- Email: [email protected]
In situ replacement of arterial prosthesis infected by bacterial biofilms: Long-term follow-up
In situ replacement of arterial prosthesis infected by bacterial biofilms: Long-term follow-up Jonathan B. Towne, MD, Gary R. Seabrook, MD, Dennis Bandyk, MD, Julie A. Freischlag, MD, and Charles E. Edmiston, MD, Milwaukee) Wis.
Purpose: Bacterial biofilm infections of vascular prostheses are characterized by an absence of systemic sepsis, a fluid-filled cavity surrounding the graft, a draining sinus tract, and microorganisms that must be removed from the fabric prosthesis for bacterial culture. Methods: Since 1987 we have treated 20 infected grafts with prosthetic excision and in situ replacement in 14 men and 6 women. The time from initial graft implantation to diagnosis of graft infection ranged from 3 months to 14 years (mean 4.5 years). The original graft (Dacron-17, polytetrafluoroethylene-3) was an aortobifemoral in 14, axillofemoral femorofemoral in 3, iliofemoral in 2, and femoropopliteal in 1 patient. Presenting symptoms were groin false aneurysm with perigraft fluid in 10, inflammatory mass in 6, and sinus tract in 4. At surgery all unincorporated graft material and the perigraft capsule were excised from a point where the proximal graft was incorporated, including debridement of vessels at the distal anastomosis. Of the 14 aortobifemoral grafts, only the femoral limbs were excised at the initial presentation ofbiofilm infection. The conduit was replaced with an in situ polytetrafluoroethylene interposition graft, which was covered with a gracilis or sartorius muscle flap when possible. Results: All surgical sites healed, all grafts remained patent, and there was no limb loss. After ultrasonic oscillation of the explanted graft, bacterial cultures recovered coagulasenegative Staphylococcus species in 14, coagulase-positive Staphylococcus species in one, both species in three, with no growth from two specimens. During follow-up, two patients have had clinical involvement in the proximal intraabdominal portion of the graft that had not been previously resected. In all grafts, the in situ replacement graft remained well incorporated. Conclusion: In situ graft replacement is effective treatment for biofilm infections of vascular prostheses. Because of the indolent nature of these infections, subsequent infection of previously uninvolved graft segments may be expected. (J VAse SURG 1994;19:226-35.)
Coagulase-negative staphylococci are a significant cause of bioprosthetic infections, because of their unique ability to adhere to and grow on implanted artificial surfaces and to secrete a biofilm or slime, which is a galactose-rich polysaccharide that provides protection from the body's host defenses. 1,2 Bandyk et aI., in a la-year review of aortofemoral graft infections, determined that Staphylococcus epidennidis From the Department of Vascular Surgery at the Medical College of Wisconsin, Milwaukee. Presented at the Forty-seventh Annual Meeting of the Society for Vascular Surgery, Washington, D.C., June 8-9, 1993. Reprint requests: Jonathan ·B. Towne, MD, Department of Vascular Surgery, 8700 W. Wisconsin Ave., Milwaukee, WI 53226. Copyright © 1994 by The Society for Vascular Surgery and International Society for Cardiovascular Surgery, North American Chapter. 0741-5214/94/$3.00 + 0 24{6{52342
226
was the infecting organism in 60% of the cases and occurred a mean of 41 months after the initial operation. 3 In a subsequent prospective study of anastomotic femoral false aneurysm after aortofemoral bypass grafting, Seabrook et al. 4 reported that 60% of anastomotic femoral false aneurysms presenting a mean of 95 months after graft implantation were infected with coagulase-negative staphylococci. Usually, vascular prostheses infected with coagulasenegative staphylococci are associated with a less virulent clinical course compared with vascular graft infections caused by coagulase-positive staphylococci or gram-negative bacteria. Infections are not associated with fever, leukocytosis, or positive blood culture results. Graft biofilm infections usually present as a graft healing complication such as absence of tissue incorporation with or without perigraft cavities and perigraft fluid, false aneurysms,
JOURNAL OF VASCULAR SURGERY Volume 19, Number 2
or graft-cutaneous sinuses. Because the bacteria are adherent to the prosthesis, they are seldom present in the periprosthetic tissue fluid. Microscopic examination of this perigraft fluid demonstrates multiple white blood cells and only rarely are bacteria seen. Because culture results of this perigraft fluid are routinely negative, for many years investigators believed that these grafts were not infected and that the lack of graft incorporation and the development of perigraft fluid was an "allergic response" to the vascular prosthesis. However, bacteria can be recovered if they are mechanically dislodged from the fabric of the graft or if the graft material itself is placed in culture media. Because the infection is confined to the prosthetic vascular conduit, the potential for graft excision and in situ replacement is possible. To evaluate this technique, a canine model was developed to reproduce all of the findings classically seen in a graft biofilm infection: lack of graft incorporation by perigraft tissue, perigraft fluid, and the ability to culture the infecting organism only from the graft fabric. 5 With this canine model the established prosthetic surface biofilm infection with S. epidermidis was treated by graft excision, parenteral cefazolin, and in situ graft replacement with a PTFE vascular prosthesis. There was normal perigraft healing, including anastomotic tensile strength 1 month after in situ replacement. Because of these laboratory findings, in 1987 we cautiously began a clinical trial of in situ replacement in patients with graft biofilm infections. An initial report confirmed low morbidity and early eradication of the biofilm infection. 6 This study presents the long-term results of this clinical series with emphasis on the outcome of both replaced and previously uninvolved graft segments. METHODS Patients. Since 1987, 20 patients with biofilm graft infections were treated with graft excision and in situ replacement. There were 14 men and six women. The time from initial graft implantation to diagnosis of infection ranges from 3 months to 14 years, with a mean of 4.5 years. Initial indications for primary grafting procedure were atherosclerosis obliterans in 16, ruptured abdominal aortic aneurysm in three, and elective aneurysm in one. The original graft was knitted Dacron in 17 and polytetrafluoroethylene (PTFE) in three and was placed initially as an aortobifemoral graft in 14, axillofemoral, femorofemoral graft in three, an iliofemoral in two, and femoropopliteal graft in one patient. Presenting symptoms of graft infection included
Towne et al.
227
groin false aneurysm with perigraft fluid in 10, inflammatory groin mass in six, and sinus tract in four. In 12 patients the graft segment involved by the suspected bacterial biofilm infection had undergone one or multiple secondary procedures, and three patients had the initial aortofemoral graft placed for ruptured abdominal aortic aneurysm. Only five patients enrolled in the series with late graft infection had undergone a single, elective, arterial reconstruction. All 20 patients were diagnosed with the clinical criteria of graft biofilm infection, which included presentation months to years after graft implantation, lack of systemic signs of infection (afebrile, no leukocytosis), and no growth on blood cultures. The perigraft tissue demonstrated only inflammation with no bacteria seen on microscopic examination. There was a perigraft cavity with fluid with an absence of graft incorporation often accompanied by an anastomotic pseudoaneurysm. The perigraft fluid gram stain showed white blood cells with no bacteria and the perigraft fluid culture showed no growth. There was no evidence of an aortic false aneurysm in any of the patients with involvement of a limb of an aortofemoral graft. Microbiologic technique. The microbiologic technique has been previously described. s,6 In summary, cultures of the perigraft tissue fluid were obtained by sterile cotton swabs and were plated on agar media. At operation 1 cm length of prosthetic material adjacent to the femoral anastomosis or within the perigraft cavity was excised and placed directly in 20 ml of glucose supplemented trypticase soy broth (Becton Dickinson Microbiology Systems, Cockeysville, Md.). The explanted graft specimen submitted for culture did not include thrombus, perigraft capsule or adjacent artery. A second graft segment of similar size was placed in trypticase soy broth and ultrasonically oscillated at 20 kHz for 10 minutes to mechanically disrupt the graft surface biofilm and disperse adherent bacteria into the broth media for growth. The broth and biofilm culture of graft material were incubated at 35° C for up to 2 weeks. Organisms were isolated and identified after standard microbiologic technique. Operative technique. Treatment of aortofemoral graft abnormalities in the groin involved vascular isolation at the femoral anastomosis. Usually proximal control was not possible in the groin necessitating a suprainguinal counter incision with retroperitoneal exposure of the more proximal limb of the aortofemoral graft. The replacement limb was sutured end to end to the incorporated proximal limb
228
Towne et al.
JOURNAL OF VASCULAR SURGERY February 1994
Fig. 1. Color-flow duplex examination demonstrates well-incorporated PTFE graft.
and to the debrided femoral vessels. All unincorporated graft material and the perigraft capsule were excised. The artery adjacent to the suture line was debrided to a normal-appearing vessel. Because the scar tissue from the previous procedure was usually intimately adherent to the perigraft capsule, this was all excised leaving only vital structures. Graft cutaneous sinus tracts were excised en bloc. The conduit was replaced in situ with a PTFE interposition graft. The replacement graft was usually covered with a gracilis or sartorius muscle flap, although on occasion, a rectus abdominal muscle flap was used. When distal grafts were involved, adjacent muscle was mobilized to cover grafts. Replacement of the entire aortofemoral graft was not required at the initial treatment in any patient. Eight of 14 aortofemoral graft limbs had sartorius or gracilis myoplasty to cover the revised distal anastomosis. In patients with infected axillofemoral grafts, the new graft was tunneled adjacent to, but separate from, the previous graft. The skin was closed primarily in all cases. Perioperative antibiotics, cefazolin or vancomycin, were administered to all patients and were continued parenterally for 7 to 10 days. Oral antibiotics were continued for 1 month after operation. Patients were monitored every 3 months with color-flow Duplex or abdominal computed tomography (Cf) scanning. The absence of
wound problems and demonstration of a wellincorporated graft by CT and ultrasonography were the criteria for absence of residual infection (Figs. 1 and 2) . RESULTS All patients recovered without signs of graft or wound infection after the in situ grafting procedure. There was no perioperative limb loss, graft thrombosis, or death. Three patients had a lymphocele after operation that was separate from the vascular repair. In two of these, wound drainage developed that was treated with operative exploration, ligation of lymphatic vessels, and primary closure over a closed suction drain. In a third patient, the lymphocele gradually was reabsorbed over 6 months. Bacterial cultures recovered only coagulasenegative staphylococci from 14 patients, coagulasepositive staphylococci from 1, and both species from 3, and no growth in 2. Of the four patients who grew coagulase-positive staphylococci, none grew it from the graft capsule. One of these patients grew a coagulase-positive staphylococci in one groin and coagulase-negative staphylococci in the other. Patients in whom the procedure consisted of resection of a femoral limb of an aortal bifemoral graft were monitored with CT scanning. Patients with lower leg grafts were monitored with duplex scanning of the
JOURNAL OF VASCULAR SURGERY Volwnc 19, Number 2
Timme et aI.
229
Fig. 2. Color-flow duplex examination demonstrates fluid surrounding graft, indicating biofilm infection.
graft looking for the development of fluid around the graft. Ten of the 20 patients died in the follow-up period for a long-term mortality rate of 50%. The average follow-up from graft replacement to death was 32 months and ranged from 3 months to 74 months. The causes of death include cardiac events in five, carcinoma of the lung in three, and stroke and kidney failure in one patient each. No patient died of treatment of graft failure or graft infection complication. Follow-up of these patients ranged from 3 months to 74 months for the entire series (average 39 months). When the patients who died were deleted, the follow-up of the remaining 10 living patients
ranged from 10 months to 67 months (average 45 months). Further graft complications occurred in four patients in the follow-up period. Two patients with aortofemoral grafts had involvement in the proximal intraabdominal portion of the graft that had not been previously resected and had not previously had clinical evidence of infection. This manifested as fluid around the limbs of the graft and, as in the initial presentation, was not associated with any systemic evidence of infection (Fig. 3). One patient had a resection of the residual graft 2 years after in situ replacement of groin biofilm infection and represents the only aortic in situ replacement. At operation, there were classic findings of a graft biofilm infection
230 Tuwne et at.
JOURNAL OF VASCULAR SURGERY February 1994
Fig. 3. Abdominal CT scan with contrast demonstrates fluid around intraabdominal previously unrejected segment of aortobifemoral graft.
with lack of incorporation of the graft and large fluid filled perigraft cavities which extended to the proximal anastomosis. There was no aortic false aneurysm. A perigraft fluid cavity developed around the sup rainguinal portion of an aortofemoral graft in the second patient 2 years after resection and in situ replacement for femoral false aneurysm associated with absence of graft incorporation. The groin repair was well incorporated, and the patient had an interposition graft placed from the proximal groin graft to the proximal iliac limb 6 cm distal to its bifurcation. Fluid has subsequently developed around the remaining residual intraabdominal graft. Because of renal and cardiac complications this patient was not a candidate for aortic limb replacement. She died 74 months after femoral graft resection and replacement. There were two major amputations in the follow-up period. One patient had an in situ femoral limb replacement that healed. A femoral posterior tibial PTFE graft was required 10 months later because of progression of lower extremity occlusive disease. This graft had multiple episodes of thrombosis requiring thrombectomy. The graft eventually became grossly infected with coagulase-positive staphylococci, requiring removal and resulting in an above-knee amputation. The PTFE interposition graft replaced in situ in the groin remains well incorporated without evidence of a perigraft fluid
cavity 2 years after amputation. The second patient had biofilm infection of the femoral limb of an iliofemoral graft. Also involved was the proximal anastomosis of a femoral above-knee popliteal PTFE graft. Arteriograms demonstrated a stenosis of the popliteal artery just beyond the distal anastomosis of the femoral popliteal graft. Mter in situ replacement of the femoral graft limb and wound healing, the patient refused an operative procedure to correct the outflow stenosis. The femoropopliteal graft thrombosed 1 year later, resulting in an above-knee amputation. Including the two patients requiring amputations described above, there were three graft occlusions in the postoperative period. The third patient had an in situ replacement for a biofilm infection of the left groin. One month after operation he required a femoral anterior tibial PTFE bypass because of persistent ischemia of the left foot. The culture at the time of graft replacement grew coagulase-positive staphylococci. Perigraft fluid subsequently developed around his tibial graft, whereas the femoral graft remained well healed. The distal PTFE graft thrombosed at 21/2 years and on removal grew both coagulase-positive and coagulase-negative staphylococci. The in situ groin repair remains healed without evidence of perigraft fluid. To date, none of the prostheses used for in situ
JOURNAL OF VASCULAR SURGERY Volume 19, Number 2
replacement for biofilm infection have shown evidence of infection as manifested by false aneurysm formation, perigraft fluid or cavity, and cutaneous sinus tracts. The two graft infections occurring in distal leg prosthetic grafts did not involve the segment of in situ replacement in the groin. In both of these patients, the muscle flap may have protected the femoral segment from the infection involving the distal graft.
DISCUSSION The low virulence of coagulase-negative staphylococci and their unique characteristics of colonization via a surface biofilm permit treatment of infected grafts by excision of the grossly involved graft segments, debridement of perigraft tissue and adjacent artery, and in situ replacement of another prosthesis. Several caveats must be emphasized in selecting this treatment option. First, it is essential that the patient has the correct diagnosis. This technique is not appropriate or recommended for cases other than biofilm infections (i.e., patients who have gram-negative bacteria infections). The role of coagulase-positive staphylococci in these infections is not yet defined. In latent, indolent infections where coagulase-positive staphylococci are cultured, in situ replacement may be successful. In this series, four such patients did not have systemic evidence of infection, and they did not grow Staphylococcus aureus from the perigraft cavity. Why some patients with coagulase-positive staphylococcal infections fail to mount the systemic reaction that is more classically associated with this infection is not understood. There may be a spectrum of virulence in these species in which some isolates act like the more benign coagulase-negative staphylococci. The key to the operative treatment of these patients is total excision of the biofilm cavity and the involved graft. Because this often results in an extensive groin dissection, there is an increased incidence of lymphatic wound complications, particularly the occurrence of lymphoceles in the postoperative period. The use of muscle coverage is important because it provides vascularized tissue to surround the graft and be interposed between the skin and the conduit if wound complications occur. The treatment of lymphoceles should be aggressive, with operative ligation and drainage if they do not promptly resolve or begin to leak through the wound creating a potential tract for secondary infection. 7 The choice of a PTFE as a replacement graft was selected because our laboratory data demonstrated
Towne et at.
231
that the bacterial adherence of slime-producing bacteria was significantly less to PTFE than to knitted or woven Dacron. 8 Slime-producing coagulasenegative staphylococci adhered to the knitted Dacron 100 times greater than to PTFE. Because bacterial adhesion is the important first step in biofilm graft infection, the relative resistance to bacterial adhesion possessed by PTFE has significant potential advantages as an in situ replacement conduit. A related facet of this study is the profile of the patients who were enrolled. Because biofilm infections occur months and years after an initial vascular reconstructive procedure, the mortality rate from associated cardiovascular disease is high. The mortality rate in this series, with follow-up to 74 months, is 50%, most of it caused by cardiovascular causes. Because of this high mortality rate we are curtailed in studying extended durability of this repair technique. However, if one accepts the thesis that the role of these procedures is to allow the patients to continue their lives with both limbs for as long as possible, that goal is accomplished. A special note of caution needs to be raised for patients who have femoral limbs of their aortal bifemoral grafts resected. In all of these patients in this series, the replacement grafts have healed well and have demonstrated no evidence of recurrent infection. However, the residual aortic graft segment in the patient may subsequently develop changes suggestive of coagulase-negative staphylococcus infection manifested by the development of a perigraft cavity and fluid. This, in all likelihood, is the clinical presentation of a latent infection that existed at the time of the initial resection. It is safe to expect that eventually the total graft will need to be excised if the patient lives long enough. At this point, we have done an intraabdominal resection on only one of these grafts, and the follow-up is now 6 months. Certainly, replacement of the intraabdominal portion of the graft is fraught with the greatest potential complications, because of the risk for development of a graft enteric fistula or false aneurysm with rupture. 9 Given these potential catastrophes, we have been very cautious in applying this technique for the aortic segment. In the absence of long-term evaluation of this technique at the aortic level, infrarenal aortic ligation and extraanatomic bypass must still be considered the standard of care for an aortic graft infection. The source of contamination of graft biofilms infections may be twofold. Slime-producing coagulase-negative staphylococci are often present on the skin at time of hospitalization. Levy et al. 10
232 Tuwne et at.
identified at least one slime-producing staphylococcal species as part of skin flora in 80% of 41 consecutive patients admitted to the vascular surgery service. A significant increase in antimicrobial resistance was present on cultures taken on the fifth postoperative day when compared with preoperative cultures. Because the patient's skin is the probable source of the bacteria, precise surgical technique with avoidance of contact of prosthetic graft is a good prophylactic measure to attempt to minimize the incidence of these infections. The second factor is that 75% of the patients in this series underwent multiple operative procedures or emergency procedures, indicating that the risk of graft contamination is greater with repeated operative dissection or urgent procedures. Durham et al. 11 noted that positive arterial culture results were more common after secondary vascular procedures and were associated with subsequent graft infections. Of particular interest is the possibility that colonization of the vascular repair may be an etiologic factor in its failure. The concept that bacterial colonization can cause complications such as thrombosis or stenosis is entirely speculative but an intriguing possibility. This speculation requires that the bacteria secrete some compound that would make the flow surface of the graft relatively hypercoagulable. The potential for the development of a fabric prosthesis made resistant to bacterial infection by the bonding of antibiotics to the conduit has been encouraged by animal studies with antibiotic-bonded grafts. 12,13 If the favorable results obtained in animals can be reproduced in man, this potentially could result in a marked decrease in long-term graft infections. It is becoming increasingly apparent that the most important factor in selecting treatment regimens for vascular graft infections is the proper analysis of the patient. The role of the infecting organism is primary in selecting less radical treatment of in situ repair as opposed to the traditional treatment of graft excision and extraanatomic bypass to restore distal flow. The spectrum of pathologic changes caused by different organisms with varying host virulence was demonstrated by Geary et al.,14 who used an animal graft infection model to study the effects of S. epidermidis and Pseudomonas aeruginosa comparing both vein grafts and PTFE grafts. In the S. epidermidis group there was evidence of infection in 80% of animals on histologic elevation but no overt signs of infection. In the P. aeruginosa group all grafts were infected with all vein grafts disrupted (vein and anastomosis) and 60% of PTFE grafts had the anastomosis disrupted. 14
JOURNAL OF VASCULAR SURGERY February 1994
It is critical for the surgeon to identifY patients with biofilm infections during follow-up surveillance. Because of the indolent nature of the infection and the low virulence at the organism, the diagnosis can rarely be made definitely before graft removal and culture. The clinical presentation is typical enough so that a· tentative diagnosis can be reliably made. The principle components of this diagnosis include an infection occurring at a time remote from the previous vascular procedure (at least 4 months) and the absence of systemic evidence of toxicity. The use of CT scanning and duplex scanning is valuable to detect perigraft fluid and the presence of intraabdominal false aneurysms. We do not recommend in situ replacement for gram-negative infection and for coagulase-positive staphylococcus infections where the patient has systemic effects of infection and bacteria can be seen on gram stains of the perigraft fluid. In these situations traditional techniques of graft removal and autogenous or extraanatomic revascularization are still the treatments of choice. REFERENCES 1. TojoM, YamashitaN, Goldmann DA, Pier GB. Isolation and characterization of a capsular polysaccharide adhesion from Staphylococcus epidermidis. I Infect Dis 1988;157:713-22. 2. Christenson GD, Barker LP, Mawhinney TP, Baddour LM, Simpson W A. Identification of an antigenic marker of slime production for Staphylococcus epidermidis. Infect Immun 1990; 58:2906-11. 3. Bandyk DF, Berni GA, Thiele BL, Towne lB. Aortofemoral graft infection due to Staphylococcus epidermidis. Arch Surg 1984;119:102-8. 4. Seabrook GR, Schmitt DD, Bandyk DF, Edmiston CE, Krepel CJ, Towne lB. Anastomotic femoral pseudoaneurysm: an investigation of occult infection as an etiologic factor. I VAse SURG 1990;11:629-34. 5. Bergamini TM, Bandyk DF, Govastis D, Kaebnick HW, Towne lB. Infection of vascular prostheses caused by bacterial biofilms. I VAse SURG 1988;7:21-30. 6. Bandyk DF, Bergamini TM, Kinney EV, Seabrook GR, Towne IB: In situ replacement of vascular prostheses infected by bacterial biofilms. I VAse SURG 1991;13:575-83. 7. Kwaan IHM, Bernstein 1M, Connolly IE. Management of lymph fistula in the groin after arterial reconstruction. Arch Surg 1979;114:1416-8. 8. Schmitt DD, Bandyk DF, Pequet AT, Towne IE. Bacterial adherence to vascular prostheses. I VAse SURG 1986;3:73240. 9. Kleinman LH, Towne JB, Bernard VM. A diagnostic and therapeutic approach to aortoenteric fistulas. Surgery 1979; 86:868-80. 10. Levy MF, Schmidt DD, Edminston CE, Bandyk DF, Seabrook GR, Towne lB. Sequential analysis of staphylococcal colonization of body surfaces of patients undergoing vascular surgery. I Clin MicrobioI1990;28:664-9. 11. Durham IR, Malone JM, Bernhard VM. The impact of
JOURNAL OF VASCULAR SURGERY Volume 19, Number 2
multiple operations on the importance of arterial wall cultures. J VASC SURG 1987;5:160-9. 12. Colburn MD, Moore WS, Chvapil M, Gelabert AA, Quifiones-Baldrich WJ. Use of an antibiotic-bonded graft for in situ reconstruction. J VASC SURG 1992;16:651-8. 13. Kinney EV, Bandyk DF, Seabrook GA, Kelly HM, Towne JB. Antibiotic-bonded silver antibiotic on bioactivity following implantation. J Surg Res 1991;50:430-5.
Towne et at.
233
14. Geary KJ, Tomkiewicz ZM, Harrison lIN, et al. Differential effects of a gram-negative and gram-positive infection on autogenous and prosthetic grafts. J V ASC SURG 1990; 11: 33947.
Submitted June 10, 1993; accepted Oct. 22, 1993.
DISCUSSION Dr. Linda M. Reilly (San Francisco, Calif.). The authors have collected data about 20 patients over 6 years who were believed to be candidates for this method of treatment. What proportion did this represent of all the patients with graft infection treated by your group during this time interval? In our series of graft infections, the rate of single-agent infections was 32%, and of single-agent S. epidermis graft infection about 20% to 25%, which is well less than the 60% referred to in the series. The authors state that it is essential that the patient have the correct diagnosis. But how did the authors do this? How did they identifY these patients before operation when this diagnosis rests on definitive culture results of an explanted segment of graft? What preoperative tests allowed the authors to select patients for or exclude them from this treatment? The authors characterize these infections as ones that occur at least 4 months after the latest vascular procedure and without associated leukocytosis or fever. In our experience, 44% of patients had no systemic signs of graft infection, but our frequency of single agent S. epidermis graft infection was only about 25%. This suggests that the authors' criteria for diagnosis are not specific enough, and I emphasize that this is diagnosis before the time of treatment. In fact, the authors report what is essentially a 40% rate of error in diagnosis, by which I mean that their final culture results were not single agent S. epidermidis. Do you believe that this is a reasonable rate, recognizing that your expertise almost certainly means that this is the best rate that can be achieved? Were there any patients whom the authors excluded incorrectly? Similarly, were there any patients initially treated this way whose subsequent operative culture result showed gram-negative or more virulent gram-positive organisms that then necessitated convention treatment? Were there any consequences of the resultant delay or additional operations? The title of this study suggests that the technique involved is simply removing one graft and inserting another; but in fact, the treatment in this series of patients is really graft excision, vigorous perigraft capsule, and soft tissue excision, graft reimplantation, and muscle flap coverage whenever feasible. Because it has been well described that muscle flaps can salvage infected grafts without removing and replacing them, would you tell us
how many times muscle flap coverage was used and how you controlled for the influence of the muscle flap on their outcome? The authors state, with regard to the 14 aortic graft infections, that the indolent nature of the organism is such that one can expect that all of the infected aortic grafts will ultimately need to be removed if the patient lives long enough. This means that these patients require indefinite close surveillance. In this era of cost-consciousness, do you have any information on the nature of the surveillance that is needed and its cost, in comparison to the cost of conventional treatment of graft infection where close surveillance is usually needed for about 6 to 12 months at most? My principal concern with this study is that it be recognized for what it is - the successful use of a technique in a small but carefully selected subgroup of patients, with low virulence graft infections, many of which were nonaortic grafts. Nonaortic graft infection is, in general, not a life-threatening complication. The 14 aortic grafts treated in this series only involved partial graft limb excision, and with the exception of one case, there were no patients who had the complete aortic graft removed and reimplanted with a Gore-tex graft. * This study should not be interpreted to show that any graft infection can be treated by in situ replacement. The authors state this precaution repeatedly, and it is important that we recognize the limits that they are setting. The simplicity of this approach, however, is very seductive and we must be careful that wishful thinking does not allow this approach to become inappropriately generalized to polymicrobial, gram-negative, or other gram-positive graft infections without the data to justifY it and at the grave risk of once again worsening the outcome of the treatment of prosthetic graft infection. Dr. William D. Turnipseed (Madison, Wis.). We have had similar experience with more aggressive forms of peri graft wound infections and have used the muscle flap as a means of treating this condition. I think all of us would agree that the management of perigraft infections has significantly changed in concept, with an emphasis over the
*Gore-tex is a trademark ofW. L. Gore & Associates, Elkton, Md.
234 T(JWne et at.
past few years on graft preservation when patency and anastomotic integrity persist in the presence of a local infection. I believe improved survival has been documented, as shown by Dr. Towne, and reduced morbidity rates, when this general approach is taken. The great hazard, of course, is either recurrent infection or extended infection in the graft segments that are left in place. The basic issues necessary for using this technique have been brought to light, and those are the aggressive use of soft tissue debridement, circumferential excision of the pseudocapsule, or, in the case of biofilms, the use of in situ replacement, and the circumferential coverage with viable muscle tissue. In our experience, failure to remove enough graft or failure to completely surround the new graft or the debulked graft with viable muscle has been the primary cause for our failures. We tend to be very aggressive about removing the pelvic segment of the graft, and oftentimes the retroperitoneal incision goes all the way to the bifurcation to replace it. We also use the rectus femoris muscle as our graft cover. How do you decide how much of pelvic limb graft to remove? And how do you get adequate coverage with a muscle like the sartorius or the gracilis, which has small bulk, a short arc of rotation, and a segmental blood supply that's very susceptible to occlusion or ischemia when proximal or cephalad rotation is used? Dr. DavidC. Brewster (Boston, Mass.). In view of the apparent indolent nature of many of these infections, would you recommend graft replacement in all instances when graft complications such as anastomotic disruption are, in fact, absent? Second, regarding the absence of firm bacteriologic diagnosis in all instances, and in view of the potential consequences of failure to adequately close to aortic stump with initial graft resection if the proximal graft is involved, this reflects the fact that in many cases that we manage, the surgeon has one good opportunity at secure infrarenal aortic stump closure. In view of these difficulties, do you always recommend in situ replacement if the proximal graft or proximal graft anastomosis is clearly involved? Dr. Frank T. Padberg (East Orange, N.J.). How many centimeters from the last disincorporated site do you insist on when leaving a residual portion of graft in situ? Second, what explanation do you offer for the two graft segments that had no growth despite maximal culture technique? And third, what recommendations do you have for that residual graft stump that returns an unexpectedly positive culture result in the explanted segments several days after your operation? Dr. John J. Ricotta (Buffalo, N.Y.). I'm concerned by the fact that this study began in 1987, so that your follow-up is relatively limited in a situation where you're going to be dealing with indolent infection. Would you
JOURNAL OF VASCULAR SURGERY February 1994
comment on any life-table or projection that you might have about late infections? Dr. Safuh Attar (Baltimore, Md.). As a thoracic surgeon, I have had a lot of experience with combating infections in the mediastinum and in our bronchopleural fistulas. These have been treated successfully with the transposition of muscle flaps into the mediastinum, and over the tracheobronchial anastomosis. The same applies for the infections of the aorta or other blood vessels. The important point in the management, besides the eradication of infection - and I want to interject that it's immaterial whether the infection is gram-positive or gram-negative - is the presence of neovascularity by the muscle that will combat the infection and will provide a long-term success. I believe this is really the crux of success in this operation. Dr. Brian Thiele (Hershey, Pa.). Do you have any information on the patients who died? Was there any evidence of infection by autopsy? Because the second conclusion is that of 10 patients who remained alive, four required reoperation for a continuing problem, Is that correct or incorrect? Dr. Jonathan B. Towne. We have reoperated on one patient for progression of the biofilm infection disease. Three other patients required procedures for progression of lower extremity vascular occlusive disease. In our area, biofilm infections are a common cause of problems. With regard to the correct diagnosis, these patients are afebrile, they are not systemically toxic, and results of blood cultures are negative. If there's any question at operation about the type of infection, the perigraft fluid is sent for gram staining, which should demonstrate white blood cells and no bacteria. I urge you, if there is any question of diagnosis, use traditional techniques of graft resection and extraanatomic revascularization. Weare concerned about using this technique on the abdominal aorta, because of the life-threatening complications of aortic disruption and aortoenteric fistulas. I've put muscle flaps on Pseudomonas infections and had the arterial repair disrupt 3 days later, demonstrating that muscle flaps are not always curative. We use the muscle flap in the groin to cover the femoral vessels, because by the time you totally excise the perigraft capsule, there is no good tissue to cover the repair. I think it is probably the vascularity of the muscle more than anything else that is helpful. We use the sartorius primarily because it is easier to use than the gracilis. If we use a rectus flap, we involve a plastic surgeon. One technique that we have learned is to take the sartorius off the anterior-superior iliac spine and roll it over, because its blood supply comes in from the deep femoral artery, and if you mobilize that medial border, you'll devascularize it. Any series studying graft infection shows high mortality rates and high limb loss rates. For surveillance, we use duplex imaging.
JOURNAL OF VASCULAR SURGERY Volume 19, Number 2
With regard to aortic involvement, if it isn't clearly a biofilm infection (e.g., the aorta is well incorporated but the graft isn't) we will excise the graft and do an extraanatomic bypass. Dr. Padberg, the reason that nothing grew in two of these patients is that gram-negative staphylococci are sensitive to antibiotics. With regard
Towne et al.
235
to these patients in whom nothing grew, we have all the secondary evidence that they were graft biofilm infections. There are only 10 patients who are alive an average of 44 months. We do not believe this is a large enough group to construct a life-table.
Purpose: Bacterial biofilm infections of vascular prostheses are characterized by an absence of systemic sepsis, a fluid-filled cavity surrounding the graft, a draining sinus tract, and microorganisms that must be removed from the fabric prosthesis for bacterial culture. Methods: Since 1987 we have treated 20 infected grafts with prosthetic excision and in situ replacement in 14 men and 6 women. The time from initial graft implantation to diagnosis of graft infection ranged from 3 months to 14 years (mean 4.5 years). The original graft (Dacron-17, polytetrafluoroethylene-3) was an aortobifemoral in 14, axillofemoral femorofemoral in 3, iliofemoral in 2, and femoropopliteal in 1 patient. Presenting symptoms were groin false aneurysm with perigraft fluid in 10, inflammatory mass in 6, and sinus tract in 4. At surgery all unincorporated graft material and the perigraft capsule were excised from a point where the proximal graft was incorporated, including debridement of vessels at the distal anastomosis. Of the 14 aortobifemoral grafts, only the femoral limbs were excised at the initial presentation ofbiofilm infection. The conduit was replaced with an in situ polytetrafluoroethylene interposition graft, which was covered with a gracilis or sartorius muscle flap when possible. Results: All surgical sites healed, all grafts remained patent, and there was no limb loss. After ultrasonic oscillation of the explanted graft, bacterial cultures recovered coagulasenegative Staphylococcus species in 14, coagulase-positive Staphylococcus species in one, both species in three, with no growth from two specimens. During follow-up, two patients have had clinical involvement in the proximal intraabdominal portion of the graft that had not been previously resected. In all grafts, the in situ replacement graft remained well incorporated. Conclusion: In situ graft replacement is effective treatment for biofilm infections of vascular prostheses. Because of the indolent nature of these infections, subsequent infection of previously uninvolved graft segments may be expected. (J VAse SURG 1994;19:226-35.)
Coagulase-negative staphylococci are a significant cause of bioprosthetic infections, because of their unique ability to adhere to and grow on implanted artificial surfaces and to secrete a biofilm or slime, which is a galactose-rich polysaccharide that provides protection from the body's host defenses. 1,2 Bandyk et aI., in a la-year review of aortofemoral graft infections, determined that Staphylococcus epidennidis From the Department of Vascular Surgery at the Medical College of Wisconsin, Milwaukee. Presented at the Forty-seventh Annual Meeting of the Society for Vascular Surgery, Washington, D.C., June 8-9, 1993. Reprint requests: Jonathan ·B. Towne, MD, Department of Vascular Surgery, 8700 W. Wisconsin Ave., Milwaukee, WI 53226. Copyright © 1994 by The Society for Vascular Surgery and International Society for Cardiovascular Surgery, North American Chapter. 0741-5214/94/$3.00 + 0 24{6{52342
226
was the infecting organism in 60% of the cases and occurred a mean of 41 months after the initial operation. 3 In a subsequent prospective study of anastomotic femoral false aneurysm after aortofemoral bypass grafting, Seabrook et al. 4 reported that 60% of anastomotic femoral false aneurysms presenting a mean of 95 months after graft implantation were infected with coagulase-negative staphylococci. Usually, vascular prostheses infected with coagulasenegative staphylococci are associated with a less virulent clinical course compared with vascular graft infections caused by coagulase-positive staphylococci or gram-negative bacteria. Infections are not associated with fever, leukocytosis, or positive blood culture results. Graft biofilm infections usually present as a graft healing complication such as absence of tissue incorporation with or without perigraft cavities and perigraft fluid, false aneurysms,
JOURNAL OF VASCULAR SURGERY Volume 19, Number 2
or graft-cutaneous sinuses. Because the bacteria are adherent to the prosthesis, they are seldom present in the periprosthetic tissue fluid. Microscopic examination of this perigraft fluid demonstrates multiple white blood cells and only rarely are bacteria seen. Because culture results of this perigraft fluid are routinely negative, for many years investigators believed that these grafts were not infected and that the lack of graft incorporation and the development of perigraft fluid was an "allergic response" to the vascular prosthesis. However, bacteria can be recovered if they are mechanically dislodged from the fabric of the graft or if the graft material itself is placed in culture media. Because the infection is confined to the prosthetic vascular conduit, the potential for graft excision and in situ replacement is possible. To evaluate this technique, a canine model was developed to reproduce all of the findings classically seen in a graft biofilm infection: lack of graft incorporation by perigraft tissue, perigraft fluid, and the ability to culture the infecting organism only from the graft fabric. 5 With this canine model the established prosthetic surface biofilm infection with S. epidermidis was treated by graft excision, parenteral cefazolin, and in situ graft replacement with a PTFE vascular prosthesis. There was normal perigraft healing, including anastomotic tensile strength 1 month after in situ replacement. Because of these laboratory findings, in 1987 we cautiously began a clinical trial of in situ replacement in patients with graft biofilm infections. An initial report confirmed low morbidity and early eradication of the biofilm infection. 6 This study presents the long-term results of this clinical series with emphasis on the outcome of both replaced and previously uninvolved graft segments. METHODS Patients. Since 1987, 20 patients with biofilm graft infections were treated with graft excision and in situ replacement. There were 14 men and six women. The time from initial graft implantation to diagnosis of infection ranges from 3 months to 14 years, with a mean of 4.5 years. Initial indications for primary grafting procedure were atherosclerosis obliterans in 16, ruptured abdominal aortic aneurysm in three, and elective aneurysm in one. The original graft was knitted Dacron in 17 and polytetrafluoroethylene (PTFE) in three and was placed initially as an aortobifemoral graft in 14, axillofemoral, femorofemoral graft in three, an iliofemoral in two, and femoropopliteal graft in one patient. Presenting symptoms of graft infection included
Towne et al.
227
groin false aneurysm with perigraft fluid in 10, inflammatory groin mass in six, and sinus tract in four. In 12 patients the graft segment involved by the suspected bacterial biofilm infection had undergone one or multiple secondary procedures, and three patients had the initial aortofemoral graft placed for ruptured abdominal aortic aneurysm. Only five patients enrolled in the series with late graft infection had undergone a single, elective, arterial reconstruction. All 20 patients were diagnosed with the clinical criteria of graft biofilm infection, which included presentation months to years after graft implantation, lack of systemic signs of infection (afebrile, no leukocytosis), and no growth on blood cultures. The perigraft tissue demonstrated only inflammation with no bacteria seen on microscopic examination. There was a perigraft cavity with fluid with an absence of graft incorporation often accompanied by an anastomotic pseudoaneurysm. The perigraft fluid gram stain showed white blood cells with no bacteria and the perigraft fluid culture showed no growth. There was no evidence of an aortic false aneurysm in any of the patients with involvement of a limb of an aortofemoral graft. Microbiologic technique. The microbiologic technique has been previously described. s,6 In summary, cultures of the perigraft tissue fluid were obtained by sterile cotton swabs and were plated on agar media. At operation 1 cm length of prosthetic material adjacent to the femoral anastomosis or within the perigraft cavity was excised and placed directly in 20 ml of glucose supplemented trypticase soy broth (Becton Dickinson Microbiology Systems, Cockeysville, Md.). The explanted graft specimen submitted for culture did not include thrombus, perigraft capsule or adjacent artery. A second graft segment of similar size was placed in trypticase soy broth and ultrasonically oscillated at 20 kHz for 10 minutes to mechanically disrupt the graft surface biofilm and disperse adherent bacteria into the broth media for growth. The broth and biofilm culture of graft material were incubated at 35° C for up to 2 weeks. Organisms were isolated and identified after standard microbiologic technique. Operative technique. Treatment of aortofemoral graft abnormalities in the groin involved vascular isolation at the femoral anastomosis. Usually proximal control was not possible in the groin necessitating a suprainguinal counter incision with retroperitoneal exposure of the more proximal limb of the aortofemoral graft. The replacement limb was sutured end to end to the incorporated proximal limb
228
Towne et al.
JOURNAL OF VASCULAR SURGERY February 1994
Fig. 1. Color-flow duplex examination demonstrates well-incorporated PTFE graft.
and to the debrided femoral vessels. All unincorporated graft material and the perigraft capsule were excised. The artery adjacent to the suture line was debrided to a normal-appearing vessel. Because the scar tissue from the previous procedure was usually intimately adherent to the perigraft capsule, this was all excised leaving only vital structures. Graft cutaneous sinus tracts were excised en bloc. The conduit was replaced in situ with a PTFE interposition graft. The replacement graft was usually covered with a gracilis or sartorius muscle flap, although on occasion, a rectus abdominal muscle flap was used. When distal grafts were involved, adjacent muscle was mobilized to cover grafts. Replacement of the entire aortofemoral graft was not required at the initial treatment in any patient. Eight of 14 aortofemoral graft limbs had sartorius or gracilis myoplasty to cover the revised distal anastomosis. In patients with infected axillofemoral grafts, the new graft was tunneled adjacent to, but separate from, the previous graft. The skin was closed primarily in all cases. Perioperative antibiotics, cefazolin or vancomycin, were administered to all patients and were continued parenterally for 7 to 10 days. Oral antibiotics were continued for 1 month after operation. Patients were monitored every 3 months with color-flow Duplex or abdominal computed tomography (Cf) scanning. The absence of
wound problems and demonstration of a wellincorporated graft by CT and ultrasonography were the criteria for absence of residual infection (Figs. 1 and 2) . RESULTS All patients recovered without signs of graft or wound infection after the in situ grafting procedure. There was no perioperative limb loss, graft thrombosis, or death. Three patients had a lymphocele after operation that was separate from the vascular repair. In two of these, wound drainage developed that was treated with operative exploration, ligation of lymphatic vessels, and primary closure over a closed suction drain. In a third patient, the lymphocele gradually was reabsorbed over 6 months. Bacterial cultures recovered only coagulasenegative staphylococci from 14 patients, coagulasepositive staphylococci from 1, and both species from 3, and no growth in 2. Of the four patients who grew coagulase-positive staphylococci, none grew it from the graft capsule. One of these patients grew a coagulase-positive staphylococci in one groin and coagulase-negative staphylococci in the other. Patients in whom the procedure consisted of resection of a femoral limb of an aortal bifemoral graft were monitored with CT scanning. Patients with lower leg grafts were monitored with duplex scanning of the
JOURNAL OF VASCULAR SURGERY Volwnc 19, Number 2
Timme et aI.
229
Fig. 2. Color-flow duplex examination demonstrates fluid surrounding graft, indicating biofilm infection.
graft looking for the development of fluid around the graft. Ten of the 20 patients died in the follow-up period for a long-term mortality rate of 50%. The average follow-up from graft replacement to death was 32 months and ranged from 3 months to 74 months. The causes of death include cardiac events in five, carcinoma of the lung in three, and stroke and kidney failure in one patient each. No patient died of treatment of graft failure or graft infection complication. Follow-up of these patients ranged from 3 months to 74 months for the entire series (average 39 months). When the patients who died were deleted, the follow-up of the remaining 10 living patients
ranged from 10 months to 67 months (average 45 months). Further graft complications occurred in four patients in the follow-up period. Two patients with aortofemoral grafts had involvement in the proximal intraabdominal portion of the graft that had not been previously resected and had not previously had clinical evidence of infection. This manifested as fluid around the limbs of the graft and, as in the initial presentation, was not associated with any systemic evidence of infection (Fig. 3). One patient had a resection of the residual graft 2 years after in situ replacement of groin biofilm infection and represents the only aortic in situ replacement. At operation, there were classic findings of a graft biofilm infection
230 Tuwne et at.
JOURNAL OF VASCULAR SURGERY February 1994
Fig. 3. Abdominal CT scan with contrast demonstrates fluid around intraabdominal previously unrejected segment of aortobifemoral graft.
with lack of incorporation of the graft and large fluid filled perigraft cavities which extended to the proximal anastomosis. There was no aortic false aneurysm. A perigraft fluid cavity developed around the sup rainguinal portion of an aortofemoral graft in the second patient 2 years after resection and in situ replacement for femoral false aneurysm associated with absence of graft incorporation. The groin repair was well incorporated, and the patient had an interposition graft placed from the proximal groin graft to the proximal iliac limb 6 cm distal to its bifurcation. Fluid has subsequently developed around the remaining residual intraabdominal graft. Because of renal and cardiac complications this patient was not a candidate for aortic limb replacement. She died 74 months after femoral graft resection and replacement. There were two major amputations in the follow-up period. One patient had an in situ femoral limb replacement that healed. A femoral posterior tibial PTFE graft was required 10 months later because of progression of lower extremity occlusive disease. This graft had multiple episodes of thrombosis requiring thrombectomy. The graft eventually became grossly infected with coagulase-positive staphylococci, requiring removal and resulting in an above-knee amputation. The PTFE interposition graft replaced in situ in the groin remains well incorporated without evidence of a perigraft fluid
cavity 2 years after amputation. The second patient had biofilm infection of the femoral limb of an iliofemoral graft. Also involved was the proximal anastomosis of a femoral above-knee popliteal PTFE graft. Arteriograms demonstrated a stenosis of the popliteal artery just beyond the distal anastomosis of the femoral popliteal graft. Mter in situ replacement of the femoral graft limb and wound healing, the patient refused an operative procedure to correct the outflow stenosis. The femoropopliteal graft thrombosed 1 year later, resulting in an above-knee amputation. Including the two patients requiring amputations described above, there were three graft occlusions in the postoperative period. The third patient had an in situ replacement for a biofilm infection of the left groin. One month after operation he required a femoral anterior tibial PTFE bypass because of persistent ischemia of the left foot. The culture at the time of graft replacement grew coagulase-positive staphylococci. Perigraft fluid subsequently developed around his tibial graft, whereas the femoral graft remained well healed. The distal PTFE graft thrombosed at 21/2 years and on removal grew both coagulase-positive and coagulase-negative staphylococci. The in situ groin repair remains healed without evidence of perigraft fluid. To date, none of the prostheses used for in situ
JOURNAL OF VASCULAR SURGERY Volume 19, Number 2
replacement for biofilm infection have shown evidence of infection as manifested by false aneurysm formation, perigraft fluid or cavity, and cutaneous sinus tracts. The two graft infections occurring in distal leg prosthetic grafts did not involve the segment of in situ replacement in the groin. In both of these patients, the muscle flap may have protected the femoral segment from the infection involving the distal graft.
DISCUSSION The low virulence of coagulase-negative staphylococci and their unique characteristics of colonization via a surface biofilm permit treatment of infected grafts by excision of the grossly involved graft segments, debridement of perigraft tissue and adjacent artery, and in situ replacement of another prosthesis. Several caveats must be emphasized in selecting this treatment option. First, it is essential that the patient has the correct diagnosis. This technique is not appropriate or recommended for cases other than biofilm infections (i.e., patients who have gram-negative bacteria infections). The role of coagulase-positive staphylococci in these infections is not yet defined. In latent, indolent infections where coagulase-positive staphylococci are cultured, in situ replacement may be successful. In this series, four such patients did not have systemic evidence of infection, and they did not grow Staphylococcus aureus from the perigraft cavity. Why some patients with coagulase-positive staphylococcal infections fail to mount the systemic reaction that is more classically associated with this infection is not understood. There may be a spectrum of virulence in these species in which some isolates act like the more benign coagulase-negative staphylococci. The key to the operative treatment of these patients is total excision of the biofilm cavity and the involved graft. Because this often results in an extensive groin dissection, there is an increased incidence of lymphatic wound complications, particularly the occurrence of lymphoceles in the postoperative period. The use of muscle coverage is important because it provides vascularized tissue to surround the graft and be interposed between the skin and the conduit if wound complications occur. The treatment of lymphoceles should be aggressive, with operative ligation and drainage if they do not promptly resolve or begin to leak through the wound creating a potential tract for secondary infection. 7 The choice of a PTFE as a replacement graft was selected because our laboratory data demonstrated
Towne et at.
231
that the bacterial adherence of slime-producing bacteria was significantly less to PTFE than to knitted or woven Dacron. 8 Slime-producing coagulasenegative staphylococci adhered to the knitted Dacron 100 times greater than to PTFE. Because bacterial adhesion is the important first step in biofilm graft infection, the relative resistance to bacterial adhesion possessed by PTFE has significant potential advantages as an in situ replacement conduit. A related facet of this study is the profile of the patients who were enrolled. Because biofilm infections occur months and years after an initial vascular reconstructive procedure, the mortality rate from associated cardiovascular disease is high. The mortality rate in this series, with follow-up to 74 months, is 50%, most of it caused by cardiovascular causes. Because of this high mortality rate we are curtailed in studying extended durability of this repair technique. However, if one accepts the thesis that the role of these procedures is to allow the patients to continue their lives with both limbs for as long as possible, that goal is accomplished. A special note of caution needs to be raised for patients who have femoral limbs of their aortal bifemoral grafts resected. In all of these patients in this series, the replacement grafts have healed well and have demonstrated no evidence of recurrent infection. However, the residual aortic graft segment in the patient may subsequently develop changes suggestive of coagulase-negative staphylococcus infection manifested by the development of a perigraft cavity and fluid. This, in all likelihood, is the clinical presentation of a latent infection that existed at the time of the initial resection. It is safe to expect that eventually the total graft will need to be excised if the patient lives long enough. At this point, we have done an intraabdominal resection on only one of these grafts, and the follow-up is now 6 months. Certainly, replacement of the intraabdominal portion of the graft is fraught with the greatest potential complications, because of the risk for development of a graft enteric fistula or false aneurysm with rupture. 9 Given these potential catastrophes, we have been very cautious in applying this technique for the aortic segment. In the absence of long-term evaluation of this technique at the aortic level, infrarenal aortic ligation and extraanatomic bypass must still be considered the standard of care for an aortic graft infection. The source of contamination of graft biofilms infections may be twofold. Slime-producing coagulase-negative staphylococci are often present on the skin at time of hospitalization. Levy et al. 10
232 Tuwne et at.
identified at least one slime-producing staphylococcal species as part of skin flora in 80% of 41 consecutive patients admitted to the vascular surgery service. A significant increase in antimicrobial resistance was present on cultures taken on the fifth postoperative day when compared with preoperative cultures. Because the patient's skin is the probable source of the bacteria, precise surgical technique with avoidance of contact of prosthetic graft is a good prophylactic measure to attempt to minimize the incidence of these infections. The second factor is that 75% of the patients in this series underwent multiple operative procedures or emergency procedures, indicating that the risk of graft contamination is greater with repeated operative dissection or urgent procedures. Durham et al. 11 noted that positive arterial culture results were more common after secondary vascular procedures and were associated with subsequent graft infections. Of particular interest is the possibility that colonization of the vascular repair may be an etiologic factor in its failure. The concept that bacterial colonization can cause complications such as thrombosis or stenosis is entirely speculative but an intriguing possibility. This speculation requires that the bacteria secrete some compound that would make the flow surface of the graft relatively hypercoagulable. The potential for the development of a fabric prosthesis made resistant to bacterial infection by the bonding of antibiotics to the conduit has been encouraged by animal studies with antibiotic-bonded grafts. 12,13 If the favorable results obtained in animals can be reproduced in man, this potentially could result in a marked decrease in long-term graft infections. It is becoming increasingly apparent that the most important factor in selecting treatment regimens for vascular graft infections is the proper analysis of the patient. The role of the infecting organism is primary in selecting less radical treatment of in situ repair as opposed to the traditional treatment of graft excision and extraanatomic bypass to restore distal flow. The spectrum of pathologic changes caused by different organisms with varying host virulence was demonstrated by Geary et al.,14 who used an animal graft infection model to study the effects of S. epidermidis and Pseudomonas aeruginosa comparing both vein grafts and PTFE grafts. In the S. epidermidis group there was evidence of infection in 80% of animals on histologic elevation but no overt signs of infection. In the P. aeruginosa group all grafts were infected with all vein grafts disrupted (vein and anastomosis) and 60% of PTFE grafts had the anastomosis disrupted. 14
JOURNAL OF VASCULAR SURGERY February 1994
It is critical for the surgeon to identifY patients with biofilm infections during follow-up surveillance. Because of the indolent nature of the infection and the low virulence at the organism, the diagnosis can rarely be made definitely before graft removal and culture. The clinical presentation is typical enough so that a· tentative diagnosis can be reliably made. The principle components of this diagnosis include an infection occurring at a time remote from the previous vascular procedure (at least 4 months) and the absence of systemic evidence of toxicity. The use of CT scanning and duplex scanning is valuable to detect perigraft fluid and the presence of intraabdominal false aneurysms. We do not recommend in situ replacement for gram-negative infection and for coagulase-positive staphylococcus infections where the patient has systemic effects of infection and bacteria can be seen on gram stains of the perigraft fluid. In these situations traditional techniques of graft removal and autogenous or extraanatomic revascularization are still the treatments of choice. REFERENCES 1. TojoM, YamashitaN, Goldmann DA, Pier GB. Isolation and characterization of a capsular polysaccharide adhesion from Staphylococcus epidermidis. I Infect Dis 1988;157:713-22. 2. Christenson GD, Barker LP, Mawhinney TP, Baddour LM, Simpson W A. Identification of an antigenic marker of slime production for Staphylococcus epidermidis. Infect Immun 1990; 58:2906-11. 3. Bandyk DF, Berni GA, Thiele BL, Towne lB. Aortofemoral graft infection due to Staphylococcus epidermidis. Arch Surg 1984;119:102-8. 4. Seabrook GR, Schmitt DD, Bandyk DF, Edmiston CE, Krepel CJ, Towne lB. Anastomotic femoral pseudoaneurysm: an investigation of occult infection as an etiologic factor. I VAse SURG 1990;11:629-34. 5. Bergamini TM, Bandyk DF, Govastis D, Kaebnick HW, Towne lB. Infection of vascular prostheses caused by bacterial biofilms. I VAse SURG 1988;7:21-30. 6. Bandyk DF, Bergamini TM, Kinney EV, Seabrook GR, Towne IB: In situ replacement of vascular prostheses infected by bacterial biofilms. I VAse SURG 1991;13:575-83. 7. Kwaan IHM, Bernstein 1M, Connolly IE. Management of lymph fistula in the groin after arterial reconstruction. Arch Surg 1979;114:1416-8. 8. Schmitt DD, Bandyk DF, Pequet AT, Towne IE. Bacterial adherence to vascular prostheses. I VAse SURG 1986;3:73240. 9. Kleinman LH, Towne JB, Bernard VM. A diagnostic and therapeutic approach to aortoenteric fistulas. Surgery 1979; 86:868-80. 10. Levy MF, Schmidt DD, Edminston CE, Bandyk DF, Seabrook GR, Towne lB. Sequential analysis of staphylococcal colonization of body surfaces of patients undergoing vascular surgery. I Clin MicrobioI1990;28:664-9. 11. Durham IR, Malone JM, Bernhard VM. The impact of
JOURNAL OF VASCULAR SURGERY Volume 19, Number 2
multiple operations on the importance of arterial wall cultures. J VASC SURG 1987;5:160-9. 12. Colburn MD, Moore WS, Chvapil M, Gelabert AA, Quifiones-Baldrich WJ. Use of an antibiotic-bonded graft for in situ reconstruction. J VASC SURG 1992;16:651-8. 13. Kinney EV, Bandyk DF, Seabrook GA, Kelly HM, Towne JB. Antibiotic-bonded silver antibiotic on bioactivity following implantation. J Surg Res 1991;50:430-5.
Towne et at.
233
14. Geary KJ, Tomkiewicz ZM, Harrison lIN, et al. Differential effects of a gram-negative and gram-positive infection on autogenous and prosthetic grafts. J V ASC SURG 1990; 11: 33947.
Submitted June 10, 1993; accepted Oct. 22, 1993.
DISCUSSION Dr. Linda M. Reilly (San Francisco, Calif.). The authors have collected data about 20 patients over 6 years who were believed to be candidates for this method of treatment. What proportion did this represent of all the patients with graft infection treated by your group during this time interval? In our series of graft infections, the rate of single-agent infections was 32%, and of single-agent S. epidermis graft infection about 20% to 25%, which is well less than the 60% referred to in the series. The authors state that it is essential that the patient have the correct diagnosis. But how did the authors do this? How did they identifY these patients before operation when this diagnosis rests on definitive culture results of an explanted segment of graft? What preoperative tests allowed the authors to select patients for or exclude them from this treatment? The authors characterize these infections as ones that occur at least 4 months after the latest vascular procedure and without associated leukocytosis or fever. In our experience, 44% of patients had no systemic signs of graft infection, but our frequency of single agent S. epidermis graft infection was only about 25%. This suggests that the authors' criteria for diagnosis are not specific enough, and I emphasize that this is diagnosis before the time of treatment. In fact, the authors report what is essentially a 40% rate of error in diagnosis, by which I mean that their final culture results were not single agent S. epidermidis. Do you believe that this is a reasonable rate, recognizing that your expertise almost certainly means that this is the best rate that can be achieved? Were there any patients whom the authors excluded incorrectly? Similarly, were there any patients initially treated this way whose subsequent operative culture result showed gram-negative or more virulent gram-positive organisms that then necessitated convention treatment? Were there any consequences of the resultant delay or additional operations? The title of this study suggests that the technique involved is simply removing one graft and inserting another; but in fact, the treatment in this series of patients is really graft excision, vigorous perigraft capsule, and soft tissue excision, graft reimplantation, and muscle flap coverage whenever feasible. Because it has been well described that muscle flaps can salvage infected grafts without removing and replacing them, would you tell us
how many times muscle flap coverage was used and how you controlled for the influence of the muscle flap on their outcome? The authors state, with regard to the 14 aortic graft infections, that the indolent nature of the organism is such that one can expect that all of the infected aortic grafts will ultimately need to be removed if the patient lives long enough. This means that these patients require indefinite close surveillance. In this era of cost-consciousness, do you have any information on the nature of the surveillance that is needed and its cost, in comparison to the cost of conventional treatment of graft infection where close surveillance is usually needed for about 6 to 12 months at most? My principal concern with this study is that it be recognized for what it is - the successful use of a technique in a small but carefully selected subgroup of patients, with low virulence graft infections, many of which were nonaortic grafts. Nonaortic graft infection is, in general, not a life-threatening complication. The 14 aortic grafts treated in this series only involved partial graft limb excision, and with the exception of one case, there were no patients who had the complete aortic graft removed and reimplanted with a Gore-tex graft. * This study should not be interpreted to show that any graft infection can be treated by in situ replacement. The authors state this precaution repeatedly, and it is important that we recognize the limits that they are setting. The simplicity of this approach, however, is very seductive and we must be careful that wishful thinking does not allow this approach to become inappropriately generalized to polymicrobial, gram-negative, or other gram-positive graft infections without the data to justifY it and at the grave risk of once again worsening the outcome of the treatment of prosthetic graft infection. Dr. William D. Turnipseed (Madison, Wis.). We have had similar experience with more aggressive forms of peri graft wound infections and have used the muscle flap as a means of treating this condition. I think all of us would agree that the management of perigraft infections has significantly changed in concept, with an emphasis over the
*Gore-tex is a trademark ofW. L. Gore & Associates, Elkton, Md.
234 T(JWne et at.
past few years on graft preservation when patency and anastomotic integrity persist in the presence of a local infection. I believe improved survival has been documented, as shown by Dr. Towne, and reduced morbidity rates, when this general approach is taken. The great hazard, of course, is either recurrent infection or extended infection in the graft segments that are left in place. The basic issues necessary for using this technique have been brought to light, and those are the aggressive use of soft tissue debridement, circumferential excision of the pseudocapsule, or, in the case of biofilms, the use of in situ replacement, and the circumferential coverage with viable muscle tissue. In our experience, failure to remove enough graft or failure to completely surround the new graft or the debulked graft with viable muscle has been the primary cause for our failures. We tend to be very aggressive about removing the pelvic segment of the graft, and oftentimes the retroperitoneal incision goes all the way to the bifurcation to replace it. We also use the rectus femoris muscle as our graft cover. How do you decide how much of pelvic limb graft to remove? And how do you get adequate coverage with a muscle like the sartorius or the gracilis, which has small bulk, a short arc of rotation, and a segmental blood supply that's very susceptible to occlusion or ischemia when proximal or cephalad rotation is used? Dr. DavidC. Brewster (Boston, Mass.). In view of the apparent indolent nature of many of these infections, would you recommend graft replacement in all instances when graft complications such as anastomotic disruption are, in fact, absent? Second, regarding the absence of firm bacteriologic diagnosis in all instances, and in view of the potential consequences of failure to adequately close to aortic stump with initial graft resection if the proximal graft is involved, this reflects the fact that in many cases that we manage, the surgeon has one good opportunity at secure infrarenal aortic stump closure. In view of these difficulties, do you always recommend in situ replacement if the proximal graft or proximal graft anastomosis is clearly involved? Dr. Frank T. Padberg (East Orange, N.J.). How many centimeters from the last disincorporated site do you insist on when leaving a residual portion of graft in situ? Second, what explanation do you offer for the two graft segments that had no growth despite maximal culture technique? And third, what recommendations do you have for that residual graft stump that returns an unexpectedly positive culture result in the explanted segments several days after your operation? Dr. John J. Ricotta (Buffalo, N.Y.). I'm concerned by the fact that this study began in 1987, so that your follow-up is relatively limited in a situation where you're going to be dealing with indolent infection. Would you
JOURNAL OF VASCULAR SURGERY February 1994
comment on any life-table or projection that you might have about late infections? Dr. Safuh Attar (Baltimore, Md.). As a thoracic surgeon, I have had a lot of experience with combating infections in the mediastinum and in our bronchopleural fistulas. These have been treated successfully with the transposition of muscle flaps into the mediastinum, and over the tracheobronchial anastomosis. The same applies for the infections of the aorta or other blood vessels. The important point in the management, besides the eradication of infection - and I want to interject that it's immaterial whether the infection is gram-positive or gram-negative - is the presence of neovascularity by the muscle that will combat the infection and will provide a long-term success. I believe this is really the crux of success in this operation. Dr. Brian Thiele (Hershey, Pa.). Do you have any information on the patients who died? Was there any evidence of infection by autopsy? Because the second conclusion is that of 10 patients who remained alive, four required reoperation for a continuing problem, Is that correct or incorrect? Dr. Jonathan B. Towne. We have reoperated on one patient for progression of the biofilm infection disease. Three other patients required procedures for progression of lower extremity vascular occlusive disease. In our area, biofilm infections are a common cause of problems. With regard to the correct diagnosis, these patients are afebrile, they are not systemically toxic, and results of blood cultures are negative. If there's any question at operation about the type of infection, the perigraft fluid is sent for gram staining, which should demonstrate white blood cells and no bacteria. I urge you, if there is any question of diagnosis, use traditional techniques of graft resection and extraanatomic revascularization. Weare concerned about using this technique on the abdominal aorta, because of the life-threatening complications of aortic disruption and aortoenteric fistulas. I've put muscle flaps on Pseudomonas infections and had the arterial repair disrupt 3 days later, demonstrating that muscle flaps are not always curative. We use the muscle flap in the groin to cover the femoral vessels, because by the time you totally excise the perigraft capsule, there is no good tissue to cover the repair. I think it is probably the vascularity of the muscle more than anything else that is helpful. We use the sartorius primarily because it is easier to use than the gracilis. If we use a rectus flap, we involve a plastic surgeon. One technique that we have learned is to take the sartorius off the anterior-superior iliac spine and roll it over, because its blood supply comes in from the deep femoral artery, and if you mobilize that medial border, you'll devascularize it. Any series studying graft infection shows high mortality rates and high limb loss rates. For surveillance, we use duplex imaging.
JOURNAL OF VASCULAR SURGERY Volume 19, Number 2
With regard to aortic involvement, if it isn't clearly a biofilm infection (e.g., the aorta is well incorporated but the graft isn't) we will excise the graft and do an extraanatomic bypass. Dr. Padberg, the reason that nothing grew in two of these patients is that gram-negative staphylococci are sensitive to antibiotics. With regard
Towne et al.
235
to these patients in whom nothing grew, we have all the secondary evidence that they were graft biofilm infections. There are only 10 patients who are alive an average of 44 months. We do not believe this is a large enough group to construct a life-table.