- Email: [email protected]
Diabetics Are Less Likely to Develop Thoracic Aortic Dissection: A 10-Year Single-Center Analysis
Diabetics Are Less Likely to Develop Thoracic Aortic Dissection: A 10-Year Single-Center Analysis Nada Selva Theivacumar, Matthew A. Stephenson, Hiren Mistry, and Domenico Valenti, London, United Kingdom
Background: Diabetes mellitus (DM) is an acknowledged risk factor for atherosclerosis, and diabetics are more likely to have hypertension. Atherosclerosis and hypertension are risk factors for aortic dissection. However, recent studies have shown that DM is associated with changes in aortic wall collagen. In this retrospective study we assess the relationship between DM and thoracic aortic dissection (TAD). Methods: Patients with a diagnosis of thoracic aortic dissection during the last 10 years were identified from our hospital records. The prevalence of DM in Stanford type A and B TAD was compared with that of two age- and gender-matched control groups. For every diabetic dissection case, 10 controls were selected from the hospital data. Results: Two hundred nineteen patients (median age 61 years, male:female ratio 145:74) were identified with TAD, comprising 131 type A dissections and 88 type B dissections. Only 3 of 131 (2.3%) type A aortic dissections were diabetics, whereas, in control group 1, 241 of 1310 (18.4%) were diabetics and, in control group 2, 116 of 1310 (8.9%) were diabetics [odds ratios: 0.1 (0.03e0.32) and 0.24 (0.07e0.76), respectively] (P ¼ 0.0001 and 0.007, respectively). Similarly, only 2 of 88 (2.3%) type B aortic dissections were diabetics, whereas 228 of 880 (26.0%) and 102 of 880 (11.6%) were diabetics in groups 1 and 2 [odds ratios: 0.07 (0.02e0.27) and 0.18 (0.04e0.73), respectively] (P ¼ 0.0001 and 0.0035, respectively). All these odds ratios were statistically significant (P < 0.01). Conclusions: Patients with thoracic aortic dissection are less likely to be diabetic. Although we identified association only, not causality, it is possible that DM, or its treatment, has a protective effect against aortic dissection.
INTRODUCTION Diabetes mellitus (DM) is a common disease, affecting 5.5% of the overall population in England.1
Department of Vascular Surgery, King’s Health Partners, King’s College Hospital, London, United Kingdom. Correspondence to: Domenico Valenti, MD, Department of Vascular Surgery, King’s College Hospital, Denmark Hill, London SE5 9RS, UK; E-mail: [email protected] Ann Vasc Surg 2014; 28: 427–432 http://dx.doi.org/10.1016/j.avsg.2013.03.024 Ó 2014 Elsevier Inc. All rights reserved. Manuscript received: October 9, 2012; manuscript accepted: March 12, 2013; published online: December 20, 2013.
Varying incidence and prevalence rates have been reported among different populations and age groups. Its prevalence is higher with increasing age, affecting up to 30% of those >65 years of age in the United States.2 DM is an acknowledged risk factor for atherosclerosis.3 Further, diabetics are more likely to develop systemic hypertension and their blood pressure control is more complex, often requiring polypharmaceutical agents.4 DM has therefore generally been assumed to be a risk factor for thoracic aortic dissection (TAD).5 However, a recent study has shown that DM increases cross-linkage of collagen in the aortic wall, leading to biologic changes in the aortic wall.6 Also, studies have shown that diabetics are less likely to develop aortic aneurysmal disease.79 427
428 Theivacumar et al.
Annals of Vascular Surgery
In this study we assess the relationship between DM and aortic dissection.
METHODS All patients who were admitted to King’s College Hospital, London, with a diagnosis of aortic dissection during the period January 2002 to April 2012, were included. Cases were identified from the hospital electronic information records using the International Classification of Diseases, tenth revision (ICD-10). All patients with the diagnosis code I710 (aortic dissection) were selected. The search was conducted with a filter for hypertension and diabetes mellitus (including E100eE149 codes) to check which cases were diabetic (type 1, 2, or dietcontrolled). Individual case notes were reviewed for the exact diagnosis of Stanford classification10 and other risk factors, such as collagen disorders and bicuspid aortic valve, along with smoking history, hypertension, hypertension medications, coronary heart disease, peripheral vascular disease, and race. The cases were divided into two groups: group A (patients with Stanford type A dissection) and group B (patients with Stanford type B dissection). No single control group could perfectly represent the population from which the dissection patients were drawn. Therefore, for purposes of comparison, we created two control groups. Control group 1 comprised all those admitted to the hospital for any reason. Control group 2 consisted of patients admitted for any inpatient orthopedic procedure. Age- and gender-matched control groups were selected randomly from these groups. For every case, 10 age- and gender-matched controls were randomly selected from the hospital database from the same year of admission as the individual cases. Ten controls were chosen because we had the luxury of large control groups, and ratios >5 have been shown to be advantageous, particularly when there is a low prevalence of exposure among controls.11 To assess matched controls with other comorbidities would have inevitably been highly problematic given the complex interactions between DM and other risk factors for TAD; therefore, the higher ratio of age- and gender-matched controls would help solve this problem. The corresponding controls for group A were controls 1A and 2A and for group B were controls 2A and 2B. The odds ratio of DM in group A (TAD type A) was compared with that of both controls 1A and 2A. Similarly, group B (TAD type B) was compared with 1B and 2B. Finally, both TAD (groups A and
Fig. 1. Proportion of DM cases by age in the control groups, the community from which our patients are drawn, and the TAD group.
B) groups were compared with controls 1 and 2. The prevalence of DM in the TAD group and the control groups was also compared with the community data, obtained from the southeastern England population from which our hospital receives patients. These data were obtained from published key statistics on DM in 2010.12 The percentage prevalence of DM was calculated in each age group along with male:female proportions. We did not distinguish between acute or chronic dissection because we were interested in the development of thoracic aortic dissection per se, rather than the stage of the presentation. Statistical Methods The prevalence of DM among the study groups (patients with type A and B aortic dissection) and controls 1 and 2 were compared. Group comparisons were made using nonparametric statistical testing. Odds ratios (ORs) were obtained, and the chi-squared test or Fisher’s exact test was used to assess the statistical significance of the OR. Confidence intervals (95%) were also obtained (expressed in parentheses in what follows). Age is given presented median years (range).
RESULTS We first established how well our control group matched that of the community prevalence with regard to DM. The percentages of DM in the TAD group and control groups 1 and 2 were compared in different age groups, and also compared with the community diabetic prevalence (Fig. 1). It is worth noting that control group 2 (those who had elective orthopedic procedures) had a very similar prevalence distribution of DM to the community
Vol. 28, No. 2, February 2014
Diabetes and TAD 429
Table I. Demographic details of cases and controls Cases/controls
Median age (range)
Male:female ratio
Total
TAD type A (group A) TAD type B (group B) TAD (A and B) Age- and gender-matched from control 1 (all admissions) Age- and gender-matched from control 2 (patients who had orthopedic procedures)
58 63 61 61 61
87:45 58:30 145:74 1450:740 1450:740
131 88 219 2190 2190
data, unlike control 1 (all hospital admission) in which diabetics were overrepresented (Fig. 1). During the last 10 years, there were 219 patients with a median age of 61 (range 14e92) years presenting with TAD, who were admitted to King’s College Hospital, London. The demographic details are given in Table I. Their breakdown of dissection type and risk factors are given in Table II. Only 3 of 131 (2.2%) type A aortic dissections were diabetics in group A, whereas 241 of 1310 (18.4%) and 116 of 1310 (10.2%) were diabetics in control groups 1 and 2 [OR: 0.1 (0.03e0.32) and 0.24 (0.07e0.76), respectively], which was highly statistically significant (P ¼ 0.0001 and P ¼ 0.007, respectively). Similarly, only 2 of 88 (2.3%) type B aortic dissections were diabetics in group B, whereas 228 of 880 (26.0%) and 102 of 880 (11.6%) were diabetics in groups 1 and 2 [OR: 0.07 (0.02e0.27) and 0.18 (0.04e0.73), respectively] (P ¼ 0.0001 and P ¼ 0.0035, respectively). When all aortic dissections (both A and B) were assessed together as one group, 5 of 214 cases (2.3%) were diabetic, whereas 469 of 2190 (21.4%) and 218 of 2190 (10.0%) were diabetic in control groups 1 and 2 [OR: 0.08 (0.06e0.2) and 0.21 (0.08e0.51), respectively] (P ¼ 0.0001 and P ¼ 0.0003, respectively). These data are also summarized in Table III. All diabetic patients who had TAD (both type A and B) were hypertensive, whereas, among nondiabetics, 82 of 128 (64%) with type A and 69 of 86 (80%) with type B aortic dissection were hypertensive. We found no significant difference in blood pressure levels or numbers of antihypertensives between the diabetic and nondiabetic groups (see Table II). Collagen diseases (Marfan’s syndrome or EhlerseDanlos syndrome) were found in 4 patients in group A and 2 in group B, all of whom were nondiabetics. Bicuspid aortic valve was a risk factor in 2 patients (nondiabetics) in group A, whereas this was not found in group B. No patient had a known history of illicit drug use. When we compared the mortality associated with TAD, 1 diabetic patient died due to TAD
(14e76) (42e92) (14e92) (14e92) (14e92)
compared with 41 deaths among nondiabetics during the last 10 years. Because the mortality numbers were very small in the diabetic group, it was difficult to prove whether diabetics were more or less likely to die from TAD compared with nondiabetics (Table IV).
DISCUSSION This study suggests that diabetic patients are significantly less likely to have aortic dissection, yet it is known from previous studies that diabetics suffer more frequently from hypertension. In our study, however, there was no significant difference between the diabetic and nondiabetic groups in terms of blood pressure levels or numbers of antihypertensive medications.4 High-velocity spiral flow in the proximal aorta constantly challenges the aortic wall. The curved morphology of the aortic arch places the convex aortic wall at persistent risk of injury during each systolic phase. Patients with collagen disorders are therefore at high risk of aortic dissection as they have poor aortic wall integrity. Hypertension is a well-accepted risk factor for aortic dissection.10 Patients with bicuspid aortic valve are likely to produce alternative turbulent blood flow patterns in the proximal aorta, leading to a higher risk of dissection.10 One recent study indicated a similar finding.13 Prakash et al. used a similar epidemiologic approach using national statistics in the United States over a 2year period, with randomly selected controls. They found that DM was associated with a 40e80% reduction in hospitalization due to TAD. They found this to be the case in patients with uncomplicated DM and those with chronic complicated DM, although the negative association was greater with the latter. In this study, the odds ratios of DM in the TAD groups were compared with those of two different control groups who were age- and gendermatched on a case-by-case basis. Control group 1 was taken from all hospital admissions for any
430 Theivacumar et al.
Annals of Vascular Surgery
Table II. Types of TAD and their risk factors in the diabetic and nondiabetic groups Thoracic aortic dissection type (TAD)
Type A Total type A TAD Risk factors Hypertension Systemic blood pressure on admission (median mm Hg) Number of antihypertensives (mode) Collagen disorders Bicuspid valve Hyperlipidemia Median body mass index Coronary heart disease Peripheral vascular disease History of current smoking White British Black African Race Black Caribbean Asian/Chinese Others/nonspecified Type B Total type B TAD Risk factors Hypertension Systemic blood pressure on admission (median mm Hg) Number of antihypertensives (mode) Collagen disorders Bicuspid valve Hyperlipidemia Median body mass index Coronary heart disease Peripheral vascular disease History of current smoking White British Black African Black Caribbean Race Asian/Chinese Others/nonspecified Both type A and B TAD
DM
Non-DM
Total (n)
P
3
128
131
d
3 (100%) 146/90 2 0 0 2 (66.6%) 29.4 1 (33.3%) 0 0 1 (33.3%) 1 (33.3%)
82 (64.1%) 150/88 2 4 (3.1%) 3 (2.3%) 54 (42.2%) 29.6 34 (26.6%) 11 (8.6) 22 (17.2) 98 (76.6%) 11 (8.6%)
85 (42.0%) 150/88 2 4 (3.1%) 3 (2.3%) 56 (42.7%) 29.7 35 (26.7%) 11 (8.4%) 22 (16.8%) 99 (75.6%) 12 (92.4%)
0.55 NA NA NA NA 0.57
1 (33.3%) 0 0
13 (10.1%) 2 (1.6%) 4 (3.1%)
14 (10.7%) 2 (1.5%) 4 (3.1%)
0.29 NA NA
2
86
88
2 (100%) 136/82 2 0 0 2 (100%)
69 (80.2%) 140/84 1 2 (2.3%) 0 62 (72.1%)
71 (80.7%) 140/84 1 2 (2.3%) 0 64 (72.7%)
28.5 2 (100%) 1 (50%) 1 (50%) 2 (100%) 0 0
29.2 49 (57.0%) 22 (25.6%) 24 (28.0%) 62 (72.1%) 9 (10.5%) 8 (9.3%)
29.1 51 (58.0%) 23 (26.1%) 25 (28.4%) 64 (72.7%) 9 (10.2%) 8 (9.1%)
0 0 5
2 (2.3%) 5 (5.8%) 214
2 (2.3%) 5 (5.6%) 219
1 NA NA 0.14 0.25
0.9 NA NA NA 0.9
0.9 0.5 0.45 0.9 NA NA NA NA
NA, not available.
presenting problem and control group 2 was from patients who have had inpatient orthopedic procedures, including knee and hip surgery (either joint replacement or arthroscopy). This latter group was found to represent the DM prevalence of the population that the hospital covers (Fig. 1). Recent studies have shown that diabetics are less likely to develop aortic aneurysms, and the reported incidence of DM in patients with aortic aneurysmal disease is about 13% overall,79 which matches with our unpublished hospital data. This prevalence is much higher than that of diabetics developing TAD in this study.
Patients with DM are more likely to have hypertension, and blood pressure control is more difficult than in nondiabetics. There are several reasons for the higher incidence of hypertension in diabetics.4 Hypertension, particularly when uncontrolled, significantly increases the risk of aortic dissection. Further, DM is an acknowledged risk factor for atherosclerosis, which is a risk factor for aortic dissection.10 Thus, it has been reasonable to assume that DM is a risk factor for aortic dissection and it is generally believed that this is the case.5 However, the present study suggests otherwise. A similar prevalence rate (2e6%) of DM in TAD patients was
Vol. 28, No. 2, February 2014
Diabetes and TAD 431
Table III. Comparison of TAD groups A, B and A + B with controls 1 and 2 TAD groups/control types
Total
DM
Non-DM
Or (CI)
P
Control 1A Type A TAD (group A) Control 2A Control 1B Type B TAD (group B) Control 2B Control 1 Both A and B TAD (groups A + B) Control 2
1310 131 1310 880 88 880 2190 219 2190
241 3 116 228 2 102 469 5 218
1069 128 1194 652 86 778 1721 214 1972
0.1 (0.03e0.32) 0.24 (0.07e0.76)
0.0001 0.007
0.07 (0.02e0.27) 0.18 (0.04e0.73)
0.0001 0.0035
0.08 (0.06e0.20) 0.21 (0.08e0.51)
0.0001 0.0003
Table IV. TAD-related mortality on index admission
Mortality in group A (type A TAD) Mortality in group B (type B TAD) Mortality in groups A and B
DM
Non-DM
1 of 3 0 of 2 1 of 5
31 of 128 10 of 86 41 of 214
reported recently.1416 Although diabetics are more likely to develop atherosclerosis and hypertension, the low odds of DM in TAD indicates that DM may actually have a protective role against TAD. This may be due to the fact that diabetics are less likely to develop aortic root dilation as suggested in a recent Chinese study with echocardiographic measurement of the aortic root.17 The other possible explanation is the introduction of tighter blood pressure targets in diabetics and other secondary prevention efforts being more concerted in diabetics. The third possibility is that the biologic changes in the aortic wall produced by DM may be protective against aortic dissection. A recent study suggested that hyperglycemia associated with DM induced cross-linking of the collagen network in the aortic wall media, and this crosslinking resists proteolysis and inhibits secretion of the matrix metalloproteinases (MMPs) thought to mediate aortic aneurysm formation.6 Further, DM is also known to suppress plasmin, which activates the MMPs.18 These effects could decrease aortic wall degradation directly and may also explain the thicker abdominal aortic wall observed in diabetics.19 If it is proven that a protective benefit is imbued by hyperglycemia on collagen crosslinkage, then this may provide a new pharmacologic target for therapy, particularly in susceptible patients with connective tissue disorder. It is also possible that an unknown mechanism or combination of the aforementioned factors is responsible for this protective effect. However, it is
likely that DM has an either direct or indirect protective effect on aortic dissection. A recent study by the International Registry of Aortic Dissection (IRAD) compared the factors that may have an influence on aortic diameter expansion. Of the many assumed risk factors, DM had the least ‘‘risk’’ as an influencing factor on diameter expansion, with an OR of 0.9. Not surprisingly, presence of blood flow in the false lumen was the most influential factor determining aortic diameter growth.5 Although DM is a risk factor for most cardiovascular disease, with regard to TAD and aortic aneurysm development,8,9 DM instead appears to provide a protective effect. In this study we have included only patients who survived into hospitalization; if diabetic patients with TAD were unlikely to even survive to the emergency department, then this could skew the data. However, this neither stands to reason on hypothetical grounds nor is it supported by the finding that mortality rates in the diabetic and nondiabetic groups were only 20% and 19%, respectively, in our patients. Furthermore, to make the difference in TAD between diabetic and nondiabetic patients statistically insignificant, we calculated that this would have required a further 8 diabetic patients to have died before hospitalization, compared with zero nondiabetic patients. Because of the very low prevalence of DM in the TAD group, it is difficult to make valid statements about the role of ethnicity in TAD, although the analysis presented in Table II shows no correlation. Prakash et al. struggled to obtain complete ethnicity data, but they did find there was no difference in observed frequency of dissection between whites and nonwhites. In conclusion, contrary to current dogma, DM appears to have a protective effect against thoracic aortic dissection, although the exact mechanism is not yet known. Further studies are needed to elucidate the exact reason for this apparent protective effect of DM against aortic dissection if we are to
432 Theivacumar et al.
further understand and exploit any advantages in this group. REFERENCES 1. Report 2010, Diabetes UK, Key Statistics on Diabetics. http:// www.diabetes.org.uk/Documents/Reports/Diabetes_in_the_ UK_2010.pdf. 2. Report 2011, the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), National Institutes of Health (NIH). http://diabetes.niddk.nih.gov/dm/pubs/statistics/. 3. Creager MA, Loscalzo J. Diseases of the aorta. In: Fauci AS, Braunwald E, Kasper DL (eds), Harrison’s Principles of Internal Medicine, 17th ed. New York: McGraw-Hill. 4. Bakris GL. Maximizing cardiorenal benefit in the management of hypertension: achieve blood pressure goals. J Clin Hypertens (Greenwich) 1999;1:141e7. 5. Sueyoshi E, Sakamoto I, Hayashi K, et al. Growth rate of aortic diameter in patients with type B aortic dissection during the chronic phase. Circulation 2004;110(Suppl. 1):256e61. 6. Golledge J, Karan M, Moran CS, et al. Reduced expansion rate of abdominal aortic aneurysms in patients with diabetes may be related to aberrant monocyteematrix interactions. Eur Heart J 2008;29:665e72. 7. T€ ornwall ME, Virtamo J, Haukka JK, et al. Life-style factors and risk for abdominal aortic aneurysm in a cohort of Finnish male smokers. Epidemiology 2001;12:94e100. 8. Lederle FA. The strange relationship between diabetes and abdominal aortic aneurysm. Eur J Vasc Endovasc Surg 2012;43:254e6. 9. Lederle FA, Johnson GR, Wilson SE, et al. The Aneurysm Detection and Management Study screening program: vali-
Annals of Vascular Surgery
10. 11.
12.
13.
14.
15.
16.
17.
18.
19.
dation cohort and final results. Arch Intern Med 2000;160: 1425e30. LeMaire SA, Russell L. Epidemiology of thoracic aortic dissection. Nat Rev Cardiol 2011;8:103e13. Hennessy S, Bilker W, Berlin J, et al. Factors influencing the optimal control-to-case ratio in matched caseecontrol studies. Am J Epidemiol 1999;149:195e7. National and Primary Care Trust diabetic data. http://www. diabetes.org.uk/Documents/Reports/Diabetes_in_the_UK_ 2010. Prakash S, Pedroza C, Khalil Y, et al. Diabetes and Reduced Risk for Thoracic Aortic Aneurysms and Dissections: a nationwide caseecontrol study. J Am Heart Assoc 2012;1:1e8. Nienaber CA, Fattori R, Mehta RH, et al. Gender-related differences in acute aortic dissection. Circulation 2004;109: 3014e21. Trimarchi S, Eagle KA, Nienaber CA, et al. Role of age in acute type A aortic dissection outcome: report from the international Registry of Acute Aortic Dissection (iRAD). J Thorac Cardiovasc Surg 2010;140:784e9. Achneck H, Modi B, Shaw C, et al. Ascending thoracic aneurysms are associated with decreased systemic atherosclerosis. Chest 2005;128:1580e6. Jiang JJ, Chen XF, Liu XM, et al. Diabetes mellitus: is it protective against aortic root dilatation? Cardiology 2009;112:138e43. Dua MM, Miyama N, Azuma J, et al. Hyperglycemia modulates plasminogen activator inhibitor-1 expression and aortic diameter in experimental aortic aneurysm disease. Surgery 2010;148:429e35. Astrand H, Ryden-Ahlgren A, Sundkvist G, et al. Reduced aortic wall stress in diabetes mellitus. Eur J Vasc Endovasc Surg 2007;33:592e8.
Background: Diabetes mellitus (DM) is an acknowledged risk factor for atherosclerosis, and diabetics are more likely to have hypertension. Atherosclerosis and hypertension are risk factors for aortic dissection. However, recent studies have shown that DM is associated with changes in aortic wall collagen. In this retrospective study we assess the relationship between DM and thoracic aortic dissection (TAD). Methods: Patients with a diagnosis of thoracic aortic dissection during the last 10 years were identified from our hospital records. The prevalence of DM in Stanford type A and B TAD was compared with that of two age- and gender-matched control groups. For every diabetic dissection case, 10 controls were selected from the hospital data. Results: Two hundred nineteen patients (median age 61 years, male:female ratio 145:74) were identified with TAD, comprising 131 type A dissections and 88 type B dissections. Only 3 of 131 (2.3%) type A aortic dissections were diabetics, whereas, in control group 1, 241 of 1310 (18.4%) were diabetics and, in control group 2, 116 of 1310 (8.9%) were diabetics [odds ratios: 0.1 (0.03e0.32) and 0.24 (0.07e0.76), respectively] (P ¼ 0.0001 and 0.007, respectively). Similarly, only 2 of 88 (2.3%) type B aortic dissections were diabetics, whereas 228 of 880 (26.0%) and 102 of 880 (11.6%) were diabetics in groups 1 and 2 [odds ratios: 0.07 (0.02e0.27) and 0.18 (0.04e0.73), respectively] (P ¼ 0.0001 and 0.0035, respectively). All these odds ratios were statistically significant (P < 0.01). Conclusions: Patients with thoracic aortic dissection are less likely to be diabetic. Although we identified association only, not causality, it is possible that DM, or its treatment, has a protective effect against aortic dissection.
INTRODUCTION Diabetes mellitus (DM) is a common disease, affecting 5.5% of the overall population in England.1
Department of Vascular Surgery, King’s Health Partners, King’s College Hospital, London, United Kingdom. Correspondence to: Domenico Valenti, MD, Department of Vascular Surgery, King’s College Hospital, Denmark Hill, London SE5 9RS, UK; E-mail: [email protected] Ann Vasc Surg 2014; 28: 427–432 http://dx.doi.org/10.1016/j.avsg.2013.03.024 Ó 2014 Elsevier Inc. All rights reserved. Manuscript received: October 9, 2012; manuscript accepted: March 12, 2013; published online: December 20, 2013.
Varying incidence and prevalence rates have been reported among different populations and age groups. Its prevalence is higher with increasing age, affecting up to 30% of those >65 years of age in the United States.2 DM is an acknowledged risk factor for atherosclerosis.3 Further, diabetics are more likely to develop systemic hypertension and their blood pressure control is more complex, often requiring polypharmaceutical agents.4 DM has therefore generally been assumed to be a risk factor for thoracic aortic dissection (TAD).5 However, a recent study has shown that DM increases cross-linkage of collagen in the aortic wall, leading to biologic changes in the aortic wall.6 Also, studies have shown that diabetics are less likely to develop aortic aneurysmal disease.79 427
428 Theivacumar et al.
Annals of Vascular Surgery
In this study we assess the relationship between DM and aortic dissection.
METHODS All patients who were admitted to King’s College Hospital, London, with a diagnosis of aortic dissection during the period January 2002 to April 2012, were included. Cases were identified from the hospital electronic information records using the International Classification of Diseases, tenth revision (ICD-10). All patients with the diagnosis code I710 (aortic dissection) were selected. The search was conducted with a filter for hypertension and diabetes mellitus (including E100eE149 codes) to check which cases were diabetic (type 1, 2, or dietcontrolled). Individual case notes were reviewed for the exact diagnosis of Stanford classification10 and other risk factors, such as collagen disorders and bicuspid aortic valve, along with smoking history, hypertension, hypertension medications, coronary heart disease, peripheral vascular disease, and race. The cases were divided into two groups: group A (patients with Stanford type A dissection) and group B (patients with Stanford type B dissection). No single control group could perfectly represent the population from which the dissection patients were drawn. Therefore, for purposes of comparison, we created two control groups. Control group 1 comprised all those admitted to the hospital for any reason. Control group 2 consisted of patients admitted for any inpatient orthopedic procedure. Age- and gender-matched control groups were selected randomly from these groups. For every case, 10 age- and gender-matched controls were randomly selected from the hospital database from the same year of admission as the individual cases. Ten controls were chosen because we had the luxury of large control groups, and ratios >5 have been shown to be advantageous, particularly when there is a low prevalence of exposure among controls.11 To assess matched controls with other comorbidities would have inevitably been highly problematic given the complex interactions between DM and other risk factors for TAD; therefore, the higher ratio of age- and gender-matched controls would help solve this problem. The corresponding controls for group A were controls 1A and 2A and for group B were controls 2A and 2B. The odds ratio of DM in group A (TAD type A) was compared with that of both controls 1A and 2A. Similarly, group B (TAD type B) was compared with 1B and 2B. Finally, both TAD (groups A and
Fig. 1. Proportion of DM cases by age in the control groups, the community from which our patients are drawn, and the TAD group.
B) groups were compared with controls 1 and 2. The prevalence of DM in the TAD group and the control groups was also compared with the community data, obtained from the southeastern England population from which our hospital receives patients. These data were obtained from published key statistics on DM in 2010.12 The percentage prevalence of DM was calculated in each age group along with male:female proportions. We did not distinguish between acute or chronic dissection because we were interested in the development of thoracic aortic dissection per se, rather than the stage of the presentation. Statistical Methods The prevalence of DM among the study groups (patients with type A and B aortic dissection) and controls 1 and 2 were compared. Group comparisons were made using nonparametric statistical testing. Odds ratios (ORs) were obtained, and the chi-squared test or Fisher’s exact test was used to assess the statistical significance of the OR. Confidence intervals (95%) were also obtained (expressed in parentheses in what follows). Age is given presented median years (range).
RESULTS We first established how well our control group matched that of the community prevalence with regard to DM. The percentages of DM in the TAD group and control groups 1 and 2 were compared in different age groups, and also compared with the community diabetic prevalence (Fig. 1). It is worth noting that control group 2 (those who had elective orthopedic procedures) had a very similar prevalence distribution of DM to the community
Vol. 28, No. 2, February 2014
Diabetes and TAD 429
Table I. Demographic details of cases and controls Cases/controls
Median age (range)
Male:female ratio
Total
TAD type A (group A) TAD type B (group B) TAD (A and B) Age- and gender-matched from control 1 (all admissions) Age- and gender-matched from control 2 (patients who had orthopedic procedures)
58 63 61 61 61
87:45 58:30 145:74 1450:740 1450:740
131 88 219 2190 2190
data, unlike control 1 (all hospital admission) in which diabetics were overrepresented (Fig. 1). During the last 10 years, there were 219 patients with a median age of 61 (range 14e92) years presenting with TAD, who were admitted to King’s College Hospital, London. The demographic details are given in Table I. Their breakdown of dissection type and risk factors are given in Table II. Only 3 of 131 (2.2%) type A aortic dissections were diabetics in group A, whereas 241 of 1310 (18.4%) and 116 of 1310 (10.2%) were diabetics in control groups 1 and 2 [OR: 0.1 (0.03e0.32) and 0.24 (0.07e0.76), respectively], which was highly statistically significant (P ¼ 0.0001 and P ¼ 0.007, respectively). Similarly, only 2 of 88 (2.3%) type B aortic dissections were diabetics in group B, whereas 228 of 880 (26.0%) and 102 of 880 (11.6%) were diabetics in groups 1 and 2 [OR: 0.07 (0.02e0.27) and 0.18 (0.04e0.73), respectively] (P ¼ 0.0001 and P ¼ 0.0035, respectively). When all aortic dissections (both A and B) were assessed together as one group, 5 of 214 cases (2.3%) were diabetic, whereas 469 of 2190 (21.4%) and 218 of 2190 (10.0%) were diabetic in control groups 1 and 2 [OR: 0.08 (0.06e0.2) and 0.21 (0.08e0.51), respectively] (P ¼ 0.0001 and P ¼ 0.0003, respectively). These data are also summarized in Table III. All diabetic patients who had TAD (both type A and B) were hypertensive, whereas, among nondiabetics, 82 of 128 (64%) with type A and 69 of 86 (80%) with type B aortic dissection were hypertensive. We found no significant difference in blood pressure levels or numbers of antihypertensives between the diabetic and nondiabetic groups (see Table II). Collagen diseases (Marfan’s syndrome or EhlerseDanlos syndrome) were found in 4 patients in group A and 2 in group B, all of whom were nondiabetics. Bicuspid aortic valve was a risk factor in 2 patients (nondiabetics) in group A, whereas this was not found in group B. No patient had a known history of illicit drug use. When we compared the mortality associated with TAD, 1 diabetic patient died due to TAD
(14e76) (42e92) (14e92) (14e92) (14e92)
compared with 41 deaths among nondiabetics during the last 10 years. Because the mortality numbers were very small in the diabetic group, it was difficult to prove whether diabetics were more or less likely to die from TAD compared with nondiabetics (Table IV).
DISCUSSION This study suggests that diabetic patients are significantly less likely to have aortic dissection, yet it is known from previous studies that diabetics suffer more frequently from hypertension. In our study, however, there was no significant difference between the diabetic and nondiabetic groups in terms of blood pressure levels or numbers of antihypertensive medications.4 High-velocity spiral flow in the proximal aorta constantly challenges the aortic wall. The curved morphology of the aortic arch places the convex aortic wall at persistent risk of injury during each systolic phase. Patients with collagen disorders are therefore at high risk of aortic dissection as they have poor aortic wall integrity. Hypertension is a well-accepted risk factor for aortic dissection.10 Patients with bicuspid aortic valve are likely to produce alternative turbulent blood flow patterns in the proximal aorta, leading to a higher risk of dissection.10 One recent study indicated a similar finding.13 Prakash et al. used a similar epidemiologic approach using national statistics in the United States over a 2year period, with randomly selected controls. They found that DM was associated with a 40e80% reduction in hospitalization due to TAD. They found this to be the case in patients with uncomplicated DM and those with chronic complicated DM, although the negative association was greater with the latter. In this study, the odds ratios of DM in the TAD groups were compared with those of two different control groups who were age- and gendermatched on a case-by-case basis. Control group 1 was taken from all hospital admissions for any
430 Theivacumar et al.
Annals of Vascular Surgery
Table II. Types of TAD and their risk factors in the diabetic and nondiabetic groups Thoracic aortic dissection type (TAD)
Type A Total type A TAD Risk factors Hypertension Systemic blood pressure on admission (median mm Hg) Number of antihypertensives (mode) Collagen disorders Bicuspid valve Hyperlipidemia Median body mass index Coronary heart disease Peripheral vascular disease History of current smoking White British Black African Race Black Caribbean Asian/Chinese Others/nonspecified Type B Total type B TAD Risk factors Hypertension Systemic blood pressure on admission (median mm Hg) Number of antihypertensives (mode) Collagen disorders Bicuspid valve Hyperlipidemia Median body mass index Coronary heart disease Peripheral vascular disease History of current smoking White British Black African Black Caribbean Race Asian/Chinese Others/nonspecified Both type A and B TAD
DM
Non-DM
Total (n)
P
3
128
131
d
3 (100%) 146/90 2 0 0 2 (66.6%) 29.4 1 (33.3%) 0 0 1 (33.3%) 1 (33.3%)
82 (64.1%) 150/88 2 4 (3.1%) 3 (2.3%) 54 (42.2%) 29.6 34 (26.6%) 11 (8.6) 22 (17.2) 98 (76.6%) 11 (8.6%)
85 (42.0%) 150/88 2 4 (3.1%) 3 (2.3%) 56 (42.7%) 29.7 35 (26.7%) 11 (8.4%) 22 (16.8%) 99 (75.6%) 12 (92.4%)
0.55 NA NA NA NA 0.57
1 (33.3%) 0 0
13 (10.1%) 2 (1.6%) 4 (3.1%)
14 (10.7%) 2 (1.5%) 4 (3.1%)
0.29 NA NA
2
86
88
2 (100%) 136/82 2 0 0 2 (100%)
69 (80.2%) 140/84 1 2 (2.3%) 0 62 (72.1%)
71 (80.7%) 140/84 1 2 (2.3%) 0 64 (72.7%)
28.5 2 (100%) 1 (50%) 1 (50%) 2 (100%) 0 0
29.2 49 (57.0%) 22 (25.6%) 24 (28.0%) 62 (72.1%) 9 (10.5%) 8 (9.3%)
29.1 51 (58.0%) 23 (26.1%) 25 (28.4%) 64 (72.7%) 9 (10.2%) 8 (9.1%)
0 0 5
2 (2.3%) 5 (5.8%) 214
2 (2.3%) 5 (5.6%) 219
1 NA NA 0.14 0.25
0.9 NA NA NA 0.9
0.9 0.5 0.45 0.9 NA NA NA NA
NA, not available.
presenting problem and control group 2 was from patients who have had inpatient orthopedic procedures, including knee and hip surgery (either joint replacement or arthroscopy). This latter group was found to represent the DM prevalence of the population that the hospital covers (Fig. 1). Recent studies have shown that diabetics are less likely to develop aortic aneurysms, and the reported incidence of DM in patients with aortic aneurysmal disease is about 13% overall,79 which matches with our unpublished hospital data. This prevalence is much higher than that of diabetics developing TAD in this study.
Patients with DM are more likely to have hypertension, and blood pressure control is more difficult than in nondiabetics. There are several reasons for the higher incidence of hypertension in diabetics.4 Hypertension, particularly when uncontrolled, significantly increases the risk of aortic dissection. Further, DM is an acknowledged risk factor for atherosclerosis, which is a risk factor for aortic dissection.10 Thus, it has been reasonable to assume that DM is a risk factor for aortic dissection and it is generally believed that this is the case.5 However, the present study suggests otherwise. A similar prevalence rate (2e6%) of DM in TAD patients was
Vol. 28, No. 2, February 2014
Diabetes and TAD 431
Table III. Comparison of TAD groups A, B and A + B with controls 1 and 2 TAD groups/control types
Total
DM
Non-DM
Or (CI)
P
Control 1A Type A TAD (group A) Control 2A Control 1B Type B TAD (group B) Control 2B Control 1 Both A and B TAD (groups A + B) Control 2
1310 131 1310 880 88 880 2190 219 2190
241 3 116 228 2 102 469 5 218
1069 128 1194 652 86 778 1721 214 1972
0.1 (0.03e0.32) 0.24 (0.07e0.76)
0.0001 0.007
0.07 (0.02e0.27) 0.18 (0.04e0.73)
0.0001 0.0035
0.08 (0.06e0.20) 0.21 (0.08e0.51)
0.0001 0.0003
Table IV. TAD-related mortality on index admission
Mortality in group A (type A TAD) Mortality in group B (type B TAD) Mortality in groups A and B
DM
Non-DM
1 of 3 0 of 2 1 of 5
31 of 128 10 of 86 41 of 214
reported recently.1416 Although diabetics are more likely to develop atherosclerosis and hypertension, the low odds of DM in TAD indicates that DM may actually have a protective role against TAD. This may be due to the fact that diabetics are less likely to develop aortic root dilation as suggested in a recent Chinese study with echocardiographic measurement of the aortic root.17 The other possible explanation is the introduction of tighter blood pressure targets in diabetics and other secondary prevention efforts being more concerted in diabetics. The third possibility is that the biologic changes in the aortic wall produced by DM may be protective against aortic dissection. A recent study suggested that hyperglycemia associated with DM induced cross-linking of the collagen network in the aortic wall media, and this crosslinking resists proteolysis and inhibits secretion of the matrix metalloproteinases (MMPs) thought to mediate aortic aneurysm formation.6 Further, DM is also known to suppress plasmin, which activates the MMPs.18 These effects could decrease aortic wall degradation directly and may also explain the thicker abdominal aortic wall observed in diabetics.19 If it is proven that a protective benefit is imbued by hyperglycemia on collagen crosslinkage, then this may provide a new pharmacologic target for therapy, particularly in susceptible patients with connective tissue disorder. It is also possible that an unknown mechanism or combination of the aforementioned factors is responsible for this protective effect. However, it is
likely that DM has an either direct or indirect protective effect on aortic dissection. A recent study by the International Registry of Aortic Dissection (IRAD) compared the factors that may have an influence on aortic diameter expansion. Of the many assumed risk factors, DM had the least ‘‘risk’’ as an influencing factor on diameter expansion, with an OR of 0.9. Not surprisingly, presence of blood flow in the false lumen was the most influential factor determining aortic diameter growth.5 Although DM is a risk factor for most cardiovascular disease, with regard to TAD and aortic aneurysm development,8,9 DM instead appears to provide a protective effect. In this study we have included only patients who survived into hospitalization; if diabetic patients with TAD were unlikely to even survive to the emergency department, then this could skew the data. However, this neither stands to reason on hypothetical grounds nor is it supported by the finding that mortality rates in the diabetic and nondiabetic groups were only 20% and 19%, respectively, in our patients. Furthermore, to make the difference in TAD between diabetic and nondiabetic patients statistically insignificant, we calculated that this would have required a further 8 diabetic patients to have died before hospitalization, compared with zero nondiabetic patients. Because of the very low prevalence of DM in the TAD group, it is difficult to make valid statements about the role of ethnicity in TAD, although the analysis presented in Table II shows no correlation. Prakash et al. struggled to obtain complete ethnicity data, but they did find there was no difference in observed frequency of dissection between whites and nonwhites. In conclusion, contrary to current dogma, DM appears to have a protective effect against thoracic aortic dissection, although the exact mechanism is not yet known. Further studies are needed to elucidate the exact reason for this apparent protective effect of DM against aortic dissection if we are to
432 Theivacumar et al.
further understand and exploit any advantages in this group. REFERENCES 1. Report 2010, Diabetes UK, Key Statistics on Diabetics. http:// www.diabetes.org.uk/Documents/Reports/Diabetes_in_the_ UK_2010.pdf. 2. Report 2011, the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), National Institutes of Health (NIH). http://diabetes.niddk.nih.gov/dm/pubs/statistics/. 3. Creager MA, Loscalzo J. Diseases of the aorta. In: Fauci AS, Braunwald E, Kasper DL (eds), Harrison’s Principles of Internal Medicine, 17th ed. New York: McGraw-Hill. 4. Bakris GL. Maximizing cardiorenal benefit in the management of hypertension: achieve blood pressure goals. J Clin Hypertens (Greenwich) 1999;1:141e7. 5. Sueyoshi E, Sakamoto I, Hayashi K, et al. Growth rate of aortic diameter in patients with type B aortic dissection during the chronic phase. Circulation 2004;110(Suppl. 1):256e61. 6. Golledge J, Karan M, Moran CS, et al. Reduced expansion rate of abdominal aortic aneurysms in patients with diabetes may be related to aberrant monocyteematrix interactions. Eur Heart J 2008;29:665e72. 7. T€ ornwall ME, Virtamo J, Haukka JK, et al. Life-style factors and risk for abdominal aortic aneurysm in a cohort of Finnish male smokers. Epidemiology 2001;12:94e100. 8. Lederle FA. The strange relationship between diabetes and abdominal aortic aneurysm. Eur J Vasc Endovasc Surg 2012;43:254e6. 9. Lederle FA, Johnson GR, Wilson SE, et al. The Aneurysm Detection and Management Study screening program: vali-
Annals of Vascular Surgery
10. 11.
12.
13.
14.
15.
16.
17.
18.
19.
dation cohort and final results. Arch Intern Med 2000;160: 1425e30. LeMaire SA, Russell L. Epidemiology of thoracic aortic dissection. Nat Rev Cardiol 2011;8:103e13. Hennessy S, Bilker W, Berlin J, et al. Factors influencing the optimal control-to-case ratio in matched caseecontrol studies. Am J Epidemiol 1999;149:195e7. National and Primary Care Trust diabetic data. http://www. diabetes.org.uk/Documents/Reports/Diabetes_in_the_UK_ 2010. Prakash S, Pedroza C, Khalil Y, et al. Diabetes and Reduced Risk for Thoracic Aortic Aneurysms and Dissections: a nationwide caseecontrol study. J Am Heart Assoc 2012;1:1e8. Nienaber CA, Fattori R, Mehta RH, et al. Gender-related differences in acute aortic dissection. Circulation 2004;109: 3014e21. Trimarchi S, Eagle KA, Nienaber CA, et al. Role of age in acute type A aortic dissection outcome: report from the international Registry of Acute Aortic Dissection (iRAD). J Thorac Cardiovasc Surg 2010;140:784e9. Achneck H, Modi B, Shaw C, et al. Ascending thoracic aneurysms are associated with decreased systemic atherosclerosis. Chest 2005;128:1580e6. Jiang JJ, Chen XF, Liu XM, et al. Diabetes mellitus: is it protective against aortic root dilatation? Cardiology 2009;112:138e43. Dua MM, Miyama N, Azuma J, et al. Hyperglycemia modulates plasminogen activator inhibitor-1 expression and aortic diameter in experimental aortic aneurysm disease. Surgery 2010;148:429e35. Astrand H, Ryden-Ahlgren A, Sundkvist G, et al. Reduced aortic wall stress in diabetes mellitus. Eur J Vasc Endovasc Surg 2007;33:592e8.