- Email: [email protected]
Rifampin-Bonded Vascular Grafts and Postoperative Infections
General Review Rifampin-Bonded Vascular Grafts and Postoperative Infections Olivier Go€eau-Brissonniere,1,2 Isabelle Javerliat,1,2 Fabien Koskas,3,4 Marc Coggia,1,2 and Jean-Claude Pechere,5 Boulogne-Billancourt, Versailles and Paris, France; and Geneva, Switzerland
Postoperative wound and graft infections remain a major challenge for vascular surgeons. The bonding of antimicrobial substances on the graft material has been considered for many years, but the demonstration of safety and efficacy of these techniques is far from evident. Among the different proposed options, bonding of rifampin to the grafts has been the most evaluated technique, both experimentally and clinically. The objective of this review was to present and analyze the available data on rifampin-bonding and the possible evolutions of this technique to improve the resistance of vascular prostheses.
INTRODUCTION Infection of an aortic graft is one of the most feared complications in vascular surgery and continues to challenge the vascular surgeon. Many measures have been advocated to decrease the frequency of its occurrence. One of them is to bind antibiotics to vascular grafts. The development of infectionresistant vascular grafts has considerable appeal to prevent an infection of the aortic graft. Many experimental studies have demonstrated the efficacy of antibiotic-bonded grafts to reduce the incidence of 1 Department of Vascular Surgery, Ambroise Pare University Hospital, Assistance Publique-H^ opitaux de Paris, Boulogne-Billancourt, Paris, France.
infection of the graft.1 The advent of gelatin-sealed polyester grafts has raised the possibility of using the sealant as vehicle for the local delivery of the antibiotic.2 Rifampin was a good candidate for bonding because of its antimicrobial spectrum and physicochemical characteristics. Favorable experimental results3 were the rationale for a prospective randomized study which evaluated the efficacy of rifampin-bonded graft to prevent the occurrence of early wound and/or graft infection after prosthetic aorto-ilio-femoral reconstruction. The results of this randomized study, combined with those of two other European randomized controlled trials (RCT),4-6 led us to evaluate a manufactured graft prebonded with antibiotics.7,8
2
Faculte de Medecine Paris-Ile de France-Ouest, Versailles Saint Quentin en Yvelines University, Versailles, France. 3
Department of Vascular Surgery, Pitie-Salpetriere Hospital, Assistance Publique-H^ opitaux de Paris, Paris, France. 4 Faculte de Medecine Pitie-Salpetriere, Paris 7 University, Paris, France. 5
Department of Microbiology, H^ opitaux Universitaires de Geneve, Suisse. Correspondence to: Olivier Go€eau-Brissonniere, MD, PhD, Department of Vascular Surgery, Ambroise Pare University Hospital, 9 avenue Charles de Gaulle, 92104 Boulogne Cedex, France, E-mail: olivier. [email protected] Ann Vasc Surg 2011; 25: 134-142 DOI: 10.1016/j.avsg.2010.09.002 Ó Annals of Vascular Surgery Inc.
134
INITIAL APPROACH TO RIFAMPINBONDING Why Rifampin-bonding? An infection-resistant prosthesis should comply with several prerequisites. The antimicrobial agent bonded to the prosthesis should have a bactericidal effect against the bacteria involved in graft infections. It should be nonallergenic and have a minimal risk of toxicity. The duration of the antibacterial activity of the bonded graft should be as long as possible to allow a satisfactory graft healing without
Vol. 25, No. 1, January 2011
Rifampin-bonded grafts and postoperative infections 135
infection. Finally, the technique used to bind the antimicrobial agent should be easy to accomplish in the usual clinical setting. Rifampin was then a good candidate for bonding, with a strong affinity for gelatin-sealed polyester grafts. Rifampin demonstrates a wide range of antibacterial activity against most aerobic gram-positive cocci, notably with remarkable antistaphylococcal potency, and against many aerobic gram-negative organisms that cause vascular graft infection. Overall, it is a well-tolerated drug, especially after the parenteral administration of a single dose slowly released in the bloodstream. An in vitro study had shown that rifampin activity is present on the graft surface for 4 days.2 Chervu et al.9 demonstrated that rifampin-bonded grafts had bactericidal activity against Staphylococcus aureus (S. aureus) and Staphylococcus epidermidis (S. epidermidis) for 3 weeks in vitro. In these experiments, the release of the antibiotic into the surrounding tissues took place as the collagen sealant was gradually released. Experimental Data With the use of a dog model of bacteremia, we demonstrated the resistance of rifampin-bonded gelatin-sealed grafts to an early postoperative methicillin-resistant S. aureus bacteremia (Table I).3 In these experiments, 19 dogs underwent a thoracoabdominal aortic bypass with a gelatin-sealed knitted polyester prosthesis. Seven dogs (group I) received a rifampin-treated graft (1 g/L), six (group II) received an untreated gelatin-coated graft, and six (group III) received an uncoated polyester graft. After 2 days, bacteremic challenge was produced by rapid injection of 5 105 colony forming units (CFU/cm2) of methicillin-resistant S. aureus. Grafts Table I. Summary of the experimental data
Dogs (n) Infection of grafts (n) Bacterial count on grafts (CFU/cm2) Positive blood cultures (n) Positive organs culturesa (n)
Group I
Group II
Group III
7 0* NA
6 6 2.5 105
6 6 4 104
0
5
5
0
22 of 24
23 of 24
Group I: Dogs receiving rifampin-treated grafts. Group II: Dogs receiving untreated gelatin-coated grafts. Group III: Dogs receiving uncoated polyester grafts. NA, not applicable. a Organ cultures included liver, spleen, kidney, and lung specimens for each dog. *p < 0.05.
were harvested 5 days after this challenge and cut into 10 fragments, each submitted for bacterial counts. Results were expressed as CFU/cm2 of graft material. In group I, no graft was infected, whereas all grafts in groups II and III were infected ( p < 0.05). Median bacterial counts from the infected fragments were similar in groups II (2.5 105 CFU/cm2, range: 0-107) and III (4 104 CFU/cm2, range: 70-107). Blood cultures that were performed when the dogs were killed were negative in all dogs in group I and positive in five of six dogs in groups II and III. Culture tests of liver, spleen, kidneys, and lungs were always negative in group I and positive in 22 of 24 specimens in group II and 23 of 24 specimens in group III. These results demonstrated the experimental efficacy of rifampin-bonded gelatin-sealed grafts to prevent early bacteremic graft infection and secondary foci of infection. At the time of final analysis, frozen parts of graft fragments were also submitted to a microbiological assay using Sarcina lutea ATCC 9341 as test organism incorporated into Antibiotic Medium I (Difco, Chemie Brunshwig, Basel, Switzerland). After 7 days of implantation, the 40 prosthetic fragments assayed exhibited significant antibiotic activities ranging from 0.1 to 0.9 mg/g of graft material. Clinical Data: Three RCTs These experimental results were the rationale for a prospective randomized study evaluating the efficacy of rifampin-bonding of knitted polyester grafts to prevent early (3 months) infections of wounds and grafts after scheduled prosthetic aorto-iliofemoral replacement (Table II). This study, the Rifampin Bonded Graft European Trial (RBGET), was a multicenter, prospective, and randomized study performed in 90 centers in three European countries (France, Belgium, and Germany) between September 1991 and December 1993. In this study, 2,610 enrolled patients were scheduled to undergo aorto-ilio-femoral reconstruction for abdominal aortic aneurysm (AAA), or aorto-iliac occlusive disease (AIOD), or both (AAA and AIOD). Patients were randomized to receive a rifampin-bonded gelatin-sealed knitted polyester graft (R+ ¼ 1,318) or an untreated gelatin-sealed knitted polyester graft (R ¼ 1,292). The two groups were similar in terms of infectious risk factors, operative indication, nature of systemic antibiotic prophylaxis, or presence of at least a groin incision. Predisposing risk factors of infection were the hospitalization for >48 hours, redo surgery, gangrene, diabetes, and denutrition. Exclusion criteria included patients who required emergent aortic surgery, those with grafts infection, and those
136 Go€eau-Brissonni ere et al.
Annals of Vascular Surgery
Table II. Summary of the data from the three RCTs
Period of study Enrolled patients (n) Arterial disease Repartition of patients into the groups Duration of hospitalization (days) Perioperative mortality (%) Overall infection rates (%)
European trial5,6
European trial4
RBGET
1998 257 AIODa R+ 123 10
March 1991-July 1994 600 AAA/AIOD R+ R 296 304
September 1991-December 1993 2,610 AAA/AIOD R+ R 1,318 1,292 16.5 ± 10.5 16.1 ± 10.6
9 15
R 134 10 5 21
2.6 3.19
Postoperative wound complications in patients with groin incisions (%) Time elapsed between intervention and infection (days) Graft infection rate at 1 year (%)
4.59
1.7
Graft infection rate at 2 years (%)
1.7
Overall mortality rate (%) at 1 year Overall mortality rate (%) at 2 years
4
8.5 ± 4.61
4.5
2.9 5.03 p < 0.005 6.89 p < 0.005 12.6 ± 11.06 p ¼ 0.53
2 p ¼ ns 2.3 p ¼ ns
7.5
R+: Group of patients who received rifampin-bonded grafts. R: Group of patients who received unbonded grafts. AIOD, aorto-iliac occlusive disease; AAA, abdominal aortic aneurysm. a Only extra-anatomical revascularizations for AIOD, no aortic revascularizations.
known to be allergic to rifampin. In this trial, chronic use of steroids was not noted. For randomization, each center received a list specifying the allocation of 30 consecutive patients between the R+ and Re grafts. The centers which involved more than 30 patients received an additional list of randomization. All patients provided written informed consent. Two gelatin-sealed polyester grafts were used for this study (GelsoftÒ and GelsealÒ, VascutekTerumo, Inchinnan, Scotland). In the R+ group, antibiotic-bonding was obtained by perioperatively soaking the grafts for 15 minutes in a 37 C-solution containing 1 mg/mL of rifampin. In the R group, untreated knitted polyester grafts were used. All patients received three perioperative doses of an intravenous antibiotic according to the local protocol. Surgical technique was standardized within each center. Postoperative follow-up was performed daily until the patient was discharged from the hospital, and at 1 and 3 months after the operation. At 1 and 3 months, follow-up included a clinical examination and ultrasound study in all the patients. The outcome was prospectively recorded on specially designed data collecting sheets. The main criterion of evaluation
was the occurrence of early (<3 months) wounds and/or graft infection, according to Szilagyi classification.10 Grade I infections were superficial wound infections, involving only the dermis. Grade II infections were deep wound infections, which extended into the subcutaneous tissue without invading the graft. Grade III infections involved the graft. The other criteria of judgment were the patency of revascularization and the perioperative and the early mortality rates. The two groups of patients were similar in terms of age, sex-ratio, predisposing risk factors of infection, perioperative antibiotic regimen, and operative indications. Mean duration of hospitalization was 16.5 ± 10.5 days for R+ patients and 16.1 ± 10.6 days (CI: 0.63) for R patients, with no significant differences. The perioperative mortality rate was 2.6% in R+ patients and 2.9% in R patients. The overall infection rates in this European study were 3.19% and 5.03% ( p < 0.05) in R+ and R patients, respectively. Grade I infections were found in 2.20% and 3.48% ( p ¼ ns) in R+ and R patients, respectively; grade II in 0.61% and 0.93% ( p ¼ ns) in R+ and R patients, respectively; and grade III in 0.38% and 0.62% ( p ¼ ns) in R+ and R patients,
Vol. 25, No. 1, January 2011
respectively. When combining grade I and grade II, R+ patients developed significantly less wound infections as compared with R patients (2.81% vs. 4.41%, respectively, p < 0.05). Moreover, R+ patients with groin incision developed significantly less postoperative wound infections than R patients (4.59% vs. 6.89%, respectively, p < 0.05). In patients without a groin incision, there were less postoperative wound infection among R+ patients as compared with R patients (0.71% vs. 1.56%, p ¼ ns). The mean time elapsed between intervention and wound infection was 8.5 ± 4.61 days and 12.6 ± 11.06 days in R+ and R patients, respectively ( p ¼ 0.53, ns). Early occlusion of the graft occurred in 0.8% in R+ patients versus 1.5% in R patients ( p ¼ ns). Mortality between the first and the third postoperative months was 0.8% and 0.7% ( p ¼ ns) in R+ and R patients, respectively. A second European RCT was conducted by D’Addato et al.4 in Italy between March 1991 and July 1994, to assess whether rifampin-bonded grafts were effective in preventing graft infections in patients undergoing aorto-femoral bypass. This study enrolled 600 patients suffering from AIOD and/or AAA, who were randomly distributed into two groups. In all, 296 patients received a rifampin-bonded graft (R+ patients) and 304 patients received an untreated graft. In both the groups, gelatin-sealed polyester grafts (Gelseal) were used. Rifampin-bonding was obtained by immersion of the grafts for 15 minutes in a solution containing 1 mg/mL of rifampin. Clinical investigations were performed at 1, 6, 12, and 24 months. Vascular graft patency and presence or absence of skin infection were noted. When infection was suspected, complementary examinations were performed to allow the diagnosis of grade III wound infection, according to Szilagyi classification.10 Between the two groups, there were more AAAs (45% vs. 34%, p ¼ 0.01) and emergent surgeries (7% vs. 4%, p ¼ 0.05) in R patients compared with R+ patients, which were obviously an Achille’s heel of this trial from a methodological point of view. At 1-year follow-up, the total infection rate was 1.8%. Graft infection developed in 1.7% and 2.0% of R+ and R patients, respectively ( p ¼ ns). At the 2-year followup, total graft infection rate was 2% and graft infection rate was 2.3% in R patients and 1.7% in R+ patients, respectively ( p ¼ ns). Patients who developed infection had a significantly higher incidence of lymphatic complications, fever, and early revision surgery ( p ¼ 0.001, p ¼ 0.005, and p ¼ 0.001, respectively). Infection always developed in the groin. Onset of infection was always within the first postoperative month, except in two cases. The overall patient mortality rate was 4% and 7.5%, at 1 and 2-year
Rifampin-bonded grafts and postoperative infections 137
follow-up, respectively. Although incidence of thrombosis in the series was 5%, differences between the two groups were not reported. The authors concluded that there was no significant benefit from the use of rifampin-bonded grafts to prevent early wound infections. In the current published data, another European RCT was published in 1998.5 This study was performed on behalf of the Joint Vascular Research Group in 14 centers in the United Kingdom. The aim of this study was to determine whether the routine use of rifampin-bonded grafts would reduce the incidence of early and late graft infections. Consecutive patients who required an extra-anatomic bypass (cross ilio-femoral, femoro-femoral, or axillofemoral) were enrolled in this study. Preoperative risk factors of infection were steroids, obesity, redo surgery, gangrene, diabetes, and emergent intervention. In all, 257 patients were randomized into the following two groups: 123 patients received rifampin-bonded grafts (R+ patients) and 134 patients received untreated grafts (R patients). Two gelatincoated polyester grafts were used for this study (Gelsoft and Gelseal). For R+ patients, the grafts were soaked in a 1 mg/mL solution of rifampin for 15 minutes before insertion. Outcome was recorded prospectively. Infectious complications were graded according to Szilagyi.10 Patients were reviewed at the first postoperative month and then every 6 months for 2 years. The median duration of hospitalization was 10 days in both the groups. The perioperative mortality rate was 9% and 5% in R+ and R patients, respectively. The total rate of infectious complications was similar in both the groups (15% and 21% in R+ and R patients, respectively). There was only one graft infection in R patients. Earnshaw et al.6 reported the 2-year results of this study in 2000. Overall infection rate was 4.5% at 2-year follow-up (six R+ patients and four R patients developed infection). All infections except one occurred by the sixth postoperative month. All infections involved the groin. The authors concluded that there were no significant advantages for rifampinbonding. In the RBGET, the incidence of wound infection was significantly reduced in patients receiving a rifampin-bonded graft in association with perioperative antimicrobial prophylaxis, especially when groin incisions were performed. The trials conducted in Italy4 and the United Kingdom5,6 did not show significant differences to prevent wound infections between patients receiving antibiotic-bonded or untreated grafts. However, in these two trials there was a trend toward a benefit for rifampin-bonded grafts to prevent wound infections. The RBGET and
138 Go€eau-Brissonni ere et al.
the Italian trial4 both demonstrated a significantly higher incidence of groin complications in the R groups. In the UK trial,5,6 most infections occurred in wounds that never healed properly, with probable extension of a local sepsis to the graft6and all the patients with complications related to infections had groin incisions.5,6 This underlines one of the differences between the three trials. The UK trial5,6 only enrolled those patients who underwent extra-anatomical revascularizations (cross iliofemoral, femorofemoral, or axillofemoral graft), with groin incisions in all cases. This difference in terms of indication probably explains the difference we observe in the early overall infection rate between the RBGET (3.19% and 5.03% in R+ and R patients, respectively, p < 0.005) and the UK trial (15% and 21% in R+ and R patients, respectively).5,6 This may also be the reason for a higher mortality rate that was observed in patients with bonded grafts in the UK trial5,6 as compared with the RBGET. This higher mortality rate in the UK trial5,6 was probably because of the higher surgical risk that patients operated for an extra-anatomic revascularization face. The three RCTs were too small to obtain significant differences regarding the rates of grade III infections. In particular, the single case of infection observed in the UK trial suggests that it would be difficult to sufficiently power subsequent studies to adequately address a real and small benefit. Another Achilles’ heel of the RBGET was to combine aortic revascularizations with and without groin incisions. However, in the RBGET, a significant difference on overall infectious complications and both grade I and grade II wound infections was observed. Moreover, there was a positive trend toward significant differences in all three trials. Larger registries or RCT should be used to significantly prove the efficacy of rifampinbonded grafts in prevention of early wound and/or graft infection. However, the RBGET only studied these cases in the early postoperative period, without a long-term follow-up. Earnshaw et al.6 published their 2-year results in 2000. However, almost one-third of the patients had died before the 2-year follow-up because of which the study group became smaller in size. From these results, the technique of soaking vascular grafts in a rifampin solution has been widely used by vascular surgeons. The following two objections could be raised to the systematic use of rifampin-bonded grafts: (a) the theoretical risk of perioperative contamination during handling of the manipulations to soak the graft in the rifampin solution, and (b) the theoretical risk of emergence of resistant strains of staphylococci. Regarding the first
Annals of Vascular Surgery
objection, the availability of manufactured rifampin prebonded grafts in the market would decrease the perioperative risk of contamination. The combination of rifampin with another antimicrobial agent, as done in systemic prophylaxis, could limit the problem of bacterial resistance. Moreover, acquired resistance has not been observed after rifampin-bonding to vascular prostheses.11 In fact, because wound infection is a predisposing factor to graft infection,12,13 the results of the RBGET probably justify the use of rifampin-bonded grafts in routine practice, especially in patients with high risks of infection as suggested by Strachan.14 Manufactured Prebonded Antibiotic Vascular Grafts The local use of a single antibiotic on vascular prostheses was criticized because emergence of resistance to rifampin that could occur in these patients.15 Although acquired resistance has not been observed after rifampin-bonding of vascular grafts,11,16,17 the adjunction of a second antibiotic could be worthwhile to avoid resistance to rifampin. From the experience obtained with rifampin-soaked gelatin grafts and from the fact that gram-negative bacteria play an increasing role in graft infections, a new graft has recently been developed, in which two antibiotics, rifampin and tobramycin, were loaded during the manufacturing process. The manufacturing process for this new antibiotic-loaded graft is substantially the same as that for the currently available gelatin-sealed grafts. The antibiotics are incorporated into the gelatin sealant during manufacture. The gelatin impregnation process is described in US patent 4,747,848. Briefly, a partially succinylated mixture of United States Pharmacopeia ossein gelatin is dissolved in water at a concentration of 10%. After filtration, this solution is used to vacuum impregnate the polyester substrate. The solution forms a gel, which is cross-linked with 20% aqueous formaldehyde. The excess fixative is removed by multiple rinsing in purified water, followed by plasticization with glycerol. At this stage, an extra step is added for incorporation of the antibiotics. The grafts are immersed in a solution of 13 g/L of rifampin and 1.7 g/L of tobramycin in a mixture of propanol, water, and glycerol. After drying, the grafts are sterilized in the normal, low-temperature, 100% ethylene oxide cycle at a temperature of 37 C. This regime gives a loading of 1.22 mg of rifampin and 0.282 mg tobramycin per centimeter of the 6 mm grafts that are used in the animal model. This level was chosen to ensure that the largest clinically used graft would contain
Vol. 25, No. 1, January 2011
Rifampin-bonded grafts and postoperative infections 139
less than the normal daily dose of each drug. Both the gelatin and the antibiotics are unaffected by this process. The gelatin retains its rapid resorption time of 2 weeks. In vitro evaluations have shown that the grafts retain a high bactericidal level of activity for 3 days, with most of the drug removed by 7 days. Gelatin-sealed polyester grafts were used for this study (Gelsoft).7 Healing and Toxicity of the Prebonded Graft We first evaluated the healing, the toxicity, and the antibiotic delivery of the new vascular graft preloaded with rifampin and tobramycin (Figs. 1 A and B).7 Sixteen dogs underwent infrarenal aortic bypass and were divided into three groups. In test group 1 (n ¼ 8), dogs received grafts loaded with a standard concentration of antibiotics. In test group 2 (n ¼ 4), dogs received grafts loaded with twice the standard concentration of antibiotics. A control group (n ¼ 4) received a commercial gelatin-sealed graft. Grafts were harvested after different periods of time and submitted for histologic evaluation and determination of the antibiotic dose. Liver and kidney toxicities were evaluated from dosages that were performed on serum samples taken at different periods between graft implantation and harvesting. The healing of the antibiotic-loaded grafts was similar to that of commercial grafts, without any signs of toxicity. Graft samples were studied with analytical methods, using a high-performance liquid chromatographic system with a Hypersil 5 ODS column and ultraviolet detection at wavelength of 254 nm. The mobile phase used was a 70:30 0.2 M phosphate buffer (pH 6.0): acetonitrile. CARE Bristol ascertained the limit of detection for this method as 0.1 mg/kg graft. Six test grafts were evaluated. Rifampin was detectable in small quantities on the grafts at up to and including day 21. Significant amounts of rifampin were found on days 7 and 14. At day 21, a very small amount of rifampin was detectable. A significant amount of tobramycin was detectable after 7 days of implantation. In one test graft, a small amount of tobramycin was found at day 14. These results are in accordance with previous rifampin assays performed in gelatin-sealed grafts after 3 days of implantation in the dog.18 By day 21, only traces of antibiotics remain which suggest that the progressive disappearance of rifampin is probably the results of resorption of the gelatin sealant. The persistence of significant amounts of rifampin on the grafts during the perioperative healing period should allow protection during the immediate perioperative stage, which represents the maximum risk for vascular infection.
Fig. 1. Healing of grafts prebonded with rifampin and tobramycin. A Graft harvested at week 3, showing the absence of any fibrous organization, mild inflammation, neutrophil polynuclears, and mural thrombus. The structure of the prosthesis is normal. There is no gelatin layer (100). B Graft harvested at week 12, showing a regular fibrous organization with an adherent peripheral fibrous capsule. A noncircumferential neointima is present, with mural thrombus at sites without intima. The structure of the prosthesis is normal (200).
The presence of tobramycin at day 7 should also contribute to protection at the time of higher risk of graft infection. Experimental Resistance of Prebonded Grafts to Infection These results suggested the necessity of an evaluation of the resistance to infection of these prebonded grafts in an animal model.8 In these experiments, the infrarenal aorta of 12 dogs was replaced using knitted polyester vascular 6-mm diameter grafts. In six dogs, the aorta was replaced with a conventional gelatin-sealed graft (Gelsoft PlusÒ) (control group). In the other six dogs, we used gelatin-sealed grafts
140 Go€eau-Brissonni ere et al.
bonded with two antibiotics, rifampin and tobramycin, during manufacture. On each operative session, two dogs were operated consecutively, one receiving a control graft, and one receiving a test graft. However, because of the yellow color of the test grafts, operators could not be blinded. All dogs were scheduled to be killed 7 days after the first procedure. Bacterial Challenge. The bacterial strain used in this study was S. aureus A980142 obtained from the French National Reference Center. This strain was chosen because we used it in previous experiments19 and it is susceptible to rifampin but resistant to methicillin (heterogenous type) and tobramycin. Bacteria were stored at 80 C using cryobeads. Challenge inoculums were prepared as follows: strain A980142 was cultured on sheep blood agar for 18 hours at 37 C. Colonies were picked and emulsified in normal sterile saline to obtain a turbidity of 0.5 MacFarland (Densimat apparatus, Biom erieux, Marcy l’Etoile, France). The suspension (z5 107 CFU/mL) was diluted appropriately in normal sterile saline to obtain the required bacterial concentration. Local graft contamination was established by topical delivery of a 50 mL volume of challenge inoculum with a sterile pipette. Bacterial concentration of the challenge inoculum was checked by the viable plate count method. The bacterial solution was instilled directly over the graft. The graft was not sponge-dried after contamination. The graft was covered by direct closure of the posterior parietal peritoneum. The dogs did not receive postoperative antibiotic prophylaxis or blood transfusion. Explantation of Grafts. Before the dogs were killed, a venous blood sample was collected for blood culture tests and white blood cells count. A midline laparotomy was used to perform harvesting so as to avoid contamination from the first procedure. Any macroscopic signs of prosthetic and general infection were noted. Specimens for bacteriological studies were collected from various organs (liver, spleen, lung, and left kidney) using sterile techniques. The graft was then harvested taking care to maintain sterile conditions. The remaining blood was gently expressed from the graft. Grafts and pieces of organs for bacteriological studies were stored in separate sterile containers. Bacteriological Studies. Graft and tissue specimens were weighed and then aseptically treated in a bead-beater (Mixer Mill MM301 Retsch GmbH, Haan, Germany) with 1 mm glass beads and 30 mL normal saline, for 5 minutes, at a frequency of 30 oscillations/sec. The number of viable bacteria in this sample was determined by plating serial 5-fold
Annals of Vascular Surgery
dilutions on Mueller-Hinton agar plates and counting the CFUs after a 48-hour incubation at 37 C. Results were expressed as the number of CFU/cm2 of surface of the graft (minimal detection limit: 30 CFU per sample) or per gram of tissue specimen (minimal detection limit: 3.102 CFU per sample). A more sensitive qualitative method was also used to check the sterility of samples (minimal detection limit: 30 CFU per sample). A 13-mL volume of Schaedler broth was inoculated with 1 mL of each suspension. Aliquots were subcultured on sheep blood agar on day 2, and between days 2 and 7, if turned turbid. The samples were considered positive if S. aureus A980142 was recovered from any subcultures.
RESULTS The mean inoculum size was 1.83 104 CFU/mL (range: 1 104-2 104 CFU/mL) in the test group and 1.82 104 (range: 6 103-4 104 CFU/mL) in the control group, without significant differences between the two groups of dogs. All dogs were alive when they were killed. Five of the six control grafts had S. aureus A980142 growing on them at the time of graft removal, whereas none of the six antibioticbonded grafts were infected. The mean bacterial count of infected grafts was 2 103 CFU/cm2 (range: 3 102-1010). This experimental study confirmed the efficacy of gelatin-sealed grafts manufactured with prebonded rifampin and tobramycin to resist infection from S. aureus in dogs. A limitation of this study may be the absence of several other control groups, such as unbonded grafts with or without gelatin sealing, unsealed grafts soaked in rifampin or tobramycin, gelatin-sealed grafts soaked in rifampin or tobramycin, and gelatin-sealed grafts soaked in rifampin and tobramycin. However, we felt that these groups were not required for several reasons. First, the primary focus of the study was resistance to infection of the manufactured grafts prebonded with antibiotics. Second, as discussed earlier in the text, we previously demonstrated (a) the inability of unsealed polyester grafts to bind significant amounts of rifampin,18 and (b) the constant contamination of unbonded gelatin-sealed grafts and the resistance of rifampin-bonded gelatin-sealed grafts when they were exposed to a bacteremic challenge or a local contamination in dogs.3,20,21 On the basis of these previous results, and for ethical reasons, we decided to focus on the comparison of the prebonded graft with the unbonded gelatin-sealed grafts. Several considerations are important when designing grafts preloaded with antibiotics during
Vol. 25, No. 1, January 2011
the manufacturing process. Prebonding is not only a matter of convenience over point-of-use loading, it also has to ensure a consistent standardized dose. Activity must be high at implant and perioperatively, but the antibiotics should not persist for longer than a few days, to minimize the risk of bacterial resistance. It is essential for the manufacturing process, including sterilization, not to change the chemical nature of either the drugs or the substrate used as a carrier. Untoward reactions could affect the resorption of the sealant and impair healing. The bonding of nonantibiotic antimicrobial substances is a different approach to the prevention and/or treatment of graft infection, and grafts containing silver and collagen are available in Europe.22 However, with the use of an animal model, it has been demonstrated that silverand collagen-coated grafts were less resistant to bacteremic infection with S. aureus than rifampin gelatin-sealed grafts.19 Schmacht et al.23 also demonstrated that rifampin-soaked grafts exhibited a higher initial resistance to local contamination with S. aureus than silver- and collagen-coated grafts, but with a secondary growth curve. Bonding has been extensively investigated with different antimicrobial agents and different techniques. However, as yet, since the introduction of rifampin-soaking, no manufactured graft has been made commercially available prebonded with antibiotics. This is mostly because of the technical difficulties regarding the homogeneity of the loading, the stability of the antimicrobial agents during the sterilization process and over the time. One of the greatest advantages of grafts prebonded with antibiotics would be their availability in any circumstances, especially in emergency, or in patients at high risk of infection. In this new antibiotic prebonded graft, tobramycin was added to rifampin because it widens the antimicrobial spectrum, with activity against gram-positive cocci and gram-negative strains. This association should also contribute in preventing the emergence of resistance to rifampin.16,24 Tobramycin-glue has already been studied by Shenk et al. in 1989.25 In this study, none of the dogs with expanded polytetrafluoroethylene (PTFE) grafts, treated with antibiotic glue, developed graft infection. The amount of rifampin used to coat the grafts in our study was 13 mg/mL. This was lower than the amount used in previous studies.19 However, we have previously showed the efficacy in preventing vascular graft infection of a gelatin-sealed graft soaked in a solution containing rifampin at 1 mg/mL.3 Koshiko et al.26 implanted gelatin grafts that were soaked in a 1 mg/mL rifampin solution in dogs. They failed to demonstrate the efficacy of these grafts
Rifampin-bonded grafts and postoperative infections 141
in cases of virulent organisms like S. aureus. However, in this study, the microbial concentration (1 108 organisms/mL) was higher than our previous study, and very far from the number of microorganisms involved in a perioperative contamination. These results indicate that this new gelatin-sealed graft manufactured prebonded with antibiotics resists infection caused by S. aureus graft contamination in a dog model. However, further clinical studies are definitely required to confirm these experimental findings, and antibiotic-bonded grafts should not wave from the use of systemic antibiotics and/or omentum wrapping in a contaminated field.
CONCLUSION In conclusion, experimental data and results of the RBGET are in favor of the efficacy and the safety of rifampin-bonded grafts to contribute in the prevention of infections of the early wound and/or vascular graft, especially in patients with groin incisions, or with risk factors of infection (diabetic patients, patients with gangrene or ulcers, redo surgery, emergent surgery, chronic use of steroids). However, rifampin-bonded grafts should not wave from using perfect surgical techniques to prevent lymphatic complications and the need for early revisions, or attempting endovascular treatment before open operative treatment to reduce the risk of infection again. In the near future, the availability of manufactured antibiotic-prebonded grafts could be useful to easily develop the use of antibioticbonded grafts in these situations, in routine practice. REFERENCES 1. Gelabert HA, Colburn MD. Development and results of antibiotic-impregnated vascular grafts. In: Moore WS, Gelabert HA, eds. Antibiotic Impregnated Vascular Grafts. Austin, TX: R.G. Landes, 1992. pp 88-100. 2. Ashton TR, Cunningham JD, Paton D, et al. Antibiotic loading of vascular grafts. Paper presented at: Proceedings of the 16th Annual Meeting of the Society for Biomaterials, May 20-23, Charleston, SC, 1990. p235. 3. Go€eau-Brissonniere O, Leport C, Bacourt F, et al. Prevention of vascular graft infection by rifampin bonding to a gelatinsealed Dacron graft. Ann Vasc Surg 1991;5:408-412. 4. D’Addato M, Curti T, Freyrie A, et al. Prophylaxis of graft infection with rifampicin-bonded Gelseal graft: 2-year follow-up of a prospective clinical trial. Cardiovasc Surg 1996;4:200-204. 5. Braithwaite BD, Davies B, Heather BP, et al. Early results of a randomized trial of rifampicin-bonded Dacron grafts for extra-anatomic vascular reconstruction. Br J Surg 1998;85: 1378-1381. 6. Earnshaw JJ, Whitman B, Heather BP. Two-year results of a randomized controlled trial of rifampicin-bonded extraanatomic Dacron grafts. Br J Surg 2000;87:758-759. 7. Javerliat I, Go€eau-Brissonniere O, Bruneval P, et al. Experimental study of a new vascular graft prebonded with antibiotic:
142 Go€eau-Brissonni ere et al.
8.
9.
10.
11.
12.
13. 14.
15.
16.
17.
healing, toxicity and antibiotic retention. Ann Vasc Surg 2007;21:603-610. Javerliat I, Go€ eau-Brissonniere O, Sivadon-Tardy V, et al. Prevention of Staphylococcus aureus graft infection by a new gelatin-sealed vascular graft prebonded with antibiotics. J Vasc Surg 2007;46:1026-1031. Chervu A, Moore WS, Chvapil M, et al. Efficacy and duration of antistaphylococcal activity comparing three antibiotics bonded to Dacron vascular grafts with a collagen release system. J Vasc Surg 1991;13:897-901. Szylagyi DE, Smith RF, Elliott JP, et al. Infection in arterial reconstruction with synthetic grafts. Ann Surg 1972;176: 321-332. Sardelic F, Ao PY, Fletcher JP. Rifampicin impregnated Dacron grafts: no development of rifampicin resistance in an animal model. Eur J Vasc Endovasc Surg 1995;9: 314-318. Jones L, Braithwaite BD, Davies B, et al. Mechanism of late prosthetic vascular graft infection. Cardiovasc Surg 1997;5: 486-489. Edwards WH, Martin RS, Jenkins JM, et al. primary graft infections. J Vasc Surg 1987;6:235-239. Strachan CJL. Antibiotic prophylaxis in peripheral vascular and in ortopaedic prosthetic surgery. J Antimicrobial Chemother 1993;31(Suppl. B):65-78. Sande SA. The use of rifampin in the treatment of non tuberculous infections: an overview. Rev Infect Dis 1983;5: 399-411. Kunin CM, Brandt D, Wood H. Bacteriologic studies of rifampin, a new semisynthetic antibiotic. J Infect Dis 1969;119:132-137. Mandell GL, Moorman DR. Treatment of experimental staphylococcal infections: rifampin alone and in combination on
Annals of Vascular Surgery
18.
19.
20.
21.
22.
23.
24.
25.
26.
development of rifampin resistance. Antimicrob Agents Chemother 1980;17:658-662. Malassiney P, Go€eau-Brissonniere O, Coggia M, et al. Rifampicin loading of vascular grafts. J Antimicrob Chemother 1996;37:1121-1129. Go€eau-Brissonniere O, Fabre D, Leflon-Guibout V, et al. Comparison of the resistance to infection of rifampin-bonded gelatin-sealed and silver/collagen-coated polyester prostheses. J Vasc Surg 2002;35:1260-1263. Go€eau-Brissonniere O, Mercier F, Nicolas MH, et al. Treatment of vascular graft infection by in situ replacement with rifampin-bonded gelatine-sealed Dacron graft. J Vasc Surg 1994;19:739-744. Coggia M, Go€eau-Brissonniere O, Leflon V, et al. Experimental treatment of vascular graft infection due to Staphylococcus epidermidis by in situ replacement with a rifampin-bonded polyester graft. Ann Vasc Surg 2001;15:421-429. Ricco JB. Intergard Silver Study Group. Intergard silver bifurcated graft: features and results of a multicenter clinical study. J Vasc Surg 2006;44:339-346. Schmacht D, Armstrong P, Johnson B, et al. Graft infectivity of rifampin and silver-bonded polyester grafts to MRSA contamination. Vasc Endovasc Surg 2005;39:411-420. Mandell GL, Keit Vest T. Killing of intraleukocytic Staphylococcus aureus by rifampin: in-vitro and in-vivo studies. J Infect Dis 1972;125:486-490. Shenk JS, Ney AL, Tsukayama DT, et al. Tobramycin-adhesive in preventing and treating PTFE vascular graft infections. J Surg Res 1989;47:487-492. Koshiko S, Sasajima T, Muraki S, et al. Limitations in the use of rifampicin-gelatin grafts against virulent organisms. J Vasc Surg 2002;35:779-785.
Postoperative wound and graft infections remain a major challenge for vascular surgeons. The bonding of antimicrobial substances on the graft material has been considered for many years, but the demonstration of safety and efficacy of these techniques is far from evident. Among the different proposed options, bonding of rifampin to the grafts has been the most evaluated technique, both experimentally and clinically. The objective of this review was to present and analyze the available data on rifampin-bonding and the possible evolutions of this technique to improve the resistance of vascular prostheses.
INTRODUCTION Infection of an aortic graft is one of the most feared complications in vascular surgery and continues to challenge the vascular surgeon. Many measures have been advocated to decrease the frequency of its occurrence. One of them is to bind antibiotics to vascular grafts. The development of infectionresistant vascular grafts has considerable appeal to prevent an infection of the aortic graft. Many experimental studies have demonstrated the efficacy of antibiotic-bonded grafts to reduce the incidence of 1 Department of Vascular Surgery, Ambroise Pare University Hospital, Assistance Publique-H^ opitaux de Paris, Boulogne-Billancourt, Paris, France.
infection of the graft.1 The advent of gelatin-sealed polyester grafts has raised the possibility of using the sealant as vehicle for the local delivery of the antibiotic.2 Rifampin was a good candidate for bonding because of its antimicrobial spectrum and physicochemical characteristics. Favorable experimental results3 were the rationale for a prospective randomized study which evaluated the efficacy of rifampin-bonded graft to prevent the occurrence of early wound and/or graft infection after prosthetic aorto-ilio-femoral reconstruction. The results of this randomized study, combined with those of two other European randomized controlled trials (RCT),4-6 led us to evaluate a manufactured graft prebonded with antibiotics.7,8
2
Faculte de Medecine Paris-Ile de France-Ouest, Versailles Saint Quentin en Yvelines University, Versailles, France. 3
Department of Vascular Surgery, Pitie-Salpetriere Hospital, Assistance Publique-H^ opitaux de Paris, Paris, France. 4 Faculte de Medecine Pitie-Salpetriere, Paris 7 University, Paris, France. 5
Department of Microbiology, H^ opitaux Universitaires de Geneve, Suisse. Correspondence to: Olivier Go€eau-Brissonniere, MD, PhD, Department of Vascular Surgery, Ambroise Pare University Hospital, 9 avenue Charles de Gaulle, 92104 Boulogne Cedex, France, E-mail: olivier. [email protected] Ann Vasc Surg 2011; 25: 134-142 DOI: 10.1016/j.avsg.2010.09.002 Ó Annals of Vascular Surgery Inc.
134
INITIAL APPROACH TO RIFAMPINBONDING Why Rifampin-bonding? An infection-resistant prosthesis should comply with several prerequisites. The antimicrobial agent bonded to the prosthesis should have a bactericidal effect against the bacteria involved in graft infections. It should be nonallergenic and have a minimal risk of toxicity. The duration of the antibacterial activity of the bonded graft should be as long as possible to allow a satisfactory graft healing without
Vol. 25, No. 1, January 2011
Rifampin-bonded grafts and postoperative infections 135
infection. Finally, the technique used to bind the antimicrobial agent should be easy to accomplish in the usual clinical setting. Rifampin was then a good candidate for bonding, with a strong affinity for gelatin-sealed polyester grafts. Rifampin demonstrates a wide range of antibacterial activity against most aerobic gram-positive cocci, notably with remarkable antistaphylococcal potency, and against many aerobic gram-negative organisms that cause vascular graft infection. Overall, it is a well-tolerated drug, especially after the parenteral administration of a single dose slowly released in the bloodstream. An in vitro study had shown that rifampin activity is present on the graft surface for 4 days.2 Chervu et al.9 demonstrated that rifampin-bonded grafts had bactericidal activity against Staphylococcus aureus (S. aureus) and Staphylococcus epidermidis (S. epidermidis) for 3 weeks in vitro. In these experiments, the release of the antibiotic into the surrounding tissues took place as the collagen sealant was gradually released. Experimental Data With the use of a dog model of bacteremia, we demonstrated the resistance of rifampin-bonded gelatin-sealed grafts to an early postoperative methicillin-resistant S. aureus bacteremia (Table I).3 In these experiments, 19 dogs underwent a thoracoabdominal aortic bypass with a gelatin-sealed knitted polyester prosthesis. Seven dogs (group I) received a rifampin-treated graft (1 g/L), six (group II) received an untreated gelatin-coated graft, and six (group III) received an uncoated polyester graft. After 2 days, bacteremic challenge was produced by rapid injection of 5 105 colony forming units (CFU/cm2) of methicillin-resistant S. aureus. Grafts Table I. Summary of the experimental data
Dogs (n) Infection of grafts (n) Bacterial count on grafts (CFU/cm2) Positive blood cultures (n) Positive organs culturesa (n)
Group I
Group II
Group III
7 0* NA
6 6 2.5 105
6 6 4 104
0
5
5
0
22 of 24
23 of 24
Group I: Dogs receiving rifampin-treated grafts. Group II: Dogs receiving untreated gelatin-coated grafts. Group III: Dogs receiving uncoated polyester grafts. NA, not applicable. a Organ cultures included liver, spleen, kidney, and lung specimens for each dog. *p < 0.05.
were harvested 5 days after this challenge and cut into 10 fragments, each submitted for bacterial counts. Results were expressed as CFU/cm2 of graft material. In group I, no graft was infected, whereas all grafts in groups II and III were infected ( p < 0.05). Median bacterial counts from the infected fragments were similar in groups II (2.5 105 CFU/cm2, range: 0-107) and III (4 104 CFU/cm2, range: 70-107). Blood cultures that were performed when the dogs were killed were negative in all dogs in group I and positive in five of six dogs in groups II and III. Culture tests of liver, spleen, kidneys, and lungs were always negative in group I and positive in 22 of 24 specimens in group II and 23 of 24 specimens in group III. These results demonstrated the experimental efficacy of rifampin-bonded gelatin-sealed grafts to prevent early bacteremic graft infection and secondary foci of infection. At the time of final analysis, frozen parts of graft fragments were also submitted to a microbiological assay using Sarcina lutea ATCC 9341 as test organism incorporated into Antibiotic Medium I (Difco, Chemie Brunshwig, Basel, Switzerland). After 7 days of implantation, the 40 prosthetic fragments assayed exhibited significant antibiotic activities ranging from 0.1 to 0.9 mg/g of graft material. Clinical Data: Three RCTs These experimental results were the rationale for a prospective randomized study evaluating the efficacy of rifampin-bonding of knitted polyester grafts to prevent early (3 months) infections of wounds and grafts after scheduled prosthetic aorto-iliofemoral replacement (Table II). This study, the Rifampin Bonded Graft European Trial (RBGET), was a multicenter, prospective, and randomized study performed in 90 centers in three European countries (France, Belgium, and Germany) between September 1991 and December 1993. In this study, 2,610 enrolled patients were scheduled to undergo aorto-ilio-femoral reconstruction for abdominal aortic aneurysm (AAA), or aorto-iliac occlusive disease (AIOD), or both (AAA and AIOD). Patients were randomized to receive a rifampin-bonded gelatin-sealed knitted polyester graft (R+ ¼ 1,318) or an untreated gelatin-sealed knitted polyester graft (R ¼ 1,292). The two groups were similar in terms of infectious risk factors, operative indication, nature of systemic antibiotic prophylaxis, or presence of at least a groin incision. Predisposing risk factors of infection were the hospitalization for >48 hours, redo surgery, gangrene, diabetes, and denutrition. Exclusion criteria included patients who required emergent aortic surgery, those with grafts infection, and those
136 Go€eau-Brissonni ere et al.
Annals of Vascular Surgery
Table II. Summary of the data from the three RCTs
Period of study Enrolled patients (n) Arterial disease Repartition of patients into the groups Duration of hospitalization (days) Perioperative mortality (%) Overall infection rates (%)
European trial5,6
European trial4
RBGET
1998 257 AIODa R+ 123 10
March 1991-July 1994 600 AAA/AIOD R+ R 296 304
September 1991-December 1993 2,610 AAA/AIOD R+ R 1,318 1,292 16.5 ± 10.5 16.1 ± 10.6
9 15
R 134 10 5 21
2.6 3.19
Postoperative wound complications in patients with groin incisions (%) Time elapsed between intervention and infection (days) Graft infection rate at 1 year (%)
4.59
1.7
Graft infection rate at 2 years (%)
1.7
Overall mortality rate (%) at 1 year Overall mortality rate (%) at 2 years
4
8.5 ± 4.61
4.5
2.9 5.03 p < 0.005 6.89 p < 0.005 12.6 ± 11.06 p ¼ 0.53
2 p ¼ ns 2.3 p ¼ ns
7.5
R+: Group of patients who received rifampin-bonded grafts. R: Group of patients who received unbonded grafts. AIOD, aorto-iliac occlusive disease; AAA, abdominal aortic aneurysm. a Only extra-anatomical revascularizations for AIOD, no aortic revascularizations.
known to be allergic to rifampin. In this trial, chronic use of steroids was not noted. For randomization, each center received a list specifying the allocation of 30 consecutive patients between the R+ and Re grafts. The centers which involved more than 30 patients received an additional list of randomization. All patients provided written informed consent. Two gelatin-sealed polyester grafts were used for this study (GelsoftÒ and GelsealÒ, VascutekTerumo, Inchinnan, Scotland). In the R+ group, antibiotic-bonding was obtained by perioperatively soaking the grafts for 15 minutes in a 37 C-solution containing 1 mg/mL of rifampin. In the R group, untreated knitted polyester grafts were used. All patients received three perioperative doses of an intravenous antibiotic according to the local protocol. Surgical technique was standardized within each center. Postoperative follow-up was performed daily until the patient was discharged from the hospital, and at 1 and 3 months after the operation. At 1 and 3 months, follow-up included a clinical examination and ultrasound study in all the patients. The outcome was prospectively recorded on specially designed data collecting sheets. The main criterion of evaluation
was the occurrence of early (<3 months) wounds and/or graft infection, according to Szilagyi classification.10 Grade I infections were superficial wound infections, involving only the dermis. Grade II infections were deep wound infections, which extended into the subcutaneous tissue without invading the graft. Grade III infections involved the graft. The other criteria of judgment were the patency of revascularization and the perioperative and the early mortality rates. The two groups of patients were similar in terms of age, sex-ratio, predisposing risk factors of infection, perioperative antibiotic regimen, and operative indications. Mean duration of hospitalization was 16.5 ± 10.5 days for R+ patients and 16.1 ± 10.6 days (CI: 0.63) for R patients, with no significant differences. The perioperative mortality rate was 2.6% in R+ patients and 2.9% in R patients. The overall infection rates in this European study were 3.19% and 5.03% ( p < 0.05) in R+ and R patients, respectively. Grade I infections were found in 2.20% and 3.48% ( p ¼ ns) in R+ and R patients, respectively; grade II in 0.61% and 0.93% ( p ¼ ns) in R+ and R patients, respectively; and grade III in 0.38% and 0.62% ( p ¼ ns) in R+ and R patients,
Vol. 25, No. 1, January 2011
respectively. When combining grade I and grade II, R+ patients developed significantly less wound infections as compared with R patients (2.81% vs. 4.41%, respectively, p < 0.05). Moreover, R+ patients with groin incision developed significantly less postoperative wound infections than R patients (4.59% vs. 6.89%, respectively, p < 0.05). In patients without a groin incision, there were less postoperative wound infection among R+ patients as compared with R patients (0.71% vs. 1.56%, p ¼ ns). The mean time elapsed between intervention and wound infection was 8.5 ± 4.61 days and 12.6 ± 11.06 days in R+ and R patients, respectively ( p ¼ 0.53, ns). Early occlusion of the graft occurred in 0.8% in R+ patients versus 1.5% in R patients ( p ¼ ns). Mortality between the first and the third postoperative months was 0.8% and 0.7% ( p ¼ ns) in R+ and R patients, respectively. A second European RCT was conducted by D’Addato et al.4 in Italy between March 1991 and July 1994, to assess whether rifampin-bonded grafts were effective in preventing graft infections in patients undergoing aorto-femoral bypass. This study enrolled 600 patients suffering from AIOD and/or AAA, who were randomly distributed into two groups. In all, 296 patients received a rifampin-bonded graft (R+ patients) and 304 patients received an untreated graft. In both the groups, gelatin-sealed polyester grafts (Gelseal) were used. Rifampin-bonding was obtained by immersion of the grafts for 15 minutes in a solution containing 1 mg/mL of rifampin. Clinical investigations were performed at 1, 6, 12, and 24 months. Vascular graft patency and presence or absence of skin infection were noted. When infection was suspected, complementary examinations were performed to allow the diagnosis of grade III wound infection, according to Szilagyi classification.10 Between the two groups, there were more AAAs (45% vs. 34%, p ¼ 0.01) and emergent surgeries (7% vs. 4%, p ¼ 0.05) in R patients compared with R+ patients, which were obviously an Achille’s heel of this trial from a methodological point of view. At 1-year follow-up, the total infection rate was 1.8%. Graft infection developed in 1.7% and 2.0% of R+ and R patients, respectively ( p ¼ ns). At the 2-year followup, total graft infection rate was 2% and graft infection rate was 2.3% in R patients and 1.7% in R+ patients, respectively ( p ¼ ns). Patients who developed infection had a significantly higher incidence of lymphatic complications, fever, and early revision surgery ( p ¼ 0.001, p ¼ 0.005, and p ¼ 0.001, respectively). Infection always developed in the groin. Onset of infection was always within the first postoperative month, except in two cases. The overall patient mortality rate was 4% and 7.5%, at 1 and 2-year
Rifampin-bonded grafts and postoperative infections 137
follow-up, respectively. Although incidence of thrombosis in the series was 5%, differences between the two groups were not reported. The authors concluded that there was no significant benefit from the use of rifampin-bonded grafts to prevent early wound infections. In the current published data, another European RCT was published in 1998.5 This study was performed on behalf of the Joint Vascular Research Group in 14 centers in the United Kingdom. The aim of this study was to determine whether the routine use of rifampin-bonded grafts would reduce the incidence of early and late graft infections. Consecutive patients who required an extra-anatomic bypass (cross ilio-femoral, femoro-femoral, or axillofemoral) were enrolled in this study. Preoperative risk factors of infection were steroids, obesity, redo surgery, gangrene, diabetes, and emergent intervention. In all, 257 patients were randomized into the following two groups: 123 patients received rifampin-bonded grafts (R+ patients) and 134 patients received untreated grafts (R patients). Two gelatincoated polyester grafts were used for this study (Gelsoft and Gelseal). For R+ patients, the grafts were soaked in a 1 mg/mL solution of rifampin for 15 minutes before insertion. Outcome was recorded prospectively. Infectious complications were graded according to Szilagyi.10 Patients were reviewed at the first postoperative month and then every 6 months for 2 years. The median duration of hospitalization was 10 days in both the groups. The perioperative mortality rate was 9% and 5% in R+ and R patients, respectively. The total rate of infectious complications was similar in both the groups (15% and 21% in R+ and R patients, respectively). There was only one graft infection in R patients. Earnshaw et al.6 reported the 2-year results of this study in 2000. Overall infection rate was 4.5% at 2-year follow-up (six R+ patients and four R patients developed infection). All infections except one occurred by the sixth postoperative month. All infections involved the groin. The authors concluded that there were no significant advantages for rifampinbonding. In the RBGET, the incidence of wound infection was significantly reduced in patients receiving a rifampin-bonded graft in association with perioperative antimicrobial prophylaxis, especially when groin incisions were performed. The trials conducted in Italy4 and the United Kingdom5,6 did not show significant differences to prevent wound infections between patients receiving antibiotic-bonded or untreated grafts. However, in these two trials there was a trend toward a benefit for rifampin-bonded grafts to prevent wound infections. The RBGET and
138 Go€eau-Brissonni ere et al.
the Italian trial4 both demonstrated a significantly higher incidence of groin complications in the R groups. In the UK trial,5,6 most infections occurred in wounds that never healed properly, with probable extension of a local sepsis to the graft6and all the patients with complications related to infections had groin incisions.5,6 This underlines one of the differences between the three trials. The UK trial5,6 only enrolled those patients who underwent extra-anatomical revascularizations (cross iliofemoral, femorofemoral, or axillofemoral graft), with groin incisions in all cases. This difference in terms of indication probably explains the difference we observe in the early overall infection rate between the RBGET (3.19% and 5.03% in R+ and R patients, respectively, p < 0.005) and the UK trial (15% and 21% in R+ and R patients, respectively).5,6 This may also be the reason for a higher mortality rate that was observed in patients with bonded grafts in the UK trial5,6 as compared with the RBGET. This higher mortality rate in the UK trial5,6 was probably because of the higher surgical risk that patients operated for an extra-anatomic revascularization face. The three RCTs were too small to obtain significant differences regarding the rates of grade III infections. In particular, the single case of infection observed in the UK trial suggests that it would be difficult to sufficiently power subsequent studies to adequately address a real and small benefit. Another Achilles’ heel of the RBGET was to combine aortic revascularizations with and without groin incisions. However, in the RBGET, a significant difference on overall infectious complications and both grade I and grade II wound infections was observed. Moreover, there was a positive trend toward significant differences in all three trials. Larger registries or RCT should be used to significantly prove the efficacy of rifampinbonded grafts in prevention of early wound and/or graft infection. However, the RBGET only studied these cases in the early postoperative period, without a long-term follow-up. Earnshaw et al.6 published their 2-year results in 2000. However, almost one-third of the patients had died before the 2-year follow-up because of which the study group became smaller in size. From these results, the technique of soaking vascular grafts in a rifampin solution has been widely used by vascular surgeons. The following two objections could be raised to the systematic use of rifampin-bonded grafts: (a) the theoretical risk of perioperative contamination during handling of the manipulations to soak the graft in the rifampin solution, and (b) the theoretical risk of emergence of resistant strains of staphylococci. Regarding the first
Annals of Vascular Surgery
objection, the availability of manufactured rifampin prebonded grafts in the market would decrease the perioperative risk of contamination. The combination of rifampin with another antimicrobial agent, as done in systemic prophylaxis, could limit the problem of bacterial resistance. Moreover, acquired resistance has not been observed after rifampin-bonding to vascular prostheses.11 In fact, because wound infection is a predisposing factor to graft infection,12,13 the results of the RBGET probably justify the use of rifampin-bonded grafts in routine practice, especially in patients with high risks of infection as suggested by Strachan.14 Manufactured Prebonded Antibiotic Vascular Grafts The local use of a single antibiotic on vascular prostheses was criticized because emergence of resistance to rifampin that could occur in these patients.15 Although acquired resistance has not been observed after rifampin-bonding of vascular grafts,11,16,17 the adjunction of a second antibiotic could be worthwhile to avoid resistance to rifampin. From the experience obtained with rifampin-soaked gelatin grafts and from the fact that gram-negative bacteria play an increasing role in graft infections, a new graft has recently been developed, in which two antibiotics, rifampin and tobramycin, were loaded during the manufacturing process. The manufacturing process for this new antibiotic-loaded graft is substantially the same as that for the currently available gelatin-sealed grafts. The antibiotics are incorporated into the gelatin sealant during manufacture. The gelatin impregnation process is described in US patent 4,747,848. Briefly, a partially succinylated mixture of United States Pharmacopeia ossein gelatin is dissolved in water at a concentration of 10%. After filtration, this solution is used to vacuum impregnate the polyester substrate. The solution forms a gel, which is cross-linked with 20% aqueous formaldehyde. The excess fixative is removed by multiple rinsing in purified water, followed by plasticization with glycerol. At this stage, an extra step is added for incorporation of the antibiotics. The grafts are immersed in a solution of 13 g/L of rifampin and 1.7 g/L of tobramycin in a mixture of propanol, water, and glycerol. After drying, the grafts are sterilized in the normal, low-temperature, 100% ethylene oxide cycle at a temperature of 37 C. This regime gives a loading of 1.22 mg of rifampin and 0.282 mg tobramycin per centimeter of the 6 mm grafts that are used in the animal model. This level was chosen to ensure that the largest clinically used graft would contain
Vol. 25, No. 1, January 2011
Rifampin-bonded grafts and postoperative infections 139
less than the normal daily dose of each drug. Both the gelatin and the antibiotics are unaffected by this process. The gelatin retains its rapid resorption time of 2 weeks. In vitro evaluations have shown that the grafts retain a high bactericidal level of activity for 3 days, with most of the drug removed by 7 days. Gelatin-sealed polyester grafts were used for this study (Gelsoft).7 Healing and Toxicity of the Prebonded Graft We first evaluated the healing, the toxicity, and the antibiotic delivery of the new vascular graft preloaded with rifampin and tobramycin (Figs. 1 A and B).7 Sixteen dogs underwent infrarenal aortic bypass and were divided into three groups. In test group 1 (n ¼ 8), dogs received grafts loaded with a standard concentration of antibiotics. In test group 2 (n ¼ 4), dogs received grafts loaded with twice the standard concentration of antibiotics. A control group (n ¼ 4) received a commercial gelatin-sealed graft. Grafts were harvested after different periods of time and submitted for histologic evaluation and determination of the antibiotic dose. Liver and kidney toxicities were evaluated from dosages that were performed on serum samples taken at different periods between graft implantation and harvesting. The healing of the antibiotic-loaded grafts was similar to that of commercial grafts, without any signs of toxicity. Graft samples were studied with analytical methods, using a high-performance liquid chromatographic system with a Hypersil 5 ODS column and ultraviolet detection at wavelength of 254 nm. The mobile phase used was a 70:30 0.2 M phosphate buffer (pH 6.0): acetonitrile. CARE Bristol ascertained the limit of detection for this method as 0.1 mg/kg graft. Six test grafts were evaluated. Rifampin was detectable in small quantities on the grafts at up to and including day 21. Significant amounts of rifampin were found on days 7 and 14. At day 21, a very small amount of rifampin was detectable. A significant amount of tobramycin was detectable after 7 days of implantation. In one test graft, a small amount of tobramycin was found at day 14. These results are in accordance with previous rifampin assays performed in gelatin-sealed grafts after 3 days of implantation in the dog.18 By day 21, only traces of antibiotics remain which suggest that the progressive disappearance of rifampin is probably the results of resorption of the gelatin sealant. The persistence of significant amounts of rifampin on the grafts during the perioperative healing period should allow protection during the immediate perioperative stage, which represents the maximum risk for vascular infection.
Fig. 1. Healing of grafts prebonded with rifampin and tobramycin. A Graft harvested at week 3, showing the absence of any fibrous organization, mild inflammation, neutrophil polynuclears, and mural thrombus. The structure of the prosthesis is normal. There is no gelatin layer (100). B Graft harvested at week 12, showing a regular fibrous organization with an adherent peripheral fibrous capsule. A noncircumferential neointima is present, with mural thrombus at sites without intima. The structure of the prosthesis is normal (200).
The presence of tobramycin at day 7 should also contribute to protection at the time of higher risk of graft infection. Experimental Resistance of Prebonded Grafts to Infection These results suggested the necessity of an evaluation of the resistance to infection of these prebonded grafts in an animal model.8 In these experiments, the infrarenal aorta of 12 dogs was replaced using knitted polyester vascular 6-mm diameter grafts. In six dogs, the aorta was replaced with a conventional gelatin-sealed graft (Gelsoft PlusÒ) (control group). In the other six dogs, we used gelatin-sealed grafts
140 Go€eau-Brissonni ere et al.
bonded with two antibiotics, rifampin and tobramycin, during manufacture. On each operative session, two dogs were operated consecutively, one receiving a control graft, and one receiving a test graft. However, because of the yellow color of the test grafts, operators could not be blinded. All dogs were scheduled to be killed 7 days after the first procedure. Bacterial Challenge. The bacterial strain used in this study was S. aureus A980142 obtained from the French National Reference Center. This strain was chosen because we used it in previous experiments19 and it is susceptible to rifampin but resistant to methicillin (heterogenous type) and tobramycin. Bacteria were stored at 80 C using cryobeads. Challenge inoculums were prepared as follows: strain A980142 was cultured on sheep blood agar for 18 hours at 37 C. Colonies were picked and emulsified in normal sterile saline to obtain a turbidity of 0.5 MacFarland (Densimat apparatus, Biom erieux, Marcy l’Etoile, France). The suspension (z5 107 CFU/mL) was diluted appropriately in normal sterile saline to obtain the required bacterial concentration. Local graft contamination was established by topical delivery of a 50 mL volume of challenge inoculum with a sterile pipette. Bacterial concentration of the challenge inoculum was checked by the viable plate count method. The bacterial solution was instilled directly over the graft. The graft was not sponge-dried after contamination. The graft was covered by direct closure of the posterior parietal peritoneum. The dogs did not receive postoperative antibiotic prophylaxis or blood transfusion. Explantation of Grafts. Before the dogs were killed, a venous blood sample was collected for blood culture tests and white blood cells count. A midline laparotomy was used to perform harvesting so as to avoid contamination from the first procedure. Any macroscopic signs of prosthetic and general infection were noted. Specimens for bacteriological studies were collected from various organs (liver, spleen, lung, and left kidney) using sterile techniques. The graft was then harvested taking care to maintain sterile conditions. The remaining blood was gently expressed from the graft. Grafts and pieces of organs for bacteriological studies were stored in separate sterile containers. Bacteriological Studies. Graft and tissue specimens were weighed and then aseptically treated in a bead-beater (Mixer Mill MM301 Retsch GmbH, Haan, Germany) with 1 mm glass beads and 30 mL normal saline, for 5 minutes, at a frequency of 30 oscillations/sec. The number of viable bacteria in this sample was determined by plating serial 5-fold
Annals of Vascular Surgery
dilutions on Mueller-Hinton agar plates and counting the CFUs after a 48-hour incubation at 37 C. Results were expressed as the number of CFU/cm2 of surface of the graft (minimal detection limit: 30 CFU per sample) or per gram of tissue specimen (minimal detection limit: 3.102 CFU per sample). A more sensitive qualitative method was also used to check the sterility of samples (minimal detection limit: 30 CFU per sample). A 13-mL volume of Schaedler broth was inoculated with 1 mL of each suspension. Aliquots were subcultured on sheep blood agar on day 2, and between days 2 and 7, if turned turbid. The samples were considered positive if S. aureus A980142 was recovered from any subcultures.
RESULTS The mean inoculum size was 1.83 104 CFU/mL (range: 1 104-2 104 CFU/mL) in the test group and 1.82 104 (range: 6 103-4 104 CFU/mL) in the control group, without significant differences between the two groups of dogs. All dogs were alive when they were killed. Five of the six control grafts had S. aureus A980142 growing on them at the time of graft removal, whereas none of the six antibioticbonded grafts were infected. The mean bacterial count of infected grafts was 2 103 CFU/cm2 (range: 3 102-1010). This experimental study confirmed the efficacy of gelatin-sealed grafts manufactured with prebonded rifampin and tobramycin to resist infection from S. aureus in dogs. A limitation of this study may be the absence of several other control groups, such as unbonded grafts with or without gelatin sealing, unsealed grafts soaked in rifampin or tobramycin, gelatin-sealed grafts soaked in rifampin or tobramycin, and gelatin-sealed grafts soaked in rifampin and tobramycin. However, we felt that these groups were not required for several reasons. First, the primary focus of the study was resistance to infection of the manufactured grafts prebonded with antibiotics. Second, as discussed earlier in the text, we previously demonstrated (a) the inability of unsealed polyester grafts to bind significant amounts of rifampin,18 and (b) the constant contamination of unbonded gelatin-sealed grafts and the resistance of rifampin-bonded gelatin-sealed grafts when they were exposed to a bacteremic challenge or a local contamination in dogs.3,20,21 On the basis of these previous results, and for ethical reasons, we decided to focus on the comparison of the prebonded graft with the unbonded gelatin-sealed grafts. Several considerations are important when designing grafts preloaded with antibiotics during
Vol. 25, No. 1, January 2011
the manufacturing process. Prebonding is not only a matter of convenience over point-of-use loading, it also has to ensure a consistent standardized dose. Activity must be high at implant and perioperatively, but the antibiotics should not persist for longer than a few days, to minimize the risk of bacterial resistance. It is essential for the manufacturing process, including sterilization, not to change the chemical nature of either the drugs or the substrate used as a carrier. Untoward reactions could affect the resorption of the sealant and impair healing. The bonding of nonantibiotic antimicrobial substances is a different approach to the prevention and/or treatment of graft infection, and grafts containing silver and collagen are available in Europe.22 However, with the use of an animal model, it has been demonstrated that silverand collagen-coated grafts were less resistant to bacteremic infection with S. aureus than rifampin gelatin-sealed grafts.19 Schmacht et al.23 also demonstrated that rifampin-soaked grafts exhibited a higher initial resistance to local contamination with S. aureus than silver- and collagen-coated grafts, but with a secondary growth curve. Bonding has been extensively investigated with different antimicrobial agents and different techniques. However, as yet, since the introduction of rifampin-soaking, no manufactured graft has been made commercially available prebonded with antibiotics. This is mostly because of the technical difficulties regarding the homogeneity of the loading, the stability of the antimicrobial agents during the sterilization process and over the time. One of the greatest advantages of grafts prebonded with antibiotics would be their availability in any circumstances, especially in emergency, or in patients at high risk of infection. In this new antibiotic prebonded graft, tobramycin was added to rifampin because it widens the antimicrobial spectrum, with activity against gram-positive cocci and gram-negative strains. This association should also contribute in preventing the emergence of resistance to rifampin.16,24 Tobramycin-glue has already been studied by Shenk et al. in 1989.25 In this study, none of the dogs with expanded polytetrafluoroethylene (PTFE) grafts, treated with antibiotic glue, developed graft infection. The amount of rifampin used to coat the grafts in our study was 13 mg/mL. This was lower than the amount used in previous studies.19 However, we have previously showed the efficacy in preventing vascular graft infection of a gelatin-sealed graft soaked in a solution containing rifampin at 1 mg/mL.3 Koshiko et al.26 implanted gelatin grafts that were soaked in a 1 mg/mL rifampin solution in dogs. They failed to demonstrate the efficacy of these grafts
Rifampin-bonded grafts and postoperative infections 141
in cases of virulent organisms like S. aureus. However, in this study, the microbial concentration (1 108 organisms/mL) was higher than our previous study, and very far from the number of microorganisms involved in a perioperative contamination. These results indicate that this new gelatin-sealed graft manufactured prebonded with antibiotics resists infection caused by S. aureus graft contamination in a dog model. However, further clinical studies are definitely required to confirm these experimental findings, and antibiotic-bonded grafts should not wave from the use of systemic antibiotics and/or omentum wrapping in a contaminated field.
CONCLUSION In conclusion, experimental data and results of the RBGET are in favor of the efficacy and the safety of rifampin-bonded grafts to contribute in the prevention of infections of the early wound and/or vascular graft, especially in patients with groin incisions, or with risk factors of infection (diabetic patients, patients with gangrene or ulcers, redo surgery, emergent surgery, chronic use of steroids). However, rifampin-bonded grafts should not wave from using perfect surgical techniques to prevent lymphatic complications and the need for early revisions, or attempting endovascular treatment before open operative treatment to reduce the risk of infection again. In the near future, the availability of manufactured antibiotic-prebonded grafts could be useful to easily develop the use of antibioticbonded grafts in these situations, in routine practice. REFERENCES 1. Gelabert HA, Colburn MD. Development and results of antibiotic-impregnated vascular grafts. In: Moore WS, Gelabert HA, eds. Antibiotic Impregnated Vascular Grafts. Austin, TX: R.G. Landes, 1992. pp 88-100. 2. Ashton TR, Cunningham JD, Paton D, et al. Antibiotic loading of vascular grafts. Paper presented at: Proceedings of the 16th Annual Meeting of the Society for Biomaterials, May 20-23, Charleston, SC, 1990. p235. 3. Go€eau-Brissonniere O, Leport C, Bacourt F, et al. Prevention of vascular graft infection by rifampin bonding to a gelatinsealed Dacron graft. Ann Vasc Surg 1991;5:408-412. 4. D’Addato M, Curti T, Freyrie A, et al. Prophylaxis of graft infection with rifampicin-bonded Gelseal graft: 2-year follow-up of a prospective clinical trial. Cardiovasc Surg 1996;4:200-204. 5. Braithwaite BD, Davies B, Heather BP, et al. Early results of a randomized trial of rifampicin-bonded Dacron grafts for extra-anatomic vascular reconstruction. Br J Surg 1998;85: 1378-1381. 6. Earnshaw JJ, Whitman B, Heather BP. Two-year results of a randomized controlled trial of rifampicin-bonded extraanatomic Dacron grafts. Br J Surg 2000;87:758-759. 7. Javerliat I, Go€eau-Brissonniere O, Bruneval P, et al. Experimental study of a new vascular graft prebonded with antibiotic:
142 Go€eau-Brissonni ere et al.
8.
9.
10.
11.
12.
13. 14.
15.
16.
17.
healing, toxicity and antibiotic retention. Ann Vasc Surg 2007;21:603-610. Javerliat I, Go€ eau-Brissonniere O, Sivadon-Tardy V, et al. Prevention of Staphylococcus aureus graft infection by a new gelatin-sealed vascular graft prebonded with antibiotics. J Vasc Surg 2007;46:1026-1031. Chervu A, Moore WS, Chvapil M, et al. Efficacy and duration of antistaphylococcal activity comparing three antibiotics bonded to Dacron vascular grafts with a collagen release system. J Vasc Surg 1991;13:897-901. Szylagyi DE, Smith RF, Elliott JP, et al. Infection in arterial reconstruction with synthetic grafts. Ann Surg 1972;176: 321-332. Sardelic F, Ao PY, Fletcher JP. Rifampicin impregnated Dacron grafts: no development of rifampicin resistance in an animal model. Eur J Vasc Endovasc Surg 1995;9: 314-318. Jones L, Braithwaite BD, Davies B, et al. Mechanism of late prosthetic vascular graft infection. Cardiovasc Surg 1997;5: 486-489. Edwards WH, Martin RS, Jenkins JM, et al. primary graft infections. J Vasc Surg 1987;6:235-239. Strachan CJL. Antibiotic prophylaxis in peripheral vascular and in ortopaedic prosthetic surgery. J Antimicrobial Chemother 1993;31(Suppl. B):65-78. Sande SA. The use of rifampin in the treatment of non tuberculous infections: an overview. Rev Infect Dis 1983;5: 399-411. Kunin CM, Brandt D, Wood H. Bacteriologic studies of rifampin, a new semisynthetic antibiotic. J Infect Dis 1969;119:132-137. Mandell GL, Moorman DR. Treatment of experimental staphylococcal infections: rifampin alone and in combination on
Annals of Vascular Surgery
18.
19.
20.
21.
22.
23.
24.
25.
26.
development of rifampin resistance. Antimicrob Agents Chemother 1980;17:658-662. Malassiney P, Go€eau-Brissonniere O, Coggia M, et al. Rifampicin loading of vascular grafts. J Antimicrob Chemother 1996;37:1121-1129. Go€eau-Brissonniere O, Fabre D, Leflon-Guibout V, et al. Comparison of the resistance to infection of rifampin-bonded gelatin-sealed and silver/collagen-coated polyester prostheses. J Vasc Surg 2002;35:1260-1263. Go€eau-Brissonniere O, Mercier F, Nicolas MH, et al. Treatment of vascular graft infection by in situ replacement with rifampin-bonded gelatine-sealed Dacron graft. J Vasc Surg 1994;19:739-744. Coggia M, Go€eau-Brissonniere O, Leflon V, et al. Experimental treatment of vascular graft infection due to Staphylococcus epidermidis by in situ replacement with a rifampin-bonded polyester graft. Ann Vasc Surg 2001;15:421-429. Ricco JB. Intergard Silver Study Group. Intergard silver bifurcated graft: features and results of a multicenter clinical study. J Vasc Surg 2006;44:339-346. Schmacht D, Armstrong P, Johnson B, et al. Graft infectivity of rifampin and silver-bonded polyester grafts to MRSA contamination. Vasc Endovasc Surg 2005;39:411-420. Mandell GL, Keit Vest T. Killing of intraleukocytic Staphylococcus aureus by rifampin: in-vitro and in-vivo studies. J Infect Dis 1972;125:486-490. Shenk JS, Ney AL, Tsukayama DT, et al. Tobramycin-adhesive in preventing and treating PTFE vascular graft infections. J Surg Res 1989;47:487-492. Koshiko S, Sasajima T, Muraki S, et al. Limitations in the use of rifampicin-gelatin grafts against virulent organisms. J Vasc Surg 2002;35:779-785.