- Email: [email protected]
Bacterial adherence to vascular prostheses
Bacterial adherence
vascular post
A determinant of graft David D. Schmitt, M.D., Dennis P. Bandyk, M.D., Jonathan B. Towne, M.D., M&W&~, Wis.
Arch J. Pequet, M.
An in vitro model was developed to quantitatively measure bacterial adherence to the surface of prosthetic vascular graft material. Four strains of bacteria (Staphylococw aweus, nonmucin-producing S. epidevmidis [SP-21, mucin-producing S. epidemnids p-121, and Eschevicbia coli) were used to inoculate expanded polytetrafluoroethylene (ePTl?E), woven Dacron, and velour knitted Dacron graft material. After graft specimens were incubated in a lo7 suspension of bacteria, they were washed to remove nonadherent organisms and ultrasonically oscillated to dislodge adherent organisms. Quantitative culture of the sonication effluent was used to calculate bacterial adherence, expressed as the number of colony-forming units found in each square centimeter of graft material per 10’ inoculum. All bacterial strains had a greater affinity to velour knitted Dacron graft than to ePTFE (p < 0.025). E. w 1i and S. atirew adhered to velour knitted Dacron in greater numbers than to woven Dacron (p < 0.04). The production of extracellular polysaccharide (mu&) by the U-12 strain significantly increased adherence to both ePTFE and Dacron grafts compared with the other three bacterial strains tested (p < 0.04). Although E. wli was less adherent to ePTFE than nonmucin-producing staphylococcal strains (S. awyem and SP-2), no difference in adherence to knitted or woven Dacron graft material was demonstrated. The differential adherence of bacteria to prosthetic vascular grafts pays an important role in the pathogenesis of graft sepsis and determines relative graft infectivity. The in vitro model developed is well suited for further study of the mechanisms by which bacteria adhere to and colonize vascular grafts. (J VAX SURG 1986; 3:732-40.)
The adhesion of bacteria to the surface of prosthetic implants is recognized as an important initial step of an infectious process.1-3 Marine microbiologists have long emphasized the importance of bacterial adhesion in the colonization of moving streams of water. Bacteria that fail to attach also fail to flourish and are washed away. Similarly in human beings, the relative adhesiveness of bacteria to implanted medical devices such as cardiac valves,4 cerebrospinal fluid shunts,5,6 and intravascular catheters7Tshas correlated with the predisposition of certain bacteria to produce clinical infection. Despite the serious consequences of prosthetic vascular graft infection, few reports concerning the bacterial adhesion onto vascular graft material are available. Improved prophylaxis against graft infection should result if the mechFrom the Department of Surgery, The Medical College of Wisconsin. Presented at the Ninth Annual Meeting of the Midwestern Vascular Surbxl Societv. Chicano. III.. Scot. 6-7. 1985. Reprint requuests: Dent& F. B&&k, M.6., Department of Surgery, Medical College of Wisconsin, 8700 W. Wisconsin Ave., Milwaukee. WI 53226.
732
anisms by which bacteria adhere to graft surfaces and establish microcolonies are better understood. Bacterial adherence ro biomaterials has been shown to be dependent on many factors, including the bacterial species, physical properties and chemical composition of the material, the duration of cxposure, and the protein biolayer that forms on all prosthetic surfaces after graft implantation.9~10 Bacterial attachment and the formation of microcolonies are mediated by a mass of tangled polysaccharide fibers that extends from the bacterial cell surface and forms a felt-like glycocalyx surrounding an individual cell or a colony of cells. In a competitive natural environment, the glycocalyx is essential for biologic suscess by favoring a particular species over others in the population, by protecting the microorganisms from phagocytes and biosides, and by acting as an ion exchange resin for the transport of organic nutrients.1’-‘3 Protection by the glycocalyx results in the formation of microbiologic reservoirs that can then produce chronic or late-appearing infections. Although virtually any organism can infect a vascular graft, the staphylococcal species (StapkyLococms
Volume 3 Number 5 May 1986
Bacmial
aweus and S. epidemzidis) and the gram-negative organism, Eschevichia coli, are the most common pathogens reported. 14-16The cell wall structures of grampositive and gram-negative bacteria differ, thereby influencing their adherence to biomaterials. S. aureew has been demonstrated to adhere to suture material in as high as 100 times greater numbers than E. coli.l7 A difference in bacterial affinity may also have a significant effect on tissue susceptibility to infection in the presence of various types of vascular grafts. Recently, S. epidemidis strains that exhibit mucoid growth have been implicated as the prevalent pathogen causing late aortofemoral graft infection.Q8 The production of an extracellular mucinoid substance by these microorganisms may promote the sequestration of bacteria in the interstices of the graft material and account for the late postoperative appearance and inherent antibiotic resistance of this graft infection. Vascular graft composition and construction can also influence bacterial adherence. Graft fiber surface characteristics, the relative degree of material hydrophobicity, and the presence of anionic vs. cationic surface charge all affect initial bacterial adherence.‘,4 Dacron vascular grafts have been shown to have a greater propensity for bacterial adherence than expanded polytetrafluoroethylene (ePTFE) grafts.‘~” As new vascular grafts are developed, graft characteristics such as low bacterial affinity and resistance to colonization may be important in reducing the incidence of both early and late graft sepsis. The purpose of this study was to examine the differential adherence of various bacterial species to three types of commonly used prosthetic vascular grafts. An in vitro model was developed to quantitatively measure the bacterial adherence to vascular graft material. The methodology developed should prove useful for the further delineation of mechanisms by which bacteria adhere to prosthetic surfaces. MATERIAL
AND
METHODS
Bacterial adherence to the vascular graft surface was determined by a quantitative culture technique that calculated the number of adherent, viable bacteria dislodged from the graft surface after ultrasonic oscillation. Scanning electron microscopy (SEM) and graft material culture were used to verify the effectiveness of bacterial removal by the sonication process, Vascular grafts. Two types of commercially obtained sterile vascular grafts were used: ePTFE (Gore-Tex, W. L. Gore & Associates, Inc., Elkton, Md.), woven Dacron graft, and single-velour knitted
adherence to vascular prostheses 733
Dacron (Meadox Medicals, Inc., Oakland, N.J.). The graft specimens used in the calculation of bacterial adherence were cut into 1 cm squares and sterilized. Both surfaces of the graft specimens were vulnerable to bacterial exposure during immersion in a bacterial suspension. Bacterial adherence was calculated for each square centimeter of graft material. Bacteria. Two sources of bacterial specimens were used. These were standard American Type Culture Collection (ATCC) strains of S. utiyegs (ATCC No. 25923) andE. coli (ATCC No. 25922), and two clinical isolates of S. epidemidis. The S. epidemidis strains were identified with the STAPHIDENT system (Analytab Products, Plainview, N.Y.) and included a mucin-producing strain (RP-12) and a nonmu&-producing strain (SP-2). The two S. epidefpmidis strains were catalase-positive, coagulasenegative, and sensitive to novobiocin and fermented glucose, but not mannitol. Only the RP-12 strain produced mucoid colonies on glucose-supplemented tryptic soy agar that stained with Alcian blue. The staining of RP-12 colonies with Alcian blue demonstrates the production of an extracellular polysaccharide (mucin) by th.e bacteria. Bacterial adherence determination. The four strains of bacteria were plated on blood agar for 18hour growth. Five colonies were then transferred with a sterile wire loop to 10 ml of trypticase soy broth (TSB) with 0.25% dextrose and incubated for 18 hours at 35” C. Ten ml of SP-2 broth, 5 ml of RI’-12 and S. aulreus broth, and 1 ml ofE. coli broth were diluted with 40 ml of TSB with 0.25% dextrose and incubated for 4 hours at 35” 6. This produced bacterial inocula in eariy logarithmic growth phase. The bacterial suspensions were then centrifuged at 3500 rpm for 10 minutes and resuspended three times in 0.01 M phosphate-buffered saline (PBS) solution at pH 7.4. The third resuspension was gently oscillated ultrasonically (Fischer Sonic Dismembrator, Artek Systems Corp., Farmingdale, N.Y.) at 20 kHz for 90 seconds to disrupt bacterial clusters. Optical density of the suspension was measured at 660 nm with a spectrophotometer (Spectronic 21, Bausch & Lomb Incorporated Rochester, N.Y.) to adjust the concentration of the inoculating solution to 1 X lo7 organisms (colonyforming units [CFU]) per milliliter. Inoculum concentrations were verified by backplating three times at a lo- 5 dilution on tryptic soy agar. The final inoculating solutions were composed of one of the four bacterial strains added to the PBS solution with 0.25% dextrose. The dextrose was used to provide
734
Journd of VASCULAR SURGERY
Schnitt et al.
Fig. 1. Scanning electron microscopy shows effect of washing and sonication on removal of Staphylococczu epidempzidis (RI?-12) organisms from surfaceof ePTFE graft ( x 2000). A, Surface before nonadherent organisms are washed off. B, Surface after washing but before ultrasonic oscillation to dislodge adherent organisms. C, Graft surface after sonication with no organisms visible.
substrate for bacterial viability and for mucin production by the RP-12 strain. Graft specimens were immersed in the inoculating solutions for 24 hours at room temperature. After the incubation, the inoculum was decanted and the graft specimens were washed 10 times with 50 ml of PBS solution. Organisms remaining on the graft specimens after washing were judged to be adherent although further washing would remove additional small numbers of organisms. Each graft specimen was placed in 40 ml of PBS solution and sonicated at 20 kHz for 10 minutes to dislodge adherent organisms from the graft material. The sonication effluent was then backplated at 10e3 and 10m4 dilutions on tryptic soy agar. The quantitative culture plates were read at 24 and 48 hours. Bacterial adherence was determined as the number of CFUs for each centimeter of graft material dislodged from the specimen and was calculated by multiplying the num-
ber of CFUs for each milliliter of sonication effluent times 40 (the volume of the sonication effluent). To determine the differential adherence between strains tested, colony counts were standardized at an inoculum of IO7 organisms (CFUs) per milliliter. Final results were expressed as the number of CFUs found in each square centimeter of graft material per lo7 inoculum. The efficiency of ultrasonic oscillation in dislodging bacterial colonies from the graft material was assessed by both graft culture and SEM. After sonication, graft specimens were touched directly onto tryptic soy agar plates 10 times. Plates were incubated for 24 hours at 35” C, and the number of bacterial colonies was visually counted. Similar graft specimens were stained with Al&m blue or ruthenium red to detect the presence of mucin production and placed in 2% giutaraldehyde solution for SEM studies.
Volume 3 Number 5 May 1986
Bacterial adijeren,
8. epldermidlr (RP-12) n=6
GRAFT TYPE
PTFE
(l.4*o.3)xlo5
VELOUR KNITTED DACRON
E. cotI
S. aureux
S. epldermldis (SP-2) n=6
(1.8*p.2)x104
n=5
“X5
(9.4y.4)xlo3
(t.6yl5)xld
.
piz-p&q~ .
D 40.025
I I’ WOVEN DACRON
ND
to vasculm prostheses 735
pGq D 4 0.01
ND
,
II
( (8.7+_1.7)xl$
11 (6.6*p.7)Xl$
p 4 0.01
pI
1 0.01
Fig. 2. Quantitative bacterial adherence: Graft type variation. Data are expressedas number of colony-forming units per square centimeter of graft material per 10’ inoculum; data in boxed areas compared for statistical significance, with inner boxes indicating statistically significant differences; p value determined by analysisof variance. ND = not determined.
GRAFT
8.
PTFE
l pldermldl8 (RP-12)
TYPE
&,
S. epldermldls
s. aweus
E. cdl
(SP-2)
~q.*~:~xqo’
uJ.4:::xqo3
(q.6..j ip (0.04
VELOW) KNITTED DACRON
1(2.9*0.8)x1oIl
(3.3+0.3)x105
(3.4+o.2)xlo5
(5.1~o.s)xlos p 49.94
WOVEN DACRON
ND
ND
(8.7*_1.7)x10’
(6.6~0.7)Xld
Fig. 3. Quantitative bacterial adherence: Speciesvariation. Data are expressed as number of colony-forming units per square centimeter of graft material per 10’ inoculum; data in boxed areas compared for statistical significance, with inner boxes indicating statistically significant differences; p value determined by analysisof variance. ND = not determined.
Specimens for SEM were prepared as follows, After glutaraldehyde fixation for a minimum of 48 hours, each graft specimen was rinsed in PBS solution and placed in 2% phosphate-buffered osmium followed by three rinses in distilled water. Graft materials were then serially dehydrated in ethanol and critical-point dried with carbon dioxide, mounted on metal stubs, and coated with gold palladium. Representative graft specimens were viewed in a scanning electron microscope (model 100, Advanced Metallurgical Research, Bedford, Mass.). Data from bacterial adherence studies were compared for statistical significance with the two-tailed Student t test and analysis of variance, with p < 0.05 considered significant.
RESULTS Efficiency of ultrasonic oscillation in removal of adherent bacteria. Sonication of the graft specimens removed essentially all adherent organisms from the surface of the graft fibers. Although cultures of the graft specimens after sonication were consistently positive, the total number of colonies counted from the 10 graft-touch imprints varied from a total of 800 colonies for velour Dacron to less than 10 colonies for ePTFE. Before sonication, velour and woven graft specimens imprinted onto tryptic soy agar produced conflulent growth after 24 hours of incubation. Graft specimens examined by SEM before washing with PBS solution and before and after sonication verified both the effect of washing in re-
736
Journd of ‘VASCULAR SURGERU
Schmitt et al.
Fig. 4. Scanning electron microscopy of adherent (A) muck-producing RP- 12, (B) nonmucinproducing Q-2 S. epiderunidis,(C), S. aweus, and (D) E. colimicroorganisms on ePTFE graft surface after 1 hour of incubation ( x 10,000). Morphology of attachment is similar. Al1 strains attach as single bacterium, distinctly outlined. No evidence of extracellular glycocalyx production is visible.
moving nonadherent bacteria and the efficiency of sonication in dislodging adherent bacterial colonies (Fig. I). After ultrasonic oscillation, no bacteria were visible on the graft material surface. Quantitative adherence studies. Figs. 2 and 3 depict the adherence of the four strains of bacteria to el?TFE, woven Dacron, and velour knitted Dacron vascular graft material expressed as the number of CFUs for each square centimeter of graft material at an immersion inoculum of lo7 organisms per milliliter. In these studies the incubation time of the graft material in the bacterial suspensions was 24 hours. Bacterial adherence was dependent on both the bacterial strain and the graft material tested. All bacterial strains had greater affinity to velour knitted Dacron graft compared with ePTFE (BP-12, p < 0.025; S. auvetis, SP-2, E. coli, p < 0.01). In adherence studies with S. aureus and E. coli strains, woven Dacron graft material was included. Both S. aG?egs and E. coli or-
ganisms adhered in greater numbers to velour knitted Dacron fibers compared with woven Dacron, p < 0.04 and p < O.OI, respectively. The bacterial strains also demonstrated differential adherence to the two types of vascular graft material. The mucin-producing S. epidekdis strain (BP-X?) had greater adherence to both ePTFE and velour knitted Dacron grafts compared with nonmu&n-producing S. epidemidis (SP-2) (p < 0.02), S. aureus (p < 0.04), andE. wli (p < 0.04) (Fig. 3). No significant differential adherence to knitted Dacron grafts was found between S. uuyetis, SP-2, and E. cob. The increased adherence of the BP-12 strain of S. epidemidis was due to the production of an extracellular mucinoid substance that stained with both Alcian blue and ruthenium
red. The production
of
mucin was dependent on the duration of incubation time. After 1 hour of incubation, the morphology of
Volume 3 Number 5 iMay 1986
Bactetial
adhvence
to vmcula~ prostheses 737
adhesion to the graft surface of the four bacterial strains was similar (Fig. 4). Bacteria adhered as single organisms, and the BP-12 strain demonstrated no evidence of mucin production by Alcian blue or SEM examination. No significant difference in quantitative bacterial adherence could be demonstrated between the two S. epiden?zidis strains after 1 hour of incubation. After 24 and 48 hours of incubation, colonies of BP-12 adherent to the graft specimens demonstrated extensive production of an extracellular mucinoid substance that stained with Alcian blue and ruthenium red (Fig. 5). SEM of the graft material showed the BP-12 in colonies surrounded by an amorphous extracellular substance. No similar substance could be demonstrated by either histochemical staining or SEM graft specimens immersed in suspensions of Q-2, S. aweus, or E. coli for 24 hours. DISCUSSION All prosthetic vascular grafts are susceptible in varying degrees to infection via direct contamination during implantation or bacteremia after operation. The mechanisms by which bacteria adhere to the graft surface and establish microcolonies are a fundamental step in the pathogenesis of graft infection. Although bacterial adhesion to the surface of implanted prosthetic devices is believed to play an important role in the pathogenesis of infections involving foreign bodies, the process of colonization and the subsequent emergence of clinical infection remain poorly defined. Bacterial adherence to prosthetic vascular grafts was demonstrated in the present study to be dependent on the bacterial species, the type of graft material, the method of graft construction, and bacterial production of an extracellular polysaccharide. On the basis of available data concerning the pathologic features of graft infection in human beings, the differential adherence of bacteria to vascular grafts is undoubtedly important not only in the pathogenesis of graft infection but also in its prevention, recognition, and treatment. Previous studies of bacterial adherence to vascular prostheses have used in vitro models and radiolabeled bacteria to quantitatively measure the differential adherence of graft materials and bacterial species.‘,19 The experimental model developed in this study yielded reproducible results both in the quantitative calculations of bacterial adherence and qualitatively by SEM. The use of ultrasonic oscillation is advantageous over radiolabeled methods because only organisms that remain viable on the graft surface are included in the adherence measurements, It is unknown whether the process of radiolabeling alters
Fig. 5. Scanning electron microscopy of adherent mucinproducing (RP-12) microorganisms on velour Dacron graft surface after (A) 24 hours of incubation (X 5000) and (B) 48 hours of incubation ( x 2000). RP-12 organisms are seen in greater numbers and form microcolonies buried in an amorphous ground substance produced by bacteria.
the bacterial cell wall properties or the ability of organisms to produce their extracellular glycocalyx important for adhesion and survival in a natural environment. SEM of the graft specimen surface after ultrasonic oscillation documented convincingly the efficacy of sonication in removing adherent microcolonies that remained after repeated washing. The in vitro method developed should prove useful in evaluating bacterial adherence characteristics of newly developed graft material and identifying bacterial-graft interactions that might be interrupted or favorably altered. When in vitro cultures are used to study bacterial adherence, several limitations should be emphasized. In vitro culture of single species, devoid of antibacterial factors, may favor more mobile “swarmer” cells over the glycocalyx-enclosed cells typically found in natural environments. In contrast, the inclusion of antibiotics, prosthetic surfaces, or specific nutrient
JoumaP oi VASCULAR SURGERY
conditions may promote glycocalyx production and cell division primarily within the hydrated exopolysaccharide matrix, producing microcolonies of morphologically identical cells.” Growth conditions that favor glycocalyx production will mediate both microcolony formation and adhesion, with the eventual formation of an adherent biofilm. At present it is unknown whether a reduction in bacterial adherence will result in a concomitant decrease in prosthetic infectivity. Bacterial adherence onto vascular graft material was observed to depend on both graft material and construction in this study. Bacterial adherence was greatest to velour knitted Dacron, less to woven Dacron, and least to ePTFE graft material. Surface area differences between the ePTFE graft material and the two types of Dacron grafts may account for some of the measured difference in bacterial attachment. Graft material of ePTFE is relatively nonporous compared with the multifilamented Dacron graft fabric. The chemical structure of ePTFE grafts can also contribute to a difference in bacterial affinity. PTFE is more hydrophobic than Dacron and therefore less likely to form bonds with those bacteria in which the cell walls have hydrophobic properties. The concept of the relative hydrophobicity of biomaterial and its influence on bacterial adhesion has been demonstrated in studies of bacterial adherence to suture material” and to intravascular catheters.3 Staphylococcal species were observed both quantitatively (radiolabeled bacteria) and qualitatively (SEM) to adhere to sutures and intravascular catheters in greater numbers than did E. coli. A gram-positive ce!l wall, such as that of S. @@yeus,consists of a single rigid layer of peptidoglycan, whereas a gram-negative cell, such as that of E. coli, is a multilayered, complex structure and consists of additional layers of liposaccharide and protein. The E. coli microorganisms also possessed a special organelle of motility, the flagellum. This property may also influence the dynamics of adsorption and desorption of organisms on the graft surface. Bacterial adhesion is initially a reversible process but develops into an irreversible process as the bacteria produce an extracellular adhesive. Morphologic observations by SEM demonstrated that the mucin-producing S. epide?vnidis strain adhered to graft surfaces in clusters whereas E. coli and nonmucin-producing staphylococcal strains tended to adhere as a single microorganism. This difference in attachment correlated with the production of an exopolysaccharide matrix that selectively stained with Alcian blue and ruthenium red. The method of Dacron graft construction was also demonstrated to alter the adherence of S. LUW~US
and E. cob. Both strains adhered in greater numbers to the velour knitted graft. The addition of the velour pile in graft construction is to provide anchorage of the prosthesis to the perigraft tissue. This graft characteristic produces an improved pseudointimal development on the luminal surface that has been shown experimentally to reduce the incidence of graft contamination from a bacteremic challenge.22 The porous nature of velour grafts increases the fiber surface area avaiable for bacterial attachment and may account for the differential adherence measured in this study. Because of the porosity of knitted grafts, preclotting with blood is required for clinical use to avoid hemorrhage through the interstices. The porous nature of these grafts in combination with specific bacteria properties, such as mucin production, may potentiate infection by promoting the sequestration of bacteria in areas inaccessible to local hosr defenses or antibacterial agents. Studies on bacterial adherence to nonabsorbable suture material have clearly demonstrated an increased adherence and slower removal of adhered bacteria by tissue defense factors from braided compared with monofilament suture.17~21 An important observation of this study was the correlation of the production of an extracellular polysaccharide material (mucin) by the BP-12 strain with an increased bacterial adherence. After a short incubation time (1 hour), no mucin production was apparent by Alcian blue staining and the adherence of S. aweus and the two strains of S. epidemidti were similar. After a longer incubation time (24 hours), the BP-12 strain produced mucin and the measured adherence to both ePTFE and velour knitted Dacron grafts was increased by 10 to 100 times compared with S. agreus and the nonmucin-producing S. epidemidis strains. Bacteria in aquatic systems live predominantly in exopolysaccharide-enclosed microcolonies adherent to surfaces. After initial colonization, production of this amorphous ground substance allows the bacterial microcolonies to develop into a continuous biofilm and an extensive fibrous anionic matrix of exopolysaccharide glycocalyx that contains more than 99.99% of the bacteria. This bacterial biofilm formation can persist despite intact host defenses and the presence of antibiotics.23 The glycocalyx matrix has been demonstrated to be a protective factor in vitro against antibodies, antibiotics, and phagocytes. “Jo The basic bacterial strategy is to live within protected adherent microcolonies in nutritionally favorable environments and to dispatch mobile celis to seek out other favorable niches and to establish new adherent microcolonies. Strains of S. epidemidis that produce a mucinoid
Volume 3 Number 5 May 1986
extracellular material have been implicated in infection involving implanted foreign devices, such as prosthetic heart valves,4 cerebrospinal fluid shunts,6 orthopedic appliances, 9 intravascular catheters,* and recently aortofemoral Dacron grafts.” Layers of mucinoid deposits are found not only in infections caused by S. epidemidis, but also in S. aulreus and Pseudornonm aew&osa infections. 121z4 The glycocalyx produced by specific bacteria appears to be important both in the initial adhesion to a foreign body surface and in the establishment of an adherent bacterial biofilm. Because of their stationary mode of growth, the bacteria that colonize a prosthetic surface may initially cause no overt symptoms or signs of infection.25 The bacterial populations may then persist for extended periods of time and act as a microbiologic reservoir. If local host defenses are activated, the resulting indolent inflammatory response can, in time, produce clinical signs of infection.26’27 In the case of prosthetic vascular grafts, the result can be a loss of suture line tensile strength and the formation of anastomotic aneurysms. The epidemiology of prosthetic vascular graft infection has changed during the past two decades. The incidence of graft sepsis has declined and early postoperative infection involving the graft itself is a rare occurrence (less than 1%) . The recognition of graft infection now occurs later in the postoperative period and less virulent staphylococcal species and gramnegative bacteria are replacing S. auyetis as the prevalent pathogens. 14-16In a lo-year review of aortofemoral graft infection, we found that only 3 of 30 (10%) infections were treated within 4 months of graft implantation. In the remaining 27 infections S. epidemidis was documented as the infecting organism in 18 cases (66%). The present-day predominance of late over early graft infection is difficult to explain, but the suppression of infection with prophylactic perioperative antibiotic administration and ‘a change in the virulence of the infecting organisms may be important factors. The present study suggests that the production of exopolysaccharide by some strains of bacteria is a virulence factor that promotes the adherence to vascular graft surfaces and the formation of a persistent bacterial biofilm. The contact of mucin-producing bacteria with a vascular graft during implantation or via bacteremia after operation may be an important initiating step in the pathogenesis of present-day, late graft infection. Our experience with aortofemoral graft infection confirms the ability of S. epidemidis to persist as a bacterial biofilm and to produce graft infection.” Diagnosis of graft infection caused by S. epidemidis required a high index of suspicion because of the
Bacterial
adherence to vascuLay prostheses
739
inability to identify a microorganism by routine wound culture or by Gram stain either before or at operation. A mucinoid fluid was typically found surrounding the graft with many polymorphonuclear leukocytes but no bacteria demonstrated by Gram stain. Culture of the graft fabric in trypticase soy broth confirmed that S. epidemidis was the infecting organism in 15 of 18 cases (83%), whereas cultures of the perigraft fluid or tissue were positive in only two cases (11%). Sampling error in culturing the graft biofilm, in which. most bacteria are buried, results in the occurrence of negative cultures. Accurate identification of the primary pathogen requires culture of the disrupted graft biofilm. Ultrasonic oscillation can be used to break up aggregates containing bacteria, their exopolysaccharide glycocalyx, and organic detritus such as erythrocytes and fibrin, thereby freeing glycocalyx-enclosed adherent microorganisms for growth in culture media. Bacterial contamination of prosthetic vascular grafts during implantation is a frequent occurrence. Potential sources of contamination include skin, contaminated lymphatic vessels, organisms harbored in diseased arteries, and breaks in surgical technique during the operation. In a recent report by Macbeth et al.,28 positive cultures were obtained from 38 of 88 (43%) arterial walls during clean, elective vascular reconstructive procedures, including carotid endarterectomy and bypass of arterial occlusive disease. The most common organism culture was S. epideymidis, recovered in 71% of positive cultures. Of particular importance was the finding that all three postoperative graft infections reported in this study occurred in patients whose arterial wall cultures were positive at the time of the original operation, and in each case the infecting organism was S. epidevmidis. This study suggests that the diseased arterial wall is an important source of graft contamination and harbors a microflora similar to that identified as causing later aortofemoral graft infection. By interfering with bacterial adhesion and the formation of adherent microcolonies within the vascular graft surface biofilm, the potential for graft infection can be reduced. The in vitro model devela’ped in this study should prove useful in further defining the mechanisms in which bacteria interact with, prosthetic vascular grafts and in the development of methods that interfere with the initial step of the infectious process, bacterial adhesion. REFERENCES 1. Sugarman B. In vitro adherence of bacteria grafts. Infection 1982; 10:9-12. 2. Moore WS, Malone JM, Keown K. Prosthetic material. Arch Surg 1980; 115:1379-83.
to prosthetic arterial
graft
740
Schnitt
et al.
3. Hogt AH, Dankert J, DeVries JA, Feijen J. Adhesion of coagulase negative staphylococci to biomaterials. J Gen Microbiol 1983; 129:2959-68. 4. Slaughter L, Morris JE, Starr A. Prosthetic valvular endocarditis: A 12-year review. Circulation 1973; 47:1319-25. 5. Bayson R, Penny SR. Excessive production of mucoid substance in StaphylococcusSIIA: A possible factor in colonization of Holter shunts. Dev Med Child Nemo1 1972; 14:25-g. 6. Schoenbaum SC, Gardner P, Shillito J. Infection of cerebrospinal fluid shunts: Epidemiology, clinical manifestation, and therapy. J Infect Dis 1975; 131:543-51. 7. Peters G, Locci R, Pulverer G. Adherence and growth of coagulase-negative staphylococci on surfaces of intravenous catheters. J Invest Dis 1982; 146:479-82. 8. Sheth NK, Rose HD, Franson TR, Buckmire FL, Sohnle PG. In vitro quantitative adherence of bacteria to intravascular catheters. J Surg Res 1983; 34:213.-g. 9. Christenson GD, Simpson WA, Bisno AL, Beachey EH. Adherence of slime-producing strains of Staphylococcusepidemidis to smooth serfaces. Infect Immm 1982; 37:318-26. 10. Costerton JW, Geesey GG, Cheng K-J. How bacteria stick. Sci Am 1978;238:86-95. 11. Baltimore RS, Mitchell M. Immunologic investigation of mucoid strains of Pseua&ondc aerzzginosa.Comparison of susceptibility of opsonic antibody in mucoid and non-mucoid strains. J Infect Dis. 1980; 141:238-47. 12. Schwarzmann S, Boring III JR. Antiphagocytic effect ofslime from a mucoid strain of Pseudomonasaem&nosa. Comparison of susceptibility of opsonic antibody in mucoid and nonmucoid strains. J Infect Dis 1980; 141:238-47. 13. Ladd TI, Costerton JW, Geesey GG. Determination of the heterotrophic activity of epilithic microbial population. In: Costerton JW, Colwell RR, eds. Native aquatic bacteria: Enumeration activity and ecology. Technical Publication No. 695. Philadelphia: American Society of Testing Materials, 1979: 180-95. 14. Szilagyi DE, Smith DF, Elliott JP, Vrandecic MP. Infection in arterial reconstruction with synthetic grafts. Ann Surg 1972; 176:321-33. 15. Goldstone J, Moore WS. Infection in vascular prostheses. Am J Surg 1974; 128:225-33. 16. Lickweg Jr WF, Greenfield LF. Vascular prosthetic infection:
17. 18. 19. 20.
21. 22. 23.
24. 25. 26. 27.
28.
Collected experience and results of treatment. Surgery 1977; 81:335-42. Chih-Chang C, Williams DF. Effects af physical configuration and chemical structure of materials on bacterial adhesion. Am J Surg 1984; 147:197-204. Bandvk DF, Berni GA, Thiele BL, Towne JR. Aortofemoral graft .infection due to Staphylococcusepidemidti. Arch Surg 1984; 119:102-g. Costerton JW, Irvin RT, Cheng K-J. The role of bacterial surface structure in pathogenesis. CRC Crit Rev Microbial 1980; 8:303-g. Govan JRW, Fyfe JM. Mucoid pseudomonas aeruginosa and . cystic fibrosis: Resistance of the mucoid form to carbenicillin, Aoxicillin, tobramycin and the isolation of mucoid variants in vitro. J Antimicrob Chemother 1978; 4:233-40. Katz S, Izhar M, Mirelman D. Bacterial adherence ro surgical sutures: A possible factor in suture-induced infection. Ann Surg 1981; 194:35-41. Roon A, Malone JB, Moore WS, Bean B, Campagna G. Bacteremic infectability: A function of vascular graft material and design. J Surg Res 1977; 22:489-98. Marrie TJ, Noble MA, Costerton JW. Examination of the morpholoby of bacteria adhering to peritoneal dialysis cath eters by scanning and transmission electron microscopy. J Clm Microbial 1983; IS:138898. Gristina AG, Costerton JW, McGanity PI,. Bacteria-laden biofilms: A hazard to orthopedic prostheses. Infect Surg 1984; 3:655-62. Marrie TJ, Nelligan J, Costerton JW. A scanning and transmission electron microscopic study of infected endocardial pacemaker lead. Circulation 1982; 66:133941. Dougherty SH, Simmons RL. Infections in bionic man: The pathobioloar of infection in prosthetic devices-parts I & II. Curr Probi Surg 1982; 19:1-314. Archer GL. Antimicrobial susceptibility and selection of resistance among Staphylococm epidennidti isolates recovered from patients with indwelling foreign devices. Antimicrob Agents Chemother 1978; 14:353-9. Macbeth GA, Rubin JR, McIntyre Jr KE, Goidstone J, Malone JM. The relevance of arterial wall microbiology to the treatment of prosthetic graft infections. J VASC SURC 1984; 1:750-6.
vascular post
A determinant of graft David D. Schmitt, M.D., Dennis P. Bandyk, M.D., Jonathan B. Towne, M.D., M&W&~, Wis.
Arch J. Pequet, M.
An in vitro model was developed to quantitatively measure bacterial adherence to the surface of prosthetic vascular graft material. Four strains of bacteria (Staphylococw aweus, nonmucin-producing S. epidevmidis [SP-21, mucin-producing S. epidemnids p-121, and Eschevicbia coli) were used to inoculate expanded polytetrafluoroethylene (ePTl?E), woven Dacron, and velour knitted Dacron graft material. After graft specimens were incubated in a lo7 suspension of bacteria, they were washed to remove nonadherent organisms and ultrasonically oscillated to dislodge adherent organisms. Quantitative culture of the sonication effluent was used to calculate bacterial adherence, expressed as the number of colony-forming units found in each square centimeter of graft material per 10’ inoculum. All bacterial strains had a greater affinity to velour knitted Dacron graft than to ePTFE (p < 0.025). E. w 1i and S. atirew adhered to velour knitted Dacron in greater numbers than to woven Dacron (p < 0.04). The production of extracellular polysaccharide (mu&) by the U-12 strain significantly increased adherence to both ePTFE and Dacron grafts compared with the other three bacterial strains tested (p < 0.04). Although E. wli was less adherent to ePTFE than nonmucin-producing staphylococcal strains (S. awyem and SP-2), no difference in adherence to knitted or woven Dacron graft material was demonstrated. The differential adherence of bacteria to prosthetic vascular grafts pays an important role in the pathogenesis of graft sepsis and determines relative graft infectivity. The in vitro model developed is well suited for further study of the mechanisms by which bacteria adhere to and colonize vascular grafts. (J VAX SURG 1986; 3:732-40.)
The adhesion of bacteria to the surface of prosthetic implants is recognized as an important initial step of an infectious process.1-3 Marine microbiologists have long emphasized the importance of bacterial adhesion in the colonization of moving streams of water. Bacteria that fail to attach also fail to flourish and are washed away. Similarly in human beings, the relative adhesiveness of bacteria to implanted medical devices such as cardiac valves,4 cerebrospinal fluid shunts,5,6 and intravascular catheters7Tshas correlated with the predisposition of certain bacteria to produce clinical infection. Despite the serious consequences of prosthetic vascular graft infection, few reports concerning the bacterial adhesion onto vascular graft material are available. Improved prophylaxis against graft infection should result if the mechFrom the Department of Surgery, The Medical College of Wisconsin. Presented at the Ninth Annual Meeting of the Midwestern Vascular Surbxl Societv. Chicano. III.. Scot. 6-7. 1985. Reprint requuests: Dent& F. B&&k, M.6., Department of Surgery, Medical College of Wisconsin, 8700 W. Wisconsin Ave., Milwaukee. WI 53226.
732
anisms by which bacteria adhere to graft surfaces and establish microcolonies are better understood. Bacterial adherence ro biomaterials has been shown to be dependent on many factors, including the bacterial species, physical properties and chemical composition of the material, the duration of cxposure, and the protein biolayer that forms on all prosthetic surfaces after graft implantation.9~10 Bacterial attachment and the formation of microcolonies are mediated by a mass of tangled polysaccharide fibers that extends from the bacterial cell surface and forms a felt-like glycocalyx surrounding an individual cell or a colony of cells. In a competitive natural environment, the glycocalyx is essential for biologic suscess by favoring a particular species over others in the population, by protecting the microorganisms from phagocytes and biosides, and by acting as an ion exchange resin for the transport of organic nutrients.1’-‘3 Protection by the glycocalyx results in the formation of microbiologic reservoirs that can then produce chronic or late-appearing infections. Although virtually any organism can infect a vascular graft, the staphylococcal species (StapkyLococms
Volume 3 Number 5 May 1986
Bacmial
aweus and S. epidemzidis) and the gram-negative organism, Eschevichia coli, are the most common pathogens reported. 14-16The cell wall structures of grampositive and gram-negative bacteria differ, thereby influencing their adherence to biomaterials. S. aureew has been demonstrated to adhere to suture material in as high as 100 times greater numbers than E. coli.l7 A difference in bacterial affinity may also have a significant effect on tissue susceptibility to infection in the presence of various types of vascular grafts. Recently, S. epidemidis strains that exhibit mucoid growth have been implicated as the prevalent pathogen causing late aortofemoral graft infection.Q8 The production of an extracellular mucinoid substance by these microorganisms may promote the sequestration of bacteria in the interstices of the graft material and account for the late postoperative appearance and inherent antibiotic resistance of this graft infection. Vascular graft composition and construction can also influence bacterial adherence. Graft fiber surface characteristics, the relative degree of material hydrophobicity, and the presence of anionic vs. cationic surface charge all affect initial bacterial adherence.‘,4 Dacron vascular grafts have been shown to have a greater propensity for bacterial adherence than expanded polytetrafluoroethylene (ePTFE) grafts.‘~” As new vascular grafts are developed, graft characteristics such as low bacterial affinity and resistance to colonization may be important in reducing the incidence of both early and late graft sepsis. The purpose of this study was to examine the differential adherence of various bacterial species to three types of commonly used prosthetic vascular grafts. An in vitro model was developed to quantitatively measure the bacterial adherence to vascular graft material. The methodology developed should prove useful for the further delineation of mechanisms by which bacteria adhere to prosthetic surfaces. MATERIAL
AND
METHODS
Bacterial adherence to the vascular graft surface was determined by a quantitative culture technique that calculated the number of adherent, viable bacteria dislodged from the graft surface after ultrasonic oscillation. Scanning electron microscopy (SEM) and graft material culture were used to verify the effectiveness of bacterial removal by the sonication process, Vascular grafts. Two types of commercially obtained sterile vascular grafts were used: ePTFE (Gore-Tex, W. L. Gore & Associates, Inc., Elkton, Md.), woven Dacron graft, and single-velour knitted
adherence to vascular prostheses 733
Dacron (Meadox Medicals, Inc., Oakland, N.J.). The graft specimens used in the calculation of bacterial adherence were cut into 1 cm squares and sterilized. Both surfaces of the graft specimens were vulnerable to bacterial exposure during immersion in a bacterial suspension. Bacterial adherence was calculated for each square centimeter of graft material. Bacteria. Two sources of bacterial specimens were used. These were standard American Type Culture Collection (ATCC) strains of S. utiyegs (ATCC No. 25923) andE. coli (ATCC No. 25922), and two clinical isolates of S. epidemidis. The S. epidemidis strains were identified with the STAPHIDENT system (Analytab Products, Plainview, N.Y.) and included a mucin-producing strain (RP-12) and a nonmu&-producing strain (SP-2). The two S. epidefpmidis strains were catalase-positive, coagulasenegative, and sensitive to novobiocin and fermented glucose, but not mannitol. Only the RP-12 strain produced mucoid colonies on glucose-supplemented tryptic soy agar that stained with Alcian blue. The staining of RP-12 colonies with Alcian blue demonstrates the production of an extracellular polysaccharide (mucin) by th.e bacteria. Bacterial adherence determination. The four strains of bacteria were plated on blood agar for 18hour growth. Five colonies were then transferred with a sterile wire loop to 10 ml of trypticase soy broth (TSB) with 0.25% dextrose and incubated for 18 hours at 35” C. Ten ml of SP-2 broth, 5 ml of RI’-12 and S. aulreus broth, and 1 ml ofE. coli broth were diluted with 40 ml of TSB with 0.25% dextrose and incubated for 4 hours at 35” 6. This produced bacterial inocula in eariy logarithmic growth phase. The bacterial suspensions were then centrifuged at 3500 rpm for 10 minutes and resuspended three times in 0.01 M phosphate-buffered saline (PBS) solution at pH 7.4. The third resuspension was gently oscillated ultrasonically (Fischer Sonic Dismembrator, Artek Systems Corp., Farmingdale, N.Y.) at 20 kHz for 90 seconds to disrupt bacterial clusters. Optical density of the suspension was measured at 660 nm with a spectrophotometer (Spectronic 21, Bausch & Lomb Incorporated Rochester, N.Y.) to adjust the concentration of the inoculating solution to 1 X lo7 organisms (colonyforming units [CFU]) per milliliter. Inoculum concentrations were verified by backplating three times at a lo- 5 dilution on tryptic soy agar. The final inoculating solutions were composed of one of the four bacterial strains added to the PBS solution with 0.25% dextrose. The dextrose was used to provide
734
Journd of VASCULAR SURGERY
Schnitt et al.
Fig. 1. Scanning electron microscopy shows effect of washing and sonication on removal of Staphylococczu epidempzidis (RI?-12) organisms from surfaceof ePTFE graft ( x 2000). A, Surface before nonadherent organisms are washed off. B, Surface after washing but before ultrasonic oscillation to dislodge adherent organisms. C, Graft surface after sonication with no organisms visible.
substrate for bacterial viability and for mucin production by the RP-12 strain. Graft specimens were immersed in the inoculating solutions for 24 hours at room temperature. After the incubation, the inoculum was decanted and the graft specimens were washed 10 times with 50 ml of PBS solution. Organisms remaining on the graft specimens after washing were judged to be adherent although further washing would remove additional small numbers of organisms. Each graft specimen was placed in 40 ml of PBS solution and sonicated at 20 kHz for 10 minutes to dislodge adherent organisms from the graft material. The sonication effluent was then backplated at 10e3 and 10m4 dilutions on tryptic soy agar. The quantitative culture plates were read at 24 and 48 hours. Bacterial adherence was determined as the number of CFUs for each centimeter of graft material dislodged from the specimen and was calculated by multiplying the num-
ber of CFUs for each milliliter of sonication effluent times 40 (the volume of the sonication effluent). To determine the differential adherence between strains tested, colony counts were standardized at an inoculum of IO7 organisms (CFUs) per milliliter. Final results were expressed as the number of CFUs found in each square centimeter of graft material per lo7 inoculum. The efficiency of ultrasonic oscillation in dislodging bacterial colonies from the graft material was assessed by both graft culture and SEM. After sonication, graft specimens were touched directly onto tryptic soy agar plates 10 times. Plates were incubated for 24 hours at 35” C, and the number of bacterial colonies was visually counted. Similar graft specimens were stained with Al&m blue or ruthenium red to detect the presence of mucin production and placed in 2% giutaraldehyde solution for SEM studies.
Volume 3 Number 5 May 1986
Bacterial adijeren,
8. epldermidlr (RP-12) n=6
GRAFT TYPE
PTFE
(l.4*o.3)xlo5
VELOUR KNITTED DACRON
E. cotI
S. aureux
S. epldermldis (SP-2) n=6
(1.8*p.2)x104
n=5
“X5
(9.4y.4)xlo3
(t.6yl5)xld
.
piz-p&q~ .
D 40.025
I I’ WOVEN DACRON
ND
to vasculm prostheses 735
pGq D 4 0.01
ND
,
II
( (8.7+_1.7)xl$
11 (6.6*p.7)Xl$
p 4 0.01
pI
1 0.01
Fig. 2. Quantitative bacterial adherence: Graft type variation. Data are expressedas number of colony-forming units per square centimeter of graft material per 10’ inoculum; data in boxed areas compared for statistical significance, with inner boxes indicating statistically significant differences; p value determined by analysisof variance. ND = not determined.
GRAFT
8.
PTFE
l pldermldl8 (RP-12)
TYPE
&,
S. epldermldls
s. aweus
E. cdl
(SP-2)
~q.*~:~xqo’
uJ.4:::xqo3
(q.6..j ip (0.04
VELOW) KNITTED DACRON
1(2.9*0.8)x1oIl
(3.3+0.3)x105
(3.4+o.2)xlo5
(5.1~o.s)xlos p 49.94
WOVEN DACRON
ND
ND
(8.7*_1.7)x10’
(6.6~0.7)Xld
Fig. 3. Quantitative bacterial adherence: Speciesvariation. Data are expressed as number of colony-forming units per square centimeter of graft material per 10’ inoculum; data in boxed areas compared for statistical significance, with inner boxes indicating statistically significant differences; p value determined by analysisof variance. ND = not determined.
Specimens for SEM were prepared as follows, After glutaraldehyde fixation for a minimum of 48 hours, each graft specimen was rinsed in PBS solution and placed in 2% phosphate-buffered osmium followed by three rinses in distilled water. Graft materials were then serially dehydrated in ethanol and critical-point dried with carbon dioxide, mounted on metal stubs, and coated with gold palladium. Representative graft specimens were viewed in a scanning electron microscope (model 100, Advanced Metallurgical Research, Bedford, Mass.). Data from bacterial adherence studies were compared for statistical significance with the two-tailed Student t test and analysis of variance, with p < 0.05 considered significant.
RESULTS Efficiency of ultrasonic oscillation in removal of adherent bacteria. Sonication of the graft specimens removed essentially all adherent organisms from the surface of the graft fibers. Although cultures of the graft specimens after sonication were consistently positive, the total number of colonies counted from the 10 graft-touch imprints varied from a total of 800 colonies for velour Dacron to less than 10 colonies for ePTFE. Before sonication, velour and woven graft specimens imprinted onto tryptic soy agar produced conflulent growth after 24 hours of incubation. Graft specimens examined by SEM before washing with PBS solution and before and after sonication verified both the effect of washing in re-
736
Journd of ‘VASCULAR SURGERU
Schmitt et al.
Fig. 4. Scanning electron microscopy of adherent (A) muck-producing RP- 12, (B) nonmucinproducing Q-2 S. epiderunidis,(C), S. aweus, and (D) E. colimicroorganisms on ePTFE graft surface after 1 hour of incubation ( x 10,000). Morphology of attachment is similar. Al1 strains attach as single bacterium, distinctly outlined. No evidence of extracellular glycocalyx production is visible.
moving nonadherent bacteria and the efficiency of sonication in dislodging adherent bacterial colonies (Fig. I). After ultrasonic oscillation, no bacteria were visible on the graft material surface. Quantitative adherence studies. Figs. 2 and 3 depict the adherence of the four strains of bacteria to el?TFE, woven Dacron, and velour knitted Dacron vascular graft material expressed as the number of CFUs for each square centimeter of graft material at an immersion inoculum of lo7 organisms per milliliter. In these studies the incubation time of the graft material in the bacterial suspensions was 24 hours. Bacterial adherence was dependent on both the bacterial strain and the graft material tested. All bacterial strains had greater affinity to velour knitted Dacron graft compared with ePTFE (BP-12, p < 0.025; S. auvetis, SP-2, E. coli, p < 0.01). In adherence studies with S. aureus and E. coli strains, woven Dacron graft material was included. Both S. aG?egs and E. coli or-
ganisms adhered in greater numbers to velour knitted Dacron fibers compared with woven Dacron, p < 0.04 and p < O.OI, respectively. The bacterial strains also demonstrated differential adherence to the two types of vascular graft material. The mucin-producing S. epidekdis strain (BP-X?) had greater adherence to both ePTFE and velour knitted Dacron grafts compared with nonmu&n-producing S. epidemidis (SP-2) (p < 0.02), S. aureus (p < 0.04), andE. wli (p < 0.04) (Fig. 3). No significant differential adherence to knitted Dacron grafts was found between S. uuyetis, SP-2, and E. cob. The increased adherence of the BP-12 strain of S. epidemidis was due to the production of an extracellular mucinoid substance that stained with both Alcian blue and ruthenium
red. The production
of
mucin was dependent on the duration of incubation time. After 1 hour of incubation, the morphology of
Volume 3 Number 5 iMay 1986
Bactetial
adhvence
to vmcula~ prostheses 737
adhesion to the graft surface of the four bacterial strains was similar (Fig. 4). Bacteria adhered as single organisms, and the BP-12 strain demonstrated no evidence of mucin production by Alcian blue or SEM examination. No significant difference in quantitative bacterial adherence could be demonstrated between the two S. epiden?zidis strains after 1 hour of incubation. After 24 and 48 hours of incubation, colonies of BP-12 adherent to the graft specimens demonstrated extensive production of an extracellular mucinoid substance that stained with Alcian blue and ruthenium red (Fig. 5). SEM of the graft material showed the BP-12 in colonies surrounded by an amorphous extracellular substance. No similar substance could be demonstrated by either histochemical staining or SEM graft specimens immersed in suspensions of Q-2, S. aweus, or E. coli for 24 hours. DISCUSSION All prosthetic vascular grafts are susceptible in varying degrees to infection via direct contamination during implantation or bacteremia after operation. The mechanisms by which bacteria adhere to the graft surface and establish microcolonies are a fundamental step in the pathogenesis of graft infection. Although bacterial adhesion to the surface of implanted prosthetic devices is believed to play an important role in the pathogenesis of infections involving foreign bodies, the process of colonization and the subsequent emergence of clinical infection remain poorly defined. Bacterial adherence to prosthetic vascular grafts was demonstrated in the present study to be dependent on the bacterial species, the type of graft material, the method of graft construction, and bacterial production of an extracellular polysaccharide. On the basis of available data concerning the pathologic features of graft infection in human beings, the differential adherence of bacteria to vascular grafts is undoubtedly important not only in the pathogenesis of graft infection but also in its prevention, recognition, and treatment. Previous studies of bacterial adherence to vascular prostheses have used in vitro models and radiolabeled bacteria to quantitatively measure the differential adherence of graft materials and bacterial species.‘,19 The experimental model developed in this study yielded reproducible results both in the quantitative calculations of bacterial adherence and qualitatively by SEM. The use of ultrasonic oscillation is advantageous over radiolabeled methods because only organisms that remain viable on the graft surface are included in the adherence measurements, It is unknown whether the process of radiolabeling alters
Fig. 5. Scanning electron microscopy of adherent mucinproducing (RP-12) microorganisms on velour Dacron graft surface after (A) 24 hours of incubation (X 5000) and (B) 48 hours of incubation ( x 2000). RP-12 organisms are seen in greater numbers and form microcolonies buried in an amorphous ground substance produced by bacteria.
the bacterial cell wall properties or the ability of organisms to produce their extracellular glycocalyx important for adhesion and survival in a natural environment. SEM of the graft specimen surface after ultrasonic oscillation documented convincingly the efficacy of sonication in removing adherent microcolonies that remained after repeated washing. The in vitro method developed should prove useful in evaluating bacterial adherence characteristics of newly developed graft material and identifying bacterial-graft interactions that might be interrupted or favorably altered. When in vitro cultures are used to study bacterial adherence, several limitations should be emphasized. In vitro culture of single species, devoid of antibacterial factors, may favor more mobile “swarmer” cells over the glycocalyx-enclosed cells typically found in natural environments. In contrast, the inclusion of antibiotics, prosthetic surfaces, or specific nutrient
JoumaP oi VASCULAR SURGERY
conditions may promote glycocalyx production and cell division primarily within the hydrated exopolysaccharide matrix, producing microcolonies of morphologically identical cells.” Growth conditions that favor glycocalyx production will mediate both microcolony formation and adhesion, with the eventual formation of an adherent biofilm. At present it is unknown whether a reduction in bacterial adherence will result in a concomitant decrease in prosthetic infectivity. Bacterial adherence onto vascular graft material was observed to depend on both graft material and construction in this study. Bacterial adherence was greatest to velour knitted Dacron, less to woven Dacron, and least to ePTFE graft material. Surface area differences between the ePTFE graft material and the two types of Dacron grafts may account for some of the measured difference in bacterial attachment. Graft material of ePTFE is relatively nonporous compared with the multifilamented Dacron graft fabric. The chemical structure of ePTFE grafts can also contribute to a difference in bacterial affinity. PTFE is more hydrophobic than Dacron and therefore less likely to form bonds with those bacteria in which the cell walls have hydrophobic properties. The concept of the relative hydrophobicity of biomaterial and its influence on bacterial adhesion has been demonstrated in studies of bacterial adherence to suture material” and to intravascular catheters.3 Staphylococcal species were observed both quantitatively (radiolabeled bacteria) and qualitatively (SEM) to adhere to sutures and intravascular catheters in greater numbers than did E. coli. A gram-positive ce!l wall, such as that of S. @@yeus,consists of a single rigid layer of peptidoglycan, whereas a gram-negative cell, such as that of E. coli, is a multilayered, complex structure and consists of additional layers of liposaccharide and protein. The E. coli microorganisms also possessed a special organelle of motility, the flagellum. This property may also influence the dynamics of adsorption and desorption of organisms on the graft surface. Bacterial adhesion is initially a reversible process but develops into an irreversible process as the bacteria produce an extracellular adhesive. Morphologic observations by SEM demonstrated that the mucin-producing S. epide?vnidis strain adhered to graft surfaces in clusters whereas E. coli and nonmucin-producing staphylococcal strains tended to adhere as a single microorganism. This difference in attachment correlated with the production of an exopolysaccharide matrix that selectively stained with Alcian blue and ruthenium red. The method of Dacron graft construction was also demonstrated to alter the adherence of S. LUW~US
and E. cob. Both strains adhered in greater numbers to the velour knitted graft. The addition of the velour pile in graft construction is to provide anchorage of the prosthesis to the perigraft tissue. This graft characteristic produces an improved pseudointimal development on the luminal surface that has been shown experimentally to reduce the incidence of graft contamination from a bacteremic challenge.22 The porous nature of velour grafts increases the fiber surface area avaiable for bacterial attachment and may account for the differential adherence measured in this study. Because of the porosity of knitted grafts, preclotting with blood is required for clinical use to avoid hemorrhage through the interstices. The porous nature of these grafts in combination with specific bacteria properties, such as mucin production, may potentiate infection by promoting the sequestration of bacteria in areas inaccessible to local hosr defenses or antibacterial agents. Studies on bacterial adherence to nonabsorbable suture material have clearly demonstrated an increased adherence and slower removal of adhered bacteria by tissue defense factors from braided compared with monofilament suture.17~21 An important observation of this study was the correlation of the production of an extracellular polysaccharide material (mucin) by the BP-12 strain with an increased bacterial adherence. After a short incubation time (1 hour), no mucin production was apparent by Alcian blue staining and the adherence of S. aweus and the two strains of S. epidemidti were similar. After a longer incubation time (24 hours), the BP-12 strain produced mucin and the measured adherence to both ePTFE and velour knitted Dacron grafts was increased by 10 to 100 times compared with S. agreus and the nonmucin-producing S. epidemidis strains. Bacteria in aquatic systems live predominantly in exopolysaccharide-enclosed microcolonies adherent to surfaces. After initial colonization, production of this amorphous ground substance allows the bacterial microcolonies to develop into a continuous biofilm and an extensive fibrous anionic matrix of exopolysaccharide glycocalyx that contains more than 99.99% of the bacteria. This bacterial biofilm formation can persist despite intact host defenses and the presence of antibiotics.23 The glycocalyx matrix has been demonstrated to be a protective factor in vitro against antibodies, antibiotics, and phagocytes. “Jo The basic bacterial strategy is to live within protected adherent microcolonies in nutritionally favorable environments and to dispatch mobile celis to seek out other favorable niches and to establish new adherent microcolonies. Strains of S. epidemidis that produce a mucinoid
Volume 3 Number 5 May 1986
extracellular material have been implicated in infection involving implanted foreign devices, such as prosthetic heart valves,4 cerebrospinal fluid shunts,6 orthopedic appliances, 9 intravascular catheters,* and recently aortofemoral Dacron grafts.” Layers of mucinoid deposits are found not only in infections caused by S. epidemidis, but also in S. aulreus and Pseudornonm aew&osa infections. 121z4 The glycocalyx produced by specific bacteria appears to be important both in the initial adhesion to a foreign body surface and in the establishment of an adherent bacterial biofilm. Because of their stationary mode of growth, the bacteria that colonize a prosthetic surface may initially cause no overt symptoms or signs of infection.25 The bacterial populations may then persist for extended periods of time and act as a microbiologic reservoir. If local host defenses are activated, the resulting indolent inflammatory response can, in time, produce clinical signs of infection.26’27 In the case of prosthetic vascular grafts, the result can be a loss of suture line tensile strength and the formation of anastomotic aneurysms. The epidemiology of prosthetic vascular graft infection has changed during the past two decades. The incidence of graft sepsis has declined and early postoperative infection involving the graft itself is a rare occurrence (less than 1%) . The recognition of graft infection now occurs later in the postoperative period and less virulent staphylococcal species and gramnegative bacteria are replacing S. auyetis as the prevalent pathogens. 14-16In a lo-year review of aortofemoral graft infection, we found that only 3 of 30 (10%) infections were treated within 4 months of graft implantation. In the remaining 27 infections S. epidemidis was documented as the infecting organism in 18 cases (66%). The present-day predominance of late over early graft infection is difficult to explain, but the suppression of infection with prophylactic perioperative antibiotic administration and ‘a change in the virulence of the infecting organisms may be important factors. The present study suggests that the production of exopolysaccharide by some strains of bacteria is a virulence factor that promotes the adherence to vascular graft surfaces and the formation of a persistent bacterial biofilm. The contact of mucin-producing bacteria with a vascular graft during implantation or via bacteremia after operation may be an important initiating step in the pathogenesis of present-day, late graft infection. Our experience with aortofemoral graft infection confirms the ability of S. epidemidis to persist as a bacterial biofilm and to produce graft infection.” Diagnosis of graft infection caused by S. epidemidis required a high index of suspicion because of the
Bacterial
adherence to vascuLay prostheses
739
inability to identify a microorganism by routine wound culture or by Gram stain either before or at operation. A mucinoid fluid was typically found surrounding the graft with many polymorphonuclear leukocytes but no bacteria demonstrated by Gram stain. Culture of the graft fabric in trypticase soy broth confirmed that S. epidemidis was the infecting organism in 15 of 18 cases (83%), whereas cultures of the perigraft fluid or tissue were positive in only two cases (11%). Sampling error in culturing the graft biofilm, in which. most bacteria are buried, results in the occurrence of negative cultures. Accurate identification of the primary pathogen requires culture of the disrupted graft biofilm. Ultrasonic oscillation can be used to break up aggregates containing bacteria, their exopolysaccharide glycocalyx, and organic detritus such as erythrocytes and fibrin, thereby freeing glycocalyx-enclosed adherent microorganisms for growth in culture media. Bacterial contamination of prosthetic vascular grafts during implantation is a frequent occurrence. Potential sources of contamination include skin, contaminated lymphatic vessels, organisms harbored in diseased arteries, and breaks in surgical technique during the operation. In a recent report by Macbeth et al.,28 positive cultures were obtained from 38 of 88 (43%) arterial walls during clean, elective vascular reconstructive procedures, including carotid endarterectomy and bypass of arterial occlusive disease. The most common organism culture was S. epideymidis, recovered in 71% of positive cultures. Of particular importance was the finding that all three postoperative graft infections reported in this study occurred in patients whose arterial wall cultures were positive at the time of the original operation, and in each case the infecting organism was S. epidevmidis. This study suggests that the diseased arterial wall is an important source of graft contamination and harbors a microflora similar to that identified as causing later aortofemoral graft infection. By interfering with bacterial adhesion and the formation of adherent microcolonies within the vascular graft surface biofilm, the potential for graft infection can be reduced. The in vitro model devela’ped in this study should prove useful in further defining the mechanisms in which bacteria interact with, prosthetic vascular grafts and in the development of methods that interfere with the initial step of the infectious process, bacterial adhesion. REFERENCES 1. Sugarman B. In vitro adherence of bacteria grafts. Infection 1982; 10:9-12. 2. Moore WS, Malone JM, Keown K. Prosthetic material. Arch Surg 1980; 115:1379-83.
to prosthetic arterial
graft
740
Schnitt
et al.
3. Hogt AH, Dankert J, DeVries JA, Feijen J. Adhesion of coagulase negative staphylococci to biomaterials. J Gen Microbiol 1983; 129:2959-68. 4. Slaughter L, Morris JE, Starr A. Prosthetic valvular endocarditis: A 12-year review. Circulation 1973; 47:1319-25. 5. Bayson R, Penny SR. Excessive production of mucoid substance in StaphylococcusSIIA: A possible factor in colonization of Holter shunts. Dev Med Child Nemo1 1972; 14:25-g. 6. Schoenbaum SC, Gardner P, Shillito J. Infection of cerebrospinal fluid shunts: Epidemiology, clinical manifestation, and therapy. J Infect Dis 1975; 131:543-51. 7. Peters G, Locci R, Pulverer G. Adherence and growth of coagulase-negative staphylococci on surfaces of intravenous catheters. J Invest Dis 1982; 146:479-82. 8. Sheth NK, Rose HD, Franson TR, Buckmire FL, Sohnle PG. In vitro quantitative adherence of bacteria to intravascular catheters. J Surg Res 1983; 34:213.-g. 9. Christenson GD, Simpson WA, Bisno AL, Beachey EH. Adherence of slime-producing strains of Staphylococcusepidemidis to smooth serfaces. Infect Immm 1982; 37:318-26. 10. Costerton JW, Geesey GG, Cheng K-J. How bacteria stick. Sci Am 1978;238:86-95. 11. Baltimore RS, Mitchell M. Immunologic investigation of mucoid strains of Pseua&ondc aerzzginosa.Comparison of susceptibility of opsonic antibody in mucoid and non-mucoid strains. J Infect Dis. 1980; 141:238-47. 12. Schwarzmann S, Boring III JR. Antiphagocytic effect ofslime from a mucoid strain of Pseudomonasaem&nosa. Comparison of susceptibility of opsonic antibody in mucoid and nonmucoid strains. J Infect Dis 1980; 141:238-47. 13. Ladd TI, Costerton JW, Geesey GG. Determination of the heterotrophic activity of epilithic microbial population. In: Costerton JW, Colwell RR, eds. Native aquatic bacteria: Enumeration activity and ecology. Technical Publication No. 695. Philadelphia: American Society of Testing Materials, 1979: 180-95. 14. Szilagyi DE, Smith DF, Elliott JP, Vrandecic MP. Infection in arterial reconstruction with synthetic grafts. Ann Surg 1972; 176:321-33. 15. Goldstone J, Moore WS. Infection in vascular prostheses. Am J Surg 1974; 128:225-33. 16. Lickweg Jr WF, Greenfield LF. Vascular prosthetic infection:
17. 18. 19. 20.
21. 22. 23.
24. 25. 26. 27.
28.
Collected experience and results of treatment. Surgery 1977; 81:335-42. Chih-Chang C, Williams DF. Effects af physical configuration and chemical structure of materials on bacterial adhesion. Am J Surg 1984; 147:197-204. Bandvk DF, Berni GA, Thiele BL, Towne JR. Aortofemoral graft .infection due to Staphylococcusepidemidti. Arch Surg 1984; 119:102-g. Costerton JW, Irvin RT, Cheng K-J. The role of bacterial surface structure in pathogenesis. CRC Crit Rev Microbial 1980; 8:303-g. Govan JRW, Fyfe JM. Mucoid pseudomonas aeruginosa and . cystic fibrosis: Resistance of the mucoid form to carbenicillin, Aoxicillin, tobramycin and the isolation of mucoid variants in vitro. J Antimicrob Chemother 1978; 4:233-40. Katz S, Izhar M, Mirelman D. Bacterial adherence ro surgical sutures: A possible factor in suture-induced infection. Ann Surg 1981; 194:35-41. Roon A, Malone JB, Moore WS, Bean B, Campagna G. Bacteremic infectability: A function of vascular graft material and design. J Surg Res 1977; 22:489-98. Marrie TJ, Noble MA, Costerton JW. Examination of the morpholoby of bacteria adhering to peritoneal dialysis cath eters by scanning and transmission electron microscopy. J Clm Microbial 1983; IS:138898. Gristina AG, Costerton JW, McGanity PI,. Bacteria-laden biofilms: A hazard to orthopedic prostheses. Infect Surg 1984; 3:655-62. Marrie TJ, Nelligan J, Costerton JW. A scanning and transmission electron microscopic study of infected endocardial pacemaker lead. Circulation 1982; 66:133941. Dougherty SH, Simmons RL. Infections in bionic man: The pathobioloar of infection in prosthetic devices-parts I & II. Curr Probi Surg 1982; 19:1-314. Archer GL. Antimicrobial susceptibility and selection of resistance among Staphylococm epidennidti isolates recovered from patients with indwelling foreign devices. Antimicrob Agents Chemother 1978; 14:353-9. Macbeth GA, Rubin JR, McIntyre Jr KE, Goidstone J, Malone JM. The relevance of arterial wall microbiology to the treatment of prosthetic graft infections. J VASC SURC 1984; 1:750-6.