Graft associated hemorrhage from femoropopliteal vein grafts| Dr. Ahsan Ali

Graft associated hemorrhage from femoropopliteal vein grafts

Ahsan T. Ali, MD,^aChristopher Bell, MD,^b J. Gregory Modrall, MD,^b R. James Valentine, MD,^b and G. Patrick Clagett, MD,^b Dallas, Tx

Objective: The femoropopliteal vein (FPV) graft has been used extensively for large-caliber vascular reconstructions. To date, there have been no reports of anastomotic dehiscence or rupture leading to graft-associated hemorrhage (GAH). In the present report, we review our experience with GAH from FPV grafts to determine the incidence of this problem, to better understand the etiology, and to determine potential methods to prevent this complication.

Methods: All patients undergoing arterial reconstructions with FPV grafts were entered into a registry that included demographics, operative details, complications, and follow-up information. Episodes of GAH that occurred during the period from 1990 to 2004 were studied to determine etiologic factors and outcomes.

Results: During the study period, 574 FPV grafts were used for arterial reconstructions in 364 patients. GAH occurred in 11 patients (3%). Onset of GAH ranged from 1 hour to 180 days after operation. The mean blood transfusion requirement for GAH was 10 ± 4 units. In three patients, the etiology of GAH was purely technical, resulting in a slipped or “popped” tie from a large side branch. In eight patients, the etiology was due to graft disruption secondary to uncontrolled infection and failure of anastomotic healing. Most of these patients were being treated for aortic graft infection. Special risk factors for this complication included malnutrition, ongoing polymicrobial and fungal infections, immunocompromised state, active cancer, steroid treatment, and ongoing graft contamination from gastrointestinal or pharyngeal leaks. Outcomes included four deaths and one stroke

Conclusions: GAH is a serious complication with high morbidity, mortality, and transfusion requirements. Although technical problems are preventable, FPV grafts, like all biologic grafts, can develop disruption with GAH from ongoing infection, especially in severely immunocompromised patients who are malnourished and have poor healing ability. Strategies for prevention and alternative treatment modalities are appropriate in patients at high risk for GAH. ( J Vasc Surg 2005;42:667-72.)

The femoropopliteal vein (FPV), or deep vein, has been used extensively for arterial and venous reconstructions.^1-9 Because of its large caliber and resistance to infection, most of these grafts have been placed in reoperative and infected fields involving large arteries. FPV grafts have proven to be durable, with excellent long-term patency. To date, there have been no reports of aneurysmal degeneration,6 anastomotic dehiscence, or rupture.

In our large experience with FPV grafts used for arterial reconstructions at multiple anatomic sites, we have encountered occasional graft-associated hemorrhage (GAH) in the postoperative period.GAHhas usually been caused by graft disruption from uncontrolled infection or poor healing at anastomoses. We have also, disappointingly, encountered GAH from technical errors, usually related to poorly placed ligatures on the large side branches of the FPV grafts. The present report details our experience with GAH from FPV grafts used for arterial reconstructions. We undertook this

From the University of Arkansas Medical Center Division of Vascular and Endovascular Surgery,a University of Texas Southwestern Medical Center.^b
Competition of interest: none.
Presented at the Twenty-sixth World Congress International Society for CV Surgery, Maui, Hawaii, March 21-25, 2004.
Reprint requests: G. Patrick Clagett, MD, Division of Vascular and Endovascular Surgery, University of Texas Southwestern Medical Center, 5323
Harry Hines Boulevard, Dallas, TX 75390-9157.
0741-5214/$30.00
Copyright © 2005 by The Society for Vascular Surgery.
doi:10.1016/j.jvs.2005.06.002

review to determine the incidence of this complication, to better understand the etiology, and to determine potential methods to prevent this problem.

METHODS

Study population. All arterial bypasses with FPV grafts from 1990 to 2004 at the University of Texas Southwestern Medical Center at Dallas and affiliated hospitals (Zale Lipshy University Hospital, Dallas Veterans Affairs Medical Center, and Parkland Memorial Hospital) were retrospectively reviewed. These included aortoiliac, aortofemoral, brachiocephalic, carotid, visceral/renal, and lowerextremity reconstructions. Venous reconstructions and arteriovenous grafts for angioaccess were excluded. Data were entered into a registry that included operative and anesthesia records, hospital progress notes, graft microbiology, and follow-up information.
Patients were followed at 3- to 4-month intervals the first year after operation and at 6-month intervals thereafter. Episodes of GAH from FPV grafts were noted and analyzed in further detail that included chart review, interview with surgeons involved in the case, and review of morbidity and mortality conference records. The GAH episodes that were reviewed were classified as being technical or nontechnical in etiology.

Harvest and handling of FPV grafts. Details of the FPV graft harvest and subsequent handling have been published.10 Germane to this report, all branches of the vein are ligated with silk ligatures. In contrast to handling

D’Addio et al

JOURNAL OF VASCULAR SURGERY
October 2005

Table I. Total experience with femoropopliteal vein grafts for arterial replacement

FPV graft reconstruction	Grafts	Patients
Aortoiliac/femoral	451	242
Visceral/renal	40	39
Carotid/brachiocephalic	35	36
Femorofemoral	32	32
Femoropopliteal	15	15
Total	575	364

the greater saphenous vein, ties are applied in close apposition to the deep vein wall. This is because many branches have a thin wall at their origin that often extends onto the caudad portion of the FPV.¹¹ Close apposition of ligatures at this site helps to buttress this thin area with adventitia.

In our previous experiences, when the FPV graft was used in a nonreversed configuration, venous valves were ablated with a valvulotome. We have subsequently encountered stenoses that later developed at valve sites and have noted that routine passage of a valvulotome only partially disrupts these large valves. Because of this, we now completely evert the vein graft and excise venous valves under direct vision

After the FPV grafts are harvested and prepared, they are stored for 30 minutes to 2 hours in a preservation solution consisting of Ringer’s lactate (1 L), heparin (5000 U), albumin (25 g), and papaverine (60 mg) at 4° C.

RESULTS

Over a 14-year period (1990 to 2004), 574 FPV grafts were used for arterial reconstructions in 364 patients. Follow- up ranged from 2 months to 96 months, with a mean follow-up of 56 months. Among living patients, 95% received complete follow-up; the remaining 5% were lost to follow-up. Reconstructions were classified according to anatomic sites (Table I). Most of these reconstructions were performed in the aortofemoral or aortoiliac location. When two FPV grafts were used to fashion an aortic reconstruction, these were considered separately as individual grafts (Table I).

GAH occurred in 11 patients (3%). Details of these patients and their outcomes are shown in Tables II and III. GAH occurred in six male and five female patients with a mean age of 63±4 years. The overall incidence ofGAHfor all FPV grafts was 1.9%. Onset of GAH varied from 1 hour to 180 days after operation (Fig 1). By Kaplan-Meier survival analysis, freedom from GAH was 97% at 6 months and 96% thereafter for up to 5 years, with a standard error of 2%. The mean blood transfusion requirement for GAH was 10 ± 4 units (range, 4 to 17 units).

Fig 1. Graphic representation of postoperative time of onset of graft-associated hemorrhage.

Fig 1. Large anastomotic false aneurysm at site of left iliac FPV graft limb dehiscence from aorto-right femoral FPV graft. This site was also contiguous with leak from duodenorraphy

Technical problems leading to GAH occurred in three patients (Table II). These episodes were all due to a slipped or “popped” tie from a large side branch, and all occurred within 12 hours of operation (Fig 1). Two of these patients presented with cardiac arrest, and one had hypotension and an expanding, pulsatile thigh hematoma. All three patients were rushed back to the operating room, where the bleeding side branch was rapidly repaired. All three patients survived without further complication.

Eight patients had GAH from graft disruption secondary to uncontrolled infection, failure of anastomotic healing, or both (Table III). Time to disruption was from 6 to 180 days (Fig 1). Six aortic reconstructions and two carotid replacements were performed. The carotid replacements were for patients with head and neck cancer. These patients were generally malnourished and immunocompromised (Table III). Mean serum albumen levels near the time of onset of GAH were 2.0 ± 0.8 mg/dL.

Among patients with aortic FPV reconstructions, four developed anastomotic dehiscence that led to GAH (Fig 2), and two had perforation or “blow-out” at a single site of an iliac or femoral limb. The perforations were approximately 1 cm in diameter with irregular borders. It was not clear whether either of these perforations were the site of a side branch ligature or were simply graft wall erosion (no ligatures were found at the site of perforation).

Almost all aortic reconstructions had polymicrobial infections with varying mixtures of gram-negative organisms, anaerobic bacteria, and fungal organisms (Table III).

JOURNAL OF VASCULAR SURGERY
Volume 42, Number 1

D’Addio et al

Table II. Patients with technical etiologies for graft-associated hemorrhage

Age/Sex	FPV graft reconstruction	Postoperative time of GAH	Presentation	Additional blood transfusion (units)	Outcome	ASA score
54 y/M	Femoropopliteal bypass	6 hours	Expanding thigh hematoma	6	Alive with limb salvage	3
59 y/F	Iliofemoral	1 hour	Cardiac arrest	4	Alive with limb salvage	3
64 y/M	Iliofemoral (transobturator)	12 hours	Cardiac arrest	11	Alive with limb salvage	3

FPV, Femoropopliteal vein, GAH, graft-associated hemorrhage; ASA, American Society of Anesthesiologists

The vein graft from one of the patients with perforation was specifically cultured and yielded Escherichia coli, Pseudomonas aeruginosa, and Enterococcus species.

Many of these patients had special risk factors that predisposed them to ongoing infection and an immunocompromised state (Table III). These included metastatic adenocarcinoma in the liver; treatment with high-dose steroids, gold, and methotrexate for severe rheumatoid arthritis; ongoing leak from a duodenorrhaphy in the patient with an aortoenteric fistula; severe malnutrition from refusal of enteral and parenteral alimentation; head and neck cancer; recent radiotherapy or chemotherapy, or both; and a myeloproliferative disorder. Most had a prolonged preoperative time interval between the time of diagnosis of graft infection and definitive treatment with graft excision and FPV replacement (Table III).

Four of these six patients died, two from sepsis and multisystem organ dysfunction and two from exsanguination. The patient with metastatic liver cancer required high above-knee amputation after ligation of the left limb of an FPV graft aortofemoral bypass and subsequently died from sepsis and proximal muscle ischemia

Two patients had GAH from anastomotic dehiscence after FPV graft replacement of the carotid artery for radical head and neck surgery operations in which the carotid artery was encased in tumor. One of these patients had recurrent neck squamous cell carcinoma, and both were alcoholic and malnourished. Despite coverage with pectoralis major muscle flap, oropharyngeal leaks occurred in both of these patients, and the vein grafts were bathed in saliva. Both disruptions and episodes of GAH occurred at about 1 week after operation. Both patients underwent emergency operation and survived, but one had a major hemispheric stroke after ligation of the FPV graft (Table III).

DISCUSSION

FPV GAH is a serious problem that led to four deaths, one stroke, and one lower-extremity amputation in the 11 patients in this report. Fortunately, the overall incidence of GAH is low, being 1.9% for all FPV grafts and 3.0% for all patients. If one excludes technical causes of GAH, the incidence due to ongoing infection and failure to heal is 1.4% of grafts and 2.2% of patients. The latter incidence is a realistic expectation of outcomes from using the FPV graft in hostile environments characterized by a reoperative field, ongoing and uncontrolled infection, poor wound

healing secondary to severe malnutrition, immunocompromised state, and in some cases, irradiated tissues as well as the potential for enzymatic digestion from saliva and bacterial proteases.

All technical causes were preventable and due to inadequate ligation of FPV side branches. In reviewing the details of these cases, it is noteworthy that a slipped or “popped” tie was found on re-exploration. More precisely, the tie became dislodged from a large side branch that was typically broad-based and short, such that the tie had an attenuated hold or purchase on the side branch. Because of the large diameter of the FPV, the wall tension when exposed to arterial pressure is much greater than a comparable side branch on a greater saphenous vein graft, and attention to this seemingly small detail is important, even for experienced surgeons.

Based on this experience, we have modified our technique as follows: (1) all side branches ≥1 mm in diameter are doubly ligated, (2) on side branches ≥ 3 mm in diameter a suture ligature is placed as the second, outer ligature, and (3) side-branch division is carried out at least 2 mm beyond the outer ligature to provide adequate hold or purchase that may compensate for the greater wall tension of these large-caliber grafts. Since making these technical modifications, we have not encountered this complication.

The cases of GAH due to graft disruption from ongoing infection, malnutrition, and poor healing are more problematic. Patients with infected aortic prostheses are commonly elderly, chronically ill, deconditioned, and malnourished, with multiple serious medical comorbidities. However, the patients in the present report were at the extreme end of this spectrum in that most had polymicrobial infections and a prolonged course of medical and surgical treatment for their graft infections before definitive treatment with graft excision and FPV graft in situ reconstruction. Other factors leading to malnutrition, immunocompromised state, and poor healing in the patients with aortic graft infections included metastatic adenocarcinoma, lung cancer, myeloproliferative disorder, treatment with steroids, methotrexate, and gold for severe rheumatoid arthritis, and ongoing duodenal leak after repair of an aortoduodenal fistula.

Three of these patients had infection with Candida glabrata in addition to bacterial organisms. We have found this organism to be highly associated with mortality in our overall series of infected aortic grafts (unpublished obser-

D’Addio et al

JOURNAL OF VASCULAR SURGERY
October 2005

Table III. Patients with nontechnical etiologies for graft-associated hemorrhage

Age/Sex	Time interval between diagnosis of aortic graft infection and definitive treatment	FPV graft reconstruction	ASA score	Special risk factors	Microbiology
69/F	1 month	Aortoiliac	4	1. Active steroid treatment for rheumatoid arthritis. 2. Aortoenteric erosion. 3. Leak from duodenorrhaphy after FPV aortoiliac reconstruction.	Candida glabrata and Klebsiella
52/F	2 weeks	Aortobifemoral	3	1. Morbid obesity (body mass index = 42). 2. Irreversible ischemia in both lower extremities (refused further bypass surgery and ultimately required bilateral BKAs). 3. Profound depression with refusal to eat; refused feeding tube and parental nutrition. 4. Dehiscence femoral wounds.	Candida glabrata and Escherichia coli
59/F	10 years (patient had multiple re-current infected femoral anastomotic aneurysms after Dacron aortobifemoral bypass procedure and had at least 6 femoral limb replacements).	Aortobifemoral	3	1. Postoperative delirium and depression. 2. Malnutrition from refusal to eat and refusal of feeding tube. 3. Failure of primary wound healing.	Candida glabrata and Bacteroides fragilis
69/F	7 months	Aortobifemoral	4E	1. Initial operation performed for ruptured mycotic abdominal aortic aneurysm (Dacron graft replacement). 2. Re-explored 1 month later for retroperitoneal hematoma that grew Staphylococcus epidermidis. Aortic graft left in place.	Methicillin-resistant Staphylococcus epidermidis, Haemophilus parainfluenzae, and Fusobacterium.
54/F	5 months	Aortobifemoral	4	1. Severe rheumatoid arthritis with active treatment consisting of steroids, gold, and methotrexate. 2. Lung cancer. 3. Aortoenteric fistula.	Candida spp
81/M	7 months	Aortobifemoral	4E	1. Chemotherapy for non-Hodgkin’s lymphoma. 2. Myeloproliferative disorder. 3. Metastatic adenocarcinoma to the liver (primary unknown).	Escherichia coli, Pseudomonas aeruginosa, and Enterococcus
65/M	7 months	Aortobifemoral	4E	1. Chemotherapy for non-Hodgkin’s lymphoma. 2. Myeloproliferative disorder. 3. Metastatic adenocarcinoma to the liver (primary unknown).	Escherichia coli, Pseudomonas aeruginosa, and Enterococcus
65/M	N/A	Carotid	4	1. Status post bilateral radical neck dissection and laryngectomy for squamous cell carcinoma of the larynx. 2. Developed carotid blowout from oropharyngeal leak; first treated with coil embolization of external carotid artery and covered carotid stent. 3. Developed recurrent oropharyngeal leak and wound breakdown that was subsequently treated with carotid resection, FPV graft reconstruction, and muscle flap coverage. 4. Recurrent oropharyngeal leak with contamination of vein graft with saliva and oral flora.	Unknown
53/M	N/A	Carotid	4	1. Extensive squamous cell carcinoma involving the larynx. At the time of laryngectomy and bilateral radical neck dissection, carotid involvement was found. Reconstruction with FPV performed. 2. Developed oropharyngeal leak; vein graft contaminated with saliva and oral flora.	Unknown

FPV, Femoropopliteal vein; ASA, American Society of Anesthesiologists; GAH, graft-associated hemorrhage; pRBC, packed red blood cells; BKA, below knee amputation; AKA, above knee amputation
*Normal albumin range is 3.1 to 4.3 mg/dL

JOURNAL OF VASCULAR SURGERY
Volume 42, Number 4

Ali et al

Table III. Continued

Albumen (mg/dL)*	Time of Onset GAH	Additional blood transfusion requirement (units pRBC)	Site of GAH	Outcome
2.4	32 days	17	Dehiscence left iliac limb (Fig 2)	Death
1.7	30 days	10	Exposed, infected right distal limb of aortobifemoral reconstruction and nonhealing femoral wound. Anastomotic dehiscence.	Death
2.1	45 days	N/A (patient died at home)	Femoral wound (probable femoral anastomotic dehiscence).	Death
1.9	23 days	8	Left limb blowout at site of infected retroperitoneal hematoma	Alive and well after ligation left limb and femoral crossover bypass
2.1	1. At 11 days after operation, patient had bleeding from femoral anastomotic disruption that was repaired. 2. At 180 days after operation, patient had aortic anastomotic false aneurysm from ongoing retroperitoneal infection.	7	Femoral and aortic anastomotic dehiscences.	Survived aortic ligation with axillofemoral bypass, but died later from lung cance
2.1	14 days	6	Left limb of aortofemoral reconstruction (blowout).	Initially survived ligation of left limb of aortofemoral reconstruction but required AKA and subsequently died from multisystem organ failure.
1.6	2 weeks	12	Dehiscence of carotid FPV anastomosis.	Ligation of FPV graft and subsequent stroke
3.3	6 days	14	Carotid blowout, site unknown	Replaced with another FPV graft with muscle coverage and subsequent survival.

Ali et al

JOURNAL OF VASCULAR SURGERY
October 2005

vations). It is unlikely that this organism possesses special virulence factors. Rather, we hypothesize that growth of this organism is reflective of a prolonged treatment course with multiple antibiotics in a debilitated and immunocompromised, chronically septic patient with diminished likelihood of survival.

In our earlier reports, we stated that FPV grafts resist infection.^1,3,6 Clearly, this is not absolute; like all biologic grafts, FPV grafts can develop disruption with GAH from ongoing infection, especially when coupled with malnutrition and poor healing ability. We suggest that poor healing is, in part, etiologic in our cases of GAH with aortic graft infection because of the debilitated and malnourished state of these patients and because many had poor wound healing and anastomotic disruption.

Because of these considerations, we have modified our standard procedure for FPV graft reconstruction after removal of an infected aortic graft as follows:

1. Intraoperative anastomotic sites (aortic and iliac limb anastomosis) are covered with omentum whenever possible. This is particularly important in cases of aortoenteric fistula/erosion.
2. We avoid placement of iliac/femoral limbs in grossly purulent tunnels that are left after removal of infected limbs. In this circumstance, new retroperitoneal tunnels are created or FPV grafts are routed around areas of gross purulence.
3. Consideration is given for prolonged intravenous antibiotics (4 to 6 weeks) in patients who are severely immunocompromised and debilitated, particularly if graft cultures yield Candida organisms.
4. Femoral anastomotic sites receive complete tissue coverage after thorough débridement of infected tissues. This may require muscle flap coverage in some instances.
5. In patients at the extreme, disadvantaged end of the spectrum of malnutrition, immunocompromised state, and debilitation, we consider alternative methods to treat aortic graft infection, especially if multiple medical comorbidities are present.^12,13

Attempts to repair anastomotic dehiscences and perforations in the patients with poor healing and on-going infection were, for the most part, futile because of tissue friability and inability to hold sutures. A simple femoral anastomotic bleeding site was successfully resutured (patient 5 in Table III). Another patient had successful FPV graft replacement (patient 8 in Table III). In both cases, contamination was minimal or controlled, and the tissues appeared relatively healthy.

Because we have had excellent results using the FPV graft for carotid and brachiocephalic reconstructions,4 we have extended this experience to include carotid replacement after radical head and neck surgery for cancer encasing the carotid artery. Arteriography and balloon occlusion with nuclear brain imaging are routinely done preoperatively to determine absolute requirements for revascularization.

We have replaced the common and internal carotid arteries with FPV grafts in nine patients over the past 5 years. GAH from anastomotic dehiscence occurred in two patients (28%). All patients had oropharyngeal or laryngeal contamination of the operative field and all underwent preoperative radiotherapy. However, the two patients with GAH in this report were unique in that they had uncontrolled oropharyngeal salivary leaks that communicated directly with FPV grafts despite pectoralis major muscle coverage.

We postulate that enzymatic digestion from salivary secretions coupled with ongoing infection from oral flora led to anastomotic disruption and GAH. Because of this experience, we insist on early exploration to control oropharyngeal leaks, and when this cannot be achieved and the FPV graft continues to be exposed to salivary secretions, ligation should be undertaken if preoperative balloon occlusion testing demonstrates a low risk of stroke.

REFERENCES

1. Clagett GP, Bowers BL, Lopez-Viego MA, Rossi MB, Valentine RJ, Myers SI, et al. Creation of a neo-aortoiliac system from lower extremity deep and superficial veins. Ann Surg 1993;218:239-49.
2. Hagino RT, Bengtson TD, Fosdick DA, Valentine RJ, Clagett GP. Venous reconstructions using the superficial femoral-popliteal vein. J Vasc Surg 1997;26:829-37.
3. Gordon LL, Hagino RT, Jackson MR, Modrall JG, Valentine RJ, Clagett GP. Complex aortofemoral prosthetic infections: the role of autogenous superficial femoropopliteal vein reconstruction. Arch Surg 1999;134:615-21.
4. Modrall JG, Joiner DR, Seidel SA, Jackson MR, Valentine RJ, Clagett GP. Superficial femoral-popliteal vein as a conduit for brachiocephalic arterial reconstructions. Ann Vasc Surg 2002;16:17-23.
5. Modrall JG, Sadjadi J, Joiner DR, Ali A, Welborn MB III, Jackson MR, et al. Comparison of superficial femoral vein and saphenous vein as conduits for mesenteric arterial bypass. J Vasc Surg 2003;37:362-6.
6. Clagett GP, Valentine RJ, Hagino RT. Autogenous aorto-iliac/femoral reconstruction from superficial femoral-popliteal veins: feasibility and durability. J Vasc Surg 1997;25:255-70.
7. Jackson MR, Ali AT, Bell C, Modrall JG, Welborn MB, Scoggins E, et al. Aortofemoral bypass in young patients with premature atherosclerosis: is superficial femoral vein superior to Dacron? J Vasc Surg 2004; 40:17-23.
8. Huber TS, Hirneise CM, Lee WA, Flynn TC, Seeger JM. Outcome after autogenous brachial-axillary translocated superficial femoropopliteal vein hemodialysis access. J Vasc Surg 2004;40:311-8.
9. Schulman ML, Badhey MR, Yatco R. Superficial femoral-popliteal veins and reversed saphenous veins as primary femoropopliteal bypass grafts: a randomized comparative study. J Vasc Surg 1987;6:1-10.
10. Clagett GP. Update on surgical management of infected aortic grafts. In: Pearce WH, Matsumura JS, Yao JST, editors. Trends in vascular surgery 2003. Evanston: Reischling Press, 2003; p. 199-211.
11. Seidel SA, Modrall JG, Jackson MR, Valentine RJ, Clagett GP. The superficial femoral-popliteal vein graft: a reliable conduit for largecaliber arterial and venous reconstructions. Perspect Vasc Surg Endovasc Ther 2001;14:57-80.
12. Sladen JG, Chen JC, Reid JDS. An aggressive local approach to vascular graft infection. Am J Surg 1998;176:222-5.
13. Calligaro KD, Veith FJ, Yuan JG, Gargiulo NJ, Dougherty MJ. Intraabdominal aortic graft infection: complete or partial graft preservation in patients at very high risk. J Vasc Surg 2003;38:1199-205.

Submitted Mar 31, 2005; accepted Jun 3, 2005.

Graft-associated hemorrhage from femoropopliteal vein grafts

Summary