Introduction

The aorta connects the heart with the peripheral organs and plays a central role in the cardiovascular system. Aortic conditions such as thoracic or abdominal aneurysm and dissection are common and progress over time, often becoming life-threatening with a need for elective or emergency therapy [1–3]. Because these conditions are often associated with an increase in aortic diameter, the latter is an important parameter in deciding when and how to treat these patients [1]. Moreover, there is evidence that an increased baseline diameter of the infrarenal aorta is a strong and independent risk factor for the development of abdominal aortic aneurysm [4]. Therefore, it is crucial to have reference values for the different aortic segments.

Regardless of the imaging modality used (CT or MRI), the identification of reference values is hampered by the fact that aortic diameter depends on many physiologic factors including age, sex and body surface area (BSA) and is affected by a number of cardiovascular risk factors [5–7]. A positive correlation of aortic diameter and age is well established. In contrast, the association of other risk factors for cardiovascular disease (CVD) or morphologic pathologies of the aorta with aortic diameters and the cumulative and interactive effects of multiple risk factors on different aortic segments are not fully understood [5, 6, 8–10]. In addition, further evidence is also needed with regard to how these factors affect different aortic segments. Cross-sectional imaging using magnetic resonance imaging (MRI) allows visualization of the entire aorta and reliable measurement of diameters at different levels [11, 12]. However, studies analysing aortic diameter by MRI are sparse and have some limitations such as small number of subjects, highly selected study populations or incomplete imaging of the aorta [13–15].

The aim of this study was to provide reference values for thoracic and abdominal aortic diameters derived by MRI and to evaluate associations with age and other cardiovascular risk factors in a large general population.

Methods

Study sample

A subsample of participants of the Study of Health in Pomerania (SHIP-TREND), who underwent a whole-body magnetic resonance imaging (WBMRI) examination between 2008 and 2012 (n = 1759; 872 women), were included in this study. SHIP-TREND is a cross-sectional, population-based study conducted in the northeast region of Germany. A sample of 8826 adults (20–79 years) was drawn from local population registries [16], and 4420 subjects volunteered for baseline examinations. Exclusion criteria were non-WBMRI examination (n = 2373) and missing data on aortic diameters (n = 35). Furthermore, subjects with self-reported stroke (n = 26) and myocardial infarction (n = 22) were excluded from analysis of aortic diameters. Other exclusion criteria were aortic pathologies such as thoracic (≥5 cm) or abdominal (≥3.5 cm) aneurysm and aortic dissection (n = 7). A total of 1759 participants were analysed in this study. All participants provided written informed consent. The study was approved by the Ethics Committee of University of Greifswald and complies with the Declaration of Helsinki.

MRI examination and aortic diameter measurement

WBMRI was performed on a 1.5-T MRI scanner (Magnetom Avanto; Siemens Healthcare, Erlangen, Germany). Imaging was performed using integrated coil elements and phased-array surface coils. Aortic diameters were measured on plain axial 3D T1-weighted volumetric interpolated breath-hold examination (VIBE) images. Imaging parameters for the thoracic aorta (one slab) were field of view 450 × 360 mm, TR/TE 3.1/1.1 ms, flip angle 8°, voxel size 1.8 × 1.8 × 3.0 mm, scan time 21 s and slices per scan 96. Imaging parameters for the abdominal aorta (two slabs) were field of view 450 × 360 mm, TR/TE 7.5/2.4 ms, flip angle 10°, voxel size 2.4 × 1.6 × 4.0 mm, scan time 38 s and slices per scan 96. Slice thickness was always 1.5 mm with a gap of 0.3 mm.

The outer diameters of six predefined aortic segments were measured: the ascending and descending aorta (level of the pulmonary trunk), the aortic arch (proximal to the origin of the left subclavian artery), the subdiaphragmatic aorta (level of the aortic hiatus), and the supra- and infrarenal aorta (1 cm above/below the right renal artery origin; Fig. 1). Diameters were measured on axial slices in coronal orientation from outer wall to outer wall using the OsiriX image viewing and processing software (version 3.6.1; Pixmeo Sarl, Bernex, Switzerland). Diameter measurements were carried out independently by two observers (LH, MW) and are based on either one reader’s measurement. The readers were blinded to other individual data. The method of measurement including intra- and interobserver agreement was validated in a previous study using orthogonal contrast-enhanced magnetic resonance angiography [12]. If at least one diameter measurement was missing, the whole data set was excluded from further analysis.

Fig. 1
figure 1

T1-weighted VIBE images illustrating diameter measurement in a 73-year-old male volunteer: a ascending and descending aorta, b aortic arch, c subphrenic aorta, d suprarenal aorta, e infrarenal aorta

Full size image

Before starting diameter measurements in a data set, each observer separately rated the image quality of each segment to be measured as sufficient (clear delineation of the outer aortic wall from perivascular tissue without significant artefacts (breathing, cardiac movement) hindering diameter measurement) or insufficient (significant blurring or artefacts obscuring the aortic wall). If at least one segment of the aorta was rated as insufficient the entire data set was discarded. The criteria for sufficient image quality were very strict.

Aortic diameters were analysed unadjusted and adjusted for body surface area (BSA). BSA was calculated according to the Du Bois formula [17]: (BSA = 0.007184 × (height in cm)0.725 × (weight in kg)0.425.

Risk factor measurement

Methods for measurement of baseline characteristics in SHIP-TREND have been described elsewhere [16]. Besides age and sex, other factors considered to potentially affect aortic diameter included smoking status, blood pressure (BP), HbA1c, low-density and high-density lipoprotein cholesterol (LDL-C and HDL-C) and triglyceride. Interview-assessed smoking status was categorized as never, former or current smoker. Systolic and diastolic BP were measured at the right arm of seated subjects after a 5-min rest period during the core examination. The mean of the second and third measurement was used for the present analysis. Blood samples were taken from each volunteer in the supine position between 07.00 a.m. and 04.00 p.m. and were analysed immediately.

Statistical analysis

Medians (25th and 75th percentiles) and absolute numbers (percentages) were used to summarize baseline characteristics of the male and female study sample of SHIP-TREND.

Different percentiles (5th, 25th, 50th, 75th and 95th) of aortic diameters were calculated for 10-year age groups and separately for women and men. Differences in median diameter between women and men were tested for significance using quantile regression. Age- and sex-specific reference values for BSA-adjusted aortic diameters were provided by estimated intercepts and β coefficients using quantile regression for the median and for 5th and 95th percentiles. Exemplarily, reference values were calculated for a 51-year-old man (representing median age). Additionally, the 5th and 95th percentile reference limits according to age were presented graphically using fractional polynomial regression models [18, 19].

Associations between BSA-adjusted aortic diameter and risk factors were assessed using multivariable linear regression models, and β coefficients were provided. Sex, age, smoking status, systolic and diastolic BP, HbA1c, HDL-C, LDL-C and triglycerides were considered as potential risk factors. Adjusted R 2 and partial R 2 were calculated to evaluate the fit of the model and the contribution of each risk factor. The assumption of linearity of the association between risk factors and BSA-adjusted aortic diameter was checked visually by comparing residual distributions with the normal distribution and additionally using multivariable regression spline models [20]. Interaction effects between all analysed risk factors were additionally tested. A value of p < 0.05 was considered statistically significant.

Statistical analysis was performed using Stata 12.1 (Stata Corporation, College Station, TX, USA).

Results

In the study sample, women (n = 872; 50 %) had a median age of 53 years, 22 % of them were current smokers, 8 % had diabetes and 38 % a history of hypertension. Men (n = 887) had a similar median age of 51 years, included 25 % smokers and 8 % diabetics, but had a much higher prevalence of hypertension of 49 % compared to women. Further baseline characteristics of the study sample are summarized in Table 1.

Table 1 Baseline characteristics of the study sample
Full size table

A total of 198 participants were excluded because of insufficient image quality.

Reference values for aortic diameters

The unadjusted median diameters of the different aortic segments were as follows: ascending aorta (3.20 cm for women, 3.49 cm for men), aortic arch (2.73 cm, 2.93 cm), descending aorta (2.34 cm, 2.63 cm), subphrenic aorta (2.22 cm, 2.46 cm), suprarenal aorta (2.07 cm, 2.34 cm) and infrarenal aorta (1.75 cm, 1.97 cm) with a relative reduction of 45 % for women and 44 % for men (from ascending to infrarenal aorta). Each median aortic diameter was lower in women compared to men (p < 0.001 for all aortic segments) with the relative reduction ranging between 7 % (aortic arch) and 12 % (suprarenal) (Tables 2 and 3).

Table 2 Age- and sex-specific percentiles of thoracic aortic diameter (cm) in the study sample
Full size table
Table 3 Age- and sex-specific percentiles of abdominal aortic diameter (cm) in the study sample
Full size table

Values for 5th percentile, median and 95th percentile increase with the 10-year age group.

Reference values for BSA-adjusted aortic diameters

Parameters for calculation of reference values for BSA-adjusted aortic diameters are presented in Table 4. The significant increase in diameter (p < 0.001) with each single year of age is similar for women and men for the median, 5th percentile and 95th percentile and varies between the lowest for the female infrarenal aorta (βp50 = 0.005, βp5 = 0.004, βp95 = 0.006) and the highest for the female ascending aorta (βp50 = 0.011, βp5 = 0.008, βp95 = 0.014), whereas the increase at the 95th percentile is always higher than at the 5th percentile (Table 4, Figs. 2 and 3).

Table 4 Association of age with body surface area-adjusted thoracic and abdominal aortic diameters and parameters for calculation of reference values based on the study sample
Full size table
Fig. 2
figure 2

Age-specific distributions of BSA-adjusted diameters of the thoracic ascending, the aortic arch and the descending aorta. The diameters are given as mean values with 5th and 95th percentiles for women and men. The values were calculated using fractional polynomial regression models

Full size image
Fig. 3
figure 3

Age-specific distributions of BSA-adjusted diameters of the abdominal subphrenic, suprarenal and infrarenal aorta. The diameters are given as mean values with 5th and 95th percentiles for women and men. The values were calculated using fractional polynomial regression models

Full size image

Cardiovascular risk factors and BSA-adjusted aortic diameters

For the ascending aorta (β = −0.049, p < 0.001), the aortic arch (β = −0.061, p < 0.001) and the subphrenic aorta (β = −0.018, p = 0.004), the BSA-adjusted diameters were lower in men than in women. There were no sex differences for the descending and the suprarenal aorta, while for the infrarenal aorta the diameter was higher in men than in women (β = 0.013, p = 0.013; Table 5).

Table 5 Cardiovascular risk factor model for body surface area-adjusted thoracic and abdominal aortic diameters
Full size table

As with unadjusted association, multivariable-adjusted association revealed significant increases in BSA-adjusted diameters of all six investigated aortic segments with age (p < 0.001 for each aortic segment). Current smoking was positively associated with the diameter of the descending, subphrenic, suprarenal and infrarenal aorta but not with the diameter of the ascending aorta and the aortic arch (Table 5). Consistent results for all aortic segments were observed for the positive associations of diastolic BP and HDL-C with BSA-adjusted aortic diameters and for the inverse association of systolic BP with aortic diameters. HbA1c and LDL-C were not associated with aortic diameters except for the subphrenic aorta (HbA1c: β = −0.008, p = 0.041) and the infrarenal aorta (LDL-C: β = −0.005, p = 0.042) with borderline significance. In sensitivity analysis, further adjustment for lipid-lowering medication did not substantially alter the results regarding HDL-C and LDL-C.

Furthermore, a higher triglyceride level was identified to be a potential risk factor for smaller aortic diameter, e.g. of the descending aorta (β = −0.011, p < 0.001; Table 5).

The diameters of the descending aorta (adjusted R 2 = 0.56) and the subphrenic aorta (R 2 = 0.59) were most strongly affected by the cardiovascular risk factors investigated, while the aortic arch (R 2 = 0.35) and infrarenal aorta (R 2 = 0.38) were least affected by these factors.

Interactions

A multivariable-adjusted model revealed a statistically significant interaction between sex and age with respect to the diameter of the ascending aorta. Age was more strongly associated with aortic diameter in women compared to men (women: β = 0.012, 95 % CI 0.011–0.013, p < 0.001; men: β = 0.010, 95 % CI 0.009–0.011, p < 0.001 for interaction) HbA1c was inversely associated with the subphrenic aortic diameter in women (β = −0.016, 95 % CI −0.029 to −0.002, p = 0.022) but not in men. Interactions between HDL-C or LDL-C with lipid-lowering medication were not significant.

Discussion

This is the first study presenting MRI-based age- and sex-specific reference diameters for the thoracic and abdominal aorta derived in an unselected European population. Additionally, the BSA-adjusted aortic diameters of women and men were compared and their associations with cardiovascular risk factors were evaluated. Our results suggest that the median aortic diameter decreases from ascending to infrarenal aorta for women and men and that the median diameters of all aortic segments are lower in women than men, supporting the findings of earlier studies [5, 6, 8, 14, 21]. However, most studies used CT [5, 6, 8], analysed only the thoracic aorta [5, 8, 15] or investigated a small or highly selected study sample [14, 15]. In contrast, our study assessed the whole aorta using MRI in a large unselected population.

Moreover, the median aortic diameters of all six segments as well as the 5th and 95th percentiles increased with age group. The increase in aortic diameter with age for both sexes is a well-known fact [7, 8, 21, 22]. In a study of a North-American population, Rogers et al. found an annual increase in the diameter of the ascending aorta of 0.016 cm in women and 0.02 cm in men. The results for the descending and infrarenal aorta were 0.016 cm/0.019 cm and 0.009 cm/0.013 cm, respectively [6]. Kälsch et al. presented a similar analysis for a European population, demonstrating an increase in diameter for women and men of 0.015 cm/0.015 cm per year (ascending aorta) and 0.016 cm/0.017 cm (descending aorta) [5]. We analysed the association of age with median aortic diameter for both the thoracic and abdominal aorta after adjustment for BSA. The significant increase in aortic diameter with each single year of age was similar for women and men for the median, 5th percentile and 95th percentile. The associations between age and unadjusted diameters varied between different aortic segments. The ascending aorta showed a stronger association between age and diameter for women (0.011 cm/year) compared to men (0.010 cm/year), while the association between age and the diameter of the infrarenal aorta was the same for both sexes (0.005 cm/year). This result is not surprising in view of the different functions of the aortic segments. The ascending aorta is a conduit but also has a cushion function, ensuring continuous blood perfusion of the peripheral organs [23]. The mechanical stress with rapidly alternating wall tension during the cardiac cycle, which contributes significantly to the aortic enlargement, is therefore much higher for the ascending aorta compared to the infrarenal segment. The increase in diameter with age for each segment was always higher at the 95th percentile than at the 5th percentile for both sexes, indicating that the diameter of the aorta is increasing faster with larger diameters. The Norwegian population-based Tromsø Study, which focussed on the infrarenal aorta, found similar results [24]. The faster increase of larger aortic diameters can be explained by Laplace’s law, predicting that an increasing vessel diameter leads to an increase in wall tension, resulting in a further increase in aortic diameter [25].

Our multivariable model for the assessment of cardiovascular risk factors and BSA-adjusted aortic diameters identified an inverse association with male sex for the ascending aorta, the aortic arch and the subphrenic aorta, whereas only the infrarenal aorta (p = 0.0013) revealed a positive association. In line with our results, Kälsch et al. found a significantly greater diameter for the ascending and descending thoracic aorta after BSA adjustment for women [5]. Furthermore, diastolic BP was positively and systolic BP was slightly inversely associated with diameters for all aortic segments investigated in our study. Kälsch et al. found a greater positive association for diastolic BP (β = 0.05–0.08 for women and men, per 10 mmHg) compared to systolic BP (β = 0.03–0.04) with the thoracic aortic diameter [5]. Rogers et al. showed a slightly weaker positive correlation for abdominal aortic diameter with systolic BP compared to diastolic BP [6]. However, the studied populations were probably older and both results refer to absolute diameter measurements in contrast to our BSA-adjusted results.

In a sensitivity analysis of a subgroup comparable to the populations investigated in the latter studies (aged 45–74 years, using absolute diameters), we observed similar results for systolic and diastolic BP in univariate models and for diastolic BP in multivariable-adjusted models. However, the inverse association between systolic BP and aortic diameters revealed by the multivariable-adjusted models (including diastolic BP) was consistent in these subanalyses.

The positive association of HDL-C with the diameters of all aortic segments in our study appeared paradoxical. Rogers et al. showed a predominantly inverse correlation of thoracic and abdominal aortic diameters with HDL-C [6]. When considering that an increase in aortic diameter is part of a vascular aging process with consecutive atherosclerosis and vascular dilatation and that HDL-C protects against this sequel, it would be more reasonable to find an inverse association [26]. On the other hand, LDL-C and triglycerides show more of an inverse association with aortic diameter, which points to the pathophysiologic role of HDL-C as an antagonist. However, recent data suggest that not only the amount of HDL-C in blood is critical but also its function [27].

The major strength of this study is its population-based setting including a large number of participants and the additional recording of comprehensive clinical and laboratory data. Another advantage is that the diameters of the thoracic as well as the abdominal aorta were assessed using a previously validated technique of aortic diameter measurement including radiation-free image acquisition. A limitation is that only a subgroup of the whole SHIP study population was examined by MRI. Another drawback is that comparability of our results with findings reported by other groups is limited. Aortic diameter measurement in volunteers using MRI is less widespread than the use of CT/electron beam computed tomography. Finally, information on diseases which might influence aortic diameter (large vessel arteritis, bicuspid aortic valve or connective tissue diseases) was not available for our study participants. All subjects included in our study were of European descent. Results for other ethnicities and possible ethnic variations remain to be established.

In conclusion, our study presents MRI-based reference values for the diameter of the thoracic and abdominal aorta in a general population. The median aortic diameter shows a positive association with male sex and age, though the association with sex is partially reversed after BSA adjustment. In addition, our results demonstrate that some cardiovascular risk factors such as systolic and diastolic BP, smoking and HDL-C are associated with thoracic and abdominal aortic diameters.