WINDKESSEL-PRESERVING AORTIC STENT-GRAFT ATTENUATES LEFT VENTRICULAR REMODELING AFTER TEVAR IN A PORCINE MODEL

Abstract

Objective:

Thoracic endovascular aortic repair (TEVAR) has largely replaced open surgery for aortic pathologies and trauma due to its minimally invasive nature and reduced perioperative risk. However, conventional stent-grafts are significantly stiffer than the native aorta and may impair Windkessel function, promoting adverse cardiac remodeling - particularly in younger trauma patients with longer life expectancies. Our goal was to evaluate the impact of conventional stiff versus compliant Windkessel-preserving stent-grafts on aortic biomechanics and left ventricular (LV) remodeling in a swine model of TEVAR.

Methods:

Twenty juvenile swine were divided into three groups: sham-operated controls (n=7), animals implanted with conventional stiff stent-grafts (CS-SG, n=8), and animals implanted with novel compliant nanofibrous stent-grafts (NF-SG, n=5). Longitudinal cardiac-gated CT angiography was used to quantify aortic volumes, diameter pulsatility, and LV mass over 4.3 months.

Results:

CS-SG implantation resulted in a 93% reduction in aortic pulsatility in the descending thoracic aorta and accelerated LV mass gain (4.83±2.34 g/month), approximately 70% faster than controls. In contrast, NF-SG preserved local aortic compliance and maintained Windkessel function, with systolic-to-diastolic volume and diameter ratios in the stented segment similar to baseline. LV growth in the NF-SG group (2.80±1.84 g/month) was comparable to controls (2.83±2.46 g/month), indicating attenuation of adverse remodeling.

Conclusions:

CS-SG impair Windkessel function and promote LV hypertrophy, while compliant stent-grafts preserve aortic biomechanics and mitigate cardiac remodeling. These findings highlight the importance of compliance-matched endografts in TEVAR patients and support further development of Windkessel-preserving technologies.

1. INTRODUCTION

The aorta is a highly elastic conduit artery that delivers blood from the heart to distal tissues and organs. Its defining feature is its ability to expand during systole - storing approximately 50% of the left ventricular (LV) stroke volume - due to the presence of elastic lamellae and fibers within the aortic wall. During diastole, the aorta recoils to propel blood forward and maintain perfusion, including to the coronary circulation¹. This phenomenon, known as the Windkessel effect, is central to normal cardiovascular physiology. However, aging and disease lead to extracellular matrix remodeling, characterized by fragmentation of elastic lamellae and increased deposition and cross-linking of collagen. These changes stiffen the aortic wall² and attenuate the Windkessel function³, contributing to elevated systolic blood pressure⁴, increased pulse wave velocity, and augmented LV afterload⁵. Collectively, these effects drive adverse cardiac remodeling - including LV hypertrophy, myocardial fibrosis, diastolic dysfunction, and ultimately, cardiomyopathy⁶.

A similar but more rapid process is believed to occur following implantation of conventional stiff stent-grafts or surgical grafts, though this phenomenon is less well understood. The textile materials used in these devices - typically expanded polytetrafluoroethylene (ePTFE) or polyethylene terephthalate (PET, or Dacron) - are substantially stiffer than the native aortic wall⁷. Their implantation significantly increases aortic stiffness and mimics the biomechanical effects of vascular aging^8,9. While traditional evaluations of stent-graft performance focus on short-term outcomes such as morbidity, mortality, and re-intervention rates, long-term biomechanical and cardiac effects are more difficult to detect in typical TEVAR populations. Because most recipients are older and medically complex, subtle changes in aortic compliance or LV remodeling may be masked by comorbidities or limited follow-up, resulting in fewer opportunities to study these phenomena in humans.

Several experimental efforts have attempted to develop more compliant vascular grafts or stent-grafts to better mimic native aortic biomechanics^10–13. Although multi-layer or bio-inspired designs have shown improved distensibility in computational and benchtop studies – and in some cases in acute animal experiments¹³ – none have demonstrated durable in vivo performance or advanced toward clinical use. Clinical data, however, consistently show that implantation of stiff aortic grafts is associated with adverse cardiac remodeling. Takeda et al⁹ showed a significant increase in LV mass index after endovascular repair for abdominal aortic aneurysms, and van Bakel et al¹⁴ observed similar findings in patients undergoing thoracic endovascular aortic repair (TEVAR) using echocardiography and cardiac CT. Houben et al¹⁵ reported increased LV mass index after open surgical repair of the ascending aorta with stiff polyester grafts, and Kadoglou et al⁸ demonstrated that both PTFE and polyester grafts increased aortic stiffness, with polyester having the greater effect.

As stent-grafts are increasingly used in younger trauma patients – who have longer life expectancies, greater physiological reserve, and lower compliance with long-term surveillance – the need for refined performance metrics and long-term physiological evaluation is growing. Traditional endpoints fail to capture the cumulative cardiovascular impact in this population. Emerging evidence from advanced imaging and functional assessments reveals increased vascular stiffness, elevated LV mass, and reduced exercise tolerance in patients implanted with stiff aortic prostheses^8,9,16–21. Our recent study in young trauma TEVAR patients similarly showed increased LV mass, hypertension, and accelerated expansile remodeling of the ascending aorta²². A systematic review further confirmed these trends, identifying increased post-TEVAR hypertension, greater LV mass index, and rapid aortic remodeling in patients treated for blunt thoracic aortic injury²³.

Although younger trauma patients may be most affected over the long term due to their life expectancy, preservation of Windkessel function may be equally or even more important in older, frailer patients. With reduced cardiac reserve and diminished ability to buffer increases in afterload, abrupt stiffening of the thoracic aorta may impose a disproportionate hemodynamic burden in this population. Thus, the physiological consequences of endograft stiffness have relevance across the full clinical spectrum.

Despite this growing evidence, human studies remain confounded by heterogeneity in demographics, comorbidities, aortic pathology, and injury characteristics. Identifying suitable clinical control groups is also challenging. In contrast, animal models provide a more controlled experimental setting, enabling clearer attribution of observed changes to the intervention itself. In this study, we employed a porcine model – chosen for its close resemblance to human cardiovascular physiology and aortic dimensions²⁴ – to evaluate three groups: sham-operated controls to quantify somatic growth and remodeling, animals implanted with conventional stiff stent-grafts (CS-SG), and a pilot group treated with a novel compliant nanofibrous stent-graft (NF-SG) engineered to preserve Windkessel function¹². The NF-SG served as a mechanistic research tool to test the hypothesis that a Windkessel-preserving aortic stent-graft attenuates LV remodeling following TEVAR under physiologically relevant and tightly controlled conditions. Accordingly, the intent of this study was to interrogate ventricular-arterial coupling mechanisms rather than to perform a device efficacy comparison.

2. METHODS

2.1. Animals

All procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Nebraska Medical Center and were conducted in compliance with PHS Policy, the USDA Animal Welfare Act, and AAALAC Guide for the Care and Use of Laboratory Animals. A total of 20 Yucatan mini-swine (baseline age: 13.7 ± 1.9 months; weight: 58.3 ± 4.8 kg; 2F / 18M) were evaluated over an average duration of 4.3 months. Animals were randomly assigned to one of three groups: control (n = 7), conventional stiff stent-graft (CS-SG, n = 8), and compliant nanofibrous stent-graft (NF-SG, n = 5). Three computed tomography angiographies (CTAs) were performed – at baseline before TEVAR, intermediate eight weeks after TEVAR, and pre-sacrifice. Device positioning and deployment accuracy were verified intraoperatively using high-resolution fluoroscopy and angiography. One control animal underwent baseline CTA but died following the initial sham surgery; its data were included in baseline analyses of LV mass and aortic diameters but excluded from longitudinal growth analysis. Similarly, one NF-SG animal was excluded due to a malfunction of the experimental delivery system during the stent-graft implantation.

2.2. CTA imaging

Animals were sedated with Telazol (4.4–5 mg/kg) – Ketamine (2.2–2.5 mg/kg) – Xylazine (2.2–2.5 mg/kg), then intubated and maintained on 1–5% isoflurane during imaging. All CTA scans were ECG-gated and acquired retrospectively on a 256-slice GE Revolution CT scanner (GE Healthcare, Chicago, IL), producing 512 × 512 pixel images at a resolution of 0.23 mm and an axial slice thickness of 0.625 mm. Intravenous contrast (100 mL of iopamidol; Bracco Imaging, Milan, Italy) was administered, and 20 cardiac phases were obtained at 5% increments covering 5–95% of the cardiac cycle. Tube current was automatically modulated, and explicit mAs values were not stored in the DICOM metadata. All scans were conducted at 120 kVp to optimize image quality. DICOM images were imported into Materialise Mimics v25 (Materialise, Leuven, Belgium) for 3D reconstruction using standard tools for masking, region growing, mask separation, and model generation^25,26.

Baseline and follow-up cardiac CTA scans were analyzed independently in random order by a single operator to minimize variability. Cardiac reconstruction followed a previously established protocol²², which included isolating the epicardial surface of the LV from the apex to the aortic valve and pulmonary trunk, followed by segmentation of the endocardial surface, and subtraction of the endocardial mask from the epicardial mask to generate the 3D geometry of the LV wall transected immediately below the aortic valve. Reconstructions were performed at cardiac peak systole and end diastole, and results from both phases were averaged to assess LV mass²². Stroke volume was calculated as the difference between the LV inner volume at end-diastole and peak systole.

Aortic geometry was also analyzed at peak systole and end diastole, but phases were selected based on maximal and minimal aortic diameters, as these did not always coincide with the cardiac phases. Each selected phase was segmented manually and independently. The thoracic aorta was reconstructed from the aortic valve through the ascending aorta, arch, and descending thoracic aorta to the level of the cardiac apex. 3D geometries were then used to generate centerlines and measure best-fit diameters and lengths between major branch points^22,26. Best-fit diameter denotes the equivalent circular diameter in a centerline-orthogonal cross-section, calculated from a circle with area equal to the segmented lumen. Assessments included volumetric measurements of the ascending thoracic aorta (ATA), defined as the segment from the aortic valve to the innominate artery; the aortic arch, spanning from the innominate artery to the left subclavian artery; and the descending thoracic aorta (DTA), which extended from the left subclavian artery to the base of the heart. Aortic diameters were measured midpoint of the ATA, immediately distal to the left subclavian artery, and in the DTA at the base of the heart using centerline-referenced cross-sections to account for obliqueness. Lengths of the ATA, aortic arch, and DTA were also measured along the centerline (Figure 1). Ratios of systolic to diastolic aortic volumes and diameters were used to quantify the Windkessel effect in each aortic segment, while ratios of systolic to diastolic centerline lengths characterized the axial stretch experienced by each segment during the cardiac cycle.

Figure 1.

3D reconstruction of the LV and thoracic aorta at (A) peak systole and (B) end diastole. The ATA, aortic arch, and DTA are color-coded for segmental distinction. Images are from a representative animal in the CS-SG group; the stent-graft is shown in grey.

2.3. Surgical procedures

All animals were scheduled to undergo two procedures: stent-graft implantation (CS-SG or NF-SG) or sham surgery, followed by a terminal surgical procedure followed by euthanasia 4 months later. Anesthesia was induced with Telazol (4.4–5 mg/kg) – Ketamine (2.2–2.5 mg/kg) – Xylazine (2.2–2.5 mg/kg), and maintained with 1–5% inhaled isoflurane. Aspirin (325 mg) and clopidogrel (75 mg) were given daily from 3 days before surgery until euthanasia; cefazolin (1 g) was administered preoperatively. The common carotid artery was accessed percutaneously, and stent-grafts were deployed in the DTA just distal to the left subclavian artery, using a 10% oversizing relative to the proximal landing zone diameter measured on baseline CTA.

CS-SG animals received two overlapping Gore Excluder limbs (tapered diameter 18–20 to 14–16 mm, combined length 213 ± 17.3 mm, W.L. Gore & Associates, USA, Figure 2A) with ~2–3 cm of overlap. To accelerate attainment of a fully apposed, low-pulsatility configuration within the study timeframe, three balloon-expandable stents were deployed per animal - Palmaz XL stents proximally and distally (Cordis Corporation, USA) and a CP stent in the mid-segment (B. Braun Interventional Systems Inc., USA) - each expanded to the nominal diameter of the parent endograft. This approach was used to produce a functionally stiff, minimally deformable aortic segment, reflecting the marked reduction in cyclic diameter change and loss of Windkessel behavior observed in the stented thoracic aorta during chronic follow-up after TEVAR, which in humans typically evolves over 6–12 months.²²

Figure 2.

Representative stent-graft constructs used in this study. (A) Commercially available Gore Excluder limb stent-graft (CS-SG), representative of a conventional stiff endograft. (B) Example of a compliant nanofibrous stent-graft (NF-SG) shown during deployment from a custom-designed delivery system, illustrating controlled release and full expansion of the compliant graft architecture.

The NF-SG devices were developed in-house using a tapered, elastomeric nanofibrillar fabric electrospun from biomedical-grade polyether urethanes (Pellethane^® 5863–82A and Pellethane^® 2363–55DE, Lubrizol, USA)²⁷ and integrated with a nitinol stent skeleton¹² (Figure 2B). Grafts measured 12.5 cm in length and tapered from 18 mm to 14 mm in diameter. Importantly, our prior benchtop testing demonstrated that pulse wave velocity – an integrated, length-dependent measure of vascular stiffness – was independent of NF-SG length¹², indicating that a single continuous device adequately captures the intended biomechanical behavior. After implantation, animals were recovered and monitored. At the terminal procedure (~4 months post-implantation), euthanasia was performed via exsanguination, and death was confirmed by bilateral thoracotomy.

2.4. Statistical analysis

Results are presented as mean ± standard deviation. Group comparisons at baseline were assessed using one-way ANOVA, with Levene’s test confirming variance homogeneity. Normality and equal variance assumptions were met, so non-parametric tests were not required. Longitudinal changes in LV and aortic metrics were analyzed by fitting linear models for each animal, with slope representing the rate of change. Paired two-tailed t-tests compared baseline to follow-up values; p<.05 was considered significant. Analyses were performed in SPSS v29 (IBM Corp., Armonk, NY).

3. RESULTS

3.1. LV mass and volumes

Baseline LV mass did not differ among groups (p=.44, overall mean 107.4±17.0 g). In controls (Figure 3A), LV mass increased by 2.83±2.46 g/month (11% over 4.3 months, p=.014). CS-SG animals (Figure 3B) showed 71% faster increase (4.83±2.34 g/month; 21% increase from baseline, p=.001). In NF-SG animals (Figure 3C), LV mass rose at 2.80±1.84 g/month (10% increase, p=.042), similar to controls. Although individual NF-SG animals exhibited variable LV mass slopes, the group-level LV growth rate remained equivalent to controls and significantly lower than CS-SG animals. LV systolic, diastolic, and stroke volumes for all groups are shown in Figure 3D,E,F. Systolic volume remained stable in controls and NF-SG animals (p=.998, p=.70), but increased by 37% in CS-SG animals (p=.078). At follow-up, systolic volume was significantly higher in CS-SG vs. controls (p=.022), but not in NF-SG vs. controls (p=.061). Diastolic volume rose in both control (15%, p=.011) and CS-SG animals (24%, p=.006); NS-SG animals showed a similar 22% increase (p=.051). Stroke volume increased by 24% in controls (p=.036), 18% in CS-SG (p=.050), and 38% in NF-SG animals (p=.067).

Figure 3.

Characterization of LV remodeling. Changes in LV mass (g) over time for control (A, N = 6 animals), CS-SG (B, N = 8 animals), and NF-SG (C, N = 4 animals) groups across three time points. Lower panels show systolic, diastolic, and stroke volumes at baseline (“B”) and terminal evaluation (“T”) for each group (N = 6 for control, 8 for CS-SG, and 4 for NG-SG). In the violin plots, boxes represent the interquartile range (25^th-75^th percentiles), whiskers extend to 1.5 standard deviations, and medians are indicated by hollow circles. Paired two-tailed t-tests were used to compare baseline to follow-up with ** denoting p < .001, and * denoting p < .05.

3.2. Aortic volumes

In control animals (Table 1), ATA systolic volume increased by 19% (p=.027), while diastolic volume and the systolic-to-diastolic ratio remained unchanged (p=.13, p=.25, Figure 4A). Arch diastolic volume rose by 8% (p=.009), but systolic volume (p=.10) and DTA parameters remained stable (p=.21). In CS-SG animals, both systolic and diastolic volumes increased across all segments. ATA volumes rose by 27% and 17% (p=.036, p=.006), arch volumes by 37% and 14% (p=.026, p=.019), and DTA volumes by 63% and 87% (p<.001). The DTA systolic-to-diastolic ratio dropped by 87% (from 1.18±0.03 to 1.02±0.03; p<.001, Figure 4B), indicating loss of Windkessel function. In NF-SG animals, volume increases were observed in all segments, but systolic-to-diastolic ratios were preserved (Figure 4C). ATA volumes rose by 31% and 20% (p=.007, p=.012), arch volumes by 35% and 19% (p=.041, p=.047), and DTA volumes by 19% and 23% (p=.007, p=.04). The DTA ratio remained unchanged (p=.40), and terminal systolic and diastolic DTA volumes did not differ significantly from controls (p=.19, p=.053).

Table 1.

Aortic volumes (ml), diameters (mm), and lengths (mm) measured at baseline and during terminal evaluation at peak aortic systole and end diastole for three groups of animals: controls, CS-SG, and NF-SG.

Group	Time	ATA		Arch		DTA
		Systole	Diastole	Systole	Diastole	Systole	Diastole
Aortic volumes (ml)
Controls	Baseline	20.1 ± 4.0	14.3 ± 2.0	6.7 ± 1.5	4.9 ± 1.3	18.6 ± 5.8	15.1 ± 4.6
	Terminal	24.0 ± 6.3	16.1 ± 4.8	7.5 ± 2.3	5.3 ± 1.5	18.5 ± 4.3	14.7 ± 3.8
CS-SG	Baseline	16.5 ± 2.9	12.4 ± 2.4	5.4 ± 0.8	4.2 ± 0.5	17.1 ± 2.0	14.5 ± 1.6
	Terminal	20.9 ± 3.9	14.5 ± 3.0	7.4 ± 2.0	4.8 ± 0.4	27.9 ± 4.4	27.2 ± 4.0
NF-SG	Baseline	18.9 ± 8.0	14.3 ± 5.1	6.2 ± 1.6	4.9 ± 1.1	18.6 ± 2.6	16.0 ± 2.7
	Terminal	24.7 ± 8.4	17.1 ± 5.9	8.4 ± 2.7	5.8 ± 1.3	22.1 ± 2.8	19.7 ± 2.6
Aortic diameters (mm)
Controls	Baseline	21.9 ± 1.6	18.6 ± 1.4	17.3 ± 1.8	15.6 ± 1.6	12.9 ± 1.9	11.8 ± 1.5
	Terminal	23.0 ± 1.7	19.1 ± 1.6	17.4 ± 1.2	15.2 ± 1.1	13.1 ± 1.1	11.8 ± 1.2
CS-SG	Baseline	20.8 ± 1.7	17.8 ± 1.5	16.6 ± 1.1	15.1 ± 1.0	12.5 ± 0.9	11.6 ± 0.6
	Terminal	22.0 ± 1.7	18.2 ± 0.6	18.3 ± 1.9	16.4 ± 0.8	16.5 ± 0.8	16.4 ± 0.9
NF-SG	Baseline	21.0 ± 3.0	18.9 ± 2.5	16.8 ± 1.7	15.3 ± 1.6	12.5 ± 0.9	11.8 ± 0.8
	Terminal	23.5 ± 3.5	19.5 ± 3.4	18.2 ± 2.2	16.7 ± 2.2	14.4 ± 1.1	13.4 ± 1.2
Aortic lengths (mm)
Controls	Baseline	32.8 ± 4.4	30.7 ± 1.6	14.4 ±3.5	10.8 ± 2.8	113.3 ± 20.2	113.2 ± 20.3
	Terminal	34.7 ± 4.5	31.5 ± 4.4	13.5 ±4.3	11.9 ± 2.7	114.3 ± 14.6	111.9 ± 14.9
CS-SG	Baseline	30.4 ± 3.5	28.1 ± 4.0	11.5 ±1.8	11.0 ± 2.4	113.3 ± 12.7	113.2 ± 11.4
	Terminal	33.6 ± 3.0	30.7 ± 4.7	13.1 ±2.1	10.9 ± 1.6	109.8 ± 13.2	109.2 ± 13.1
NF-SG	Baseline	33.8 ± 4.6	32.3 ± 6.1	13.9 ± 2.6	13.1 ± 1.8	123.6 ± 11.4	121.0 ± 11.3
	Terminal	36.1 ± 4.7	35.3 ± 7.7	14.9 ± 2.3	12.3 ± 0.7	118.6 ± 17.2	118.0 ± 15.6

Figure 4.

Ratios of systolic to diastolic volumes in the ATA, aortic arch, and DTA for A) control animals (N = 6), B) CS-SG (N = 8), and C) NF-SG groups (N = 4). “B” and “T” indicate baseline and terminal events, respectively. Whiskers extend to 1.5 standard deviations. Paired two-tailed t-tests were used to compare baseline to follow-up, with ** denoting p < .001.

3.3. Aortic diameters

In control animals (Table 1, Figure 5A), aortic diameters remained stable across all segments (p=.099–.21). In CS-SG animals, ATA diameters showed no significant change (p=.22, p=.40), but diameters just distal to the left subclavian artery increased by 10% in systole (p=.023) and 8% in diastole (p=.033), with no change in diameter ratio (p=.55, Figure 5B). In the DTA, where the stent-grafts were deployed, diameter remained fixed at ~16mm across the cardiac cycle, leading to a 93% reduction in systolic-to-diastolic ratio (from 1.08±0.03 to 1.01±0.01; p<.001), indicating near-total loss of Windkessel function. No differences in systolic-diastolic diameter change or volumetric pulsatility were observed between CS-SG segments reinforced with balloon-expandable stents and those supported by the stent-graft alone. In NF-SG animals, systolic ATA diameter increased by 12% (p=.004), and the diameter ratio rose by 93% (p=.020, Figure 5C). No significant changes were seen just distal to the left subclavian artery (p=.063, p=.19). In the DTA, systolic diameter increased by 15% (p=.048), but the systolic-to-diastolic ratio remained unchanged (p=.80), supporting preserved Windkessel function. DTA diameters at the terminal time point did not differ significantly between NF-SG and control animals (p=.11 systole, p=.066 diastole).

Figure 5.

Ratios of systolic to diastolic diameters in the ATA, aortic arch, and DTA for A) control animals (N = 6), B) CS-SG (N = 8), and C) NF-SG (N = 4) groups. In each split violin plot, the left side represents baseline measurements and the right side shows data from the terminal evaluation. The I-shaped box within each plot indicates the median and the 25^th and 75^th percentiles. Paired two-tailed t-tests were used to compare baseline to follow-up with ** denoting p < .001, and * denoting p < .05.

3.4. Aortic lengths and axial stretch over the cardiac cycle

In control animals (Table 1, Figure 6A), systolic ATA length increased by 6% (p=.042), while other segment lengths and axial stretch remained unchanged (p=.21–.94). In CS-SG animals, ATA length increased by 11% in systole (p=.004) and 9% in diastole (p=.005), but axial stretch did not change (p=.37, Figure 6B). Arch systolic length increased by 15% (p=.020), with no change in diastolic length or stretch (p=.87, p=.10). DTA length decreased by 3%, significant only in diastole (p=.016). In NF-SG animals, lengths of all segments and axial stretch remained stable (p=.11–.85 and p=.36–.52, respectively, Figure 6C).

Figure 6.

Axial stretch over the cardiac cycle in the ATA, aortic arch, and DTA for A) control animals (N = 6), B) CS-SG (N = 8), and C) NF-SG (N = 4) groups. Boxes bound 25^th and 75^th percentiles, whiskers extend to 1.5 standard deviations, and mean is marked with a hollow square. Paired two-tailed t-tests were used to compare baseline to follow-up.

4. DISCUSSION

This study demonstrates that CS-SGs profoundly alter aortic mechanics and cardiac load in a porcine model, leading to marked reductions in pulsatility, loss of Windkessel function, and associated changes consistent with increased LV workload. In contrast, the compliant NF-SG preserved physiological aortic behavior across all measured parameters, providing direct mechanistic evidence linking stent-graft stiffness to adverse ventricular-arterial coupling. TEVAR has become the preferred treatment for many aortic pathologies and traumatic injuries; however, its long-term consequences remain incompletely understood, particularly in younger patients with decades of expected survival²⁸. Recent clinical studies have reported increased vascular stiffness, LV mass, and reduced exercise tolerance in patients implanted with stiff aortic prostheses^{8,9,16–20,29}. Our analysis of 20 trauma TEVAR patients followed over five years, revealed a marked increase in LV mass - from 138.5 ± 39.6 g to 173.5 ± 50.1 g - at a rate of 10.03 ± 12.79 g/year²². In contrast, LV characteristics in control patients remained stable. We also observed a dramatic rise in hypertension prevalence in the TEVAR group, increasing from 5% to 50%, and a 2.4-fold faster rate of aortic growth²². While these and other studies²³ have highlighted the long-term consequences of stent-graft stiffness, key limitations remain. Baseline differences in demographics and risk factors between TEVAR and control patients introduce potential confounders, particularly in the absence of pre-implantation imaging. Clinical imaging also lacks sensitivity to functional aortic changes such as loss of Windkessel function, as repeat multi-phase cardiac CT is rarely feasible. This study addressed these gaps using a controlled porcine model with true baseline imaging and longitudinal multi-phase CTA. We also evaluated a novel Windkessel-preserving stent-graft¹² and its potential to attenuate LV and aortic remodeling after TEVAR.

Our data show that LV mass in swine treated with CS-SG increased at 4.83 ± 2.34 g/month (~58 g/year), over 70% faster than in controls and over five times the rate previously reported in TEVAR patients over five years²². This accelerated remodeling is likely influenced by somatic growth in juvenile swine, in contrast to adult humans where LV mass remains largely stable. Our findings align with prior large animal studies on aortic stiffening: in swine, external wrapping of the aortic arch to reduce compliance significantly increased systolic and pulse pressures, resulting in a 40% increase in LV mass within 60 days⁵, and Wallstents implanted into the DTA altered coronary blood flow³⁰. In dogs, stiff conduits impaired cardiac function, increasing afterload and myocardial oxygen demand^21,31,32. In goats, endovascular repair amplified reflected pressure waves, correlating with local enlargement at the stent-graft site³³.

We also observed significant changes in aortic geometry and loss of Windkessel function following CS-SG implantation. In control animals, systolic-to-diastolic volume ratios in the ATA and arch were 1.40–1.50 and 1.39–1.43, respectively, with corresponding diameter ratios (1.18–1.21 and 1.11–1.15) reflecting preserved compliance. After CS-SG implantation, these ratios in the ATA and arch remained largely unchanged, though systolic and diastolic diameters increased – suggesting compensatory dilation in proximal segments. In contrast, the DTA, where the graft was placed, showed marked changes: the volume ratio dropped by 87% (from 1.18 to 1.02), and diameter pulsatility decreased by 93% (from 1.08 to 1.01), indicating near-complete loss of Windkessel function. Axial stretch in the ATA and arch remained stable, despite modest elongation. These results confirm that stiff stent-grafts severely impair local compliance and promote LV remodeling through increased afterload and altered hemodynamics.

Importantly, our findings demonstrate that LV remodeling can be attenuated by a compliant, Windkessel-preserving stent-graft. Prior bench-top testing of the NF-SG showed preservation of pulse pressure and pulse wave velocity, with only a 13–20% reduction in mid-graft distensibility, compared to 82% with CS-SG¹². In vivo, NF-SG maintained physiological waveform profiles and pulse wave velocity with minimal distortion³⁴. Intravascular ultrasound showed preserved pulsatility at the stented site – a finding supported by our current multi-phase CTA data. Despite a small sample size, NF-SG implantation preserved DTA systolic-to-diastolic volume (1.12 vs. 1.17 at baseline) and diameter ratios (1.08 vs. 1.07), indicating maintained local compliance. This preservation likely underlies the attenuated LV growth rate (2.80 ± 1.84 g/month), which was comparable to controls (2.83 ± 2.46 g/month) and provides preliminary but encouraging evidence of reduced LV remodeling with a compliant stent-graft. The extent to which these effects translate to specific clinical pathologies may differ, as aneurysmal disease and dissection create distinct mechanical coupling conditions. Nevertheless, the physiologic principle that preserving aortic compliance reduces ventricular-arterial load is expected to apply across these scenarios.

Study Limitations.

While our findings offer encouraging early evidence that compliant, Windkessel-preserving stent-grafts may mitigate adverse cardiac remodeling, they must be interpreted in light of key limitations. First, the swine model did not include underlying aortic pathology, which may influence device behavior. Although this allowed controlled comparisons in healthy vessels, future studies should assess performance in more clinically relevant disease models. Second, the CS-SG configuration was designed to represent a high-stiffness, low-pulsatility benchmark rather than a quantitatively validated replica of a specific commercial thoracic endograft at late follow-up. While this approach captures the physiologic consequence most relevant to ventricular-arterial coupling (elimination of cyclic aortic deformation) it may overestimate stiffness contrasts relative to some clinical TEVAR configurations. Additionally, the treated aortic length and use of balloon-expandable reinforcement differed between CS-SG and NF-SG groups, which precludes direct conclusions regarding material-specific performance and reinforces that the comparison should be interpreted strictly within a mechanistic framework. Moreover, the conventional graft used here was PTFE-based, and whether similar ventricular remodeling responses occur with other stent-graft materials or stent architectures (e.g., polyester-based grafts or alternative stent designs) remains to be determined. Third, the NF-SG cohort was small due to the early prototype nature of the device and its experimental delivery system, which likely contributed to procedural variability; larger cohorts are needed to confirm these effects and enhance statistical power. Fourth, follow-up was limited to four months. In humans, LV remodeling often takes years to develop²², particularly in healthier adults. Although rapid growth in juvenile swine enables early detection of remodeling, longer-term studies are required to evaluate the durability of compliant stent-graft benefits and their implications for long-term cardiovascular health. Finally, prior experience in endovascular therapy has shown that changes in stent-graft materials or architectures can introduce important clinical challenges, including durability, sealing performance, and long-term device integrity. The present study was not designed to assess these aspects, and any translation of compliant stent-graft concepts will require rigorous evaluation of mechanical durability, fatigue resistance, and clinical efficacy in addition to hemodynamic performance.

What this paper adds: Conventional stent-grafts used in TEVAR are significantly stiffer than native aortic tissue and can impair aortic compliance and Windkessel function. In a juvenile swine model, compliance-matched stent-grafts preserved aortic biomechanics and mitigated left ventricular growth compared to conventional stiff grafts. Longer-term follow-up in preclinical models is needed to confirm durable biomechanical and cardiac benefits of the compliant stent-graft.