Bacterial adherence to vascular grafts after in vitro bacteremia

JOURNALOFSURGICALRESEARCH 38,648-655(1985)

Bacterial Adherence

to Vascular Grafts after in Vitro Bacteremia

JOHN E. ROSENMAN, M.D., WILLIAM H. PEARCE, M.D., AND RICHARD F. KEMPCZINSKI, M.D.’ Division of Vascular Surgery, Department of Surgery, University of Cincinnati Medical Center, 231 Bethesda Avenue, Cincinnati, Ohio 45267 Presented at the Annual Meeting of the Association for Academic Surgery, San Antonio, Texas, October 31-November 3, 1984 All currently used arterial prosthetics have a greater susceptibility to infection following bacteremia than does autogenous tissue. This experiment compares quantitative bacterial adherence to various prosthetic materials after bacteremia carried out in a tightly controlled and quantitative fashion. Ten centimeters long, 4 mm i.d. Dacron, umbilical vein (HUV), and polytetrafluoroethylene (PTPE) grafts, as well as PTPE grafts with a running suture line at the midportion were tested. Each graft was interposed into a pulsatile perfusion system modified from a Waters MOX 100 TM renal transplant pump. Indium-I I l-labeled Staphylococcus aureus were added to heparinized canine blood to give a mean concentration of 4.7 X lo6 bacteria/cc. This infected blood was recirculated through each graR for 30 min at a rate of 125 cc/m, 100 Torr (sys), 60 beats/min. The gamma counts/graft were used to calculate the number of bacteria/cm’ of graft surface. After nine experiments, a mean of 9.63 X lo5 bacteria/cm2 were adherent to the Dacron, 1.04 X 10’ bacteria/cm2 to the HUV, and 2.15 X 10” bacteria/cm2 to the PTPE. These differences were all significant at the 0.05 level. The addition of a suture line increased bacterial adherence to the PTPE graft by 50%. These results suggest that PTPE is the vascular graft material of choice when a prosthetic graft must be implanted despite a high risk of subsequent clinical bacteremia. Our in vitro, pulsatile perfusion model gave accurate and reproducible results, and appears well suited for further studies of bacterial, or platelet adherence to grafts, as well as the biomechanics of vascular conduits. 0 1985 Academic PRS, hc.

is most resistant to bacterial contamination [7]. There is a paucity of well-controlled data Despite major advances in diagnosis and regarding the relative infectability of the three management, grafi infection remains one of most commonly used arterial prostheses, the most dangerous and feared complications Dacron, polytetrafluoroethylene (PTFE), and of arterial prosthetic surgery [6, 16,201. Since human umbilical vein (HUV). This report the results of treatment for graft infection, describes the quantitative differences in baconce established, are poor, prophylaxis is an terial adherence to each of these three graft important tenet of vascular surgery [6, 16, types using an in vitro model. 201. All implanted prosthetics are susceptible in varying degrees to infection via direct METHODS contamination or bacteremia [7]. Thus, placement of any prosthetic graft in a patient Pulsatile perfusion system. The experiwho is at risk for bacteremia secondary to mental system used is shown in Fig. 1. A ongoing sepsis is fraught with potential for Waters MOX 100 TM renal transplant pulserious complications [7]. Unfortunately, satile pump was used to generate pulsatile, when the patient’s life or limb is in jeopardy, unidirectional flow through a Belzer pump it may be unavoidable. In such circumstances, at 60 beats/min. Latex tubing (8 mm) conit is preferable to use the graft material which nected the Belzer pump to a glass reservoir containing 250-500 cc of heparinized canine blood (1000 U/500 cc) which was maintained ’ To whom reprint requests should be addressed. INTRODUCTION

0022-4804/85 %1.50 Copyright 0 1985 by Academic Press, Inc. All rigbts of repmduction in any form reserved.

648

ROSENMAN, PEARCE, AND KEMPCZINSKI:

FIG. 1. Drawing of pulsatile perfusion system. (A) Experimental graft interposed in circuit, (B) Belzer pump head, (C) MOX 100 TMA renal transport pump, (D) reservoir with efferent and tierent tubes in place, (E) pressure monitor.

at 37°C by a water bath. Efferent blood flow was carried by latex tubing from the Belzer pump to a Plexiglas chamber containing the experimental grafts which had been interposed into the circulation. The efferent flow was then carried from the chamber back to the reservoir, completing the circuit. Constant pressure monitoring was accomplished by a standard Hewlett-Packard transducer and module connected to the afferent limb of the system via a “y” connection. The stroke output of the pulsatile pump was adjusted to deliver a flow rate of 125 cc/min. A screw clamp applied to the afferent tubing was used to create adequate “peripheral” resistance to produce a peak systolic pressure of 100 Torr. Bacterial labeling. The labeling technique was modified from Goodwin et al. [ 121. Stock frozen Staphylococcus aureus was thawed and incubated for 16 hr at 37°C in 5 cc soy tryptic nutrient broth. The bacteria were centrifuged for 10 min at 1OOOgand resuspended in approximately 1.2 cc normal saline. Quantitative cultures showed a mean yield of 1.4 X lo9 bacteria/cc (range 3.5 X lo*-2.2 X 109). This bacterial suspension (1 cc) was then incubated with approximately

GRAFTS

649

65 PCi of Indium-1 11-oxine (Amersham) at room temperature for 20 min, then centrifuged for 10 min at 1OOOg.The supernatant was discarded, and the cells were washed twice with normal saline. The final bacterial pellet was resuspended in 1 cc normal saline and counted in a dose calibrator. Labeling efficiency was 85-95%. A O.l-cc aliquot was removed for quantitative culture. In order to evaluate the reliability of the Indium-1 11 label, several samples of labeled bacteria were resuspended in 3 cc normal saline, incubated at 37°C and recentrifuged after 4 hr. The supernatant contained < 1% of the Indium- 111, thus, confirming minimal leak out of the radiolabel. Adherence of Iridium-1 11-labeled bacteria to prosthetic grafts. Four separate graft types were tested: (1) Knitted Dacron, crimped (CooleyMeadox). (2) Modified human umbilical vein (Dardik Biograft-Meadox). (3) PTFE (thin-wall Goretex). (4) Same as (3), but divided and resutured end-to-end with running simple 6-O prolene. All grafts except umbilical vein were 10 cm long with a 4 mm i.d. The umbilical vein grafts were 8 cm long with a 5 mm i.d., to standardize the surface area/graft. The umbilical vein graft was pretreated, as recommended by its manufacturer, with multiple rinses of heparinized normal saline, and then soaked in heparinized normal saline for 30 min prior to use. The Dacron grails were measured on stretch, and preclotted in standard fashion with unheparinized canine whole blood 30-60 min prior to perfusion. A total of nine separateexperiments were performed, each comparing the four different grafts. Approximately lo9 labeled S. aureus (range 3.5 X lo*-2.2 X 109)were added to the reservoir of canine blood. After 30 min of equilibration, the grafts were randomly interposed and perfused for 30 min. Care was taken to allow only the luminal surface of each graft to contact the per&sate. Upon removal of each graft, excessintraluminal blood was drained.

650

JOURNAL OF SURGICAL RESEARCH: VOL. 38, NO. 6, JUNE 1985

Severalaliquots were removedfrom the blood reservoir prior to perfusion of the first graft, and after perfusion of the final graft for measurement of partial thromboplastin times, fibrinogen level, platelet count, hematocrit, and bacterial concentration. The radioactivity of each graft and the blood samples were determined using a “Hippo” total body gamma counter and Tracer-Northern multichannel analyzer. Standard geometry was used for the counting of all samples. A standard count in the same geometry of 1 &i of indium-1 11 was performed for reference comparison. Autoradiography was performed on all grafts to assessthe distribution of labeled bacteria upon the graft surfaces. The grafts were fixed by 30 min of emersion in 1.0%gluteraldehyde, air dried, wrapped in cellophane, and layed against a sheet of Xray film (XK-1, Kodak) in a sealed cassette. Proper exposure at room temperature was determined by the method of Pearce et al. [ 191and varied from 1 to 200 hr. Scanning electron microscopy was performed on samples of each graft type after 30 min exposure to pulsatile perfusion in the artificial circulation without added bacteria. Calculations. The number of bacteria per square centimeter of each graft was derived from the gamma counts per minute per graft divided by its luminal surface area as well as the gamma counts per minute per bacterium,

The number of bacteria per square centimeter of each graft was normalized for the mean bacterial concentration of all nine experiments by dividing the mean bacterial concentration of the nine experiments by the bacterial concentration of an individual experiment, and multiplying this, in turn, by the number of bacteria per graft, BAC/cm2 (graft) = bac/cm2 (graft) X [BAC]/[bac], where bac/cm2 (graft), number of bacteria per graft; [bat], concentration of bacteria in blood of each experiment; BAG/cm’ (graft), number of bacteria per graft normalized for the mean bacterial concentration of all nine experiments; [BAC], mean bacterial concentration of all nine experiments; cpm, gamma counts per minute; mCi “‘In, microcuries of indium- 111. Statistical analysis of the data was performed using one-way analysis of variance (ANOVA) with Tukey’s studentized range test for comparison between groups. This test was adjusted to control the Type-l experimentwise error rate at the 0.05 level and analysis was done using Proc GLM of the Statistical Analysis System (SAS). RESULTS

The mean bacteremia of all nine experiments was 4.7 X lo6 bacteria/cc (range 2.62 bac/cm2 (graft) X 104-1.3 X 10’). The number of bacteria per square centimeter of each graft normalmm (graft) = ized for the mean bacteremia is shown on cm2 (graft) X mCi l”In/bac’ Table 1, and the statistical significance of X cpm (std “‘In)/mCi each comparison between the graft types is The bacterial concentration in the circu- shown in Table 2. There were no significant lating blood was derived from the average changes in the bacterial concentration, progamma counts per cubic centimeter of blood thrombin time, partial thromboplastin time, divided by the gamma counts per minute platelet count, or hematocrit over the course of the experiments. Autoradiographs demper bacterium, onstrated a generally even distribution of labeled bacteria along the graft surfaces and [bat]* corroborated the quantitative differences in cpm/0.5 cc blood (initial sample) graft activities (Fig. 2). However, there was a + cpm/OS cc blood (final sample) marked increase in adherence at the suture = lines (Figure 2B). Scanning electron microsmCi “‘In/bat X cpm (std “‘In)/mCi’

ROSENMAN, PEARCE, AND KEMPCZINSKI:

651

GRAFTS

TABLE 1 BACTERIAL ADHERENCE TO EACH GRAFT TYPE

Grafttype

N

Dacron Umbilical vein FTFE + suture line PTFE

9 9 9 9

Meanbacteti/cm2’ 9.63 x 1.04 x 3.11 x 2.15 x

lo5 lo5 lo4 lo4

SD 3.69 x 3.96 x 1.01 x 1.49 x

SE lo5 lo4 104 lo4

1.23 x 1.32 x 3.37 x 4.97 x

lo5 l@ 103 10)

LIValues normalized for standard bacteremic concentration of 4.7 X lo6 bacteria/cc.

copy revealed marked differences in the quantity of surface thrombus formation on the various grafts, with Dacron appearing most thrombogenic, followed by HUV, then PTFE. The perianastomotic region of the sutured PTFE grafts had more adherent thrombus than any areas of the unsutured PTFE (Fig. 3). DISCUSSION

The prevention of prosthetic graft infection is a basic tenet of vascular surgery. Although placement of a prosthetic arterial conduit in the face of potential sepsisis contraindicated, it may be occasionally unavoidable [6, 16, 201. In these cases, it is wise to choose the graft which is least susceptible to adherence of bacteria [7]. Of the three prosthetic materials most commonly employed, PTFE appears to have the lowest propensity for bacterial adherence. Umbilical vein harbored 5 times more bacteria, and Dacron harbored 50 times more. TABLE 2 STATISTICAL SIGNIFICANCE OF THE VARIOUS GRAFT vs GRAFT COMPARISONS

Graftscompared Dacron vs PTFE

Dacron vs sutured PTFE Dacron vs HUV HUV vs PTFE HUV vs sutured PTFE Sutured PTFE vs PTFE * Significant at the 0.05 level.

Tukey’s test

Significant* Significant* Significant* NS NS NS

Consequently, although a PTFE graft may tolerate a moderate bacteremic challenge without developing clinical infection, the markedly increased (50X) adherence of bacteria to Dacron suggestan enhanced suscep tibility to bacteremia and would appear to mandate against its use in high-risk clinical situations. The addition of a suture line appeared to increase significantly bacterial adherence to PTFE grafts. Although all the anastomoses were hemostatic, small quantities of blood did leak through needle holes. The grafts were positioned to avoid significant contamination of the graft’s external surface. However, the perianastamotic region of the grafts (l-2 mm) were blood stained, and certainly the needle holes became plugged with contaminated blood. We feel that this partially explains the increased bacterial adherence observed in sutured grafts. The specific role of various sutures in bacterial adherence may not be discounted [2, 141. One important difference between the Dacron graft and the others is its hydrophillic porosity. Even with careful preclotting, the graft behaves somewhat as a sponge, acutely absorbing infected blood into its interstices, thus providing a greater surface area for bacterial adherence. This tendency probably decreaseswith chronic implantation as the Dacron flow surface becomes less blood permeable. There are other differences in the biochemical and micromorphologic features of the grafts, such as surface charge and relative hydrophobic qualities, which may explain the differences in their bacterial adherence.

652

JOURNAL OF SURGICAL RESEARCH: VOL. 38, NO. 6, JUNE 1985

FIG. 2. (A) Autoradiograph of Dacron graft. Note the even distribution of activity over the flow surface, with slight increases at the ridges of the crimped material. (B) F’TFE graft with a prolene suture line is shown below its autoradiograph. The most significant sign of bacterial adherence is at the suture line.

However, Sugar-manhas reported no significant difference between the staphylococcal adherence to PTFE or woven Dacron when the materials are simply incubated in suspensions of bacteria in buffered saline [24]. This suggests that the interactions between the perfused blood and the various grafts provide different milieus for bacterial adherence. Microscopic examination of the graft surface after perfusion under the experimental conditions of this study revealed striking differences in the quantity of fibrin present (Fig. 3). These differences in thrombogenicity between Dacron, PTFE, and HUVG have been previously described [ 11, 131. Earlier work in our laboratory has demonstrated a specific adherence of intravenously administered staphylococci to acute and chronic thrombus covering areas of the luminal surfaces of acute and chronically implanted grafts (Fig. 4) [21]. This suggeststhat the different propensity for thrombus formation upon each type of graft surface is responsible for the marked differences in their bacterial adherence.

An in vitro system allows selection of limited variables for study. In vim, however, complex factors may affect bacteremic adherence. Different species of bacteria may have different mechanisms of adherence [23]. Staphylococci have been shown to produce mucoid substances,and teichoic acid, which facilitate adherenceto implantable prostheses, as well as to cells [3, 41. Coagulase-positive strains also produce a clumping factor that binds to circulating fibrinogen [ 11. Enterobacteriaceaeproduce Type I pili that promote cellular adherence [23]. Streptococci can produce dextran-like glucans from sucrosewhich may promote their adherence to teeth and contribute to decay [9]. Certainly data regarding staphylococcal adherencein this study cannot be extrapolated to other organisms. Chronically implanted grafts of various materials have different propensities for the development of neointima, or endothelialized flow surfaces after autologous seeding [ 18, 211, both of which are important protections against bacterial adherence and bacteremic infectability [ 17, 18, 211. Thus, data from

ROSENMAN, PEARCE, AND RBMPCZINSKI:

GRAFTS

653

FIG. 3. Scanning electron micrographs of the flow surface OE (A) Knitted Dacron; (B) HUVG, (C) PTFE; (D) PTPE (anastomosis with prolene suture line), after 30 min of pulsatile perfusion without bacteremia. Note the marked quantity of fibrin matrix and entrapped corpuscles upon the Dacron as compared to the other grafts. Also note the increased perianastomotic thrombus on the PTPE graft with the prolene suture (D).

acutely implanted grafts give limited insight grafts reported the adherence of only lOOinto the behavior of chronic grafts. 1000 bacteria/graft after intravenous injecMost studies of vascular graft infection in tions of 108-lo9 bacteria [lo]. Scanning elecanimal or in vitro models have used quali- tron micrographs of small areas of these tative or quantitative cultures of the graft as grafts, however, show hundreds of bacteria an endpoint [5, 10, 17, 18, 221. Unfortuclearly demonstrating the inaccuracy of nately, the graft material and blood products quantitative cultures of pulverized grafts [lo]. prevent adequate dispersal of the bacteria Recent work in our laboratory using injecinto an even homogenous suspension. At- tions of radiolabeled bacteria in dogs have tempts at quantitative culture in these cir- corroborated that 0. 1- 1% of intravenously cumstances may be fraught with error, gen- injected bacteria will adhere to acutely imerally tending to underestimate the true planted grafts [21]. Radioactive labeling of number of bacteria present. One recent in the bacteria greatly diminishes the inaccuravitro study of bacterial adherence on vascular cies of quantitative culture. Iridium- 111-0xine

654

JOURNAL OF SURGICAL RESEARCH: VOL. 38, NO. 6, JUNE 1985

RG. 4. Scanning electron micrograph of a FTFE graft, which had been implanted in a dog’s femoral artery, and immediately thereafter exposed to an intravenous challenge of 3 X lo* Stuphylococcus aureus. Note the adherent fibrin with entrapped corpuscles and cocci.

is an excellent label for this purpose [8]. It is relatively inexpensive, binds tightly to bacteria with very little leak out over time, and emits y irradiation which is more easily counted and localized (e.g., gamma camera or autoradiography) than are #Iparticles [ 151.Its half life is short (2.8 days) making it a relatively safe compound for laboratory use, and the labeling technique is quick and simple. This method has allowed for highly accurate and reproducible assessmentsof the true number of bacteria per graft. This information may be employed to determine size of the bacteria innoculum needed to cause clinical infection for a given graft [5]. The pulsatile perfusion system used in this study closely mimics the hemodynamics of

a single, isolated vascular bed. As such, it is well suited for tightly controlled in vitro or ex vivo studies of vascular conduits. The stability of whole heparinized blood in the system over several hours appears excellent. Additional studies involving chronically implanted grafts, endothelial seededgrafts, and native vessels are underway. In addition to bacterial perfusion and adhesion, the study of platelet adherence, endothelial products, and the biomechanics of anastamoses are also appealing applications for this model. CONCLUSION

” ‘Indium-labeled bacteria and a pulsatile perfusion system have been used to study the

ROSENMAN, PEARCE, AND KEMPCZINSKI:

GRAFTS

655

8. Friedrich, E. A., Lehman, V., Suss,R., et al. Homing adherence of bacteria to various prosthetic kinetics of Indium-1 1l-labeled bacteremia: Detection vascular grafts following bacteremia. Dacron of organ specificities as revealed by scintography. grafts were found to harbor lo-fold more 9. Gibbons, R. J., and van Home, J. Bacterial adherence bacteria than did umbilical vein, and 50-fold in oral microbial ecology. Annu. Rev. Microbial. 29: 19, 1975. more than PTFE. The addition of a suture line in a PTFE graft slightly but significantly 10. Goeau-Brissoniere, O., Pechere, J. C., Guidoin, R., et al. Experimental colonization of vascular grafts increased bacterial adherence. Autoradiograwith Staphylococcus aureus. Cunad. J. Surg. 26: phy documented even distribution of bacteria 540, 1983. over the luminal surface with preferential 11. Goldman, M., Gunson, B., Hawker, R. J., et al. adherence to suture lines. The reduced bacHuman umbilical vein and polytetratluorethylene arterial grafts compared in an artificial circulation. terial adherence to PTFE may afford protecBrit. J. Surg. 70: 4, 1983. tion against the establishment of clinical graft 12. Goodwin, D. A., Doherty, D. W., and McDougall, infection. The methods used for this study I. Clinical use of Indium- 11l-labeled white cells: An gave accurate and reproducible results, and analysis of 312 cases. In M. L. Thakur (Ed.), Proceedings of the Yale Symposium on Radiolabeled may be well suited for further studies of Blood Cellular Elements, New York: Trivium Pub]. bacterial or platelet adherence to grafts, as Co., 1982. Pp. 131-144. well as the biomechanics of vascular conduits.

ACKNOWLEDGMENTS The technical assistance of Judy Falk, Steve Lukes, and Gail Mayheld is gratefully acknowledged, and the assistanceof Michael Miller in the statistical analysis of the data is much appreciated. This work was supported in part by a grant from W. L. Gore and Associates,Inc., Flagstaff, Ariz. 86002, and DRR Clinfo Grant RR68-2 1.

REFERENCES I. Abramson, C. Staphylococcal enzymes. In J. D. Cohen (Ed.), The Stuphyiococci, New York: Wiley, 1972. Pp. 187-248. 2. Alexander, J. W., Kaplan, J. Z., and Altemeier, W. A. Role of suture materials in the development of wound infection. Ann. Surg. 165: 192, 1967. 3. Aly, R., Shinefield, H. R., Litz, C. et al. Role of teichoic acid in the binding of Staphylococcusaureus to nasal epithelial cells. J. Infect. Dis. 141: 463, 1980. 4. Bayston, R., and Penny, S. R. Excessive production of mucoid substance in Staphylococcus SIIA; A possible factor in colonization of Holter shunts. Dev. Med. Child. Neural. 27(Suppl. 14): 25, 1972. 5. Bennion, R. S., Williams, R. A., and Wilson, S. E. Comparison of infectibihty of vascular prosthetic materials by quantitation of median infective dose. Surgery 95: 22, 1984. 6. Bunt, T. J. Synthetic vascular graft infections I. Graft infections. Surgery 93: 733, 1983. 7. Daughtery, S. H., and Simmons, R. L. Infections in bionic man: The pathophysiology of.infections in prosthetic devices-Part I. Curr. Probl. Surg. 19: 221, 1982.

13. Hamlin, G. W., Rajah, S. M., Crow, M. J., et al. Evaluation of the thrombogenic potential of three types of arterial graft studied in an artificial circulation. Brit. J. Surg. 65: 272, 1978. 14. Katz, S., Izhar, M., and Mirelman, D. Bacterial adherence to surgical sutures. Ann. Surg. 194: 35, 1981. 1.5. Kishore, R. Radiolabeled microorganisms: Comparison of different radioisotope labels. Rev. Int. Dis. 3: 1179, 1981. 16. Liekweg, W. G., and Greenfield, L. J. Vascular prosthetic infections: Collected experience and results of treatment. Surgery 81: 335, 1977. 17. Malone, J. M., Moore, W. S., Campagna, G., et al. Bacteremia infectability of vascular grafts: The influence of pseudointernal integrity and duration of graft function. Surgery 78: 211, 1975. 18. Moore, W. S., Malone, J. M., and Keown, K. Prosthetic arterial graft material: Influence on neointimal healing and bacteremic infectibility. Arch. Surg. 115: 1379, 1980.

19. Pearce, W. H., Rosenman, J. E., Bjomson, S., et al. Whole graft autoradiography. Submitted for publication. 20. Rosenman, J. E., and Kempczinski, R. F. Infected prosthetic arterial grafts. Probl. Gen. Surg. 1: 647, 1984. 21. Rosenman, J. E., Kempczinski, R. F., Berlatzky, Y., et al. Bacterial adherenceto endothelial seededPTFE grafts. Surgery, in press. 22. Roon, A. J., Malone, J. M., Moore, W. S., et al. Bacteremic infectability: A function of vascular graft material and design. J. Surg. Res. 22: 489, 1977. 23. Sugarman, B. Attachment of bacteria to mamelon surfaces.Infection 8: 132, 1980. 24. Sugarman, B. In vitro adherence of bacteria to prosthetic vascutar grafts. Infection 10: 9, 1982.