- Email: [email protected]
In situ replacement of vascular prostheses infected by bacterial biofilms
In situ replacement of vascular prostheses infected by bacterial biofilms Dennis F. Bandyk, MD, Thomas M. Bergamini, MD, Edward V. Kinney, MD, Gary R. Seabrook, MD, and Jonathan B. Towne, MD, Milwaukee, Wis. Late prosthetic graft infections are commonly the result of coagulase-negative staphylococci that survive within a biofilm on prosthetic surfaces and provoke perigraft inflammation. The indolent nature and microbiologic characteristics of bacterial biofilm infections coupled with the morbidity of graft excision and extraanatomic bypass grafting prompted us to use in situ graft replacement in 15 patients admitted to the hospital with 17 infected graft segments at a mean (--.SEM) time interval of 70 -4--16 months after graft implantation (n = 6) or revision (n = 9). Since 1986, 17 grafts (14 aortofemoral, 2 axillofemoral, and 1 femoropopliteal) infected by bacterial biofilms have been treated. Signs on admission included femoral pseudoaneurysm (n = 7), perigraft abscess (n = 6), or graft-cutaneous sinus tract (n = 4). N o patient exhibited septicemia. At operation graft incorporation was absent and Gram's stain of perigraft exudate showed polymorphonuclear leukocytes but no bacteria. Culture of explanted graft material isolated coagulasenegative staphylococci (n = 12), Staphylococcus aureus (n = 1), and no growth (n = 2). All patients were successfully treated by a regimen that included parenteral antibiotics, removal of involved graft material, excision of inflamed perigraft tissue, and in situ replacement with an expanded polytetrafluoroethylene prosthesis. No deaths, graft thromboses, or deep wound infections occurred after operation. Recurrent graft infection did not develop during a follow-up interval that ranged from 5 to 50 months (mean, 21 months). Diagnosis of vascular prosthesis infection caused by bacterial biofilms can be based on signs at admission and operative findings. Complications of this perigraft infection can be eradicated by antibiotic administration, local debridement, and in situ graft replacement. (I VASC SuRG 1991;13:575-83.)
Surgeons are well aware of the importance of
Staphylococcus aureus and gram-negative bacteria as pathogens in vascular prostheses infections, but only in recent years has the role of coagulase-negative staphylococci (CNS) been emphasized in the pathogenesis of late-appearing graft infections. I"2 The anatomic and microbiologic features of late graft infections differ from "classic graft sepsis" unless there is involvement of the gastrointestinal tract. Bacteria colonization is typically confined to prosthetic surfaces, systemic signs of infection are absent, and the number and virulence of the infecting organisms are low. Although several CNS strains have demonstrated the ability to infect implanted prosthetic devices in man, Staphylococcus epidermidis From the Department of Surgery,MedicalCollegeof Wisconsin, and the Surgical Service,Clement J ZablockiVeterans Affairs Medical Center, Milwaukee. Presented at the Fourteenth Annual Meeting of the Midwestern Vascular Surgical Society,Toledo, Ohio, Sept. 14-15, 1990. Reprint requests: Dennis Bandyk,MD, Department of Surgery, MCMC, 8700 W. WisconsinAve., Milwaukee,WI 53226. 24/6/27762
has been, by far, the most common strain implicated in late graft infections. Infection of a vascular prosthesis by S. epidermidis tends to manifest months to years after graft implantation or revision as a graft-healing complication (anastomotic false aneurysm, perigraft cavity abscess, graft-enteric fistula, or graft-cutaneous sinus tract). Management of these graft-healing complications can pose a difficult diagnostic and therapeutic dilemma for the surgeon. Despite evidence by preoperative imaging studies (ultrasonography, CT scanning) of perigraft inflammation, and at operation of purulence surrounding the prosthesis, Gram's stain or culture of the perigrat~ exudate or tissue may not confirm an infectious process. As such, the surgeon may be reluctant to proceed with graft excision and extraanatomic bypass grafting, and instead opt to procrastinate and treat the suspected perigraft infection with antibiotics until more obvious signs of graft sepsis develops. Our group has previously detailed important pathobiologic aspects of graft infections caused by S. epidermidis in both experimental and clinical studies. In a canine model of aortic graft infection, 575
576 Bandyk et al.
colonization of Dacron vascular prosthesis resulted in a perigraft inflammatory process with anatomic and microbiologic characteristics similar to late graft infections in man, including the formation of anastomotic false aneurysms, s,4 O f note, the infectious process was primarily confined to the bioprosthesis as a bacteria-laden biofilm with little extension into surrounding tissue. In situ replacement of infected grafts with an expanded polytetrafluoroethylene (ePTFE) prosthesis resulted in normal perigraft healing and anastomotic tensile strength at both 1 and 3 months despite colonization of one third of the replacement grafts. 3,s Recently our group confirmed occult graft infection by CNS as an etiologic factor in approximately two thirds of patients in w h o m anastomotic femoral pseudoaneurysms developed after aortofemoral bypass grafting. 6 Despite recovery of CNS from explanted graft material, excision of the pseudoaneurysm and in situ replacement with a prosthetic graft was effective treatment. N o infections of the replacement grafts developed, and no recurrent pseudoaneurysms were observed during a mean follow-up interval of 14.5 months. Because CNS colonize vascular prostheses as a bacterial biofilm, it may be safe to manage this biomaterial infection by methods other than total graft excision and remote bypass. Several investigators have documented satisfactory results after local debridement and muscle flap coverage of perigraft infections localized to the groin. 7-~° Unfortunately, the microbiology of the graft infections treated in these reports was not routinely specifed. This report details an initial experience with in situ graft replacement to treat 15 consecutive patients with late-appearing prosthetic graft infections suspected to be caused by CNS. Patient selection for this less aggressive surgical therapy was based on clinical presentation, and the anatomic and microbiologic features of the infectious process. Our hypothesis was that local signs of graft infection can be eradicated by excision of the infected graft and replacement with a sterile prosthesis under the protection of parenteral antibiotic administration if the origin of the perigraft infection was a bacterial biofilm harboring CNS. CLINICAL MATERIAL AND METHODS OF STUDY Patients. From 1986 to 1990, 15 patients (10 men and 5 women)were admitted to the hospital without systemic signs of infection but with a perigraft process suggestive of a bacterial biofilm infection including a groin sinus tract, perigraft
Journal of VASCULAR SURGERY
inflammatory mass, or anastomotic false aneurysm with perigraft cavity. Pertinent clinical data from the 15 patients are outlined in Table I. Seventeen graft segments (two patients with bilateral aortofemoral graft limb infection) were involved with a perigraft infectious process that appeared as a femoral pseudoaneurysm (n = 7), groin or perigraft inflammatory mass (n = 6), or graft-cutaneous sinus tract (n = 4). Graft types involved included Dacron aortofemoral graft limbs (n = 14), Dacron axillofemoral graft limb (n = 2), and an ePTFE femoropopliteal below-knee bypass (n = 1). Indications for the primary grafting procedure were atherosclerosis obliterans (n -- 10), ruptured abdominal aortic aneurysm (n = 4), and limb ischemia after removal of infected aortic graft (n = 1). In eight patients the graft segments involved by the suspected bacterial biofilm infection had undergone one or multiple secondary procedures. Only two patients (nos. 2 and 8) developed signs of late graft infection after a single, elective arterial reconstruction, and in one of these patients a postoperative wound infection that did not involve the graft had been successfully treated. Diagnosis. The 15 patients selected for in situ graft replacement therapy fulfilled clinical, anatomic, and microbiologic criteria necessary to establish a diagnosis of graft infection caused by bacterial biofilms (Table II)." All patients were a~brile on admission to the hospital for treatment of suspected graft infection, and blood cultures obtained in 10 patients demonstrated no growth. Patients with early (less than 4 months) postoperative graft infections manifested by systemic graft sepsis (fever, leukocytosis, bacteremia) and evidence of bacterial invasion of the perigraft fluid and tissues (bacteria on Gram's stain and positive culture) were not considered to have a bacterial biofilm infection and thus were not candidates for in situ graft replacement. The mean (+ SEM) time interval between graft implantation (n = 6) or a secondary graft procedure (n = 9) and treatment of graft infection was 70 + 16 months (range, 5 months to 14 years). All treated graft segments had signs of a perigraft infection by preoperative imaging (ultrasonography, CT scanning) including one or all of the following: perigraft inflammation or cavity, absence of graft incorporation with surrounding tissue associated with false aneurysm formation, graft-cutaneous sinus tract (Fig. 1). Patients with suspected aortofemoral graft infection underwent both aortography and abdominal CT scanning to assess the integrity of the aorta-graft anastomosis. N o patient was identified to have an aortic false aneurysm. At operation all treated
Volume 13 Number 5
May 1991
Bacterial biofilm grafl infection
577
T a b l e I. Clinical characteristics o f 15 patients treated for prosthetic graft infection caused by bacterial biofilms
Patient
1
Sex~Age
M/60
Primary procedure
Dacron ABFG
2
M/62
3
M/70
4
M/71
PTFE Femoropopliteal bypass Dacron Femorofemoral bypass Dacron ABFG
5
M/64
Dacron ABFG
6
F/49
Dacron ABFG
7
M/68
Dacron ABFG
8
F/80
9
M/60
Dacron Axillofemoral bypass Dacron ABFG
10
F/63
11
M/64
Dacron Axillofemoral bypass Dacron ABFG
12 13
M/70 F/61
Dacron ABFG Dacron ABFG
14
F/68
Dacron ABFG
15
M/72
Dacron ABFG
Secondary procedure
Time to presentation ~
Dacron femoropopliteal bypass -
11 yrs
Dacron ABFG
14 yrs
Dacron femorofemoral bypass -
12 yrs
False aneurysm repair, thrombectomy -
Sign(s) at admission
Right femoral pseudo-
Admission WBC/oral temperature/-
8,900/afebrile
aneurysm
8 mo
11 yrs 1 yr 7 yrs
-
11 yrs
-
5 mo
-
4 yrs
Left femoral pseudoaneurysm Left femoral pseudoaneurysm Right femoral pseudoaneurysm Right femoral pseudoaneurism Inflammatory left groin mass Bilateral femoral pseudoaneurysms Graft-cutaneous sinus tract
13,100/afebrile 10,200/afebrile 1,400/afebrile 7,300/afebrile 8,500/afebrile 10,200/a febrile 7,400/a febrile
Inflammatory fight groin mass Perigraft cavity/exudate
9,500/afebrile
Bilateral groin sinus tracts, right femoral pseudoaneurysm Left groin sinus tract Inflammatory left groin mass
5,000/afebrile
5 mo
Inflammatory right groin mass
9,000/afebrilc
4 yrs
Inflammatory right groin mass, femoral pseudoaneurysm
10,100/afebrile
Thrombectomy
11 mo
Thrombectomy three times, false aneurysm repair Thrombectomy, false aneurysm repair False aneurysm repair
3 yrs 19 mo
7,200/a febrile
8,400/afebrile 8,900/afebrile
ABFG, Dacron aortobifemoral bypass graft; WBC, white blood cell count. Afebrile, less than 38 C. ~Calculated from primary or secondary operation to diagnosis of bacterial biofilm infection. ?Values prior to surgery for graft infection. graft segments were s u r r o u n d e d by a perigraft cavity containing a mucinous purulent fluid. T h e extent o f perigraft inflammation varied, but was m o s t pronounced at groin s e g m e n t o f the prosthetic bypass, or surrounding the fistulous tract between the perigraft cavity and the skin. G r a m ' s stain o f perigraft fluid revealed no bacteria but white blood cells in 14 o f 15 patients. Gram-positive cocci were seen on G r a m ' s stain o f perigraft fluid in one patient (no. 9), but cultures yielded no growth. Operative technique. Parenteral antibiotics (cefazolin or vancomycin) were administered before and after operation in all patients. T h e surgical technique o f in situ graft replacement was uniform, being p e r f o r m e d by two o f the authors (D.F.B., J.B.T.). T r e a t m e n t o f aortofemoral graft abnormalities appearing in the groin involved vascular isola-
tion o f the femoral artery anastomotic site. W h e n proximal control could not be safely obtained t h r o u g h the vertical groin incision, the graft limb was isolated above the inguinal ligament via a retroperitoneal approach by use o f a counter incision. T h e s e g m e n t o f unincorporated graft material was excised, and surrounding perigraft cavity/inflammation extensively debrided to normally appearing tissue. T h e graft-cutaneous sinus tract, if present, was excised en bloc as part o f exposure o f the aortofemoral graft limb and femoral anastomosis. T h e femoral artery wall adjacent to the anastomotic suture line was also debrided to normally appearingwall. An e P T F E vascular prosthesis (Gore-Tex; W. L. G o r e Associates Inc., Elkton, Md.) was anastomosed proximally to the divided aortofemoral graft limb and distally to the outflow vessels in end-to-end fashion. W h e n appro-
578
Journal of VASCULAR SURGERY
Bandyk et al.
Fig. 1. Contrast-enhanced C T scan of right aortofkmoral graft limb o f patient no. 9 who was admitted to the hospital with an inflammatory groin mass. Perigraft inflammation and caviw (arrow) surrounds prosthetic graft limb in groin with extension to the skin.
Table II. Diagnostic criteria for vascular prostheses infections caused by bacterial biofilms Clinical Presentation months to years after graft implantation No systemic signs of infection: Afebrile Normal white blood cell count Sterile blood culture Anatomic Adjacent tissue and artery wall inflammation Perigrai~ cavity and fluid Absence of graft incorporation Anastomotic dehiscence Microbiologic Perigraft fluid Gram stain: White blood cells No bacteria Perigraft fluid culture: No growth Graft biofilm culture: Coagulase-negative staphylococci
priate, profundaplasty or endarterectomy of the femoral arteries was used to enhance the graft outflow tract. Sartorius or gracilis muscle flap coverage was used selectively to cover the replacement ePTFE graft based on the status of perigraft tissue and the extent of tissue debridement performed. Treatment of axillofemoral graft infections involved complete replacement of the infected prostheses by use of an ePTFE graft positioned in an adjacent subcutaneous tunnel but anastomosed to the same inflow and outflow arteries. The skin was closed primarily in all cases. After in situ graft replacement,
parenteral antibiotic administration (cefazolin or vancomycin) was continued for a minimum of 14 days (range, 14 to 28 days), followed by a varied duration (6 to 12 weeks) of oral antibiotic therapy (cephalexin, ciprofloxacin, or dicloxacillin). Microbiologic technique. Cultures of perigraft tissue or fluid were obtained by a sterile cotton swab and transported to the clinical microbiology laboratory where plating on agar media was performed. At operation a 1 cm length of prosthetic material adjacent to the femoral anastomosis or within a perigraft cavity was excised and placed directly into 20 ml of glucose-supplemented trypticase soy broth (TSB). The explanted graft material was without adherent thrombus, perigraft capsule, or adjacent artery. A second graft segment of similar size was placed in TSB and ultrasonically oscillated (Fischer Sonic Dismembrator, Artek Systems Corp., Farmingdale, N.Y.) at 20 kHz for 10 minutes to mechanically disrupt the graft surface biofilm and disperse the adherent bacteria into the broth media for growth (biofilm culture). This method has previously been shown to increase the recovery of microorganisms from explanted vascular graft material. '2 The broth and biofilrn culture of graft material were incubated at 35 ° C for up to 2 weeks. After turbid growth, the broth was plated to blood agar, and isolated colonies were identified to species level. The staphylococcal isolates were identified with use of the API Staph TRAC system (Analylab Products, Plainview, N.Y.). Antibiotic susceptibility testing was performed for
Volume 13 Number 5 May 1991
each S. epidermidis isolate to cefazolin sodium, nafcillin sodium, vancomycin hydrochloride, gentamicin sulfate, and clindamycin phosphate. The recovered S. epidermidis strains were tested by means of alcian blue staining to assess their ability to produce an exopolysaccharide slime. An inoculum of the organism was transferred to 10 ml of TSB in a test tube containing a glass coverslip and allowed to incubate for 48 hours. The solution was decanted, and 10% solution of Alcian blue was added for 30 minutes. The coverslips were air dried, and mucin production was confirmed by comparison with mucin positive (ATCC 35983) and mucin negative (ATCC 35982) staphylococcal test strains. Follow-up. Patients were followed at 3- to &month intervals by clinical examination for recurrent signs of graft infection or false aneurysm formation. Fourteen of the 15 patients underwent one or serial duplex examinations of replacement ePTFE grafts to document patency and the presence of perigraft abnormalities (false aneurysm, perigraft cavity/fluid). FINDINGS
The microbiologic characteristics, operative procedure, and outcome of the 15 patients (17 graft segments) treated by in situ graft replacement are detailed in Table III. Microorganisms were recovered from 14 of the 17 graft segments studied. Thirteen of 14 isolates recovered were CNS. Staphylococcus epidermidis, the common organism recovered, was cultured from 11 explanted graft segments. Staphylococcus aureus was recovered from one graft, but culture of perigraft tissue exhibited no growth. In the three patients with graft-cutaneous sinus tracts, preoperative cultures of the perigraft fistula isolated CNS. Of note, all cultures were negative in two patients (no. 2 and 9) despite a clinical diagnosis of a bacterial biofilm infection. Eight of ten S. epidermidis isolates exhibited the ability to produce exopolysaccharide slime, a pathogenic characteristic that has been implicated in prosthetic infections. 13 Biofilm culture of explanted graft material was superior (p < 0.001, chi square) to broth culture alone in identifying a pathogen (Table IV). No broth or biofilm culture isolated more than one bacterial species. The operative procedure in all 15 patients consisted of graft replacement by use ofePTFE prosthesis, debridement of the perigraft tissues and adjacent artery, and parenteral antibiotics. A sartorius or gracilis myoplasty was used to cover the replaced segment of the aortofemoral graft limb in five cases.
Bacterial biofilmgrafi infection 579 The one femoropopliteal and two axillofemoral replacement grafts were tunneled in an adjacent subcutaneous tunnel with the proximal and distal anastomosis performed in situ to either a wellincorporated segment of previous graft or normal, debrided artery. Seven replacement grafts were covered with adjacent subcutaneous tissues. No patient died after operation, and no postoperative graft thromboses occurred. Three minor wound complications occurred including a lymphocele, wound hematoma, and a superficial wound infection. All were treated by local measures and resolved. None of the complications required operative correction or resulted in recurrent graft infection. During a follow-up period that ranged from 5 to 50 months (mean, 21 months), no recurrent graft infections or false aneurysms developed. Four of the 15 patients died of causes unrelated to graft infection. In two patients (nos. 4 and 12) autopsy examination of the intraabdominal aortofemoral graft segment demonstrated no clinical signs of infection or aortic false aneurysm formation. Duplex ultrasonography of the in situ replacement graft site was performed in 14 patients (16 in situ replacement ePTFE grafts) at a maximum postoperative interval ranging from 3 to 40 months. A residual groin lymphocele, documented by both ultrasonography and aspiration, was identified in one patient (no. 14) 3 months after repair of a noninfected femoral pseudoaneurysm. In three patients, duplex examination identified perigraft fluid surrounding the Dacron aortofemoral graft limb proximal to the replacement ePTFE graft, but no aortic or femoral false aneurysms were demonstrated. No patient had graft failure leading to lower limb amputation. Patient no. 10 required graft thrombectomy 3 years after in situ replacement of an infected axillofemoral graft. No perigraft infection was found at operation, but subsequent duplex examination of the graft revision site demonstrated a perigraft cavity. Antibiotic susceptibility testing was performed on 10 S. epidermidis isolates. The number of isolates resistant to various antibiotics were as follows: clindamycin, 3; cefazolin, 2; gentamicin, 1; nafcillin, 0; vancomycin, 0. DISCUSSION
Staphylococcus epidermidis is now the prevalent pathogen causing infections of implanted prosthetic devices in humans. Normally an organism of low pathogenicity, it has been demonstrated as the causative organism in most infections involving
Journal of VASCULAR SURGERY
Bandyk et al.
580
Table III. Microbiology, operative management, and outcome of 15 patients with bacterial biofilm graft infections treated by in situ graft replacement. Microbiology Patient
Perigraft
Biofilm culture
1
NG
S. epidermidis
2
NG
NG
3
NG
S. hominis
4
NG
S. epidermidis
5
NG
S. epidermidis
6
NG
S. hominis
7
NG
(R) S. aureus (L) S. epidermidis
8
CNS
S. epidermidis
9
NG
NG
10
NG
S. epidermidis
11
CNS
S. epidermidis
12
CNS
(R) S. epidermidis (L) NG
13
NG
S. epidermidis
14
NG
S, epiderm/d/s
15
NG
S. epidermidis
Operative procedure
Antibiotic Coverage
PTFE interposition graft PTFE obturator bypass PTFE interposition graft PTFE interposition graft PTFE interposition graft
Cefazofin/cephalexin Vancomycin/cephalexin Cefazolin/cephalexin Cefazolin/cephalexin Vancomycin/cephalexin
PTFE interposition graft Bilateral PTFE interposition grafts and sartorius muscle flaps PTFE replacement, Axillofemoral graft PTFE interposition graft, sartorius muscle flap PTFE interposition graft Bilateral PTFE interposition grafts and sartorius muscle flaps PTFE interposition graft, sartorius muscle flap PTFE interposition graft PTFE interposition graft, sartorius muscle flap PTFE interposition graft, gracilis muscle flap
Outcome
Months
Follow-up duplex scan
Alive
(50)
Normal
Died, cancer Alive
(7)
Normal
(30)
Normal
Died, cancer Died, cancer
(24)
Normal
(34)
Cefazolin/ciprofloxacin Cefazolin/cephalexin
Alive
(7)
ND, Normal clinical exam Normal
Alive
(10)
Normal
Vancomycin/ciprofloxacin
Alive
(11)
Normal
Vancomycin/ciprofloxacin
Alive
(7)
Normal
Cefazolin/dicloxacillin
Alive
(48)
Vancomycin/cephalexin
Alive
(30)
Segmental perigraft cavity Normal
Vancomycin/ciprofloxacin
Died, MI
(9)
Normal
Cefazolin/cephalexin Vancomycin/ciprofloxacin
Alive
(37)
Normal
Alive
(5)
Lymphocele
Cefazolin/cephalexin
Died, MI
(11)
Normal
NG, No growth; ND, not determined; PTFE, expanded polytetrafluoroethylene; CNS, coagulase-negative staphylococci; MI, myocardial infarction.
prosthetic heart valves, CSF shunts, hip prostheses, and prosthetic vascular grafts. 1'2"14q6 Infections of these devices are difficult to cure with antibiotics alone, but the combination of antibiotic therapy coupled with prosthesis replacement has been associated with favorable results (90% success rate) in
treating infected hip prosthesis and prosthetic heart valves. The present study confirmed that in situ graft replacement coupled with antibiotic therapy was equally successful in treating patients with late perigraft infections that manifested clinically as anastomotic false aneurysm, inflammatory groin mass, or
Volume 13 Number 5 May 1991
a graft-cutaneous sinus tract. Prosthetic graft segments located in the groin were the most commonly involved sites. Signs of graft infection were eradicated in all patients by operative management that included prolonged antibiotic administration, excision of the graft segment involved by bacterial biofilms, aggressive debridement of abnormal perigraft tissue including adjacent native artery, and followed by in situ replacement with ePTFE prosthesis. Rotational muscle flap coverage was used as an adjunct for soft tissue coverage of the arterial reconstruction in selected patients. Of significance, late postoperative imaging of 14 replacement ePTFE grafts by use of duplex ultrasonography demonstrated normal perigraft healing. Although our experience with in situ graft replacement is anectdotal, we believe this approach for treating late perigraft infections without gastrointestinal involvement resulted in less morbidity than would have occurred if total graft excision and extraanatomic revascularization had been performed. No deaths or graft failures occurred in the 15 patients selected for treatment. In a prior report, total graft excision with immediate vascular reconstruction resulted in a mortality and major amputation rate of 11%, and recurrent thrombosis of extraanatomic bypasses occurred in one third of survivors. 2 Eradication of the infectious process is an important principle in the management of graft infections. Factors that have been identified to influence the outcome of surgical management include the following: quantity and virulence of the pathogens, adequacy of local and systemic host defense mechanisms, and the extent of graft-artery involvement. Residual infection in the graft bed or artery wall is a major cause of morbidity and death, and the reason local treatment methods and in situ graft replacement procedures often fail. Local control of infection is more successful when graft infection is the result of colonization by S. epidermidis or other CNS compared with infections caused by S. aureus or gramnegative organisms, especially Pseudomonas aeruginosa. ~7 The colonization of prosthetic surfaces by CNS as a bacterial biofilm is a complex process, involving activation of host defenses by both the microorganism and their products of metabolism, and the foreign body. With time, autolysis of perigraft tissue occurs, which can culminate in anastomotic dehiscence and/or fistulization to skin or bowel. Important components of this process are: bacteria-laden biofilm on prosthetic surfaces, abnormal perigraft tissue with formation of a cavity containing a mucinous exudate, and a weakening of graft-artery anastomosis with formation of a false
Bacterial biofilm graft infection
581
Table IV. Recovery of microorganisms from the explanted graft material relative to culture technique Isolate
Broth culture
Biofilm culture ~
Coagulase-negative staphylococci S. epidermidis S. hominis S. aureus No growth
3 0 0 14
11 2 1 3
~p < 0.001 (chi square) compared to broth culture.
aneurysm (Fig. 2). The technique of in situ graft replacement described in this report attempts to treat these pathologic processes and also minimize colonization of the replacement prosthesis. Essential components of the operative procedure include aggressive debridement of both inflamed perigraft tissues and the artery wall adjacent to the anastomosis followed by graft replacement. Although perigraft cultures are typically negative, this can result from either a low number of organisms and/or the presence of normal host defenses. The adjacent artery wall is also known to be a potential source of CNS and can predispose to recurrent infection if aggressive debridement is not performed, s'~8"19 At present, we prefer an ePTFE vascular prosthesis for in situ graft replacement. This biomaterial has demonstrated reduced quantitative bacterial adherence compared to Dacron in vitro, and in animal models, was associated with improved early graft-healing after replacement of Dacron prostheses infected with slime-producing strain of S. epidermidis. 2"2° The possibility of developing a vascular prosthesis that is resistant to bacterial colonization also has considerable appeal both for the prevention and treatment of graft infections. In animal experiments binding antibiotic to graft surfaces has been demonstrated to increase graft resistance to infection ofbacteremia origin and has shown benefit in the treatment of established graft infections caused by S. aureus and Escherichia coli. 21"22 Further study is required to determine whether this approach can reliably prevent CNS from becoming sequestered within prosthetic surface biofilms, and thereby decreasing the significant problem of late graft infections. Lastly, although the use of local debridement and rotational muscle flap coverage has been shown to preserve graft function and result in local healing, we believe this therapy is inferior to in situ replacement since it does not address the underlying pathogenetic process, that is, the bacteria-laden biofilm on the graft material. For example, debride-
582
Journal of VASCULAR SURGERY
Bandyk et al.
PERIGRAFT CAVITY • chronic inflammation
• sterile exudate
/
- d4~Q~
-
~
~ W l
~
BACTERLA,LADEN SURFACE 810FILM . low m~obial corcentration • Wotect~ "sane coarse'
Fig. 2. Schematic of the anatomic and microbiologic characteristics of vascular prosthesis infection caused by bacterial biofilms. ment and myoplasty would be inadequate treatment in the presence of an anastomotic false aneurysm, a complication observed in more than one half of our patients. Expectant treatment of a bacterial biofilm graft infection or the use of antibiotic administration alone is not recommended. If the prosthesis becomes colonized by a more virulent pathogen during the period of medical treatment, the risk of"graft sepsis" and its life- and limb-threatening complications increases dramatically, and treatment by total graft excision and extraanatomic bypass becomes mandatory. Dissatisfaction with thrombosis and infection in extraanatomic bypass grafts has prompted a renewed interest in the use of in situ graft placement for a variety of vascular infections. Walker et al.2s reported no limb loss but a 26% mortality rate in a selected group of 23 patients with secondary aortoduodenal fistula using in situ replacement. Recently Robinson and Johansen2~ reported long-term survival of 11 patients treated by in situ grafting for both primary and secondary aortic sepsis. Both reviews emphasized the caveat of proper patient selection. Patients with graft-enteric erosions and minimal retroperitoneal infection fared best, particularly when treatment in~cluded aggressive aorta wall debridement and interposition Of the greater omentum between the replacement graft and bowel. The favorable outcome documented in many of these patients is not unexpected. We have confirmed the presence of a CNS biofilm infection remote from the site of bowel involvement in several patients initially seen with late aortic graft infection manifesting as a graft-enteric erosion or fistula. If the infectious process can be
eradicated by debridement and graft excision, in situ graft replacement with techniques and biomaterials that minimize bacterial colonization may emerge as an appropriate treatment option for proximal aortic graft infections. Proper selection of patients for in situ graft replacement requires sound clinical judgement, the use of diagnostic imaging techniques to assess extent of graft involvement, and sensitive microbiologic culture methods to identify the infecting microorganism. At present, the preoperative diagnosis of a vascular prosthesis infection caused by bacterial biofilms must be based solely on the clinical presentation and anatomic signs of infection. By use of the criteria of bacterial biofilm infections outlined in this study, biofilm culture of explanted graft material should isolate a microorganism, typically a CNS, in more than 80% of cases. We have not found labeled white cell scans helpful in establishing the presence or extent of a vascular prosthesis infection caused by bacterial biofilms. Contrast-enhanced CT scanning is useful for the detection of perigraft infammafion or fluid and to assess anastomotic sites for structural integrity. Of note, the authors do not recommend in situ replacement for the management of graft infections that (1) manifest in the early (less than 4 months) postoperative period with sepsis or an obvious, bacterial perigraft infection evident by positive Gram's stain and culture; or (2) appear as gastrointestinal hemorrhage caused by graft-enteric erosion or fistula. These grafts are infected by S. aureus or gram-negative organisms and require complete excision, particularly if anastomotic dehiscence has developed. 2s Careful anatomic and micro-
Volume 13 Number 5 May 1991
b i o l o g i c evaluation o f a p a t i e n t w i t h suspected graft infection enables the s u r g e o n to select, w i t h confidence, those patients w i t h bacterial biofilm infections that can be m a n a g e d b y in situ graft replacement. Patients t r e a t e d w i t h this less aggressive surgical t h e r a p y m u s t be f o l l o w e d closely b y use o f vascular i m a g i n g techniques to detect life- a n d limbt h r e a t e n i n g graft-healing c o m p l i c a t i o n s (anastom o t i c false aneurysm, graft-enteric erosion) associated w i t h recurrent o r persistent graft infection. A t present, it is n o t k n o w n if patients w i t h a bacterial biofilm graft infection can be cured b y in situ replacement. T h e results r e p o r t e d in this s t u d y indicate that in p r o p e r l y selected patients in situ r e p l a c e m e n t is safe, b u t further evaluation is warranted, especially the l o n g - t e r m f o l l o w - u p o f apparently successful cases. REFERENCES
1. Goldstone J, Moore WS. Infection in vascular prosthesis. Am J Surg 1974;128:225-33. 2. Bandyk DF, Berni GA, Thiele BL, Towne JB. Aortofemoral graft infection due to Staphylococcusepidermidis. Arch Surg 1984;119:102-8. 3. Bergamini TM, Bandyk DF, Govostis D, et al. Infection of vascular prostheses caused by bacterial biofilms. J VAsc SURG 1988;7:21-30. 4. Martin LF, Harris JM, Fehr DM, et aL Vascular prosthetic infection with Staphylococcus epidermidis: experimental study of pathogenesis and therapy. J VAsc SURG 1989; 9:464-71. 5. Vetsch R, Bandyk DF, Schmitt DD, et aL Anastomotic tensile strength following in situ replacement of an infected abdominal aortic graft. Arch Surg 1989;124:425-8. 6. Seabrook GR, Schmitt DD, Bandyk DF, et al. Anastomotic femoral pseudoaneurysm: an investigation of occult infection an etiologic factor. J VASCSURG 1990;11:629-34. 7. Samson RH, Veith FJ, Janko GS, et al. A modified classification and approach to the management of infections involving peripheral arterial prosthetic grafts. J VAsc SURG 1988; 8:147-53. 8. Mixter RC, Turnipseed WD, Smith DJ Jr, et al. Rotational muscle flap: a new technique for covering infected vascular grafts. J VAse SURG 1989;9:472-8. 9. Kretschmer G, Niederle B, Huk I, et al. Groin infections following vascular surgery: obturator bypass (BYP) versus "biologic coverage" (TRP)-a comparative analysis. Eur J Vasc Surg 1989;3:25-9.
Bacterial biofilm graft infection 583 10. Petrasek PF, Kalman PG, Martin RD. Sartorius myoplasty for deep groin wounds following vascular reconstruction. Am J Surg 1990;160:175-8. 11. Bergamini TM. Vascular prostheses infection caused by bacterial biofilms. Semin Vase Surg 1990;3:101-9. 12. Bergamini TM, Bandyk DF, Govostis D, et al. Identification of Staphylococcusepidermidisvascular graft infections: a comparison of culture techniques. J VAsc SURG 1989;9:665-70. 13. Christensen GD, Simpson WA, Younger JJ, et al. Adherence of coagulase-negative staphylococci to plastic tissue culture plates: a quantitative model for the adherence of staphylococci to medical devices. J Clin Microbiol 1985;22:996-1006. 14. Carlsson AS, Josefsson G, Lindberg L. Revision with gentarnicin-impregnated cement for deep infections in total hip arthroplasties. J Bone Joint Surg 1978;60A: 1059-64. 15. Schoenbaum SC, Gardner P, Shillito J. Infection of cerebrospinal shunts: epidemiology, clinical manifestation, therapy. J Infect Dis 1975;131:543-51. 16. Slaughter L, Morris IE, Starr A. Prosthetic valvular endocarditis: a 12-year review. Circulation 1973;47:1319-25. 17. Geary KJ, Tomkiewicz ZM, Harrison HN, et al. Differential effects of a gram-negative and a gram-positive infection on autogenous and prosthetic grafts. J VAsc SURG 1989;9:66570. 18. Macbeth GA, Rubin JR, McIntyre KE, et al. The relevance of arterial wall microbiology to the treatment of prosthetic graft infections. Graft infection vs. arterial infection. J VAse SURG 1984;1:750-6. 19. Durham JR, Malone JM, Bernhard VM. The impact of multiple operations on the importance of arterial wall cultures. J VASCSURG 1987;5:160-9. 20. Schmitt DD, Bandyk DF, Pequet AJ, Towne JB. Bacterial adherence to vascular prostheses. J VASCSURG 1986;3:73240. 21. Moore WS, Chavapil M, Sieffert G, et al. Development of an infection-resistant vascular prosthesis. Arch Surg 1981 ;116: 1403-7. 22. Shenk IS, Ney AL, Tsukayama DT, et al. Tobramycinadhesive in preventing and treating PTFE vascular graft infections. J Surg Res 1989;47:487-92. 23. Walker WE, Cooley DA, Duncan JM, et al. The management of aortoduodenal fistula by in situ replacement of the infected abdominal aortic graft. Ann Surg 1987;205:727-32. 24. Robinson JA, Johansen K. Aortic sepsis: Is there a role of in situ graft reconstruction? J VASCSURG (In press). 25. Schmitt DD, Seabrook GR, Bandyk DF, et al. Graft excision and extraanatomic reconstruction. The treatment of choice for the septic aortic prosthesis, l Cardiovasc Surg 1990;31:32732. Submitted Oct. 9, 1990; accepted Jan. 8, 1991.
Surgeons are well aware of the importance of
Staphylococcus aureus and gram-negative bacteria as pathogens in vascular prostheses infections, but only in recent years has the role of coagulase-negative staphylococci (CNS) been emphasized in the pathogenesis of late-appearing graft infections. I"2 The anatomic and microbiologic features of late graft infections differ from "classic graft sepsis" unless there is involvement of the gastrointestinal tract. Bacteria colonization is typically confined to prosthetic surfaces, systemic signs of infection are absent, and the number and virulence of the infecting organisms are low. Although several CNS strains have demonstrated the ability to infect implanted prosthetic devices in man, Staphylococcus epidermidis From the Department of Surgery,MedicalCollegeof Wisconsin, and the Surgical Service,Clement J ZablockiVeterans Affairs Medical Center, Milwaukee. Presented at the Fourteenth Annual Meeting of the Midwestern Vascular Surgical Society,Toledo, Ohio, Sept. 14-15, 1990. Reprint requests: Dennis Bandyk,MD, Department of Surgery, MCMC, 8700 W. WisconsinAve., Milwaukee,WI 53226. 24/6/27762
has been, by far, the most common strain implicated in late graft infections. Infection of a vascular prosthesis by S. epidermidis tends to manifest months to years after graft implantation or revision as a graft-healing complication (anastomotic false aneurysm, perigraft cavity abscess, graft-enteric fistula, or graft-cutaneous sinus tract). Management of these graft-healing complications can pose a difficult diagnostic and therapeutic dilemma for the surgeon. Despite evidence by preoperative imaging studies (ultrasonography, CT scanning) of perigraft inflammation, and at operation of purulence surrounding the prosthesis, Gram's stain or culture of the perigrat~ exudate or tissue may not confirm an infectious process. As such, the surgeon may be reluctant to proceed with graft excision and extraanatomic bypass grafting, and instead opt to procrastinate and treat the suspected perigraft infection with antibiotics until more obvious signs of graft sepsis develops. Our group has previously detailed important pathobiologic aspects of graft infections caused by S. epidermidis in both experimental and clinical studies. In a canine model of aortic graft infection, 575
576 Bandyk et al.
colonization of Dacron vascular prosthesis resulted in a perigraft inflammatory process with anatomic and microbiologic characteristics similar to late graft infections in man, including the formation of anastomotic false aneurysms, s,4 O f note, the infectious process was primarily confined to the bioprosthesis as a bacteria-laden biofilm with little extension into surrounding tissue. In situ replacement of infected grafts with an expanded polytetrafluoroethylene (ePTFE) prosthesis resulted in normal perigraft healing and anastomotic tensile strength at both 1 and 3 months despite colonization of one third of the replacement grafts. 3,s Recently our group confirmed occult graft infection by CNS as an etiologic factor in approximately two thirds of patients in w h o m anastomotic femoral pseudoaneurysms developed after aortofemoral bypass grafting. 6 Despite recovery of CNS from explanted graft material, excision of the pseudoaneurysm and in situ replacement with a prosthetic graft was effective treatment. N o infections of the replacement grafts developed, and no recurrent pseudoaneurysms were observed during a mean follow-up interval of 14.5 months. Because CNS colonize vascular prostheses as a bacterial biofilm, it may be safe to manage this biomaterial infection by methods other than total graft excision and remote bypass. Several investigators have documented satisfactory results after local debridement and muscle flap coverage of perigraft infections localized to the groin. 7-~° Unfortunately, the microbiology of the graft infections treated in these reports was not routinely specifed. This report details an initial experience with in situ graft replacement to treat 15 consecutive patients with late-appearing prosthetic graft infections suspected to be caused by CNS. Patient selection for this less aggressive surgical therapy was based on clinical presentation, and the anatomic and microbiologic features of the infectious process. Our hypothesis was that local signs of graft infection can be eradicated by excision of the infected graft and replacement with a sterile prosthesis under the protection of parenteral antibiotic administration if the origin of the perigraft infection was a bacterial biofilm harboring CNS. CLINICAL MATERIAL AND METHODS OF STUDY Patients. From 1986 to 1990, 15 patients (10 men and 5 women)were admitted to the hospital without systemic signs of infection but with a perigraft process suggestive of a bacterial biofilm infection including a groin sinus tract, perigraft
Journal of VASCULAR SURGERY
inflammatory mass, or anastomotic false aneurysm with perigraft cavity. Pertinent clinical data from the 15 patients are outlined in Table I. Seventeen graft segments (two patients with bilateral aortofemoral graft limb infection) were involved with a perigraft infectious process that appeared as a femoral pseudoaneurysm (n = 7), groin or perigraft inflammatory mass (n = 6), or graft-cutaneous sinus tract (n = 4). Graft types involved included Dacron aortofemoral graft limbs (n = 14), Dacron axillofemoral graft limb (n = 2), and an ePTFE femoropopliteal below-knee bypass (n = 1). Indications for the primary grafting procedure were atherosclerosis obliterans (n -- 10), ruptured abdominal aortic aneurysm (n = 4), and limb ischemia after removal of infected aortic graft (n = 1). In eight patients the graft segments involved by the suspected bacterial biofilm infection had undergone one or multiple secondary procedures. Only two patients (nos. 2 and 8) developed signs of late graft infection after a single, elective arterial reconstruction, and in one of these patients a postoperative wound infection that did not involve the graft had been successfully treated. Diagnosis. The 15 patients selected for in situ graft replacement therapy fulfilled clinical, anatomic, and microbiologic criteria necessary to establish a diagnosis of graft infection caused by bacterial biofilms (Table II)." All patients were a~brile on admission to the hospital for treatment of suspected graft infection, and blood cultures obtained in 10 patients demonstrated no growth. Patients with early (less than 4 months) postoperative graft infections manifested by systemic graft sepsis (fever, leukocytosis, bacteremia) and evidence of bacterial invasion of the perigraft fluid and tissues (bacteria on Gram's stain and positive culture) were not considered to have a bacterial biofilm infection and thus were not candidates for in situ graft replacement. The mean (+ SEM) time interval between graft implantation (n = 6) or a secondary graft procedure (n = 9) and treatment of graft infection was 70 + 16 months (range, 5 months to 14 years). All treated graft segments had signs of a perigraft infection by preoperative imaging (ultrasonography, CT scanning) including one or all of the following: perigraft inflammation or cavity, absence of graft incorporation with surrounding tissue associated with false aneurysm formation, graft-cutaneous sinus tract (Fig. 1). Patients with suspected aortofemoral graft infection underwent both aortography and abdominal CT scanning to assess the integrity of the aorta-graft anastomosis. N o patient was identified to have an aortic false aneurysm. At operation all treated
Volume 13 Number 5
May 1991
Bacterial biofilm grafl infection
577
T a b l e I. Clinical characteristics o f 15 patients treated for prosthetic graft infection caused by bacterial biofilms
Patient
1
Sex~Age
M/60
Primary procedure
Dacron ABFG
2
M/62
3
M/70
4
M/71
PTFE Femoropopliteal bypass Dacron Femorofemoral bypass Dacron ABFG
5
M/64
Dacron ABFG
6
F/49
Dacron ABFG
7
M/68
Dacron ABFG
8
F/80
9
M/60
Dacron Axillofemoral bypass Dacron ABFG
10
F/63
11
M/64
Dacron Axillofemoral bypass Dacron ABFG
12 13
M/70 F/61
Dacron ABFG Dacron ABFG
14
F/68
Dacron ABFG
15
M/72
Dacron ABFG
Secondary procedure
Time to presentation ~
Dacron femoropopliteal bypass -
11 yrs
Dacron ABFG
14 yrs
Dacron femorofemoral bypass -
12 yrs
False aneurysm repair, thrombectomy -
Sign(s) at admission
Right femoral pseudo-
Admission WBC/oral temperature/-
8,900/afebrile
aneurysm
8 mo
11 yrs 1 yr 7 yrs
-
11 yrs
-
5 mo
-
4 yrs
Left femoral pseudoaneurysm Left femoral pseudoaneurysm Right femoral pseudoaneurysm Right femoral pseudoaneurism Inflammatory left groin mass Bilateral femoral pseudoaneurysms Graft-cutaneous sinus tract
13,100/afebrile 10,200/afebrile 1,400/afebrile 7,300/afebrile 8,500/afebrile 10,200/a febrile 7,400/a febrile
Inflammatory fight groin mass Perigraft cavity/exudate
9,500/afebrile
Bilateral groin sinus tracts, right femoral pseudoaneurysm Left groin sinus tract Inflammatory left groin mass
5,000/afebrile
5 mo
Inflammatory right groin mass
9,000/afebrilc
4 yrs
Inflammatory right groin mass, femoral pseudoaneurysm
10,100/afebrile
Thrombectomy
11 mo
Thrombectomy three times, false aneurysm repair Thrombectomy, false aneurysm repair False aneurysm repair
3 yrs 19 mo
7,200/a febrile
8,400/afebrile 8,900/afebrile
ABFG, Dacron aortobifemoral bypass graft; WBC, white blood cell count. Afebrile, less than 38 C. ~Calculated from primary or secondary operation to diagnosis of bacterial biofilm infection. ?Values prior to surgery for graft infection. graft segments were s u r r o u n d e d by a perigraft cavity containing a mucinous purulent fluid. T h e extent o f perigraft inflammation varied, but was m o s t pronounced at groin s e g m e n t o f the prosthetic bypass, or surrounding the fistulous tract between the perigraft cavity and the skin. G r a m ' s stain o f perigraft fluid revealed no bacteria but white blood cells in 14 o f 15 patients. Gram-positive cocci were seen on G r a m ' s stain o f perigraft fluid in one patient (no. 9), but cultures yielded no growth. Operative technique. Parenteral antibiotics (cefazolin or vancomycin) were administered before and after operation in all patients. T h e surgical technique o f in situ graft replacement was uniform, being p e r f o r m e d by two o f the authors (D.F.B., J.B.T.). T r e a t m e n t o f aortofemoral graft abnormalities appearing in the groin involved vascular isola-
tion o f the femoral artery anastomotic site. W h e n proximal control could not be safely obtained t h r o u g h the vertical groin incision, the graft limb was isolated above the inguinal ligament via a retroperitoneal approach by use o f a counter incision. T h e s e g m e n t o f unincorporated graft material was excised, and surrounding perigraft cavity/inflammation extensively debrided to normally appearing tissue. T h e graft-cutaneous sinus tract, if present, was excised en bloc as part o f exposure o f the aortofemoral graft limb and femoral anastomosis. T h e femoral artery wall adjacent to the anastomotic suture line was also debrided to normally appearingwall. An e P T F E vascular prosthesis (Gore-Tex; W. L. G o r e Associates Inc., Elkton, Md.) was anastomosed proximally to the divided aortofemoral graft limb and distally to the outflow vessels in end-to-end fashion. W h e n appro-
578
Journal of VASCULAR SURGERY
Bandyk et al.
Fig. 1. Contrast-enhanced C T scan of right aortofkmoral graft limb o f patient no. 9 who was admitted to the hospital with an inflammatory groin mass. Perigraft inflammation and caviw (arrow) surrounds prosthetic graft limb in groin with extension to the skin.
Table II. Diagnostic criteria for vascular prostheses infections caused by bacterial biofilms Clinical Presentation months to years after graft implantation No systemic signs of infection: Afebrile Normal white blood cell count Sterile blood culture Anatomic Adjacent tissue and artery wall inflammation Perigrai~ cavity and fluid Absence of graft incorporation Anastomotic dehiscence Microbiologic Perigraft fluid Gram stain: White blood cells No bacteria Perigraft fluid culture: No growth Graft biofilm culture: Coagulase-negative staphylococci
priate, profundaplasty or endarterectomy of the femoral arteries was used to enhance the graft outflow tract. Sartorius or gracilis muscle flap coverage was used selectively to cover the replacement ePTFE graft based on the status of perigraft tissue and the extent of tissue debridement performed. Treatment of axillofemoral graft infections involved complete replacement of the infected prostheses by use of an ePTFE graft positioned in an adjacent subcutaneous tunnel but anastomosed to the same inflow and outflow arteries. The skin was closed primarily in all cases. After in situ graft replacement,
parenteral antibiotic administration (cefazolin or vancomycin) was continued for a minimum of 14 days (range, 14 to 28 days), followed by a varied duration (6 to 12 weeks) of oral antibiotic therapy (cephalexin, ciprofloxacin, or dicloxacillin). Microbiologic technique. Cultures of perigraft tissue or fluid were obtained by a sterile cotton swab and transported to the clinical microbiology laboratory where plating on agar media was performed. At operation a 1 cm length of prosthetic material adjacent to the femoral anastomosis or within a perigraft cavity was excised and placed directly into 20 ml of glucose-supplemented trypticase soy broth (TSB). The explanted graft material was without adherent thrombus, perigraft capsule, or adjacent artery. A second graft segment of similar size was placed in TSB and ultrasonically oscillated (Fischer Sonic Dismembrator, Artek Systems Corp., Farmingdale, N.Y.) at 20 kHz for 10 minutes to mechanically disrupt the graft surface biofilm and disperse the adherent bacteria into the broth media for growth (biofilm culture). This method has previously been shown to increase the recovery of microorganisms from explanted vascular graft material. '2 The broth and biofilrn culture of graft material were incubated at 35 ° C for up to 2 weeks. After turbid growth, the broth was plated to blood agar, and isolated colonies were identified to species level. The staphylococcal isolates were identified with use of the API Staph TRAC system (Analylab Products, Plainview, N.Y.). Antibiotic susceptibility testing was performed for
Volume 13 Number 5 May 1991
each S. epidermidis isolate to cefazolin sodium, nafcillin sodium, vancomycin hydrochloride, gentamicin sulfate, and clindamycin phosphate. The recovered S. epidermidis strains were tested by means of alcian blue staining to assess their ability to produce an exopolysaccharide slime. An inoculum of the organism was transferred to 10 ml of TSB in a test tube containing a glass coverslip and allowed to incubate for 48 hours. The solution was decanted, and 10% solution of Alcian blue was added for 30 minutes. The coverslips were air dried, and mucin production was confirmed by comparison with mucin positive (ATCC 35983) and mucin negative (ATCC 35982) staphylococcal test strains. Follow-up. Patients were followed at 3- to &month intervals by clinical examination for recurrent signs of graft infection or false aneurysm formation. Fourteen of the 15 patients underwent one or serial duplex examinations of replacement ePTFE grafts to document patency and the presence of perigraft abnormalities (false aneurysm, perigraft cavity/fluid). FINDINGS
The microbiologic characteristics, operative procedure, and outcome of the 15 patients (17 graft segments) treated by in situ graft replacement are detailed in Table III. Microorganisms were recovered from 14 of the 17 graft segments studied. Thirteen of 14 isolates recovered were CNS. Staphylococcus epidermidis, the common organism recovered, was cultured from 11 explanted graft segments. Staphylococcus aureus was recovered from one graft, but culture of perigraft tissue exhibited no growth. In the three patients with graft-cutaneous sinus tracts, preoperative cultures of the perigraft fistula isolated CNS. Of note, all cultures were negative in two patients (no. 2 and 9) despite a clinical diagnosis of a bacterial biofilm infection. Eight of ten S. epidermidis isolates exhibited the ability to produce exopolysaccharide slime, a pathogenic characteristic that has been implicated in prosthetic infections. 13 Biofilm culture of explanted graft material was superior (p < 0.001, chi square) to broth culture alone in identifying a pathogen (Table IV). No broth or biofilm culture isolated more than one bacterial species. The operative procedure in all 15 patients consisted of graft replacement by use ofePTFE prosthesis, debridement of the perigraft tissues and adjacent artery, and parenteral antibiotics. A sartorius or gracilis myoplasty was used to cover the replaced segment of the aortofemoral graft limb in five cases.
Bacterial biofilmgrafi infection 579 The one femoropopliteal and two axillofemoral replacement grafts were tunneled in an adjacent subcutaneous tunnel with the proximal and distal anastomosis performed in situ to either a wellincorporated segment of previous graft or normal, debrided artery. Seven replacement grafts were covered with adjacent subcutaneous tissues. No patient died after operation, and no postoperative graft thromboses occurred. Three minor wound complications occurred including a lymphocele, wound hematoma, and a superficial wound infection. All were treated by local measures and resolved. None of the complications required operative correction or resulted in recurrent graft infection. During a follow-up period that ranged from 5 to 50 months (mean, 21 months), no recurrent graft infections or false aneurysms developed. Four of the 15 patients died of causes unrelated to graft infection. In two patients (nos. 4 and 12) autopsy examination of the intraabdominal aortofemoral graft segment demonstrated no clinical signs of infection or aortic false aneurysm formation. Duplex ultrasonography of the in situ replacement graft site was performed in 14 patients (16 in situ replacement ePTFE grafts) at a maximum postoperative interval ranging from 3 to 40 months. A residual groin lymphocele, documented by both ultrasonography and aspiration, was identified in one patient (no. 14) 3 months after repair of a noninfected femoral pseudoaneurysm. In three patients, duplex examination identified perigraft fluid surrounding the Dacron aortofemoral graft limb proximal to the replacement ePTFE graft, but no aortic or femoral false aneurysms were demonstrated. No patient had graft failure leading to lower limb amputation. Patient no. 10 required graft thrombectomy 3 years after in situ replacement of an infected axillofemoral graft. No perigraft infection was found at operation, but subsequent duplex examination of the graft revision site demonstrated a perigraft cavity. Antibiotic susceptibility testing was performed on 10 S. epidermidis isolates. The number of isolates resistant to various antibiotics were as follows: clindamycin, 3; cefazolin, 2; gentamicin, 1; nafcillin, 0; vancomycin, 0. DISCUSSION
Staphylococcus epidermidis is now the prevalent pathogen causing infections of implanted prosthetic devices in humans. Normally an organism of low pathogenicity, it has been demonstrated as the causative organism in most infections involving
Journal of VASCULAR SURGERY
Bandyk et al.
580
Table III. Microbiology, operative management, and outcome of 15 patients with bacterial biofilm graft infections treated by in situ graft replacement. Microbiology Patient
Perigraft
Biofilm culture
1
NG
S. epidermidis
2
NG
NG
3
NG
S. hominis
4
NG
S. epidermidis
5
NG
S. epidermidis
6
NG
S. hominis
7
NG
(R) S. aureus (L) S. epidermidis
8
CNS
S. epidermidis
9
NG
NG
10
NG
S. epidermidis
11
CNS
S. epidermidis
12
CNS
(R) S. epidermidis (L) NG
13
NG
S. epidermidis
14
NG
S, epiderm/d/s
15
NG
S. epidermidis
Operative procedure
Antibiotic Coverage
PTFE interposition graft PTFE obturator bypass PTFE interposition graft PTFE interposition graft PTFE interposition graft
Cefazofin/cephalexin Vancomycin/cephalexin Cefazolin/cephalexin Cefazolin/cephalexin Vancomycin/cephalexin
PTFE interposition graft Bilateral PTFE interposition grafts and sartorius muscle flaps PTFE replacement, Axillofemoral graft PTFE interposition graft, sartorius muscle flap PTFE interposition graft Bilateral PTFE interposition grafts and sartorius muscle flaps PTFE interposition graft, sartorius muscle flap PTFE interposition graft PTFE interposition graft, sartorius muscle flap PTFE interposition graft, gracilis muscle flap
Outcome
Months
Follow-up duplex scan
Alive
(50)
Normal
Died, cancer Alive
(7)
Normal
(30)
Normal
Died, cancer Died, cancer
(24)
Normal
(34)
Cefazolin/ciprofloxacin Cefazolin/cephalexin
Alive
(7)
ND, Normal clinical exam Normal
Alive
(10)
Normal
Vancomycin/ciprofloxacin
Alive
(11)
Normal
Vancomycin/ciprofloxacin
Alive
(7)
Normal
Cefazolin/dicloxacillin
Alive
(48)
Vancomycin/cephalexin
Alive
(30)
Segmental perigraft cavity Normal
Vancomycin/ciprofloxacin
Died, MI
(9)
Normal
Cefazolin/cephalexin Vancomycin/ciprofloxacin
Alive
(37)
Normal
Alive
(5)
Lymphocele
Cefazolin/cephalexin
Died, MI
(11)
Normal
NG, No growth; ND, not determined; PTFE, expanded polytetrafluoroethylene; CNS, coagulase-negative staphylococci; MI, myocardial infarction.
prosthetic heart valves, CSF shunts, hip prostheses, and prosthetic vascular grafts. 1'2"14q6 Infections of these devices are difficult to cure with antibiotics alone, but the combination of antibiotic therapy coupled with prosthesis replacement has been associated with favorable results (90% success rate) in
treating infected hip prosthesis and prosthetic heart valves. The present study confirmed that in situ graft replacement coupled with antibiotic therapy was equally successful in treating patients with late perigraft infections that manifested clinically as anastomotic false aneurysm, inflammatory groin mass, or
Volume 13 Number 5 May 1991
a graft-cutaneous sinus tract. Prosthetic graft segments located in the groin were the most commonly involved sites. Signs of graft infection were eradicated in all patients by operative management that included prolonged antibiotic administration, excision of the graft segment involved by bacterial biofilms, aggressive debridement of abnormal perigraft tissue including adjacent native artery, and followed by in situ replacement with ePTFE prosthesis. Rotational muscle flap coverage was used as an adjunct for soft tissue coverage of the arterial reconstruction in selected patients. Of significance, late postoperative imaging of 14 replacement ePTFE grafts by use of duplex ultrasonography demonstrated normal perigraft healing. Although our experience with in situ graft replacement is anectdotal, we believe this approach for treating late perigraft infections without gastrointestinal involvement resulted in less morbidity than would have occurred if total graft excision and extraanatomic revascularization had been performed. No deaths or graft failures occurred in the 15 patients selected for treatment. In a prior report, total graft excision with immediate vascular reconstruction resulted in a mortality and major amputation rate of 11%, and recurrent thrombosis of extraanatomic bypasses occurred in one third of survivors. 2 Eradication of the infectious process is an important principle in the management of graft infections. Factors that have been identified to influence the outcome of surgical management include the following: quantity and virulence of the pathogens, adequacy of local and systemic host defense mechanisms, and the extent of graft-artery involvement. Residual infection in the graft bed or artery wall is a major cause of morbidity and death, and the reason local treatment methods and in situ graft replacement procedures often fail. Local control of infection is more successful when graft infection is the result of colonization by S. epidermidis or other CNS compared with infections caused by S. aureus or gramnegative organisms, especially Pseudomonas aeruginosa. ~7 The colonization of prosthetic surfaces by CNS as a bacterial biofilm is a complex process, involving activation of host defenses by both the microorganism and their products of metabolism, and the foreign body. With time, autolysis of perigraft tissue occurs, which can culminate in anastomotic dehiscence and/or fistulization to skin or bowel. Important components of this process are: bacteria-laden biofilm on prosthetic surfaces, abnormal perigraft tissue with formation of a cavity containing a mucinous exudate, and a weakening of graft-artery anastomosis with formation of a false
Bacterial biofilm graft infection
581
Table IV. Recovery of microorganisms from the explanted graft material relative to culture technique Isolate
Broth culture
Biofilm culture ~
Coagulase-negative staphylococci S. epidermidis S. hominis S. aureus No growth
3 0 0 14
11 2 1 3
~p < 0.001 (chi square) compared to broth culture.
aneurysm (Fig. 2). The technique of in situ graft replacement described in this report attempts to treat these pathologic processes and also minimize colonization of the replacement prosthesis. Essential components of the operative procedure include aggressive debridement of both inflamed perigraft tissues and the artery wall adjacent to the anastomosis followed by graft replacement. Although perigraft cultures are typically negative, this can result from either a low number of organisms and/or the presence of normal host defenses. The adjacent artery wall is also known to be a potential source of CNS and can predispose to recurrent infection if aggressive debridement is not performed, s'~8"19 At present, we prefer an ePTFE vascular prosthesis for in situ graft replacement. This biomaterial has demonstrated reduced quantitative bacterial adherence compared to Dacron in vitro, and in animal models, was associated with improved early graft-healing after replacement of Dacron prostheses infected with slime-producing strain of S. epidermidis. 2"2° The possibility of developing a vascular prosthesis that is resistant to bacterial colonization also has considerable appeal both for the prevention and treatment of graft infections. In animal experiments binding antibiotic to graft surfaces has been demonstrated to increase graft resistance to infection ofbacteremia origin and has shown benefit in the treatment of established graft infections caused by S. aureus and Escherichia coli. 21"22 Further study is required to determine whether this approach can reliably prevent CNS from becoming sequestered within prosthetic surface biofilms, and thereby decreasing the significant problem of late graft infections. Lastly, although the use of local debridement and rotational muscle flap coverage has been shown to preserve graft function and result in local healing, we believe this therapy is inferior to in situ replacement since it does not address the underlying pathogenetic process, that is, the bacteria-laden biofilm on the graft material. For example, debride-
582
Journal of VASCULAR SURGERY
Bandyk et al.
PERIGRAFT CAVITY • chronic inflammation
• sterile exudate
/
- d4~Q~
-
~
~ W l
~
BACTERLA,LADEN SURFACE 810FILM . low m~obial corcentration • Wotect~ "sane coarse'
Fig. 2. Schematic of the anatomic and microbiologic characteristics of vascular prosthesis infection caused by bacterial biofilms. ment and myoplasty would be inadequate treatment in the presence of an anastomotic false aneurysm, a complication observed in more than one half of our patients. Expectant treatment of a bacterial biofilm graft infection or the use of antibiotic administration alone is not recommended. If the prosthesis becomes colonized by a more virulent pathogen during the period of medical treatment, the risk of"graft sepsis" and its life- and limb-threatening complications increases dramatically, and treatment by total graft excision and extraanatomic bypass becomes mandatory. Dissatisfaction with thrombosis and infection in extraanatomic bypass grafts has prompted a renewed interest in the use of in situ graft placement for a variety of vascular infections. Walker et al.2s reported no limb loss but a 26% mortality rate in a selected group of 23 patients with secondary aortoduodenal fistula using in situ replacement. Recently Robinson and Johansen2~ reported long-term survival of 11 patients treated by in situ grafting for both primary and secondary aortic sepsis. Both reviews emphasized the caveat of proper patient selection. Patients with graft-enteric erosions and minimal retroperitoneal infection fared best, particularly when treatment in~cluded aggressive aorta wall debridement and interposition Of the greater omentum between the replacement graft and bowel. The favorable outcome documented in many of these patients is not unexpected. We have confirmed the presence of a CNS biofilm infection remote from the site of bowel involvement in several patients initially seen with late aortic graft infection manifesting as a graft-enteric erosion or fistula. If the infectious process can be
eradicated by debridement and graft excision, in situ graft replacement with techniques and biomaterials that minimize bacterial colonization may emerge as an appropriate treatment option for proximal aortic graft infections. Proper selection of patients for in situ graft replacement requires sound clinical judgement, the use of diagnostic imaging techniques to assess extent of graft involvement, and sensitive microbiologic culture methods to identify the infecting microorganism. At present, the preoperative diagnosis of a vascular prosthesis infection caused by bacterial biofilms must be based solely on the clinical presentation and anatomic signs of infection. By use of the criteria of bacterial biofilm infections outlined in this study, biofilm culture of explanted graft material should isolate a microorganism, typically a CNS, in more than 80% of cases. We have not found labeled white cell scans helpful in establishing the presence or extent of a vascular prosthesis infection caused by bacterial biofilms. Contrast-enhanced CT scanning is useful for the detection of perigraft infammafion or fluid and to assess anastomotic sites for structural integrity. Of note, the authors do not recommend in situ replacement for the management of graft infections that (1) manifest in the early (less than 4 months) postoperative period with sepsis or an obvious, bacterial perigraft infection evident by positive Gram's stain and culture; or (2) appear as gastrointestinal hemorrhage caused by graft-enteric erosion or fistula. These grafts are infected by S. aureus or gram-negative organisms and require complete excision, particularly if anastomotic dehiscence has developed. 2s Careful anatomic and micro-
Volume 13 Number 5 May 1991
b i o l o g i c evaluation o f a p a t i e n t w i t h suspected graft infection enables the s u r g e o n to select, w i t h confidence, those patients w i t h bacterial biofilm infections that can be m a n a g e d b y in situ graft replacement. Patients t r e a t e d w i t h this less aggressive surgical t h e r a p y m u s t be f o l l o w e d closely b y use o f vascular i m a g i n g techniques to detect life- a n d limbt h r e a t e n i n g graft-healing c o m p l i c a t i o n s (anastom o t i c false aneurysm, graft-enteric erosion) associated w i t h recurrent o r persistent graft infection. A t present, it is n o t k n o w n if patients w i t h a bacterial biofilm graft infection can be cured b y in situ replacement. T h e results r e p o r t e d in this s t u d y indicate that in p r o p e r l y selected patients in situ r e p l a c e m e n t is safe, b u t further evaluation is warranted, especially the l o n g - t e r m f o l l o w - u p o f apparently successful cases. REFERENCES
1. Goldstone J, Moore WS. Infection in vascular prosthesis. Am J Surg 1974;128:225-33. 2. Bandyk DF, Berni GA, Thiele BL, Towne JB. Aortofemoral graft infection due to Staphylococcusepidermidis. Arch Surg 1984;119:102-8. 3. Bergamini TM, Bandyk DF, Govostis D, et al. Infection of vascular prostheses caused by bacterial biofilms. J VAsc SURG 1988;7:21-30. 4. Martin LF, Harris JM, Fehr DM, et aL Vascular prosthetic infection with Staphylococcus epidermidis: experimental study of pathogenesis and therapy. J VAsc SURG 1989; 9:464-71. 5. Vetsch R, Bandyk DF, Schmitt DD, et aL Anastomotic tensile strength following in situ replacement of an infected abdominal aortic graft. Arch Surg 1989;124:425-8. 6. Seabrook GR, Schmitt DD, Bandyk DF, et al. Anastomotic femoral pseudoaneurysm: an investigation of occult infection an etiologic factor. J VASCSURG 1990;11:629-34. 7. Samson RH, Veith FJ, Janko GS, et al. A modified classification and approach to the management of infections involving peripheral arterial prosthetic grafts. J VAsc SURG 1988; 8:147-53. 8. Mixter RC, Turnipseed WD, Smith DJ Jr, et al. Rotational muscle flap: a new technique for covering infected vascular grafts. J VAse SURG 1989;9:472-8. 9. Kretschmer G, Niederle B, Huk I, et al. Groin infections following vascular surgery: obturator bypass (BYP) versus "biologic coverage" (TRP)-a comparative analysis. Eur J Vasc Surg 1989;3:25-9.
Bacterial biofilm graft infection 583 10. Petrasek PF, Kalman PG, Martin RD. Sartorius myoplasty for deep groin wounds following vascular reconstruction. Am J Surg 1990;160:175-8. 11. Bergamini TM. Vascular prostheses infection caused by bacterial biofilms. Semin Vase Surg 1990;3:101-9. 12. Bergamini TM, Bandyk DF, Govostis D, et al. Identification of Staphylococcusepidermidisvascular graft infections: a comparison of culture techniques. J VAsc SURG 1989;9:665-70. 13. Christensen GD, Simpson WA, Younger JJ, et al. Adherence of coagulase-negative staphylococci to plastic tissue culture plates: a quantitative model for the adherence of staphylococci to medical devices. J Clin Microbiol 1985;22:996-1006. 14. Carlsson AS, Josefsson G, Lindberg L. Revision with gentarnicin-impregnated cement for deep infections in total hip arthroplasties. J Bone Joint Surg 1978;60A: 1059-64. 15. Schoenbaum SC, Gardner P, Shillito J. Infection of cerebrospinal shunts: epidemiology, clinical manifestation, therapy. J Infect Dis 1975;131:543-51. 16. Slaughter L, Morris IE, Starr A. Prosthetic valvular endocarditis: a 12-year review. Circulation 1973;47:1319-25. 17. Geary KJ, Tomkiewicz ZM, Harrison HN, et al. Differential effects of a gram-negative and a gram-positive infection on autogenous and prosthetic grafts. J VAsc SURG 1989;9:66570. 18. Macbeth GA, Rubin JR, McIntyre KE, et al. The relevance of arterial wall microbiology to the treatment of prosthetic graft infections. Graft infection vs. arterial infection. J VAse SURG 1984;1:750-6. 19. Durham JR, Malone JM, Bernhard VM. The impact of multiple operations on the importance of arterial wall cultures. J VASCSURG 1987;5:160-9. 20. Schmitt DD, Bandyk DF, Pequet AJ, Towne JB. Bacterial adherence to vascular prostheses. J VASCSURG 1986;3:73240. 21. Moore WS, Chavapil M, Sieffert G, et al. Development of an infection-resistant vascular prosthesis. Arch Surg 1981 ;116: 1403-7. 22. Shenk IS, Ney AL, Tsukayama DT, et al. Tobramycinadhesive in preventing and treating PTFE vascular graft infections. J Surg Res 1989;47:487-92. 23. Walker WE, Cooley DA, Duncan JM, et al. The management of aortoduodenal fistula by in situ replacement of the infected abdominal aortic graft. Ann Surg 1987;205:727-32. 24. Robinson JA, Johansen K. Aortic sepsis: Is there a role of in situ graft reconstruction? J VASCSURG (In press). 25. Schmitt DD, Seabrook GR, Bandyk DF, et al. Graft excision and extraanatomic reconstruction. The treatment of choice for the septic aortic prosthesis, l Cardiovasc Surg 1990;31:32732. Submitted Oct. 9, 1990; accepted Jan. 8, 1991.