Quantification of the regional bioarchitecture in the human aorta

Abstract

Regional variance in human aortic bioarchitecture responsible for the elasticity of the vessel is poorly understood. The current study quantifies the elements responsible for aortic compliance, namely, elastin, collagen and smooth muscle cells, using histological and stereological techniques on human tissue with a focus on regional heterogeneity. Using donated cadaveric tissue, a series of samples were excised between the proximal ascending aorta and the distal abdominal aorta, for five cadavers, each of which underwent various staining procedures to enhance specific constituents of the wall. Using polarised light microscopy techniques, the orientation of collagen fibres was studied for each location and each tunical layer of the aorta. Significant transmural and longitudinal heterogeneity in collagen fibre orientations were uncovered throughout the vessel. It is shown that a von Mises mixture model is required accurately to fit the complex collagen fibre distributions that exist along the aorta. Additionally, collagen and smooth muscle cell density was observed to increase with increasing distance from the heart, whereas elastin density decreased. Evidence clearly demonstrates that the aorta is a highly heterogeneous vessel which cannot be simplistically represented by a single compliance value. The quantification and fitting of the regional aortic bioarchitectural data, although not without its limitations, including mean cohort age of 77.6 years, facilitates the development of next‐generation finite element models that can potentially simulate the influence of regional aortic composition and microstructure on vessel biomechanics.

The microarchitecture responsible for compliance of the human aorta is quantified and shown to be highly diverse regionally. The orientation of collagen fibres is non‐symmetric and varies significantly both axially and transmurally, and the density of collagen, elastin and smooth muscle cells is also highly heterogeneous along the aortic length. The quantification of aortic bioarchitecture facilitates the development of next‐generation finite element models that can potentially simulate the influence of regional aortic composition and microstructure on vessel biomechanics.

Introduction

Heterogeneity in regional mechanical properties along the aorta is a widely accepted phenomenon. The degree of heterogeneity, however, remains poorly characterised, and the underlying cause remains poorly understood, particularly in humans. In general, a high level of elastic compliance is required in the proximal aorta so that the kinetic energy expelled during systole can be stored as elastic strain energy in the vessel wall, leading to subsequent augmentation of pressure pulse propagation during diastole due to elastic recoil of the wall. The distal aorta, however, takes an alternative approach to blood transfer, with a higher smooth muscle cell (SMC) content and a reduced requirement for elastin. In the absence of a robust study investigating regional bioarchitecture of the human aorta using stereological techniques, this study focuses on quantifying the variation in orientations and densities of the main load‐bearing constituents along the entire vessel length, in an attempt to understand regional heterogeneity observed both in vitro and in vivo.

Significant literature on characterisation of the constituents of arteries to better understand their form and function exists, particularly with reference to the effects of ageing on the microstructural components of the aortic wall (Feldman & Glagov, 1971; Tomaszewski et al. 1976; Cattell et al. 1996; Fritze et al. 2012; Taghizadeh & Tafazzoli‐Shadpour, 2017). Hosoda et al. (1984) investigated the human thoracic aorta and found that elastin decreased with age, whereas collagen remained the same. Faber & Moller‐Hou (2009) also found that elastin decreased with age; however, they found that collagen increased. Others have focused on pathological states such as hypertension (Berry & Greenwald, 1976), Marfan syndrome (Halme et al. 1985), dissection (Yamada et al. 2015) and aneurysmal tissue (Baxter et al. 1992; Choudhury & Leask, 2009) and found significant differences compared with control groups. For further details regarding elastin and collagen microstructure in the human aorta in both ageing and disease, the reader is directed to the comprehensive review by Tsamis et al. (2013).

Much less emphasis, however, has been placed on the regional heterogeneity in the biomechanics of the aorta in health and disease. Due to both the ethics and accessibility of human tissue, animal studies surpass human studies. Harkness et al. (1957) investigated the regional elastin and collagen proportions in dogs and found that in the intrathoracic aorta there was approximately twice as much elastin as collagen in the wall; in all other vessels this relationship was reversed. Saey et al. (2015) report, for an equine population, a greater concentration of collagen in the distal thoracic aorta compared with proximal; however, there were no significant regional differences in elastin content. Results by Davidson et al. (1985) show that elastic fibres were reduced in the abdominal compared with thoracic aorta in a porcine group, and Shadwick (1999) reported higher compliance proximally than distally in whales. Despite the wealth of knowledge for animal populations, the regional biomechanics of the human aorta in terms of bioarchitecture remains relatively unknown.

This work quantifies the regional orientations of collagen fibres, in addition to the regional density of elastin, collagen and SMCs along the length of the human aorta. Stereological and polarised light microscopy (PLM) techniques provide evidence that a von Mises mixture model is required to fit the complex axially and transmurally varying fibre distributions of the vessel. Previous finite element models of the aorta assume a homogeneous (Roy et al. 2014) and symmetric (Grytsan et al. 2015) orientation of collagen fibres, which cannot capture the true anisotropy of the vessel. The regional population densities of elastin, collagen and SMCs along the length of the human aorta are quantified using high spatial resolution, intended to advance the understanding of regional aortic biomechanics. Results are presented for eight sites from the proximal ascending aorta to immediately proximal to the common iliac bifurcation for five cadavers. Histological and stereological techniques reveal significant spatial variations in constituent densities along the aorta, and morphological analyses show marked variation in wall layer thicknesses between intimal, medial and adventitial layers with increasing distance from the heart. Results presented on the regional heterogeneity of the aorta may aid in the understanding of the underlying causes for the differences in mortality following proximal compared with distal aortic stenting (Martín et al. 2008; Bischoff et al. 2016; Beach et al. 2017; Concannon et al. 2017; Conrad et al. 2017). Additionally, the quantification and fitting of regional aortic microstructural data facilitates the development of next‐generation finite element models that can potentially simulate the influence of regional aortic composition and microstructure on vessel biomechanics.

Methodology

In this paper, the regional aortic microstructure responsible for vessel compliance was quantified using histological and stereological techniques. Localised anterior tissue fractions of elastin, collagen and SMCs were quantified by taking samples at eight sites along the aorta. The dissection procedure is outlined in section on ‘Tissue harvesting’ and illustrated in Fig. 1A. Additionally, circumferential distribution of collagen and elastin in the ascending thoracic aorta was quantified, results of which can be found in the Supporting Information. Regional wall layer thicknesses are presented for each cadaver, and SMC content and layer specific collagen fibre orientations (intima, media and adventitia) were also investigated in a single cadaver for each of the eight sample sites. As endothelial cells and SMCs are considered the two major cellular components within the vessel wall which play a role in stiffness (Qi et al. 2011; Kaeberlein & Martin, 2016; Lacolley et al. 2017) and endothelial cells contribute more in a mechanotransductive nature and negligibly to the actual stiffness of the wall themselves (Williams & Wick, 2005; Karšaj & Humphrey, 2012; Gültekin et al. 2016), our cellular focus herein concentrates on SMCs.

Figure 1.

(A) Excision sites along the aorta (black squares), where the diaphragm is indicated by the dashed black line. (B) Sample orientations for Area fraction (AF) in the r‐c (radial‐circumferential) plane. (C) Sample orientations for fiber orientation (FO) in the a‐c (axial‐circumferential) plane. (D) Sections post‐staining for AF with Verhoeff's Method. (E) Sections post‐staining for FO with Picrosirius Red.

Tissue harvesting

Aortic samples were harvested from five adult human cadavers (age range 67–92 years, mean 77.6 years) with no reported history of aortic disease. Table 1 outlines the details of the donors. All cadaveric material used was bequeathed to Anatomy, School of Medicine, National University of Ireland Galway in accordance with legislation governing the practice of Anatomy in the Republic of Ireland (Medical Practitioners Act 2007). Three of the five cadavers were female. Cadavers were embalmed using a standard mixture containing formalin, glycerine, phenol and methanol (12 L water + 2.4 L of a 37–41% formalin solution + 2 L phenol + 6 L glycerine + 6 L methanol).

Table 1.

Details of donor cohort

Donor #	Gender	Age at death	Reported cause of death
1	Female	83 years	Carcinoma of urinary bladder
2	Male	67 years	Carcinoma of sigmoid colon metastatic to liver & lung
3	Male	76 years	Squamous cell carcinoma of parietal scalp with intracranial extension
4	Female	70 years	Heart failure secondary to valvular heart disease
5	Female	92 years	Cerebrovascular accident; ischaemic heart disease

An approximately 1‐cm² area of the aorta was excised at each of the following levels: (1) ascending thoracic aorta (distal to sinus of Valsalva); (2) ascending thoracic aorta (proximal to great vessels); (3) thoracic aorta (distal to great vessels); (4) thoracic aorta (start of descending); (5) thoracic aorta (mid); (6) abdominal aorta (proximal to coeliac trunk); (7) abdominal aorta (distal to superior mesenteric artery) and (8) abdominal aorta (proximal to common iliac bifurcation). Anterior sample excision sites are illustrated in Fig. 1A. Due to anatomical variance between cadavers, including length, curvature and tortuosity, histological results are presented as a function of eight anatomical sites rather than as a function of physical distance from the aortic root. Following dissection, specimens are dehydrated through a graded series of ethanol (50–100%) before being embedded in paraffin wax and sectioned at a thickness of 5 μm in order to maximise collagen to tissue contrast and limit tissue scattering (Yang et al. 2018). For area fraction (AF) analyses, sectioning is performed in the r‐c (radial‐circumferential) plane to enable investigation through the thickness of the aortic wall, while for fibre orientation (FO) analyses sectioning is performed in the a‐c (axial‐circumferential) plane on a separate sample obtained from the same site (Fig. 1B,C).

Staining procedures

In each case, the pre‐staining procedure involved wax removal from each sample using xylene and sample rehydration through a series of ethanol solutions of decreasing concentrations (100% to 50%). Separate staining procedures were performed on separate slides for quantification of each tissue constituent. For AF analyses, elastin was stained dark blue/black using Verhoeff's method (Zugun et al. 2013); collagen was stained green/blue using Masson's Trichrome (Vorkapic et al. 2016) and nuclei were stained dark blue using haematoxylin & eosin (Lo Vasco et al. 2011) protocols. For FO analyses, slides were stained using Picrosirius Red to attenuate the birefringence of collagen fibres (Junqueira et al. 1979), which subsequently appear red on a black background. Completion of each staining protocol involved dehydration through alcohols of increasing concentrations, followed by xylene clearing and slide mounting with D.P.X.

Postprocessing

Stereological analyses

Sections were obtained from the tissue blocks of eight regions from the proximal ascending to distal abdominal aorta (as outlined in section on ‘Tissue harvesting’). All AF slides were examined using a Leica DM500 light microscope, with an ICC50 HD camera attachment (Leica Microsystems Limited, Switzerland) and a 40× objective lens. Systematic random sampling procedures were employed and simple point counting methods (using a 12 × 12 grid generated by MATLAB (R2017b, MathWorks Inc., Natick, MA, USA) were used to estimate the local area fraction of collagen, elastin and SMCs (Elias et al. 1971; Mayhew, 1991; Wreford, 1995; Evanko et al. 2018). Given that there are no cells present in the media apart from SMCs (Rhodin, 1980), random sampling during SMC quantification is confined to this layer, as haematoxylin & eosin staining provides no specific distinction between cell types. The approach involves calculation of the area fraction (V _v) of each tissue component by expressing the proportion of points hitting a tissue component (P _c) as a fraction of the total number of points hitting any tissue (P _T):

$$V_v=\frac{\sum P_C}{\sum P_T}$$ (1)

The (V _v) was calculated for each image in each triplet and the average was recorded for each location. For elastin and collagen, the average (V _v) for each location across the five cadavers is shown in Figs 3 and 4, respectively, and SMC content is shown in Fig. 8.

Wall layer thickness

The individual thickness values for each layer of the aortic wall (intima, media, adventitia) were measured at each site for all five cadavers using imagej following calibration with the scale bar. The average thickness values are presented in Fig. 5. The intima/media (IM) border is defined along the internal elastic lamina while the media/adventitia (MA) border is defined by the external elastic lamella.

Polarised light microscopy

Polarised light microscopy techniques were employed at each location and wall layer for a single donor to investigate both axial and radial heterogeneity in collagen fibre orientations. Each FO slide was reviewed using a Nikon Eclipse E200 transmitted polarised light microscope with an epifluorescence attachment. Images were captured with a Nikon DS‐Fi1, 6‐megapixel camera equipped with NIS elements software. The circumferential direction of each tissue sample was oriented along the long axis of the slide to ensure consistency in determining fibre orientations relative to a universal horizontal plane. Each slide was then placed on the rotating stage with the long axis parallel to the polariser (East/West) and the centre of the sample at the crosshairs of the polariser and analyser (filters).

A detailed review of the principles surrounding PLM is outside the scope of this work but can be found in the literature, for example Török et al. (1998), Murphy (2002), Chayen (1983) and Carlton (2011). In brief, a birefringent material, such as collagen, changes the polarisation state of light, which results in a degree of altered light intensity that depends on the angle between the sample and the filters. Maximum intensity is achieved when fibres are aligned at ± 45°.

For each wall layer at each site, such that the full spectrum of fibre orientations is mapped, rotation of the stage and acquisition of 0° and 45° images resulted in the inputs of the composite image (Fig. 2A) (Gaul et al. 2017) which was formed using imagej. This image was then analysed using the OrientationJ plugin. Knowledge of the mean fibre thickness (σ = 24 μm) is required for accurate quantification of fibre orientation, which was obtained from the images via the scale bar. The angle (φ) is defined relative to the horizontal (0°) according to:

$$\phi =\frac{1}{2}arctan\left(2\frac{{〈f_x,f_y〉}_w}{{〈f_y,f_y〉}_w-{〈f_x,f_x〉}_w}\right)$$ (2)

Figure 2.

(A) Raw composite image used as input into OrientationJ plugin. (B) Result following input image where 0° indicates the circumferential axis of the sample and ± π/2d indicate the axial axis. (C) Corresponding probability density histogram with fitted von Mises mixture model to raw data in red.

where f _x and f _y are the spatial partial derivatives of the image function f(x,y) along the x and y directions, respectively, angled brackets indicate the inner product, and w(x,y) is the Gaussian weighting function that specifies the area of interest. Further detail on the theory surrounding OrientationJ can be found in Rezakhaniha et al. (2012) and Püspöki et al. (2016). Figure 2B shows the result of the input composite image Fig. 2A, where horizontal and vertical directions indicate circumferential and axial sample axes, respectively. The corresponding probability density histogram is shown in Fig. 2C, where MATLAB (R2017b, MathWorks Inc., Natick, MA, USA) based von Mises mixture models is fit to the individual datasets to estimate the number of fibre families present based on the log‐likelihood function (Hung et al. 2012; Schymura, 2019). The wavelengths of the individual collagen fibrils observed in Krasny et al. (2018), which are two to three times the fibril thickness, 60 nm (Hansen et al. 2009), are not observed at the magnification used in this study, where the pixel resolution was 500 nm. Therefore, the individual fibril undulations did not affect the von Mises distributions measured.

Statistical analysis

Statistical analysis of this research was performed in MATLAB (R2017b, MathWorks Inc.). After collecting the data, it was entered into a workspace for organisation. Data were analysed and tested using Student's t‐test for continuous variables. Statistical significance was accepted with a P‐value of less than 0.05.

Results

We employed histological and stereological techniques to investigate the orientations and densities of aortic wall constituents which contribute to the compliance of the vessel. Stereological analyses allow for quantification of the regional area fractions of elastin, collagen and SMCs within the aortic wall at each location outlined in section on ‘Tissue harvesting’, and PLM techniques allow for the quantification of fibre orientation distributions.

Elastin

Figure 3 shows the spatial change in elastin content from proximal to distal aorta. Elastin is clearly demonstrated to decrease with increasing distance from the heart. For example, in the ascending aorta, elastin contributes 27 ± 4.7% of the area, whereas in the distal abdominal aorta (immediately proximal to the common iliac bifurcation) the area made up of elastin drops to 17 ± 3.9%. Additionally, significant circumferential heterogeneity in elastin content is observed in the ascending thoracic aorta, with the highest area fraction observed on the lateral segment of the wall (see Supporting Information Fig. S1).

Figure 3.

(A) Elastin as a function of location along the aorta; (B,C) thoracic aorta (Sites 1–4), (D,E) abdominal aorta (Sites 5–8). A 36.5% decrease in elastin content is observed between the ascending thoracic and distal abdominal portions of the vessel following staining by Verhoeff's method. Error bars show the range at each site from min to max.

Collagen

Figure 4 shows the regional variance in collagen content along the aorta from proximal to distal. In contrast to elastin, the area fraction of collagen increases with increasing distance from the heart. At the level of the proximal ascending aorta, collagen constitutes 22 ± 1.6% of the vessel wall, whereas in the distal abdominal portion this is markedly increased to 53 ± 5.1%. Making use of thoracic and abdominal subgroups, split by the diaphragm, a statistically significant difference in collagen content is observed (P = 0.006). Additionally, significant circumferential heterogeneity in collagen content is observed in the ascending thoracic aorta, with the highest area fraction observed on the medial segment of the wall (see Fig. S1c).

Figure 4.

(A) Collagen as a function of location along the aorta; (B,C) thoracic aorta (Sites 1–4), (D,E) abdominal aorta (Sites 5–8). A 133% increase in collagen content is observed between the ascending thoracic and distal abdominal portions of the vessel following Masson's Trichrome stain. Error bars show the range at each site from min to max, and * indicates a statistically significant difference between thoracic and abdominal aorta subgroups (P < 0.05).

Layer thickness

Individual layer (intima, media and adventitia) thicknesses are presented for each location in Fig. 5. The intimal thickness (defined by the internal elastic lamina) increases with increasing distance from the heart from 100 ± 47 μm at the proximal ascending aorta to 130 ± 20 μm at the distal abdominal aorta. The media, bound by the internal and external elastic laminae, markedly decreases from 877 ± 123 μm proximally to 369 ± 33 μm to distally. Finally, the adventitia, defined as outside the external elastic lamina, increases from 221 ± 71 μm nearest the heart to 323 ± 12 μm immediately proximal to the common iliac bifurcation. Mean layer thickness values for the thoracic and abdominal aorta are presented in Table 2.

Figure 5.

Intima/media/adventitia layer thickness breakdown according to location along the aorta. The overall aortic wall thickness changes from 1.2 to 0.83 mm from proximal to distal portions. Error bars show the range at each site from min to max.

Table 2.

Wall layer thickness values, where (*) indicates a statistically significant difference (P < 0.05) between thoracic and abdominal subgroups

	Thoracic				Abdominal
Site	1	2	3	4	5	6	7	8
Intima (mm)	0.10	0.08	0.11	0.12	0.12	0.12	0.11	0.13
Mean (mm)	0.10				0.12
Media (mm)	0.88	0.65	0.60	0.62	0.46	0.53	0.53	0.37
Mean (*) (mm)	0.69				0.47
Adventitia (mm)	0.22	0.17	0.28	0.28	0.25	0.30	0.34	0.32
Mean (mm)	0.24				0.30

Collagen fibre orientations

Polarised light microscopy techniques are employed to quantify the regional heterogeneity in collagen fibre orientations along both the axial and radial dimensions of the aorta. As outlined in section ‘Tissue harvesting’, eight samples were analysed along the length of the aorta from proximal ascending to distal abdominal. For each sample, the orientation of the blocks was set to allow sectioning through the thickness of the wall layers, such that samples could be taken from the intima, media and adventitia at each axial location. Figure 6 illustrates the probability density histograms (dotted) for each aortic layer for each location along the vessel in ascending order from proximal to distal. Significant heterogeneity is evident both radially and axially. Von Mises mixture models (solid) fit to the data allowed quantification of the mean and standard deviation (inversely related to Kappa) of each von Mises distribution, which were taken to represent an individual fibre family. Parameters pertaining to each von Mises mixture model can be found in Table 3.

Figure 6.

Probability density histograms (dotted) for each site along the axial dimension of the aorta, with von Mises mixture model fits overlaid (solid). For each location the intima is indicated in red, media in green and adventitia in blue. 0 and ± π/2 represent the circumferential and axial axes of the sample, respectively.

Table 3.

Von Mises mixture model parameters for each site and wall layer

Location	Order	Mean (Radians)	Kappa (Radians)	Proportion
Site1_Intima	2	(−0.59), (1.05)	(3.8), (10.5)	(0.61), (0.39)
Site1_Media	2	(−0.34), (−0.35)	(4.1), (90.7)	(0.65), (0.35)
Site1_Adventitia	2	(−0.53), (1.21)	(2.3), (19.7)	(0.37), (0.63)
Site2_Intima	3	(–1.34), (1.23), (0.77)	(32.4), (34.0), (9.3)	(0.11), (0.44), (0.45)
Site2_Media	2	(−0.66), (0.73)	(5.1), (9.8)	(0.35), (0.65)
Site2_Adventitia	3	(−0.68), (0.40), (−1.34)	(51.7), (17.9), (77.4)	(0.96), (0.02), (0.02)
Site3_Intima	3	(0.92), (−0.46), (0.45)	(565.8), (12.4), (10.3)	(0.02), (0.75), (0.2)
Site3_Media	3	(−1.39), (0.46), (−0.9)	(65.5), (7.3), (11.5)	(0.08), (0.75), (0.17)
Site3_Adventitia	3	(1.08), (−0.72), (−0.21)	(8.9), (19.0), (44.4)	(0.06), (0.48), (0.46)
Site4_Intima	1	(0.08)	(3.7)	(1)
Site4_Media	2	(0.92), (−0.41)	(11.8), (10.7)	(0.06), (0.94)
Site4_Adventitia	2	(−0.84), (−0.04)	(58.2), (3.6)	(0.24), (0.76)
Site5_Intima	3	(−1.36), (0.64), (1.15)	(37.5), (30.5), (24.8)	(0.07), (0.34), (0.59)
Site5_Media	2	(−0.54), (0.94)	(5.0), (29.4)	(0.16), (0.84)
Site5_Adventitia	2	(0.61), (−1.22)	(6.0), (12.0)	(0.97), (0.03)
Site6_Intima	2	(−0.63), (−0.21)	(81.8), (52.3)	(0.12), (0.88)
Site6_Media	3	(0.47), (0.9), (0.48)	(3.1), (19.4), (80.2)	(0.13), (0.5), (0.36)
Site6_Adventitia	3	(1.48), (−1.41), (−1.38)	(224.8), (230.2), (98.5)	(0.06), (0.68), (0.26)
Site7_Intima	3	(−1.32), (1.25), (0.84)	(26.7), (35.3), (10.9)	(0.18), (0.48), (0.33)
Site7_Media	3	(−1.48), (1.36), (0.24)	(193.5), (72.3), (3.4)	(0.15), (0.33), (0.52)
Site7_Adventitia	4	(−0.09), (−0.34), (0.25), (0.15)	(20), (486), (506), (350)	(0.3), (0.19), (0.16), (0.35)
Site8_Intima	5	(−0.65), (0.01), (0.81), (−0.34), (−0.69)	(10.2), (37.8), (11.3), (386.6), (356.2)	(0.2), (0.39), (0.02), (0.23), (0.16)
Site8_Media	2	(−1.47), (1.17)	(279.7), (56.2)	(0.05), (0.95)
Site8_Adventitia	1	(−0.06)	(2.3)	(1)

Figure 7 highlights the evolution of fibre peaks along the length of the aorta for each tunical layer. Significant heterogeneity is evident in the mean angle and its standard deviation for each peak. At the majority of sites (intima, 6/8 sites; media, 5/8 sites; adventitia, 6/8 sites) peaks are observed at both a positive and negative angle to the circumferential axis. However, in general, the magnitude and standard deviation of each of the peaks at a given site are not symmetric about the circumferential axis. In the adventitial layer at sites 4, 7 and 8, strong peaks are observed near to the circumferential axis, whereas in the media no strong peak is observed on the circumferential axis. In the media, the mean angle at a site becomes more positive distally, whereas in the adventitia the mean angle at a site becomes more negative distally.

Figure 7.

Evolution of fibre peaks along the aorta for each layer (intima, media, adventitia). For each site, from proximal to distal aorta, the size of the filled circle represents the proportion of the von Mises mixture model captured by the given mean angle and spread of each peak. In each case, the standard deviation about each mean is represented by the solid vertical bars.

Smooth muscle cells

Finally, SMC content was investigated as a function of position along the aorta, results of which are presented in Fig. 8. The proximal aorta exhibited a lower density of SMCs than the distal abdominal portion, with an area fraction of 22.4% and 27.0%, respectively. Categorising the sample sites into thoracic and abdominal subgroups, a statistically significant difference in SMC content (Fig. 8A) is observed between the two groups split by the diaphragm (P = 0.009). Additionally, a higher number of SMC nuclei/mm² (Fig. 8B) are observed distally compared with proximally.

Figure 8.

(A) SMC area fraction (%). (B) Number of SMC nuclei/mm². Error bars show the range at each site from min to max, and * indicates a statistically significant difference between thoracic (Sites 1–4) and abdominal (Sites 5–8) aorta subgroups (P < 0.05).

Discussion

The focus of this paper is to investigate the regional bioarchitecture of the human aorta, specifically the constituents which contribute to the compliance of the vessel. The study quantifies the orientation and density of the main load‐bearing constituents of the wall at a number of discrete points along the aorta from proximal ascending to distal abdominal portions. It is shown that the organisation of collagen fibres is highly complex, and the quantification and fitting of the orientation distributions raises questions regarding the accuracy of models that assume collagen fibres to be homogeneously (Roy et al. 2014) and symmetrically (Grytsan et al. 2015) distributed throughout. The quantification of regional aortic bioarchitecture represents advanced through integration with advanced finite element models, incorporating the true regional microstructure of the aorta to levels of unprecedented detail. To the authors’ knowledge, no previous study to date has quantified, using histological and stereological methods, the regional variations in elastin, collagen and SMC distributions in human cadaveric aortae.

The regional density of elastin within the aortic wall from the proximal ascending to distal abdominal segments was examined. The high levels of elastin in the proximal ascending aorta (27 ± 4.7%) render it the dominant protein within the wall proximally. Additionally, significant circumferential heterogeneity in elastin content is observed in the ascending thoracic aorta, with the highest area fraction observed on the lateral segment of the wall (see Fig. S1a). From a biomechanical point of view, high levels of elasticity within the vessel wall proximally are crucial for maintenance of the Windkessel effect, left ventricular function and diastolic flow (Belz, 1995; Davies et al. 2008; Cavalcante et al. 2011). Distally, a reduction in elastin concentration of 36.5% is observed at the level immediately proximal to the common iliac bifurcation (17 ± 3.9%). This is in broad agreement with Halloran et al. (1995), who observed a decrease in elastin concentration using biochemical analysis between the thoracic and abdominal segments. As the Windkessel effect results from the compliance of aortic tissue, a decrease in elastin and increase in collagen and SMCs distally suggests a microstructural mechanism for heterogeneous aortic compliance observed previously in vitro. An inverse relationship between elastin and SMCs exists whereby SMC content is higher distally, dictating vascular tone. Haskett et al. (2010) subjected human aortic tissue samples to biaxial tension and found that the abdominal aorta was significantly stiffer than proximal regions. Kim et al. (2013) observed under bulge inflation tests of thoracic aortic rings that distal segments were stiffer than proximal segments. Without the implementation of histological and stereological techniques, however, the components of aortic tissue contributing to this increased stiffness cannot be determined.

Investigation of the regional density of collagen within the aortic wall provides insight into both its importance at specific anatomical locations and its role in the deformation of the system. In the proximal ascending aorta, collagen content is lower (22 ± 1.6%) but becomes far more pronounced with increasing distance from the heart, and distally becomes the dominant protein within the wall (53% ± 5.1%). Additionally, significant circumferential heterogeneity in collagen content is observed in the ascending thoracic aorta, with the highest area fraction observed on the medial segment of the wall (see Fig. S1c). The increased density of collagen distally observed here is also in agreement with work published on various species (Grant, 1967), although those authors rely on biochemical methods for the quantification of collagen through hydroxyproline, which is also present in elastin. The inverse relationship between elastin and collagen observed in this study is in agreement with previous work stating that the ratio of elastin to collagen decreases markedly between the aorta and femoral arteries in both porcine and murine populations (Sokolis et al. 2008; Basu et al. 2010). With collagen reported to exhibit 5000 times the tensile stiffness of elastin (Cavalcante et al. 2011), results presented here also provide microstructural insight into the increased stiffness seen in the distal aorta both in vitro (Moriwaki et al. 2011; Krüger et al. 2016) and in vivo (Mohiaddin et al. 1989; Saouti et al. 2012). Moreover, these findings suggest that the common description of aortic compliance using a single coefficient may be inaccurate (Lehmann et al. 1998; Vyas et al. 2007; Lalande et al. 2008).

Due to the collagen fibres being orders of magnitude stiffer than the other constituents of aortic tissue, the vessel walls behave like a fibre‐reinforced composite, in that the orientation of the fibres plays a role in the structural anisotropy of the vessel wall. Thus, fibre density alone cannot fully describe the compliance of the vessel, and an investigation of the orientation of collagen fibres within the aortic wall at each location was performed as outlined previously. Significant dispersion is evident in each probability density histogram, in addition to notable transmural variation in mean fibre angles between the intima, media and adventitia. Interestingly, the presence of two equally dominant and symmetric fibre families is absent.

Although two families (peaks in probability density histogram) are frequently apparent in the present data, in each case one constitutes a significantly greater area fraction than the other. O'Connell et al. (2008), Horny et al. (2010), Schriefl et al. (2012), Chow et al. (2014), Sassani et al. (2015), Weisbecker et al. (2015) and Sugita & Matsumoto (2018) report that a single family of fibres was evident; however, others have reported two (Schriefl et al. 2012; Laksari et al. 2016), three (Schriefl et al. 2012) and even four (Rezakhaniha et al. 2012; Schriefl et al. 2012) fibre families, all of which are evident here, further emphasising aortic microarchitectural heterogeneity at a local level. It is shown that a von Mises mixture model is required accurately to fit the complexity of collagen fibre orientations that exist along the aorta, which suggests that common constitutive laws implemented through finite element analysis (Holzapfel et al. 2000; Nolan et al. 2014) may not be capable of fully capturing the full anisotropy of the vessel.

Significant variance in layer thicknesses is observed along the aortic length within the intima, media and adventitia. Results show that intimal thickness increases from the proximal (100 ± 47 μm) to distal aorta (130 ± 20 μm). Similarly, adventitial layer thickness increases from the proximal (221 ± 71 μm) to distal (323 ± 12 μm) aorta. The medial layer, however, decreases from the proximal (877 ± 123 μm) to distal (369 ± 33 μm) aorta. The overall wall thickness reduces from (1200 ± 430 μm) to (825 ± 192 μm) with increasing distance from the heart. Results shown here are in agreement with work published by Rhodin (1980) and Schriefl et al. (2012), in which these authors report an increase in both intimal and adventitial thicknesses in addition to a decrease in medial thickness distally.

We also examined regional SMC content to investigate the spatial distribution of active constituents within the aortic wall. Results show that SMC content increases with increasing distance from the heart, with 22.4% of the proximal ascending aortic wall comprised of SMCs, increasing to 27.0% in the distal abdominal aorta. Although the aorta is considered to be an elastic artery, relative to smaller calibre vessels which are termed muscular (Lacolley et al. 2017), a significant difference is observed regionally between the proximal thoracic and distal abdominal portions that follow the tapering of the aorta with increased SMC density distally. This result is in agreement with reports by Lacolley et al. (2017) and the findings of Arnaud (2000) where the abdominal aorta exhibited a higher basal requirement for oxygen consumption and mitochondrial activity than its thoracic counterpart in the porcine aorta. The same authors hypothesise that the more passive elastic nature of the proximal aorta overrides the requirement for a dense active component, whereas the abdominal aorta may be subject to a more pronounced vasoactive regulation. Future work should investigate the regional variance in the three‐dimensional morphology of SMCs and nuclei, including detailed quantification of aspect ratio and cell volume.

Due to legislative constraints, cell content and collagen fibre orientation analyses were only conducted on one of the donors. Further work is warranted here to increase the sample size to gain a better insight into the true variability of such parameters. However, considerable differences are apparent even within each constituent of the cohort studied here, and so care should be taken when averaging values across donors not to lose important inter‐sample variances due to age, lifestyle and medical history. Finally, it should be noted that the average age of the donors in this study is 77.6 years, entirely due to availability, and as such reflects an elderly population. Further scope exists in extending this framework to a younger comparative cohort to characterise the effects of ageing on regional bioarchitecture.

The detailed insights into the regional bioarchitecture of the human aorta revealed in this paper are critical for the development of a more complete understanding of aortic anatomy and physiology and its complex changes throughout its length. The individual component densities and orientations of elastin, collagen and SMCs all contribute to the complex local mechanical behaviour of the aorta, ultimately governing the non‐linear elasticity and contractility of the vessel. Follow‐on 4D Flow MRI/finite element investigations are being performed to parse the individual contributions of the components of the aortic wall to the regional heterogeneity in aortic biomechanics. Significant heterogeneity between the thoracic and abdominal aorta in terms of embryology, atherosclerotic plaque deposition, protease profiles and cell signalling pathways has been reported previously (see review by Ruddy et al. 2008) in addition to heterogeneous changes in mechanical behaviour for porcine tissue (Kim et al. 2013). Results presented in the current study provide new information on the complex distributions of fibres at each section along the aorta, in addition to detailed quantification of the spatial heterogeneity of the aortic microstructure.

All values of wall thickness and fibre orientation reported in this study are for unloaded tissue. Therefore, such data can be directly input to a finite element model in the unloaded reference configuration. Simulation of the artery response to applied physiological loading (e.g. diastolic/systolic lumen pressure profiles) could readily predict heterogeneous changes in wall thickness and fibre orientation during a cardiac cycle. A follow‐on study will be performed in which the heterogeneous microstructural and geometric data will be input into a finite element framework. Finally, the data presented can facilitate the generation of the most advanced anisotropic finite element models of the human aorta available to date, highlighting the importance of integrating histological and stereological data with the field of computational science.

Conclusion

The regional bioarchitecture of the human aorta was quantified using histological and stereological techniques. Significant heterogeneity is evident in the orientation of collagen fibers both axially and transmurally within the aortic wall. Heterogeneity in elastin, collagen and SMC content is also observed along the axial dimension of the aorta. In the proximal ascending portion, elastin is the dominant protein, whereas within the wall distally, collagen is the dominant protein, rendering this area naturally less compliant. Morphological analyses show the thickness of each tunical layer to also vary significantly with increasing distance from the heart. Finally, the quantification and fitting of regional aortic bioarchitectural data can serve as a direct input into the development of next‐generation finite element models that can potentially simulate the influence of regional aortic composition and microstructure on vessel biomechanics.

Author contributions

J.C.: concept/design, acquisition of data, data analysis/interpretation, drafting of the manuscript, critical revision of the manuscript and approval of the article. P.D.: concept/design, data analysis/interpretation, critical revision of the manuscript and approval of the article. A.B.: concept/design, data analysis/interpretation, critical revision of the manuscript and approval of the article. S.S.: critical revision of the manuscript and approval of the article. N.H.: critical revision of the manuscript and approval of the article. P.McH.: concept/design, data analysis/interpretation, drafting of the manuscript, critical revision of the manuscript and approval of the article. K.M.: data analysis/interpretation, critical revision of the manuscript and approval of the article. J.P.McG.: concept/design, data analysis/interpretation, drafting of the manuscript, critical revision of the manuscript and approval of the article.

Supporting information

Fig. S1. (A) Circumferential heterogeneity in elastin area fraction (%) around the proximal ascending aorta. (B) Schematic illustrating the vessel quadrants analysed where A = anterior; L = lateral; M = Medial and P = posterior. (C) Circumferential variability in collagen content within the aortic wall.