Patient-specific hemodynamic evaluation of fenestrated physician-modified endografts in complex abdominal aortic aneurysms

Introduction

Abdominal aortic aneurysm (AAA) is a permanent dilatation of the abdominal aorta, commonly defined as a diameter of at least 3.0 cm. Degeneration of the media, chronic inflammation, and extracellular-matrix degradation weaken the wall, while increasing diameter raises wall tension and remains the principal clinical marker of rupture risk [1, 2]. Endovascular aneurysm repair (EVAR) excludes the aneurysm from systemic pressure, but standard devices may be unsuitable when the proximal landing zone is short or visceral branches are involved [3]. Fenestrated physician-modified endografts (PMEGs) extend the sealing zone while preserving target-vessel perfusion and have shown acceptable technical and patency outcomes in complex aortic disease [4–7].

The clinical importance of PMEG-related hemodynamics extends beyond the expected reduction in sac caliber. Under a fixed inflow, continuity predicts acceleration as effective flow area decreases; however, local curvature, tapering, fenestrations, branch angles, and residual cavities determine where separation, recirculation, viscous energy loss, and abnormal wall shear persist. These mechanical stimuli interact with vascular biology [8, 9]. Prior modeling has linked vortical flow and altered shear to aneurysm remodeling and thrombosis [10, 11]. Low or oscillatory wall shear can favor endothelial dysfunction and platelet residence, whereas excessive shear may also damage the wall or activate platelets [12]. TAWSS, oscillatory shear index (OSI), RRT, and ECAP therefore provide complementary spatial descriptors of low-shear exposure, directional disturbance, near-wall stasis, and a pro-thrombotic endothelial environment [13–17].

The literature does not establish these indices as direct causes of AAA progression or rupture. CTA-based simulations have demonstrated postoperative changes in branch flow, pressure loss and shear after PMEG [18], whereas longitudinal studies have reported inconsistent spatial or causal relationships between low shear, intraluminal thrombus (ILT), aneurysm growth, and rupture [19, 20]. ILT itself may reduce local wall stress while simultaneously promoting hypoxia, inflammation, and proteolytic weakening; wall inflammation and medial degeneration remain important biological determinants [21]. Thus, hemodynamics is one contributor within a multifactorial process that also includes patient-specific tissue properties and systemic factors [22, 23]. The present study is not designed to resolve this causal controversy; it evaluates changes in model-derived surrogate markers after a controlled geometric reconstruction. However, the reported relationships between hemodynamic indices and AAA progression remain inconsistent. Some longitudinal and patient-specific studies have associated low or oscillatory wall shear with ILT accumulation and aneurysm growth, whereas others have found weak, heterogeneous, or even spatially discordant relationships between these hemodynamic markers and aneurysm expansion or rupture. These discrepancies likely reflect the multifactorial nature of AAA progression, in which hemodynamic conditions interact with ILT, wall inflammation, tissue properties, and systemic risk factors. Therefore, TAWSS, RRT, and ECAP should be regarded as mechanistic surrogate markers rather than direct causal predictors of AAA progression or rupture. Importantly, the present study does not aim to resolve this causal controversy; instead, it focuses on quantifying patient-specific changes in these surrogate markers following PMEG-induced geometric reconstruction under controlled and matched loading conditions.

Existing PMEG simulations have provided valuable descriptions of flow separation, thrombus-prone recirculation, renal-branch anatomy, and vortex evolution [24–26]. However, most previous studies have focused on individual cases or limited hemodynamic descriptors, and few have systematically combined paired pre- and post-PMEG geometries with multiple complementary surrogate indices and wall-mechanical assessment. This limits the ability to distinguish intervention-related geometric effects from patient-specific anatomical variability. To address this methodological gap, we reconstructed pre- and post-PMEG CTA geometries from eight patients and applied identical inlet and outlet loading conditions within each paired comparison. We quantified velocity, pressure loss, TAWSS, RRT, and ECAP, while a representative case was further examined using FSI to evaluate wall deformation. The objective was therefore not to establish causal relationships between hemodynamic indices and AAA progression, but to determine how patient-specific PMEG reconstruction alters spatially heterogeneous flow and wall-mechanical features that cannot be characterized by aneurysm diameter reduction alone.

Materials and methods

Patient information

A total of eight patients with AAA were included in this study, as summarized in Table 1. Six cases (Cases 1, 3, 4, 6, 7 and 8) presented with primary AAA (PAAA). The remaining two patients (Cases 2 and 5) had previously undergone thoracic endovascular aortic repair (TEVAR) with untreated distal segments prior to the PMEG procedure, as illustrated in Figure 1a. It is worth noting that the primary surgical indications for PMEG encompass the standard criteria for AAA repair, specifically focusing on clinical presentation and maximum aneurysm diameter. Besides, patients with PAAA and those post-TEVAR were studied as a single cohort, as both groups were treated with PMEG and exhibited similar pre-procedural vascular characteristics. Furthermore, all subjects underwent endovascular repair using fenestrated PMEG at Nanjing Drum Tower Hospital, detailed patient profiles are summarized in Table 1. At the 1-year follow-up, no endoleaks of any type were observed across the eight cases, and all branch stents remained patent without occlusion or displacement. Furthermore, no adverse events, such as chest pain, abdominal pain or spinal cord ischemia, were reported.

TABLE 1. The specifics of patient including gender, age, diameter of aortic aneurysm (ϕ, mm), symptoms (a: abdominal pain; b: resistant hypertension; c: ischemia; d: shock).

PMEG procedure and comparator scope

Conventional infrarenal EVAR places an unfenestrated main body below the lowest renal artery and relies on an adequate, relatively nondilated infrarenal neck for proximal fixation and sealing; iliac limbs then complete aneurysm exclusion. In a juxtarenal or pararenal aneurysm, however, a short, dilated, or severely angulated neck may provide an inadequate seal. Deploying a conventional graft too distally can leave proximal type Ia endoleak, migration, or persistent sac pressurization, whereas moving an unfenestrated graft above the renal arteries risks occluding the renal or mesenteric branches. Fenestrated PMEG was therefore selected when visceral-branch involvement made standard infrarenal EVAR unsuitable. Preoperative CTA centerline planning defined the longitudinal position and circumferential orientation of each target ostium. Corresponding fenestrations were created and reinforced in a commercially available graft, which was resheathed, aligned, and deployed with its proximal seal extended into a healthier suprarenal or visceral aortic segment (Figure 1a). The target vessels were cannulated through the fenestrations and supported with bridging stents as clinically required, after which the distal components were deployed to exclude the aneurysm. This patient-specific reconstruction provides a longer effective proximal sealing zone while preserving renal and mesenteric perfusion. Theoretically, more complete sac exclusion and a smoother, lower-volume patent conduit reduce residual sac recirculation, near-wall stasis, and pressure transmission and produce more favorable shear and wall loading than attempting conventional infrarenal EVAR in anatomy without an adequate neck. Thus, for the complex anatomies studied here, PMEG has an anatomical and hemodynamic advantage over conventional infrarenal EVAR; the absence of a matched EVAR group means that the present data do not establish universal clinical superiority across all anatomies.

Modeling method

CTA data (512 × 512 matrix; 1 mm slice thickness) were segmented in Mimics to reconstruct patient-specific pre- and post-PMEG lumens. Where adjacent structures obscured the aortic boundary, the contour was corrected by approximately one pixel (0.4 mm), as illustrated in Figure 1b. Surface defects were repaired and smoothed in Geomagic Wrap to generate the analysis geometries shown in Figure 1c. For the Case 7 FSI model, CTA did not provide a reliable spatial map of wall thickness; therefore, a uniform nominal 2.0 mm wall was generated from the lumen surface with the Rhino “Offset Surface” command (Figure 1d). Thickness was not varied between normal and aneurysmal segments, and ILT and calcified plaque were not modeled as separate material domains. These assumptions should be considered when interpreting the deformation results.

Meshing and computational domain

The aneurysm analysis region was defined as shown by the blue wireframe in Figure 2a; in post-PMEG models, the excluded distal sac was identified separately from the patent flow lumen. Inlets and outlets were extended to reduce backflow-related numerical instability. Mesh sensitivity was tested in pre-PMEG Case 1 with 1.88, 3.54, and 5.46 million elements. Because the 3.54 million element solution differed little from the fine mesh, comparable settings were used for the remaining models (3.35–5.46 million elements). Five prism layers were retained at the wall to resolve near-wall velocity gradients.

Boundary conditions

A common time-varying inlet waveform with a 0.952 s cardiac period was applied to all models (Figure 2b) [27]. Three-element Windkessel (RCR) conditions represented downstream characteristic resistance, arterial compliance, and peripheral resistance at the outlets (Figure 2c) [28, 29]. The same physiologic inlet waveform and outlet parameter set were retained for each pre/post comparison to standardize systemic loading and isolate the hemodynamic consequences of PMEG-induced geometric remodeling. Accordingly, the simulated pre/post differences should be interpreted as anatomy-dependent responses to the reconstructed geometry under matched loading conditions, rather than as direct predictions of changes in postoperative systemic circulation or patient-specific physiology. All rigid-wall CFD surfaces were no-slip. Consequently, the reported pre/post differences should be interpreted as geometry-driven changes under standardized loading, not as patient-specific predictions of postoperative systemic physiology. The governing RCR equations are provided in Appendix A.

Numerical simulation

Transient simulations were performed in ANSYS Fluent. Blood was treated as an incompressible Newtonian fluid (density 1,060 kg/m³; dynamic viscosity 3.5 × 10⁻³ Pa·s) with laminar flow [13, 30, 31]. For the Case 7 FSI analysis, the wall was modeled as homogeneous, isotropic, and linearly elastic (density 1,500 kg/m³; Young’s modulus 6 × 10⁶ Pa; Poisson’s ratio 0.45) [32, 33]; the same inlet waveform and a zero-pressure outlet were used. A second-order implicit scheme, 0.01 s time step, maximum 20 sub-iterations, and RMS residual criterion of 10⁻⁵ were applied. Simulations were continued for approximately 10 cycles to reach cycle-to-cycle convergence, and the subsequent 10 cycles were used to calculate time-averaged indices. Full governing Equations A1-A9 and index definitions are listed in Appendix A.

Hemodynamic parameters

Wall shear stress (WSS) is the tangential traction exerted by blood on the endothelium. Persistently low WSS can indicate poor near-wall washout and has been associated with atherogenic endothelial signaling, whereas very high WSS may reflect localized jets or impingement [34]. TAWSS averages WSS magnitude over one cardiac cycle and summarizes the cumulative shear exposure. In this study, low TAWSS is interpreted as a surrogate marker of disturbed flow and potential ILT-prone conditions, not as a stand-alone predictor of thrombosis or rupture [35–43]. OSI measures how strongly the WSS vector reverses during the cardiac cycle. Values approaching 0 indicate predominantly unidirectional shear, whereas values approaching 0.5 indicate marked directional oscillation. High OSI can accompany disturbed flow and increased endothelial permeability [42, 43]. RRT combines shear magnitude and oscillation to estimate the persistence of blood-borne particles near the wall. Higher RRT denotes slower near-wall exchange and a greater opportunity for platelet or macrophage adhesion; it is therefore used as a surrogate of flow stasis and ILT susceptibility [44–46]. ECAP is the ratio of OSI to TAWSS proposed by Di Achille et al. [47]. It highlights regions where oscillatory flow coincides with low shear and is interpreted as a spatial marker of a pro-thrombotic endothelial environment. TAWSS, RRT, and ECAP were summarized as area-averaged values over the defined aneurysm region; their equations and variables are provided in Appendix A.

Statistical analysis

Continuous variables are reported as mean ± standard deviation, with individual patient values shown where applicable. Pre- and post-PMEG measurements were treated as paired observations. Normality of the paired differences was assessed with the Shapiro-Wilk test; a two-sided paired t-test was used for approximately normal differences and the Wilcoxon signed-rank test otherwise. Exact P values are reported, and P < 0.05 was considered statistically significant. The analyses were applied to the eight patient-level values rather than to surface nodes or time steps, thereby avoiding pseudo-replication. Given the small exploratory cohort, no multiplicity correction was applied and statistical significance was interpreted together with effect direction, inter-patient heterogeneity, and the model assumptions.

Results and discussion

Grid sensitivity analysis

Case 1 was used for mesh sensitivity testing. Area-averaged aneurysm TAWSS changed from 2.36 dyn/cm² on the 1.88 million element mesh to 2.48 and 2.52 dyn/cm² on the 3.54- and 5.46-million-element meshes, respectively (Table 2). The 1.6% difference between the middle and fine meshes supported use of the 3.54 million element configuration as the reference for the remaining models.

TABLE 2. Results of grid sensitivity analysis of the vascular model of Case 1.

Flow patterns

Peak-systolic flow was examined in all cases (Figure 3a). Before PMEG, the expanded aneurysm lumen produced slow flow, separation, and recirculating vortices, particularly near the neck (Figure 3b). Higher velocities occurred at the aortic inlet, iliac outlets, tight curvatures, and abrupt changes in caliber. These patterns are consistent with continuity and momentum balance: local contraction accelerates flow, whereas sudden expansion and curvature promote separation and energy loss. Because the simulations assumed laminar flow, these structures are described as disturbed or vortical rather than turbulent.

After PMEG, flow was directed through a smaller and generally smoother conduit, so an increase in velocity under the fixed inlet waveform was expected. Therefore, the increase in mean velocity should not be regarded as an independent demonstration of a therapeutic benefit of PMEG. Rather, it provides a quantitative reference for evaluating how the patient-specific reconstruction redistributes flow and alters residual recirculation and other local hemodynamic features. The value of the patient-specific analysis lies in the magnitude and spatial heterogeneity of this response. Case 4 showed the largest velocity increase (0.993 m/s), whereas Case 8 changed by only 0.017 m/s because its preoperative flow was already relatively fast and its geometric alteration was modest (Table 3). These outliers demonstrate that the cohort-average direction cannot be assumed to represent every reconstructed anatomy. Figure 3c shows that mean velocity increased from approximately 0.8 m/s before PMEG to 1.5 m/s after PMEG (P = 0.0021). This result quantifies the expected acceleration caused by lumen remodeling under matched loading; it should not be interpreted alone as evidence of clinical benefit. Its principal use is as a reference against which residual recirculation, branch jets, and case-specific departures such as Case 8 can be identified.

TABLE 3. The calculated average velocity of the aorta (V, m/s) and pressure drop (ΔP, Pa) of the blood flow model for these cases in pre- and post-PMEG blood vessel.

Pressure

Pressure drop (ΔP) was defined between the aortic root and the aneurysm analysis region (Figure 4a). In both pre- and post-PMEG models, pressure decreased from the inlet toward the iliac outlets (Figure 4b). Under the matched boundary conditions and rigid-wall assumption, changes in ΔP primarily reflect geometry-dependent separation, recirculation, and viscous losses; they do not represent a measured improvement in systemic blood pressure or vascular compliance. Thus, the observed reduction in pressure drop should be interpreted as a geometry-dependent computational outcome under controlled loading, rather than evidence of improved systemic blood pressure or a direct causal effect of PMEG independent of anatomical reconstruction. The largest reduction in ΔP occurred in Case 4, whereas Case 6 changed little and Case 8 showed a small increase (Table 3). The Case 8 result is clinically relevant as a model-derived flag: despite aneurysm exclusion, the reconstructed path did not reduce pressure loss under standardized loading. Such residual or adverse features may motivate closer imaging review or comparison of alternative virtual configurations, although this proposed use requires prospective validation. Across the cohort, mean ΔP decreased after PMEG (P = 0.0183; Figure 4c). This group-level result is consistent with reduced geometry-related energy loss, but the inter-patient spread and the Case 8 reversal show why statistical significance alone is insufficient to establish clinical importance.

TAWSS and WSS

Figures 5a,b show similar spatial patterns for TAWSS and peak-systolic WSS. Before PMEG, extensive low-shear regions occupied the aneurysm sac and inner curvatures, with adjacent high- and low-shear zones indicating disturbed flow. After PMEG, low-TAWSS regions generally became less extensive and the shear field along the patent lumen became more continuous, although focal peaks persisted at curvatures, diameter transitions, and outlets. Thus, the images add information not contained in diameter or mean velocity alone by locating residual low-shear zones and focal high-shear jets.

Patient-level values are summarized in Table 4. Case 1 showed the largest post-PMEG increase in mean TAWSS and WSS, whereas Case 8 showed the smallest changes. In Case 8, postoperative TAWSS remained above the cited physiological reference range, identifying a residual high-shear exposure that would be obscured by the cohort average. Mean TAWSS and WSS increased after PMEG (PTAWSS <0.0001; PWSS = 0.0002; Figures 5c,d). These indices are interpreted as hemodynamic surrogates; no threshold in this study has been validated to predict rupture or thrombosis.

TABLE 4. The calculated mean TAWSS (dynes/cm²), WSS (dynes/cm²), RRT (Pa⁻¹) and ECAP (Pa⁻¹) of the aneurysm sac wall for pre-PMEG and post-PMEG cases.

RRT and ECAP

RRT and ECAP were used to map near-wall stasis and the combined exposure to low, oscillatory shear (Figures 6a,b). Before PMEG, high values occupied much of the aneurysm sac. After exclusion, both indices decreased along most of the patent abdominal aorta, but focal elevations remained at selected curvatures and branch transitions. These residual spatial features, rather than the expected direction of the mean change, are the potentially useful patient-specific output.

Table 4 demonstrates marked inter-patient differences in the magnitude of change. Mean RRT and ECAP decreased after PMEG (PRRT = 0.0008; PECAP = 0.0063; Figures 6c,d), consistent with less modeled flow stasis and a less adverse shear environment. These changes do not prove a reduction in clinical thrombotic events. Instead, the maps can generate surveillance hypotheses by indicating where slow washout or unfavorable shear persists despite technically successful aneurysm exclusion.

Fluid–structure coupling

Case 7 was selected as a representative model for exploratory FSI analysis. Across the four cardiac phases in Figure 7, computed deformation was concentrated in the mid-aneurysm region before PMEG and was lower in adjacent normal-caliber segments. After PMEG, deformation of the reconstructed wall was substantially smaller under the same imposed loading. This spatial pattern reflects the combined influence of lumen geometry and the assigned wall model; it does not identify tissue weakness independently because wall thickness and material properties were uniform.

During early systolic acceleration (Figure 7a), deformation began in the mid-aneurysm wall, while the post-PMEG model changed little. At peak systole (Figure 7b), the pre-PMEG model reached its maximum computed deformation and the postoperative maximum shifted distally. Deformation decreased during deceleration (Figure 7c) and approached the initial configuration in diastole (Figure 7d). These phase-dependent results quantify the mechanical consequence of the reconstructed geometry, but the single-case, uniform-wall model cannot be used to infer patient-specific rupture risk.

The FSI WSS contours in Figure 8 show the same phase-dependent pattern as the rigid-wall analysis: high shear occurred near high-velocity jets and pronounced curvature, while the expanded mid-sac remained dominated by low shear. As inlet flow varied through the cycle, the extent and magnitude of these regions migrated in space. The coupled model therefore adds a qualitative view of how wall motion and shear co-evolve, while remaining subject to the simplified wall assumptions.

During early acceleration, pre-PMEG WSS remained near zero across much of the sac, with higher values at the proximal transition and distal outlet (Figure 8a). At peak systole, localized WSS increased along the lateral wall and outlet (Figure 8b); it then decreased during deceleration (Figure 8c) and became uniformly low in diastole (Figure 8d). The postoperative model showed less extensive low-shear exposure within the patent lumen. This direction of change is consistent with the paired CTA-CFD study by [18], which reported restored visceral flow, higher postoperative abdominal-aortic TAWSS, and significantly lower RRT and ECAP after PMEG. The present FSI result adds a lower computed wall deformation after reconstruction, supporting a more stable wall-mechanical response while remaining subject to the simplified wall assumptions.

Clinical interpretation and limitations

Several clinically important hemodynamic quantities cannot currently be measured directly and comprehensively in routine postoperative practice. CTA can confirm graft position, branch patency, sac exclusion, and endoleak, but it does not provide spatial distributions of WSS, RRT, ECAP, or aneurysm-wall deformation; patient-specific outlet pressure and flow waveforms are also rarely available for every branch. CFD and FSI therefore provide a necessary noninvasive means of estimating these otherwise inaccessible mechanistic variables. At present, these model-derived spatial abnormalities should be regarded as hypothesis-generating rather than clinically actionable findings; their potential use for targeted surveillance or virtual treatment planning requires prospective validation. In this cohort, the post-PMEG reduction in RRT and ECAP, the smaller low-shear region, and the lower Case 7 wall deformation provide a mechanistic basis for improved washout, a less thrombosis-prone flow environment, and more favorable wall mechanics. The agreement with the independent paired PMEG study by [18] strengthens this interpretation. Clinical follow-up provides complementary support: at 1 year, all branch stents remained patent, with no endoleak, displacement, or reported adverse event. Long-term results from a prospective physician-sponsored PMEG trial in juxtarenal AAA likewise demonstrated durable treatment with low long-term morbidity and mortality [48]. Taken together, the computational and follow-up evidence supports the favorable hemodynamic and mechanical effects of PMEG in complex anatomy. Nevertheless, the present study remains retrospective, includes eight patients and no matched conventional-EVAR group, uses standardized rather than measured patient-specific loading, assumes rigid walls in the cohort analysis, and performs FSI in one case with uniform thickness and homogeneous material properties while omitting ILT and plaque. Larger prospective comparative studies are required to quantify the extent to which these surrogate improvements reduce clinical thrombotic or rupture events.

Conclusion

This study developed a patient-specific framework for comparing pre- and post-PMEG hemodynamics in complex AAA under matched loading conditions. PMEG-associated geometric remodeling increased mean velocity, reduced pressure loss at the cohort level, reduced low-shear and high-stasis regions, and lowered computed wall deformation in the representative FSI case. These changes, together with patent branches and no endoleak or reported adverse event at 1 year, support PMEG as a favorable anatomical, hemodynamic, and wall-mechanical reconstruction for juxtarenal or pararenal anatomy unsuitable for conventional infrarenal EVAR. The lower RRT and ECAP indicate a less stasis-prone, potentially less thrombogenic environment, while reduced computed deformation indicates more stable wall mechanics. Consistency with prior paired PMEG simulations [18] and long-term clinical PMEG results [48] further supports this interpretation. Because these computational indices remain surrogate endpoints, prospective studies with measured boundary conditions, spatially resolved wall and thrombus properties, matched EVAR comparators, and longer event-based follow-up are still required to quantify the reduction in clinical thrombosis or rupture risk. Physically, these effects are primarily governed by the PMEG-induced redistribution of flow area and vascular geometry, which alters local velocity, pressure loss, near-wall shear, flow stasis, and wall loading.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Guangzhao Wang, Yangtze Normal University, China

Reviewed by: Hongyue Zhou, Jiangsu University, China

Zhang Jie, Ruijin Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, China