This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

Author's personal copy

White blood cell count and renin–angiotensin systeminhibitors for the risk of cancer in type 2 diabetes

Xilin Yang a,*, Ronald C.W. Ma a, Wing Yee So a, Gary T.C. Ko a, Alice P.S. Kong a,Hailu Zhao a, Gang Xu a, Peter C.Y. Tong a,b,c, Juliana C.N. Chan a,b,c

aDepartment of Medicine and Therapeutics, The Chinese University of Hong Kong, Shatin, Hong Kong, SAR, ChinabHong Kong Institute of Diabetes and Obesity, The Chinese University of Hong Kong, Shatin, Hong Kong, SAR, Chinac Li Ka Shing Institute of Health Sciences, The Chinese University of Hong Kong, Shatin, Hong Kong, SAR, China

1. Introduction

Type 2 diabetes mellitus (T2DM) predisposes patients to a

variety of cancers [1] such as colorectal [2], pancreatic [3] and

liver cancers [4], as well as breast [5] and endometrial cancers

[6] in women and prostate cancers in men [7], The overall

cancer incidence in T2DM is about one-third higher than that

in general population [8]. Our group has reported the moderate

accuracy of a cancer risk score consisting of lipids, white blood

cell (WBC) count, age and smoking status in predicting cancer

with an area under the receiver’s operating characteristic

curve of up to 0.71 [9].

Angiotensin-converting enzyme inhibitors (ACEI) are

widely used for cardio-renal protection in T2DM. Experimental

studies suggest that ACEI have anti-cancer properties [10,11]

although the risk association of cancer with ACEI remains

inconsistent in epidemiological surveys [12,13]. It is known

that ACEI have immunomodulatory potential, i.e., down-

regulating an overactive immune system [14,15]. To this end,

dysregulation of inflammation and immune system have been

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 8 7 ( 2 0 1 0 ) 1 1 7 – 1 2 5

a r t i c l e i n f o

Article history:

Received 12 May 2009

Received in revised form

14 October 2009

Accepted 22 October 2009

Published on line 22 November 2009

Keywords:

ACE inhibitor

White blood cell

Cancer

Type 2 diabetes

a b s t r a c t

Background: High white blood cell (WBC) predicted cancer-associated mortality and renin–

angiotensin system (RAS) inhibitors have immunomodulating effects. We hypothesize that

RAS inhibitors may reduce cancer risk associated with high WBC in type 2 diabetes mellitus

(T2DM).

Methods: A prospective cohort of 4570 Chinese T2DM patients, free of cancer at enrolment,

were analyzed. Biological interaction between WBC groups and use of RAS inhibitors was

estimated using relative excess risk due to interaction (RERI), attributable proportion due to

interaction (AP) and synergy index (S). RERI > 0, AP > 0 or S > 1 indicates biological inter-

action.

Results: During 4.89 years of follow-up, 205 (4.49%) patients developed cancer.

WBC � 8.2 � 109 counts/L plus non-use of RAS inhibitors was associated with elevated

cancer risks in multivariable models. The RERI and AP for interaction between

WBC � 8.2 � 109 counts/L and non-use of RAS inhibitors were, respectively, 1.26 (95% CI:

0.22–2.31) and 0.50 (0.23–0.78). In patients with WBC � 8.2 � 109 counts/L, use of RAS inhi-

bitors was associated with 64% (31–81%) cancer risk reduction in multivariable analysis.

Conclusions: In T2DM, increased WBC predicts cancer while use of RAS inhibitors may

reduce cancer risks associated with high WBC count.

# 2009 Elsevier Ireland Ltd. All rights reserved.

* Corresponding author at: Room 507, Li Ka Shing Institute of Health Sciences, c/o Department of Medicine and Therapeutics, The Prince ofWales Hospital, Shatin, Hong Kong, SAR, China. Tel.: +852 3763 6052; fax: +852 2637 3852.

E-mail address: yang.xilin@cuhk.edu.hk (X. Yang).

Contents lists available at ScienceDirect

Diabetes Researchand Clinical Practice

journal homepage: www.elsevier.com/locate/diabres

0168-8227/$ – see front matter # 2009 Elsevier Ireland Ltd. All rights reserved.doi:10.1016/j.diabres.2009.10.012

Author's personal copy

implicated in development and progression of cancer [16]. In

support of this notion, high circulating WBC count was

associated with increased risk of cancer mortality in large

population-based studies [17,18]. Since inflammation appears

to be an important culprit in cancer development and that the

ex vivo anti-cancer effects of ACE inhibitor may be in part

mediated by its anti-inflammatory effects, we used a

comprehensive prospective database to test two hypotheses:

1) High WBC count is associated with incident cancer in T2DM;

and 2) Use of renin–angiotensin system (RAS) inhibitors

including ACEI and angiotensin II receptor blockers (ARB)

may reduce the risk of cancer associated with high WBC.

2. Patients and methods

2.1. Patients

The Hong Kong Diabetes Registry was established at the

Prince of Wales Hospital, the teaching hospital of the Chinese

University of Hong Kong, in 1995. Sources of patients came

from general practitioners, community clinics, and other

specialty clinics and patients discharged from the Prince of

Wales Hospital or other hospitals. Enrolled patients with

hospital admissions within 6–8 weeks prior to assessment

accounted for less than 10% of all referrals. On the day of visit

to the Centre, patients underwent a 4-h assessment of

complications and risk factors on an outpatient basis, which

was modified from the European DiabCare protocol [19]. Once

a diabetic patient had undergone this comprehensive

assessment, the patient is considered to have been enrolled

and will be continuously followed till time of death. Ethical

approval was obtained from the Chinese University of Hong

Kong Clinical Research Ethics Committee and the Declaration

of Helsinki was adhered. Informed consent was obtained

from all patients for research purpose at the time of

assessment.

A trained team of personnel of the Hospital Authority (HA)

routinely coded all the hospital discharge diagnoses including

cancer and non-cancer-related hospital admissions according

to the International Classification of Diseases, Ninth Revision

(ICD-9). Admissions to all public hospitals in Hong Kong on or

before 30th July 2005 were retrieved from the Hong Kong HA

Central Computer System. In Hong Kong, public hospitals,

under the governance of HA, provide 95% of the hospital bed-

days [20]. Mortality data of the cohort during the period were

also retrieved from the Hong Kong Death Registry for cross-

checking with the mortality data from the HA Central

Computer System. Since 1 December 1996, the HA Computer

System recorded all drug dispensary data in public hospitals

including the start dates and end dates for each of the drugs of

interest. In Hong Kong, all medications are dispensed on site in

both inpatient and outpatient settings. We retrieved relevant

data on drug use in this cohort during the follow-up period

from the drug dispensary database. Using the Hong Kong

Identity Card number, a unique identification number

compulsory for all Hong Kong residents and used by

government and commercial organizations, we successfully

merged the complication assessment, drug use and clinical

outcome databases for analysis.

From 1995 to 2005, 7920 patients had been enrolled in the

registry. Among them, 6539 patients were enrolled after 1

December 1996 when drug use data were computerized by HA.

After excluding patients with type 1 diabetes (n = 261),

uncertain type of diabetes (n = 5), non-Chinese or uncertain

ethnicity (n = 40), known history of cancer (n = 160) and

missing values (n = 1503), 4570 patients were included in the

current analysis.

2.2. Endpoint definition

The cancer endpoint was defined as having an incident cancer

during follow-up period (ICD-9 Codes 140-208), including fatal

cancer and non-fatal cancer. Follow-up time was calculated as

the period in years from the date of enrolment to the date of

the first cancer event, death or 30 July 2008, whichever came

first.

2.3. Clinical measurements

Details of the 4-h assessment of complications and risk factors

have been reported [8,9,21]. Briefly, on the day of assessment,

patients attended the centre after 8 h of fasting and under-

went anthropometric measurements and laboratory investi-

gations. Body weight and height were measured with subject

wearing light clothing and without shoes. A sterile, random

spot urine sample was used to measure albumin:creatinine

ratio (ACR). The abbreviated Modification of Diet in Renal

Disease Study formula recalibrated for Chinese [22] was used

to estimate glomerular filtration rate expressed in mL min�1

1.73 m�2: eGFR = 186 � [SCR � 0.011]�1.154 � [age]�0.203 � [0.742

if female] � 1.233, where SCR is serum creatinine expressed as

mmol/L (original mg/dL converted to mmol/L) and 1.233 is the

adjusting coefficient for Chinese.

2.4. Laboratory measurements

Complete blood picture was assayed using Beckman Coulter

counter. Plasma glucose was measured by a hexokinase

method (Hitachi 911 automated analyzer, Boerhringer Man-

nheim, Mannheim, Germany). Lipids (total cholesterol, trigly-

ceride and high-density lipoprotein (HDL) cholesterol were

measured by enzymatic methods on a Hitachi 911 automated

analyzer (Boehringer Mannheim, Mannheim, Germany) using

reagent kits supplied by the manufacturer of the analyzer. The

inter-assay and intra-assay coefficients of variation have been

described previously [9]. Low-density lipoprotein (LDL) cho-

lesterol was calculated by the Friedewald’s equation [23]. The

precision performance of these assays was within the

manufacturer’s specifications.

2.5. Statistical analysis

The Statistical Analysis System (Release 9.10) was used in the

analysis (SAS Institute Inc., Cary, North Carolina, USA). Cox

proportional hazard regression was used to obtain hazard

ratio (HR) for cancer. As some of the risk factors including WBC

count are associated with cancer in curving manners [8,9], we

used restricted cubic spline [24,25] in Cox models to examine

full-range associations of WBC count with cancer. Briefly, the

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 8 7 ( 2 0 1 0 ) 1 1 7 – 1 2 5118

Author's personal copy

restricted cubic spline with 4 knots at the 5th, 35th, 65th, and

95th percentiles of the WBC count was used in Cox propor-

tional hazard regression to generate HR curves of WBC count

for cancer, with adjustment for confounders including lipids

and other covariates as previously described [8,9]. We plotted

HR curves of WBC count for cancer among RAS inhibitors users

and non-users and identified cut-off points, if any, by visual

checking.

We examined biological interaction between WBC levels

and use of RAS inhibitors [26,27] to determine if the latter was

associated with reduced risk of cancer in patients categorized

by WBC count. We used three measures to estimate biological

Table 1 – Clinical and biochemical characteristics of the study cohort.

Patients without cancer(n = 4365)

Patients with cancer(n = 205)

Median or %(n) IQR Median or %(n) IQR P value

Baseline variables

Age (year) 58 20 66 15 <.0001yMale gender 46.2%(2016) 48.3%(99) .5543zSmoking status <.0001zCurrent smoker 13.2%(578) 23.4%(48)

Ex-smoker 15.6%(681) 18.5%(38)

Alcohol drinking status .0003zCurrent drinker 7.5%326) 8.3%(17)

Ex-drinker 12.1%(529) 21.5%(44

Body mass index (kg/m2) 24.8 4.9 24.2 5.1 .0947yDuration of diabetes (year) 6 10 7 9 .1472ySystolic BP (mmHg) 135 27 138 27 .0144yDiastolic BP (mmHg) 75 14 75 15 .7380yGlycated hemoglobin (%) 7.3 2.1 7.4 2.2 .9308yLDL cholesterol (mmol/L) 3.10 1.21 3.10 1.24 .5152yHDL cholesterol (mmol/L) 1.26 0.44 1.25 0.51 .5148yTriglyceride (mmol/L) 1.35 1.02 1.23 0.76 .0036yTotal cholesterol (mmol/L) 5.10 1.32 5.0 1.52 .2008yACR (mg/mmol) 2.2 11.6 3.7 10.5 .1033yeGFR (mL min�1 1.73 m�2) 101.9 41.7 98.6 38.9 .0831yWhite blood cell count (109 counts/L) 7.0 2.4 6.9 2.8 .2813yPrior myocardial infarction 2.27%(99) 1.95%(4) .7652zPrior stroke 4.58%(200) 4.39%(9) .8979zDeath (all-cause) 6.76%(295) 50.73%(104) <.0001zDrug use at enrolment

Use of ACEI 23.6%(1029) 22.0%(45) .5923zUse of ARB 1.9%(82) 0.5%(1) .1837�

Use of ACEI or ARB

Use of statins 13.9%(608) 10.2%(21) .1345zUse of fibrates 3.3%(142) 2.4%(5) .5185zUse of lipid lowering drugs other than statins and fibrates 0.1%(6) 0.5%(1) .2749�

Use of oral antidiabetic drugs 68.0%(2969) 69.8%(143) .6019zUse of insulin 19.1%(833) 21.5%(44) .3978zUse of antihypertensive drugs other than ACEI or ARB 40.9%(1787) 46.8%(96) .0940zUse of antiplatelet drugs 11.5%(503) 11.7%(24) .9358zUse of aspirin 11.2%(488) 10.7%(22) .8421zUse of anticoagulation drugs 1.2%(51) 0.5%(1) .7298zUse of non-steroidal anti-inflammation drugs 2.3%(100) 0.5%(1) .0901�

Use of glucocorticoid drugs 1.6%(68) 0.5%(1) .3725zDrug use from enrolment to the earliest date of cancer, death or censoring

Use of ACEI 56.6%(2472) 49.8%(102) .0524zDuration of use of ACEI (year)§ 2.41 3.70 1.75 2.31

Use of ARB 9.9%(431) 4.4%(9) .0093zDuration of use of ARB (year)§ 1.81 2.89 2.05 0.56

Use of ACEI or ARB 58.6%(2558) 50.2%(103) .0177zUse of statins 38.6%(1686) 22.0%(45) <.0001zUse of fibrates 10.2%(445) 5.9%(12) .0429zUse of lipid lowering drugs other than statins and fibrates 0.5%(23) 1.0%(2) .3099�

Use of oral antidiabetic drugs 86.3%(3765) 83.9%(172) .3407zUse of insulin 38.0%(1660) 35.1%(72) .4016z

Abbreviations: IQR, interquartile range; ACR, spot urine albumin:creatinine ratio; eGFR, estimated glomerular filtration rate; LDL, low-density

lipoprotein; HDL, high-density lipoprotein; ACEI, angiotensin-converting enzyme inhibitors; ARB, angiotensin II receptor blockers.

y, Derived from Wilcoxon Two-Sample test; z, Derived from Chi-square test; �, Derived from Fisher’s exact test; §, Among users only.

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 8 7 ( 2 0 1 0 ) 1 1 7 – 1 2 5 119

Author's personal copy

interaction: 1) Relative excess risk due to interaction (RERI); 2)

Attributable proportion due to interaction (AP); and 3) Synergy

index (S). The RERI is the excess risk due to interaction relative

to the risk without exposure. AP refers to the attributable

proportion of disease which is due to interaction among

persons with both exposures. S is the excess risk from both

exposures when there is an biological interaction, relative to

the risk from both exposures without interaction [28]. PERI > 0,

AP > 0 or S > 1 indicates biological interaction. In Cox models,

the RERI is the best choice among the three measures [29]. As

testing of interaction requests that both variables are binary,

we used Yes/No coding for use of RAS inhibitors from

enrolment to the earliest date of events, death or censoring [8].

We used a structured analysis scheme to adjust for

confounding variables. Firstly, we adjusted for age, sex, body

mass index, smoking and alcohol status. Secondly, we

further adjusted for LDL cholesterol-related risk (LDL

cholesterol � 3.80 mmol/L and presence of LDL cholesterol

<2.80 mmol/L and albuminuria) [30], other metabolic vari-

ables and drug use. Thirdly, we used restricted cubic spline

in Cox models to adjust for the non-linear associations of

lipids and other variables with cancer risk in T2DM [8,9].

Additionally, we computed a propensity score [31] of use of

RAS inhibitors from enrolment to the earliest date of cancer,

death or censoring to estimate the probability of use of these

drugs in the cohort. We used a logistic regression procedure

with use of RAS inhibitors during the follow-up period as the

dependent variable, and traditional risk factors as well as

clinical outcomes as the independent variables (see notes of

Table 2 for details). Deciles of the propensity score were used

in Cox models as a stratum variable to adjust for likelihood

using RAS during the follow-up period.

As in our previous studies [8], proportional hazards were

checked using either Supremum test or plots of Log negative

Log (survival distribution function) versus log (follow-up time).

Correlations between pairs of baseline covariates were

checked using Pearson’s correlation test (correlation

coefficient < 0.60) [32].

3. Results

3.1. Characteristics of the study cohort

The median age of the cohort was 58 (25th to 75th percentiles:

48–68) years and the median duration of diabetes was 6 (2–11)

years. During 4.89 (2.39–6.96) years of follow-up, 205 (4.49%)

patients developed cancer. The incidence of cancer was 9.58

(95% CI: 8.27–10.88) per 1000 years. WBC count was similar in

cancer and non-cancer patients. Approximately half of the

patients who developed cancer died on or before 30th July

2005. From enrolment to the endpoint time, the usage of RAS

inhibitors was 58.6% in patients who did not develop cancer

and 50.2% in those who developed cancer (P = 0.0177) (Table 1).

3.2. Full-range risk associations of WBCcount with cancer

The HR curve of WBC count for cancer among RAS inhibitors

non-users exhibited a J-shape (Fig. 1). Based on these risk

curves, we chose 5.8 � 109 and 8.2 � 109 counts/L as two cut-

off points to stratify patients into different groups by usage of

RAS inhibitors. Using patients with WBC count � 5.8 � 109 but

<8.2 � 109 counts/L as the referent group, RAS inhibitor non-

users with WBC count � 8.2 � 109 counts/L had higher cancer

risk (hazard ratio [HR]: 1.73, 95%CI 1.07–2.80, P = .0262) but not

patients with WBC count <5.8 � 109 counts/L (HR: 1.26, 95%CI:

0.76–2.09, P = .3744). Among RAS inhibitor users, patients with

WBC count � 8.2 � 109 counts/L did not have (HR: 0.78, 95%CI:

0.47–1.29, P = .3350) increased cancer risk while patients with

WBC count <5.8 � 109 counts/L had increased risk of cancer

compared to the reference group (HR: 1.79, 95%CI: 1.13–2.86,

P = .0141) (see the legend of Fig. 1 for a list of the adjusted

variables). These observations indicate a need to test biological

interaction between WBC <5.8 � 109 counts/L and use of RAS

inhibitors and that between WBC � 8.2 � 109 counts/L and

non-use of RAS inhibitors for cancer risk.

3.3. WBC count, use of RAS inhibitors and risks of cancer

WBC � 8.2 � 109 counts/L plus non-use of RAS inhibitors was

associated with an elevated risk of cancer in multivariable

models, and after further adjusting for non-linear confound-

ing effects of covariates, as well as likelihood of using RAS

inhibitors. On the other hand, WBC � 8.2 � 109 counts/L plus

Fig. 1 – Hazard ratio curve of white blood cell count for all-

site cancer according to their status of use of ACEI or ARB

in type 2 diabetes.

Abbreviations: ACEI, angiotensin-converting enzyme

inhibitors; ARB, angiotensin II receptor blockers; WBC,

white blood cell; All the curves are adjusted for age, sex,

BMI, smoking status (current plus exit), alcohol drinking

status (current plus ex), LDL cholesterol-related risk,

duration of diabetes, HbA1c, systolic BP, HDL cholesterol,

triglyceride, eGFR and mediation from enrolment to

cancer, death or censoring date (oral antidiabetic drugs,

insulin, statins and fibrates) whichever came first, and use

of other antihypertensive drugs, and antiplatelet drugs at

enrolment. WBC 6.5 T 109 counts/L was used as the

reference. Spline functions of BMI, HDL cholesterol and

triglyceride were used in the adjustment for their non-

linear associations with cancer.

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 8 7 ( 2 0 1 0 ) 1 1 7 – 1 2 5120

Author's personal copy

use of RAS inhibitors was not associated with elevated risks of

cancer (Table 2). After excluding those who were followed for

less than two years, WBC � 8.2 � 109 counts/L plus non-use of

RAS inhibitors remained significant (HR: 2.13, 95%CI: 1.18–

3.83). There was also significant interaction between

WBC � 8.2 � 109 counts/L and non-use of RAS inhibitors for

the risk of cancer. The RERI and AP for this interaction were,

respectively, 1.26 (95% CI: 0.22–2.31) and 0.50 (0.23–0.78),

although S was not significant. In patients with

WBC � 8.2 � 109 counts/L, use of RAS inhibitors was asso-

ciated with 64% (31–81%) reduction in cancer risk in the cohort

after adjusting for covariates as compared to non-use of RAS

inhibitors (Table 2).

On the other hand, WBC <5.8 � 109 counts/L plus use of

RAS inhibitors was not associated with an increased cancer

risk in multivariable analysis and the interaction measures

between WBC <5.8 � 109 counts/L and use of RAS inhibitors

were also not significant (Table 3). In these analyses, use of

anticoagulation drugs, non-steroidal anti-inflammation

drugs and glucocorticoid drugs were rare (<2%), and their P

values for cancer were non-significant (P > 0.15 in all the

models). Hence, they were not included in the adjustment list

(Tables 1–3).

The Kaplan Meier plot shows that non-users of RAS

inhibitors with WBC �8.2 � 109 counts/L had the highest

cumulative incidence of cancer while users of RAS inhibitors

Table 2 – White blood cell count, use of RAS inhibitors and risks of all-site cancer in type 2 diabetes mellitus (white bloodcell count was stratified at 5.8 T 109 and 8.2 T 109 counts/L, where the hazard ratio curves for RAS inhibitor users andnon-users separated).

Exposures Number at risk Hazard ratio 95% CI P value

WBC <5.8 � 109 counts/L models

Model oneyWBC <5.8 � 109 counts/L and non-use of ACEI or ARB vs. others 563 1.06 0.67–1.68 .8117

WBC �5.8 � 109 counts/L and use of ACEI or ARB vs. others 2131 0.46 0.33–0.63 <.0001

WBC <5.8 � 109 counts/L and use of ACEI or ARB vs. others 530 0.92 0.60–1.42 .7168

Model twozWBC <5.8 � 109 counts/L and non-use of ACEI or ARB vs. others 563 1.06 0.66–1.68 .8220

WBC �5.8 � 109 counts/L and use of ACEI or ARB vs. others 2131 0.50 0.35–0.72 .0002

WBC <5.8 � 109 counts/L and use of ACEI or ARB vs. others 530 0.93 0.59–1.46 .7526

Model three�

WBC <5.8 � 109 counts/L and non-use of ACEI or ARB vs. others 563 1.05 0.66–1.68 .8253

WBC �5.8 � 109 counts/L and use of ACEI or ARB vs. others 2131 0.47 0.33–0.69 <.0001

WBC <5.8 � 109 counts/L and use of ACEI or ARB vs. others 530 0.81 0.51–1.29 .3809

WBC �8.2 � 109 counts/L models

Model oneyWBC �8.2 � 109 counts/L and use of ACEI or ARB vs. others 819 0.76 0.47–1.22 .2547

WBC <8.2 � 109 counts/L and non-use of ACEI or ARB vs. others 1495 1.41 1.01–1.96 .0431

WBC �8.2 � 109 counts/L and non-use of ACEI or ARB vs. others 414 2.69 1.76–4.11 <.0001

Model twozWBC �8.2 � 109 counts/L and use of ACEI or ARB vs. others 819 0.76 0.47–1.24 .2730

WBC <8.2 � 109 counts/L and non-use of ACEI or ARB vs. others 1495 1.35 0.94–1.94 .1028

WBC �8.2 � 109 counts/L and non-use of ACEI or ARB vs. others 414 2.50 1.59–3.91 <.0001

Model three�

WBC �8.2 � 109 counts/L and use of ACEI or ARB vs. others 819 0.78 0.48–1.27 .3137

WBC <8.2 � 109 counts/L and non-use of ACEI or ARB vs. others 1495 1.48 1.02–2.14 .0405

WBC �8.2 � 109 counts/L and non-use of ACEI or ARB vs. others 414 2.52 1.59–3.99 <.0001

Use of ACEI or ARB vs. non-use in patients with WBC count �8.2 � 109 counts/L�

Model oney 819 0.30 0.18–0.52 <.0001

Model twoz 819 0.31 0.17–0.58 .0002

Model three� 819 0.36 0.19–0.69 .0002

Abbreviations: CI, confidence interval; ACEI, angiotensin-converting enzyme inhibitors; ARB, angiotensin II receptor blockers; WBC, white blood

cell; LDL, low-density lipoprotein; HDL, high-density lipoprotein; BMI, body mass index.

y, Adjusted for age, sex, BMI, smoking status (current plus exit), and drinking status (current plus ex) [and WBC �8.2 � 109 counts/L (1 = Yes/

0 = No) for WBC<5.8 � 109 counts/L models or WBC <5.8 � 109 counts/L (1 = Yes/0 = No) for �8.2 � 109 counts/L models].

z, Further adjusted for LDL cholesterol-related risk, duration of diabetes, HbA1c, systolic BP, HDL cholesterol, triglyceride, eGFR and mediation

from enrolment to cancer, death or censoring date (oral antidiabetic drugs, insulin, statins and fibrates) whichever came first, and use of other

antihypertensive drugs, and antiplatelet drugs at enrolment.

�, Adjusted for covariates listed in z, but restricted spline functions of BMI, HDL cholesterol and triglyceride were used for better adjustment for

their non-linear associations, as well as using stratified Cox model on propensity score to adjust for the likelihood using ACEI or ARB during

follow-up period, which was calculated using a logistic procedure with covariates of sex, smoking status (current/ex), alcohol drinking status

(current/ex), age, BMI, glycated hemoglobin, LDL cholesterol, HDL cholesterol, triglyceride, systolic blood pressure, spot urinary albumin:

creatinine ratio, estimated glomerular filtration rate at enrolment, duration of diabetes, retinopathy, neuropathy, peripheral arterial disease,

myocardial infarction and stroke.

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 8 7 ( 2 0 1 0 ) 1 1 7 – 1 2 5 121

Author's personal copy

with WBC �8.2 � 109 had a cumulative incidence of cancer

similar to or lower than other two groups of patients stratified

by WBC count and use of RAS inhibitors (Fig. 2).

4. Discussion

We observed a J-shaped risk association between cancer and

WBC count in this relatively large prospective cohort of

Chinese patients with T2DM with detailed documentation of

risk factors. While high WBC count was associated with

increased cancer risk, this was markedly attenuated in

subjects using RAS inhibitors. We further observed significant

biological interaction between WBC �8.2 � 109 counts/L and

non-use of RAS inhibitors. These findings provide corrobora-

tive evidence in support of risk association of high WBC count

with cancer mortality [17,18]. Furthermore, our results

suggested that WBC count �8.2 � 109 counts/L was the

threshold for predicting cancer risk in T2DM.

The anti-cancer effect of RAS inhibitors in human remains

controversial. In a meta-analysis of randomized trials of

antihypertensive drugs, RAS inhibitors had neutral effect on

cancer risk [33]. However, several epidemiological studies

suggest that RAS inhibitors might prevent cancer. In the

Rotterdam Study (a population-based, prospective cohort

study with 7983 participants), RAS inhibitor was associated

with reduced risk of cancer in ACE DD genotype carriers. The

latter genotype is known to be associated with increased ACE

activity [34]. In a cohort analysis of the randomized VATTC

trial, RAS inhibitors was associated with reduced incidence of

keratinocyte cancers in subjects who had past history of basal

cell or squamous cell carcinoma [35]. In our previous analysis,

we found that the risk of cancer associated with low LDL

cholesterol [8] depended upon presence of albuminuria in

T2DM [30]. Given the anti-albuminuric effects of RAS

inhibitors [36], it is tempting to speculate that this may

Table 3 – Biological interactions of low and high WBCcount and use of ACEI or ARB on the risk of cancer in type2 diabetes .

Measures of biologicalinteraction

Estimate 95% confidenceinterval

Between WBC <5.8 � 109 counts/L and use of ACEI or ARB

Model oneyRERI 0.41 �0.13 to 0.95

AP 0.45 �0.10 to 1.00

S 0.16 0.00–20.25

Model twozRERI 0.38 �0.18 to 0.93

AP 0.41 �0.15 to 0.96

S 0.16 0.00–42.75

Model threey, �RERI 0.29 �0.25 to 0.82

AP 0.35 �0.29 to 0.99

S 0.40 0.06–2.60

Between WBC � 8.2 � 109 counts/L and non-use of ACEI or ARB

Model oneyRERI 1.53* 0.46–2.59

AP 0.57* 0.33–0.81

S 10.30 0.22–484.73

Model twoy, zRERI 1.38* 0.37–2.40

AP 0.55* 0.30–0.81

S 13.04 0.05–3751.66

Model threey, �RERI 1.26* 0.22–2.31

AP 0.50* 0.23–0.78

S 5.97 0.42–84.32

Abbreviations: CI, confidence interval; ACEI, angiotensin-converting

enzyme inhibitors; ARB, angiotensin II receptor blockers; WBC,

white blood cell; LDL, low-density lipoprotein cholesterol; HDL,

high-density lipoprotein; BMI, body mass index; RERI, Relative

excess risk due to interaction; AP, Attributable proportion due to

interaction; S, Synergy index.

*, Statistically significant with RERI > 0, AP > 0 or S > 1 indicating

biological interaction.

y, Adjusted for age, sex, BMI, smoking status (current plus exit),

and drinking status (current plus ex) [and WBC �8.2 � 109 counts/L

(1 = Yes/0 = No) for WBC 5.8 � 109 counts/L models or WBC

<5.8 � 109 counts/L (1 = Yes/0 = No) for �8.2 � 109 counts/L mod-

els]; z, Further adjusted for LDL cholesterol-related risk, duration of

diabetes, HbA1c, systolic BP, HDL cholesterol, triglyceride, eGFR

and mediation from enrolment to cancer, death or censoring date

(oral antidiabetic drugs, insulin, statins and fibrates) whichever

came first, and use of other antihypertensive drugs, and antiplate-

let drugs at enrolment; �, Adjusted for covariates listed in z, but

restricted spline functions of BMI, HDL cholesterol and triglyceride

were used for better adjustment for their non-linear associations,

as well as using stratified Cox model on propensity score to adjust

for the likelihood using ACEI or ARB during follow-up period,

which was calculated using a logistic procedure with covariates of

sex, smoking status (current/ex), alcohol drinking status (current/

ex), age, BMI, glycated hemoglobin, LDL cholesterol, HDL choles-

terol, triglyceride, systolic blood pressure, spot urinary albumin:

creatinine ratio, estimated glomerular filtration rate at enrolment,

duration of diabetes, retinopathy, neuropathy, peripheral arterial

disease, myocardial infarction and stroke.

Fig. 2 – Cumulative incidence of cancer stratified by white

blood cell count and use of ACEI or ARB status in type 2

diabetes.

Abbreviations: ACEI, angiotensin-converting enzyme

inhibitors; ARB, angiotensin II receptor blockers; WBC,

white blood cell (T109 counts/L); P value = .0002 for Log-

Rank test; the figure shows that the group with

WBC I 8.2 T 109 counts/L and non-use of ACEI or ARB had

the highest cumulative incidence of cancer while the

group with WBC I 8.2 T 109 counts/L and use of ACEI or

ARB had a cumulative incidence of cancer similar to or

lower than other groups.

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 8 7 ( 2 0 1 0 ) 1 1 7 – 1 2 5122

Author's personal copy

contribute to reduced risk of cancer among RAS inhibitor users

in the present analysis.

In a previous analysis, we reported that WBC

<5.8 � 109 counts/L was predictive of cancer in T2DM [9]. In

this analysis, we observed that WBC <5.8 � 109 counts/L was

associated with cancer among RAS inhibitor users but not

among RAS inhibitor non-users. However, we did not detect

significant biological interaction between WBC

<5.8 � 109 counts/L and use of RAS inhibitors. Further

explorations in larger cohorts of patients with T2DM are

needed to address this issue.

T2DM is characterized by insulin resistance and hyper-

glycaemia. Both insulin resistance and hyperglycaemia are

associated with upregulation of RAS components, including

the angiotensin II type 1 receptor and angiotensin II [37].

Activation of the RAS induces the inflammatory cascade [38],

which has been implicated in carcinogenesis [16]. In support

of the immunomodulatory effects of ACEI which can down-

regulating an overactive immune system [14,15], this class of

drugs has been shown to prevent recurrent cancer in high

risk patients [35]. In this study, high WBC levels predicted

cancer and this risk association was attenuated among users

of RAS inhibitors. Of note, the HR of WBC � 8.2 � 109 counts/

L plus non-use of ACEI or ARB decreased from 2.46 to 2.13

after excluding those who were followed for less than two

years, suggesting that subclinical or undiagnosed cancer

cannot be excluded. Taken together, we hypothesize that

abnormal metabolic milieu characterized by high blood

glucose, lipid and insulin levels may activate hydroxy-

methylglutaryl-CoA reductase, angiotensin II and insulin-

like growth factor-1 pathways [39], which interact to give rise

to many subphenotypes including albuminuria, abnormal

LDL cholesterol and WBC counts which are associated with

cancer (Fig. 3).

There are several limitations in our study. Firstly, this

study was not a randomized clinical trial and thus we cannot

establish the nature of causality of these risk associations.

Secondly, we adjusted for all confounders including the use

of spline functions for non-linearity and propensity score for

possible use of RAS inhibitors. However, effects due to other

unmeasured or unknown factors cannot be excluded.

Thirdly, 24.8% of patients were removed from the analysis

due to missing values. However, the crude rates of incident

cancer (4.5% versus 4.5%, P = 0.9505) and use of RAS

inhibitors (58.2% versus 59.5%, P = 0.3922) were similar

between the excluded and analyzed group making selection

bias less likely. Fourthly, a small number of cancer events

might have been missed. Nevertheless, given the heavily

subsidized nature of our health care system; our databases

have covered 95% of major clinical events based on hospital

discharge records. Fifthly, only a single measurement of

WBC count was used for analysis and we recognized that a

large variation of WBC count may occur over time. Sixthly,

selection of cut-off points was arbitrary based on visual

checking.

In conclusion, using the HR curve analysis and adjusting for

non-linear associations between cancer and the likelihood of

using RAS inhibitors where appropriate, we detected a

significant interaction between non-use of RAS inhibitors

and WBC � 8.2 � 109 counts/L for cancer. The study suggests

that RAS inhibitors may prevent cancer in high risk conditions

such as WBC � 8.2 � 109 counts/L.

Acknowledgements

We thank all medical and nursing staff of the Prince of

Wales Hospital Diabetes Centre in recruiting and managing

these patients. This study was supported by the Hong

Kong Foundation for Research and Development in Diabetes,

Lioa Wun Yuk Diabetes Research Fund and a direct grant of

the Chinese University of Hong Kong (Reference No.

2008.1.043).

Conflict of interest

The authors declare that they have no conflict of interest.

r e f e r e n c e s

[1] S.H. Jee, H. Ohrr, J.W. Sull, J.E. Yun, M. Ji, J.M. Samet, Fastingserum glucose level and cancer risk in Korean men andwomen, JAMA 293 (January (2)) (2005) 194–202.

Fig. 3 – Hypothesized pathways of cancer in Type 2

diabetes mellitus in relation to blood glucose (BG) and

insulin resistance (IR).

There are cross-talks among the lipid, glucose and insulin

pathways in type 2 diabetes which in turn are related to

the hydroxymethylglutaryl-CoA reductase (HMGCR),

insulin-like growth factor-1 (IGF-1) and renin–angiotensin

(AII) systems (RAS). Activation of the RAS, HMGCR and IGF-

I pathways can have direct and indirect cancer-promoting

effects. The multiple effects of these pathways on renal

(e.g. urinary albumin:creatinine ratio, ACR), inflammatory

(e.g. white blood cells, WBC) and metabolic parameters

(e.g. LDL cholesterol) may partly explain their risk

associations of cancer in type 2 diabetes in cohort studies.

Experimental and intervention studies are needed to

elucidate the underlying mechanisms of these

observations.

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 8 7 ( 2 0 1 0 ) 1 1 7 – 1 2 5 123

Author's personal copy

[2] A. Seow, J.M. Yuan, W.P. Koh, H.P. Lee, M.C. Yu, Diabetesmellitus and risk of colorectal cancer in the SingaporeChinese Health Study, J. Natl. Cancer Inst. 98 (January (2))(2006) 135–138.

[3] R. Huxley, A. Ansary-Moghaddam, A. Berrington deGonzalez, F. Barzi, M. Woodward, Type-II diabetes andpancreatic cancer: a meta-analysis of 36 studies, Br. J.Cancer 92 (June (11)) (2005) 2076–2083.

[4] M.C. Rousseau, M.E. Parent, M.N. Pollak, J. Siemiatycki,Diabetes mellitus and cancer risk in a population-basedcase-control study among men from Montreal, Canada, Int.J. Cancer 118 (April (8)) (2006) 2105–2109.

[5] L.L. Lipscombe, P.J. Goodwin, B. Zinman, J.R. McLaughlin,J.E. Hux, Diabetes mellitus and breast cancer: aretrospective population-based cohort study, Breast CancerRes. Treat. 98 (March (3)) (2006) 349–356.

[6] E. Friberg, N. Orsini, C.S. Mantzoros, A. Wolk, Diabetesmellitus and risk of endometrial cancer: a meta-analysis,Diabetologia 50 (July (7)) (2007) 1365–1374.

[7] C. Rodriguez, A.V. Patel, A.M. Mondul, E.J. Jacobs, M.J. Thun,E.E. Calle, Diabetes and risk of prostate cancer in aprospective cohort of US men, Am. J. Epidemiol. 161(January (2)) (2005) 147–152.

[8] X. Yang, W.Y. So, G.T. Ko, R.C. Ma, A.P. Kong, C.C. Chow,et al., Independent associations between low-densitylipoprotein cholesterol and cancer among patients withtype 2 diabetes mellitus, CMAJ 179 (5) (2008) 427–437.

[9] X. Yang, W.Y. So, R.C. Ma, G.T. Ko, A.P. Kong, Q. Wang,et al., Predicting values of lipids and white blood cell countfor all-site cancer in type 2 diabetes, Endocr. Relat. Cancer15 (June (2)) (2008) 597–607.

[10] H. Yoshiji, S. Kuriyama, R. Noguchi, J. Yoshii, Y. Ikenaka, K.Yanase, et al., Amelioration of carcinogenesis and tumorgrowth in the rat liver by combination of vitamin K2 andangiotensin-converting enzyme inhibitor via anti-angiogenic activities, Oncol. Rep. 15 (January (1)) (2006) 155–159.

[11] R. Yasumatsu, T. Nakashima, M. Masuda, A. Ito, Y.Kuratomi, T. Nakagawa, et al., Effects of the angiotensin-Iconverting enzyme inhibitor perindopril on tumor growthand angiogenesis in head and neck squamous cellcarcinoma cells, J. Cancer Res. Clin. Oncol. 130 (October(10)) (2004) 567–573.

[12] S. Friis, H.T. Sorensen, L. Mellemkjaer, J.K. McLaughlin, G.L.Nielsen, W.J. Blot, et al., Angiotensin-converting enzymeinhibitors and the risk of cancer: a population-based cohortstudy in Denmark, Cancer 92 (November (9)) (2001) 2462–2470.

[13] T. Sjoberg, L.A. Garcia Rodriguez, M. Lindblad, Angiotensin-converting enzyme inhibitors and risk of esophageal andgastric cancer: a nested case-control study, Clin.Gastroenterol. Hepatol. 5 (October (10)) (2007), 1160-6 e1.

[14] L.M. Godsel, J.S. Leon, D.M. Engman, Angiotensinconverting enzyme inhibitors and angiotensin II receptorantagonists in experimental myocarditis, Curr. Pharm. Des.9 (9) (2003) 723–735.

[15] J.R. Gage, G. Fonarow, M. Hamilton, M. Widawski, O.Martinez-Maza, D.L. Vredevoe, Beta blocker andangiotensin-converting enzyme inhibitor therapy isassociated with decreased Th1/Th2 cytokine ratios andinflammatory cytokine production in patients with chronicheart failure, Neuroimmunomodulation 11 (3) (2004) 173–180.

[16] O.J. Finn, Cancer immunology, N. Engl. J. Med. 358 (June(25)) (2008) 2704–2715.

[17] A. Shankar, J.J. Wang, E. Rochtchina, M.C. Yu, R. Kefford, P.Mitchell, Association between circulating white blood cellcount and cancer mortality: a population-based cohortstudy, Arch. Intern. Med. 166 (January (2)) (2006) 188–194.

[18] T.P. Erlinger, P. Muntner, K.J. Helzlsouer, WBC count andthe risk of cancer mortality in a national sample of U.S.adults: results from the Second National Health andNutrition Examination Survey mortality study, CancerEpidemiol. Biomarkers Prev. 13 (June (6)) (2004) 1052–1056.

[19] K. Piwernetz, P.D. Home, O. Snorgaard, M. Antsiferov, K.Staehr-Johansen, M. Krans, For the DiabCare MonitoringGroup of the St. Vincent Declaration Steering Committee.Monitoring the targets of the St. Vincent declaration andthe implementation of quality management in diabetescare: the DiabCare initiative, Diabetic Med. 10 (1993)371–377.

[20] Information Services Department of Hong Kong, HongKong: the Facts. Public Health, Available at: http://wwwinfogovhk/hkfacts/pubhealth-epdf Accessed on 19May 2006. 2005 October.

[21] X.L. Yang, W.Y. So, A.P. Kong, P. Clarke, C.S. Ho, C.W. Lam,et al., End-stage renal disease risk equations for Hong KongChinese patients with type 2 diabetes: Hong Kong DiabetesRegistry, Diabetologia 49 (October (10)) (2006) 2299–2308.

[22] Y.C. Ma, L. Zuo, J.H. Chen, Q. Luo, X.Q. Yu, Y. Li, et al.,Modified glomerular filtration rate estimating equation forChinese patients with chronic kidney disease, J. Am. Soc.Nephrol. 17 (October (10)) (2006) 2937–2944.

[23] W.T. Friedewald, R.I. Levy, D.S. Fredrickson, Estimation ofthe concentration of low-density lipoprotein cholesterol inplasma, without use of the preparative ultracentrifuge,Clin. Chem. 18 (June (6)) (1972) 499–502.

[24] F. Harrell, Regression Modelling Strategies withapplications to Linear Models, Logistic Regression, andSurvival Analysis, Spinger-Varlag New York, Inc., NewYork, 2001.

[25] S. Durrleman, R. Simon, Flexible regression models withcubic splines, Stat. Med. 8 (May (5)) (1989) 551–561.

[26] A. Ahlbom, L. Alfredsson, Interaction: a word with twomeanings creates confusion, Eur. J. Epidemiol. 20 (7) (2005)563–564.

[27] T. Andersson, L. Alfredsson, H. Kallberg, S. Zdravkovic, A.Ahlbom, Calculating measures of biological interaction,Eur. J. Epidemiol. 20 (7) (2005) 575–579.

[28] L. Kalilani, J. Atashili, Measuring additive interaction usingodds ratios, Epidemiol. Perspect. Innov. 3 (2006) 5.

[29] R. Li, L. Chambless, Test for additive interaction inproportional hazards models, Ann. Epidemiol. 17 (March(3)) (2007) 227–236.

[30] X. Yang, W.Y. So, R.C. Ma, G.T. Ko, A.P. Kong, H. Zhao, et al.,Low LDL cholesterol, albuminuria, and statins for the riskof cancer in type 2 diabetes: the Hong Kong diabetesregistry, Diabetes Care 32 (October (10)) (2009) 1826–1832.

[31] M.M. Joffe, P.R. Rosenbaum, Invited commentary:propensity scores, Am. J. Epidemiol. 150 (August (4)) (1999)327–333.

[32] A. Rassi Jr., A. Rassi, W.C. Little, S.S. Xavier, S.G. Rassi, A.G.Rassi, et al., Development and validation of a risk score forpredicting death in Chagas’ heart disease, N. Engl. J. Med.355 (August (8)) (2006) 799–808.

[33] C.I. Coleman, W.L. Baker, J. Kluger, C.M. White,Antihypertensive medication and their impact on cancerincidence: a mixed treatment comparison meta-analysis ofrandomized controlled trials, J. Hypertens. 26 (April (4))(2008) 622–629.

[34] R. van der Knaap, C. Siemes, J.W. Coebergh, C.M. van Duijn,A. Hofman, B.H. Stricker, Renin–angiotensin systeminhibitors, angiotensin I-converting enzyme geneinsertion/deletion polymorphism, and cancer: theRotterdam Study, Cancer 112 (February (4)) (2008) 748–757.

[35] J.B. Christian, K.L. Lapane, A.L. Hume, C.B. Eaton, M.A.Weinstock, Association of ACE inhibitors and angiotensinreceptor blockers with keratinocyte cancer prevention in

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 8 7 ( 2 0 1 0 ) 1 1 7 – 1 2 5124

Author's personal copy

the randomized VATTC trial, J. Natl. Cancer Inst. 100(September (17)) (2008) 1223–1232.

[36] The ADVANCE Collaborative Group, Intensive Blood glucosecontrol and vascular outcomes in patients with type 2diabetes, N. Engl. J. Med. 358 (June (24)) (2008) 2560–2572.

[37] J.A. Olivares-Reyes, A. Arellano-Plancarte, J.R. Castillo-Hernandez, Angiotensin II and the development of insulinresistance: implications for diabetes, Mol. Cell Endocrinol.302 (April (2)) (2009) 128–139.

[38] A. Liguori, P. Abete, J.M. Hayden, F. Cacciatore, F. Rengo, G.Ambrosio, et al., Effect of glycaemic control and age on low-density lipoprotein susceptibility to oxidation in diabetesmellitus type 1, Eur. Heart J. 22 (November (22)) (2001)2075–2084.

[39] X. Yang, H. Zhao, Y. Sui, R.C. Ma, W.Y. So, G.T. Ko, et al.,Additive interaction between renin–angiotensin systemand lipid metabolism for cancer in type 2 diabetes, Diabetes58 (April (7)) (2009) 1518–1525.

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 8 7 ( 2 0 1 0 ) 1 1 7 – 1 2 5 125