Contemporary Outcomes of Thoracofemoral Bypass

Abstract

Objectives:

Thoracofemoral bypass (TFB) is an alternative to aorto-femoral (AFB) or extra-anatomic bypass for severe aorto-iliac occlusive disease (AIOD). TFB may be particularly useful in select patients with concurrent visceral aortic branch vessel disease, infrarenal aortic occlusions or after failed AFB. However, there are few contemporary series describing the indications and outcomes for TFB. Therefore, the purpose of this analysis was to review our experience with TFB.

Methods:

All patients undergoing TFB for occlusive disease from 2002-2017 were reviewed. All subjects underwent left thoraco-retroperitoneal exposure of the supra-celiac aorta with division of the diaphragmatic crus and supra-celiac cross-clamping. An end-to-side aortic anastomosis was created and each graft limb was tunneled in the retroperitoneum to the femoral bifurcation. Adjunctive visceral/infra-inguinal revascularization was selectively performed based on symptoms, end-organ function and/or preoperative imaging. The primary end-points were major complications and 30-day mortality. Secondary end-points included limb patency, freedom from major adverse limb events (MALE), and survival. Kaplan-Meier methodology was used to characterize end-points.

Results:

Forty-one patients [age: 61±9 years; female-54%, hypercoaguable state-7%] underwent TFB. Mean preoperative ankle brachial index (ABI) was 0.4, bilaterally. Indications included: critical limb ischemia (56%), claudication (30%), acute limb ischemia (7%) and combined AIOD+mesenteric ischemia (7%). Seven (17%) had previously undergone AFB and 15(38%) underwent any prior aortic operation. Adjunctive visceral bypass occurred in 8(20%) (14 grafts: renal-6, superior mesenteric artery-5, celiac-3). Postoperative LOS was 11 [IQR 7, 16] days and 30-day mortality was 5% (n=2). Major complications occurred in 34% (n=14; pulmonary-15%, cardiac-12%, bleeding-7% [accidental splenectomy-5%], renal-5%, wound-2%). Mean postoperative ABI was 0.9, bilaterally.

At median follow-up of 7[IQR 1, 17] months, 5(12%) patients underwent some form of reintervention (graft/limb related-4[graft thrombosis-2, graft infection-2], mesenteric bypass revision-1). The estimated 3-year primary limb patency and freedom from MALE were 80±10%, and 70±10%, respectively. Estimated 5-year survival was 93±5% [median 27.3(14.5, 35.2); 95%CI 17.9-32.8].

Conclusions:

This experience represents one of the largest and most current series of retroperitoneal TFB. We demonstrate that TFB can be performed with good outcomes for patients with severe AIOD, especially if concomitant visceral/infrainguinal reconstruction is warranted. These results support a continued role for TFB in select patients.

TOC summary:

This retrospective single center study of 41 patients who underwent thoracofemoral bypass for aortoiliac occlusive disease found a 30 day mortality and morbidity of 5% and 34% respectively, a 3 year graft patency of 80%, freedom from major adverse limb events of 70% and a 5 year survival of 93%.

Introduction

Aortobifermoral bypass(AFB) remains the gold standard treatment for complex aortoiliac occlusive disease(AIOD). However, 10-20% of patients will experience some mode of graft failure which can manifest as limb stenosis, thrombosis, pseudoaneurysmal degeneration and/or graft infection^1–4. Within this cohort, there is a subset of patients(1-3%) that present with bilateral limb thrombosis^{5, 6}. These subjects can be managed with a variety of remedial procedures including thrombolysis, angioplasty/stenting, thrombectomy, extra-anatomic revascularization or redo aortic surgery. Although redo AFB can be performed safely in good risk patients^{7, 8}, some present with severe pararenal aortic neck calcification and/or thrombus that precludes the ability to safely construct a juxtarenal anastomosis^9–11. Similarly, certain patients have severe native pararenal and/or mid-visceral aortic atherosclerotic disease that, due to clinical symptoms and/or anatomic proximity, mandates concurrent reno-mesenteric arterial reconstruction if in-line aortic repair is attempted^9–11. In these scenarios, an alternative to primary or (redo)AFB is thoracofemoral bypass(TFB).

Few contemporary reports of TFB outcomes exist^{12, 13} and most commonly, historical series highlight use of a left anterolateral thoracotomy with a descending thoracic aortic end-to-side anastomosis and blind retroperitoneal tunneling through the diaphragm^14–18. This approach may lead to pulmonary, diaphragmatic and/or tunnel related complications. An alternative strategy involves use of a left thoraco-retroperitoneal incision with division of the left diaphragmatic crus to facilitate the supraceliac aortic anastomosis¹⁹. This technique obviates need anterolateral thoracotomy, facilitates direct tunneling and enables concurrent renal-mesenteric revascularization if indicated. However, there is limited data describing this specific approach or the associated outcomes.

Due to the paucity of data on indications and outcomes particularly in the endovascular era, we sought to describe our results and define utility of TFB.

Methods

Database and subjects.

After approval from the Institutional Review Board(IRB # 201701847), a retrospective review was performed at the University of Florida to identify all patients undergoing TFB from January 2002 to August 2017. The need for informed consent was waived since no direct patient contact and/or harm occurred. Patients were identified from a prospectively maintained, institutional procedural database. Elective and non-elective patients undergoing TFB for any peripheral arterial occlusive(PAD) indication were reviewed. However, patients undergoing TFB with or without antegrade visceral debranching for mycotic indications(e.g. prior aortic ligation with failed extra-anatomic bypass or subjects with mid-visceral aortic infections) were excluded.

Data collection and definitions.

Patient demographics, comorbidities, previous vascular operations, non-invasive vascular laboratory data, indications, post-operative outcomes and need for re-intervention or amputation were documented through chart review of the electronic medical record(EPIC, Verona, Wisconsin, USA). Operative details were obtained from the operative note including aortic exposure technique, need for adjunct procedures(defined as any concomitant visceral or renal bypass, re-implantation or endarterectomy, profundaplasty or infrainguinal bypass), conduit type and configuration, aortic cross-clamp location, blood loss, intraoperative crystalloid and blood product resuscitation.

Patient comorbidities and severity²⁰ were defined as any prior history of hypertension(any antihypertensive drug), coronary artery disease(angina, prior coronary artery bypass or percutaneous coronary intervention), chronic obstructive pulmonary disease(smoking history >20 pack-years, abnormal pulmonary function testing, medications), diabetes mellitus(oral hypoglycemic, insulin), congestive heart failure(New York Heart Association class II or greater), chronic renal insufficiency(creatinine concentration > 1.8mg/dL or dialysis dependence), and dyslipidemia(chart history or medication). Peripheral arterial disease(PAD) severity was categorized using ankle-brachial index(ABI), toe-brachial index(TBI) and reporting guidelines²¹. Similarly, complications were defined (Appendix Table) and graded based upon reporting standards^21–23. Graft patency was determined by the presence of a femoral pulse on examination and stable ABIs during follow-up(i.e. no change>0.15).

Clinical practice.

In our practice, TFB is used predominantly for severe(TASC IID) AIOD patients with extensive thrombo-occlusive disease and one or more of the following anatomic features: 1) severe, bulky, circumferential calcification at the level of the renal arteries(‘coral reef plaque’²⁴), 2) patients with ‘coral reef’ aorta and concomitant visceral occlusive disease or 3) patients with prior failed AFB who do not have suitable residual juxta-renal aortic anatomy to safely construct an anastomosis[either due to pararenal aortic atherosclerotic disease and/or presence of an anastomosis immediately below the renal arteries]. Most commonly, patients had chronic juxtarenal aortic occlusion with mid-visceral aortic bulky, circumferential calcification. If patients underwent concurrent renal bypass, the indication typically was due to presence of a failing renal stent and ‘coral reef aorta. Mesenteric revascularization typically occurred in subjects with acute/chronic mesenteric ischemia with flush visceral aortic branch vessel occlusion(or failed stent/not amenable to endovascular revascularization) and coral reef aorta with inadequate infrarenal aorto-iliac donor vessels for retrograde bypass. All patients undergo high-resolution(thin-cut, <2mm) arterial phase contrasted computed tomographic angiogram(CTA) to determine suitability of the supra-celiac aorta for cross clamping, as well as to assess visceral occlusive disease severity and infrainguinal runoff.

Patients requiring concomitant infrainguinal bypass to achieve adequate lower extremity reperfusion also undergo vein mapping. Patients underwent extensive preoperative evaluation which includes chest radiography, electrocardiography and transthoracic echocardiography(TTE), as well as basic laboratory studies. Pulmonary function testing is reserved for patients with oxygen(and/or oral steroid) dependent chronic obstructive pulmonary disease. Cardiology consultation is obtained at the operating surgeon’s discretion and is frequently influenced by the patient’s preoperative functional capacity(<4 METS), presence of abnormal ejection fraction on TTE, and/or presence of cardiac symptoms.

Postoperatively all patients are recovered in a dedicated cardiovascular surgical intensive care unit and subsequently transitioned to a dedicated cardiovascular ward. Care transitions are determined by the operating surgeon’s recommendation. Postoperative surveillance includes follow-up at 2 weeks followed by 1, 6 and 12 months with physical examination and ABIs. Patients with visceral and/or infrainguinal bypass grafts undergo concurrent surveillance with duplex graft ultrasound performed at 1-month, 3-months, 6-months, 12-months, 18-months, 24-months and then annually, thereafter. Need for re-intervention was determined based upon surgeon judgment.

Surgical technique.

Femoral exposure is completed using standard technique prior to the aortic exposure. TFB procedures are preferentially performed via left thoraco-retroperitoneal(RP) exposure using a curvilinear posterolateral incision from the midline halfway between the umbilicus and pubis to the 8^th or 9^th intercostal space. Exposure is facilitated with the patient positioned in a right lateral decubitus position on a bean bag with the bed flexed between the anterior superior iliac crest and caudal margin of the ribs. The hips are left flat to facilitate femoral artery access and the upper torso is rotated ~30 degrees with the left shoulder elevated and immobilized on an arm rest(Figure 1A).

Figure 1. Positioning and Anatomical Exposure for Left Thoraco-Retroperitoneal Thoracofemoral Bypass.

A, Depiction of the patient positioning for thoracofemoral bypass surgery through a left retroperitoneal aortic exposure. Note (inset) that the hips are left as flat as possible to allow access to the right femoral incision, and the patient is placed over the kidney break to facilitate maximal separation of the iliac crest and costal margin. B, Demonstration of the end-to-end proximal anastomosis of a left retroperitoneal thoracofemoral bypass that is constructed in a supraceliac location immediately above the celiac trunk. The left crus of the diaphragm has been divided to facilitate exposure and construction of the anastomosis. The Lambert-Kay side biting, partially occlusive aortic clamp is especially useful in this location. C, The retroperitoneal tunnels are created by reflecting the viscera to the right and performing blunt dissection between the right and left femoral incisions. The main body of the prosthetic graft is left as long as possible to provide enough graft length to reach the femoral incisions. The right retroperitoneal tunnel can occur in an anatomic plane if the patient has not previously undergone prior aortic surgery and if the body habitus allows, otherwise, the limb is tunneled cephalad to the bladder in the pre-peritoneal plane and allows the limb of the graft to course in a long, gentle arc. Importantly, the limbs are tunneled in a retro-ureteral position if attempting to place them in an anatomic position (on the Right side); however, if it is a re-operative field, the both limbs are placed in the pre-peritoneal plane with the right limb having a relatively long, gentle arc-like configuration.

The rectus abdominus is preserved and mobilized medially, the oblique muscles are divided and the retroperitoneal space is bluntly dissected starting at the inferior margin of the 8^th or 9^th rib. The viscera including the spleen, left kidney, gonadal vessels and left ureter are reflected to the right and a fixed retractor system is used to maintain exposure. The lumbar vein is identified and routinely divided to gain exposure to the juxtarenal aorta and also serves as an important landmark for the left renal artery. Next, dissection proceeds cranially and the left crus of the diaphragm is divided to expose the supra-celiac aorta. If visceral revascularization is anticipated(e.g. concomitant flush visceral aortic vessel occlusion/failed mesenteric stent with tandem, severe TASCIID AIOD with bulky, circumferential pararenal calcification) then the targeted visceral vessel(s) is exposed in preparation for reconstruction. In such cases, typically a 6 or 8mm Dacron graft is sewn to the main body of the TFB graft prior to performing the proximal aortic anastomosis. The main body of the graft is kept intact(i.e. not shortened) to maintain an adequate length to reach the groin for the femoral anastomosis. The supraceliac aorta is then assessed manually to confirm suitability for clamping and sewing.

The aortic anastomosis is typically performed using a partial-occluding, side-biting Lambert-Kay clamp on the supraceliac aorta(Figure 1B). An end-to-side anastomosis is fashioned using a collagen-impregnated double velour prosthetic graft(Hemashield; Atrium, Hudson, NH). The limbs of the TFB graft are tunneled through the retroperitoneum to both groins anterior to the iliac arteries(Figure 1C). Tunnels are created bluntly with one finger below the inguinal ligament from the groin incision and the other along the iliac artery from the retroperitoneal incision. In cases where patients have had a prior AFB then the right limb is tunneled in a more anterior, pre-peritoneal plane cephalad to the bladder. The femoral anastomosis is routinely extended onto the profunda femoris artery with concomitant profundaplasty if required.

Statistical analysis.

Data was managed and retained in a Microsoft Excel database(v. 14; Microsoft Corp., Redmond, WA, USA). Primary endpoints included 30-day mortality and postoperative morbidity. Secondary endpoints were all-cause mortality, need for re-intervention, and major averse limb events(MALE)²². Re-intervention was defined as any bypass related endovascular or open reoperation. MALE was defined as any major vascular re-intervention including thrombectomy, thrombolysis, bypass related jump/interposition graft revision, and/or limb amputation occurring within 1-year. Kaplan-Meier product-limit estimates were used to estimate freedom from secondary end-points. Fisher exact and Mann-Whitney tests were used to compare nominal and continuous categorical variables when appropriate. All deaths were verified using the Social Security Death Index Masterfile. The R-statistical software package(Vienna, Austria; V.2.15.0) was used for all analyses. A P-value <.05 was considered significant.

Results

Patient population.

During the study interval, 534 patients underwent either an aortic or extra-anatomic reconstruction for a PAD indication. Of these subjects, 346(64%) received some form of a primary or redo aortic operation while the remainder received extra-anatomic revascularization. We identified forty-one patients(12% of aortic PAD operations) who underwent TFB from 2002-2017. Mean age was 61±9 years, 54% were female and a majority had hypertension, dyslipidemia as well as prior/current smoking history(Table I). Mean preoperative ankle brachial index(ABI) was 0.4, bilaterally. Indications for TFB included critical limb ischemia(56%), short-distance claudication(30%), acute limb ischemia(7%) and mesenteric ischemia with concomitant AIOD(7%). Seven(17%) subjects had previously undergone AFB bypass and 15(38%) underwent any prior aortic procedure including open aortic surgery and/or prior endovascular aortic aneurysm repair(n=1).

Table I.

Demographics, comorbidities and presentation of thoracofemoral bypass patients

Feature (n=41)	No. (%)
Demographics
Age (years, mean±SD)a	61±9
Female	22 (54)
Body mass index (±SD)	26.2±6.2
Comorbidities
Coronary artery disease	16 (40)
Congestive heart failure	5 (12)
Hypertension	38 (93)
Dyslipidemia	33 (80)
Chronic pulmonary obstructive disease	15 (37)
Diabetes mellitus	9 (22)
Prior stroke/TIAb	13 (32)
Tobacco use (current/former)	38 (93)
Composite total (mean)	3.5±1.5
SVSc comorbidity score	3.0±1.4
Preoperative statin use	33 (80)
Preoperative chronic (>30 day) beta-blocker use	25 (61)
Presentation
Preoperative ABId (R/L, mean±SD)	0.41±.22/0.39±.26
Elective	26 (63)
Urgent/Emergent	15 (37)

^aSD, standard deviation;

^bTIA, transient ischemic attack;

^cSVS, Society for Vascular Surgery

^dABI, ankle brachial index

Intraoperative details.

A majority of patients(59%;n=29) underwent some form of additional adjunctive procedure at the time of the TFB. Adjunctive visceral bypass was performed in 8(20%) patients using a total of 14 visceral grafts. Visceral bypasses were created to 5 left renal arteries(1-right), five superior mesenteric arteries and three celiac axis vessels. In addition, 14 patients(34%) underwent concurrent infrainguinal and/or hypogastric artery reconstructions including: 7(17%) adjunctive femoral reconstructions, 6(15%) iliofemoral thromboendarterectomy and profundaplasties and one(2%) hypogastric artery bypass. Median blood loss and intraoperative packed red blood cell transfusion was 800mL(IQR 600, 1425) and 1 unit(IQR 0, 3), respectively. Autologous red blood cell scavenging was utilized in all patients and median auto-transfusion volume was 225mL(IQR 0, 362) (Table II).

Table II.

Operative details of thoracofemoral bypass

Feature (n=41)	No. (%)
Any intraoperative adjunct	24 (59)
-Visceral revascularization	8 (20)
Total number of visceral grafts	14
-Femoral revascularization	7 (17)
SFA bypass/jump graft (prosthetic)	5
Femoral-femoral bypass (femoral vein)	2
-femoral endarterectomy +/− profundaplasty	6 (15)
-incidental splenectomy	2 (5)
-hypogastric bypass	1 (2)
Estimated blood loss, median [IQR]a	800 [600, 1425]
Intravenous fluid resuscitation	3800 [3000, 5000]
Auto-transfusion of scavenged red blood cells	225 [0, 362]
Intraoperative packed red blood cell autologous transfusion	1 [0, 3]

^aIQR, interquartile range

Outcomes.

Postoperative length of stay(LOS) was 11[IQR 7, 16] days and 30-day mortality was 5%(n=2). One subject died in-hospital due to decompensated congestive heart failure leading to respiratory failure while the second patient died after suffering a stroke that resulted in a respiratory arrest. Major complications occurred in 14 patients(34%). These included pulmonary(15%), cardiac(12%), bleeding(7%), renal(5%) and wound related complications(2%). The rate of accidental splenectomy from iatrogenic injury was 5%. Mean ventilator days was 1±.8 while the median ICU LOS was 4(IQR 3, 5) days. A majority of patients were discharged to home with outpatient physical therapy(74%)(Table III). Among the subgroup of patients that underwent TFB and an adjunctive revascularization procedure(n=18;e.g. visceral/renal and/or femoral reconstruction), their outcomes did not differ significantly from subjects undergoing TFB without adjunctive revascularization: 30-day mortality- TFB+adjunct revascularization, 11% vs. TFB alone 6%(p=.6); any complication, 44% vs. 31%(p=.5); LOS 14±7 vs. 13±8 days(p=.7); MALE, 33% vs. 19%(p=.4).

Table III.

Outcomes of thoracofemoral bypass

Feature (n=41)	No. (%)
30-day mortality	2 (5)
Major complication (any)	14 (34)
-Clavien-Dindo classification
I/II	5 (12)
III	8 (20)
IV	0
V	2 (5)
ICUa days, median [IQR]b	4 [3, 5]
Ventilator days	1 [0, 1]
Length of stay	11 [7, 16]
Discharge disposition
-Home	29 (74)
-Rehabilitation unit	10 (26)
Postoperative ABIc (R/L, mean±SD)	0.93±.31/0.92±.28

^aTIA, transient ischemic attack;

^bIQR, interquartile rang;

^cABI, ankle brachial index

The median clinical follow-up of time was 7[IQR 1, 17] months and 5(12%) patients underwent some form of re-intervention. Four graft/limb related complications occurred which included two patients with graft thrombosis(treated by 1- femorofemoral bypass, 1-axillofemoral bypass) and two who experienced graft limb infection (both managed with axillo-bi-femoral bypass with composite autogenous and cadaveric femoral vein, sub-total graft explant and aortic stent-graft exclusion of the residual proximal anastomotic Dacron cuff). They remain reinfection free after 14 months. One patient successfully underwent a retrograde SMA bypass for SMA graft limb occlusion ~16 months postoperatively. Mean postoperative ABI was 0.9±.3, bilaterally.

The one and 3-year primary limb patency(Figure 2) was 94±4% and 80±10%, respectively. Similarly, freedom from MALE(Figure 3) at one and 3-years was 83±7% and 70±10%. A single limb amputation occurred during follow-up and it was in a subject who presented with acute aortic occlusion resulting in significant postoperative lower extremity muscle necrosis despite a patent bypass. Three year amputation free survival for CLI patients(n=29 of 41;71%) was 86±6%(100%-claudication). Median survival time was 27.3 months[IQR (14.5, 35.2);95% CI 17.9-32.8]. A total of three patients died during follow-up(two within 30-days; third patient died 6 months postoperatively from myocardial infarction). The corresponding estimated 1 and 5-year survival after TFB was 93±4% and 93±5%, respectively(Figure 4).

Figure 2.

The Kaplan-Meier curve demonstrates the freedom from any graft related reintervention. The standard error of the mean exceeds 10% at 1.3 years.

Figure 3.

The freedom from major adverse limb event after thoracofemoral bypass is depicted in the figure. MALE included any open/endovascular remediation of the thoracofemoral bypass and/or limb amputation within one year. The standard error of the mean exceeds 10% at 1.3 years.

Figure 4.

The overall estimated survival for thoracofemoral bypass is highlighted in the figure. Notably, the 5-year estimated survival was 93±5% [95%CI 17.9-32.8]. The standard error is less than 10% for all displayed intervals.

Discussion

This current series of TFB offers a unique contribution to the existing literature since it provides detailed descriptions of patient selection, technical conduct and postoperative outcomes of a relatively rare aortic operation during an era when open surgery is increasingly being replaced by endovascular procedures. Despite TFB being a complex operation as evidenced by features of the intra-operative resuscitation, occurrence of adjunctive procedures and prolonged LOS, the overall morbidity and mortality are acceptable given the anatomic complexity and indications in the patient cohort. Re-intervention and MALE rates were modest, however, limb salvage and survival were both excellent. These results further establish a role for TFB in appropriately selected patients with complex AIOD and prior failed AFB and/or concurrent renal/mesenteric occlusive disease. The significance of these findings is underscored by the fact that our study represents the largest to date highlighting utility of left thoraco-retroperitoneal aortic access during TFB.

AFB bypass has long been established as the gold standard treatment for severe AIOD due to its well documented applicability to most anatomic patterns of disease, modest morbidity, low mortality and overall superb durability. Specifically, elective major perioperative morbidity rates of 10-30% and mortality outcomes of 1-4% are routinely reported^{4, 25}. Moreover, the excellent hemodynamic impact of AFB bypass has resulted in some of the most durable outcomes for any vascular operation with primary limb patency rates of 76-95% at 5 years and 75-85% at 10 years^{26, 27}. These outcomes represent the benchmark that all other operations for complex AIOD should be compared. Importantly, the results of this analysis compare favorably with the outcomes of primary AFB with respect to postoperative complications and mortality.

Despite AFB having superior long-term outcomes for severe AIOD compared to other endovascular and/or hybrid(e.g. iliofemoral endarterectomy with retrograde iliac angioplasty/stent placement) strategies^28–30, there are specific scenarios where this approach may not be the optimal revascularization choice even in good risk patients. Specifically, these situations include patients presenting with TASC IID AIOD and severe juxtarenal/suprarenal aortic calcification or failed AFB with marginal residual native juxta-renal aorta for reconstruction. Notably, another important consideration for when AFB may not be the optimal choice for revascularization is in the unique subset of patients presenting with complex AIOD and concurrent renal and/or mesenteric atherosclerotic disease.

Prevalence studies have documented significant(>50%) renal artery stenosis in 22-59% of PAD patients with up to 50% of these subjects experiencing disease progression(>60% and/or occlusion) resulting in potentially clinically relevant sequelae^{31, 32}. Similarly, the association of mesenteric arterial occlusive disease with PAD is well described with incidence approaching 20% on duplex screening of patients over age 65^{33, 34} While clinically significant mesenteric arterial occlusive disease occurs at a lower frequency in these subjects, they often have significant atherosclerotic plaque within the aortic lumen that encroaches and/or obstructs the proximal mesenteric vessels that may require management during open juxta/suprarenal aortic surgery. In the most severe presentation, these lesions may manifest as a coral reef aorta²⁴ characterized by extreme bulky calcific plaque in the juxtarenal and suprarenal aorta. In all these scenarios, attempting primary or even redo-AFB may be dangerous leaving the remaining revascularization choices to various permutations of extra-anatomic reconstruction and/or hybrid strategies (e.g. renal/mesenteric endovascular intervention with extra-anatomic/axillo-femoral bypass).

An alternative aortic operation to AFB in these cases is TFB, which has comparable magnitude and outcomes as highlighted in this analysis. Historically, TFB was most often used as a tertiary option in our practice(initial operation, AFB; second operation, redo AFB; third operation, neo-aortoiliac system if enough residual peri-renal aorta is present to construct an anastomosis or TFB). However, similar to other centers^{16, 35}, we have increasingly used this procedure as a primary operation in patients with significant visceral aortic occlusive disease in whom an infrarenal anastomosis is not possible, for subjects with mycotic complications involving the paravisceral aorta, as well as in cases of multiple failed axillo-femoral bypass grafts performed after aortic ligation for graft infection[while mycotic indication cases were not included in this analysis, this further highlights specific scenarios where TFB may be useful].

The ‘classic’ description of the operation using an anterolateral thoracotomy has several technical demands, and our enthusiasm for this form of the procedure is tempered by three factors: (1) mid-thoracic aortic cross-clamp potentially places the lung and spinal cord at risk; (2) remedial options for an infected graft are poor; and (3) prevalence of smoking or chronic obstructive pulmonary disease history and need for thoracotomy leading to postoperative pulmonary morbidity. We readily concede that these complications are rare; however, we submit that the “true” rate of complications from any relatively rare and/or redo aortic operation for AIOD is poorly understood because of the limited published experience.

Although there is a paucity of contemporary information about TFB outcomes, there are some notable historical series(Table IV). Branchereau and colleagues reviewed publications before 1993 and concluded that results from series including 10 or more patients suggested a 10% mortality rate³⁶. However, the series that were reviewed included patients with mycotic and non-mycotic indications which undoubtedly biased the outcomes. The postoperative mortality 15 16 37 39 outcomes for elective, non-mycotic indications is 0-6%^{15, 16, 37–39}, which is similar to our experience. The most frequent complications cited are those pertaining to cardiac, pulmonary and/or bleeding related events with an overall rate of 8-41%. The rates of pulmonary complications are notable since several authors have documented pulmonary complication(e.g. re-intubation, pneumonia, mechanical ventilation>3-days) rates of 15-30%^{14–17, 36, 39}, which compares favorably to our experience. With respect to longer term outcomes, the largest published series to date by Passman et al¹⁶ reported a 5-year primary patency of 79% in 50 patients¹⁶. Koksal et al¹² reported a series of 20 patients undergoing TFB also via a thoracotomy with 5-year patency rate of 94% similar to our study.

Table IV.

Outcomes of Thoracofemoral Bypass^a

Series	Number	Aortic Access	30-day mortality	Complication rate	Primary patency
bRosenfeld-1985	10	Thoracotomy	0	Not reported	Not reported
bBowes-1985	12	Thoracotomy	0	8%	Not reported
bFeldhaus-1985	18	Thoracotomy	6%	17%	85±8%@5yrs
De Laurentis-1986	10	Thoracotomy	0	Not available	100%@2yrs
Schultz-1986	15	Retroperitoneal	7%	18%	80±10%@5yrs
bBowes-1990	26	Thoracotomy	4%	18%	86±9%@3.5yrs
bBranchereau-1991	30	Thoracotomy	10%	22%	56±9%@4yrs
bMcCarthy-1993	21	Thoracotomy	0	19%	86±7%@5yrs
bCriado-1994	32	Thoracotomy	6%	41%	73±8%@6yrs
bSapienza-1997	41	Thoracotomy	5%	15%	64±8%@10yrs
bPassman-1999	50	Thoracotomy	4%	16%	79±5%@5yrs
Reppert-2014	13	Mini-Thoracotomy	0	38%	86%@1yr
Current series	41	Retroperitoneal	5%	34%	95±3%@3yrs
Pooled estimate (95%CI)c			4% (3%-8%)	22% (17%-31%)	80% (76%-83%) @5yrs

^aincluded only series with at least 10 patients reported;

^bmycotic & non-mycotic indications;

^cpredicted exponential survival based on N-weighted averaging of the center specific log (hazard rates)when a constant hazard rate is assumed

In an attempt to reduce some of the risks of the traditional description of TFB, we have embraced the strategy of using a left thoraco-retroperitoneal aortic exposure⁴⁰. A key advantage of TFB over traditional TFB is that it avoids use of a formal anterolateral thoracotomy. Although speculative, we believe this reduces the risk of pulmonary related complications as evidenced by our outcomes compared to historical series. In addition, this approach employs an aortic partial-occluding, side-biting clamp, minimizing the renal/mesenteric ischemia time. This is distinctly different than what may be required if an AFB is attempted in the setting of significant juxtarenal/suprarenal atherosclerotic disease due to the need for a suprarenal(or mesenteric) aortic cross-clamp. Our study also highlights the versatility of a left thoraco-retroperitoneal TFB for synchronous treatment of visceral occlusive disease. The retroperitoneal exposure facilitates access to the celiac, superior mesenteric and left renal artery which can be revascularized if indicated.

A possible disadvantage to the thoraco-retroperitoneal TFB is that it mandates more proximal aortic dissection and may result in opening of the pleural space in the 8^th or 9^th interspace to facilitate exposure of the supraceliac aorta. When the pleural space is entered our practice is to attempt diaphragm closure with Valsalva or place a left tube thoracostomy if the defect is too large, which may impact post-operative pulmonary outcomes and pain. Although this approach provides flexibility to concomitantly address significant visceral occlusive disease, it is best suited to treat focal proximal/orificial left renal and superior mesenteric artery occlusive disease. The right renal artery can be difficult to revascularize and if the pattern of occlusive disease extends into the mesenteric artery, the peritoneum may need to be opened to gain access to the distal SMA which can also be very challenging.

Other important limitations include the difficulty in managing graft infection after TFB. Two patients in our series experienced Szilagyi Grade III infection localized to the femoral anastomosis and were managed with graft excision and extra-anatomic revascularization. Although they did not experience residual/persistent infectious complications after remediation, the potential for proximal seeding of the graft and involvement of the thoracic aorta is a concern so frequently the procedure is used as a tertiary option in selected patients with unique patterns of complex paravisceral aortic occlusive disease. In addition, we found a 5% accidental splenectomy rate during the conduct of the operation. As a result, during the study period, our practice evolved such that we now routinely open the peritoneal cavity prior to closing and inspect the spleen and colon. Another potential drawback of TFB is that it requires familiarity with the retroperitoneal dissection and open aortic surgery. This is notable since the outcomes observed in this series may not be broadly reproducible over time as vascular surgeons perform fewer open aortic operations. Nonetheless, given the anatomic complexity of these patients, we advocate that they be treated at facilities with experience in both advanced endovascular, hybrid and open aortic surgery procedures.

There are several important limitations to this study including its retrospective design, small sample size and inherent selection bias. Therefore, it is difficult to make definitive conclusions from a single-center experience. However, this is one of the largest series in the literature on TFB and it offers specific examples of where this operation may fit in contemporary practice. Importantly, our short clinical follow-up time limits interpretation about what the anticipated long-term patency and re-intervention rates should be after this operation. Follow-up in this study was significantly influenced by our regional practice referral patterns. A majority of patients were referred from distant geographic locations and many patients preferred follow-up with local vascular surgeons(who were frequently the referring providers) rather than at our institution despite all efforts to schedule them in follow-up.

Conclusion

This experience represents one of the largest and most current series of TFB. We demonstrate that a left thoraco-retroperitoneal TFB is a technically demanding operation; however, it can be performed safely with good outcomes for patients with severe AIOD. TFB may be particularly useful for AIOD patients requiring concomitant visceral/infrainguinal reconstruction. These results support a continued role for TFB as a primary and/or remedial operation in selected patients.

Type of Research:

Retrospective analysis of prospectively collected single center registry data

Key Findings:

Thoraco-femoral bypass in 41 patients with severe aorto-iliac occlusive disease (AIOD) resulted in 30 day mortality and morbidity of 5% and 34% respectively, with a 3 year graft patency of 80%, 3 year freedom from major adverse limb event of 70% and a 5 year survival of 93%.

Take Home Message:

This study suggests that thoracofemoral bypass is a safe and effective option for patients with severe AIOD

Appendix Table. Complication definitions

Complication	Definition
Acute kidney injury	50% change in pre- vs. postoperative eGFR and/or new hemodialysis requirement
Bleeding	≥3 unit transfusion postoperatively and/or return to the OR for bleeding
Cardiac	Postoperative MI (NSTEMI or ST elevation diagnosed by ECG, clinically and/or by imaging), any new arrhythmia, new CHF requirement treatment
Pulmonary	Re-intubation and/or ventilator dependence > 72 hours, pneumothorax, pneumonia, pleural effusion necessitating thoracentesis/thoracostomy
Gastrointestinal	Inadvertent visceral injury (e.g. iatrogenic splenectomy), protracted ileus, bowel obstruction
Wound	Deep (Szilagyi Grade III) surgical site infection
Neurologic	New stroke/TIA, protracted delirium, seizure
Ischemia	New postoperative limb ischemia necessitating amputation and/or reintervention