Polyester Vascular Graft Material and Risk for Intracavitary Thoracic Vascular Graft Infection

Prosthetic vascular graft infections of the thoracic aorta are rare but can be fatal. Our comparison of collagen- and gelatin-coated grafts showed that collagen-coated grafts were associated with increased biofilm formation and bacterial adherence in vitro and with higher rates of perioperative vascular graft infections in vivo.

DISPATCHES P rosthetic vascular graft infections (PVGIs) of the thoracic aorta occur in 1%-3% of patients, but lethality rates are >20% (1,2). Because of an aging population with multiple medical conditions, more vascular grafts are being implanted, resulting in more PVGIs. Infection often occurs during the perioperative period (3) as a consequence of inoculation with bacteria mostly originating from the patient's own skin flora. PVGIs are biofilm-associated infections in which the matrix around the bacteria impairs chances of treatment success (4). Hence, the primary aim is to prevent perioperative infections by identifying risk factors, such as type of prosthesis.
The few comparative studies reported have focused mostly on use of antibiotic-bonded grafts to reduce the risk for PVGIs in vitro and in vivo (5)(6)(7). However, lack of approval by regulatory authorities, reduced commercial availability, and lack of longterm follow-up data on infection-free survival should be considered (8)(9)(10). Furthermore, selection pressure from the use of topical antibiotics might lead to resistance. Staphylococcus aureus has been shown to colonize rifampin-bonded grafts 7 days after implantation (10). Hence, implanted grafts are usually coated with proteinaceous solutions only, allowing for quick integration into host tissue. We compared susceptibility of 2 graft materials to biofilm formation in vitro and rates of infections in vivo.
For our in vitro study, we compared the susceptibility of 2 thoracic vascular woven polyester grafts with different coatings-collagen (collagen graft, InterGard Hemabridge, https://www.getinge.com) and gelatin (gelatin graft, Terumo Aortic, Gelweave, https://terumoaortic.com)-to biofilm formation. The collagen graft is coated with a highly purified form of cross-linked bovine type 1 collagen. The gelatin graft is coated with a modified mammalian gelatin. Gelatin is derived mainly from type 1 collagen by heat denaturation, a process during which collagen loses its native triple helical structure. The resorption time for collagen is 4-8 weeks and for gelatin, 14 days. For our in vivo study, we investigated the rate of infections associated with the 2 grafts among prospective patients undergoing open-chest cardiac surgery at the University Hospital Zurich (Zurich, Switzerland).

The Study
For the in vitro experiments, we dissected the grafts into 5 × 5 mm square pieces and inoculated them with bacterial strains representing pathogens implicated in thoracic PVGI in our patient cohort. These were derived from either Vascular Graft Cohort Study (VASGRA) patients or laboratory strains (Appendix Table 1, Figure 1 Prosthetic vascular graft infections of the thoracic aorta are rare but can be fatal. Our comparison of collagenand gelatin-coated grafts showed that collagen-coated grafts were associated with increased biofilm formation and bacterial adherence in vitro and with higher rates of perioperative vascular graft infections in vivo. soy broth (Becton Dickinson, https://www.bd.com) at 37°C. Graft patches were incubated with bacteria in tryptic soy broth-glucose solution (glucose concentration 8 mmol/L) at 37°C for 72 h, and medium was exchanged every 24 h. The patches were washed with phosphate-buffered saline (PBS), sonicated at 44 khz, and the resulting optical density at a wavelength of 600 was measured in a microplate reader. All bacteria, apart from Pseudomonas aeruginosa strain 2, showed increased biofilm formation on the collagen graft compared with the gelatin graft patches ( Figure, panel A).
Biofilms of selected strains were stained with SYTO 9 of the LIVE/DEAD BacLight Bacterial Viability Kit (ThermoFisher Scientific, https://www. thermofisher.com) according to the manufacturer's instructions. The graft patches were placed in 8-well microslides (ibidi, https://ibidi.com) and visualized by confocal laser scanning microscopy with a Leica TCS SP8 inverted microscope (https://www.leicamicrosystems.com) under a 63×/1.4 oil immersion objective. We selected 2 representative spots per graft patch, providing a stack of horizontally acquired images (512 × 512 pixels representing an area of 244.8 µm × 244.8 µm) with a z-step size of 0.12 µm. We processed the obtained stacks by using Imaris 9.2.0 software (Bitplane; Oxford Instruments, https://imaris. oxinst.com/support/imaris-release-notes/9-2-0). Biofilm height and volume were determined as previously described (11). This approach illustrated the increased biofilm formation on collagen graft patches (Appendix Figure 2). Quantitative analysis from the obtained confocal laser scanning microscopy images corroborated the initial findings because biofilm grown on collagen graft patches displayed increased total biofilm mass volume as well as maximal biofilm height ( Figure, panel B).
One possible explanation for the increased susceptibility to biofilm formation could be the distinct coatings of the grafts. Hence, we coated well plates overnight at 4°C with either rat tail collagen 1 (10 µg/mL; ThermoFisher Scientific) or type B gelatin solution (10 µg/mL; Sigma-Aldrich, https://www. sigmaaldrich.com). The plates were incubated with bacteria at 37°C for 30 min. Bacteria were washed, stained with 0.1% crystal violet (Fluka; Sigma Aldrich, https://www.sigmaaldrich.com), solubilized in 95% ethanol, and the resulting optical density at 570 nm was measured. The tested strains adhered substantially better to collagen (Figure, panel C). Our findings are supported by studies demonstrating the potential of gram-positive bacteria to adhere to collagen, whereas only minor affinity was observed for gelatin (12,13).
To assess the effects of these findings in vivo,  Software, https://www.graphpad.com), we used nonparametric tests (Fisher exact or Wilcoxon rank-sum, as appropriate). When normalized to the total number of control patients (n = 384 who had undergone the cardiac surgery but had no PVGI), the calculated percentage of intracavitary thoracic PVGI (n = 28 VASGRA patients who had undergone the cardiac surgery and had PVGI) was higher for patients in the collagen-graft (10.8%) versus the gelatin-graft group (3.52%; p<0.005).

Conclusions
We found more biofilm formation on collagencoated polyester vascular grafts than on gelatincoated grafts, possibly because the tested strains adhered substantially better to collagen than to gelatin. When we analyzed prospective patients with PVGI and contemporary controls, the percentage of PVGI associated with collagen-coated grafts was also higher. The strengths of our study include use of patientderived strains of bacteria and the prospective collection of patients with incident PVGIs and controls. Our study has some limitations. First, it was a single-center study, resulting in low patient numbers, and the role of the graft material as a risk factor for PVGI is difficult to prove because of the rarity of the infection.
Furthermore, data for patients with PVGI have to be interpreted with caution because some patients have additional foreign material in the heart. Second, publicly available information on the exact type of coating as well as the application procedures used is lacking. In addition, the experiments were performed in a static experimental setup instead of a flow chamber.
However, because we were interested in direct bacterial adherence to the graft material, simulating a scenario in which infection would occur as consequence of unintentional inoculation during the perioperative period, we believe that the setup used is adequate. We did not reproduce a gram-positive and gram-negative mixed biofilm formation experiment because a monomicrobial setup enabled us to determine which material and coating was more susceptible to bacterial adherence in a more controllable fashion.
In conclusion, biofilm formation was increased on collagen-coated vascular grafts compared with gelatin-coated grafts in vitro. As opposed to another risk factor analysis from the VASGRA study (3), in our study, the graft material was associated with the PVGI rate. Parameters such as vascularization potential, secure pseudointima growth, and reduced thrombogenicity are perceived as affecting successful integration and functionality of prosthetic vascular grafts (14). Further parameters should be considered in the future design and development of vascular prostheses to reduce the emerging trend of PVGI. Zimmermann, A. S. Zinkernagel.

Acknowledgments
We are grateful to our patients for their participation in the study. We thank study nurses Caroline Mueller and Simone Bürgin, and we thank Christine Laich and Christine Voegtli for administrative assistance.