Identification of Staphylococcus epidermidis vascular graft infections: A comparison of culture techniques☆☆☆★
Article Outline
Abstract
Culture of prosthetic material is routinely used to exclude or implicate infection in the pathogenesis of late-appearing graft complications. In a canine model of aortic graft infection caused by a bacterial biofilm, the influence of culture media (blood agar and tryptic soy broth) and mechanical surface biofilm disruption (tissue grinding and ultrasonic oscillation) on microorganism recovery was determined. Dacron prostheses colonized in vitro with Staphylococcus epidermidis were implanted in the infrarenal aortas of 36 dogs. After 3 weeks an infection with anatomic characteristics of late graft infection in humans was present. Explantation (± surface biofilm disruption) of infected grafts showed broth culture was superior (p < 0.001) to agar media in confirming infection. The recovery rate of S. epidermidis was 30% with agar media, was 72% with broth media alone, and was 83% with broth media plus biofilm disruption. In situ replacement of infected grafts plus parenteral antibiotics resulted in early (1 month) healing of 31 grafts without signs of infection. All replacement grafts were sterile when cultured in broth media alone, but the addition of biofilm disruption isolated the study strain from eight (22%) of 36 grafts (p < 0.01). Biofilm disruption by tissue grinding or sonication increased bacteria recovery equally. When biofilm bacterial concentration was less than 100 colony-forming units/cm2 of graft, only culture in broth media reliably recovered microorganisms. In the absence of perigraft inflammation, microbiologic recovery techniques that identify bacterial biofilms are necessary to exclude infection in studies concerning the pathogenesis of late graft complications or the treatment of S. epidermidis prosthetic infections. (J Vasc Surg 1989;9:665–70.)
The nature of infections involving vascular prostheses is changing. Although staphylococcal species remain the most important pathogens, Staphylococcus epidermidis has displaced S. aureus as the prevalent organism infecting biomaterials implanted in the vascular system of man. 1, 2, 3 S. epidermidis tends to be less virulent than S. aureus and produces infections of a chronic, indolent nature. As such, these infections may persist for months before an accurate diagnosis can be made.
Standard clinical signs of vascular graft infection include sepsis with bacteremia, fever, leukocytosis, absence of graft incorporation with perigraft inflammation and exudate, a draining pulsatile mass in the groin, or a vasculoenteric fistula. 4, 5, 6 In most of these infections, especially when they occur within 4 months after graft implantation, the offending microorganism is easily identified by routine Gram's stain and culture of perigraft drainage. When graft healing complications (perigraft inflammation with sinus tract formation and anastomotic aneurysms) appear months to years after implantation, systemic signs of infection are usually absent and Gram's stain of the graft or perigraft exudate or both typically shows many polymorphonuclear leukocytes but no bacteria.2 Routine microbiologic techniques will frequently not isolate a microorganism from perigraft tissue, and culture of the prosthetic material is necessary to implicate or exclude infection.
The pathogenesis of late-appearing graft infection involves the ability of certain bacterial strains to adhere to prosthetic surfaces and grow as a continuous biofilm. The foreign body and bacteria-laden biofilm act together as coinflammatory stimuli to activate the immune system, a process that has tissue-damaging effects. The sequestration of bacteria within a surface biofilm and its adherent nature to the prosthesis, the absence of perigraft tissue invasion, and low virulence of the organisms account for the absence of graft sepsis. These anatomic and microbiologic characteristics are typical of prosthetic infections caused by slime-producing S. epidermidis.2 Recovery of viable microorganisms from the biofilm that forms on implanted prosthetic surfaces is necessary to confirm infection. Work from our laboratory has shown that mechanical disruption of prosthetic surface biofilms can increase the recovery of microorganisms from explanted grafts when compared to routine culture methods.7 Use of culture techniques that include broth media or mechanical tissue disruption or both has recovered S. epidermidis from arteries and lymph nodes of one third of patients undergoing peripheral arterial reconstruction, 3, 8 from 15 (75%) of 21 prostheses revised for anastomotic pseudoaneurysm,9 and from 18 (66%) of 27 aortofemoral prostheses removed for infection.2
We have recently developed a canine model of vascular graft infection caused by bacterial biofilms that mimics the clinical characteristics of late graft infection in humans.10 This animal model has been used to assess the efficacy of in situ graft replacement as a treatment option for these infections. It has also been valuable for comparing the sensitivity of various microbiologic techniques in the recovery of microorganisms from explanted graft material. Use of accurate microbiologic recovery techniques that identify surface bacterial biofilms is important in studies concerning the pathobiologic characteristics of late vascular graft complications or when treatment of S. epidermidis prosthetic infections is analyzed. In this study we report the influence of culture media (blood agar and tryptic soy broth, TSB) and mechanical surface biofilm disruption (tissue grinding and ultrasonic oscillation) on the recovery of S. epidermidis from grafts suspected of being infected. Culture techniques applicable for processing explanted vascular graft material were compared by use of both in vivo (canine aortic graft infection model) and in vitro experimental models. In the animal experiments culture results were correlated with presence or absence of anatomic signs of infection.
Material and methods
Three microbiologic culture techniques were compared with respect to the ability to isolate microorganisms from grafts suspected of being colonized by a biofilm containing S. epidermidis. Graft material cultured measured approximately 1 cm2.
Culture on agar media
Explanted graft material was imprinted (external and internal surfaces) onto tryptic soy agar with 5% sheep's blood. Agar plates were incubated at 35° C.
Culture in broth media
Graft material was placed in glucose-supplemented (0.25% dextrose) TSB. The culture tube containing broth media and graft specimen was incubated at 35° C.
Broth media culture plus surface biofilm disruption (biofilm culture)
Two methods of mechanical surface biofilm disruption were used and included ultrasonic oscillation and tissue grinding. Graft specimens to be sonicated were placed in TSB, and the sterilized metal probe of the sonic dismembranator (Artak Systems Corp., Farmingdale, N.Y.) was inserted into the culture tube. Ultrasonic oscillation of the broth media containing the graft specimen was performed at 20 kHz for 10 minutes. During the sonication process the culture tube was immersed in ice to prevent heating of the broth media. Tissue of the graft surfaces was ground by placing the graft specimen in a sterilized Tenbroeck tissue homogenizer containing TSB for 5 minutes. After the sonication or the mechanical grinding process, graft specimens were transferred to a second culture tube containing TSB. Broth culture tubes containing disrupted biofilm or graft specimen were incubated at 35° C, and growth in either tube was counted as a positive biofilm culture.
Canine model
The details of the canine model of aortic graft infection caused by S. epidermidis have been previously described.10 In brief, sterile surgical technique was used to implant Dacron grafts (6 mm diameter, 5 cm length) colonized in vitro with a slime-producing strain of S. epidermidis (American Type Culture Collection No. 35983) into the infrarenal aorta of 36 female conditioned mongrel dogs (weight, 20 to 25 kg). A commercially available double-velour Dacron graft (Meadox Medicals Inc., Oakland, N.J.) was used. The in vitro graft colonization technique consisted of immersion of graft segments in a phosphate-buffered saline solution with 0.25% dextrose that contained 7.0 log10 colony-forming units (CFU)/ml of S. epidermidis for 24 hours at room temperature. During the incubation period a bacteria-laden biofilm forms on the prosthetic surfaces with a mean bacterial concentration of approximately 5 × 106 CFU/cm2 of graft material. All animals received humane care in compliance with the “Principles of Laboratory Animal Care” and the “Guide for the Care and Use of Laboratory. Animals” (NIH Publication No. 80-23, revised 1978). No antibiotics were administered to animals during the perioperative period.
One month after the grafts were implanted, the infected aortic grafts were exposed with sterile surgical technique through the previous midline incision. Cefazolin sodium (25 mg/kg) was administered intravenously 1 hour before the operation. After proximal and distal control of the graft aorta was obtained, the Dacron graft was inspected for signs of infection that included perigraft inflammation/fluid, perigraft cavity, and anastomotic dehiscence. The graft, including anastomotic sites, was then totally excised, and 1 cm length graft segments were cultured by the three microbiologic bacteria recovery techniques described above. After the graft was excised, the aorta was debrided to grossly normal artery wall, and continuity was established by randomized reimplantation of double-velour Dacron or expanded polytetrafluoroethylene (PTFE, Gore-Tex, W.L. Gore & Associates, Inc., Elkton, Md.) vascular prostheses. All animals underwent in situ graft replacement by means of similar graft-artery anastomotic technique with 6-0 continuous polypropylene suture. If a perigraft cavity surrounded the infected Dacron graft, it was debrided to adjacent retroperitoneal structures (duodenum and inferior vena cava). Animals in groups of 18 received 1 or 14 days of cefazolin sodium (25 mg/kg three times a day) intravenously after in situ graft replacement.
One month after grafts were replaced in situ, the aortic grafts were explanted by use of identical operative technique as described for excision of the primarily infected Dacron grafts. Replacement grafts were inspected in situ for signs of infection or perigraft inflammation and were cultured by means of the three microbiologic bacteria recovery techniques.
In vitro model
This experimental model was designed to measure the minimum bacterial concentration that will be detected by means of microbiologic recovery techniques that use mechanical surface biofilm disruption (sonication and tissue grinding) and broth or agar culture media. Double-velour, knitted Dacron grafts were cut into 1 cm squares and sterilized. The graft specimens were inoculated with the S. epidermidis study strain to form a bacteria-laden surface biofilm. Groups of 20 graft specimens were immersed in 10 ml of one of three inoculating solutions containing different concentrations of S. epidermidis (9.2 × 104 CFU/ml, 1.6 × 102 CFU/ml, and 2.1 × 101 CFU/ml) for 18 hours at room temperature. After incubation was complete, the inoculum was decanted and the graft specimens were washed 10 times with 10 ml of phosphate-buffered saline solution. Organisms remaining on the graft specimens were judged to be adherent to the surface and within the formed biofilm. Each graft specimen was then placed in 25 ml of TSB. Sixty graft specimens (20 exposed to each concentration of S. epidermidis) were sonicated (20 kHz, 5 minutes) to dislodge adherent microcolonies from the graft surfaces. The remaining 60 graft specimens (20 exposed to each inoculating solution of S. epidermidis) were ground for 5 minutes in a Ten Broch tissue grinder that contained TSB. One milliliter of the effluent from sonication or tissue grinding was cultured on tryptic soy agar with 5% sheep's blood at 35° C for 72 hours. One milliliter of the effluent from sonication and tissue grinding was also incubated in 25 ml of TSB at 35° C for 72 hours. All broth and agar cultures with no growth by 72 hours were considered sterile.
Data analysis
All cultures of grafts explanted from animal experiments without growth after 14 days were considered sterile. Bacterial isolates were identified and staphylococcal species isolated with the API STAPH-IDENT (Analylab Products, Plainview, N.Y.) system. S. epidermidis isolates were evaluated for slime production to verify recovery of the study strain. The microbiologic data from the various culture techniques were compared by means of chi-square analysis, where p value less than 0.05 was considered significant.
Results
Recovery of S. epidermidis from canine grafts
All 36 study animals implanted with colonized Dacron grafts had anatomic and histologic evidence of graft infection. The retroperitoneum covering the grafts was inflamed and had poor graft incorporation, and a perigraft cavity containing an exudate surrounded 28 grafts. All infected grafts were patent and had no evidence of anastomotic dehiscence. No wound infections occurred. The S. epidermidis study strain was isolated from 30 (83%) of 36 explanted primary Dacron grafts (Table I). Culture of graft material in broth media was superior (p < 0.001) to culture in agar media in confirming infection. The recovery rate of S. epidermidis was similar in graft culture in broth media alone (72%) compared to broth media plus biofilm disruption (83%). No microorganisms were isolated from three grafts despite clinical signs of infection. S. hominis was recovered from the biofilm of three primary infected grafts whose cultures were negative for S. epidermidis.
Table I. Recovery of S. epidermidis from canine grafts relative to microbiologic culture technique
| Recovery of S. epidermidis (no. positive/no. grafts) | ||
|---|---|---|
| Culture technique | Primary infected Dacron graft | In situ replacement graft |
| Agar media | 1/18* | ND |
| Broth media | 26/36 | 0/36† |
| Broth media plus biofilm disruption | 30/36 | 8/36 |
| * p < 0.001 versus broth media alone and broth media plus biofilm disruption. † p < 0.01 versus media plus biofilm disruption. | ||
One month after in situ graft replacement, graft healing without signs of infection was seen in 18 of 18 PTFE and 13 of 18 Dacron replacement grafts. All replacement grafts were patent and had no signs of anastomotic healing complications. Perigraft inflammation was observed surrounding five (28%) of 18 Dacron replacement grafts, and S. epidermidis was recovered from three of these grafts. The recovery of S. epidermidis from replacement grafts required use of the biofilm culture technique (Table I). When replacement grafts were cultured in broth media alone, all explanted graft material showed no growth. In contrast, biofilm culture recovered S. epidermidis from five (28%) of 18 PTFE grafts and three (17%) of 18 Dacron replacement grafts (p > 0.05). Mechanical biofilm disruption by sonication or tissue grinding resulted in similar recovery rates of S. epidermidis from both primary infected and replacement graft surfaces (Table II).
Table II. Comparison biofilm disruption methods in recovery of S. epidermidis from canine grafts cultured in broth media
| Recovery of S. epidermidis (no. positive/no. grafts) | ||
|---|---|---|
| Method of surface biofilm disruption | Primary infected Dacron graft | In situ replacement graft |
| Tissue grinding | 8/14 | 4/17 |
| Ultrasonic oscillation | 11/14 | 3/17 |
 p value (chi square) | NS | NS |
Culture results relative to anatomic signs of infection
When perigraft inflammation was present, culture of graft material in broth media alone was similar to the biofilm culture technique in the isolation of a pathogen. For the 36 primary infected grafts and five replacement Dacron grafts with clinical signs of infection, S. epidermidis was recovered by broth culture from 26 of 41 grafts and by biofilm culture from 31 of 41 grafts (p > 0.05). It is important to note that biofilm culture did not detect bacterial colonization of 10 grafts despite clinical signs of infection. When grafts appeared to be healed and incorporated within surrounding tissue, graft culture in broth media plus surface biofilm disruption was required to detect graft colonization by S. epidermidis. For the 31 replacement grafts without signs of recurrent infection, biofilm culture isolated S. epidermidis from seven grafts, but all grafts were judged sterile by broth culture. Other bacteria recovered from explanted canine grafts included coagulase-negative staphylococcal species (S. hominis, S. warneri, S. capitus, S. intermedius), corynebacterium, Moraxella species, bacillus, and enterococcus. Most of the second isolates (seven of nine) were recovered only from the biofilm cultures. It is of note that S. hominis was the second most common isolate recovered from explanted canine grafts; it was the only isolate from three primary infected Dacron grafts. This finding is of importance because the API STAPH-IDENT system will misidentify phosphatase-negative S. epidermidis as S. hominis. Only six of 41 grafts with clinical signs of infection were judged to be sterile by all culture methods compared to 20 of 31 grafts judged clinically to be free of infection (p < 0.001).
In vitro comparison of agar and broth media
Recovery of S. epidermidis from grafts cultured after surface biofilm disruption in either agar or broth media was dependent on bacteria concentration within the biofilm (Table III).
Table III. In vitro comparison of culture media in recovery of S. epidermidis from grafts relative to biofilm bacteria concentration
| Recovery of S. epidermidis (no. positive/no. grafts) | ||
|---|---|---|
| Biofilm concentration | Agar media | Broth media |
| 101 | 10/20* | 20/20 |
| 102 | 20/20 | 20/20 |
| 104 | 20/20 | 20/20 |
| *p < 0.001 versus broth media. | ||
Discussion
As innovative methods for processing explanted biomaterial for culture have been developed, the epidemiologic and microbiologic characteristics of prosthetic device infections have been clarified. The most prevalent microorganisms isolated from explanted medical devices are coagulase-negative staphylococci (CNS). These organisms are the primary pathogens that cause infections of a variety of indwelling biomaterials such as intravascular catheters and grafts, orthopedic appliances, and cerebrospinal fluid shunts. 1, 11, 12, 13 The predilection of CNS for foreign body infection is not well understood, but pathogenic strains are characterized by increased quantitative adherence to prosthetic surfaces, production of a viscid extracellular “mucoid” material in vitro, and antimicrobial resistance. 13, 14
The present study confirmed prior clinical observations made in our laboratory that culture of explanted vascular prostheses in broth media allows accurate identification of pathogens when signs of inflammation are present. 2, 9 Routine microbiologic culture methods, which typically involve swabs taken from perigraft tissue or biomaterial surfaces, transported in thioglycollate media to the laboratory, and imprinted unto agar plates, are subject to sampling error because of the low concentration of microorganisms involved in prosthetic device infections caused by S. epidermidis. Although the broth and agar media used in this study have essentially an identical biochemical composition, the nutrient and growth environment afforded by broth media exerts less stress for continued growth of an inoculum containing small numbers of microorganisms. The addition of mechanical surface biofilm disruption to the broth culture technique did not significantly increase the recovery of S. epidermidis from grafts with anatomic signs of perigraft inflammation. The superiority of broth media over agar plating in recovery of pathogenic isolates and in predicting clinical infection has been documented when both culture techniques were used to monitor the microbiologic characteristics of surgical wounds.15 Similarly, positive artery wall cultures (mechanically ground tissue incubated in broth media) adjacent to infected vascular prostheses were reported by Malone et al.16 to be predictive of arterial/aortic disruption, and were used to identify a group of patients who might benefit from long-term antibiotic therapy.
Poor graft incorporation with surrounding tissue, perigraft inflammation, or adjacent arteritis are anatomic signs of a bacterial biofilm infection and should not be attributed to an immune-mediated foreign body reaction. As documented in the canine model of graft infection, microorganisms will be isolated from more than 80% of prostheses with these clinical signs despite perigraft Gram's stains and culture without evidence of bacterial infection. Conversely, the absence of inflammation surrounding an implanted bioprosthesis does not exclude the presence of bacterial colonization. Mechanical disruption of surface biofilms of these grafts is critical in processing the explanted biomaterial for culture in either broth or agar media. Based on data from the in vitro experiments, surface biofilm disruption will recover S. epidermidis from vascular prostheses by means of agar media to biofilm concentration less than 1000 CFU/cm2 of graft, and with broth media to biofilm concentration less than 100 CFU/cm2 of graft. Tissue grinding and ultrasonic oscillation were equivalent as methods for mechanical biofilm disruption based on incidence of bacteria biofilm identification in both in vivo and in vitro experiments. Both methods are applicable to processing of graft specimens in clinical microbiology laboratories. Mechanical tissue grinders are present in all hospital-based microbiology laboratories, and personnel are familiar with the equipment. Although ultrasonic disruption is not used commonly in clinical microbiology, and the instrument costs approximately $1000, the sonication technique is well suited for processing of biomaterial for culture. Solid objects (metal appliances, catheters, vascular prostheses) can be inserted in the instrument and can have adherent biologic tissues disrupted and displaced from their surfaces. The clinical value of surface biofilm disruption resides in the identification of biomaterial colonization when overt signs of infection are absent. Despite normal healing of 31 in situ replacement grafts in our canine model, S. epidermidis was isolated from seven grafts by biofilm culture. Without this culture methodology, in situ replacement plus antibiotic administration would have been judged erroneously to be successful treatment in approximately one fifth of the animals. The late outcome of prosthesis colonization with a bacterial biofilm but no anatomic evidence of infection is unknown. Using our canine model, we have measured normal anastomotic tensile strength in grafts with persistent colonization 3 months after in situ replacement of a graft with an established S. epidermidis infection.
The taxonomy of CNS has been clarified over the past decade. 17, 18 More than two dozen new species have been characterized. Many species that previously would have been classified S. epidermidis were reclassified as S. hominis, S. capitus, S. warneri, and S. haemolyticus. In general, strains of S. epidermidis are more virulent than strains of S. hominis and the other CNS. The kit used in this study for identifying species of CNS will misidentify S. epidermidis as S. hominis.13 The API STAPH-IDENT system discriminates between these two species on the basis of a single biochemical reaction—phosphatase. In our janine model it is probable that the S. hominis strains recovered as the only isolate from canine grafts were in fact the S. epidermidis study strain. This error in CNS identification can be minimized by use of the API STAPH-TRAC (Analylab Products, Plainview, N.Y.) system, which will not misidentify phosphatase-negative S. epidermidis as S. hominis. Most (greater than 80%) CNS species isolated from infected prosthetic devices (cerebrospinal fluid shunts, vascular prostheses) adhere to plastic tissue culture plates and produce mucin or “slime.” The precise role of slime production in the pathogenesis of these infections is unknown, but it is hypothesized that this growth characteristic promotes adherence and colonization to implanted biomaterials, and it may also interfere with cell-mediated immunity19 and bactericidal activity of polymorphonuclear leukocytes.20
The localized nature of S. epidermidis graft infections, the confinement of microorganisms within an adherent surface biofilm, and the low virulence and concentration of the infecting organisms are important characteristics, not present with S. aureus or gram-negative infections, that permit in situ graft replacement to be considered as a treatment option. Implementation of this treatment in our canine model resulted in healing of the replacement grafts, particularly when grafts constructed of PTFE were used. The prolonged (14 days) administration of antibiotics also appeared to decrease formation of bacterial biofilms on the in situ replacement PTFE grafts. The use of microbiologic processing techniques capable of identifying formation and persistence of bacterial biofilms on implanted prostheses is critical in investigations of this nature. By routine use of culture techniques that reliably identify CNS biofilm colonization of biomaterials, exclusion or implication of infection in the pathogenosis of late-appearing graft complication, especially anastomotic aneurysm formation, will be facilitated.
Acknowledgements
We thank W.L. Gore & Associates, Elkton, Md., and Meadox Medicals Inc., Oakland, N.J., for the generous donation of vascular graft material. We acknowledge the excellent technical support provided by Holly Kelly and Candace Krepel.
References
- . Infections in bionic man: the pathobiology of infection in prosthetic devices. Part I and II. Curr Probl Surg. 1982;19:1–314
- . Aortofemoral graft infection due to Staphylococcus epidermidis. Arch Surg. 1984;119:102–108
- The relevance of arterial wall microbiology to the treatment of prosthetic graft infections: graft infection vs. arterial infection. J Vasc Surg. 1984;1:750–756
- Infections in arterial reconstructions with synthetic grafts. Ann Surg. 1972;176:321–323
- . Vascular prosthetic infections: collected experience and results of treatment. Surgery. 1977;81:335–342
- . Synthetic graft infections: I. Graft infections. Surgery. 1983;93:733–746
- Surface biofilm disruption: enhanced recovery of microorganisms from vascular prostheses. Arch Surg. 1986;122:38–43
- . Incidence of positive inguinal lymph node cultures during peripheral revascularization. Am Surg. 1984;10:522–523
- . The microbiology of explanted vascular prostheses. Surgery. 1987;102:756–762
- . Infection of vascular prostheses caused by bacterial biofilms. J Vasc Surg. 1988;7:21–30
- . Bacterial-laden biofilms: a hazard to orthopedic prostheses. Infect Surg. 1984;3:655–662
- . The confusing and tenacious coagulase-negative staphylococci. Adv Intern Med. 1987;32:177–192
- Coagulase-negative staphylococci isolated from cerebrospinal fluid shunts: importance of slime production, species identification, and shunt removal to clinical outcome. J Infect Dis. 1987;156:548–555
- . Influence of bacterial adherence to intravascular catheters on in-vitro antibiotic susceptibility. Lancet. 1985;2:1266–1268
- . Microbiologic prediction of abdominal surgical wound infection. Arch Surg. 1987;122:33–37
- The necessity of long-term antibiotic therapy with positive arterial wall cultures. J Vasc Surg. 1988;8:262–267
- Characterization of clinically significant strains of coagulase-negative staphylococci. J Clin Microbiol. 1983;18:258–269
- . Simplified scheme for routine identification of human Staphylococcus species. J Clin Microbiol. 1975;1:82–88
- Effect of extracellular slime substance on the human cellular immune response. Lancet. 1984;1:365–367
- Interference with granulocyte function by Staphylococcus epidermidis slime. Infect Immun. 1986;54:13–20
☆ Supported by the Veterans Administration.
☆☆ Reprint requests: Dennis F. Bandyk, MD, Department of Surgery, MCMC, 8700 W. Wisconsin Ave., Milwaukee, WI 53226.
★ J Vasc Surg 1989;9:665–70.
PII: S0741-5214(89)70037-9
© 1989 Society for Vascular Surgery and International Society for Cardiovascular Surgery, North American Chapter. Published by Elsevier Inc. All rights reserved.
