Advanced complicated diabetes mellitus is associated with a reduced risk of thoracic and abdominal aortic aneurysm rupture: a population-based cohort study.

Advanced complicated diabetes

mellitus is associated

with a reduced risk of thoracic and

abdominal aortic

aneurysm rupture: a

population-based cohort study

Chung-Lin Tsai

Cheng-Li Lin

Yi-Ying Wu

Dong-Chen Shieh

Fung-Chang Sung

*

Chia-Hung Kao

*

Introduction

Diabetes mellitus (DM) may carry some lethal complications for patients even following intensive blood sugar control. The epidemic database revealed that DM has become an epidemic in many populations worldwide [1]. Abdominal aortic aneurysm (AAA) is generally considered as the disease of ageing, mainly occurring in the elderly. AAA has been ranked as the tenth leading cause of death in the elderly [2]. Type 2 diabetes could contribute 10–15% in the risk of developing AAAs [3–6]. However, a benefcial effect of diabetes on AAA was frst observed in a US study in 1997 [7]. The same research team recruited a new group of 52 745 subjects for a similar study, and the negative association between diabetes and AAA remains [8]. Thoracic aortic

aneurysm (TAA) is an aorta aneurysm with high fatality but less common than AAA. Studies reported that these two diseases have distinct clinical and genetic profles; however, diabetes also has been associated with decreased atherosclerosis and other cardiovascular risks

[9–16]. These evidences provide challenges to the traditional view of AAA and TAA as a manifestation of atherosclerosis.

Series of researches have explored on the

protective effect of the paradoxical confuting phenomena. All these could be associated with changes in the aortic wall and characteristics of the mural thrombus [17–19]. Most of previous studies on the association between diabetes and TAA and AAA are cross-sectional and case– control designs and mainly for Western populations. The objective of this study was to examine the subjects of an Asian population using retrospective cohort study using the Taiwan National Health Insurance Research Database (NHIRD). The insurance programme has been reformed

since 1995, with approximately 99% of 23.74 million population covered by the end of 2009. This study measured

the pooled incidence of AAA and TAA in patients with diabetes and in general population. The severity of hyperglycaemia may have various effects on biologic changes in the aortic wall and fbrolytic characteristics. This study would further examine how the severity of diabetes is associated with TAA and AAA with and without rupture.

Research design and methods

Data source

The National Health Research Institute (NHRI) was responsible to provide the medical claims data. After receiving the approval for this study from NHRI, we were able to use the scrambled patient identifcation numbers to link fles, including inpatient care claims and the registry for benefciaries. This study was exempted from a full institutional review (CMU-REC-101-012). The NHRI maintains

and updates the NHIRD. The insurance programme maintains contracts with 97% of hospitals and clinics in Taiwan

[20]. The accuracy and high validity of diagnoses in the NHIRD have also been evaluated [21,22]. The International Classifcation of Diseases, Ninth Revision (ICD-9), was used for the diagnosis codes in this study. This study used the inpatient claims data for all 23 740 000 insured people to established study cohorts.

Study participants

the two study cohorts. Patients with the primary diagnosis of type 2 DM (ICD-9 code 250) identifed as new-onset cases during the 1998–2008 period were included as the DM cohort. The index date for patients with type 2 DM was the frst medical visit date. We excluded patients with a history of TAA and AAA (ICD-9 codes 441.1, 441.2, 441.3 and 441.4, respectively) diagnosed before the index date, those with missing information on age or sex and those younger than 54 years of age. We used a simple random sampling method to select four insured people without diabetes history for each type 2 DM patient as the non-DM cohort, frequency matched by age (5 years), sex and index year. We adopted a 1 : 4 matching design method to increase the statistical power.

Outcome measures

The follow-up time began on the index date and until the date of the TAA and AAA diagnoses, withdrawal from the insurance system, death or 31 December 2010, whichever came frst. The patients diagnosed with TAA and AAA were identifed using the hospital discharge records. The baseline comorbidities of hypertension (ICD-9 codes 401–405), chronic kidney disease (ICD-9 codes 582.9, 585.3–585.9, 586), ischaemic heart disease (ICD-9 codes 412, 414.00–414.07, 414.2–414.9), stroke (ICD-9 codes 430–438), retinopathy (ICD-9 codes 362.0) and blindness (ICD-9 codes 369) were identifed from the claims of inpatients records.

Statistical analysis

We compared the categorical distributions of sex, age and baseline comorbidities between the type 2 DM cohort and the non-DM comparison cohort using the chi-square test. The mean age between both cohorts was measured and tested using t-test. We estimated the pooled incidence rate of TAA and AAA, compared between the two cohorts by sex, age and comorbidity. A Poisson regression model

was used to estimate the DM cohort to non-DM cohort incidence rate ratios and 95% confdence intervals (CIs) of

regression analysis to calculate the hazard ratios (HRs) of TAA and AAA and 95% CIs. Multivariate models were also used to assess the adjusted HRs (aHRs) of developing

TAA and AAA associated with sociodemographic factors and comorbidities. A further analysis assessed the aHRs of

TAA and AAA with and without rupture by the severity level of type 2DM. The severity of diabetic patientswas partitioned into uncomplicated and advanced groups according to existing DM-related diseases; they were in ICD-9 codes of 250.0–250.3 and 250.4–250.9, respectively. Compared with non-DM cohort, aHRs associated with uncomplicated type 2 DM and advanced type 2 DM were estimated. The statistical signifcance level was set at p<0.05 (SAS _{software, version} 9.1, SAS Institute Inc, Cary, NC, USA).

Results

This study consisted of 160 391 patients in the type 2 DM cohort and 646 710 subjects in the non-DM comparison cohort. Both cohorts were similar in sex and age distributions, mainly the elderly (70.7%) (Table 1). More type 2

DM patients resided in the lower urbanized areas (51.0% vs 48.8%) and held blue-collar jobs (50.0% vs 45.8%) with lower income, compared with non-DM subjects. Baseline comorbidities, including hypertension,

chronic kidney disease, ischaemic heart disease, stroke, retinopathy and blindness, were more prevalent in the type 2

DMcohort than in the non-DMcohort (p<0.0001) (Table 2). Table 3 shows that the overall incidence rate of TAA and AAA was 15% lower in the type 2 DM cohort than

in the non-DM cohort (3.85 vs 4.51 per 10 000 personyears), with an aHR of 0.65 (95% CI 0.56–0.74) (Table 3).

The incidence was higher in men than in women but with an aHR of 0.76 for women and of 0.59 for men. The agespecifc analysis showed that the incidence increased with

age in both cohorts. But the HRs showed that the benefcial effect of diabetes was relatively better for the older groups. The incidence of TAA and AAA was higher for patients with comorbidity, except those with retinopathy and blindness. The analysis of Kaplan–Meiermodel showed a lower cumulative

incidence of TAA and AAA in the type 2 DM cohort than in the non-DM cohort. The two curves became wider starting the ffth year of follow-up and signifcantly different by the end of follow-up (Figure 2).

Further Cox model measured the HRs associated with specifc variables for the whole study population (Table 4). Compared with 55- to 64-year-old subjects, those aged 75 years and above had an aHR of 5.24 in the multivariable analysis. Higher hazard was found signifcant for

women and low-income people (aHRs=2.79, p<0.001 and 1.24, p<0.05, respectively). Hypertension, chronic kidney disease and ischaemic heart disease were associated with elevated hazards. The overall reduced aHR in

association with diabetes remained.

Table 5 shows the HRs of TAA and AAA with and without rupture by severity of type 2 DM. The advanced type 2 DM functioned as a signifcant benefcial factor for TAA and AAA without rupture with reduced aHRs of 0.50 (95% CI 0.35–0.71) and 0.53 (95% CI 0.40–0.69),

respectively. The uncomplicated type 2 DM patients also had a protective association with AAA without rupture (aHR=0.58, 95% CI 0.45–0.74). The aHRs shown in

this table failed to demonstrate the signifcant relationship between diabetes and rupture cases.

Discussion

The increasing trends of incidence and prevalence of AAA

in patients with DM have been documented, and the disease has become the tenth most lethal cause in older

men [23,24]. Studies found no difference in genetic predispositions of AAA between Caucasian and other ethnic

groups. Tilsonwas the frst to illustrate the fundamental issue in 1992 [25]. In 1998, a study inHong Kong demonstrated an increasing trend for the disease, with a pattern similar to that reported for the Western population [26]. An updated

prospective data from South-East Asian state of Sarawak also reported in 2003 that AAAwas common in Asian population, comparable with the fnding of Western study [27].

determine the true incidence of the disease. The reason is that aorta is distinctwith several segments in the embryonic development. They are undiagnosed in clinic. Patients with vasculitis,

such as Takayasu’s arteritis, can also develop TAAs. It

primarily affects youngwomen and occurs most often in Asian ethnicity [28]. Patients younger than 54 years associatedwith cystic medial degeneration of thoracic aorta were excluded from the present study, to assure no misclassifcation in degenerating pathology. With regard to TAA rupture, Clouse et al. indicated that aortic aneurysm dissection and ruptured degenerative TAA occurred with similar frequency but less commonly than ruptured AAA [29].

Diabetes is a well-known predisposing disorder of

atherosclerosis with a high causal role in the occlusive vascular disease. In the Aneurysm Detection and Management

Veterans Affairs Cooperative Study, Lederle et al. frst provided the evidence of inverse relationship between diabetes

and aortic aneurysmdevelopment in 1997 [7]. Several articles have reported later that the AAA enlargement

progresses more slowly in diabetic patients than in non-diabetic subjects [30–34]. In TAA diabetic patients,

studies also found that diabetes is associated with decreased risk for atherosclerosis and cardiovascular disorders [13–16]. Documented articles revealed that

TAA diabetes is associated with increased circulating concentrations of insulin-like growth factor 1, an endocrine

peptide with potent anti-diabetic effects [35,36].

Our fndings from the Oriental population seem consistent with the Western study results. With an aHR of 0.65 (95% CI 0.56–0.74) for AAA and TAA, our study does show that diabetes plays a signifcant protective role in reducing the aortic aneurysm growth.

From these population-based data, the sex-specifc

analysis revealed that the incidence rates of TAA and AAA were not different between the type 2 DM cohort and the

non-DM cohort for women (2.63 vs 2.60 per 10 000 person-years, respectively). On the other hand, the incidence of TAA and AAA was signifcantly lower in the

type 2 DM cohort than in the non-DM cohort for men with an incidence rate ratio of 0.79 (95% CI 0.76–0.810). Hence, women had an HR of 2.79 compared with men, indicating that the role of diabetes in reducing the aortic aneurysm growth is evident in men. Previous studies have also documented similar contract between men and women

[37–40]. The oestrogen levels in women may play an important role in the worse prognosis of aortic aneurysm [41–43].

Among the subjects with comorbidity in this study, the incidence of TAA and AAAwas higher in those with the comorbidity

than those without the comorbidity in the same

cohort. But the increased incidence in the type 2 DM cohort was much lower than that in the non-DM cohort. Therefore, for those with comorbidity, the benefcial effect of diabetes remained; the DM cohort to non-DM cohort aHR ranged from 0.47 for those with chronic kidney disease to 0.56 for those with hypertension. The presence of chronic kidney disease has a role in the increased incidence of aortic aneurysms

in the non-DM cohort but much less in the type 2 DMcohort (19.1 vs 7.89 per 10 000 person-years). Hypertension is a well-known factor contributing a remarkable risk in the development of aortic aneurysms. Our data show that the protective effect of diabetes is even associated with the reduced TAA and AAA risks in those with the comorbidity of

hypertension. These fndings could command the protective effect of DM on reducing aortic aneurysms.

The Kaplan–Meier model-measured cumulative incidence curves of TAA and AAA in this study showed that the benefcial effect of diabetes had a 5-year lag. This fnding is consistent with a previous fnding in the healthy men study [38]. Le et al. found that the risk of AAA declined with the follow-up time for patients with diabetes. An odds ratio of 0.50 appears in 3–5 years of follow-up, which declines further to 0.37 after 12 years of follow-up.

This study is strengthened by using large population data. The fnding on how DM is related to the reduced rupture hazards of aortic aneurysms, that is, thoracic aortic aneurysm rupture, thoracic aortic aneurysm without

rupture, abdominal aortic aneurysm rupture and

abdominal aortic aneurysm without rupture, in this study is of particular importance. We excluded type 1 DM in this study because it has more distinct clinical effects and performances than type II DM. Our data analysis has

further stratifed DM patients into advanced and uncomplicated DM groups. We found that the benefcial effect of

DM appeared between advanced type 2 diabetes and TAA and AAA without rupture. The risks of TAA and AAA in

patients with advanced type 2 diabetes cohort were reduced for near a half. No benefcial effect was found for ruptured cases. Uncomplicated DM was also associated with reduced risk of AAA but not with TAA. No previous study has demonstrated this paradoxical situation.

The advanced diabetic patients react to much potent

medication. Thompson et al. found that diabetes medication could reduce the AAA growth rate for 56%, strongly

associated with taking angiotensin-receptor blockers and aldosterone-receptor blockers. This healthy men study has reported a negative association between fasting glucose

and aortic diameter in 2859 non-diabetic subjects [38]. Investigators at Stanford also addressed that hyperglycaemia

was associated with slower AAA enlargement than insulin therapy [44]. The ‘hyperglycaemia itself’ retarded aneurysm progression rather than its treatment. These studies

provide a powerful basis of theory for our study. This study has few limitations. First, we used the total population inpatients claims data to select study cohorts because the incident cases of TAA and AAA in outpatient data were not large enough for this study. We, therefore, may miss some cases in the outpatient claims. Second, the HbA1c _{value is used as a predictor of further complication}

of diabetes at clinics [45–47]. The HbA1c _{values were}

unavailable in our claims data for measuring the severity of diabetes for our study population. Therefore, classifcation error may exist for measuring the association

between the severity level of type 2 DM and the hazards of TAA and AAA with and without rupture. Third, information

on smoking is not available in the insurance records. We, therefore, were unable to control for smoking in the data analysis. Fourth, some patients with ruptured aneurysms might not be included in this study because they

could not reach hospital alive. The sudden deaths that occurred in patients with severe diabetes are more likely to be attributed to cardiac causes than ruptured aneurysms. The ‘competing risk’ may underestimate the ruptured cases [7]. Our study used large population-based data with adequate power to date and has provided meaningful fndings in thoracic and abdominal aortic aneurysms for patients with diabetes in an Asian population.

Conclusion

In this population-based study, the overall incidence rate of TAA and AAA was 15% lower in the type 2 DM cohort than in the general population. This relationship is also greater for men than for women. Patients with more advanced diabetes have more notable reduced hazards of thoracic and abdominal aortic aneurysms. This database of the study is the largest and robust to date and could describe a meaningful observation between aortic aneurysm development

and diabetes in Asian population. Attentions should further approach to determine the biological pathways of aneurysm protective effect in advanced hyperglycaemia condition.