Long term results of the treatment of aortic graft infection by in situ replacement with femoral popliteal vein grafts

      recent meta-analysis from pooled data demonstrated several advantages for in situ graft replacement over extraanatomic bypass.⁶ There are currently three options for in situ graft replacement for AGI: cryo-preserved allograft, antibiotic soaked synthetic graft, and the autogenous femoral popliteal vein (FPV) graft. In an early series with limited follow-up, Rifampin soaked Dacron was shown to be effective treatment for AGI with low virulent and limited S epidermidis infections.⁵ Recently, two large series in which AGI was treated with excision and in situ revascularization with either cryo-preserved allograft or Rifampinsoaked Dacron grafts have reported good results.^7,8

      The third option for in situ replacement is to use femoral popliteal vein grafts. This operation, termed the neo-aortoiliac system (NAIS) procedure represents the only durable autogenous reconstruction. Initially reported by Schulman^9-11 as a conduit for femoral popliteal bypass, FPV grafts were first described for AGI treatment in 1993.¹² In 1997, an expanded series with moderate follow-up was reported.¹³ Since then, the NAIS procedure has been widely practiced and reported with similar results in several small series.^5,14,15

      The purpose of the current study is to provide a comprehensive and robust, long-term follow-up in a large cohort of patients treated for AGI with NAIS reconstruction.

the FPVs at one setting, followed by vein graft harvest, excision of the infected graft, and NAIS reconstruction one day later, as described by Ali et al.21 In the postoperative period, patients were routinely treated with subcutaneous low dose heparin and calf intermittent pneumatic compression for venous thromboembolism prophylaxis.

      Vein grafts were usually oriented in a nonreversed fashion to allow the larger proximal end to be comfortably anastomosed to the aorta. Most proximal anastomoses were end-to-end. However, if the diameter of the aorta greatly exceeded that of the FPV graft, a “pantaloon” configuration was employed whereby two FPV grafts were joined to increase the diameter of the conduit.¹³ Purulent fluid, grossly infected tissue, thrombus, and graft material were routinely cultured for bacteria and fungi by standard microbiologic techniques.

      Data collection and analysis. Charts were reviewed for basic demographics, associated comorbidities, clinical presentation, and culture results. Operative data included total operative time, intravenous fluid, and transfusion requirements and American Society of Anesthesiologists (ASA) patient class. Adjunctive surgical procedures were recorded as either vascular or nonvascular. Vascular procedures included concomitant infrainguinal bypass or a visceral/renal revascularization. Nonvascular procedures included placement of a feeding tube, tracheostomy, muscle flap transfer for coverage in the groin, and duodenal repair or bowel resection.
Patients were divided into three groups according to the location of the infected graft: aortofemoral bypass grafts (AFBG); aortoiliac or aortic tube graft (AIBG); and axillofemoral bypass grafts (AxFBG) grafts. The latter group represented patients who had undergone previous treatment for AGI by AFBG excision and an extra anatomic bypass.
Operative complications were divided into surgical and medical. Surgical complications were specific to the procedure such as limb loss, surgical bleeding, compartment syndrome requiring fasciotomy, mesenteric ischemia, and graft thrombosis. Medical complications included pneumonia, respiratory failure, myocardial infarction, and other complications that might be expected to follow major surgery.
Standard 30-day mortality was recorded. However, because some patients died as a result of the operation later in the hospital course, at a rehabilitation unit or at home, we also assessed procedure-related mortality regardless of the postoperative time span.² ICU time and total hospital length of stay was recorded. Information was also collected about discharge disposition to home, rehabilitation facility, or to a long-term skilled nursing facility.
The standard follow-up of patients who were geographically close consisted of duplex graft surveillance and clinical examination every 4 months during the first year and then every 6 to 12 months thereafter. If indicated, imaging was obtained using either a conventional arteriogram or a CTA. For out of state patients, follow-up was done through a telephone inquiry to the patients or the

Table III. Morbidity and mortality

Total (n = 187%)

• Mortality

• 30-day (n)19 (10%)

• Procedure related (n)27 (14%)^a

• Discharge disposition

• Home99 (53%)

• Rehab facility43 (23%)

• Nursing home18 (10%)

• Length of stay

• (mean)21 ± 8

• (median) 14

• Surgical complications

• Wound complications63 (34%)

• Fasciotomy46 (25%)

• Amputations14 (7.4%)^a

• Surgical bleeding10 (5.3%)

• Mesh closure (abdomen)8 (4.3%)

• Mesenteric ischemia3 (1.6%)

• Delayed graft disruptions10 (5.3%)

• Medical complications

• Pulmonary 25 (13%)

• Tracheostomy 17 (9%)

• Acute renal failure 22 (12%)

• Coagulopathy 22 (12%)

• Myocardial infarction 8 (4.3%)

• Stroke3 (1.6%)

• Paralysis3 (1.6%)

• Pulmonary embolism2 (1.1%)

---

^aProcedure related mortality includes death beyond 30 days.

Immediate outcomes. The 30-day mortality was 10% and the procedure-related death rate was 14% (Table III). Details for all deaths are listed in Table IV. The mean length of stay was 21 ± 8 days, however, the median length of stay was 14 days. Out of 187 patients, 27 (14%) did not survive, 99 (53%) were discharged home, 43 (23%) went to a rehabilitation facility and subsequently were discharged home, and 18 (10%) were transferred to a long-term skilled facility.

Limb loss occurred in 14 (7.4%) patients. Five had irreversible ischemia on presentation; four had no autogenous conduit in the presence of grossly infected wounds; two had ligation of the FPV limb as a result of postoperative graft disruption; and one each had limb ischemia from compartment syndrome with extensive myonecrosis, failed distal bypass, and cardiac embolism.

Other complications are listed in Table III. Wound complications occurred in 63 patients (34%). These included all patients who required surgical treatment for groin wounds and vein harvest sites that were treated with muscle flap for coverage, incision and drainage, skin grafting, and intravenous antibiotic therapy. Almost half (44%) of these patients had a purulent groin sinus or septic femoral wounds on presentation. Acute postoperative bleeding occurred in 10 patients (5%) necessitating return to the operating room for control. Two patients had acute postoperative bleeding caused by a slipped tie from one of the FPV grafts.

Table V. Microbiology

• No growth31 (17%)

• Single organism81 (45%)

• Multiple organism67 (37%)^a

• No data available7 (4%)

• Data available180 (96%)

• Type^a43 (23%)

• Gram positive100 (55%)

• Gram negative57 (32%)

• Anaerobes24 (13%)

• Fungal 32 (18%)

---

^aProcedure related mortality includes death beyond 30 days.aMutually exclusive with more than one type of bacteria was cultured in 37% of the patients

during long-term follow-up occurred because of progression of distal occlusive disease. Overall survival at 5 years was 52% with most patients dying of cardiovascular disease (Fig 3). Patients who presented with AE fistula did worse compared with those without AE fistula for perioperative period and on long term follow-up (log-rank P = .006, Fig 4). Multivariate independent predictors of death or poor out come were reported for the immediate, perioperative period and 12 months after surgery (Table VI). The perioperative factors included microorganism: Candida species with Candida glabrata in particular, Klebsiella pneumoniae, and Bacteroides fragilis infections. Additionally, perioperative sepsis was a predictor of poor

Fig 3. Survival. The survival at 5 years is 52.4%. The numbers at risk are noted at each time point. Standard error (in parentheses) at 6 years is 10%.

Fig 4. Survival for patients with aortoenteric (AE) fistula. Survival differed significantly between patients with and without AE fistula (18.4% vs. 56.7%). Standard errors (in parentheses) exceeded 10% at 7 years for patients without AE fistula and all time points for patients with AE fistula.

DISCUSSION

morbidity (CEAP Class C₃-C₅) on clinic examination at a mean follow-up of 70 months.^26,27 In that study, we described two limbs with more advanced venous morbidity. One limb had brawny edema (CEAP Class C₄) and one limb with a healed ulceration (CEAP Class C₅). To date, those two limbs are the only limbs with CEAP Class C4 or greater venous morbidity in over 400 limbs from which FPV has been harvested for a variety of operations at our institutions. Although early postoperative swelling of harvested limbs is common, swelling typically subsides over the ensuing weeks. It is rare that we need to prescribe compression stockings at late follow-up of patients after FPV harvest. In addition to NAIS procedures for AGI, we and others have demonstrated the beneficial features of this large caliber vein graft in multiple vascular beds.^9,28-31
Acute venous morbidity resulting in the need for fasciotomy remains a problem with harvesting FPVs in some patients. In a previous report, we demonstrated that preexisting advanced lower extremity ischemia, concomitant harvest of ipsilateral GSV, and excessive crystalloid administration during operation were independent risk factors for compartment syndrome and the need for fasciotomy.³² The overall rate of fasciotomy in the current series was 25%. This high rate was due to multiple factors. Most of the fasciotomies were prophylactic and not due to the development of compartment syndrome. They were carried out at the end of the NAIS procedure because of asymmetric calf fullness or other subjective findings. Although we generally have a low threshold to perform fasciotomy at our institution, it is the senior author’s opinion that many of the fasciotomies in this series were unnecessary. It is noteworthy that no fasciotomies were performed in the series of NAIS reconstructions for AGI reported by Nevelsteen et al.^14,15

Other factors that contributed to the high rate of fasciotomy in this series were that most of the patients had occlusive disease (many had advanced ischemia) and that anesthetic techniques rooted in a philosophy of administering large amounts of crystalloid were prevalent earlier in the series. We have recently demonstrated an overall reduction in complications following the NAIS operations by restricting the amount of intravenous crystalloids administered.³³ The rate of fasciotomy has declined in recent years to approximately 12%, and this is most likely due to improved anesthetic techniques and more stringent criteria for the performance of fasciotomy

Much has been made of the postulate that the NAIS procedure is long, technically demanding, and may require two teams. At our institution, we perform this operation with a single team that most often consists of the attending surgeon, vascular fellow, a second or third year general surgery resident, and a medical student. As we document in this report, the operation is long taking 7 to 11 hours on average. This could possibly be shortened with two separate surgical teams (one harvesting veins while the other excises the aortic graft); however, we have not found this necessary or feasible in our practice. Ali et al recently reported shortening the NAIS operation by staging FPV harvest.21 The femoral veins are dissected and left in situ and the aortic graft excision and NAIS reconstruction are carried out 24 hours later.

As far as the technical demands, they are well within the capability of a well-trained vascular surgeon. Throughout our experience, we have noted no “learning curve” in performing this operation. It is noteworthy that our overall mortality has not decreased over time. In fact, it has actually increased. In our previous report, the mortality for NAIS for AGI was 8% and currently it is 10% for 30-day mortality and 14% for procedure-related mortality. This change most likely reflects the more realistic appraisal of mortality that is realized with a larger, more robust series. In addition, we have subjectively noted that referral of sicker patients has