Superficial femoropopliteal vein (SFPV) aortofemoral reconstructions. A, Unilateral SFPV aortofemoral bypass with femoral crossover (n=7); B, SFPV aortobifemoral bypasses (n=6); C, SFPV from prosthetic aortofemoral bypass limb to femoral bypass (n=2); D, Unilateral SFPV aortofemoral bypass with contralateral aortoiliac bypass (n=1); and E, SFPV aortoiliac bypass with iliofemoral crossover (n=1). Unilateral SFPV aortofemoral bypass (n=1) is not shown.
Gordon LL, Hagino RT, Jackson MR, Modrall JG, Valentine RJ, Clagett GP. Complex Aortofemoral Prosthetic InfectionsThe Role of Autogenous Superficial Femoropopliteal Vein Reconstruction. Arch Surg. 1999;134(6):615-621. doi:10.1001/archsurg.134.6.615
With increasing experience, we have encountered patients with complex aortofemoral prosthetic infections in whom extra-anatomic bypass (EAB) is not an option.
Autogenous superficial femoropopliteal vein (SFPV) aortic reconstruction provides a limb-saving and lifesaving alternative with acceptable morbidity and mortality.
University-based county, private, and Veterans Affairs hospitals.
Seventeen patients with infected aortofemoral bypasses in whom conventional EAB was impossible because of infection of previously placed EAB, massive groin and/or thigh sepsis, or both.
Main Outcome Measures
Morbidity and mortality.
Multiple previous operations were common (mean, 4 per patient) and included EAB (n=11), replacement aortofemoral bypass (n=4), prosthetic femoropopliteal bypass (n=7), and thoracobifemoral bypass (n=1); all bypasses became infected. Overall, 11 patients had sepsis at the time of presentation. Of the patients with massive groin infection, 7 had extensive deep infections involving most of the proximal thighs or retroperitoneum, 4 had enterocutaneous fistulae, and 2 had necrotizing fasciitis of the lower abdomen and thigh. Polymicrobial infections were common (n=9). Four patients (24%) died in the perioperative period, 8 (47%) suffered major complications, and 4 (24%) underwent major amputations. Mortality in this group of patients was 3 times that of all other patients undergoing autogenous SFPV aortic reconstruction for prosthetic infection (8%). Amputation rates were also increased (24% vs 6%). The mean ± SD follow-up time is 23 ± 21 months. All patients maintained patent SFPV reconstructions.
In the setting of complex aortofemoral prosthetic infections, autogenous SFPV aortic reconstruction is a useful option for patients in whom EAB is impossible and limb loss and/or death would be inevitable without revascularization.
AUTOGENOUS superficial femoropopliteal vein (SFPV) reconstruction has successfully been used as an alternative to extra-anatomic bypass (EAB) in the treatment of infected aortic prostheses.1- 4 Superficial femoropopliteal vein reconstruction offers many advantages including resistance to reinfection, durability, excellent long-term patency in the face of compromised outflow, low amputation rates, minimal lower extremity venous morbidity, and no risk for aortic stump blowout.2,5 In earlier reports from our group, all patients underwent SFPV reconstruction regardless of their eligibility for EAB.1,2 As experience has accrued, we have encountered patients with more complex and advanced disease in whom EAB was not feasible, leaving autogenous SFPV aortic replacement as the only solution. These patients presented with infection of previously placed EABs or massive lower abdominal, groin, and thigh infection that precluded safe EAB. This subgroup of desperately ill patients with aortic prosthetic infection treated using autogenous SFPV aortic reconstruction is the subject of our report. Our purpose is to determine the feasibility and usefulness of SFPV reconstruction in this unusual circumstance.
Data were collected on all patients undergoing SFPV reconstruction for aortic prosthetic infection since January 1, 1990. We identified 17 patients in whom conventional EAB was impossible because of infection of previously placed EAB (n=11) and/or massive groin and/or thigh sepsis (n=9). Three patients had both conditions. Six of the 11 patients with infected EABs had concomitant graft occlusion.
Eighteen procedures were performed in 17 patients. All patients underwent single-staged procedures consisting of SFPV harvest, infected graft removal, and SFPV aortofemoral reconstruction. Using a 2-team approach, 1 team to harvest SFPV and 1 team to expose the aorta, the time of operation is greatly reduced. The SFPV harvest is performed through an incision placed along the lateral border of the sartorius muscle. Reflection of the sartorius muscle medially exposes the SFPV. The proximal transection should be placed flush with the profunda femoris–common femoral vein junction to prevent significant venous hypertension. The length of required reconstruction determines the extent of distal dissection. Valvulotomy is necessary to allow placement of the vein graft in a nonreversed direction. All infected prosthetic material is removed, septic tissue is debrided, and aortofemoral reconstruction is performed.1 Surviving patients were observed at 6-month intervals with clinical examination and noninvasive testing that included ankle-brachial indexes, toe pressures, and duplex studies of the SFPV reconstruction and remaining deep venous system.
Continuous data are expressed as mean ±SD. The Society for Vascular Surgery/International Society for Cardiovascular Surgery (SVS/ISCVS) standards were used to calculate runoff scores to assess extent of distal disease.6
Included in this study were 12 men and 5 women with a mean age of 58 ± 10 years (range, 40-70 years). Fifteen (94%) of 16 patients were past or present cigarette smokers with a mean of 54 ± 31 pack-years. Other pertinent risk factors included coronary artery disease (n=8, 47%), hypertension (n=11, 65%), diabetes mellitus (n=2, 12%), hyperlipidemia (n=4, 24%), obesity (n=4, 24%), chronic obstructive pulmonary disease (n=3, 18%), and renal disease (creatinine level, >176.8 µmol/L [2.0 mg/dL]) (n=2, 12%).
Information on individual patients is presented in Table 1. The original indication for operation was aortoiliac occlusive disease in 12 patients (71%), aortoiliac aneurysmal disease in 3 (18%), and unknown in 2 (12%). Initial operations consisted of 11 aortobifemoral bypasses (AFBs), 3 AFBs with simultaneous femoropopliteal bypasses (14 total AFBs), and 3 axillobifemoral bypasses. All AFBs had been performed with Dacron prostheses; 2 axillobifemoral and all femoropopliteal bypasses had been performed using expanded polytetrafluoroethylene (ePTFE) prostheses. One axillobifemoral bypass was a Dacron implant. In addition, 4 patients underwent subsequent femoropopliteal bypasses, 3 of these with ePTFE and 1 with in situ greater saphenous vein (GSV).
A mean of 52 ± 59 months (range, 2 weeks to 216 months) elapsed between primary revascularization and documentation of prosthetic infection. Once prosthetic infection was diagnosed, 13 patients underwent multiple procedures in an attempt to eradicate infection and/or restore graft patency (Table 1). The mean number of operations (excluding primary procedure) for these 13 patients was 4.3 ± 2.6 (range, 1-9) and included abscess drainage and wound debridement with local and systemic antibiotics (n=9), EAB with removal of infected AFB (n=8), prosthetic AFB replacement (n=4), thrombectomy (n=5), repair of femoral false aneurysms (n=4), and thrombolytic therapy followed by thrombectomy (n=1). Prosthetic AFB replacements were performed after EAB failure.
Presenting signs and symptoms exemplified the severity of disease. Eleven patients had sepsis at the time of presentation as defined by elevated white blood cell count, fever, and tachycardia. The 5 blood cultures performed yielded positive results; 1 of these patients had diabetic ketoacidosis. In the 9 patients with massive groin infection, 7 had extensive deep infections involving most of the proximal thighs or retroperitoneum; 4 had enterocutaneous fistulae, 2 had necrotizing fasciitis of the lower abdomen and thigh, and 2 had large, expanding, and cellulitic femoral false aneurysms. All 4 patients with enterocutaneous fistulae had contiguous involvement of their aortic prosthesis. Two of these patients had fistulae of the duodenum; 1 patient, of the jejunum; and 1 patient, of the colon. All 4 had gastrointestinal tract fluid and pus draining from the groin. Other presenting factors included rest pain (n=7), gastrointestinal tract bleeding (n=1), exposed prosthetic material (n=2), and psoas abscess (n=1).
Overall, infection of previously placed EABs developed in 11 patients, and 6 of these had coexistent thrombosis. Three axillobifemoral bypasses were performed as primary revascularization procedures in 3 patients with rest pain and/or impending tissue loss; all 3 patients had had a documented myocardial infarction within the previous 6 months. Nine EABs were performed in 8 patients as secondary procedures for treatment of aortic prosthetic infections. These 9 procedures included 4 bilateral axillofemoral bypasses, 1 axillobifemoral bypass, 1 thoracobifemoral bypass, 1 unilateral transobturator iliopopliteal bypass, 1 unilateral axilloprofunda bypass, and 1 unilateral axillopopliteal bypass. In addition, all 7 patients with prosthetic femoropopliteal bypasses had contiguous involvement with infection.
At the time of autogenous SFPV aortic reconstruction, polymicrobial infections were common (n=9), as seen in the following tabulation:
Cultures from 6 patients yielded single organisms only. Final cultures yielded no growth in 2 patients, but clinical signs of graft infection were seen at the time of operation, including nonincorporation of the graft with surrounding turbid fluid. All organisms are listed in the following tabulation:
Bilateral SFPV was harvested simultaneously in 12 patients and sequentially in the 1 patient who underwent 2 separate procedures. Unilateral SFPV was harvested in the remaining 4 patients. Greater saphenous vein was also harvested for supplemental procedures in 7 patients; 1 patient underwent bilateral harvest of SFPV and GSV; 1 patient, contralateral harvest of GSV; and all others, harvest of SFPV and GSV in an ipsilateral lower extremity.
Reconstructions consisted of 7 unilateral SFPV aortofemoral bypasses (6 SFPV femoral crossovers and 1 GSV femoral crossover), 6 SFPV AFBs, 1 SFPV aortofemoral and contralateral aortoiliac bypass with GSV, 1 unilateral aortofemoral bypass, 2 SFPV from prosthetic AFB limb to femoral bypasses, and 1 aortoiliac bypass with iliofemoral crossover bypass (Figure 1).
Fifteen proximal anastomoses were performed in an end-to-end fashion and 2 in an end-to-side fashion. One proximal anastomosis required splicing 2 SFPVs together to fit end-to-end to a large native aorta. Another proximal anastomosis was fashioned by partitioning an aneurysmal native aorta into 2 orifices for separate SFPV end-to-end anastomoses. The supraceliac aorta was the site of 1 proximal anastomosis, whereas all others were located in an infrarenal position.
There were a total of 33 distal anastomoses. Most were to the common femoral arteries (n=13, 39%) or profunda femoris arteries (n=8, 24%). Twenty (61%) of all distal anastomoses were performed in an end-to-side fashion. The mean SVS/ISCVS runoff score was 4.7 ± 2.5, with a minimum value of 1 (normal runoff) and a maximum value of 10 (no runoff).
Within the first 30 days of operation, there were 3 deaths. Two patients died secondary to myocardial infarction and the third patient of respiratory failure. One patient died in the hospital after 30 days due to multiple-system organ failure, for an overall in-hospital mortality of 24%. Eight patients (47%) had significant postoperative complications. Four patients (24%) required 5 amputations (3 above-knee and 1 below-knee amputations and 1 hip disarticulation due to preexisting irreversible ischemia at the time of SFPV reconstruction). Six patients (35%) underwent fasciotomies, with 4 being prophylactic and 2 for compartment syndrome. One patient experienced a postoperative myocardial infarction and acute renal failure requiring dialysis. Acute renal failure that did not require dialysis developed in another patient. Wound infections requiring debridement developed in 4 patients (24%), and 1 of these required gracilis muscle flap for wound closure. Unilateral lower-extremity paresis secondary to distal spinal cord and/or nerve root ischemia developed in 1 patient (6%). One patient (6%) ruptured a pseudoaneurysm at the site of GSV-to-SFPV anastomosis, necessitating ligation of the SFPV reconstruction and above-knee amputation. Multiple-system organ failure developed in this patient, who died 2 months postoperatively. Other complications included pneumonia and respiratory failure in 2 additional patients and graft thrombosis in 1 patient.
Mean follow-up among the 13 surviving patients was 23 ± 21 months. The crude primary patency rate was 92%, with 100% secondary patency rate. Two patients died of comorbid factors greater than 1 year after autogenous SFPV aortic reconstruction. One patient experienced unilateral SFPV limb occlusion at 49 months due to an anastomotic stenosis at an end-to-end anastomosis to the superficial femoral artery. Patency was successfully restored with thrombectomy and anastomotic revision.
There have been no notable long-term venous morbidities, although 6 patients (35%) had transient mild-to-moderate lower-extremity edema treated with compression stockings. Results of physiologic testing demonstrated mild venous outflow obstruction, rich collateral development, and valvular competence in donor limbs.5 No chronic venous insufficiency was seen, as evidenced by the lack of increased skin pigmentation, venous claudication, and venous ulceration.
The patients in this series represent the most desperately ill of those presenting with vascular prosthetic infections. Most had sepsis at the time of presentation; failure due to infection, with or without concomitant thrombosis, of secondary and tertiary procedures; or massive groin, thigh, and lower-abdominal soft tissue infections with deep extension into the thigh along supplementary prosthetic infrainguinal bypasses. Extra-anatomic bypass with staged removal of infected vascular prostheses would be the preferred approach, offering the prospect of less physiologic stress and improved morbidity and mortality.7 However, EAB was not an option in these patients because of previous failure of EABs and the lack of clean, uninfected tissue planes.
Autogenous aortofemoral reconstruction remains the only reasonable option in dealing with this problem.8 We have championed the use of SFPVs for aortofemoral reconstruction because of their resistance to infection from a wide variety of gram-positive and gram-negative bacteria and fungal organisms, their durability and lack of long-term deterioration or aneurysm formation, and their excellent patency in the face of multiple-level occlusive disease and poor runoff.1,2 In a larger series, the 5-year cumulative patency of autogenous aortofemoral SFPV reconstruction was 83% for primary patency and 100% for secondary or assisted patency.2 Other autogenous reconstructions include those fashioned from endarterectomized arterial segments and GSV. We and others have found these reconstructions to be less desirable because their high failure rates due to progressive neointimal hyperplasia.1,2,9
Aortic allograft replacement is another option and has been used successfully to treat aortic prosthetic infections.10- 12 The main advantages of this approach include convenience and a decreased operating time, obviating the need to harvest and prepare autogenous conduits. However, allografts are subject to reinfection and short-term and long-term deterioration, often with disastrous consequences. Many consider autogenous allograft replacement to be a "bridge" procedure until a more definitive reconstruction can be undertaken.9,11
Five patients in this study were included in a larger, previous report.2 Data from these 5 patients were removed from the larger group for analysis of morbidity and mortality comparisons between the groups. The in-hospital mortality in our present subgroup was 24%, 3 times the mortality of 8% in the larger series of autogenous SFPV aortofemoral reconstructions, usually performed as a secondary procedure after infection or failure of a primary AFB.2 In addition, the acute amputation rate was also higher, 24% vs 6%, and other major complications were common in our present subgroup. This is not surprising, given the severity of illness of these patients at the time of presentation and the fact that most had already undergone multiple secondary and tertiary operations to treat aortic prosthetic infection.
This raises another issue relative to reporting vascular prosthetic infections. A wide range of incidences for mortalities and other outcomes have been reported with multiple operative approaches.13 These are heterogeneous patients with varying levels of illness dependent on extent and virulence of infection, severity of underlying occlusive disease, presence and types of medical comorbidities, types of initial aortic reconstruction, presence of additional prosthetic infrainguinal bypasses involved with infection, presence of aortoenteric erosion or fistula, and performance of previous secondary or tertiary operations to treat primary aortic prosthetic infection. These details are frequently lacking in many reports, making it difficult to determine relative effectiveness of various approaches.
Despite the high mortality and morbidity rates among patients in our series, intermediate follow-up outcomes have been good with sustained patency, limb salvage, freedom from reinfection, and minimal lower-extremity venous morbidity.5 This disadvantaged subgroup of patients with aortic prosthetic infections should enjoy the same longer-term outlook documented in the larger series of patients with autogenous SFPV aortofemoral reconstruction.2 We conclude that the SFPV aortofemoral reconstruction is feasible and useful for this condition.
Presented at the 106th Scientific Session of the Western Surgical Association, Indianapolis, Ind, November 16, 1998.
Reprints: G. Patrick Clagett, MD, Division of Vascular Surgery, University of Texas Southwestern Medical Center, 5323 Harry Hines Blvd, Dallas, TX 75235-9157.
Bruce L. Gewertz, MD, Chicago, Ill: I approach this podium with some trepidation. Bill Fry, a terrific surgeon, member of the Western Surgical, as well as father of Will Fry, a new member here today, had the mixed blessing of training both Pat Clagett and me. Bill Fry taught me many things, but one thing he taught me that I have remembered and up to this day have adhered to is never give a paper or a discussion on infected grafts.
The time-tested principles of management of arterial graft infections are very well delineated in this nice presentation. They include removal of prosthetic material, debridement of devitalized tissue, and revascularization as needed. The innovative approach, which was outlined by the Parkland group and Dr Clagett in 1993, accomplishes all 3 of these objectives. The adverse consequences of utilizing superficial femoral veins as conduits, which concerned many of us when this was first reported, seem to be well controlled by support hose. This more focused report centers on a specific group of these patients, the very worst patients with massive soft tissue infections and sepsis. Although the mortality and morbidity are fairly high (24% and 47%, respectively), in truth these results are more than respectable, given the poor condition of the patients involved.
I have the following 3 questions for the authors. They make a valid point that the heterogeneity of patients and the general diagnosis of arterial graft infection makes comparison of series quite difficult. This is emphasized in their manuscript. Can they suggest some sort of classification or stratification of infection that might improve our ability to gauge the severity of disease? Based on their overall experience, what historical or immunologic characteristics might explain why these patients develop aortic graft infections when the vast majority of our patients, more than 98%, are unburdened by this devastating complication? Is groin wound infection at the first operation still the greatest risk factor? Or, are more subtle predispositions at work? One could even suggest that the occurrence of groin wound infection is an effect, not the cause of the problem.
Finally, the authors did an excellent job of documenting the type of organisms involved. When considering their entire experience, was any specific pathogen or combination of bacteria particularly virulent and portending of bad outcome?
Gary W. Lemmon, MD, Dayton, Ohio: I have 3 questions. First, are you continuing long-term antibiotics for these patients? Second, most of these procedures were done for occlusive disease; have you considered performing an aortoiliac endarterectomy rather than an NAIS procedure? Lastly, what do you do when the saphenous vein has either been harvested or previously used in a bypass procedure?
Fred A. Weaver, MD, Los Angeles, Calif: In your overall experience, how often do these grafts develop the intrinsic vein graft stenoses that we see with saphenous vein bypasses for lower-extremity reconstruction?
Dr Claggett: Dr Gewertz reiterates Dr Bill Fry's admonition that presentation of a paper on infected grafts will result in an increase in referrals of patients with this problem. This is exactly what happened following our original presentations and publications dealing with in situ replacement of infected aortic grafts. Not only have the referrals of patients with this problem increased, they have become more complex. We are no longer seeing straightforward patients (if there is such a thing) with aortic graft infections. Most of the patients we now see have very advanced disease and complex situations, and this, in part, spurred us to develop this series.
Dr Gewertz raises the issue of stratification of these patients and whether or not this might be helpful in determining prognoses. This is a very important point. If one looks at the literature on this topic, one is impressed with the heterogeneity of patients in various series as well as the wide variance in morbidity and mortality. Comparing patients in different series and outcomes from these series is like comparing apples and oranges. There are several factors that portend a poor outcome, such as the presence of aortoenteric fistula or erosion, sepsis at the time of initial presentation, extent of prosthetic involvement, status of runoff, the numbers and types of previous vascular operations, and medical comorbidities. All of these are important factors in determining outcomes, and the more risk factors present, the worse the outcome will be. We definitely need a system to stratify these patients. We have reporting standards in other areas in vascular surgery, and we need to also develop reporting standards for this problem.
Dr Gewertz asked if these patients with advanced disease were immunologically disadvantaged. Some of these patients had obvious comorbidities that predisposed them to severe and virulent infections. These included diabetes mellitus, malnourishment, and, in one case, chronic steroid use because of chronic obstructive pulmonary disease. In many patients, it is difficult to sort out whether the patients' multiple operations disadvantaged them and predisposed them to further complications, or whether they had some underlying condition that led to the multiple operations. Dr Gewertz also asked whether distal gangrene or wound infections at the time of primary operation may have predisposed these patients to problems. Most of these patients were referred patients, and information with regard to their primary operations was somewhat sketchy, so I can't answer this question with completeness.
As far as pathogens are concerned, gram-negative infections are generally considered to be more virulent and serious. Some of the gram-negative organisms secrete exoproteases that can actually dissolve autogenous tissue, and this has been a particular concern of ours in developing this reconstruction. However, we have noted no difference in outcomes among the many different types of organisms that we have encountered. Many of the patients in our original series had S epidermidis infections, and these infections are easier to control and eradicate by removal of prosthetic material. However, we have also had many gram-negative infections, including Pseudomonas species. In the present series, most patients had polymicrobial infections with mixed gram-negative and gram-positive organisms.
Dr Lemmon asked about long-term antibiotics. We keep these patients on antibiotics for as long as invasive lines are in, and this is approximately 4 to 5 days. He also asked about the possibility of endarterectomizing arterial segments, such as the native aortoiliac system, and using this for reconstruction. In my experience, the retroperitoneum of these patients is like concrete, and it is almost impossible to dig out the old aortoiliac system intact. Others have reported using endarterectomized vascular segments but have experienced a high incidence of occlusion from neointimal hyperplasia.
The other question deals with the potential problem of harvesting the deep vein when the saphenous vein is absent. In many of our patients, the saphenous vein had been harvested previously, or we have had to use it at the time of this reconstruction for an additional reconstruction such as a distal bypass. We do not hesitate to harvest the saphenous vein when we need it. It is interesting that the venous morbidity is minimal in these patients. We have followed up to 60 patients prospectively for short- and long-term venous morbidity after harvesting the deep vein, and it does not appear to make any difference whether or not the ipsilateral saphenous vein is present. Some of these patients have transient swelling that is controlled with compression stockings, but in fact very few of them need long-term compression stocking treatment.
Dr Weaver asked whether or not these large vein grafts were predisposed to neointimal hyperplasia. They certainly can develop neointimal hyperplasia, but this is less likely to lead to occlusion or failure in comparison with saphenous vein grafts and other smaller conduits. The difference is that these are large conduits, and a little bit of neointimal hyperplasia really makes no difference in patency. We have had significant neointimal hyperplasia at anastomotic sites, and we have had to revise anastomoses in 2 of our patients on long-term follow-up. We do not get the problem of neointimal hyperplasia at valve sites such as one sees with saphenous veins.