Use of an antibiotic-bonded graft for in situ reconstruction after prosthetic graft infections Michael D. Colburn, M D , Wesley S. Moore, M D , Milos Chvapil, M D , PhD, H u g h A. Gelabert, M D , and W'flliam J. Quifiones-Baldrich, MD, Los Angeles, Calif. We have developed an infection resistant vascular prosthesis by bonding rifampin to Dacron grafts with the use of a collagen matrix release system. The purpose of this study was to determine the efficacy of this antibiotic-bonded graft in resisting infection after an in situ reconstruction of a previously infected prosthetic bypass. Eighty-three adult mongrel dogs underwent implantation of a 3 cm untreated Dacron graft into the infrarenal aorta. This initial graft was deliberately infected, at the time of operation, with 102 organisms of Staphylococcusaureus by direct inoculation. One week later, the dogs were reexplored, the retropeHtoneum debrided, and the animals randomiTed to undergo an end-to-end in situ graft replacement with either one of two types of prosthetic grafts: group I (coUagen, n = 36) received control collagen-impregnated knitted Dacron grafts; group II (rifampin, n = 47) received experimental collagen-rifampin-bonded Dacron grafts. Each group of animals was then subdivided to receive one of four treatment protocols: (a) no antibiotic therapy, (b) cephalosporin peritoneal irrigation solution (cefazolin 500 mg/1000 ml) during operation and two doses of cephalosporin (cefazolin, 500 mg intramuscularly) postoperatively, (c) treatment as in protocol group b plus 1 week ofcephalosporin (cefazolin, 500 mg intramuscularly, twice daily), and (d) treatment as in protocol group b plus 2 weeks of cephalosporin (cefazolin, 500 mg intramuscularly, twice daily). All grafts were sterilely removed between 3 and 4 weeks after implantation. There were no anastomotic disruptions and all grafts were patent at the time of removal. Cultures were obtained from the grafts and peri-graft tissues separately. Analysis included determination of the culture positivity of tissue and graft samples combined and of the graft segments alone irrespective of the results of the surrounding tissue cultures. Results were expressed as the percentage of animals that had culture positive results at sacrifice. For group I, collagen graft animals, the rate ofinfeetion in graft segments alone was 100%, 87.5%, 100%, and 80% when the animals received supplemental antibiotic therapy according to groups a through d as described, respectively. For group II, rifampin graft animals, the corresponding values were 50%, 50%, 20%, and 20%. When overall infection rates were evaluated, samples from dogs in group I were culture positive in 100%, 87.5%, 100%, and 80% of the animals in each of the supplemental antibiotic therapy groups, respectively. In group II animals, the corresponding positivity rates for overall infection were 62.5%, 62.5%, 60%, and 20%. In all four treatment groups the reduction of positive graft cultures after in situ replacement of a previously infected aortic prosthesis with a collagen-rifampin-bonded graft was statistically significant. When overall infection rates were evaluated, this reduction was statistically significant only in the subset of animals treated with 2 weeks of supplemental antibiotics. We conclude that the collagenrifampin-bonded graft reduces the incidence of graft colonization after in situ replacement of an infected graft. A course of supplemental antibiotics is required to sterilize the surrounding peri-graft tissues. (J VAsc SURG 1992;16:651-60.)
From the Department of Surgery, Sectionof Vascular Surgery, UCLA Center for the Health Sciences,Los Angeles. Supported in part by funding from a Veterans Administration Merit Reviewgrant and the JOASH Foundation. Presentedat the SeventhAnnualMeetingof the WesternVascular Society,Mani, Hawaii,Jan. 11-15, 1992. Reprint requests: WesleyS. Moore, MD, Professor of Surgery, Chief, Section of Vascular Surgery, UCLA Center for the Health Sciences,Los Angeles,CA 90024-6904. 2416139117
The problem of prosthetic graft infections persists despite the development and use of powerful broad spectrum antibiotics. The combination of antibiotic prophylaxis and optimal surgical technique has been able to reduce graft infections to the range of 1.0% to 5%. 1 This incidence has remained stable and recently no further reduction in graft infections has been achieved. Furthermore, the complications of 651
652 Colburn et al.
graft infections have remain unchanged: sepsis, anastomotic disruption, graft thrombosis, and limb loss still haunt us. Once a prosthetic graft becomes infected, it will almost always require excision and extraanatomic reconstruction as treatment. Unfortunately, the results of these reconstructions are characterized by low patency rates and a significant incidence of infections. Our efforts have focused on the development of a bacterial resistant vascular prosthesis by the bonding of an antibiotic to Dacron grafts with the use of a collagen matrix release system. 2 Previously we have demonstrated in vitro the antibacterial efficacy of a collagen-rifampin-bonded release system, s Also, in vivo experiments have demonstrated the ability of this graft to resist infection, after an intravenous bacteremic challenge, for periods lasting up to 7 days. 4 The purpose of this study was to test the efficacy of the collagen-rifampin-bonded graft in inhibiting infection after the in situ replacement of an infected vascular prosthesis. MATERIALS Graft material. Three different prosthetic Dacron grafts (Meadox Medicals, Inc., Oakland, N. J.) were used in this study. Nonantibiotic bonded grafts used were an untreated double-velour knitted Dacron prosthesis and a collagen-coated double-velour knitted Dacron Hemashield graft. The experimental antibiotic bonded graft was prepared from a knitted double-velour Dacron prosthesis. The graft was impregnated with a collagen-rifampin mixture and minimally cross-linked with formalin. Rifampin is one of a limited number of hydrophobic antibiotics and has been shown to provide prolonged antimicrobial activity after bonding to a prosthetic graft, s Also, rifampin has a broad antimicrobial specmma and, in addition to its well-known activity against mycobacteria, is active against several gram-negative and gram-positive organisms including Staphylococcus aureus. Details regarding the preparation of this antibiotic-bonded vascular graft have been previously described, s Bacterial solution. S. aureus was obtained as a clinical isolate from the UCLA School of Medicine, Department of Microbiology. The strain was coagulase positive, catalase and 13-1actamasenegative, and was sensitive to both oxacillin (minimum inhibitory concentration 2.0 Ixg/ml) and rifampin (minimum inhibitory concentration 3.0 I±g/ml). This clinical isolate was stored on 4 m m glass beads at -70* C. One day before use, a single bead was sterilely removed, placed in 5 ml of brain heart infusion
Journal of VASCULAR SURGERY
(BHI) broth, and incubated at 37 ° C for 24 hours. The following day, an aliquot was removed and the organisms counted. The quantified organisms were diluted with normal saline solution to produce a standardized bacterial solution containing 2 x 102 organisms/ml. Bacterial plates. Agar plates that consisted of tryptic soy containing 5% sheep blood were obtained sterile (Clinical Standards Laboratories, Inc., Rancho Dominguez, Calif.) and used for all bacteriologic studies. METHODS Surgical protocol. Eighty-three adult mongrel dogs, weighing 20 kg each, were entered into the study. To produce a graft infection, each animal underwent an initial laparotomy. During this procedure a 6 m m untreated Dacron prosthesis, 3.0 cm in length, was inserted into the infrarenal aorta in a double end-to-end fashion with 6-0 polypropylene suture. This graft was subsequently infected by direct application of 0.5 ml of the saline suspension containing S. aureus at a concentration of 2 x 102 organisms/ml. Once inoculated, the graft was covered by reapproximating the retroperitoneum, and the abdomen was closed in layers with standard surgical technique. One week after the initial graft had been placed and contaminated, each animal underwent reoperation. The infected graft was removed, cultured, and the retroperitoneum debrided. The animals were then randomly assigned to one of two experimental groups: group I (collagen, n = 36) received control collagen-impregnated knitted Dacron grafts; group II (rifampin, n = 47) received experimental collagen-rifampin-bonded Dacron prostheses. In each case the replacement graft was implanted in situ, in an end-to-end fashion, with 6-0 polypropylene suture. These two groups were then subdivided and treated with one of four different supplemental antibiotic protocols: (a) no adjunct antibiotic therapy (No Abx) (b) cephalosporin peritoneal irrigation (cefazolin 500 mg/1000 ml) during operation and two doses of cephalosporin (cefazolin, 500 mg intramuscularly) postoperatively (Peri-op), (c) Peri-op plus 1 week ofcephalosporin (cefazolin, 500 mg intramuscularly, twice daily) (One wk), and (d) Peri-op plus 2 weeks ofcephalosporin (cefazolin, 500 mg intramuscularly, twice daily) (Two wks). The abdomen was again dosed in layers. Three to four weeks after the second operation all animals were again anesthetized to recover specimens for bacteriologic study. Cultures were obtained from
Volume 16 Number 4 October 1992
Antibiotic-bonded graft after graft infictions 653
I 100
100. 90. 80.
ll
Collagen Graft
[]
Rifampin Graft
100
100
.oC 70, 6o
~- 50.
84o. N 80. 20. 10. O. No Abx
Pefi-op One wk Two wk~ Supplemental Antibiotic Treatment Protocol
Fig. 1. Chart shows effect of supplemental antibiotics on graft incorporation. Results are expressed as percentage of grafts that were well incorporated at sacrifice. Data are separated into the four supplemental antibiotic treatment groups: no adjunct antibiotics (No Abx), perioperative antibiotics only (P~/-W), perioperative plus 1 week of postoperative antibiotics (One ruk), and perioperative plus 2 weeks of postoperative antibiotics (Two wks).
Tab!¢ I. Intraoperative evaluation of graft incorporation and gross purulence No Abx
Collagen Rifampin
Peri-op
1 wk
2 wk
Incorp
Purulence
Incorp
Purulence
Incorp
Purulence
Incorp
Purulence
n
(%)
(~)
n
(%)
(~)
n
(%)
(~)
n
(%)
(%)
5 16
0.0 62.5
40.0 12.5
16 16
68.7 100.0
18.7 0.0
5 5
40.0 100.0
20.0 0.0
10 10
60.0 100.0
10.0 0.0
NoAbx, No adjunct antibiotics; Per/-0p, perioperative antibiotics only; 1 wk, perioperative plus 1 week of postoperative antibiotics; 2 wk, perioperative plus 2 weeks of postoperative antibiotics; Incorp, graft incorporation.
the grafts and peri-grafr tissues separately. In addition to the bacteriologic data, information regarding graft patency, graft incorporation, and the presence of any gross purulence was recorded. Surgeons making these determinations were blinded with regards to the animal's cohort. The animals were killed at the end of this operation. The data from this experiment were analyzed with chi-square analysis and, when appropriate, the onetailed Fishers exact test for small groups. Statistical significance was assigned whenp values were less than 0.05. All animal care complied with the "Principles of Laboratory Animal Care" and the "Guide for the Care and Use of Laboratory Animals" (NIH Publication No. 80-23, revised 1985).
Bacteriologic studies. A total of five specimens were collected at each operation (both for the initially infected graft and the in situ replacement graft). These included three segments of graft (proximal, midportion, and distal), one swab specimen from the exterior of the graft (abscess cavity), and one specimen from the peri-graft tissue. The swab specimens were directly applied onto the surface of sheep-blood agar plates. Both the graft segments and the tissue specimens were initially placed in brain heart infusion (BHI) broth and incubated at 37 ° C for 24 hours. The BHI broth was then plated onto sheep-blood agar plates and allowed to incubate for periods of up to 2 weeks. If colonies ofS. aureus grew in these plates, then the specimen was considered positive. If S. aureus failed to grow over the 2-week
Journal of VASCULAR SURGERY
654 Coltmrn et al.
l
100 oC 0
1
Collagen Graft
[]
Rifampin Graft
80
0.
40 J=
10
0
,_ll0
Ped-op One wk Supplemental AntibioticTreatment Protocol
No Abx
Two wks
Fig. 2. Chart shows effect of supplemental antibiotics on occurrence of gross purulence. Results are expressed as percentage of grafts that were grossly purulent at sacrifice. Groups as indicated in Fig. 1. Table 1I. Bacteriologic evaluation of graft infection No Abx
Collagen Rifampin p value
5 16
Peri-o#
TOG Cx
Graft Cx
(%)
(%)
100 62.5 NS
100 50 < 0.05
n
16 16
I wk
TOG Cx
Graft Cx
(%)
(%)
87.5 62.5 NS
87.5 50.0 < 0.05
n
5 5
2 wk
T&G Cx
Graft Cx
(%)
(%)
100 60 NS
100 20 < 0.05
n
10 10
T6aG Cx
Graft Cx
(%)
(%)
80 20 < 0.02
80 20 < 0.02
NoAbx, No adjunct antibiotics; Per/-op, perioperative antibiotics only; 1 wk, perioperative plus 1 week of postoperative antibiotics; 2 wk, perioperative plus 2 weeks of postoperative antibiotics; T&G Cx, culture Positivityof tissue and graft samples combined; Graft Cx, culture positivity of graft segments alone, irrespective of the results of the surrounding tissue cultures; NS, not significant.
period, they were considered negative. Analysis included determination of the culture positivity of both the tissue and graft samples combined (T&G Cx) and of the graft segments alone (Graft Cx), irrespective of the results of the surrounding perigraft tissue cultures. Results are expressed as the percentage of animals that were culture positive at the time they were killed. RESULTS All of the initially implanted prosthetic grafts and their Peri-graft tissues were infected with S. aurcus at the time of the second operation. All of these grafts were patent, but each showed poor incorporation and gross evidence of purulence. There were no anastomotic disruptions and no animals died. Furthermore, all animals survived the secondary in situ graft replacement operation and were accounted for
at the conclusion of the experiment. At sacrifice, all 83 grafts were patent and there were no anastomotic disruptions. The percentage of grafts with good incorporation or evidence of gross purulence is shown in Table I. Good incorporation was found in none of the five (0%) collagen (group I) grafts that received no adjunct antibiotics (No Abx), and gross purulence was present in two of five (40%). The percent incorporation increased to 68.7%, 40%, and 60% and the presence of purulence decreased to 18.7%, 20%, and 10% when the animals received either Peri-op, One wk, or Two wks of supplemental antibiotic therapy, respectively. The rifampin grafts (group II) without supplemental antibiotics (No Abx) were incorporated in 10 of 16 (62.5%), and gross ptmalence was detectable in two of 16 (12.5%). Rifampin grafts were all well incorporated (100%,
Volume 16 Number 4 ~ October 1992
Antibiotic-bonded graj~ after ffraft infctiom 655
If~,'t
100
•
Collagen Graft
[]
Rifampin Graft
100
o o == o
0 A
~+
No Abx
Peri-op One wk Supplemental Antibiotic Treatment Protocol
Two wks
Fig. 3. Chart shows effect of supplemental antibiotics on subsequent culture positive rate of graft samples alone. Results are expressed as percentage of grafts that remained culture positive at sacrifice. Groups as indicated in Fig. 1. 100%, and 100%) and showed no evidence of gross purulence (0%, 0%, and 0%) when the animals received Peri-op, One wk, or Two wks of therapy (Figs. 1 and 2). The results of the bacteriologic cultures are shown in Table ]I. When the graft segments were analyzed alone (Graft Cx), the collagen-rifampin-bonded graft reduced the incidence of culture positivity in all four treatment groups. For group I, Collagen graft animals, this rate was 100%, 87.5%, 100%, and 80% when the animals received No Abx, Peri-op, One wk, and Two wks of supplemental antibiotic therapy, respectively. For group II, Rifampin graft animals, the corresponding values were 50%, 50%, 20%, and 20%. This reduction in culture positivity, seen in animals in which the aorta was reconstructed with a rifampin-impregnated graft, was statistically significant in all four treatment cohorts. When overall infection rates (T&G Cx) were evaluated, the collagen-rifampin-bonded graft again reduced the incidence of culture positivity in all four treatment groups. T h e rate in group I animals was 100%, 87.5%, 100%, and 80% for the No Abx, Peri-op, One wk, and Two wks of supplemental antibiotic therapy groups, respectively. In group II animals, the corresponding T&G Cx positivity rates were 62.5%, 62.5%, 60%, and 20%. Unlike that found in the analysis of graft cultures alone (Graft Cx), the reduction in T&G Cx culture positivity observed in the rifampin graft cohorts was only statistically
significant in the subset of animals treated with two weeks (Two wks) of supplemental antibiotics (Figs. 3 and 4). DISCUSSION The current management of prosthetic graft infectious is based on several precepts of proven value. The prosthetic graft must be removed. Continuity of arterial blood supply is restored by means of an extraanatomic bypass that circumvents the area of infection. In addition, patients are treated with intravenous antibiotics for prolonged periods of time. Some patients are placed on lifelong regimens of suppressive oral antibiotics. Despite this systematic approach, several problems remain associated with this management. The best reported primary patency rates for axillobifemoral bypass grafts are between 62% and 72% at 5 years, s,e These studies, however, only include patients in whom the indication for bypass was either aneurysmal or occlusive disease. When extraanatomic graft patency is evaluated in patients in whom the primary indication for operation was an aortic graft infection, the primary graft patency rate at 3 years is 43%, which can be improved to 65% after secondary procedures. 7 Subsequent limb loss associated with a failed axiUobifemoral graft is as high as 34%. 7 Furthermore, the recurrence of infection in an extraanatomic bypass graft is significant at about 20%. 7 Lastly, the incidence of fatalities during the
656
Journal of VASCULAR SURGERY
Colburn et al.
[ ] Collagen Graft 100
o
*6
100
[]
Rifampin Graft
90,
~ 70. 50. ¢9 40, 30, o
~" 10, & O, No Abx
Ped-op One wk Supplemental AntibioticTreatment Protocol
Two wks
Fig. 4. Chart shows effect of supplemental antibiotics on subsequent culture positive rate of graft and tissue samples combined. Results are expressed as percentage of grafts and/or tissues that remained culture positive at sacrifice. Groups as indicated in Fig. 1. course of these reconstructive efforts has been reported to be as high as 25%. 7.9 In one report, 82% of patients treated for an aortic graft infection were dead at 5 years. 1° In situ reconstruction of primary vascular infections along with surgical debridement and prolonged antibiotic administration has yielded some success. In one study by Brown et al.,n the survival rate after in situ reconstruction in patients with aortic mycotic aneurysms was 63% with a reinfection rate of 19% and an incidence of rupture of 11%. In situ reconstruction after prosthetic graft infections has been reported sporadically, however, the results have been variable. In a review in 1983, Bunt 12 reported a mortality rate of 56% after in situ reconstructions in patients with graft-enteric fistulas. More recently, Jacobs et alj3 reported their experience with in situ graft replacement in 18 patients with infected aortic grafts. In this study, 12 of the 18 patients were classified as having low-grade graft infections with negative blood and peri-graft cultures. All 12 of these patients were alive at a mean follow-up period of 8 years. The remaining six patients had severe graft infections with positive blood and peri-graft cultures. In this group, five patients died within 1 month of operation and the remaining patient required conversion to an extraanatornic bypass because of recurrent infection. The authors concluded that there may be a role for in situ reconstruction in patients who have low-grade aortic graft infections. A similar conclusion was reached recently in a report by
Robinson and Johansen.14 In this paper, the authors reviewed their experience and that of the literature regarding in situ replacement of infected aortic grafts. Their findings suggested that the operative mortality rate in these patients is as low as 19% and that this procedure may be appropriate in a select group of patients with low-grade graft infections. Lastly, Bandyk et aLas recently reported their experience with in situ graft replacement in 15 patients with low-grade graft infections secondary to Staphylococcus ep/derm/d/s. In this series, no deaths, graft thromboses, or recurrent infections occurred after a mean follow-up period of 21 months. The authors also suggested that a reevaluation of the use of in situ graft replacement for infected vascular prostheses seems warranted. The development of a graft that is capable of resisting bacterial contamination would have several advantageous uses. It would be expected to reduce the current incidence of graft infections. It may be able to reduce the need for perioperative antibiotic coverage, and it may improve the results after an in situ reconstruction in an infected arterial bed. The characteristics of such a graft should include a broad-spectrum antibacterial effect, with particular efficacy against the most common pathogens. The duration of the antibacterial effect should be as prolonged as possible and have minimal risk of toxicity. Finally, the antibacterial agent should not degrade the graft or make its technical use less desirable.
Volume 16 Number 4 October 1992
Past experiments in the development of a bacterial resistant prosthetic graft have used several antibiotic agents combined with many modes of application. These efforts have included passive soaking of grafts with antimicrobial d r u g s , 16 ionic bonding of antibiotics with various bonding agents, 17"21 and topical application of antibiotic sealants. 22'23 Unfortunately, differences in antibiotic dosages, tested organisms, and methods of evaluation make comparisons of these techniques difficult. Several properties of rifampin make it a useful agent for this purpose. It has a rather broad spectrum of antimicrobial activity and is particularly active against gram-positive cocci. Further, it is relatively hydrophobic and thus will not dissolve into the bloodstream as rapidly as other antibiotics. This low aqueous solubility, combined with the binding of the agent by a collagen matrix, results in the gradual release of the antibiotic and therefore a prolonged local antibacterial action. In vitro studies have demonstrated activity of collagenrifampin bonded grafts in inhibiting S. aureus growth for up to 3 weeks. 3 In addition, in vivo experiments have shown antibacterial activity in this graft after an intravenous challenge for as long as 7 days after implantation. 4 The present experiment has demonstrated that the collagen-rifampin prosthesis is capable of inhibiting bacterial growth on the graft for periods of up to 3 weeks. The in situ replacement of an infected arterial prosthesis with this antibiotic-bonded graft, supplemented with systemic antibiotics, resulted in no gross evidence of infection after a 3-week period. Attempts to culture bacteria from the graft after this 3-week period were unsuccessful in 50% of cases when no adjunct antibiotic:therapy was administered. This was improved to an 80% culture negative rate when the graft was supplemented with 2 weeks of systemic antibiotic therapy. This represented a significant improvement over the collagen (group I) grafts. This is all the more remarkable given the implantation of the prosthetic grafts into grossly infected arterial beds with demonstrated abscesses. Clearly a 20% subsequent infection rate after the in situ replacement of a previously infected bypass graft remains prohibitively high. Therefore, use of this prosthesis in a comparable clinical setting is not warranted at this time. On the other hand, as mentioned previously, in situ graft replacement with unprotected prosthetic grafts in the setting of lowgrade graft infections is currently being performed with considerable success./31s It is therefore not unreasonable to assume that the results of this study, which tested the antibiotic-bonded graft in a highly
Antibiotic-bondedgraft after graft infections
657
challenging model, would have likely been improved had the grafts been used to replace prostheses containing only a low-grade infection. Although the infection rate of the rifampinimpregnated grafts was significantly lower than that of the collagen grafts in all four treatment groups when graft cultures were considered alone, when the results of the cultures on the peri-graft tissues were included, this difference was lost in the No Abx, Peri-op, and One wk treatment groups. Two weeks of supplemental antibiotic therapy, however, was able to sterilize the surrounding peri-graft tissues and significantly reduce the overall (T&G Cx) culture positive rate. The persistence of organisms in the peri-graft tissues suggests that the antimicrobial effect of the graft alone is not able to penetrate the adjacent tissues and eradicate the bacteria. The use of prolonged perioperative antibiotics (as would normally be done in most instances of graft infection) clearly improves the success of this technique. We conclude that the coUagen-rifampin-bonded Dacron graft is able to reduce the incidence of subsequent bacterial growth on a prosthesis after implantation into a grossly infected arterial bed. The addition of perioperative parenteral antibiotic supplementation, however, is required to sterilize the surrounding peri-graft tissues. REFERENCES
1. Goldstone J, Moore WS. Infection in vascular prostheses: clinical manifestations and surgical management. Am J Surg 1974;128:225-33. 2. Moore WS, Chvapil M, Sieffert G, Keown K. Development of an infection resistant vascular prosthesis. Arch Surg 1981; 116:1403-7. 3. Chervu A, Moore WS, Chvapil M, Henderson T. Efficacyand duration of antistaphylococcal activity comparing three antibiotics bonded to Dacron vascular grafts with a collagen release system. J VASCSUgG 1991;13:897-901. 4. Chervu A, Moore WS, Gelabert HA, Colburn MD, Chvapil M. Prevention of graft infection by use of prostheses bonded with a rifampin/collagen release system. J V~sc SURG 1991;14:521-5. 5. Rutherford RB, Patt A, Pearce WH. Extraanatomic bypass: a closer view. J V ~ c SURG 1987;6:437-46. 6. Hepp W, Jonge K, Pallua N. Late results following extraanatomic bypass procedures for chronic aortoiliac occlusive disease. J Cardiovasc Surg (Torino) 1988;29:181-5. 7. Quifiones-BaldrichWI, Hernandezll, MooreWS. Long-term results following surgical management of aortic graft infection. Arch Surg 1991;126:507-11. 8. Reilly LM, Stoney RJ, Goldstone J, Ehrenfeld WK. Improved management of aortic graft infection: the influence of operation sequence and staging. J VASCSURG1987; 5:412- 31. 9. Yeager RA, Moneta GL, Taylor LM, McConnell DB, Porter JM. Can prosthetic graft infection be avoided? If not, how do we treat it? Acta Chir Scand Suppl 1990;555:155-63.
658
Journal of VASCULAR SURGERY
Colburn et al.
10. O'Hara PJ, Hertzer NR, Beven EG, Krajewski LP. Surgical management of infected abdominal aortic grafts: review of a 25-year experience. J VASCSURG 1986;3:725-31. 11. Brown SL, Busuttil RW, Baker ID, Machleder HI, Moore WS, Barker WF. Bacteriologic and surgical determinants of survival in patients with mycotic aneurysms. J VASC SURG 1984;1:541-7. 12. Bunt TJ. Synthetic vascular graft infections, II: graft-enteric erosions and graft-enteric fistulas. Surgery 1983;94:1-9. 13. Jacobs MJ, Reul GJ, Gregoric I, Cooley DA. In-situ replacement and extra-anatomic bypass for the treatment of infected abdominal aortic grafts. Eur J Vasc Surg 1991;5:83-6. 14. Robinson JA, Johansen K. Aortic sepsis: is there a role for in situ graft reconstruction? J VASESURG 1991;13:677-84. 15. Bandyk DF, Bergamini TM, Kinney EV, Seabrook GR, Towne JB. In situ replacement of vascular prostheses infected by bacterial biofilms. J VASC SURG 1991;13:575-83. 16. Powell TW, Burnham SJ, Johnson G. A passive system using rifampin to create an infection-resistant vascular prosthesis. Surgery 1983;94:765-9. 17. Benvenisty A, Tannenbaum G, Ahlborn TN, et al. Control of prosthetic bacterial infection: evaluation of an easily incorporated, tightly bound, silver antibiotic PTFE graft. J Surg Res 1986;44:1-7.
18. Sobinsky KR, Flanigan DP. Antibiotic binding to polytettafluoroethylene via glucosaminoglycan-kerarin luminal coating. Surgery 1986;100:629-34. 19. Greco RS, Harvey RA, Smilow PC, Tesoriero JV, Prevention of vascular prosthetic infection by a benzalkonium-oxacillin bonded polytetrafluoroethylene graft. Surg Gynecol Obstet 1982;155:28-32. 20. Shu¢ WB, Worosilo SC, Donetz AP, Trooskin SZ, Harvey RA, Greco RS. Prevention of vascular prosthetic infection with an antibiotic-bonded Dacron graft. J VASC SURG 1988;8:600-5. 21. Kinney EV, Bandyk DF, Seabrook GA, Kelly HM, Town JB. Antibiotic-bonded PTFE vascular grafts: the effect of silver antibiotic on bioactivity following implantation. J Surg Res 1991;50:430-5. 22. Shenk JS, Ney AL, Tsukayama DT, Olson ME, Bubrick MP. Tobramycin-adhesive in preventing and treating PTFE vascular graft infections. J Surg Res 1989;47:487-92. 23. Ney AL, Kelly PH, Tsukayama DT, Bubrick MP. Fibrin glue-antibiotic suspension in the prevention of prosthetic graft infection. J Trauma 1990;30:1000-6. Submitted Feb. 13, 1992; accepted May 1, 1992.
DISCUSSION Dr. Kenneth E. McIntyre, Jr. (Tucson, Ariz.). This study is another in a long line of accomplishments by the UCLA Vascular Surgery group whose goal has been to develop an infection-resistant aortic prosthesis. The current study that was presented today offers a new twist, however. Dr. Gelabert and associates have tackled a clinical problem that has plagued all of us who do vascular surgery: Is there an alternative to extraanatomic arterial reconstruction after excision of an infected aortic prosthetic graft? In the past, we have accepted the mediocre long-term patency rates of axillobifemoral bypass grafts, which were placed to provide infrainguinal perfusion after excision of an infected aortic graft. In situ graft replacement has been a current "hot topic" of late as an alternative to these dismal results. Although several anecdotal reports abound, this is the first prospective trial done in canines to address this problem. The authors evaluated a knitted Dacron velour graft impregnated with collagen bound to an antibiotic, which was implanted in an infected surgical field. As in their previous study, which was reported before this society, rifampin was bonded to the collagen before being applied to the graft. Rifampin was chosen became of its low solubility in water and its gradual release from the collagen matrix, which allowed for continued bacteriocidal levels of antibiotic over time. Mongrel dogs underwent laparotomy with a 3 cm aortic tube graft replacement and direct contamination of the aortic prosthesis with S. aureus bacteria. One week later, another laparotomy was performed to remove the infected aortic grafts. The dogs were
then randomized into one of two groups on the basis of their subsequent treatment. Excision of the infected aortic graft with replacement by a knitted Dacron double-velour collagen-coated aortic prosthesis was performed in 36 animals, and 47 dogs underwent infected aortic graft excision with replacement by an identical graft except that the collagen was bonded to rifampin. Each group was then subdivided into one of four treatment protocols: (1) no antibiotic therapy, (2) topical cephalosporin antibiotic peritoneal irrigation plus two doses of parenteral cephalosporin (Peri-op), (3) Peri-op plus 1 week of parenteral cephalosporin, or (4) Peri-op plus 2 weeks of parenteral cephalosporin. All the grafts were removed with sterile technique between 3 and 4 weeks after implantation and cultures were obtained from the grafts and peri-graft tissues separately. There was a direct correlation between surgeon evaluation of graft incorporation at the time of graft excision and presence or absence of purulence. Generally, the more the graft was incorporated, the smaller the incidence of purulence. It is not surprising that the collagen-bonded rifampin grafts demonstrated a greater resistance to graft infection than the grafts lined with collagen alone. However, the effect of rifampin was only apparent in the graft itself. The peri-graft tissue surrounding the infected graft remained infected despite the presence of the rifampin-bonded collagen graft. This result is not surprising considering that rifanapin is slowly released from the collagen over time. Therefore statistical significance was only obtained between the two groups
Volume 16 Number 4 October 1992
when graft cultures alone were considered. However, when overall infection rates were evaluated in both tissue and graft cultures, statistical significance was only obtained in that subset of animals that were treated with 2 weeks of supplemental antibiotics. With this model, therefore, although the collagen-rifampin-bonded graft reduced the incidence of graft colonization after replacement of an infected graft, additional antibiotics were required to help sterilize surrounding peri-graft tissues. I have no problem with either the study design or methods, but I would like to pose several questions to the authors. Why did the rifampin-collagen grafts seem to incorporate better? Do they incorporate better than collagen-coated grafts without rifampin in the absence of infection? Does incorporation alone in this case influence infectability? Because of its low solubility, rifampin was dissolved in ethanol to facilitate its homogeneous bonding to and distribution within the collagen. Can you speculate as to whether residual ethanol may have influenced either graft incorporation or resistance to infection? Third, since the graft was not effective in treatment of peri-graft infection, do you think in situ graft replacement can ever be successful when there is severe contamination of peri-graft tissue? Fourth, were any cultures obtained of the aortic wall itself?. We have found in our studies of graft infections in humans that when the aortic wall itself is infected, eradication of infection by any means is unlikely. Finally, will this new graft be subjected to appropriate clinical trials to evaluate its efficacy or will it suddenly appear on shelves as the next expensive wonder device before an adequate evaluation has been completed? I enjoyed the manuscript and recommend it for your review. Dr. H u g h A. Gelabert. There are several points that I would like to emphasize. First, we consider this one of the most severe bacterial challenges: placing a graft into what one of our residents refers to as a "puddle of pus." This required an extensive debridement and antibiotic irrigation of the retroperitoneum. I think it is impressive that we were able to reduce the subsequent infection rate. The second point is that the graft itself has often been considered the anchor of the infection. We have been taught that what causes a graft infection to persist is the affinity between the organisms and the prosthetic graft material. Further, the biologic tissues surrounding this area were thought to be innocently subjected to infection and to play a passive role in the infectious process. In this experiment we have succeeded in making a graft that will clear itself of the bacterial infection. We discovered that the surrounding biologic tissues served to anchor the infection, and that is where prolonged treatment with antibiotics is helpful. I mention this because it gets back to several important points that Dr. Mclntyre brought up in his discussion. Specifically, the questions of the role of this graft in a clinically severely infected patient and the issue of involvement of the artery itself. We did not specifically remove segments of the arteries
Antibiotic-bondedgra~ ajqm"gruft infections 659 for microbiologic study. We did, however, incorporate the proximal and distal ends of the arteries along with the specimen as it was excised during the second operation. This specimen was subsequently divided into proximal, middle, and distal portions. These were, in turn, analyzed for bacterial contamination. There was no significant difference between any segment with regards to infection: they were all infected. Reconstruction with the experimental graft required resection of the aorta back to a point where the arterial wall consisted of grossly healthy, viable tissue. The resected perianastomotic portion of the aortas was inflamed and edematous. It would not have been adequate for anastomosis. As to the question of whether this graft is ready for implantation in humans: That remains to be established. We are certainly working in that direction. The question regarding the relationship of incorporation to infection and the effect of the collagen binding is provocative. Clearly, infection will inhibit incorporation, and a poorly incorporated graft is thought to be more likely to become infected. At times it is impossible to tell which comes first. The effect of the rifampin binding could be answered by comparing the rifampin-bound graft against a collagen graft in a noninfected model. Unfortunately, we have not done this. The question of the residual presence of ethanol in the experimental graft is unanswered. In the process of graft construction, the graft is heated in a formalin vapor to provide minimal collagen cross-linkage. It has been our presumption that the ethanol evaporates in the course of this process. We do not, however, have any specific, conclusive evidence to that effect. Finally, as to whether the graft will suddenly appear on the shelf next week, your speculation is about as good as mine. I doubt that this will happen, since this is a new means of applying and delivering the antibiotic. As far as I understand, the Food and Drug Administration will require a set of trials before it can appear on the shelf. Thank you very much for your insightful comments. Dr. Linda M. Reilly (San Francisco, Calif.). One interpretation of your data could be that the effect of intravenous antibiotics does not begin to manifest itself until at least 2 weeks of treatment have elapsed. Have you considered extending this experiment for a longer period of time, particularly to show the effect of the antibiotics in your control group? One other question: Had you considered using a non-collagen-impregnated graft as your true control? Although there is an increasing use of collagen impregnated grafts, I think that for most of us the more common graft is a non-collagen-bonded graft. Dr. Gelabert. There is no question about it. I think there are several variations in this experiment that would be helpful and that would be fun to do. Extending the period of time certainly would be of great interest to see how durable this effect is, first of all, and secondly to see whether
660
Journal of VASCULAR SURGERY
Colburn et al.
or not the outcome of the animals changes. Unfortunately, within the limits of time and money we have not done that yet, but that is something that we have been considering. In response to the second question as to the use of a non-collagen-bound graft, we actually did use noncollagen-bound grafts in previous experiments. As a matter of fact, when developing this model we went through a whole set of developmental experiments in which we were trying to determine the minimal infected bacterial concen-
trations. The first graft that we placed was a non-collagenbound graft. We found that exceedingly easy to infect. In addition, there's a tremendous historic,~l background that has demonstrated that the infections persist when these grafts are used in animals. So we didn't really think it was absolutely necessary to do that. Finally, to provide a good comparison with the rifampin-bound graft, which is bound in a collagen matrix, we thought it would be best to use the collagen-bound graft.
SOCIETY F O R V A S C U L A R S U R G E R Y LIFELINE
FOUNDATION
GRANT
AWARD
The Lifeline Foundation of the Society for Vascular Surgery invites grant applications for funding of meritorious research by young surgical investigators. The awards are intended for surgeons who have completed their formal surgical education in general surgery and who have completed or are in an advanced training program in vascular surgery. To be considered for selection a candidate: i. Should be certified by the American Board of Surgery or have completed the requirements for certification 2. Should submit an application within 3 years of completion of an approved residency training program 3. Must have either a faculty appointment in an approved medical school in the United States or Canada or have received an academic appointment within the guidelines of the applicant's institution Grant awards are not intended to supplement salary, which will remain the responsibility of the institution in which the awardee holds an appointment. The awardee is expected to devote a significant amount of time to the funded project. A progress report will be presented by the investigators during the annual meeting of the Society for Vascular Surgery. A grant awards committee will review competitive applications. It is anticipated that two grants will be awarded annually totaling $50,000 each to include indirect costs. Each award will be for 1 year with the option to extend for an additional year. Grant applications may be obtained from: The Lifeline Foundation Society for Vascular Surgery Thirteen E l l St. Manchester, MA 01944 (508) 526-8330
The deadline for receiving applications in the Foundation office is January 15, 1993. Funds will be awarded by July 1, 1993.
From the Department of Surgery, Sectionof Vascular Surgery, UCLA Center for the Health Sciences,Los Angeles. Supported in part by funding from a Veterans Administration Merit Reviewgrant and the JOASH Foundation. Presentedat the SeventhAnnualMeetingof the WesternVascular Society,Mani, Hawaii,Jan. 11-15, 1992. Reprint requests: WesleyS. Moore, MD, Professor of Surgery, Chief, Section of Vascular Surgery, UCLA Center for the Health Sciences,Los Angeles,CA 90024-6904. 2416139117
The problem of prosthetic graft infections persists despite the development and use of powerful broad spectrum antibiotics. The combination of antibiotic prophylaxis and optimal surgical technique has been able to reduce graft infections to the range of 1.0% to 5%. 1 This incidence has remained stable and recently no further reduction in graft infections has been achieved. Furthermore, the complications of 651
652 Colburn et al.
graft infections have remain unchanged: sepsis, anastomotic disruption, graft thrombosis, and limb loss still haunt us. Once a prosthetic graft becomes infected, it will almost always require excision and extraanatomic reconstruction as treatment. Unfortunately, the results of these reconstructions are characterized by low patency rates and a significant incidence of infections. Our efforts have focused on the development of a bacterial resistant vascular prosthesis by the bonding of an antibiotic to Dacron grafts with the use of a collagen matrix release system. 2 Previously we have demonstrated in vitro the antibacterial efficacy of a collagen-rifampin-bonded release system, s Also, in vivo experiments have demonstrated the ability of this graft to resist infection, after an intravenous bacteremic challenge, for periods lasting up to 7 days. 4 The purpose of this study was to test the efficacy of the collagen-rifampin-bonded graft in inhibiting infection after the in situ replacement of an infected vascular prosthesis. MATERIALS Graft material. Three different prosthetic Dacron grafts (Meadox Medicals, Inc., Oakland, N. J.) were used in this study. Nonantibiotic bonded grafts used were an untreated double-velour knitted Dacron prosthesis and a collagen-coated double-velour knitted Dacron Hemashield graft. The experimental antibiotic bonded graft was prepared from a knitted double-velour Dacron prosthesis. The graft was impregnated with a collagen-rifampin mixture and minimally cross-linked with formalin. Rifampin is one of a limited number of hydrophobic antibiotics and has been shown to provide prolonged antimicrobial activity after bonding to a prosthetic graft, s Also, rifampin has a broad antimicrobial specmma and, in addition to its well-known activity against mycobacteria, is active against several gram-negative and gram-positive organisms including Staphylococcus aureus. Details regarding the preparation of this antibiotic-bonded vascular graft have been previously described, s Bacterial solution. S. aureus was obtained as a clinical isolate from the UCLA School of Medicine, Department of Microbiology. The strain was coagulase positive, catalase and 13-1actamasenegative, and was sensitive to both oxacillin (minimum inhibitory concentration 2.0 Ixg/ml) and rifampin (minimum inhibitory concentration 3.0 I±g/ml). This clinical isolate was stored on 4 m m glass beads at -70* C. One day before use, a single bead was sterilely removed, placed in 5 ml of brain heart infusion
Journal of VASCULAR SURGERY
(BHI) broth, and incubated at 37 ° C for 24 hours. The following day, an aliquot was removed and the organisms counted. The quantified organisms were diluted with normal saline solution to produce a standardized bacterial solution containing 2 x 102 organisms/ml. Bacterial plates. Agar plates that consisted of tryptic soy containing 5% sheep blood were obtained sterile (Clinical Standards Laboratories, Inc., Rancho Dominguez, Calif.) and used for all bacteriologic studies. METHODS Surgical protocol. Eighty-three adult mongrel dogs, weighing 20 kg each, were entered into the study. To produce a graft infection, each animal underwent an initial laparotomy. During this procedure a 6 m m untreated Dacron prosthesis, 3.0 cm in length, was inserted into the infrarenal aorta in a double end-to-end fashion with 6-0 polypropylene suture. This graft was subsequently infected by direct application of 0.5 ml of the saline suspension containing S. aureus at a concentration of 2 x 102 organisms/ml. Once inoculated, the graft was covered by reapproximating the retroperitoneum, and the abdomen was closed in layers with standard surgical technique. One week after the initial graft had been placed and contaminated, each animal underwent reoperation. The infected graft was removed, cultured, and the retroperitoneum debrided. The animals were then randomly assigned to one of two experimental groups: group I (collagen, n = 36) received control collagen-impregnated knitted Dacron grafts; group II (rifampin, n = 47) received experimental collagen-rifampin-bonded Dacron prostheses. In each case the replacement graft was implanted in situ, in an end-to-end fashion, with 6-0 polypropylene suture. These two groups were then subdivided and treated with one of four different supplemental antibiotic protocols: (a) no adjunct antibiotic therapy (No Abx) (b) cephalosporin peritoneal irrigation (cefazolin 500 mg/1000 ml) during operation and two doses of cephalosporin (cefazolin, 500 mg intramuscularly) postoperatively (Peri-op), (c) Peri-op plus 1 week ofcephalosporin (cefazolin, 500 mg intramuscularly, twice daily) (One wk), and (d) Peri-op plus 2 weeks ofcephalosporin (cefazolin, 500 mg intramuscularly, twice daily) (Two wks). The abdomen was again dosed in layers. Three to four weeks after the second operation all animals were again anesthetized to recover specimens for bacteriologic study. Cultures were obtained from
Volume 16 Number 4 October 1992
Antibiotic-bonded graft after graft infictions 653
I 100
100. 90. 80.
ll
Collagen Graft
[]
Rifampin Graft
100
100
.oC 70, 6o
~- 50.
84o. N 80. 20. 10. O. No Abx
Pefi-op One wk Two wk~ Supplemental Antibiotic Treatment Protocol
Fig. 1. Chart shows effect of supplemental antibiotics on graft incorporation. Results are expressed as percentage of grafts that were well incorporated at sacrifice. Data are separated into the four supplemental antibiotic treatment groups: no adjunct antibiotics (No Abx), perioperative antibiotics only (P~/-W), perioperative plus 1 week of postoperative antibiotics (One ruk), and perioperative plus 2 weeks of postoperative antibiotics (Two wks).
Tab!¢ I. Intraoperative evaluation of graft incorporation and gross purulence No Abx
Collagen Rifampin
Peri-op
1 wk
2 wk
Incorp
Purulence
Incorp
Purulence
Incorp
Purulence
Incorp
Purulence
n
(%)
(~)
n
(%)
(~)
n
(%)
(~)
n
(%)
(%)
5 16
0.0 62.5
40.0 12.5
16 16
68.7 100.0
18.7 0.0
5 5
40.0 100.0
20.0 0.0
10 10
60.0 100.0
10.0 0.0
NoAbx, No adjunct antibiotics; Per/-0p, perioperative antibiotics only; 1 wk, perioperative plus 1 week of postoperative antibiotics; 2 wk, perioperative plus 2 weeks of postoperative antibiotics; Incorp, graft incorporation.
the grafts and peri-grafr tissues separately. In addition to the bacteriologic data, information regarding graft patency, graft incorporation, and the presence of any gross purulence was recorded. Surgeons making these determinations were blinded with regards to the animal's cohort. The animals were killed at the end of this operation. The data from this experiment were analyzed with chi-square analysis and, when appropriate, the onetailed Fishers exact test for small groups. Statistical significance was assigned whenp values were less than 0.05. All animal care complied with the "Principles of Laboratory Animal Care" and the "Guide for the Care and Use of Laboratory Animals" (NIH Publication No. 80-23, revised 1985).
Bacteriologic studies. A total of five specimens were collected at each operation (both for the initially infected graft and the in situ replacement graft). These included three segments of graft (proximal, midportion, and distal), one swab specimen from the exterior of the graft (abscess cavity), and one specimen from the peri-graft tissue. The swab specimens were directly applied onto the surface of sheep-blood agar plates. Both the graft segments and the tissue specimens were initially placed in brain heart infusion (BHI) broth and incubated at 37 ° C for 24 hours. The BHI broth was then plated onto sheep-blood agar plates and allowed to incubate for periods of up to 2 weeks. If colonies ofS. aureus grew in these plates, then the specimen was considered positive. If S. aureus failed to grow over the 2-week
Journal of VASCULAR SURGERY
654 Coltmrn et al.
l
100 oC 0
1
Collagen Graft
[]
Rifampin Graft
80
0.
40 J=
10
0
,_ll0
Ped-op One wk Supplemental AntibioticTreatment Protocol
No Abx
Two wks
Fig. 2. Chart shows effect of supplemental antibiotics on occurrence of gross purulence. Results are expressed as percentage of grafts that were grossly purulent at sacrifice. Groups as indicated in Fig. 1. Table 1I. Bacteriologic evaluation of graft infection No Abx
Collagen Rifampin p value
5 16
Peri-o#
TOG Cx
Graft Cx
(%)
(%)
100 62.5 NS
100 50 < 0.05
n
16 16
I wk
TOG Cx
Graft Cx
(%)
(%)
87.5 62.5 NS
87.5 50.0 < 0.05
n
5 5
2 wk
T&G Cx
Graft Cx
(%)
(%)
100 60 NS
100 20 < 0.05
n
10 10
T6aG Cx
Graft Cx
(%)
(%)
80 20 < 0.02
80 20 < 0.02
NoAbx, No adjunct antibiotics; Per/-op, perioperative antibiotics only; 1 wk, perioperative plus 1 week of postoperative antibiotics; 2 wk, perioperative plus 2 weeks of postoperative antibiotics; T&G Cx, culture Positivityof tissue and graft samples combined; Graft Cx, culture positivity of graft segments alone, irrespective of the results of the surrounding tissue cultures; NS, not significant.
period, they were considered negative. Analysis included determination of the culture positivity of both the tissue and graft samples combined (T&G Cx) and of the graft segments alone (Graft Cx), irrespective of the results of the surrounding perigraft tissue cultures. Results are expressed as the percentage of animals that were culture positive at the time they were killed. RESULTS All of the initially implanted prosthetic grafts and their Peri-graft tissues were infected with S. aurcus at the time of the second operation. All of these grafts were patent, but each showed poor incorporation and gross evidence of purulence. There were no anastomotic disruptions and no animals died. Furthermore, all animals survived the secondary in situ graft replacement operation and were accounted for
at the conclusion of the experiment. At sacrifice, all 83 grafts were patent and there were no anastomotic disruptions. The percentage of grafts with good incorporation or evidence of gross purulence is shown in Table I. Good incorporation was found in none of the five (0%) collagen (group I) grafts that received no adjunct antibiotics (No Abx), and gross purulence was present in two of five (40%). The percent incorporation increased to 68.7%, 40%, and 60% and the presence of purulence decreased to 18.7%, 20%, and 10% when the animals received either Peri-op, One wk, or Two wks of supplemental antibiotic therapy, respectively. The rifampin grafts (group II) without supplemental antibiotics (No Abx) were incorporated in 10 of 16 (62.5%), and gross ptmalence was detectable in two of 16 (12.5%). Rifampin grafts were all well incorporated (100%,
Volume 16 Number 4 ~ October 1992
Antibiotic-bonded graj~ after ffraft infctiom 655
If~,'t
100
•
Collagen Graft
[]
Rifampin Graft
100
o o == o
0 A
~+
No Abx
Peri-op One wk Supplemental Antibiotic Treatment Protocol
Two wks
Fig. 3. Chart shows effect of supplemental antibiotics on subsequent culture positive rate of graft samples alone. Results are expressed as percentage of grafts that remained culture positive at sacrifice. Groups as indicated in Fig. 1. 100%, and 100%) and showed no evidence of gross purulence (0%, 0%, and 0%) when the animals received Peri-op, One wk, or Two wks of therapy (Figs. 1 and 2). The results of the bacteriologic cultures are shown in Table ]I. When the graft segments were analyzed alone (Graft Cx), the collagen-rifampin-bonded graft reduced the incidence of culture positivity in all four treatment groups. For group I, Collagen graft animals, this rate was 100%, 87.5%, 100%, and 80% when the animals received No Abx, Peri-op, One wk, and Two wks of supplemental antibiotic therapy, respectively. For group II, Rifampin graft animals, the corresponding values were 50%, 50%, 20%, and 20%. This reduction in culture positivity, seen in animals in which the aorta was reconstructed with a rifampin-impregnated graft, was statistically significant in all four treatment cohorts. When overall infection rates (T&G Cx) were evaluated, the collagen-rifampin-bonded graft again reduced the incidence of culture positivity in all four treatment groups. T h e rate in group I animals was 100%, 87.5%, 100%, and 80% for the No Abx, Peri-op, One wk, and Two wks of supplemental antibiotic therapy groups, respectively. In group II animals, the corresponding T&G Cx positivity rates were 62.5%, 62.5%, 60%, and 20%. Unlike that found in the analysis of graft cultures alone (Graft Cx), the reduction in T&G Cx culture positivity observed in the rifampin graft cohorts was only statistically
significant in the subset of animals treated with two weeks (Two wks) of supplemental antibiotics (Figs. 3 and 4). DISCUSSION The current management of prosthetic graft infectious is based on several precepts of proven value. The prosthetic graft must be removed. Continuity of arterial blood supply is restored by means of an extraanatomic bypass that circumvents the area of infection. In addition, patients are treated with intravenous antibiotics for prolonged periods of time. Some patients are placed on lifelong regimens of suppressive oral antibiotics. Despite this systematic approach, several problems remain associated with this management. The best reported primary patency rates for axillobifemoral bypass grafts are between 62% and 72% at 5 years, s,e These studies, however, only include patients in whom the indication for bypass was either aneurysmal or occlusive disease. When extraanatomic graft patency is evaluated in patients in whom the primary indication for operation was an aortic graft infection, the primary graft patency rate at 3 years is 43%, which can be improved to 65% after secondary procedures. 7 Subsequent limb loss associated with a failed axiUobifemoral graft is as high as 34%. 7 Furthermore, the recurrence of infection in an extraanatomic bypass graft is significant at about 20%. 7 Lastly, the incidence of fatalities during the
656
Journal of VASCULAR SURGERY
Colburn et al.
[ ] Collagen Graft 100
o
*6
100
[]
Rifampin Graft
90,
~ 70. 50. ¢9 40, 30, o
~" 10, & O, No Abx
Ped-op One wk Supplemental AntibioticTreatment Protocol
Two wks
Fig. 4. Chart shows effect of supplemental antibiotics on subsequent culture positive rate of graft and tissue samples combined. Results are expressed as percentage of grafts and/or tissues that remained culture positive at sacrifice. Groups as indicated in Fig. 1. course of these reconstructive efforts has been reported to be as high as 25%. 7.9 In one report, 82% of patients treated for an aortic graft infection were dead at 5 years. 1° In situ reconstruction of primary vascular infections along with surgical debridement and prolonged antibiotic administration has yielded some success. In one study by Brown et al.,n the survival rate after in situ reconstruction in patients with aortic mycotic aneurysms was 63% with a reinfection rate of 19% and an incidence of rupture of 11%. In situ reconstruction after prosthetic graft infections has been reported sporadically, however, the results have been variable. In a review in 1983, Bunt 12 reported a mortality rate of 56% after in situ reconstructions in patients with graft-enteric fistulas. More recently, Jacobs et alj3 reported their experience with in situ graft replacement in 18 patients with infected aortic grafts. In this study, 12 of the 18 patients were classified as having low-grade graft infections with negative blood and peri-graft cultures. All 12 of these patients were alive at a mean follow-up period of 8 years. The remaining six patients had severe graft infections with positive blood and peri-graft cultures. In this group, five patients died within 1 month of operation and the remaining patient required conversion to an extraanatornic bypass because of recurrent infection. The authors concluded that there may be a role for in situ reconstruction in patients who have low-grade aortic graft infections. A similar conclusion was reached recently in a report by
Robinson and Johansen.14 In this paper, the authors reviewed their experience and that of the literature regarding in situ replacement of infected aortic grafts. Their findings suggested that the operative mortality rate in these patients is as low as 19% and that this procedure may be appropriate in a select group of patients with low-grade graft infections. Lastly, Bandyk et aLas recently reported their experience with in situ graft replacement in 15 patients with low-grade graft infections secondary to Staphylococcus ep/derm/d/s. In this series, no deaths, graft thromboses, or recurrent infections occurred after a mean follow-up period of 21 months. The authors also suggested that a reevaluation of the use of in situ graft replacement for infected vascular prostheses seems warranted. The development of a graft that is capable of resisting bacterial contamination would have several advantageous uses. It would be expected to reduce the current incidence of graft infections. It may be able to reduce the need for perioperative antibiotic coverage, and it may improve the results after an in situ reconstruction in an infected arterial bed. The characteristics of such a graft should include a broad-spectrum antibacterial effect, with particular efficacy against the most common pathogens. The duration of the antibacterial effect should be as prolonged as possible and have minimal risk of toxicity. Finally, the antibacterial agent should not degrade the graft or make its technical use less desirable.
Volume 16 Number 4 October 1992
Past experiments in the development of a bacterial resistant prosthetic graft have used several antibiotic agents combined with many modes of application. These efforts have included passive soaking of grafts with antimicrobial d r u g s , 16 ionic bonding of antibiotics with various bonding agents, 17"21 and topical application of antibiotic sealants. 22'23 Unfortunately, differences in antibiotic dosages, tested organisms, and methods of evaluation make comparisons of these techniques difficult. Several properties of rifampin make it a useful agent for this purpose. It has a rather broad spectrum of antimicrobial activity and is particularly active against gram-positive cocci. Further, it is relatively hydrophobic and thus will not dissolve into the bloodstream as rapidly as other antibiotics. This low aqueous solubility, combined with the binding of the agent by a collagen matrix, results in the gradual release of the antibiotic and therefore a prolonged local antibacterial action. In vitro studies have demonstrated activity of collagenrifampin bonded grafts in inhibiting S. aureus growth for up to 3 weeks. 3 In addition, in vivo experiments have shown antibacterial activity in this graft after an intravenous challenge for as long as 7 days after implantation. 4 The present experiment has demonstrated that the collagen-rifampin prosthesis is capable of inhibiting bacterial growth on the graft for periods of up to 3 weeks. The in situ replacement of an infected arterial prosthesis with this antibiotic-bonded graft, supplemented with systemic antibiotics, resulted in no gross evidence of infection after a 3-week period. Attempts to culture bacteria from the graft after this 3-week period were unsuccessful in 50% of cases when no adjunct antibiotic:therapy was administered. This was improved to an 80% culture negative rate when the graft was supplemented with 2 weeks of systemic antibiotic therapy. This represented a significant improvement over the collagen (group I) grafts. This is all the more remarkable given the implantation of the prosthetic grafts into grossly infected arterial beds with demonstrated abscesses. Clearly a 20% subsequent infection rate after the in situ replacement of a previously infected bypass graft remains prohibitively high. Therefore, use of this prosthesis in a comparable clinical setting is not warranted at this time. On the other hand, as mentioned previously, in situ graft replacement with unprotected prosthetic grafts in the setting of lowgrade graft infections is currently being performed with considerable success./31s It is therefore not unreasonable to assume that the results of this study, which tested the antibiotic-bonded graft in a highly
Antibiotic-bondedgraft after graft infections
657
challenging model, would have likely been improved had the grafts been used to replace prostheses containing only a low-grade infection. Although the infection rate of the rifampinimpregnated grafts was significantly lower than that of the collagen grafts in all four treatment groups when graft cultures were considered alone, when the results of the cultures on the peri-graft tissues were included, this difference was lost in the No Abx, Peri-op, and One wk treatment groups. Two weeks of supplemental antibiotic therapy, however, was able to sterilize the surrounding peri-graft tissues and significantly reduce the overall (T&G Cx) culture positive rate. The persistence of organisms in the peri-graft tissues suggests that the antimicrobial effect of the graft alone is not able to penetrate the adjacent tissues and eradicate the bacteria. The use of prolonged perioperative antibiotics (as would normally be done in most instances of graft infection) clearly improves the success of this technique. We conclude that the coUagen-rifampin-bonded Dacron graft is able to reduce the incidence of subsequent bacterial growth on a prosthesis after implantation into a grossly infected arterial bed. The addition of perioperative parenteral antibiotic supplementation, however, is required to sterilize the surrounding peri-graft tissues. REFERENCES
1. Goldstone J, Moore WS. Infection in vascular prostheses: clinical manifestations and surgical management. Am J Surg 1974;128:225-33. 2. Moore WS, Chvapil M, Sieffert G, Keown K. Development of an infection resistant vascular prosthesis. Arch Surg 1981; 116:1403-7. 3. Chervu A, Moore WS, Chvapil M, Henderson T. Efficacyand duration of antistaphylococcal activity comparing three antibiotics bonded to Dacron vascular grafts with a collagen release system. J VASCSUgG 1991;13:897-901. 4. Chervu A, Moore WS, Gelabert HA, Colburn MD, Chvapil M. Prevention of graft infection by use of prostheses bonded with a rifampin/collagen release system. J V~sc SURG 1991;14:521-5. 5. Rutherford RB, Patt A, Pearce WH. Extraanatomic bypass: a closer view. J V ~ c SURG 1987;6:437-46. 6. Hepp W, Jonge K, Pallua N. Late results following extraanatomic bypass procedures for chronic aortoiliac occlusive disease. J Cardiovasc Surg (Torino) 1988;29:181-5. 7. Quifiones-BaldrichWI, Hernandezll, MooreWS. Long-term results following surgical management of aortic graft infection. Arch Surg 1991;126:507-11. 8. Reilly LM, Stoney RJ, Goldstone J, Ehrenfeld WK. Improved management of aortic graft infection: the influence of operation sequence and staging. J VASCSURG1987; 5:412- 31. 9. Yeager RA, Moneta GL, Taylor LM, McConnell DB, Porter JM. Can prosthetic graft infection be avoided? If not, how do we treat it? Acta Chir Scand Suppl 1990;555:155-63.
658
Journal of VASCULAR SURGERY
Colburn et al.
10. O'Hara PJ, Hertzer NR, Beven EG, Krajewski LP. Surgical management of infected abdominal aortic grafts: review of a 25-year experience. J VASCSURG 1986;3:725-31. 11. Brown SL, Busuttil RW, Baker ID, Machleder HI, Moore WS, Barker WF. Bacteriologic and surgical determinants of survival in patients with mycotic aneurysms. J VASC SURG 1984;1:541-7. 12. Bunt TJ. Synthetic vascular graft infections, II: graft-enteric erosions and graft-enteric fistulas. Surgery 1983;94:1-9. 13. Jacobs MJ, Reul GJ, Gregoric I, Cooley DA. In-situ replacement and extra-anatomic bypass for the treatment of infected abdominal aortic grafts. Eur J Vasc Surg 1991;5:83-6. 14. Robinson JA, Johansen K. Aortic sepsis: is there a role for in situ graft reconstruction? J VASESURG 1991;13:677-84. 15. Bandyk DF, Bergamini TM, Kinney EV, Seabrook GR, Towne JB. In situ replacement of vascular prostheses infected by bacterial biofilms. J VASC SURG 1991;13:575-83. 16. Powell TW, Burnham SJ, Johnson G. A passive system using rifampin to create an infection-resistant vascular prosthesis. Surgery 1983;94:765-9. 17. Benvenisty A, Tannenbaum G, Ahlborn TN, et al. Control of prosthetic bacterial infection: evaluation of an easily incorporated, tightly bound, silver antibiotic PTFE graft. J Surg Res 1986;44:1-7.
18. Sobinsky KR, Flanigan DP. Antibiotic binding to polytettafluoroethylene via glucosaminoglycan-kerarin luminal coating. Surgery 1986;100:629-34. 19. Greco RS, Harvey RA, Smilow PC, Tesoriero JV, Prevention of vascular prosthetic infection by a benzalkonium-oxacillin bonded polytetrafluoroethylene graft. Surg Gynecol Obstet 1982;155:28-32. 20. Shu¢ WB, Worosilo SC, Donetz AP, Trooskin SZ, Harvey RA, Greco RS. Prevention of vascular prosthetic infection with an antibiotic-bonded Dacron graft. J VASC SURG 1988;8:600-5. 21. Kinney EV, Bandyk DF, Seabrook GA, Kelly HM, Town JB. Antibiotic-bonded PTFE vascular grafts: the effect of silver antibiotic on bioactivity following implantation. J Surg Res 1991;50:430-5. 22. Shenk JS, Ney AL, Tsukayama DT, Olson ME, Bubrick MP. Tobramycin-adhesive in preventing and treating PTFE vascular graft infections. J Surg Res 1989;47:487-92. 23. Ney AL, Kelly PH, Tsukayama DT, Bubrick MP. Fibrin glue-antibiotic suspension in the prevention of prosthetic graft infection. J Trauma 1990;30:1000-6. Submitted Feb. 13, 1992; accepted May 1, 1992.
DISCUSSION Dr. Kenneth E. McIntyre, Jr. (Tucson, Ariz.). This study is another in a long line of accomplishments by the UCLA Vascular Surgery group whose goal has been to develop an infection-resistant aortic prosthesis. The current study that was presented today offers a new twist, however. Dr. Gelabert and associates have tackled a clinical problem that has plagued all of us who do vascular surgery: Is there an alternative to extraanatomic arterial reconstruction after excision of an infected aortic prosthetic graft? In the past, we have accepted the mediocre long-term patency rates of axillobifemoral bypass grafts, which were placed to provide infrainguinal perfusion after excision of an infected aortic graft. In situ graft replacement has been a current "hot topic" of late as an alternative to these dismal results. Although several anecdotal reports abound, this is the first prospective trial done in canines to address this problem. The authors evaluated a knitted Dacron velour graft impregnated with collagen bound to an antibiotic, which was implanted in an infected surgical field. As in their previous study, which was reported before this society, rifampin was bonded to the collagen before being applied to the graft. Rifampin was chosen became of its low solubility in water and its gradual release from the collagen matrix, which allowed for continued bacteriocidal levels of antibiotic over time. Mongrel dogs underwent laparotomy with a 3 cm aortic tube graft replacement and direct contamination of the aortic prosthesis with S. aureus bacteria. One week later, another laparotomy was performed to remove the infected aortic grafts. The dogs were
then randomized into one of two groups on the basis of their subsequent treatment. Excision of the infected aortic graft with replacement by a knitted Dacron double-velour collagen-coated aortic prosthesis was performed in 36 animals, and 47 dogs underwent infected aortic graft excision with replacement by an identical graft except that the collagen was bonded to rifampin. Each group was then subdivided into one of four treatment protocols: (1) no antibiotic therapy, (2) topical cephalosporin antibiotic peritoneal irrigation plus two doses of parenteral cephalosporin (Peri-op), (3) Peri-op plus 1 week of parenteral cephalosporin, or (4) Peri-op plus 2 weeks of parenteral cephalosporin. All the grafts were removed with sterile technique between 3 and 4 weeks after implantation and cultures were obtained from the grafts and peri-graft tissues separately. There was a direct correlation between surgeon evaluation of graft incorporation at the time of graft excision and presence or absence of purulence. Generally, the more the graft was incorporated, the smaller the incidence of purulence. It is not surprising that the collagen-bonded rifampin grafts demonstrated a greater resistance to graft infection than the grafts lined with collagen alone. However, the effect of rifampin was only apparent in the graft itself. The peri-graft tissue surrounding the infected graft remained infected despite the presence of the rifampin-bonded collagen graft. This result is not surprising considering that rifanapin is slowly released from the collagen over time. Therefore statistical significance was only obtained between the two groups
Volume 16 Number 4 October 1992
when graft cultures alone were considered. However, when overall infection rates were evaluated in both tissue and graft cultures, statistical significance was only obtained in that subset of animals that were treated with 2 weeks of supplemental antibiotics. With this model, therefore, although the collagen-rifampin-bonded graft reduced the incidence of graft colonization after replacement of an infected graft, additional antibiotics were required to help sterilize surrounding peri-graft tissues. I have no problem with either the study design or methods, but I would like to pose several questions to the authors. Why did the rifampin-collagen grafts seem to incorporate better? Do they incorporate better than collagen-coated grafts without rifampin in the absence of infection? Does incorporation alone in this case influence infectability? Because of its low solubility, rifampin was dissolved in ethanol to facilitate its homogeneous bonding to and distribution within the collagen. Can you speculate as to whether residual ethanol may have influenced either graft incorporation or resistance to infection? Third, since the graft was not effective in treatment of peri-graft infection, do you think in situ graft replacement can ever be successful when there is severe contamination of peri-graft tissue? Fourth, were any cultures obtained of the aortic wall itself?. We have found in our studies of graft infections in humans that when the aortic wall itself is infected, eradication of infection by any means is unlikely. Finally, will this new graft be subjected to appropriate clinical trials to evaluate its efficacy or will it suddenly appear on shelves as the next expensive wonder device before an adequate evaluation has been completed? I enjoyed the manuscript and recommend it for your review. Dr. H u g h A. Gelabert. There are several points that I would like to emphasize. First, we consider this one of the most severe bacterial challenges: placing a graft into what one of our residents refers to as a "puddle of pus." This required an extensive debridement and antibiotic irrigation of the retroperitoneum. I think it is impressive that we were able to reduce the subsequent infection rate. The second point is that the graft itself has often been considered the anchor of the infection. We have been taught that what causes a graft infection to persist is the affinity between the organisms and the prosthetic graft material. Further, the biologic tissues surrounding this area were thought to be innocently subjected to infection and to play a passive role in the infectious process. In this experiment we have succeeded in making a graft that will clear itself of the bacterial infection. We discovered that the surrounding biologic tissues served to anchor the infection, and that is where prolonged treatment with antibiotics is helpful. I mention this because it gets back to several important points that Dr. Mclntyre brought up in his discussion. Specifically, the questions of the role of this graft in a clinically severely infected patient and the issue of involvement of the artery itself. We did not specifically remove segments of the arteries
Antibiotic-bondedgra~ ajqm"gruft infections 659 for microbiologic study. We did, however, incorporate the proximal and distal ends of the arteries along with the specimen as it was excised during the second operation. This specimen was subsequently divided into proximal, middle, and distal portions. These were, in turn, analyzed for bacterial contamination. There was no significant difference between any segment with regards to infection: they were all infected. Reconstruction with the experimental graft required resection of the aorta back to a point where the arterial wall consisted of grossly healthy, viable tissue. The resected perianastomotic portion of the aortas was inflamed and edematous. It would not have been adequate for anastomosis. As to the question of whether this graft is ready for implantation in humans: That remains to be established. We are certainly working in that direction. The question regarding the relationship of incorporation to infection and the effect of the collagen binding is provocative. Clearly, infection will inhibit incorporation, and a poorly incorporated graft is thought to be more likely to become infected. At times it is impossible to tell which comes first. The effect of the rifampin binding could be answered by comparing the rifampin-bound graft against a collagen graft in a noninfected model. Unfortunately, we have not done this. The question of the residual presence of ethanol in the experimental graft is unanswered. In the process of graft construction, the graft is heated in a formalin vapor to provide minimal collagen cross-linkage. It has been our presumption that the ethanol evaporates in the course of this process. We do not, however, have any specific, conclusive evidence to that effect. Finally, as to whether the graft will suddenly appear on the shelf next week, your speculation is about as good as mine. I doubt that this will happen, since this is a new means of applying and delivering the antibiotic. As far as I understand, the Food and Drug Administration will require a set of trials before it can appear on the shelf. Thank you very much for your insightful comments. Dr. Linda M. Reilly (San Francisco, Calif.). One interpretation of your data could be that the effect of intravenous antibiotics does not begin to manifest itself until at least 2 weeks of treatment have elapsed. Have you considered extending this experiment for a longer period of time, particularly to show the effect of the antibiotics in your control group? One other question: Had you considered using a non-collagen-impregnated graft as your true control? Although there is an increasing use of collagen impregnated grafts, I think that for most of us the more common graft is a non-collagen-bonded graft. Dr. Gelabert. There is no question about it. I think there are several variations in this experiment that would be helpful and that would be fun to do. Extending the period of time certainly would be of great interest to see how durable this effect is, first of all, and secondly to see whether
660
Journal of VASCULAR SURGERY
Colburn et al.
or not the outcome of the animals changes. Unfortunately, within the limits of time and money we have not done that yet, but that is something that we have been considering. In response to the second question as to the use of a non-collagen-bound graft, we actually did use noncollagen-bound grafts in previous experiments. As a matter of fact, when developing this model we went through a whole set of developmental experiments in which we were trying to determine the minimal infected bacterial concen-
trations. The first graft that we placed was a non-collagenbound graft. We found that exceedingly easy to infect. In addition, there's a tremendous historic,~l background that has demonstrated that the infections persist when these grafts are used in animals. So we didn't really think it was absolutely necessary to do that. Finally, to provide a good comparison with the rifampin-bound graft, which is bound in a collagen matrix, we thought it would be best to use the collagen-bound graft.
SOCIETY F O R V A S C U L A R S U R G E R Y LIFELINE
FOUNDATION
GRANT
AWARD
The Lifeline Foundation of the Society for Vascular Surgery invites grant applications for funding of meritorious research by young surgical investigators. The awards are intended for surgeons who have completed their formal surgical education in general surgery and who have completed or are in an advanced training program in vascular surgery. To be considered for selection a candidate: i. Should be certified by the American Board of Surgery or have completed the requirements for certification 2. Should submit an application within 3 years of completion of an approved residency training program 3. Must have either a faculty appointment in an approved medical school in the United States or Canada or have received an academic appointment within the guidelines of the applicant's institution Grant awards are not intended to supplement salary, which will remain the responsibility of the institution in which the awardee holds an appointment. The awardee is expected to devote a significant amount of time to the funded project. A progress report will be presented by the investigators during the annual meeting of the Society for Vascular Surgery. A grant awards committee will review competitive applications. It is anticipated that two grants will be awarded annually totaling $50,000 each to include indirect costs. Each award will be for 1 year with the option to extend for an additional year. Grant applications may be obtained from: The Lifeline Foundation Society for Vascular Surgery Thirteen E l l St. Manchester, MA 01944 (508) 526-8330
The deadline for receiving applications in the Foundation office is January 15, 1993. Funds will be awarded by July 1, 1993.
