A propensity-matched comparison of fenestrated endovascular aneurysm repair and open surgical repair of pararenal and paravisceral aortic aneurysms

CLINICAL RESEARCH STUDIES A propensity-matched comparison of fenestrated endovascular aneurysm repair and open surgical repair of pararenal and paravisceral aortic aneurysms Giovanni Tinelli, MD, PhD,a Maria Antonietta Crea, MD,b Chiara de Waure, MD, MSc, PhD,c Gian Luca Di Tanna, PhD, MPhil, MSc,d Jean-Pierre Becquemin, MD,e Jonathan Sobocinski, MD, PhD,f Francesco Snider, MD,a and Stéphan Haulon, MD, PhD,g Rome, Italy; and Champigny sur Marne, Lille, and Paris, France

ABSTRACT Objective: This study investigated the outcomes of a current series of patients treated with fenestrated and branched endovascular aneurysm repair (F-BEVAR) or open surgical repair (OSR) for pararenal abdominal aortic aneurysms (pr-AAAs), including juxtarenal, suprarenal, and type IV thoracoabdominal aneurysms. This study compares the outcomes of these procedures from two high-volume centers without the bias induced by a learning curve. Methods: All patients with pr-AAAs undergoing repair at two centers between January 2010 and June 2016 were included in a prospective database. Patients undergoing F-BEVAR and OSR were propensity matched for age, sex, anatomic criteria (aortic clamp site), coronary artery disease, chronic obstructive pulmonary disease, diabetes, smoking, chronic kidney disease, aneurysm diameter, and previous aortic surgery. The primary end points were mortality and dialysis. Secondary end points included any myocardial ischemia, respiratory and early procedural complications, acute kidney injury (AKI) according to RIFLE criteria (Risk, Injury, Failure, Loss of kidney function, and End-stage renal failure), spinal cord ischemia, a composite of these complications, and postoperative intensive care unit length of stay. During follow-up, all-cause survival and freedom from reintervention were compared, as was the patency of stented vessels and renal and visceral bypasses. Late renal function deterioration was evaluated. Results: In this period, 157 F-BEVAR patients and 119 OSR patients were operated on. After 1:1 propensity matching, the study cohort consisted of 102 F-BEVARs and 102 OSRs. In the matched population, an average of 2.5 vessels were treated per patient. Univariate analysis demonstrated no significant difference in 30-day mortality (2.9% vs 2.0%; P ¼ .68), dialysis (4.9% vs 3.9%; P ¼ 1), cardiac ischemic complications (3.8% vs 5.9%; P ¼ .52), pulmonary complications (5.9% vs 5.9%; P ¼ 1), or any complications (28.4% vs 30.4%; P ¼ .63) in the F-BEVAR and OSR groups, respectively. AKI was significantly lower in the F-BEVAR group than in the OSR group (19.6% vs 52%; P < .001), as was severe AKI (>50% decrease in glomerular filtration rate, 6.9% vs 16.7%; P ¼ .03). There was no spinal cord ischemia. The median intensive care unit length of stay was 1 day in both groups (P ¼ .33). During follow-up, we found occlusions of five stented vessels and three surgical bypasses. Late renal function deterioration was comparable between the two groups. According to Kaplan-Meier estimates, allcause survival at 24, 48, and 72 months was 85.6%, 66.8%, and 55.8% after F-BEVAR and 90.5%, 82.9%, and 68.5% after OSR (P ¼ .04). Rates of freedom from reintervention were 97.6% vs 97.5% at 24 months, 90.1% vs 93.4% at 48 months, and 63.9% vs 93.4% at 72 months in the F-BEVAR and OSR groups (P ¼ .05), respectively. Thus, both all-cause survival and freedom from reintervention were lower in the F-BEVAR group. Conclusions: This propensity score analysis in patients with pr-AAA undergoing F-BEVAR or OSR suggests no difference in terms of 30-day mortality, dialysis, or organ-specific postoperative complications, with the exception of AKI. Postoperative AKI was significantly higher after OSR, although most patients had recovered before discharge. Our data suggest similar outcomes after F-BEVAR or OSR for pr-AAA. (J Vasc Surg 2018;68:659-68.)

Extensive experience in treating complex aortic aneurysms with fenestrated endografting has led to improvements in endovascular aneurysm repair regarding preoperative planning, selection of

patients, implantation techniques, and perioperative care.1,2 Pararenal abdominal aortic aneurysms (pr-AAAs) are included in the definition of complex aneurysms. The

From the Vascular Unit, Cardiovascular Area,a and Department of Anesthesi-

Correspondence: Giovanni Tinelli, MD, PhD, UOC di Chirurgia Vascolare Polo

ology and Intensive Care,b Fondazione Policlinico Universitario A. Gemelli,

CardioVascolare e Toracico Fondazione Policlinico Universitario A. Gemelli,

Rome; the Department of Experimental Medicine, University of Perugia, Peru-

Largo Agostino Gemelli 8, 00168 Rome, Italy (e-mail: giovanni.tinelli@

giac; the Centre for Primary Care and Public Health, Queen Mary University of

policlinicogemelli.it).

London, Londond; the Institut Vasculaire Paris Est (IVPE), Hôpital Paul D’Egine,

The editors and reviewers of this article have no relevant financial relationships to

Champigny sur Marnee; the Aortic Center, CHU Lille, Lillef; and the Aortic Cen-

disclose per the JVS policy that requires reviewers to decline review of any

ter, Hôpital Marie Lannelongue, Le Plessis Robinson, INSERM UMR_S 999, Université Paris Sud, Paris.g Author conflict of interest: S.H. is a consultant for Cook Medical, Bentley, and GE Healthcare.

manuscript for which they may have a conflict of interest. 0741-5214 Copyright Ó 2018 by the Society for Vascular Surgery. Published by Elsevier Inc. https://doi.org/10.1016/j.jvs.2017.12.060

659

660

Journal of Vascular Surgery

Tinelli et al

September 2018

pr-AAAs include juxtarenal AAAs, suprarenal AAAs, and type IV thoracoabdominal aneurysms.3-5 Open surgical repair (OSR) for pr-AAA is the standard of care for surgically fit patients.5,6 Endovascular repair with custom-made fenestrated and branched endografts for pr-AAAs is associated with encouraging early outcomes, including low mortality and excellent technical success rates. Longterm follow-up now also shows favorable outcomes.1,7,8 Comparisons between fenestrated and branched endovascular aneurysm repair (F-BEVAR) and OSR were previously based on historical nonrandomized studies.9 Only one propensity score-matched study was published, showing significantly worse outcomes for F-BEVAR compared with OSR.10 This study reflects the current experience after overcoming the learning curve of F-BEVAR and OSR in two high-volume centers. This study analyzed the outcomes of these two procedures with propensity score matching.

METHODS Study population This retrospective cohort study compared the outcomes of F-BEVAR and OSR for pr-AAA by analyzing prospectively collected data from two centers: the Aortic Center (ACL; Lille, France) and the Vascular Unit of Fondazione Policlinico Universitario Gemelli (FPUG; Rome, Italy) between January 2010 and June 2016. The study was performed in accordance with the Institutional Ethics Committee rules, and individual consent for this retrospective analysis was waived. All patients provided consent for intervention. A high volume of F-BEVAR procedures has been performed in the ACL and of OSR procedures in the FPUG.6,11 During the study period, no F-BEVARs were performed at the FPUG. In this study, all F-BEVARs were performed at the ACL and all OSRs at the FPUG by a single experienced operator (S.H. and F.S., respectively). Patients were observed with regular postoperative appointments. Computed tomography (CT) angiography was performed at 1 month and 6 months postoperatively and yearly thereafter following F-BEVAR. The OSR group was followed up with abdominal ultrasound examination at 3 months and yearly thereafter. CT was performed, when possible, if abnormalities were found on ultrasound examination. Survival assessment was completed by phone interview. All patients with a pr-AAA requiring suprarenal or supravisceral proximal clamping were included in the study. For OSR, the clamp site was determined during surgery. For F-BEVAR, the operating surgeon defined the anticipated clamp site after reviewing the preoperative CT image.10 All F-BEVAR patients were deemed unsuitable for OSR after multidisciplinary evaluation because of high-risk comorbidities. The study excluded patients treated for

ARTICLE HIGHLIGHTS d

d

d

Type of Research: Retrospective analysis of data of a multicenter prospective registry Take Home Message: Propensity score matching in 204 patients who underwent either fenestrated and branched endovascular aneurysm repair or open surgical repair (OSR) of pararenal or paravisceral aortic aneurysms did not reveal any difference in 30-day mortality or postoperative complications between groups with the exception of acute kidney injury, which was more frequent after OSR, although most patients recovered before discharge. Recommendation: The data suggest similar outcomes after fenestrated and branched endovascular aneurysm repair or OSR for paravisceral aortic aneurysms.

extent I to III thoracoabdominal aneurysms, ruptured or symptomatic aneurysms, and dissections or connective tissue disorder aneurysms. End points The primary end points of this study were mortality and dialysis. Secondary end points included any respiratory or cardiac complications, acute kidney injury (AKI), spinal cord ischemia (SCI), any early secondary procedures (and the composite of these complications or death recorded at 30 days or in the hospital when they occurred during hospitalization beyond 30 days), and intensive care unit length of stay. Respiratory insufficiency included any prolonged intubation (>72 hours), the need for reintubation, and pneumonia. Cardiac complications were defined as myocardial ischemia diagnosed by electrocardiography or troponin change. A postoperative deterioration of the estimated glomerular filtration rate (eGFR) by 25% within 1 week according to the RIFLE classification (Risk, Injury, Failure, Loss of kidney function, and Endstage renal failure) was defined as AKI.12 The eGFR was calculated preoperatively and at the serum creatinine concentration peak within 1 week after the procedure. The eGFR was determined by the abbreviated Modification of Diet in Renal Disease study equation (eGFR [mL x min-1 x 1.73 m-2 ] ¼ 186 x [serum creatinine]-1.154 x [age]-0.203 x [0.704 if female] x [1.210 if African American]). Only severe AKI (>50% decrease in eGFR) was considered in the composite end point. SCI was defined as any neurologic deficit related to SCI regardless of severity and duration. Early secondary procedures included procedural and graft complications #30 days postoperatively. Late deterioration of renal function (in terms of a >25% and >50% decrease in eGFR) was evaluated using the last creatinine level during follow-up. All-cause survival, freedom from reintervention, and patency of target vessels were compared in the two groups.

Journal of Vascular Surgery

Tinelli et al

661

Volume 68, Number 3

Statistical analysis Patients in the OSR group were propensity score matched to patients in the endovascular group. In calculating the propensity score, prespecified sets of covariates were included as confounders in a logistic regression model to predict the treatment of interest without including the outcome. The covariates included were age, sex, aneurysm diameter, clamp level, previous aortic surgery, coronary artery disease, chronic obstructive pulmonary disease, chronic kidney disease (determined by serum creatinine level), diabetes mellitus, and smoking. The logistic regression model was used to generate the score and a score-based matched control group using a caliper method with a threshold of 2 standard deviations of the difference in propensity score; the balance assessment was made using various tests and checking quantile-quantile plots. Unmatched and matched cohorts were compared with respect to the set of covariates used to generate the propensity score through a t-test and c2 test for quantitative and qualitative variables, respectively. The mean and standard deviation or the median with interquartile range were used to describe quantitative variables, and absolute and relative frequencies were used to report qualitative variables.13,14 Propensity score-matched cohorts were compared with respect to end points. The c2 test, Fisher exact test, and t-test were used to perform the comparison. Overall survival and time to reintervention were analyzed with Kaplan-Meier curves and the log-rank test. To adjust for the development of any complications after the procedure in calculating midterm mortality results, a multivariable Cox regression model was run after checking the proportional hazard assumptions with the GrambschTherneau test. Results were reported as the hazard ratio with a 95% confidence interval (CI). The P value was set at .05. Stata 14 (StataCorp LP, College Station, Tex) and SPSS 22 (IBM Corp, Armonk, NY) were used for the analysis. Operative techniques OSR. A retroperitoneal approach was usually performed with a left flank incision according to Williams et al15 and Sicard et al.16 A transperitoneal approach was used less frequently. Briefly, the retroperitoneal exposure of the aorta was achieved through an oblique incision from the tip of the eleventh rib to the lateral rectus border at the paraumbilical level. The anatomic access to the aorta was completed with the position of the left kidney (anterior or posterior) during the procedure, depending on the level of aortic clamping and the left renal vein anatomy. Exposure of the proximal abdominal aorta was obtained by division of the left diaphragmatic crus. When the procedure also included renal artery surgery, we performed a direct reimplantation or a bypass with an 8-mm Dacron graft interposition. In this situation, a

selective perfusion of the renal arteries was achieved through the infusion of cold renal perfusion (lactated Ringer solution and mannitol 10%). F-BEVAR. All patients in the F-BEVAR group were treated with custom-made stent grafts within the instructions for use from the manufacturer (Cook Medical, Bloomington, Ind). The procedures required bilateral femoral access. General anesthesia was used in all patients with an open arterial exposure or a percutaneous approach when feasible. No preoperative cerebrospinal fluid drainage was performed. The bridge between the fenestrations and their respective target vessels was performed with a covered balloon-expandable stent (Advanta V12 [Atrium Medical Corporation, Hudson, NH] or BeGrafts [Bentley InnoMed, Hechingen, Germany]); relining with bare-metal stents (Luminex; Bard, Murray Hill, NJ) was selectively used in scenarios of kinking or to secure the bridging stent in the target vessels. Directional branches were used if the aortic lumen was >45 mm at the level of the origin of the target vessels. Technical success was defined by placement of the aortic stent and all intended side branches, the absence of type I and type III endoleaks, and patent target vessels.

RESULTS Demographics. During the study period, 281 procedures for pr-AAA were identified. Five patients from the OSR group were excluded because of an emergency setting. A total of 157 patients, referred to the ACL, underwent F-BEVAR treatment; 119 patients, referred to the FPUG, concurrently underwent OSR. Demographics and baseline characteristics for both groups are shown in Table I. Stent graft configurations in the F-BEVAR group and intraoperative aortic clamping level in the OSR group are shown in Table II. Before the propensity-matching process, the patients in the F-BEVAR group were more affected by coronary artery disease, diabetes mellitus, anticipated supravisceral clamp level, and previous aortic surgery; the patients in the OSR group were more frequently affected by active smoking and chronic kidney disease. After propensity matching, there were 102 patients in the F-BEVAR group and 102 patients in the OSR group with all differences in baseline characteristics corrected (Table I). Procedure data. In the matched endovascular group, the mean operative time was 160.9 6 65.3 minutes. The total volume of contrast agent averaged 108.8 6 42.6 mL, fluoroscopy time averaged 31.8 6 22.3 minutes, and indirect dose-area product averaged 57.84 6 39.4$Gy$cm2. Open arterial exposure was performed in 95 (93.1%) patients. Technical success was achieved in 102 patients (100%). There were 255 renal and visceral arteries incorporated by 245 fenestrations as well as 10 branches, with a mean of 2.5 stented vessels per patient.

662

Journal of Vascular Surgery

Tinelli et al

September 2018

Table I. Clinical and demographic features of unmatched and matched cohorts Unmatched cohorts F-BEVAR (n ¼ 157) Age, years, mean (SD)

Propensity score-matched cohorts

OSR (n ¼ 119)

P

F-BEVAR (n ¼ 102)

OSR (n ¼ 102)

P

71.8 (8.0)

71.7 (7.0)

.96

71.7 (6.8)

.55

Male

151 (96.2)

109 (91.6)

.11

97 (95.1)

94 (92.2)

.39

CAD

76 (48.4)

46 (38.7)

.11

43 (42.2)

39 (38.2)

.57

COPD

62 (39.5)

46 (38.7)

.89

41 (40.2)

39 (38.2)

.77

CKD

32 (20.4)

34 (28.6)

.11

25 (24.5)

28 (27.5)

.63

DM

41 (26.1)

12 (10.1)

<.01

13 (12.7)

12 (11.8)

.83

Smoking

37 (23.6)

45 (37.8)

.01

30 (29.4)

35 (34.3)

.45

Aneurysm diameter, mm, mean (SD)

72.3 (7.7)

58.3 (8.6)

64.2 (13.5)

<.001

59.8 (8.8)

60.6 (9.3)

.52

Clamp level Suprarenal Supravisceral Previous aortic surgery

51 (32.5)

70 (58.8)

106 (67.5)

49 (41.2)

19 (12.1)

4 (3.4)

<.001 <.01

48 (47.1)

57 (55.9)

54 (52.9)

45 (44.1)

.21

7 (6.9)

4 (3.9)

.35

CAD, Coronary artery disease; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; DM, diabetes mellitus; F-BEVAR, fenestrated and branched endovascular aneurysm repair; OSR, open surgical repair; SD, standard deviation. Values are reported as number (%) unless otherwise indicated. Significant results are listed in bold.

Table II. Stent graft configurations in the fenestrated and branched endovascular aneurysm repair (F-BEVAR) group and intraoperative aortic level clamping in the open surgical repair (OSR) group (unmatched population) Visceral artery involvement

F-BEVAR, graft configuration (n ¼ 157), No. (%)

One fenestration, above one renal artery

OSR, level of clamping (n ¼ 119), No. (%)

4 (2.5)

12 (10.0)

53 (33.7)

58 (48.7)

Three fenestrations, above SMA

88 (56.0)

24 (20.1)

Four fenestrations, above CT

12a (7.6)

25 (21)

Two fenestrations, above renal arteries

CT, Celiac trunk; SMA, superior mesenteric artery. a Five devices with two branches and two fenestrations.

In the OSR matched group, the mean procedure and proximal aortic clamping times were 311.4 6 81.4 minutes and 27.5 6 8.3 minutes, respectively. The surgical approach was retroperitoneal in 89 (87.3%) patients and transperitoneal in 13 (12.7%) patients. We performed 27 associated renal and visceral procedures in 21 patients (20.5%): 15 renal artery reimplantations (including polar renal arteries), 6 aortorenal bypasses, 2 aortovisceral bypasses (superior mesenteric artery [SMA] and common hepatic artery), 2 renal angioplasties (inclusive proximal anastomosis with a beveled graft), and 2 renal transaortic endarterectomies. Perioperative results. The characteristics of patients excluded by the matching process are reported in Table III. Table IV shows perioperative results. In the propensity score-matched cohorts, 30-day mortality was 2.9% for F-BEVAR vs 2.0% for OSR (P ¼ .68). Two patients in the F-BEVAR group died of mesenteric infarction, and one died of multiple organ failure, all caused by cholesterol emboli. Two patients died after OSR of myocardial infarction and multiorgan failure, respectively. In-hospital mortality was 3.9% after F-BEVAR and 2.9% after OSR

(P ¼ 1); one additional patient in each group died of mesenteric infarction (cholesterol embolism) and myocardial infarction, respectively. Nine patients (five in the F-BEVAR group and four in the OSR group) required dialysis immediately after the procedure (4.9% vs 3.9%; P ¼ 1), of whom three (F-BEVAR) vs two (OSR) patients (2.9% vs 2.0%; P ¼ .68) required permanent dialysis. The proportion of pulmonary, cardiac, and any complications was not statistically significant in the matched groups (Table IV). AKI according to the RIFLE classification was significantly lower in the F-BEVAR group (18.6% vs 52%; P < .001); the same was true for severe AKI (eGFR reduction >50%, 6.9% vs 16.7%; P ¼ .03; Table V). In the 53 patients with AKI after OSR, 33 (62%) returned to preoperative eGFR levels before discharge. In the F-BEVAR group, no renal function data were available at discharge. No SCI occurred in the matched or unmatched groups. Early secondary procedures were more frequent in the F-BEVAR group (11.8%) than in the OSR group (3.9%; P ¼ .04). There were 12 reinterventions in the F-BEVAR group: 3 femorofemoral crossover bypasses for iliac

Journal of Vascular Surgery

Tinelli et al

663

Volume 68, Number 3

Table III. Characteristics and outcomes of patients excluded from propensity score matching compared with the matched groups F-BEVAR Unmatched (n ¼ 55) Age, years, mean (SD)

73.1 (7.1)

OSR

Matched (n ¼ 102)

P

Unmatched (n ¼ 17)

Matched (n ¼ 102)

P

71.8 (8.0)

.35

71.9 (5.7)

71.7 (7.0)

.90

Sex, male

54 (98.2)

97 (95.1)

.67a

15 (88.2)

94 (92.2)

.63a

CAD

33 (60.0)

43 (42.2)

.03

7 (41.2)

39 (38.2)

.82

COPD

21 (38.2)

41 (40.2)

.81

7 (41.2)

39 (38.2)

.82

CKD

7 (12.7)

25 (24.5)

.08

6 (35.3)

28 (27.5)

.57

DM

28 (50.9)

13 (12.7)

<.001

0 (0)

12 (11.8)

.21

7 (12.7)

30 (29.4)

.02

10 (58.8)

35 (34.3)

.05

<.01

85.4 (15.8)

Current smokers Aneurysm diameter, mm, mean (SD)

55.4 (7.4)

59.8 (8.8)

3 (5.5)

48 (47.1)

52 (94.5)

54 (52.9)

60.6 (9.3)

<.001

Clamp level Suprarenal Supravisceral Previous aortic surgery

<.001 <.01

13 (76.5)

57 (55.9)

4 (23.5)

45 (44.1)

.10 a

12 (21.8)

7 (6.9)

0 (0)

4 (3.9)

1

30-day mortality

3 (5.5)

3 (2.9)

.42a

0 (0)

2 (2.0)

1a

In-hospital mortality

3 (5.5)

4 (3.9)

.70a

0 (0)

3 (2.9)

1a

Any complication

4 (25.5)

29 (28.4)

Dialysis

2 (3.6)

5 (4.9)

1a

0 (0)

4 (3.9)

1a

Definite dialysis

1 (1.8)

3 (2.9)

1a

0 (0)

2 (2.0)

1a

AKI

14 (25.5)

20 (19.6)

Cardiac complications

2 (3.6)

4 (3.9)

Pulmonary complications

2 (3.6)

6 (5.9)

Early reinterventions

8 (14.5)

12 (11.8)

.69

.40

10 (58.8)

31 (30.4)

.02

11 (64.7)

53 (52.0)

.33

3 (17.6)

6 (5.9)

.12a

.72a

8 (47.1)

6 (5.9)

<.001a

.62

3 (17.6)

4 (3.9)

.06a

1a

AKI, Acute kidney injury; CAD, Coronary artery disease; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; DM, diabetes mellitus; F-BEVAR, fenestrated and branched endovascular aneurysm repair; OSR, open surgical repair; SD, standard deviation. Values are reported as number (%) unless otherwise indicated. Significant results are listed in bold. a Fisher exact test.

Table IV. Outcomes comparison of fenestrated and branched endovascular aneurysm repair (F-BEVAR) and open surgical repair (OSR) in the matched groups F-BEVAR (n ¼ 102)

OSR (n ¼ 102)

P

30-Day mortality

3 (2.9)

2 (2.0)

.68a

In-hospital mortality

4 (3.9)

3 (2.9)

Any complications

29 (28.4)

31 (30.4)

Definite and transient dialysis

5 (4.9)

4 (3.9)

Definite dialysis

3 (2.9)

2 (2.0)

AKI

20 (19.6)

53 (52)

1a .63 1a .68a <.001

Severe AKI (>50% decrease in GFR)

7 (6.9)

17 (16.7)

.03

Cardiac complications

4 (3.9)

6 (5.9)

.52

6 (5.9)

6 (5.9)

12 (11.8)

4 (3.9)

.04

1 (1)

.33

Pulmonary complications Early reinterventions ICU, days, median (IQR)

1 (1)

1

AKI, Acute kidney injury; GFR, glomerular filtration rate; ICU, intensive care unit; IQR, interquartile range. Values are reported as number (%) unless otherwise indicated. Significant results are listed in bold. a Fisher exact test.

limb thrombosis, 3 artery restentings (SMA and two renal arteries), 4 surgical revisions for retroperitoneal hematoma bleeding (2) and wound dehiscence (2), and 2 bowel

resections for intestinal ischemia. There were four reinterventions in the OSR group: three wound dehiscences (surgical revision) and one retroperitoneal hematoma

664

Journal of Vascular Surgery

Tinelli et al

September 2018

Table V. Acute kidney injury (AKI) by RIFLE classification in matched group F-BEVAR (102)

OSR (102)

Preoperative eGFR, mL/min/1.73 m2

78.8 6 25

68.3 6 20.7

Postoperative eGFR, mL/min/1.73 m2

73.0 6 28.9

48.8 6 20.9

RIFLE

19 (18.6)

53 (52)

Risk

12 (11.8)

36 (35.3)

Injury

1 (1)

12 (11.8)

Failure

1 (1)

1 (1)

Loss

2 (2)

2 (2)

ESKD

3 (2.9)

2 (2)

Severe AKI (>50% decrease in eGFR)

7 (6.9)

17 (16.7)

P

<.001

.03

eGFR, Estimated glomerular filtration rate; ESKD, end-stage kidney disease; F-BEVAR, fenestrated and branched endovascular aneurysm repair; OSR, open surgical repair; RIFLE, Risk, Injury, Failure, Loss of kidney function, and End-stage kidney disease. Categorical variables are presented as number (%). Continuous variables are presented as mean 6 standard deviation. Significant results are listed in bold.

(surgical revision). No renal and visceral stent or bypass occlusions occurred during hospitalization. In addition, intensive care unit stay was similar in both groups with a median of 1 day (range, 0-286 days and 1-11 days for the F-BEVAR and OSR groups, respectively; P ¼ .33). Midterm results. The matched population in this study had a median follow-up of 38.9 months (interquartile range, 41.3 months) and a median of 38.48 months (46.59 months) and 39.02 (37.60 months) after F-BEVAR and OSR, respectively (P ¼ NS). During follow-up, five target vessels were occluded in the F-BEVAR group (four renal arteries, one SMA) and three in the OSR group (1 renal bypass, 1 renal, and 1 polar renal reimplantation into the aorta) in three patients. No reinterventions were performed for the SMA stent occlusion in the absence of intestinal symptoms because of an important collateral circulation from the celiac trunk. Late renal function deterioration (>25% and >50% decrease in eGFR) occurred in 17.6% vs 11.8% (P ¼ .85) and 4.9% vs 3.9% (P ¼ 1) of patients in the F-BEVAR and OSR groups, respectively. During follow-up, new permanent dialysis was required in four patients: three (2.9%) vs one (1%; P ¼ .62) in the F-BEVAR and OSR groups, respectively. Forty-nine deaths (24%) were observed during followup, none of which were aorta related. The Kaplan-Meier analysis estimated that at 24 months, overall survival was 85.6% (95% CI, 78.5%-92.7%) and 90.5% (95% CI, 84.6%-96.4%) after F-BEVAR and OSR, respectively. Overall survival was 66.8% (95% CI, 55.8%-77.8%) and 82.9% (95% CI, 74.5%-91.3%) at 48 months and 55.8% (95% CI, 40.9%-70.7%) and 68.5% (95% CI, 53.8%-83.2%) at 72 months in the F-BEVAR and OSR groups, respectively. Survival during follow-up was higher after OSR (Breslow test, P ¼ .08; log-rank test, P ¼ .04; Fig 1). The multivariable Cox regression analysis showed that F-BEVAR was associated with a higher risk of death than OSR (hazard

ratio, 1.86; 95% CI, 1.04-3.33) also in adjusting for any complications. The Grambsch-Therneau test did not show violation of the proportional hazards assumption. The causes of late death included cardiac disease (n ¼ 9), stroke (n ¼ 8), cancer (n ¼ 7), respiratory failure (n ¼ 3), and other (n ¼ 4) in the F-BEVAR group and cardiac disease (n ¼ 7), cancer (n ¼ 6), stroke (n ¼ 3), and other (n ¼ 2) in the OSR group. Rates of freedom from reintervention were 97.6% (94.3%-100%) vs 97.5% (94.0%-100%) at 24 months, 90.1% (82.3%-97.9%) vs 93.4% (87.7%-100%) at 48 months, and 63.9% (41.6%-86.2%) vs 93.4% (87.7%100%) at 72 months in the F-BEVAR group vs the OSR group, respectively. Freedom from reintervention was better after OSR (Breslow test, P ¼. 22; log-rank test, P ¼ .05; Fig 2). During follow-up, 11 reinterventions were performed in the F-BEVAR group, and four were performed in the OSR group. In the endovascular group, five patients underwent embolization for type II endoleak with concomitant iliac extension cuff for type IB endoleak (two cases); two patients underwent three bridging stent realignments for type III endoleak; two patients required an iliac extension cuff and an iliac relining for type IB and type III endoleak, respectively; and two patients underwent crossover femorofemoral bypass for iliac leg occlusion. In the OSR group, two patients required surgical revision for distal anastomotic pseudoaneurysm. One patient required iliac stenting for anastomosis hyperplasia. One patient has been operated on for XI left intercostal neuroma at the surgical access site.

DISCUSSION We report a comparison of early and midterm outcomes after F-BEVAR and OSR for pr-AAAs performed in matched comparable patients. This study considers the results of two high-volume centers without a learning curve bias. Most if not all previously published series included their initial experience with these

Journal of Vascular Surgery

Tinelli et al

665

Volume 68, Number 3

Fig 1. Survival after fenestrated and branched endovascular aneurysm repair (F-BEVAR; black line) and open surgical repair (OSR; gray line).

Fig 2. Freedom from reintervention after fenestrated and branched endovascular aneurysm repair (F-BEVAR; black line) and open surgical repair (OSR; gray line).

techniques, that is, learning curves affected selection of patients, procedure performance, and periprocedural management, especially in the endovascular group.10

In our study, the patients in both the F-BEVAR and OSR groups had similar rates of perioperative mortality (2.9% vs 2.0%), dialysis (4.9% vs 3.9%), and definitive dialysis

666

Tinelli et al

Journal of Vascular Surgery September 2018

(2.9% vs 2.0%). The principal cause of mortality was intestinal ischemia in the F-BEVAR group and cardiac complications in the OSR group. These results confirm the risk of cholesterol embolism and SMA dissection after F-BEVAR for pr-AAA; the first issue can be addressed by excluding all patients with a shaggy aortic wall, the second by careful wire manipulation in the SMA, especially of stiff wires (we have stopped using the Amplatz wire in the SMA). These data indicate that the endovascular group, at “high risk” for open repair, includes a high-risk subgroup for endovascular procedures because of atherothrombotic parietal disease in the descending and visceral aorta.17 The treatment of “good surgical risk” patients and the exclusion of these “high-risk aortas” are associated with outcomes similar to or better than those in the most favorable open series.1,18 After open surgery, cardiac stress from proximal aortic clamping is associated with a myocardial infarction risk.19,20 Preoperative risk assessment with cardiovascular stress testing reliably predicts perioperative and longterm cardiac events.21 In the study of Raux et al,10 a similar comparison with propensity score matching showed a higher mortality rate in the F-BEVAR group (9.2% vs 2%; P ¼ .05) with a fivefold increased 30-day mortality risk (odds ratio, 5.1; 95% CI, 1.1-24; P ¼ .04). Raux et al compared the learning curve with F-BEVAR performed in high-risk patients with open repair performed by highly experienced operators. Similar to our study, the endovascular group was from a center that also performed open repair of complex aneurysms. There is thus a significant trend to propose F-BEVAR to the most fragile patients and open surgery to the fitter patients. Propensity scores are performed to limit that bias in comparing the techniques; however, the general assessment of the patient’s physiologic status by an experienced operator cannot be accurately quantified. In addition, in a center that is performing both techniques, high-risk patients can be offered a complex endovascular repair with clear information about the death and dialysis risks. In a center performing only open repair, these high-risk patients are often turned down for surgery. We believe that high-volume aortic centers that perform both open and endovascular repairs and that have decades of experience, as well as a team approach that uses dedicated protocols, are key in achieving excellent early and midterm outcomes.1,22 It is difficult to evaluate the number of cases required to progress beyond the learning curve. We have observed a significant change in our outcomes when our annual volume increased to >30 cases/year.23 The renal morbidity rate is another critical issue after treatment for pr-AAA with both techniques. In our series, the incidence of postoperative renal failure was significantly higher after OSR (52%) compared with F-BEVAR (19.6%; P < .001). Albeit numerically inferior, severe AKI

(>50% decrease in eGFR) was significantly higher in the surgical group (6.9% vs 16.7% for F-BEVAR and OSR, respectively; P ¼ .03). The 18.6% postoperative AKI observed for endovascular repair is comparable with previously published rates.11,17 During OSR, the incidence of postoperative renal failure reported in the literature ranges from 12% to 31%.24-27 Comparing published OSR-related results on this feature is difficult, given the heterogeneous criteria adopted to define renal failure. We have analyzed AKI using more sensible criteria, such as eGFR according to the RIFLE classification.12,28 The rate of AKI was 52% in OSR; however, 62% of these patients returned to normal preoperative renal function levels by discharge after medical therapy. This can be explained by the principal cause of AKI in OSR, that is, acute postischemic tubular necrosis according to proximal clamping time.29-31 According to the RIFLE classification, Kabbani et al5 described a comparable result for open repair in pr-AAA with a 60% rate of postoperative AKI, with most (72%) recovering before discharge. The rate of definitive dialysis (2.9% vs 2.0% in the F-BEVAR and OSR groups, respectively) remained low compared with the literature.26,32 Renal function deterioration during follow-up was comparable between the two groups. The complication rate after either endovascular or open repair in this study was not significantly different in terms of other organ-specific injury or the any-complication composite end point. Unexpectedly, the pulmonary complication rates for F-BEVAR and OSR were similar in the matched group (both 5.9% in the matched group; P ¼ 1). However, a significant difference was shown between F-BEVAR and OSR in the unmatched group (5.1% vs 11.8%, respectively; P ¼ .04). This specific point can be explained because we lost eight patients with pulmonary complications in the OSR group during matching. Notably, the prevalence of current smokers among the unmatched patients in the OSR group was higher than among the matched patients. This result was expected because smoking is associated with pulmonary complications (14.6% vs 5.2% pulmonary complications among current smokers vs ex-smokers and nonsmokers; P < .01) and was used to build the propensity score.6,31,33 SCI was not observed after either treatment in the matched and unmatched groups, confirming that this severe complication is nearly exclusively associated with more extensive aortic repairs for type I to type III thoracoabdominal aneurysms.2 The number of median intensive care unit days was similar in both groups (limited to 1 day), showing the importance of a postoperative dedicated intensive care team. In contrast to the experience of Raux et al,10 we observed a comparable rate of composite end points for any complication in both groups (28.4% vs 30.4% in F-BEVAR and OSR); this was significantly higher (nearly

Journal of Vascular Surgery

Tinelli et al

667

Volume 68, Number 3

double) after endovascular repair in their initial comparison. The early postoperative reintervention rate remains higher after endovascular repair (11.8% vs 3.9% in F-BEVAR and OSR); this is not found when we compare the endovascular group with the unmatched open group. Most of these reinterventions were performed by endovascular means. During follow-up, secondary reinterventions were more frequent in the endovascular group (11.8% vs 3.9%); most were also performed by endovascular means. The reintervention rate after F-BEVAR is decreasing; this will continue because of the evolution of endograft design and planning, the availability of purpose-built bridging stents, and the routine assessment of technical success with three-dimensional imaging (completion cone beam CT). Pararenal aneurysm repair was initially performed with two-fenestration endografts but is now routinely performed with three or four fenestrations because a posterior bulge at the level of the visceral aorta can dilate and compromise proximal fixation and seal during follow-up. The more complex three- and fourfenestration designs no longer represent a technical challenge but have the advantage of providing long and stable platforms for proximal fixation and seal. Furthermore, the long fixation zone inherent to these designs minimizes the risk of movement of the fenestrated main body and its potential consequence of bridging stent compromise and secondary procedures in the target vessel.34 Oderich et al1 described their outcomes after endovascular repair of complex aneurysms using custom-made fenestrated and branched endografts; 70% of patients enrolled had pararenal or Crawford classification type IV thoracoabdominal aortic aneurysms. This prospective, single-center cohort study is the first to report that F-BEVAR can be performed without early mortality, dialysis, ruptures, or conversions and with a very low SCI rate.18 These outstanding results demonstrate the mature state of complex endovascular treatment with F-BEVAR and emphasize that F-BEVAR is associated with excellent midterm outcomes in “low-risk” patients. Our experience with pr-AAA open treatment was also associated with favorable early and midterm results, confirming that open surgery remains an excellent approach for fit patients. Study limitations. As previously stated, this study should be interpreted with caution because the propensity analysis did not allow a completely balanced comparison and the retrospective analysis was only in part balanced by prospective data collection. The number of comparable patients decreased the power of detecting a difference in treatment efficacy. The unmatched patients had higher rates of any complications, cardiac complications, pulmonary complications, kidney complications, and early reinterventions than the matched

open group (and significantly higher rates than those in the F-BEVAR group). This is a limitation. However, assuming a noninferiority margin of 0.075 and a of .05 with a 30-day mortality of 3.5% and 2.5% in the F-BEVAR and OSR groups, respectively, our 102 included patients (with an allocation ratio of 1) gave us a power of 0.80. In addition, the overall hospital or center effect, as in any multicenter study, may not have been captured. Furthermore, this study showed the one experienced surgeon result in each group, and this may be a limitation for reproducibility in a larger group of operators. Notably, readmission for bowel obstruction, secondary procedures for hernia, time spent in rehabilitation, and quality of life were not analyzed in this study.

CONCLUSIONS This study suggests similar mortality and dialysis rates in matched populations of patients with pr-AAA after endovascular and surgical repair. Among the organspecific postoperative complications, AKI is significantly higher in OSR, and most of these patients returned to preoperative renal function levels before discharge. The early and midterm results are favorable in both groups. Both techniques should be available in high-volume centers to enable tailored treatments in line with the patient’s physiologic status and anatomy.

AUTHOR CONTRIBUTIONS Conception and design: GT, CDW, GDT, JB, JS, FS, SH Analysis and interpretation: GT, MC, CDW, GDT, JB, JS, FS, SH Data collection: GT, MC, JS, FS, SH Writing the article: GT, MC, CDW, GDT, JS, SH Critical revision of the article: GT, JB, JS, FS, SH Final approval of the article: GT, MC, CDW, GDT, JB, JS, FS, SH Statistical analysis: CDW, GDT Obtained funding: Not applicable Overall responsibility: GT

REFERENCES 1. Oderich GS, Ribeiro M, Hofer J, Wigham J, Cha S, Chini J, et al. Prospective, nonrandomized study to evaluate endovascular repair of pararenal and thoracoabdominal aortic aneurysms using fenestrated-branched endografts based on supraceliac sealing zones. J Vasc Surg 2017;65:1249-59. e10. 2. Ferrer C, Cao P, De Rango P, Tshomba Y, Verzini F, Melissano G, et al. A propensity-matched comparison for endovascular and open repair of thoracoabdominal aortic aneurysms. J Vasc Surg 2016;63:1201-7. 3. Nypaver TJ, Shepard AD, Reddy DJ, Elliott JP, Ernst CB. Supraceliac aortic cross-clamping: determinants of outcome in elective abdominal aortic reconstruction. J Vasc Surg 1993;17:868-75; discussion: 875. 4. Allen BT, Anderson CB, Rubin BG, Flye MW, Baumann DS, Sicard GA. Preservation of renal function in juxtarenal and suprarenal abdominal aortic aneurysm repair. J Vasc Surg 1993;17:948-58; discussion: 958. 5. Kabbani LS, West CA, Viau D, Nypaver TJ, Weaver MR, Barth C, et al. Survival after repair of pararenal and

668

Tinelli et al

Journal of Vascular Surgery September 2018

6.

7.

8.

9.

10.

11.

12.

13.

14. 15.

16.

17.

18.

19.

20.

paravisceral abdominal aortic aneurysms. J Vasc Surg 2014;59:1488-94. Ferrante AM, Moscato U, Colacchio EC, Snider F. Results after elective open repair of pararenal abdominal aortic aneurysms. J Vasc Surg 2016;63:1443-50. Cross J, Gurusamy K, Gadhvi V, Simring D, Harris P, Ivancev K, et al. Fenestrated endovascular aneurysm repair. Br J Surg 2012;99:152-9. Roy IN, Millen AM, Jones SM, Vallabhaneni SR, Scurr JR, McWilliams RG, et al. Long-term follow-up of fenestrated endovascular repair for juxtarenal aortic aneurysm. Br J Surg 2017;104:1020-7. Nordon IM, Hinchliffe RJ, Holt PJ, Loftus IM, Thompson MM. Modern treatment of juxtarenal abdominal aortic aneurysms with fenestrated endografting and open repairda systematic review. Eur J Vasc Endovasc Surg 2009;38:35-41. Raux M, Patel VI, Cochennec F, Mukhopadhyay S, Desgranges P, Cambria RP, et al. A propensity-matched comparison of outcomes for fenestrated endovascular aneurysm repair and open surgical repair of complex abdominal aortic aneurysms. J Vasc Surg 2014;60:858-63; discussion: 863. Martin-Gonzalez T, Pinçon C, Maurel B, Hertault A, Sobocinski J, Spear R, et al. Renal outcomes following fenestrated and branched endografting. Eur J Vasc Endovasc Surg 2015;50:420-30. Bellomo R, Ronco C, Kellum JA, Mehta RL, Palevsky P; Acute Dialysis Quality Initiative workgroup. Acute renal failured definition, outcome measures, animal models, fluid therapy and information technology needs: the Second International Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group. Crit Care 2004;8:R204-12. Austin PC. A critical appraisal of propensity-score matching in the medical literature between 1996 and 2003. Stat Med 2008;27:2037-49. Stuart EA. Matching methods for causal inference: a review and a look forward. Stat Sci 2010;25:1-21. Williams GM, Ricotta J, Zinner M, Burdick J. The extended retroperitoneal approach for treatment of extensive atherosclerosis of the aorta and renal vessels. Surgery 1980;88:846-55. Sicard GA, Freeman MB, VanderWoude JC, Anderson CB. Comparison between the transabdominal and retroperitoneal approach for reconstruction of the infrarenal abdominal aorta. J Vasc Surg 1987;5:19-27. Ribeiro M, Oderich GS, Macedo T, Vrtiska TJ, Hofer J, Chini J, et al. Assessment of aortic wall thrombus predicts outcomes of endovascular repair of complex aortic aneurysms using fenestrated and branched endografts. J Vasc Surg 2017;66: 1321-33. Haulon S. Fenestrated and branched endovascular aortic repair has reached a state of maturity [editorial]. J Vasc Surg 2017;65:1247. Lala S, Knowles M, Timaran D, Baig MS, Valentine J, Timaran C. Superior mesenteric artery outcomes after fenestrated endovascular aortic aneurysm repair. J Vasc Surg 2016;64:692-7. Lim S, Halandras PM, Saqib NU, Ching YA, Villella E, Park T, et al. Comparison of supramesenteric aortic cross-clamping with supraceliac aortic cross-clamping for aortic reconstruction. J Vasc Surg 2016;64:941-7.

21. Fleisher LA, Fleischmann KE, Auerbach AD, Barnason SA, Beckman JA, Bozkurt B, et al. 2014 ACC/AHA guideline on perioperative cardiovascular evaluation and management of patients undergoing noncardiac surgery: a report of the American College of Cardiology/American Heart Association Task Force on practice guidelines. J Am Coll Cardiol 2014;64: e77-137. 22. Haulon S, Barillà D, Tyrrell M, Tsilimparis N, Ricotta JJ. Debate: whether fenestrated endografts should be limited to a small number of specialized centers. J Vasc Surg 2013;57:875-82. 23. Maurel B, Delclaux N, Sobocinski J, Hertault A, MartinGonzalez T, Moussa M, et al. The impact of early pelvic and lower limb reperfusion and attentive peri-operative management on the incidence of spinal cord ischemia during thoracoabdominal aortic aneurysm endovascular repair. Eur J Vasc Endovasc Surg 2015;49:248-54. 24. Shortell CK, Johansson M, Green RM, Illig KA. Optimal operative strategies in repair of juxtarenal abdominal aortic aneurysms. Ann Vasc Surg 2003;17:60-5. 25. Tallarita T, Sobreira ML, Oderich GS. Results of open pararenal abdominal aortic aneurysm repair: tabular review of the literature. Ann Vasc Surg 2011;25:143-9. 26. Jongkind V, Yeung KK, Akkersdijk GJ, Heidsieck D, Reitsma JB, Tangelder GJ, et al. Juxtarenal aortic aneurysm repair. J Vasc Surg 2010;52:760-7. 27. West CA, Noel AA, Bower TC, Cherry KJ, Gloviczki P, Sullivan TM, et al. Factors affecting outcomes of open surgical repair of pararenal aortic aneurysms: a 10-year experience. J Vasc Surg 2006;43:921-7; discussion: 927. 28. Mehta RL, Kellum JA, Shah SV, Molitoris BA, Ronco C, Warnock DG, et al. Acute Kidney Injury Network: report of an initiative to improve outcomes in acute kidney injury. Crit Care 2007;11:R31. 29. Svensson LG, Coselli JS, Safi HJ, Hess KR, Crawford ES. Appraisal of adjuncts to prevent acute renal failure after surgery on the thoracic or thoracoabdominal aorta. J Vasc Surg 1989;10:230-9. 30. Miller DC, Myers BD. Pathophysiology and prevention of acute renal failure associated with thoracoabdominal or abdominal aortic surgery. J Vasc Surg 1987;5:518-23. 31. Chiesa R, Marone EM, Brioschi C, Frigerio S, Tshomba Y, Melissano G. Open repair of pararenal aortic aneurysms: operative management, early results, and risk factor analysis. Ann Vasc Surg 2006;20:739-46. 32. Shahverdyan R, Majd MP, Thul R, Braun N, Gawenda M, Brunkwall J. F-EVAR does not impair renal function more than open surgery for juxtarenal aortic aneurysms: single centre results. Eur J Vasc Endovasc Surg 2015;50:432-41. 33. Jeyabalan G, Park T, Rhee RY, Makaroun MS, Cho JS. Comparison of modern open infrarenal and pararenal abdominal aortic aneurysm repair on early outcomes and renal dysfunction at one year. J Vasc Surg 2011;54:654-9. 34. Martin-Gonzalez T, Mastracci T, Carrell T, Constantinou J, Dias N, Katsargyris A, et al. Mid-term outcomes of renal branches versus renal fenestrations for thoraco-abdominal aneurysm repair. Eur J Vasc Endovasc Surg 2016;52:141-8.

Submitted Jul 29, 2017; accepted Dec 17, 2017.