Imperial College London · Web viewDagfinn Aune 1,2,3, Sabrina Schlesinger 4, Teresa Norat 1, Elio Riboli 1 Affiliations 1 Department of Epidemiology and Biostatistics, School of

1

Diabetes mellitus and the risk of abdominal aortic aneurysm:

a systematic review and meta-analysis of prospective studies.

Dagfinn Aune 1,2,3, Sabrina Schlesinger 4, Teresa Norat 1, Elio Riboli 1

Affiliations

1 Department of Epidemiology and Biostatistics, School of Public Health, Imperial College

London, London, United Kingdom

2 Bjørknes University College, Oslo, Norway

3 Department of Endocrinology, Morbid Obesity and Preventive Medicine,

Oslo University Hospital, Oslo, Norway

4 Institute for Biometry and Epidemiology, German Diabetes Center, Leibniz Institute for

Diabetes Research at the Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany

Short title: Diabetes and abdominal aortic aneurysm

Word count abstract: 216

Word count text: 1627

Correspondence to: Dagfinn Aune, Department of Epidemiology and Biostatistics,

School of Public Health, Imperial College London, St. Mary's Campus, Norfolk Place,

Paddington, London W2 1PG, UK.

2

Telephone: +44 (0) 20 7594 8478

Fax: +44(0) 20 7594 0768

E-mail: [email protected]

3

Abstract

Background: Diabetes mellitus has been associated with reduced risk of abdominal aortic

aneurysm in a number of epidemiological studies, however, until recently little data from

prospective studies have been available. We therefore conducted a systematic review and

meta-analysis of prospective studies to quantify the association.

Material and methods: Two investigators searched the PubMed and Embase databases for

studies of diabetes and abdominal aortic aneurysm up to May 8th 2018. Prospective studies

were included if they reported adjusted relative risk (RR) estimates and 95% confidence

intervals (95% CIs) of abdominal aortic aneurysm associated with a diabetes diagnosis.

Summary relative risks were estimated by use of a random effects model.

Results: We identified 16 prospective studies with 16 572 cases among 4 563 415

participants that could be included in the meta-analysis. The summary RR for individuals

with diabetes compared to individuals without diabetes was 0.58 (95% CI: 0.51-0.66,

I2=40.4%, pheterogeneity=0.06). The results persisted when stratified by sex, duration of follow-

up, and in most of the subgroup analyses. There was no evidence of publication bias with

Egger's test, p=0.64 or by inspection of the funnel plots.

Conclusions: These results suggest that individuals with diabetes mellitus are at a reduced

risk of abdominal aortic aneurysm, however, whether pharmacological agents for diabetes

mellitus explain this observation needs to be clarified in future studies.

4

Keywords: Diabetes mellitus, abdominal aortic aneurysm, systematic review, meta-analysis.

5

Introduction

Aortic aneurysms are localized dilatations of the aorta which when ruptured have very high

mortality rates with 80% of those reaching hospital and 50% of those undergoing surgery for

ruptured aortic aneurysms dying as a consequence (1;2). Globally aortic aneurysms

accounted for 168200 deaths and 2.9 million disability-adjusted life years in 2015 (3;4). Most

aortic aneurysms are abdominal aortic aneurysms. The incidence of new abdominal aortic

aneurysms is 0.4-0.67% in Western populations (5-7), and ten-fold lower in Asian

populations (8). Some of the established risk factors for abdominal aortic aneurysm include

age, male sex, hypertension, coronary heart disease, peripheral artery disease, chronic

obstructive pulmonary disease and smoking (9;10).

Although individuals with diabetes mellitus are at increased risk of several

cardiovascular diseases (11;12), evidence from epidemiological studies have paradoxically

suggested that a history of diabetes mellitus is associated with a reduced risk of abdominal

aortic aneurysms (10;13-24), although not all of these studies found a statistically significant

association (14;15;19;20;23). Although the underlying mechanisms are not firmly established

it has been suggested that individuals with diabetes mellitus have excess vascular matrix

which might protect against the loss of the arterial wall matrix that is typically seen in

abdominal aortic aneurysms. A previous meta-analysis suggested an inverse association

between diabetes mellitus and abdominal aortic aneurysms, however, most of the studies

included had a cross-sectional design and there was a limited number of cohort studies

included (25). Four cohort studies (three publications) (16;17;19) may have been missed in

the previous meta-analysis and at least six additional cohort studies (five publications) (20-

24) have since been published on the topic. For this reason we conducted an up-to-date

6

systematic review and meta-analysis of diabetes mellitus and abdominal aortic aneurysm to

provide updated data on the association using data from a stronger study design (prospective

studies) and to investigate sources of heterogeneity by conducting subgroup and sensitivity

analyses.

7

Material and methods

Search strategy

Two investigators (DA, SS) searched the Pubmed, and Embase databases up to July 20th 2017

for eligible studies and the search was later updated to May 8th 2018. The search terms used

are found in Supplementary Text. We followed standard MOOSE criteria for reporting meta-

analyses (26). In addition, we searched the reference lists of the identified publications for

further studies.

Study selection and inclusion criteria

We included published retrospective and prospective cohort studies and nested case-control

studies within cohorts that investigated the association between diabetes mellitus and the risk

of abdominal aortic aneurysm. Adjusted estimates of the relative risk (RR) had to be

available with the 95% confidence intervals (CIs) in the publication. A list of the excluded

studies can be found in Supplementary Table 1.

Data extraction

The following data were extracted from each study: The first author’s last name, publication

year, country where the study was conducted, study period, sample size, number of cases and

participants, exposure and subgroup, RRs and 95% CIs and variables adjusted for in the

analysis. Data was extracted by one author (DA) and checked for accuracy by a second

author (SS).

8

Statistical methods

We calculated summary RRs and 95% CIs of abdominal aortic aneurysm for individuals with

diabetes mellitus compared with individuals without diabetes mellitus using the random-

effects model by DerSimonian and Laird (27) which takes into account both within and

between study variation (heterogeneity). The average of the natural logarithm of the RRs was

estimated and the RR from each study was weighted using random effects weights.

Heterogeneity between studies was evaluated using Q and I2 statistics (28). I2 is a

measure of how much of the heterogeneity is due to between study variation rather than

chance. We conducted main analyses (all studies combined) and stratified by study

characteristics including sex, outcome type (incidence, mortality), duration of follow-up,

geographic location, number of cases, study quality and by adjustment for confounding

factors to investigate potential sources of heterogeneity. Study quality was assessed using the

Newcastle Ottawa scale which rates studies according to selection, comparability and

outcome assessment with a score range from 0 to 9 (29). Influence analyses were conducted

excluding one study at a time from the analysis and assessing the influence of each study on

the summary estimate. Results from these influence analyses are reported excluding the two

studies with the most positive and negative impact on the summary estimates.

Publication bias was assessed using Egger’s test (30) and Begg-Mazumdar’s test (31)

and by inspection of funnel plots. The statistical analyses were conducted using the software

package Stata, version 13.1 software (StataCorp, Texas, US).

9

Results

We identified 16 cohort studies (13 publications, 14 risk estimates) (10;13-24) that could be

included in the analysis of diabetes mellitus and risk of abdominal aortic aneurysm including

16 572 cases among 4 563 415 participants (Figure 1, Table 1). Eight of the studies were

from the USA, seven were from Europe and one was from Asia. Three publications

(19;23;24) contained data from two studies each and two of these publications only provided

a combined risk estimate for two studies each (23;24). The summary relative risk for patients

with diabetes mellitus versus patients without diabetes was 0.58 (95% CI: 0.51-0.66, I2=40%,

pheterogeneity=0.06) (Figure 2). There was no evidence of publication bias with Egger’s test,

p=0.64, or with Begg’s test, p=0.19 (Supplementary Figure 1). In sensitivity analyses

excluding one study at a time from the analysis, the summary RR ranged from 0.57 (95% CI:

0.50-0.64) when excluding the study by Wang et al (16) to 0.60 (95% CI: 0.53-0.69) when

excluding the study by Shah et al (21) (Supplementary Table 2).

Subgroup and sensitivity analyses

Inverse associations were observed in all subgroup analyses defined by sex, outcome type,

duration of follow-up, geographic location, number of cases, study quality and adjustment for

confounding factors (including age, education, alcohol, smoking, BMI, and physical activity,

hypertension, hypercholesterolemia, cholesterol, coronary heart disease, stroke, and chronic

obstructive pulmonary disease) (Table 2). With meta-regression analyses there was no

evidence that the results differed between these subgroups. The heterogeneity was in general

low to moderate, and there was little or no heterogeneity among the studies when stratified by

sex, among studies with longer duration of follow-up and among the studies with adjustment

10

for education, physical activity, and cholesterol (Table 2). The mean (median) study quality

score was 6.9 (7.0).

Discussion

This meta-analysis of 16 cohort studies including >16 000 cases of abdominal aortic

aneurysms among 4.5 million participants suggest that individuals with a diagnosis of

diabetes mellitus have a reduction in risk of abdominal aortic aneurysms by 42%. Inverse

associations were observed in both men and women and among studies in Europe, America

and Asian, however, no data were available from other geographic locations. The inverse

association was consistent in several other subgroup and sensitivity analyses. The findings are

consistent with a previous meta-analysis of 5 cohort studies (5099 cases, 595839

participants), but had more than 3 times the number of cases and more than 7 times the

number of participants (25).

The underlying mechanism(s) that may explain the inverse association between

diabetes mellitus and abdominal aortic aneurysms are not firmly established. It has been

observed that there is loss of vascular matrix and destruction of the aortic media in aortic

aneurysms (32) which is driven by matrix metalloproteinases (33). In contrast, individuals

with diabetes mellitus have been found to have excess vascular matrix (increased synthesis

and reduced degradation of matrix) via advanced glycation end products (34-36) and thicker

aortic walls which may reduce wall stress and thereby protect against the development of

abdominal aortic aneurysms (37). Experimental studies suggested that hyperglycemia may

lead to less aortic mural macrophage infiltration, elastolysis, and neovascularization and

limited the enlargement of abdominal aortic aneurysms compared to euglycemic mice, while

insulin treatment reduced serum glucose and reversed these effects (38;39). Although

11

epidemiological data are limited on the use of diabetes medications and risk of abdominal

aortic aneurysms, some animal studies have suggested that thiazolidinediones may protect

against abdominal aortic aneurysms (40). Rosiglitazone may decrease the expression of

matrix metalloproteinase in the aortic wall and may also reduce the development of aortic

aneurysms (41). It has been shown that metformin may improve aortic wall elasticity and

decrease matrix metalloproteinase and smooth muscle cell proliferation in the aortic wall

(42;43). An analysis of three cohort studies with 1697 abdominal aortic aneurysm patients

found a reduced risk of median or greater growth of abdominal aortic aneurysms among

patients who were prescribed metformin for the treatment of diabetes with odds ratios of 0.59

(95% CI: 0.39-0.87), 0.38 (95% CI: 0.18-0.80) and 0.13 (95% CI: 0.03-0.61) across the three

cohort studies, respectively (43), and another study found similar results (36). Consistent with

this are the results of a nested case-control study from Taiwan which found that use of

metformin, sulfonylurea and thiazolidinedione were associated with 18-28% reductions in

risk of aortic aneurysms (44), however, a Danish study reported only a non-significant

reduction in risk of ruptured abdominal aortic aneurysms with used of metformin (odds

ratio=0.84, 95% CI: 0.61-1.17) (45).

The current systematic review and meta-analysis may have been affected by

limitations of the individual studies that were included in the analysis. We cannot exclude the

possibility that confounding by other risk factors could have affected the findings. We found

that the results persisted in numerous subgroup analyses when stratified by adjustment for

age, education, alcohol intake, smoking status, BMI or obesity, physical activity,

hypertension, hypercholesterolemia, serum cholesterol, coronary heart disease, stroke or

chronic obstructive pulmonary disease and there was no evidence of heterogeneity between

these subgroup analyses with meta-regression analyses. Although residual confounding by

12

for example smoking is possible, smoking is associated with increased risk of both type 2

diabetes and abdominal aortic aneurysms so any confounding by smoking would likely

attenuate rather than exaggerate the observed associations. Similarly, residual confounding by

the other risk factors evaluated in subgroup analyses would also most likely attenuate rather

than exaggerate the inverse association because of the paradoxical association between

diabetes mellitus and abdominal aortic aneurysms. Therefore it seems less likely that

confounding by the above-mentioned risk factors could explain the observed association,

however, the possibility that other risk factors could confound the association cannot be

entirely excluded.

Although these data suggest a reduction in risk of abdominal aortic dissection among

individuals with diabetes mellitus compared to individuals without diabetes mellitus, this

benefit is greatly outweighed by adverse effects of diabetes mellitus upon the risk of many

other chronic diseases including coronary heart disease, stroke, heart failure, atrial

fibrillation, several cancers, digestive diseases, kidney stones, infections, respiratory diseases,

neurological disorders and all-cause mortality (11;12;46-48). Because of these wide-ranging

complications of diabetes mellitus efforts to reduce the incidence of diabetes mellitus should

continue worldwide. Therefore, the public health implications of the current results are

limited other than trying to clarify the underlying mechanisms behind the reduced risk of

abdominal aortic aneurysms, which potentially could result in preventive therapeutic

interventions in the future.

Conclusion

In conclusion, this meta-analysis suggest that individuals with diabetes mellitus have a 42%

reduction in the relative risk of developing abdominal aortic aneurysm. Whether or not

13

pharmacological agents for diabetes mellitus explain this observation needs to be clarified in

future studies. As most of the current studies were from the US and Europe, additional studies

are needed from other geographic locations and any further studies should adjust for more

confounding factors.

Acknowledgement: DA conducted the literature search and analyses and wrote the first draft

of the paper. SS and DA contributed to the literature screening. All authors interpreted the

data, revised the subsequent drafts for important intellectual content, read and approved the

final manuscript. This work has been supported by funding from the South East

Regional Health Authority of Norway, the School of Public Health of

Imperial College London, and the Imperial College National Institute of

Health Research (NIHR) Biomedical Research Centre (BRC).

14

15

16

Figure 1. Flow-chart of study selection

17

Figure 2. Diabetes mellitus and abdominal aortic aneurysm

Relative Risk .1 .25 .5 .75 1 1.5 2

Study Relative Risk (95% CI)

Larsson, 2018 0.57 ( 0.40, 0.82) Wang, 2017 0.79 ( 0.57, 1.11) Tang, 2016 0.52 ( 0.36, 0.75) Jahangir, 2015 0.75 ( 0.53, 1.05) Shah, 2015 0.46 ( 0.35, 0.59) Tsai, 2015 0.60 ( 0.51, 0.71) Sode, 2013, CCHS 0.50 ( 0.20, 1.10) Sode, 2013, CGPS 1.10 ( 0.70, 2.00) Campbell, 2012 0.60 ( 0.50, 0.72) Ohrlander, 2012 0.39 ( 0.26, 0.58) Lederle, 2008 0.29 ( 0.13, 0.68) Iribarren, 2007 0.62 ( 0.36, 1.05) Wong, 2007 0.55 ( 0.26, 1.17) Tornwall, 2001 0.43 ( 0.16, 1.15)

Overall 0.58 ( 0.51, 0.66)

18

Table 1. Prospective studies of diabetes mellitus and abdominal aortic aneurysm

First author, publication year, country

Study name or description

Study period Number of participants, number of cases

Type of diabetes mellitus, subgroup

Comparison Relative risk (95% confidence interval)

Adjustment for confounders

Tornwall ME et al, 2001, Finland

Alpha-Tocopherol, Beta-Carotene Study

1985-1988 - NA, 5.8 years follow-up

29133 male smokers, age 50-69 years: 181 AAA cases

Diabetes mellitus Yes vs. no 0.43 (0.16-1.15) Age, cigarettes per day, years of smoking, BMI, SBP, DBP, total cholesterol, HDL cholesterol, education, exercise

Iribarren C et al, 2007, USA

Kaiser Multiphasic Health Checkup Cohort Study

1965-1970 - 2003, 13 years follow-up

104813 men and women, age ≥18 years: 605 AAA cases

Diabetes, allDiabetes, menDiabetes, women

Yes vs. noYes vs. noYes vs. no

0.62 (0.36-1.05)0.65 (0.36-1.15)0.43 (0.10-1.75)

Age, sex, race, education, alcohol, smoking status and packs per day, height, weight, sagittal abdominal diameter, hypertension, serum total cholesterol, white blood cell count, history of coronary heart disease, COPD, stroke, intermittent claudication, estimated GFR, HT

Wong DR et al, 2007, USA

Health Professionals Follow-up Study

1986-2002, 14.6 years follow-up

39352 men, age 40-75 years: 376 AAA cases

Diabetes Yes vs. no 0.55 (0.26-1.17) Age, smoking, hypertension, physical activity, hypercholesterolemia, BMI

Lederle FA et al, 2008, USA

Women's Health Initiative

1993-1998 -2004-2005, 7.8 years follow-up

161808 women, age 50-79 years: 184 AAA cases

Diabetes mellitus Yes vs. no 0.29 (0.13-0.68) Age, ethnicity/race, height, weight, smoking status, pack-years, hypertension, drugs for high cholesterol, coronary artery disease, cerebrovascular disease, peripheral artery disease, venous thromboembolism, COPD, non-skin cancer, previous diagnosis of aortic aneurysm, HT, alcohol

Campbell PT et al, 2012, USA

Cancer Prevention Study 2

1982-2008, 26 years follow-up

1053831 men and women, age ≥30 years: 1404/2873 AA deaths

Diabetes mellitus, menDiabetes mellitus, women

Yes vs. noYes vs. no

0.59 (0.48-0.73)0.62 (0.44-0.87)

Age, education, BMI, smoking, alcohol, vegetables, red meat, physical activity, aspirin use

Ohrlander T et al, 2012, Sweden

Longitudinal Multilevel

1991 - 2003, 13 years

243223 men and women, age 60-

Diabetes mellitus, menDiabetes mellitus, women

Yes vs. noYes vs. no

0.38 (0.24-0.61)0.41 (0.19-0.88)

Age, income, hypertensive diseases, ischemic heart disease,

19

Analysis in Scania (LOMAS)

follow-up 90 years: 3335 AAA cases

cerebrovascular disease, atherosclerosis, chronic lower respiratory disease, and for women also adjustment for: inflammatory polyarthropathies, systemic connective tissue disorder, polymyalgia rheumatica, diseases of oral cavity and jaws, non-infective enteritis and colitisand for men also adjustment for: arthrosis, other diseases of intestines, diseases of peritoneum

Sode BF et al, 2013, Denmark

Copenhagen City Heart study

1976-1978 - 2010, 23 years follow-up

15072 men and women, age 20-≥80 years: 335 AAA cases

Diabetes mellitus Yes vs. no 0.5 (0.2-1.1) Age, sex, smoking, hypertension, BMI, hypercholesterolemia, alcohol

Sode BF et al, 2013, Denmark

Copenhagen General Population Study

2003 - 2010, 4 years follow-up

56211 men and women, age 20-≥80 years: 169 AAA cases

Diabetes mellitus Yes vs. no 1.1 (0.7-2.0) Age, sex, smoking, hypertension, BMI, hypercholesterolemia, alcohol

Tsai CL et al, 2015, Taiwan

The National Health Research Institute

1998-2008 - 2010, 6.2 years follow-up

807101 men and women, age ≥55 years: 1454 AAA cases

Type 2 diabetes mellitus, AAA without rupture

Type 2 diabetes, mellitus, AAA with rupture

NoUncomplicated DM2Advanced DM2NoUncomplicated DM2Advanced DM2

1.000.58 (0.45-0.74)0.53 (0.40-0.69)1.001.17 (0.70-1.95)0.57 (0.30-1.09)

Age, sex, hypertension, chronic kidney disease, ischemic heart disease, stroke, retinopathy, blindness

Jahangir E et al, 2015, USA

Southern Community Cohort Study

2002-2009, 4.94 years follow-up

18782 men and women, age ≥65 years: 281 cases

Diabetes, allDiabetes, menDiabetes, women

Yes vs. noYes vs. noYes vs. no

0.75 (0.53-1.05)0.68 (0.43-1.10)0.81 (0.49-1.34)

Age, sex, race, education, BMI, smoking status, history of MI/CABG, history of high blood pressure, history of high cholesterol

Shah AD et al, 2015, England

The CALIBER Programme

1998-2010, 5.5 years follow-up

1921260 men and women, age ≥30 years: 3113 AAA cases

Diabetes mellitus Yes vs. no 0.46 (0.35-0.59) Age, sex, BMI, deprivation, HDL cholesterol, total cholesterol, SBP, smoking status, statin, antihypertensive drug prescriptions

20

Tang W et al, 2016, USA

Atherosclerosis Risk in Communities Study

1987-1989 - 2013, 24 years follow-up

15703 men and women, age 45-64 years: 588 AAA cases

Diabetes mellitus Yes vs. no 0.52 (0.36-0.75) Age, sex, race, height, smoking status, pack-years of smoking, alcohol, HDL-cholesterol, LDL-cholesterol, triglycerides, hypertension, peripheral artery disease, total cholesterol

Wang L et al, 2017, USA

Physicians' Health Study I and Physicians' Health Study II

1982 - 20121996/1997 - 2012, 10.4 years follow-up

25554 men, age 40-84 years: 471 AAA cases

Diabetes mellitus Yes vs. no 0.79 (0.57-1.11) Age, race, randomized treatment assignment, BMI, smoking status, alcohol, vigorous exercise, hypertension, hypercholesterolemia, cardiovascular disease

Larsson SC et al, 2018, Sweden

Swedish Mammography Cohort Study and Cohort of Swedish Men

1997 - 2014, 28.1 years follow-up

71483 men and women, mean age 58.9/61.1 years: 1201 AAA cases

Type 2 diabetes mellitus Yes vs. no 0.57 (0.40-0.82) Age, sex, BMI, education, FH - MI, smoking status, pack-years of smoking, aspirin use, exercise, walking/bicycling, hypertension, hypercholesterolemia, alcohol, total energy, DASH diet score

AAA=abdominal aortic aneurysm, BMI=body mass index, CABG=coronary artery bypass grafting, CAD=coronary artery disease, COPD=chronic

obstructive pulmonary disease, DASH=Dietary Approaches to Stop Hypertension, FH=family history, GFR=glomerular filtration rate, HDL=high-

density lipoprotein, HT=hormone therapy, LDL=low-density lipoprotein, MI=myocardial infarction, NA=not available, PAD=peripheral artery disease,

SBP=systolic blood pressure

21

Table 2. Subgroup analyses of diabetes mellitus and abdominal aortic aneurysm

Diabetes mellitus and abdominal aortic aneurysm

n Relative risk (95% CI) I2 (%) Ph1 Ph

2

All studies 14 0.58 (0.51-0.66) 40.4 0.06

Gender

Men 7 0.60 (0.51-0.72) 16.1 0.31 0.74/

0.79 Women 5 0.56 (0.40-0.77) 27.1 0.24

Men, women 6 0.58 (0.48-0.70) 46.3 0.10

Outcome type

Incidence 13 0.58 (0.50-0.67) 44.8 0.04 0.87

Mortality 1 0.60 (0.50-0.72)

Follow-up

<10 years 6 0.60 (0.46-0.77) 64.4 0.02 0.78

≥10 years 8 0.58 (0.51-0.66) 9.5 0.36

Geographic location

Europe 6 0.54 (0.40-0.71) 54.7 0.05 0.45

America 7 0.66 (0.57-0.76) 54.9 0.03

Asia 1 0.60 (0.51-0.71)

Number of cases

<250 3 0.54 (0.22-1.33) 75.4 0.02 0.20

250-<1000 6 0.66 (0.55-0.79) 0 0.56

≥1000 5 0.54 (0.47-0.63) 38.7 0.16

Study quality

0-3 stars 0 0.83

4-6 stars 5 0.59 (0.42-0.82) 31.6 0.21

7-9 stars 9 0.58 (0.50-0.67) 48.1 0.05

Adjustment for confounding factors3

Age Yes 14 0.58 (0.51-0.66) 40.4 0.06 NC

22

No 0

Education Yes 4 0.62 (0.54-0.71) 0 0.68 0.45

No 10 0.56 (0.46-0.67) 53.4 0.02

Alcohol Yes 8 0.62 (0.52-0.75) 38.1 0.13 0.34

No 6 0.54 (0.44-0.66) 44.7 0.11

Smoking Yes 12 0.60 (0.52-0.70) 37.7 0.09 0.43

No 2 0.50 (0.33-0.76) 73.6 0.05

BMI or obesity Yes 11 0.61 (0.52-0.72) 41.4 0.07 0.33

No 3 0.52 (0.41-0.67) 49.3 0.14

Physical activity Yes 5 0.62 (0.54-0.71) 0 0.56 0.53

No 9 0.56 (0.46-0.68) 55.3 0.02

Hypertension Yes 12 0.58 (0.50-0.68) 48.5 0.03 0.95

No 2 0.59 (0.50-0.71) 0 0.52

Hypercholesterolemia Yes 8 0.62 (0.48-0.79) 57.8 0.02 0.40

No 6 0.57 (0.51-0.64) 0 0.45

Cholesterol No 2 0.57 (0.36-0.91) 0 0.52 0.91

Yes 12 0.58 (0.51-0.67) 48.6 0.03

Coronary heart disease Yes 6 0.59 (0.47-0.74) 57.0 0.04 0.78

No 8 0.57 (0.48-0.67) 26.8 0.22

Stroke Yes 4 0.50 (0.37-0.67) 52.8 0.10 0.26

No 10 0.61 (0.52-0.72) 38.7 0.10

Chronic obstructive

pulmonary disease

Yes 2 0.45 (0.22-0.94) 56.2 0.13 0.49

No 12 0.59 (0.52-0.67) 42.0 0.06

n denotes the number of risk estimates

1P for heterogeneity within each subgroup

2 P for heterogeneity between subgroups with meta-regression analysis

2 P for heterogeneity between men and women (excluding studies with both genders) with

23

meta-regression analysis

BMI, body mass index

NC, not calculable because no studies were present in one of the subgroups.

24

Supplementary Text. Search terms for the PubMed search

diabetes OR diabetes[MeSH] OR glucose OR glucose[MeSH] OR "medical history" OR medical history[MeSH]

AND

"aortic aneurysm" OR aortic aneurysm[MeSH]

AND

"case-control" OR cohort OR prospective OR longitudinal OR retrospective OR "follow-up" OR "cross-sectional" OR "hazard ratio" OR "hazard ratios" OR "relative risk" OR "relative risks" OR "incidence rate ratio" OR "incidence rate ratios" OR "odds ratio" OR odds ratios OR incidence

Search terms for the Embase search

((diabetes or glucose or medical history).ab,ti. or diabetes/ or glucose/ or medical history/)

and

(aortic aneurysm.ab,ti. or aortic aneurysm/)

and

(case-control or cohort or prospective or longitudinal or retrospective or follow-up or cross-sectional or hazard ratio or hazard ratios or relative risk or relative risks or incidence rate ratio or incidence rate ratios or odds ratio or odds ratios or incidence).af.

25

Supplementary Table 1. List of excluded studies and exclusion reasons

Exclusion reason Reference numberAbstract (1)Aortic dissection (2)Case-control study (3-9)Case only study (10)Commentary, editorial (11-13)Cross-sectional study (14-22)Duplicates (23-25)HbA1c as exposure (26)Medications as exposure (27)Meta-analysis (28-34)No risk estimates (35-37)Not relevant exposure (38)Not relevant outcome (39)Reviews (40-45)Survival (46)Thoracic aortic aneurysm as outcome (47;48)

Reference List

(1) Shah A.D., Langenberg C, Rapsomaniki E, Denaxas S, Pujades-Rodriguez M, Gale CP, Deanfield J, Smeeth L, Timmis A, Hemingway H. Type 2 diabetes and incidence of a wide range of cardiovascular diseases: a cohort study in 1.9 million people. Lancet 2015 Feb 26;385 Suppl 1:S86.

(2) He X, Liu X, Liu W, Wang B, Liu Y, Li Z, Wang T, Tan R, Gao B, Zeng H. Association between Diabetes and Risk of Aortic Dissection: A Case-Control Study in a Chinese Population. PLoS One 2015;10(11):e0142697.

(3) Blanchard JF, Armenian HK, Friesen PP. Risk factors for abdominal aortic aneurysm: results of a case-control study. Am J Epidemiol 2000 Mar 15;151(6):575-83.

(4) Madaric J, Vulev I, Bartunek J, Mistrik A, Verhamme K, De BB, Riecansky I. Frequency of abdominal aortic aneurysm in patients >60 years of age with coronary artery disease. Am J Cardiol 2005 Nov 1;96(9):1214-6.

(5) Wanhainen A, Bergqvist D, Boman K, Nilsson TK, Rutegard J, Bjorck M. Risk factors associated with abdominal aortic aneurysm: a population-based study with historical and current data. J Vasc Surg 2005 Mar;41(3):390-6.

(6) Smelser DT, Tromp G, Elmore JR, Kuivaniemi H, Franklin DP, Kirchner HL, Carey DJ. Population risk factor estimates for abdominal aortic aneurysm from electronic medical records: a case control study. BMC Cardiovasc Disord 2014 Dec 4;14:174.

26

(7) Yuan H, Han X, Jiao D, Zhou P. A Case-Control Study of Risk Factors of Abdominal Aortic Aneurysm. Heart Surg Forum 2016 Oct 21;19(5):E224-E228.

(8) Takeuchi H, Okuyama M, Uchida HA, Kakio Y, Umebayashi R, Okuyama Y, Fujii Y, Ozawa S, Yoshida M, Oshima Y, Sano S, Wada J. Chronic Kidney Disease Is Positively and Diabetes Mellitus Is Negatively Associated with Abdominal Aortic Aneurysm. PLoS One 2016;11(10):e0164015.

(9) Lamorte WW, Scott TE, Menzoian JO. Relationship of cardiovascular risk factors to racial differences in femoral bypass surgery and abdominal aortic aneurysmectomy in Massachusetts. Ann N Y Acad Sci 1996 Nov 18;800:25-35.

(10) Theivacumar NS, Stephenson MA, Mistry H, Valenti D. Diabetes mellitus and aortic aneurysm rupture: a favorable association? Vasc Endovascular Surg 2014 Jan;48(1):45-50.

(11) Golledge J, Norman PE. Diabetes and aortic aneurysm. Am J Cardiol 2008 Jun 1;101(11):1680-1.

(12) Dua MM, Dalman RL. Identifying abdominal aortic aneurysm risk factors in postmenopausal women. Womens Health (Lond) 2009 Jan;5(1):33-7.

(13) Lederle FA. The strange relationship between diabetes and abdominal aortic aneurysm. Eur J Vasc Endovasc Surg 2012 Mar;43(3):254-6.

(14) Chun KC, Teng KY, Chavez LA, Van Spyk EN, Samadzadeh KM, Carson JG, Lee ES. Risk factors associated with the diagnosis of abdominal aortic aneurysm in patients screened at a regional Veterans Affairs health care system. Ann Vasc Surg 2014 Jan;28(1):87-92.

(15) Nicholls EA, Norman PE, Lawrence-Brown MM, Goodman MA, Pedersen B. Screening for abdominal aortic aneurysms in Western Australia. Aust N Z J Surg 1992 Nov;62(11):858-61.

(16) Krohn CD, Kullmann G, Kvernebo K, Rosen L, Kroese A. Ultrasonographic screening for abdominal aortic aneurysm. Eur J Surg 1992 Oct;158(10):527-30.

(17) Smith FC, Grimshaw GM, Paterson IS, Shearman CP, Hamer JD. Ultrasonographic screening for abdominal aortic aneurysm in an urban community. Br J Surg 1993 Nov;80(11):1406-9.

(18) Kanagasabay R, Gajraj H, Pointon L, Scott RA. Co-morbidity in patients with abdominal aortic aneurysm. J Med Screen 1996;3(4):208-10.

(19) Mattes E, Davis TM, Yang D, Ridley D, Lund H, Norman PE. Prevalence of abdominal aortic aneurysms in men with diabetes. Med J Aust 1997 Jun 16;166(12):630-3.

(20) Kang SS, Littooy FN, Gupta SR, Johnson GR, Fisher SG, Cote WL, Steffen GF, Mansour MA, Labropoulos N, Maggio JC. Higher prevalence of abdominal aortic aneurysms in patients with carotid stenosis but without diabetes. Surgery 1999 Oct;126(4):687-91.

(21) Lederle FA, Johnson GR, Wilson SE. Abdominal aortic aneurysm in women. J Vasc Surg 2001 Jul;34(1):122-6.

27

(22) Badger SA, O'Donnell ME, Sharif MA, Boyd CS, Hannon RJ, Lau LL, Lee B, Soong CV. Risk factors for abdominal aortic aneurysm and the influence of social deprivation. Angiology 2008 Oct;59(5):559-66.

(23) Robson JC, Kiran A, Maskell J, Hutchings A, Arden N, Dasgupta B, Hamilton W, Emin A, Culliford D, Luqmani RA. The relative risk of aortic aneurysm in patients with giant cell arteritis compared with the general population of the UK. Ann Rheum Dis 2015 Jan;74(1):129-35.

(24) Chiu HY, Lo PC, Huang WF, Tsai YW, Tsai TF. Increased risk of aortic aneurysm (AA) in relation to the severity of psoriasis: A national population-based matched-cohort study. J Am Acad Dermatol 2016 Oct;75(4):747-54.

(25) Stackelberg O, Wolk A, Eliasson K, Hellberg A, Bersztel A, Larsson SC, Orsini N, Wanhainen A, Bjorck M. Lifestyle and Risk of Screening-Detected Abdominal Aortic Aneurysm in Men. J Am Heart Assoc 2017 May 10;6(5).

(26) Kristensen KL, Dahl M, Rasmussen LM, Lindholt JS. Glycated Hemoglobin Is Associated With the Growth Rate of Abdominal Aortic Aneurysms: A Substudy From the VIVA (Viborg Vascular) Randomized Screening Trial. Arterioscler Thromb Vasc Biol 2017 Apr;37(4):730-6.

(27) Hsu CY, Su YW, Chen YT, Tsai SH, Chang CC, Li SY, Huang PH, Chen JW, Lin SJ. Association between use of oral-antidiabetic drugs and the risk of aortic aneurysm: a nested case-control analysis. Cardiovasc Diabetol 2016 Sep 1;15(1):125.

(28) Cornuz J, Sidoti PC, Tevaearai H, Egger M. Risk factors for asymptomatic abdominal aortic aneurysm: systematic review and meta-analysis of population-based screening studies. Eur J Public Health 2004 Dec;14(4):343-9.

(29) Shantikumar S, Ajjan R, Porter KE, Scott DJ. Diabetes and the abdominal aortic aneurysm. Eur J Vasc Endovasc Surg 2010 Feb;39(2):200-7.

(30) De RP, Farchioni L, Fiorucci B, Lenti M. Diabetes and abdominal aortic aneurysms. Eur J Vasc Endovasc Surg 2014 Mar;47(3):243-61.

(31) Takagi H, Umemoto T. A contemporary meta-analysis of the association of diabetes with abdominal aortic aneurysm. Int Angiol 2015 Aug;34(4):375-82.

(32) Xiong J, Wu Z, Chen C, Wei Y, Guo W. Association between diabetes and prevalence and growth rate of abdominal aortic aneurysms: A meta-analysis. Int J Cardiol 2016 Oct 15;221:484-95.

(33) Takagi H, Umemoto T. Negative association of diabetes with rupture of abdominal aortic aneurysm. Diab Vasc Dis Res 2016 Sep;13(5):341-7.

(34) Takagi H, Umemoto T. Negative Association of Diabetes With Thoracic Aortic Dissection and Aneurysm. Angiology 2017 Mar;68(3):216-24.

(35) Hammond EC, Garfinkel L. Coronary heart disease, stroke, and aortic aneurysm. Factors in the etiology. Arch Environ Health 1969 Aug;19(2):167-82.

28

(36) Lopez-de-Andres A, Jimenez-Trujillo I, Jimenez-Garcia R, Hernandez-Barrera V, de Miguel-Yanes JM, Mendez-Bailon M, Perez-Farinos N, Salinero-Fort MA, Carrasco-Garrido P. National trends in incidence and outcomes of abdominal aortic aneurysm among elderly type 2 diabetic and non-diabetic patients in Spain (2003-2012). Cardiovasc Diabetol 2015 May 7;14:48.

(37) Wierzba W, Sliwczynski A, Pinkas J, Jawien A, Karnafel W. Diabetes mellitus increases the risk of ruptured abdominal aortic aneurysms. Diab Vasc Dis Res 2017 May 1;1479164117710391.

(38) Strachan DP. Predictors of death from aortic aneurysm among middle-aged men: the Whitehall study. Br J Surg 1991 Apr;78(4):401-4.

(39) Treiman GS, Treiman RL, Foran RF, Cossman DV, Cohen JL, Levin PM, Wagner WH, Davidson MB. The influence of diabetes mellitus on the risk of abdominal aortic surgery. Am Surg 1994 Jun;60(6):436-40.

(40) Wilmink AB, Quick CR. Epidemiology and potential for prevention of abdominal aortic aneurysm. Br J Surg 1998 Feb;85(2):155-62.

(41) Blanchard JF. Epidemiology of abdominal aortic aneurysms. Epidemiol Rev 1999;21(2):207-21.

(42) Norman PE, Davis TM, Le MT, Golledge J. Matrix biology of abdominal aortic aneurysms in diabetes: mechanisms underlying the negative association. Connect Tissue Res 2007;48(3):125-31.

(43) Torsney E, Pirianov G, Cockerill GW. Diabetes as a negative risk factor for abdominal aortic aneurysm - does the disease aetiology or the treatment provide the mechanism of protection? Curr Vasc Pharmacol 2013 May;11(3):293-8.

(44) Pafili K, Gouni-Berthold I, Papanas N, Mikhailidis DP. Abdominal aortic aneurysms and diabetes mellitus. J Diabetes Complications 2015 Nov;29(8):1330-6.

(45) Radak D, Tanaskovic S, Katsiki N, Isenovic ER. Protective Role of Diabetes Mellitus on Abdominal Aortic Aneurysm Pathogenesis: Myth or Reality? Curr Vasc Pharmacol 2016;14(2):196-200.

(46) Sasaki S, Takigami K, Kunihara T, Shiiya N, Murashita T, Matsui Y, Yasuda K. Abdominal aortic aneurysms in aged patients: analysis of risk factors in non-ruptured cases. J Cardiovasc Surg (Torino) 1999 Feb;40(1):1-5.

(47) Jimenez-Trujillo I, Gonzalez-Pascual M, Jimenez-Garcia R, Hernandez-Barrera V, de Miguel-Yanes JM, Mendez-Bailon M, de Miguel-Diez J, Salinero-Fort MA, Perez-Farinos N, Carrasco-Garrido P, Lopez-de-Andres A. Type 2 Diabetes Mellitus and Thoracic Aortic Aneurysm and Dissection: An Observational Population-Based Study in Spain From 2001 to 2012. Medicine (Baltimore) 2016 May;95(18):e3618.

(48) Theivacumar NS, Stephenson MA, Mistry H, Valenti D. Diabetics are less likely to develop thoracic aortic dissection: a 10-year single-center analysis. Ann Vasc Surg 2014 Feb;28(2):427-32.

29

Supplementary Figure 1. Funnel plot of diabetes mellitus and abdominal aortic aneurysm

0.1

.2.3

.4.5

s.e.

of l

ogrr

-1.5 -1 -.5 0 .5logrr

Funnel plot with pseudo 95% confidence limits

30

Supplementary Figure 2. Influence analysis of diabetes mellitus and abdominal aortic aneurysm

0.50 0.58 0.51 0.66 0.69

Larsson, 2018

Wang, 2017

Tang, 2016

Jahangir, 2015

Shah, 2015

Tsai, 2015

Sode, 2013, CCHS

Sode, 2013, CGPS

Campbell, 2012

Ohrlander, 2012

Lederle, 2008

Iribarren, 2007

Wong, 2007

Tornwall, 2001

Study ommited Meta-analysis random-effects estimates (exponential form)

------------------------------------------------------------------------------ Study omitted | e^coef. [95% Conf. Interval]-------------------+---------------------------------------------------------- Larsson, 2018 | 0.58324391 0.50654268 0.67155927 Wang, 2017 | 0.5663572 0.49781173 0.64434099 Tang, 2016 | 0.58796215 0.5115422 0.67579854 Jahangir, 2015 | 0.5691492 0.49799052 0.6504758 Shah, 2015 | 0.60129404 0.5275467 0.68535066 Tsai, 2015 | 0.57854909 0.49564162 0.6753248 Sode, 2013, CCHS | 0.58429837 0.51099461 0.66811776 Sode, 2013, CGPS | 0.56865907 0.50740427 0.63730866 Campbell, 2012 | 0.57877702 0.49675906 0.67433667 Ohrlander, 2012 | 0.6001026 0.53054148 0.67878413 Lederle, 2008 | 0.59149033 0.52226812 0.66988736 Iribarren, 2007 | 0.58044702 0.50619096 0.66559613 Wong, 2007 | 0.58322543 0.50961477 0.66746867 Tornwall, 2001 | 0.58535528 0.51258343 0.66845858-------------------+---------------------------------------------------------- Combined | 0.58257625 0.51194192 0.66295625

31

------------------------------------------------------------------------------