The study was conducted in accordance with the Guide for Care and Use of Laboratory Animals and the ARRIVE guidelines (Animal Research: Reporting of In Vivo Experiments). The study protocol was approved by the Institutional Animal Care and Use Committee (IACUC) and the Ethical Committee of the American Preclinical Services (Minneapolis, MN, United States) (Protocols IQI001-IS02, IQI005-IS02) before study initiation and performed in regulation and compliance to guidelines.

Sixteen Suffolk sheep with a body weight of approximately 60 kg were used in this study: thirteen were implanted with a restorative Xeltis coronary artery bypass graft (XABG), and three were implanted with an SVG. Only one graft was implanted per animal to connect the descending aorta to the LAD to bypass the proximal LAD, as previously described (Shofti et al., 2004). All animals were given aspirin and clopidogrel. Implantation, angiography, and termination were performed under general anesthesia using 2–8 mg/kg propofol and 0%–5% isoflurane. The restorative SDVGs or autologous SVGs were implanted under cardiopulmonary bypass, followed by ligation of the LAD at a few millimeters upstream of the graft distal anastomosis.

Animals were initially intended to be terminated after 6-month survival, and the safety of the restorative SDVG was to be compared to the SVG control. However, at 6 months, there were few complications, and angiography showed promising results. Therefore, the study was extended to 12 months. Humane euthanasia was ensured by intravenous administration of an appropriate barbiturate per animal study site protocol.

Serial angiography was performed at the time of implant and 1-month, 3-month, 6-month, 9-month (XABG only), and 12-month follow-up. Quantitative coronary angiography with edge detection and videodensitometry, volumetric flow velocity, and optical coherence tomography (OCT) was performed at 3 months, 6 months, 9 months, and 12 months according to standard procedures described elsewhere (Wu et al., 2022) and was analyzed by an independent core laboratory (CORRIB Corelab, National University of Ireland Galway, Galway, Ireland).

The Student’s t-test was used to analyze the significance of differences for continuous variables with normal distributions, whereas comparisons of variables with non-parametric distribution were assessed by the Kruskal–Wallis test. All analyses were performed using JMP software (version 15.0, SAS Institute, Inc., NC, United States). A p-value of <0.05 was considered statistically significant.

Supplementary Figure S2 summarizes the follow-up of Suffolk sheep (n = 16) with either XABG (N = 13) or SVG control (N = 3). Two animals with XABG did not survive the perioperative phase. Two animals with XABG were electively sacrificed after the 3-month angiographic assessment. One graft was occluded, and the other revealed distal stenosis and was sacrificed for exploratory learning (graft patent). Subsequently, two more animals with XABG were electively sacrificed as a result of the 6-month angiography findings. One graft showed focal aneurysmal change, and another showed dissection with neointimal perforation due to OCT catheterization, and both were patent at sacrifice. More details were described previously (Ono et al., 2023). The remaining seven animals with XABG survived until the exploratory 12-month time point; however, two unscheduled deaths occurred at 8.5 months and 9.5 months due to acute myocardial infarction. Finally, five animals survived with patent XABGs until the 12-month follow-up. All three SVG controls were patent at 12 months.

XABGs were generally characterized by uniform lumen profiles and stable mean diameters over time, while SVG controls had a more irregular profile with progressive dilatation (Figure 2). XABG flow velocity increased between 1 month and 3 months and remained stable with time, while flow velocity for SVG remained low through the study follow-up period. Note that detailed angiographic analysis has been reported separately (Ono et al., 2023; Poon et al., 2023).

Explanted XABGs and SVGs underwent microCT image acquisition, confirming angiographic findings with a generally uniform lumen for XABGs and a more irregular and dilated luminal profile for SVGs (Figure 3). The graph in Figure 3 and Supplementary Table S3 shows the average result of morphometric analysis of the whole graft length. The intima-graft or intima-media border area, which represents a cross-sectional area of the vessels, was much greater in SVG than in XABG. More extensive neointima formation (i.e., neointimal area) can be observed in SVG than XABG [mean ± standard error (SE); SVG: 18.32 ± 2.41 mm² vs. XABG: 7.26 ± 0.36 mm², p = 0.0253], which is more prominent in the distal segment of the SVG group due to thrombus formation (Supplementary Table S4). Percent area stenosis was comparable between the two groups (41.42% ± 6.15% vs 55.51% ± 4.76%, p = 0.3).

Grossly, all XABG devices demonstrated epicardial adhesions and coverage by epicardial fat and fibrous tissue. The grafts were fully expanded and remained uniformly circular without evidence of graft kinking. One XABG showed ectasia, while another showed the presence of an aneurysm in the middle of the graft.

The polymer absorption and matrix deposition increased over time (Table 1). The area of resorption was replaced by smooth muscle cells surrounded by proteoglycan-rich collagen matrix with angiogenesis (Figure 4). At 12 months, greater degradation of the polymer with focal areas of polymer resorption was seen than at the earlier time points (Supplementary Figure S3).

The inflammation increased over time up to 9 months and decreased thereafter; the score was not significant but trending (p = 0.0975) (Table 1). The inflammatory cells consisted predominantly of macrophages and foreign body giant cells and were almost always accompanied by angiogenesis. This is consistent with CD68 macrophage stains performed in earlier studies on the same polymer (Cramer et al., 2022). Note that macrophage IHC stains of the polymer were not performed in the current study due to plastic embedding limitations. Throughout the study, there were no indications of local or systemic adverse tissue response to the XABG conduit.

Endothelialization of the XABG surface was incomplete in early (up to 6 months) sacrificed animals with luminal thrombus, while in 12-month animals, almost complete endothelialization was observed (Table 1; Figure 5). SEM (N = 5 at 12 months) confirmed near confluent coverage with endothelial-like cells with focal thrombus, observed only in one of five grafts, and TEM (N = 1 at 12 months) showed endothelial cells on the luminal surface of the XABGs (Figure 5) with the presence of endothelial junctions. Paraffin-embedded sections of the carefully removed neointimal tissue from XABG (N = 2 at 12 months) showed positive vWF staining consistent with endothelial cell coverage (Figure 5). The neointima was rich in aSMA-positive cells consistent with a smooth muscle cell phenotype (data not shown).

Histomorphometry was performed in the animals with XABG (N = 5, 15 sections) and SVG (N = 3, 9 sections) (Supplementary Table S5). Percent area stenosis of both grafts was similar (mean ± SE; XABG 53.7% ± 5.1% vs SVG 41.4% ± 6.6%, p = 0.14); however, SVG showed greater vessel cross-sectional area than XABG (intima-media border area in SVG vs intima-graft border area in XABG; 60.8 ± 5.8 mm² vs. 12.6 ± 4.5 mm², p < 0.0001). The magnitude of neointima formation (i.e., neointimal area) was greater in SVG than XABG (23.7 ± 5.0 mm² vs. 6.6 ± 3.9 mm², p < 0.0001), especially in the distal segments of SVG due to luminal thrombus formation (Figures 6A–H, J; Supplementary Table S6). When comparing vessel size of graft sections from the proximal, mid, distal, and distal anastomosis, SVG was significantly dilated compared to XABG (Figure 6I). Therefore, the percent area stenosis of the two groups appears similar, although there is more neointimal tissue formation on the SVGs than on XABGs (Supplementary Tables S5, S6). These histomorphometry results are consistent with the results of microCT morphometric analysis.

Although prosthetic vascular conduits were introduced almost 70 years ago, no small-diameter (<6 mm) synthetic conduit has been approved for use as a CABG. Biological materials (e.g., bovine mammary arteries) were also used for SDVG; however, no studies showed successful results (Alexander et al., 2005). Although ePTFE has been successfully surgically implanted in patients with peripheral vascular disease and for hemodialysis access (Tzchori et al., 2018), rapid failure of ePTFE has been reported in CABG surgery (Chard et al., 1987). Importantly, ePTFE grafts are known to never fully endothelialize beyond ∼2 cm of the anastomotic site (Chard et al., 1987). To the best of our knowledge, this study is the first to evaluate a 4 mm SDVG, 15 cm in length, in an ovine model with a 12-month follow-up. Even with less challenging arterial bypass models (e.g., carotid interposition, femoral-femoral, or iliac-femoral) and relatively short graft lengths (typically 5–7 cm), patency of small-diameter synthetic grafts (≤4 mm) has reported 0%–25% patency in studies extending beyond 1 month follow-up (Fang et al., 2021). Our study, on the other hand, achieved 73% patency at 6 months, with near-complete endothelialization at 1 year, implanted in a low-flow environment of CABG.

The vein-graft-to-native-artery anastomosis mismatch in both diameter and compliance are known contributors to SVG failure (Ghista and Kabinejadian, 2013). Various approaches to refining the patency rate of SVG have been conducted. Only external support devices for SVGs have gained momentum in clinical utility to date (Taggart et al., 2015). The experimental work by Taggart et al. demonstrated that by using an external stent to prevent dilation of SVGs, there is a significant reduction in intimal hyperplasia by attenuating hemodynamic disturbances caused by irregular lumen geometry (Taggart et al., 2015). However, the VEST trial failed to show the superiority of the device in the primary endpoint, which was the neointimal hyperplasia area as assessed by intravascular ultrasound at 12 months (Goldstein et al., 2022). XABG showed mild neointimal growth with surface endothelialization and less diameter and compliance mismatch, which may result in better long-term outcomes (Zilla et al., 2007; Taggart et al., 2015).

Thrombosis remains a major cause of SDVG failure. Various types of modifications have been performed, and most have embraced those possessing antithrombotic properties, such as agents (e.g., heparin, nitric oxide, tissue-type plasminogen activator, and thrombomodulin) or recellularization by in vitro repopulation of a scaffold with endothelial cells (Fang et al., 2021; Tatterton et al., 2012). However, studies with high clinical relevance (use of large animal models) are still limited in number. The fundamental reason for thrombosis is the failure of the graft lumen to heal with a confluent endothelium. Although ePTFE grafts never fully endothelialize beyond ∼2 cm of the anastomotic site (Zilla et al., 2007), there are reports of small-diameter ePTFE grafts implanted in humans following pre-seeding with endothelial cells prior to implantation that can achieve chronic patency when used as CABG even as late as 27.7 months (Laube et al., 2000). Thus, if it were possible to line the lumen of a synthetic graft with endothelium, the thrombosis problem could be avoided. ePTFE (and other synthetic) graft architecture inhibits transmural endothelial cell migration, which is needed for the natural development of a confluent endothelium (Zilla et al., 2007; Bezuidenhout et al., 2002). Transmural microvessel connectivity within the graft, from the abluminal surface to the lumen, is essential to supply enough endothelial cells for a dense, confluent endothelium that prevents thrombosis (Zilla et al., 2007). Endothelial progenitor cell (EPC) recruitment is another possibility that has been shown to work with CD 34 antibody deposition on stents with limited neointimal growth at 9 months in humans (Aoki et al., 2005).

In general, the potential of fallout endothelialization in humans is considered limited compared to animal models (Zilla et al., 2007; Sánchez et al., 2018). The current sheep study does not separate fallout endothelialization from transmural ingrowth. Studies that separate both mechanisms by preventing abluminal ingrowth have demonstrated that fallout endothelialization is possible in various animal models (Talacua et al., 2015; Smith et al., 2020; Shi et al., 1994), although Pennel et al. demonstrated that endothelialization was far less when an abluminal wrap was used to prevent transmural ingrowth in a porous graft in a rat loop model (Pennel et al., 2018).

Our research has some limitations. Not all animals survived for 1 year. Repeated angiography and intravascular OCT imaging may have contributed to early failures, which could have been avoided if intravascular imaging had not been part of the experimental setup. Our study was carried out in a healthy ovine model and did not include aged and diseased states, which exist in patients who require CABG. In the future, safety and efficacy will have to be demonstrated in a heterogeneous and diseased patient population. However, to our knowledge, the ovine CABG model has never been reported to survive more than 1 year. XABG was almost completely endothelialized and had less graft/coronary anastomotic mismatch (diameter and compliance) than the control SVGs; however, the 12-month cumulative patency rate was better with SVG than with XABG.