1+.
Table 2. Colorectal Cancer Incidence Among Diabetes and Non-Diabetes
Groups
Diabetes
CRC
Incidence/
PY of CRC 10,000 PY
n Follow-up Cases (95% CI) *
Both sexes
All individuals 25,304 144,427 414 28.7 (25.9-31.4)
Age at baseline (years)
30-44 5111 30,821 21 6.8 (3.9-9.7)
45-54 6357 38,863 78 20.1 (15.6-24.5)
55-64 5919 36,150 125 34.6 (28.5-40.6)
65-74 4965 27,335 130 47.6 (39.4-55.7)
75+ 2952 11,258 60 53.3 (39.8-66.8)
Severity of
illness ([dagger])
Not severe 18,466 110,753 307 27.7 (24.6-30.8)
Severe 6838 33,675 107 31.8(25.8-37.8)
Males
Total 12,667 71,676 241 33.6 (29.4-37.9)
Age at baseline (years)
30-44 2644 15,665 14 8.9 (4.3-13.6)
45-54 3494 21,449 44 20.5 (14.5-26.6)
55-64 3137 18,778 80 42.6 (33.3-51.9)
65-74 2315 12,061 73 60.5 (46.6-74.4)
75+ 1077 3724 30 80.6 (51.7-109.4)
Severity of
illness ([dagger])
Not severe 9331 55,650 178 32.0 (27.3-36.7)
Severe 3336 16,026 63 39.3 (29.6-49.0)
Females
Total 12,637 72,751 173 23.8(20.2-27.3)
Age at baseline (years)
30-44 2467 15,155 7 4.6 (1.2-8.0)
45-54 2863 17,414 34 19.5 (13.0-26.1)
55-64 2782 17,372 45 25.9 (18.3-33.5)
65-74 2650 15,274 57 37.3 (27.6-47.0)
75+ 1875 7534 30 39.8 (25.6-54.1)
Severity of
illness ([dagger])
Not severe 9135 55,102 129 23.4 (19.4-27.5)
Severe 3502 17,648 44 24.9 (17.6-32.3)
No Diabetes
CRC
Incidence/
PY of CRC 10,000 PY
n Follow-up Cases (95% CI) *
Both sexes
All individuals 96,924 607,135 1249 20.6 (19.4-21.7)
Age at baseline (years)
30-44 20,212 124,102 56 4.5 (3.3-5.7)
45-54 24,724 157,063 202 12.9 (11.1-14.6)
55-64 22,554 148,230 356 24.0 (21.5-26.5)
65-74 18,648 120,898 434 35.9 (32.5-39.3)
75+ 10,786 56,843 201 35.4 (30.5-40.2)
Severity of
illness ([dagger])
Not severe 81,480 521,193 1023 19.6 (18.4-20.8)
Severe 15,444 85,942 226 26.3 (22.9-29.7)
Males
Total 48,489 299,162 749 25.0 (23.2-26.8)
Age at baseline (years)
30-44 10,478 63,482 40 6.3 (4.3-8.3)
45-54 13,596 86,549 134 15.5 (12.9-18.1)
55-64 11,940 77,094 237 30.7 (26.8-34.7)
65-74 8637 53,742 261 48.6 (42.7-54.5)
75+ 3838 18,294 77 42.1 (32.7-51.5)
Severity of
illness ([dagger])
Not severe 40,924 258,178 615 23.8 (21.9-25.7)
Severe 7565 40,984 134 32.7 (27.2-38.2)
Females
Total 48,435 307,973 500 16.2 (14.8-17.7)
Age at baseline (years)
30-44 9734 60,619 16 2.6 (1.3-3.9)
45-54 11,128 70,514 68 9.6 (7.4-11.9)
55-64 10,614 71,135 119 16.7 (13.7-19.7)
65-74 10,011 67,156 173 25.8 (21.9-29.6)
75+ 6948 38,549 124 32.2 (26.5-37.8)
Severity of
illness ([dagger])
Not severe 40,556 263,015 408 15.5 (14.0-17.0)
Severe 7879 44,958 92 20.5 (16.3-24.6)
PY = person-years, CRC = colorectal cancer, CI = confidence intervals.
* 95% CIs for the CRC incidence rates were calculated based on the
Poisson distribution.
([dagger]) Calculated based on the Charlson Comorbidity Index. (31)
Not severe: 0; severe: 1+.
Table 3. Diabetes and HRs of Incident Colorectal Cancer Overall and
by Cancer Subsite
Males
Diabetes No Diabetes p-value
Colorectal cancer
n * 241 749
Person-years 71,676 299,162
Incidence/10,000 (95% CI) 33.6 25.0
([dagger]) (29.4-37.8) (23.2-26.8)
Multivariate HR ([double 1.38 1.00
dagger]) (95% CI) ([dagger]) (1.19-1.60) (reference) <0.0001
Colon cancer
n * 161 464
Person-years 71,868 299,829
Incidence/10,000 (95% CI) 22.4 15.5
([dagger]) (18.9-25.9) (14.1-16.9)
Multivariate HR ([double 1.49 1.00
dagger]) (95% CI) ([dagger]) (1.24-1.78) (reference) <0.0001
Proximal colon cancer
n 78 246
Person-years 72,078 300,346
Incidence/10,000 (95% CI) 10.8 8.2
([dagger]) (8.4-13.2) (7.2-9.2)
Multivariate HR ([double 1.35 1.00
dagger]) (95% CI) ([dagger]) (1.05-1.78) (reference) 0.023
Distal colon cancer
n 66 174
Person-years 72,119 300,495
Incidence/10,000 (95% CI) 9.2 5.8
([dagger]) (6.9-11.4) (4.9-6.7)
Multivariate HR ([double 1.61 1.00
dagger]) (95% CI) ([dagger]) (1.21-2.15) (reference) 0.001
Rectal cancer
n 79 281
Person-years 72,101 300,253
Incidence/10,000 (95% CI) 11.0 9.5
([dagger]) (8.5-13.4) (8.3-10.5)
Multivariate HR ([double 1.19 1.00
dagger]) (95% CI) ([dagger]) (0.93-1.53) (reference) 0.165
Females
Diabetes No Diabetes p-value
Colorectal cancer
n * 173 500
Person-years 72,751 307,973
Incidence/10,000 (95% CI) 23.8 16.2
([dagger]) (20.3-27.3) (14.8-17.6)
Multivariate HR ([double 1.52 1.00
dagger]) (95% CI) ([dagger]) (1.27-1.80) (reference) <0.0001
Colon cancer
n * 127 379
Person-years 72,888 308,286
Incidence/10,000 (95% CI) 17.4 12.3
([dagger]) (14.4-20.5) (11.1-13.5)
Multivariate HR ([double 1.47 1.00
dagger]) (95% CI) ([dagger]) (1.20-1.80) (reference) 0.0002
Proximal colon cancer
n 84 237
Person-years 72,967 308,622
Incidence/10,000 (95% CI) 11.5 7.7
([dagger]) (9.1-14.0) (6.7-8.7)
Multivariate HR ([double 1.58 1.00
dagger]) (95% CI) ([dagger]) (1.22-2.02) (reference) 0.0004
Distal colon cancer
n 30 109
Person-years 73,072 308,952
Incidence/10,000 (95% CI) 4.1 3.5
([dagger]) (2.6-5.6) (2.9-4.2)
Multivariate HR ([double 1.19 1.00
dagger]) (95% CI) ([dagger]) (0.79-1.79) (reference) 0.398
Rectal cancer
n 43 120
Person-years 73,011 308,916
Incidence/10,000 (95% CI) 5.9 3.9
([dagger]) (4.1-7.6) (3.2-4.6)
Multivariate HR ([double 1.56 1.00
dagger]) (95% CI) ([dagger]) (1.10-2.22) (reference) 0.012
HR = hazard ratio; CI = confidence intervals.
* The numbers of proximal colon, distal colon, and rectal cancers do
not add up to the total number of CRCs, and the numbers of proximal
colon and distal colon cancers do not add to the total number of
colon cancers, because, in some cases, information on the specific
site was unknown.
([dagger]) 95% CIs for the CRC incidence rates were calculated based
on the Poisson distribution.
([double dagger]) Adjusted for age (in years), and severity of
co-morbid illness.
