- Email: [email protected]
Adherence to endovascular aortic aneurysm repair device instructions for use guidelines has no impact on outcomes
From the Western Vascular Society
Adherence to endovascular aortic aneurysm repair device instructions for use guidelines has no impact on outcomes Joy Walker, MD,a Lue-Yen Tucker, BA,b Philip Goodney, MD,c Leah Candell, MD,d Hong Hua, MD,e Steven Okuhn, MD,e Bradley Hill, MD,f and Robert W. Chang, MD,g San Francisco, Oakland, and Santa Clara, Calif; and Lebanon, NH Objective: Prior reports have suggested unfavorable outcomes after endovascular aortic aneurysm repair (EVAR) performed outside of the recommended instructions for use (IFU) guidelines. We report our long-term EVAR experience in a large multicenter registry with regard to adherence to IFU guidelines. Methods: Between 2000 and 2010, 489 of 1736 patients who underwent EVAR had preoperative anatomic measurements obtained from the M2S, Inc, imaging database (West Lebanon, NH). We examined outcomes in these patients with regard to whether they had met the device-specific IFU criteria. Primary outcomes were all-cause mortality and aneurysmrelated mortality. Secondary outcomes were endoleak status, adverse events, reintervention, and aneurysm sac size change. Results: The median follow-up for the 489 patients was 3.1 years (interquartile range, 1.6-5.0 years); 58.1% (n [ 284) had EVAR performed within IFU guidelines (IFU-adherent group), and 41.9% (n [ 205) had EVAR performed outside of IFU guidelines (IFU-nonadherent group). Preoperative anatomic data showed that 62.4% of the IFU-nonadherent group had short neck length, 10.2% had greater angulation than recommended, 7.3% did not meet neck diameter criteria, and 20% had multiple anatomic issues. A small portion (n [ 49; 10%) of the 489 patients were lost to follow-up because of leaving membership enrollment (n [ 28), moving outside the region (n [ 10), or discontinuing image surveillance (n [ 11). There was no significant difference in any of the primary or secondary outcomes between the IFUadherent and IFU-nonadherent groups. Aneurysm sac size change at any time point during follow-up also did not differ significantly between the two groups. A Cox proportional hazard model showed that IFU nonadherence was not predictive of all-cause mortality (hazard ratio, 1.0; P [ .91). Similarly, IFU nonadherence was not identified as a risk factor for aneurysm-related mortality or adverse events in stepwise Cox proportional hazards models. Conclusions: In our cohort of EVAR patients with detailed preoperative anatomic information and long-term follow-up, overall mortality and aneurysm-related mortality were unaffected by IFU adherence. In addition, rates of endoleak and reintervention after initial EVAR were similar, suggesting that lack of IFU-based anatomic suitability was not a driver of outcomes. (J Vasc Surg 2015;61:1151-9.)
Since its first description in 1991, endovascular aortic aneurysm repair (EVAR) has become a commonplace and well-accepted alternative to open abdominal aortic aneurysm (AAA) repair.1,2 Proper patient selection based on anatomic From the Division of Surgery, University of California, San Francisco, San Franciscoa; the Division of Research, Kaiser Permanente, Oaklandb; the Division of Vascular Surgery, Dartmouth-Hitchcock Medical Center, Lebanonc; the Division of Surgery, University of California, San Francisco-East Bay Medical Center, Oaklandd; the Division of Vascular Surgery, The Permanente Medical Group, San Franciscoe; the Division of Vascular Surgery, The Permanente Medical Group, Santa Claraf; and the Division of Vascular Surgery, The Permanente Medical Group, South San Francisco.g Author conflict of interest: none. Presented at Scientific Session V of the Twenty-ninth Annual Meeting of the Western Vascular Society, Coronado, Calif, September 20-23, 2014. Reprint requests: Robert W. Chang, MD, Vascular and Endovascular Surgery, Kaiser Permanente, South San Francisco, 1200 El Camino Real, South San Francisco, CA 94080 (e-mail: [email protected]). The editors and reviewers of this article have no relevant financial relationships to disclose per the JVS policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. 0741-5214 Copyright Ó 2015 by the Society for Vascular Surgery. Published by Elsevier Inc. http://dx.doi.org/10.1016/j.jvs.2014.12.053
criteria has become critical to ensuring satisfactory long-term results. To that end, each endovascular device manufacturer publishes an instructions for use (IFU) guideline that specifies anatomic criteria for “correct” use of the EVAR device. These recommendations are made on the basis of preclinical engineering assessments and clinical study results. The guidelines specify appropriate aortic neck diameter, aortic neck length, aortic neck angle, and iliac artery morphology. Many clinicians believe that outcomes after EVAR largely depend on whether the devices are used in accordance with the IFU guidelines. A recent paper documented the incidence of IFU nonadherence in registry data sets but had notable gaps in availability of device-specific and patient outcome data.3 We analyzed outcomes of EVAR patients in a longitudinal registry for whom detailed preoperative anatomic data were available, with the objective of determining whether IFU adherence affects outcomes after EVAR. METHODS Kaiser Permanente Northern California (KPNC) is an integrated health care delivery system that offers multispecialty care for more than 3 million members. Implementation of 1151
1152 Walker et al
digital health records has allowed access to all arenas of clinical information. Data for 1736 patients who underwent EVAR procedures in 17 KPNC medical centers were collected in a clinical registry from 2000 to 2010. This study protocol was approved by the KPNC Institutional Review Board and was funded for one calendar year in 2011 by the KPNC Community Benefit Research Grant Program. This in-depth evaluation of IFU adherence was funded by the KPNC Residency Programs. Informed consent was waived by the local Institutional Review Board, given that the study was retrospective and the data de-identified for analysis. Data collection and follow-up. Beginning in 2000, relevant clinical information for patients undergoing EVAR was collected by trained research nurses, with December 31, 2010, as the last follow-up date. Data collected from electronic medical records included baseline preoperative demographic data and clinical characteristics (sex, age, aneurysm sac size, comorbidities [coronary artery disease, diabetes mellitus, hypertension, hyperlipidemia, and peripheral vascular disease], smoking status, and statin history). Device type and operative details were collected from the operative report and device entry forms. Information on adjunctive maneuvers (placement of additional aortic cuffs or stents, graft limb extensions, exploratory laparotomy/conversion, renal stenting/snorkel, and femoral-femoral bypass) was also collected. Decisions regarding indications for surgery, suitability for endovascular repair, device selection, and need for secondary intervention were made at the discretion of the operating surgeon. Data recorded in our registry during the follow-up period were also collected and included aneurysm rupture, major adverse event (ie, conversion to open repair, major embolic event, graft infection requiring explantation, device migration, loss of device patency requiring reintervention, and other miscellaneous complications that substantially affected clinical outcome), types of reintervention, AAA sac size, endoleak, leaving KPNC membership, moving outside the region, and mortality. Postoperative follow-up varied across medical centers but generally involved a 1-month postoperative computed tomography (CT) scan followed by serial CT imaging at regular intervals ranging from every 3 months to every 12 months as dictated by clinical circumstances. Imaging was accompanied by clinical follow-up. As follow-up progressed, there was a considerable amount of variability regarding the timing, modality, and use of contrast material in CT imaging. However, the preoperative and first postoperative imaging generally consisted of CT scans with and without contrast material and arterial and venous phases with 1.25-mm slices. In the absence of sac growth or endoleak, intravenous contrast material was sometimes withheld for subsequent examinations at the discretion of the treating physician. Device migration was reported if it required intervention or if adequate seal was lost, usually when reduced to <10 mm of the circumferential apposition length. Endoleak was classified according to established reporting standards4 and was typically detected by CT scan, confirmatory angiography, or, more rarely, ultrasound. Determination of adherence to IFU guidelines. Our clinical EVAR registry was cross-referenced with data from
JOURNAL OF VASCULAR SURGERY May 2015
the M2S, Inc, imaging repository (West Lebanon, NH). The M2S anatomic registry (hereafter termed M2S) comprehensively assesses detailed anatomic data from CT scans submitted to them from the clinician. Using standardized algorithms, M2S creates three-dimensional computer models from CT images of aortic aneurysms with semiautomated quantification of multiple measurements of interest. Measurements of interest corresponded with major determinants of IFU adherence and therefore primarily involved the proximal infrarenal aortic neck (eg, neck length, diameter, and angulation). This in-depth study was a subset analysis limited to patients whose initial EVAR procedures had relevant preoperative M2S analysis of CT imaging. These anatomic data and corresponding measurements were then used to determine adherence to IFU guidelines. Patients whose initial EVAR procedures were performed within the IFU guidelines were classified as the IFU-adherent group, and those outside the IFU guidelines as the IFU-nonadherent group. The evaluation of specific guidelines varied according to the specific device implanted. Outcome variables. The primary outcomes were allcause mortality and aneurysm-related mortality (ARM); the latter was defined as death within 30 days of the initial EVAR or of a secondary procedure related to aneurysm rupture or a major adverse event. Secondary outcome variables examined were type I or type III endoleak, major adverse events, need for reintervention, and change in aneurysm sac size over time. Sac size was assessed at several time points during follow-up (ie, 2 months, 6 months, 9 months, 12 months, 18 months, 2 years, 3 years, 4 years, and 5 years after the initial EVAR). Sac size measurements were accepted if performed within 1.5 months before or after each time point within the first 12 months or within 3 months before or after time points after 12 months. Statistical methods. Rates of categorical demographic (sex, age groups, racial/ethnic groups) and clinical characteristics (comorbidities, smoking status, statin history, preoperative embolization, adjunctive maneuver or bifurcated graft during the initial EVAR procedure, and aneurysm sac growth status at various time points during follow-up) were compared between the IFU-adherent and IFUnonadherent groups with c2 tests or Fisher exact tests as appropriate. Preoperative AAA sac size, age at initial EVAR, and M2S anatomic data (ie, neck length, neck diameter, neck angle, and femoral diameter) were not normally distributed; therefore, nonparametric WilcoxonMann-Whitney tests were used to compare medians. Survival analysis was performed with the Kaplan-Meier method to estimate the survival function in 489 patients with M2S stratified by IFU guideline status (adherence vs nonadherence), and survival functions were compared between these two groups by the log-rank test. Before fitting of the multivariable regression models, bivariate analyses were performed to evaluate the association of risk factors of interest (IFU adherence, sex, age, preoperative AAA sac size, history of statin, comorbidities, smoking status, preoperative embolization, operative
JOURNAL OF VASCULAR SURGERY Volume 61, Number 5
Walker et al 1153
Table I. Baseline preoperative demographic and clinical characteristics of 489 patients with preoperative anatomic data at the initial endovascular aortic aneurysm repair (EVAR) procedure, stratified by instructions for use (IFU) guideline status Characteristics Female Age, years Mean 6 SD Median [IQR] Age group, years #79 $80 Preoperative AAA size, cm Mean 6 SD Median [IQR] Preoperative AAA sac size $5.5 cm EVAR device Gore Excluder Medtronic AneuRx Medtronic Talent Cook Zenith Guidant Ancure Terumo Anaconda Coronary artery disease Diabetes Preoperative embolization Hyperlipidemia Hypertension Peripheral vascular disease Smoking Never Former Current Treated with statin Operative adjunctive maneuver Bifurcated graft
Total (N ¼ 489; 100%)a
IFU adherent (n ¼ 284; 58.1%)
IFU nonadherent (n ¼ 205; 41.9%)
P valueb
49 (10.2)
19 (6.7)
30 (14.6)
74.7 6 7.6 75.0 [70.0-80.0]
74.8 6 8.0 75.0 [70.0-81.0]
74.6 6 7.0 75.0 [70.0-79.0]
<.01 .77c
351 (71.8) 138 (28.2)
196 (69.0) 88 (31.0)
155 (75.6) 50 (24.4)
5.8 6 1.0 5.6 [5.2-6.2] 298 (60.9)
5.7 6 0.9 5.5 [5.1-6.0] 158 (55.6)
6.0 6 1.2 5.8 [5.3-6.5] 140 (68.3)
.11
107 143 13 224 1 1 235 120 39 374 416 103
(21.9) (29.2) (2.7) (45.8) (0.2) (0.2) (78.3) (25.8) (8.0) (76.5) (85.1) (21.1)
55 101 6 121 0 1 126 76 29 217 235 58
(19.4) (35.6) (2.1) (42.6) (0.0) (0.4) (74.1) (27.4) (10.2) (76.4) (82.8) (20.4)
52 42 7 103 1 0 109 44 10 157 181 45
(25.4) (20.5) (3.4) (50.2) (0.5) (0.0) (83.9) (23.4) (4.9) (76.6) (88.3) (22.0)
313 81 95 340 19 412
(64.0) (16.6) (19.4) (69.5) (3.9) (84.3)
190 40 54 196 10 235
(66.9) (14.1) (19.0) (69.0) (3.5) (82.8)
123 41 41 144 9 177
(60.0) (20.0) (20.0) (70.2) (4.4) (86.3)
<.001c <.01 <.01d
.04 .33 .03 .96 .09 .68 .18
.77 .62 .28
AAA, Abdominal aortic aneurysm; IQR, interquartile range; SD, standard deviation. a Values are given as number (%) unless otherwise noted. Because of rounding, group percentages may not total 100. b For comparisons between the IFU-adherent and IFU-nonadherent groups. c Comparison result using Wilcoxon-Mann-Whitney test. d P value denoted for significance in AneuRx group distribution.
adjunctive maneuver and bifurcated graft, intraoperative endoleak, and postoperative type I or type III endoleak) with adverse events, all-cause mortality, and ARM. Having any reintervention during the follow-up period was also considered a risk factor for all-cause mortality and ARM. Cox proportional hazards models were used to identify factors predictive of adverse events, overall mortality, or ARM, with the threshold of significance set at P < .05. Stepwise Cox proportional hazards models were used to identify risk factors for adverse events and ARM because of the small number of patients with the outcomes of interest (adverse events, 48; ARM, 10). The significance level to enter and to remain in both stepwise models was set at P ¼ .05. All analyses were performed with SAS 9.3 (SAS Institute Inc, Cary, NC) with the threshold of significance set at P < .05. RESULTS Baseline demographics and clinical characteristics. The median follow-up for the entire study cohort of 1736 patients was 2.7 years (interquartile range [IQR],
1.2-4.4 years). This included 1595 patients (91.9%) who had documented follow-up visits and 141 patients (8.1%) who were lost to follow-up (79 left the Kaiser Permanente health plan, 18 moved outside the KPNC region, and 44 discontinued imaging surveillance). During the study entry period, 489 patients in the cohort of 1736 (28.2%) had anatomic data available from before the initial EVAR procedure, including 284 (58.1%) in the IFU-adherent group and 205 (41.9%) in the IFUnonadherent group. The median follow-up time for these 489 patients was 3.1 years (IQR, 1.6-5.0 years), and the median follow-up times for the IFU-adherent group and IFU-nonadherent group were 3.2 years (IQR, 1.6-5.2 years) and 2.9 years (IQR, 1.5-4.8 years), respectively (P ¼ .46). Of these 489 patients, 49 (10.0%) were lost to follow-up, including 28 who left KPNC, 10 who moved outside the region, and 11 who discontinued imaging surveillance. Bivariate comparisons of the baseline demographic and clinical characteristics of interest are shown in Table I. All comparisons showed that the two groups were similar,
JOURNAL OF VASCULAR SURGERY May 2015
1154 Walker et al
Table II. Anatomic instructions for use (IFU) criteria and rate of IFU adherence by endovascular aortic aneurysm repair (EVAR) device types in 489 patients with preoperative anatomic data at the initial EVAR procedure
Anatomic criteria Neck length, mm Neck angle, degrees Neck diameter, mm Iliac fixation length, mm Iliac diameter, mm IFU nonadherent by device type
Cook Zenitha Medtronic AneuRxb (n ¼ 224; 45.8%) (n ¼ 143; 29.2%) $15 #60 18-28 $15 10-20 103 (46.0%)
Gore Excluderc (n ¼ 107; 21.9%)
Medtronic Talentd (n ¼ 13; 2.7%)
Guidant Ancuree (n ¼ 1; 0.2%)
Terumo Anacondaf (n ¼ 1; 0.2%)
$15 #60 19-26 $10 10-18.5 52 (48.6%)
$10 #60 18-32 $15 8-22 7 (53.9%)
$15 NS 18-26 $20 <13.5 1 (100.0%)
$15 #90 16-31 $20 8.5-21.0 0 (0.0%)
$15 #45 18-25 NS NS 42 (29.4%)
NS, Not specified. a https://www.cookmedical.com/data/IFU_PDF/T_ZAAAF_REV4.PDF. b http://www.accessdata.fda.gov/cdrh_docs/pdf/P990020c.pdf. c http://www.goremedical.com/resources/dam/assets/AH0313-ML4_EN_US.pdf. d http://manuals.medtronic.com/wcm/groups/mdtcom_sg/@emanuals/@era/@cardio/documents/documents/m707213b001_cont_20080415.pdf. e http://www.accessdata.fda.gov/cdrh_docs/pdf/P990017c.pdf. f http://www.vascutek.com/Downloads/IFUs/Anaconda/301-130_2%20Anaconda%20full%20language%20IFU.pdf#page¼4.
Table III. Anatomic characteristics of the 489 patients with available anatomic data Median (IQR) Anatomic criterion Neck length, mm Neck angle, degrees Neck diameter (D2), mm Iliac landing zone Multiplec
Reason for nonadherence, No. (%)a 128 21 15 0 41
(62.4) (10.2) (7.3) (0.0) (20.0)
Adherent
Nonadherent
P valueb
26.5 (20.7-35.0) 35.5 (26.4-44.2) 24.5 (22.5-26.7)
9.9 (6.0-13.9) 43.3 (31.8-55.5) 23.5 (21.5-25.9)
<.001 <.001 <.01
IQR, Interquartile range. a Percentage of nonadherence group only (n ¼ 205). b Wilcoxon-Mann-Whitney test. c Multiple was defined as two or more of the anatomic criteria listed in the table.
with four notable exceptions: female patients made up 14.6% of the IFU-nonadherent group and 6.7% of the IFU-adherent group (P < .01); aneurysms in the IFUnonadherent group had a larger median diameter, 5.8 cm (IQR, 5.3-6.5 cm) vs 5.5 cm (IQR, 5.1-6.0 cm) (P < .001); the IFU-adherent group had a higher proportion of patients with preoperative hypogastric embolization than the IFU-nonadherent group (10.2% vs 4.9%; P ¼ .03); and the IFU adherent group had a lower proportion with coronary artery disease than the IFU-nonadherent group (74.1% vs 83.9%; P ¼ .04). The rate of intraoperative adjunctive maneuvers was similar between the two groups (3.5% vs 4.4%, respectively; P ¼ .62). Determinants of IFU adherence. Six devices were implanted, with the majority being Cook Zenith (n ¼ 224), Medtronic AneuRx (n ¼ 143), or Gore Excluder (n ¼ 107) (Table I). The specific guidelines varied according to the device implanted (Table II). Overall, 42% of the implantations were performed in patients whose anatomy did not meet the IFU guidelines (Tables I and II). The nonadherence rates were slightly higher for the Gore Excluder (49%), Medtronic Talent (54%), and Cook Zenith (46%) and lower for the Medtronic AneuRx (29%).
The single implantation of a Terumo Anaconda was done in accordance with the IFU, whereas the single implantation of a Guidant Ancure was not. Of the 205 cases that were performed outside of IFU guidelines, 128 patients (62.4%) had short infrarenal aortic neck length, 21 (10.2%) had greater than recommended aortic neck angulation, and 15 (7.3%) did not meet aortic neck diameter criteria (Table III). In addition, 41 patients (20%) had multiple anatomic issues all relating to the infrarenal neck. Bilateral femoral diameters were statistically smaller in the IFU-nonadherent group: median on the left was 7.0 mm in the adherent group vs 6.8 mm in the nonadherent group (P < .001); on the right, it was 7.1 mm vs 6.7 mm (P ¼ .01). Outcomes. There was no difference between the two groups in either of the primary outcomes of all-cause mortality and ARM. The crude all-cause mortality was 21.1% and 21.5% (P ¼ .93) in the IFU-adherent and IFUnonadherent groups, respectively, and ARM was 2.8% and 1.0% (P ¼ .20), respectively (Table IV). Unadjusted allcause and aneurysm-related survival curves comparing these two groups are shown in Figs 1 and 2. There was no difference in the rates of type I or type III endoleak,
JOURNAL OF VASCULAR SURGERY Volume 61, Number 5
Walker et al 1155
Table IV. Primary and secondary outcomes of interest during follow-up in 489 patients with preoperative anatomic data at the initial endovascular aortic aneurysm repair (EVAR) procedure Outcomes Primary All-cause mortality ARM Secondary Type I or III endoleak Adverse events Reintervention AAA sac size changeb 6 months Decrease Increase No change 1 year Decrease Increase No change 2 years Decrease Increase No change 3 years Decrease Increase No change 4 years Decrease Increase No change 5 years Decrease Increase No change
Total (N ¼ 489; 100%)
IFU adherent (n ¼ 284; 58.1%)
IFU nonadherent (n ¼ 205; 41.9%)
104 (21.3) 10 (2.0)
60 (21.1) 8 (2.8)
44 (21.5) 2 (1.0)
.93 .20
19 (3.9) 48 (9.8) 74 (15.1)
10 (3.5) 25 (8.8) 38 (13.4)
9 (4.4) 23 (11.2) 36 (17.6)
.62 .38 .20
175 (67.3) 66 (25.4) 19 (7.3)
111 (68.9) 37 (23.0) 13 (8.1)
64 (64.7) 29 (29.3) 6 (6.1)
140 (72.5) 38 (16.7) 15 (7.8)
84 (74.3) 23 (20.4) 6 (5.3)
56 (70.0) 15 (18.8) 9 (11.3)
132 (74.2) 32 (18.0) 14 (7.9)
81 (74.3) 20 (18.4) 8 (7.3)
51 (73.9) 12 (17.4) 6 (8.7)
97 (77.0) 27 (21.4) 2 (1.6)
60 (76.9) 17 (21.8) 1 (1.3)
37 (77.1) 10 (20.8) 1 (2.1)
66 (75.0) 18 (20.5) 4 (4.6)
42 (76.4) 11 (20.0) 2 (3.6)
24 (72.7) 7 (21.2) 2 (6.1)
41 (71.9) 14 (24.6) 3 (3.5)
29 (70.7) 10 (24.4) 2 (4.9)
12 (75.0) 4 (25.0) 0 (0.0)
P valuea
.48
.32
.94
>.99
.85
>.99
AAA, Abdominal aortic aneurysm; ARM, aneurysm-related mortality; IFU, instructions for use. Values are given as number (%). a For comparisons between the IFU-adherent and IFU-nonadherent groups. b AAA sac size change was defined as sac size at the time point of interest (eg, 1 year after the initial EVAR) minus the sac size before the initial EVAR. Therefore, this is limited to patients with AAA sac size measurement during follow-up imaging.
adverse events, or reintervention after the initial EVAR between the IFU-adherent and IFU-nonadherent groups (Table IV). The aneurysm sac size change over time appeared similar between the two groups (Fig 3), with no significant difference in sac size change at any time point during follow-up (Table IV). In a multivariable analysis, age $80 years was predictive of adverse events, all-cause mortality, and ARM (Table V). Other risk factors predictive of adverse events included adjunctive maneuver and endoleak during the initial EVAR procedure and type I or type III endoleak during follow-up. For all-cause mortality, additional predictors included preoperative aneurysm sac size $5.5 cm, diabetes, and peripheral vascular disease. Type I or type III endoleak after the initial EVAR was also predictive of ARM. Last, IFU nonadherence was not predictive of adverse events, all-cause mortality, or ARM (Table V). DISCUSSION As the utilization of EVAR relative to open surgery expands and the number of trained operators continues to
increase, so do the number of patients considered for EVAR who fall outside of the conventional IFU guidelines. It has been estimated that 20% of AAA patients have neck morphology that would exclude them from standard EVAR placement on the basis of IFU guidelines.5 Intuitively, one might believe that off-label use of EVAR devices would result in higher complication rates and worse outcomes. However, the importance and clinical impact of strict implantation within these guidelines have been debated in the literature. This study demonstrates that in a large population of EVAR patients undergoing repair with commercially available devices during a recent 10year period, adherence or nonadherence to IFU guidelines did not have a measurable impact on all-cause mortality, ARM, or adverse events. These findings differ from those of previous studies, in which devices placed outside of IFU guidelines were associated with worse outcomes. In their study of 565 EVAR patients, Abbruzzese et al6 showed that violation of at least one IFU parameter did not have an impact on perioperative mortality but did result in higher ARM at 1 year and
1156 Walker et al
JOURNAL OF VASCULAR SURGERY May 2015
Fig 1. Effect of instructions for use (IFU) guidelines on all-cause mortality. Kaplan-Meier survival curves for 489 endovascular aortic aneurysm repair (EVAR) patients with M2S anatomic data at the initial EVAR procedure stratified by IFU guideline status (P ¼ .82, log-rank test). All standard errors were <10%.
Fig 2. Effect of instructions for use (IFU) guidelines on aneurysm-related mortality (ARM). Kaplan-Meier survival curves for 489 endovascular aortic aneurysm repair (EVAR) patients with M2S anatomic data at the initial EVAR procedure stratified by IFU guideline status (P ¼ .17, log-rank test). All standard errors were <10%.
5 years, which was mainly attributed to graft thrombosis in those devices placed outside of IFU guidelines. There was no effect on the rate of aneurysm rupture, conversion to open repair, graft migration, or kinking. They found an
incremental effect with an increasing number of parameters outside of IFU guidelines leading to increased rates of reintervention. Similarly, Fulton et al7 showed that patients whose neck anatomy fell outside of IFU guidelines had
JOURNAL OF VASCULAR SURGERY Volume 61, Number 5
Walker et al 1157
Fig 3. Abdominal aortic aneurysm (AAA) sac size change in 489 endovascular aortic aneurysm repair (EVAR) patients with M2S anatomic data at the initial EVAR procedure: overall and stratified by instructions for use (IFU) guideline status. All standard errors were <10%.
Table V. Multivariable predictors of all-cause mortality and aneurysm-related mortality (ARM) in 489 patients with preoperative anatomic data at the initial endovascular aortic aneurysm repair (EVAR) procedure
Predictors IFU adherence Female Age $80 years Preoperative AAA sac size $5.5 cm Treated with statin Coronary artery disease Diabetes Hyperlipidemia Hypertension Peripheral vascular disease Preoperative embolization Smoking Neverb Current Former Operative adjunctive maneuver Bifurcated graft Any intraoperative endoleak Any postoperative type I or III endoleak Any reinterventionc
Adverse eventsa (n ¼ 48)
All-cause mortality (n ¼ 104)
HR (95% CI)
HR (95% CI)
P value
1.0 1.2 1.9 2.4 1.2 0.8 1.6 0.7 1.4 1.6 1.4
(0.7-1.5) (0.6-2.3) (1.2-2.8) (1.4-3.8) (0.8-2.1) (0.5-1.3) (1.04-2.6) (0.4-1.2) (0.8-2.6) (1.04-2.6) (0.7-3.0)
.91 .60 <.01 <.001 .39 .38 .03 .15 .27 .04 .33
0.7 1.1 0.8 1.3 1.0 1.3 0.9
1.0 (0.3-1.2) (0.6-1.8) (0.3-2.4) (0.7-2.3) (0.5-1.8) (0.6-3.0) (0.5-1.5)
.18 .85 .72 .36 .88 .52 .62
2.2 (1.2-4.0)
P value
<.01
1.0 4.2 (1.4-12.2)
<.01
0.1 (0.01-0.6) 8.8 (4.5-17.2) N/A
.02 <.001
ARMa (n ¼ 10) HR (95% CI)
P value
6.7 (1.7-25.9)
<.01
1.0
8.6 (2.2-33.6)
<.01
AAA, Abdominal aortic aneurysm; CI, confidence interval; HR, hazard ratio; IFU, instructions for use; N/A, not applicable. a Because of the small number of patients with adverse events and ARM, a “stepwise” method was used in the multivariable regression model; therefore, the results included in the table are limited to predictors with effects meeting the significance level of .05 to enter and to remain in the model. b Referent group. c Used in the multivariable regression models for all-cause mortality and ARM only.
higher rates of graft migration, device-related complications, and reintervention. Schanzer et al3 demonstrated in their series that nonadherence to IFU guidelines was associated with higher rate of sac enlargement, but this study was hampered by lack of information about device use, clinical complications, and outcomes as well as an unknown incidence of reintervention.
Several recent studies have shown that patients who undergo EVAR outside of IFU guidelines have favorable outcomes. Lee et al8 reported results of 218 patients, one third of whom fell outside of IFU criteria, and showed that these patients had no difference in morbidity, mortality, rate of proximal endoleak, or reintervention after a mean follow-up of 35 months (range,
JOURNAL OF VASCULAR SURGERY May 2015
1158 Walker et al
12-72 months). Similarly, the patients in the current study, 42% of whom fell outside of the IFU guidelines for the specific device they received, exhibited no significant difference in any of the primary or secondary outcomes. All of the criteria that determined IFU nonadherence were related to the morphology of the infrarenal neck, which has been considered to be the primary driver of EVAR durability (ie, no isolated iliac sizing or landing zone issues were observed). Thus, despite no increase in intraoperative adjunctive maneuvers, there was also no observed increase in type I or type III endoleaks due to these suboptimal infrarenal aortic necks. Moreover, the IFU-nonadherent group had significantly more women and larger preoperative aneurysm sac size, two modifiers that are known to increase morbidity.9 This is possibly a reflection of growing operator experience over time and careful patient selection, which are of paramount importance in this population. Given the differing results of these recent studies, it is not surprising that a recent meta-analysis by Antoniou et al10 found that there was insufficient clinical information to provide solid evidence for or against the safety of EVAR in patients with neck anatomy that did not fall within the parameters of the IFU. In their analysis of 1559 patients, neck anatomy did not significantly affect perioperative mortality, but those with non-IFU neck anatomy did experience higher perioperative morbidity. The authors speculated that inclusion of patients unable to tolerate an open operation may have influenced the outcomes in the nonIFU group. Our results may reflect selection bias in that not all clinicians in our health system use M2S for preoperative planning, thereby explaining the modest 28% proportion of the overall cohort available for this analysis. M2S utilization in this cohort may range from routine clinical use to selective case-by-case submission, but we do not have user-specific preferences on M2S or other three-dimensional image reconstruction modalities and thus cannot explore this area for further analysis. Furthermore, despite detailed aneurysm morphologic data, anatomic considerations such as number and size of lumbar vessels, inferior mesenteric artery patency and size, calcification, and thrombus burden were lacking in our data set, and we were therefore unable to analyze the effect of these variables. The decision-making of each clinician to determine the suitability of EVAR, the device and specific size to be used, and whether and when to reintervene represents an additional source of variation, although our findings reflect a “real-world” community-based practice pattern over a broad geographic region. In addition, although graft type was known, the data on graft sizing were incomplete, which limited our ability to specifically and fairly assess device performance. We also recognize it is possible that the differences observed in the proportions of patients with endoleak, adverse events, or reintervention may be due to type II error. A post hoc analysis of sample size showed that patient cohorts of 7573, 2535, and 1212 in each group, respectively, were needed to show a statistical
difference. Finally, it is possible that not all postoperative events pertaining to IFU adherence occurred within our follow-up period. We therefore might be underestimating the true impact of nonadherence to IFU guidelines. Our ongoing stent graft surveillance registry will allow us to capture events that occur with a longer delay. Although the 10% loss to follow-up rate was low, a worst-case analysis (high rate of events in the group lost to follow-up), although unlikely, leaves open the possibility of different findings. CONCLUSIONS In our cohort of EVAR patients with detailed preoperative anatomic information and long-term follow-up, all-cause mortality and ARM were unaffected by IFU adherence, despite a higher proportion of female patients and larger aneurysms in the nonadherent group. In addition, rates of late endoleak and reintervention were similar, suggesting that lack of IFU-based anatomic suitability was not a driver of outcomes. We thank Ann Rhoades, RN, for her assistance in all of the data collection and preparation of this manuscript and the Kaiser Permanente National Implant Registries Group for their assistance in data collection and analysis. AUTHOR CONTRIBUTIONS Conception and design: JW, RC Analysis and interpretation: JW, LT, PG, LC, HH, SO, BH, RC Data collection: JW, LT Writing the article: JW Critical revision of the article: JW, LT, PG, LC, HH, SO, BH, RC Final approval of the article: JW, LT, RC Statistical analysis: JW, LT Obtained funding: RC Overall responsibility: RC REFERENCES 1. Parodi JC, Palmaz JC, Barone HD. Transfemoral intraluminal graft implantation for abdominal aortic aneurysms. Ann Vasc Surg 1991;5: 491-9. 2. Schermerhorn ML, O’Malley AJ, Jhaveri A, Cotterill P, Pomposelli F, Landon BE. Endovascular vs. open repair of abdominal aortic aneurysms in the Medicare population. N Engl J Med 2008;358: 464-74. 3. Schanzer A, Greenberg RK, Hevelone N, Robinson WP, Eslami MH, Goldberg RJ, et al. Predictors of abdominal aortic aneurysm sac enlargement after endovascular repair. Circulation 2011;123:2848-55. 4. Chaikof EL, Blankensteijn JD, Harris PL, White GH, Zarins CK, Bernhard VM, et al. Reporting standards for endovascular aortic aneurysm repair. J Vasc Surg 2002;35:1048-60. 5. Malina M, Resch T, Sonesson B. EVAR and complex anatomy: an update on fenestrated and branched stent grafts. Scand J Surg 2008;97: 195-204. 6. Abbruzzese TA, Kwolek CJ, Brewster DC, Chung TK, Kang J, Conrad MF, et al. Outcomes following endovascular abdominal aortic aneurysm repair (EVAR): an anatomic and device-specific analysis. J Vasc Surg 2008;48:19-28.
JOURNAL OF VASCULAR SURGERY Volume 61, Number 5
7. Fulton JJ, Farber MA, Sanchez LA, Godshall CJ, Marston WA, Mendes R, et al. Effect of challenging neck anatomy on mid-term migration rates in AneuRx endografts. J Vasc Surg 2006;44:932-7; discussion: 937. 8. Lee JT, Ullery BW, Zarins CK, Olcott C, Harris EJ Jr, Dalman RL. EVAR deployment in anatomically challenging necks outside the IFU. Eur J Vasc Endovasc Surg 2013;46:65-73. 9. Chang RW, Goodney P, Tucker LY, Okuhn S, Hua H, Rhoades A, et al. Ten-year results of endovascular abdominal aortic aneurysm
Walker et al 1159
repair from a large multicenter registry. J Vasc Surg 2013;58: 324-32. 10. Antoniou GA, Georgiadis GS, Antoniou SA, Kuhan G, Murray D. A meta-analysis of outcomes of endovascular abdominal aortic aneurysm repair in patients with hostile and friendly neck anatomy. J Vasc Surg 2013;57:527-38.
Submitted Sep 10, 2014; accepted Dec 18, 2014.
Adherence to endovascular aortic aneurysm repair device instructions for use guidelines has no impact on outcomes Joy Walker, MD,a Lue-Yen Tucker, BA,b Philip Goodney, MD,c Leah Candell, MD,d Hong Hua, MD,e Steven Okuhn, MD,e Bradley Hill, MD,f and Robert W. Chang, MD,g San Francisco, Oakland, and Santa Clara, Calif; and Lebanon, NH Objective: Prior reports have suggested unfavorable outcomes after endovascular aortic aneurysm repair (EVAR) performed outside of the recommended instructions for use (IFU) guidelines. We report our long-term EVAR experience in a large multicenter registry with regard to adherence to IFU guidelines. Methods: Between 2000 and 2010, 489 of 1736 patients who underwent EVAR had preoperative anatomic measurements obtained from the M2S, Inc, imaging database (West Lebanon, NH). We examined outcomes in these patients with regard to whether they had met the device-specific IFU criteria. Primary outcomes were all-cause mortality and aneurysmrelated mortality. Secondary outcomes were endoleak status, adverse events, reintervention, and aneurysm sac size change. Results: The median follow-up for the 489 patients was 3.1 years (interquartile range, 1.6-5.0 years); 58.1% (n [ 284) had EVAR performed within IFU guidelines (IFU-adherent group), and 41.9% (n [ 205) had EVAR performed outside of IFU guidelines (IFU-nonadherent group). Preoperative anatomic data showed that 62.4% of the IFU-nonadherent group had short neck length, 10.2% had greater angulation than recommended, 7.3% did not meet neck diameter criteria, and 20% had multiple anatomic issues. A small portion (n [ 49; 10%) of the 489 patients were lost to follow-up because of leaving membership enrollment (n [ 28), moving outside the region (n [ 10), or discontinuing image surveillance (n [ 11). There was no significant difference in any of the primary or secondary outcomes between the IFUadherent and IFU-nonadherent groups. Aneurysm sac size change at any time point during follow-up also did not differ significantly between the two groups. A Cox proportional hazard model showed that IFU nonadherence was not predictive of all-cause mortality (hazard ratio, 1.0; P [ .91). Similarly, IFU nonadherence was not identified as a risk factor for aneurysm-related mortality or adverse events in stepwise Cox proportional hazards models. Conclusions: In our cohort of EVAR patients with detailed preoperative anatomic information and long-term follow-up, overall mortality and aneurysm-related mortality were unaffected by IFU adherence. In addition, rates of endoleak and reintervention after initial EVAR were similar, suggesting that lack of IFU-based anatomic suitability was not a driver of outcomes. (J Vasc Surg 2015;61:1151-9.)
Since its first description in 1991, endovascular aortic aneurysm repair (EVAR) has become a commonplace and well-accepted alternative to open abdominal aortic aneurysm (AAA) repair.1,2 Proper patient selection based on anatomic From the Division of Surgery, University of California, San Francisco, San Franciscoa; the Division of Research, Kaiser Permanente, Oaklandb; the Division of Vascular Surgery, Dartmouth-Hitchcock Medical Center, Lebanonc; the Division of Surgery, University of California, San Francisco-East Bay Medical Center, Oaklandd; the Division of Vascular Surgery, The Permanente Medical Group, San Franciscoe; the Division of Vascular Surgery, The Permanente Medical Group, Santa Claraf; and the Division of Vascular Surgery, The Permanente Medical Group, South San Francisco.g Author conflict of interest: none. Presented at Scientific Session V of the Twenty-ninth Annual Meeting of the Western Vascular Society, Coronado, Calif, September 20-23, 2014. Reprint requests: Robert W. Chang, MD, Vascular and Endovascular Surgery, Kaiser Permanente, South San Francisco, 1200 El Camino Real, South San Francisco, CA 94080 (e-mail: [email protected]). The editors and reviewers of this article have no relevant financial relationships to disclose per the JVS policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. 0741-5214 Copyright Ó 2015 by the Society for Vascular Surgery. Published by Elsevier Inc. http://dx.doi.org/10.1016/j.jvs.2014.12.053
criteria has become critical to ensuring satisfactory long-term results. To that end, each endovascular device manufacturer publishes an instructions for use (IFU) guideline that specifies anatomic criteria for “correct” use of the EVAR device. These recommendations are made on the basis of preclinical engineering assessments and clinical study results. The guidelines specify appropriate aortic neck diameter, aortic neck length, aortic neck angle, and iliac artery morphology. Many clinicians believe that outcomes after EVAR largely depend on whether the devices are used in accordance with the IFU guidelines. A recent paper documented the incidence of IFU nonadherence in registry data sets but had notable gaps in availability of device-specific and patient outcome data.3 We analyzed outcomes of EVAR patients in a longitudinal registry for whom detailed preoperative anatomic data were available, with the objective of determining whether IFU adherence affects outcomes after EVAR. METHODS Kaiser Permanente Northern California (KPNC) is an integrated health care delivery system that offers multispecialty care for more than 3 million members. Implementation of 1151
1152 Walker et al
digital health records has allowed access to all arenas of clinical information. Data for 1736 patients who underwent EVAR procedures in 17 KPNC medical centers were collected in a clinical registry from 2000 to 2010. This study protocol was approved by the KPNC Institutional Review Board and was funded for one calendar year in 2011 by the KPNC Community Benefit Research Grant Program. This in-depth evaluation of IFU adherence was funded by the KPNC Residency Programs. Informed consent was waived by the local Institutional Review Board, given that the study was retrospective and the data de-identified for analysis. Data collection and follow-up. Beginning in 2000, relevant clinical information for patients undergoing EVAR was collected by trained research nurses, with December 31, 2010, as the last follow-up date. Data collected from electronic medical records included baseline preoperative demographic data and clinical characteristics (sex, age, aneurysm sac size, comorbidities [coronary artery disease, diabetes mellitus, hypertension, hyperlipidemia, and peripheral vascular disease], smoking status, and statin history). Device type and operative details were collected from the operative report and device entry forms. Information on adjunctive maneuvers (placement of additional aortic cuffs or stents, graft limb extensions, exploratory laparotomy/conversion, renal stenting/snorkel, and femoral-femoral bypass) was also collected. Decisions regarding indications for surgery, suitability for endovascular repair, device selection, and need for secondary intervention were made at the discretion of the operating surgeon. Data recorded in our registry during the follow-up period were also collected and included aneurysm rupture, major adverse event (ie, conversion to open repair, major embolic event, graft infection requiring explantation, device migration, loss of device patency requiring reintervention, and other miscellaneous complications that substantially affected clinical outcome), types of reintervention, AAA sac size, endoleak, leaving KPNC membership, moving outside the region, and mortality. Postoperative follow-up varied across medical centers but generally involved a 1-month postoperative computed tomography (CT) scan followed by serial CT imaging at regular intervals ranging from every 3 months to every 12 months as dictated by clinical circumstances. Imaging was accompanied by clinical follow-up. As follow-up progressed, there was a considerable amount of variability regarding the timing, modality, and use of contrast material in CT imaging. However, the preoperative and first postoperative imaging generally consisted of CT scans with and without contrast material and arterial and venous phases with 1.25-mm slices. In the absence of sac growth or endoleak, intravenous contrast material was sometimes withheld for subsequent examinations at the discretion of the treating physician. Device migration was reported if it required intervention or if adequate seal was lost, usually when reduced to <10 mm of the circumferential apposition length. Endoleak was classified according to established reporting standards4 and was typically detected by CT scan, confirmatory angiography, or, more rarely, ultrasound. Determination of adherence to IFU guidelines. Our clinical EVAR registry was cross-referenced with data from
JOURNAL OF VASCULAR SURGERY May 2015
the M2S, Inc, imaging repository (West Lebanon, NH). The M2S anatomic registry (hereafter termed M2S) comprehensively assesses detailed anatomic data from CT scans submitted to them from the clinician. Using standardized algorithms, M2S creates three-dimensional computer models from CT images of aortic aneurysms with semiautomated quantification of multiple measurements of interest. Measurements of interest corresponded with major determinants of IFU adherence and therefore primarily involved the proximal infrarenal aortic neck (eg, neck length, diameter, and angulation). This in-depth study was a subset analysis limited to patients whose initial EVAR procedures had relevant preoperative M2S analysis of CT imaging. These anatomic data and corresponding measurements were then used to determine adherence to IFU guidelines. Patients whose initial EVAR procedures were performed within the IFU guidelines were classified as the IFU-adherent group, and those outside the IFU guidelines as the IFU-nonadherent group. The evaluation of specific guidelines varied according to the specific device implanted. Outcome variables. The primary outcomes were allcause mortality and aneurysm-related mortality (ARM); the latter was defined as death within 30 days of the initial EVAR or of a secondary procedure related to aneurysm rupture or a major adverse event. Secondary outcome variables examined were type I or type III endoleak, major adverse events, need for reintervention, and change in aneurysm sac size over time. Sac size was assessed at several time points during follow-up (ie, 2 months, 6 months, 9 months, 12 months, 18 months, 2 years, 3 years, 4 years, and 5 years after the initial EVAR). Sac size measurements were accepted if performed within 1.5 months before or after each time point within the first 12 months or within 3 months before or after time points after 12 months. Statistical methods. Rates of categorical demographic (sex, age groups, racial/ethnic groups) and clinical characteristics (comorbidities, smoking status, statin history, preoperative embolization, adjunctive maneuver or bifurcated graft during the initial EVAR procedure, and aneurysm sac growth status at various time points during follow-up) were compared between the IFU-adherent and IFUnonadherent groups with c2 tests or Fisher exact tests as appropriate. Preoperative AAA sac size, age at initial EVAR, and M2S anatomic data (ie, neck length, neck diameter, neck angle, and femoral diameter) were not normally distributed; therefore, nonparametric WilcoxonMann-Whitney tests were used to compare medians. Survival analysis was performed with the Kaplan-Meier method to estimate the survival function in 489 patients with M2S stratified by IFU guideline status (adherence vs nonadherence), and survival functions were compared between these two groups by the log-rank test. Before fitting of the multivariable regression models, bivariate analyses were performed to evaluate the association of risk factors of interest (IFU adherence, sex, age, preoperative AAA sac size, history of statin, comorbidities, smoking status, preoperative embolization, operative
JOURNAL OF VASCULAR SURGERY Volume 61, Number 5
Walker et al 1153
Table I. Baseline preoperative demographic and clinical characteristics of 489 patients with preoperative anatomic data at the initial endovascular aortic aneurysm repair (EVAR) procedure, stratified by instructions for use (IFU) guideline status Characteristics Female Age, years Mean 6 SD Median [IQR] Age group, years #79 $80 Preoperative AAA size, cm Mean 6 SD Median [IQR] Preoperative AAA sac size $5.5 cm EVAR device Gore Excluder Medtronic AneuRx Medtronic Talent Cook Zenith Guidant Ancure Terumo Anaconda Coronary artery disease Diabetes Preoperative embolization Hyperlipidemia Hypertension Peripheral vascular disease Smoking Never Former Current Treated with statin Operative adjunctive maneuver Bifurcated graft
Total (N ¼ 489; 100%)a
IFU adherent (n ¼ 284; 58.1%)
IFU nonadherent (n ¼ 205; 41.9%)
P valueb
49 (10.2)
19 (6.7)
30 (14.6)
74.7 6 7.6 75.0 [70.0-80.0]
74.8 6 8.0 75.0 [70.0-81.0]
74.6 6 7.0 75.0 [70.0-79.0]
<.01 .77c
351 (71.8) 138 (28.2)
196 (69.0) 88 (31.0)
155 (75.6) 50 (24.4)
5.8 6 1.0 5.6 [5.2-6.2] 298 (60.9)
5.7 6 0.9 5.5 [5.1-6.0] 158 (55.6)
6.0 6 1.2 5.8 [5.3-6.5] 140 (68.3)
.11
107 143 13 224 1 1 235 120 39 374 416 103
(21.9) (29.2) (2.7) (45.8) (0.2) (0.2) (78.3) (25.8) (8.0) (76.5) (85.1) (21.1)
55 101 6 121 0 1 126 76 29 217 235 58
(19.4) (35.6) (2.1) (42.6) (0.0) (0.4) (74.1) (27.4) (10.2) (76.4) (82.8) (20.4)
52 42 7 103 1 0 109 44 10 157 181 45
(25.4) (20.5) (3.4) (50.2) (0.5) (0.0) (83.9) (23.4) (4.9) (76.6) (88.3) (22.0)
313 81 95 340 19 412
(64.0) (16.6) (19.4) (69.5) (3.9) (84.3)
190 40 54 196 10 235
(66.9) (14.1) (19.0) (69.0) (3.5) (82.8)
123 41 41 144 9 177
(60.0) (20.0) (20.0) (70.2) (4.4) (86.3)
<.001c <.01 <.01d
.04 .33 .03 .96 .09 .68 .18
.77 .62 .28
AAA, Abdominal aortic aneurysm; IQR, interquartile range; SD, standard deviation. a Values are given as number (%) unless otherwise noted. Because of rounding, group percentages may not total 100. b For comparisons between the IFU-adherent and IFU-nonadherent groups. c Comparison result using Wilcoxon-Mann-Whitney test. d P value denoted for significance in AneuRx group distribution.
adjunctive maneuver and bifurcated graft, intraoperative endoleak, and postoperative type I or type III endoleak) with adverse events, all-cause mortality, and ARM. Having any reintervention during the follow-up period was also considered a risk factor for all-cause mortality and ARM. Cox proportional hazards models were used to identify factors predictive of adverse events, overall mortality, or ARM, with the threshold of significance set at P < .05. Stepwise Cox proportional hazards models were used to identify risk factors for adverse events and ARM because of the small number of patients with the outcomes of interest (adverse events, 48; ARM, 10). The significance level to enter and to remain in both stepwise models was set at P ¼ .05. All analyses were performed with SAS 9.3 (SAS Institute Inc, Cary, NC) with the threshold of significance set at P < .05. RESULTS Baseline demographics and clinical characteristics. The median follow-up for the entire study cohort of 1736 patients was 2.7 years (interquartile range [IQR],
1.2-4.4 years). This included 1595 patients (91.9%) who had documented follow-up visits and 141 patients (8.1%) who were lost to follow-up (79 left the Kaiser Permanente health plan, 18 moved outside the KPNC region, and 44 discontinued imaging surveillance). During the study entry period, 489 patients in the cohort of 1736 (28.2%) had anatomic data available from before the initial EVAR procedure, including 284 (58.1%) in the IFU-adherent group and 205 (41.9%) in the IFUnonadherent group. The median follow-up time for these 489 patients was 3.1 years (IQR, 1.6-5.0 years), and the median follow-up times for the IFU-adherent group and IFU-nonadherent group were 3.2 years (IQR, 1.6-5.2 years) and 2.9 years (IQR, 1.5-4.8 years), respectively (P ¼ .46). Of these 489 patients, 49 (10.0%) were lost to follow-up, including 28 who left KPNC, 10 who moved outside the region, and 11 who discontinued imaging surveillance. Bivariate comparisons of the baseline demographic and clinical characteristics of interest are shown in Table I. All comparisons showed that the two groups were similar,
JOURNAL OF VASCULAR SURGERY May 2015
1154 Walker et al
Table II. Anatomic instructions for use (IFU) criteria and rate of IFU adherence by endovascular aortic aneurysm repair (EVAR) device types in 489 patients with preoperative anatomic data at the initial EVAR procedure
Anatomic criteria Neck length, mm Neck angle, degrees Neck diameter, mm Iliac fixation length, mm Iliac diameter, mm IFU nonadherent by device type
Cook Zenitha Medtronic AneuRxb (n ¼ 224; 45.8%) (n ¼ 143; 29.2%) $15 #60 18-28 $15 10-20 103 (46.0%)
Gore Excluderc (n ¼ 107; 21.9%)
Medtronic Talentd (n ¼ 13; 2.7%)
Guidant Ancuree (n ¼ 1; 0.2%)
Terumo Anacondaf (n ¼ 1; 0.2%)
$15 #60 19-26 $10 10-18.5 52 (48.6%)
$10 #60 18-32 $15 8-22 7 (53.9%)
$15 NS 18-26 $20 <13.5 1 (100.0%)
$15 #90 16-31 $20 8.5-21.0 0 (0.0%)
$15 #45 18-25 NS NS 42 (29.4%)
NS, Not specified. a https://www.cookmedical.com/data/IFU_PDF/T_ZAAAF_REV4.PDF. b http://www.accessdata.fda.gov/cdrh_docs/pdf/P990020c.pdf. c http://www.goremedical.com/resources/dam/assets/AH0313-ML4_EN_US.pdf. d http://manuals.medtronic.com/wcm/groups/mdtcom_sg/@emanuals/@era/@cardio/documents/documents/m707213b001_cont_20080415.pdf. e http://www.accessdata.fda.gov/cdrh_docs/pdf/P990017c.pdf. f http://www.vascutek.com/Downloads/IFUs/Anaconda/301-130_2%20Anaconda%20full%20language%20IFU.pdf#page¼4.
Table III. Anatomic characteristics of the 489 patients with available anatomic data Median (IQR) Anatomic criterion Neck length, mm Neck angle, degrees Neck diameter (D2), mm Iliac landing zone Multiplec
Reason for nonadherence, No. (%)a 128 21 15 0 41
(62.4) (10.2) (7.3) (0.0) (20.0)
Adherent
Nonadherent
P valueb
26.5 (20.7-35.0) 35.5 (26.4-44.2) 24.5 (22.5-26.7)
9.9 (6.0-13.9) 43.3 (31.8-55.5) 23.5 (21.5-25.9)
<.001 <.001 <.01
IQR, Interquartile range. a Percentage of nonadherence group only (n ¼ 205). b Wilcoxon-Mann-Whitney test. c Multiple was defined as two or more of the anatomic criteria listed in the table.
with four notable exceptions: female patients made up 14.6% of the IFU-nonadherent group and 6.7% of the IFU-adherent group (P < .01); aneurysms in the IFUnonadherent group had a larger median diameter, 5.8 cm (IQR, 5.3-6.5 cm) vs 5.5 cm (IQR, 5.1-6.0 cm) (P < .001); the IFU-adherent group had a higher proportion of patients with preoperative hypogastric embolization than the IFU-nonadherent group (10.2% vs 4.9%; P ¼ .03); and the IFU adherent group had a lower proportion with coronary artery disease than the IFU-nonadherent group (74.1% vs 83.9%; P ¼ .04). The rate of intraoperative adjunctive maneuvers was similar between the two groups (3.5% vs 4.4%, respectively; P ¼ .62). Determinants of IFU adherence. Six devices were implanted, with the majority being Cook Zenith (n ¼ 224), Medtronic AneuRx (n ¼ 143), or Gore Excluder (n ¼ 107) (Table I). The specific guidelines varied according to the device implanted (Table II). Overall, 42% of the implantations were performed in patients whose anatomy did not meet the IFU guidelines (Tables I and II). The nonadherence rates were slightly higher for the Gore Excluder (49%), Medtronic Talent (54%), and Cook Zenith (46%) and lower for the Medtronic AneuRx (29%).
The single implantation of a Terumo Anaconda was done in accordance with the IFU, whereas the single implantation of a Guidant Ancure was not. Of the 205 cases that were performed outside of IFU guidelines, 128 patients (62.4%) had short infrarenal aortic neck length, 21 (10.2%) had greater than recommended aortic neck angulation, and 15 (7.3%) did not meet aortic neck diameter criteria (Table III). In addition, 41 patients (20%) had multiple anatomic issues all relating to the infrarenal neck. Bilateral femoral diameters were statistically smaller in the IFU-nonadherent group: median on the left was 7.0 mm in the adherent group vs 6.8 mm in the nonadherent group (P < .001); on the right, it was 7.1 mm vs 6.7 mm (P ¼ .01). Outcomes. There was no difference between the two groups in either of the primary outcomes of all-cause mortality and ARM. The crude all-cause mortality was 21.1% and 21.5% (P ¼ .93) in the IFU-adherent and IFUnonadherent groups, respectively, and ARM was 2.8% and 1.0% (P ¼ .20), respectively (Table IV). Unadjusted allcause and aneurysm-related survival curves comparing these two groups are shown in Figs 1 and 2. There was no difference in the rates of type I or type III endoleak,
JOURNAL OF VASCULAR SURGERY Volume 61, Number 5
Walker et al 1155
Table IV. Primary and secondary outcomes of interest during follow-up in 489 patients with preoperative anatomic data at the initial endovascular aortic aneurysm repair (EVAR) procedure Outcomes Primary All-cause mortality ARM Secondary Type I or III endoleak Adverse events Reintervention AAA sac size changeb 6 months Decrease Increase No change 1 year Decrease Increase No change 2 years Decrease Increase No change 3 years Decrease Increase No change 4 years Decrease Increase No change 5 years Decrease Increase No change
Total (N ¼ 489; 100%)
IFU adherent (n ¼ 284; 58.1%)
IFU nonadherent (n ¼ 205; 41.9%)
104 (21.3) 10 (2.0)
60 (21.1) 8 (2.8)
44 (21.5) 2 (1.0)
.93 .20
19 (3.9) 48 (9.8) 74 (15.1)
10 (3.5) 25 (8.8) 38 (13.4)
9 (4.4) 23 (11.2) 36 (17.6)
.62 .38 .20
175 (67.3) 66 (25.4) 19 (7.3)
111 (68.9) 37 (23.0) 13 (8.1)
64 (64.7) 29 (29.3) 6 (6.1)
140 (72.5) 38 (16.7) 15 (7.8)
84 (74.3) 23 (20.4) 6 (5.3)
56 (70.0) 15 (18.8) 9 (11.3)
132 (74.2) 32 (18.0) 14 (7.9)
81 (74.3) 20 (18.4) 8 (7.3)
51 (73.9) 12 (17.4) 6 (8.7)
97 (77.0) 27 (21.4) 2 (1.6)
60 (76.9) 17 (21.8) 1 (1.3)
37 (77.1) 10 (20.8) 1 (2.1)
66 (75.0) 18 (20.5) 4 (4.6)
42 (76.4) 11 (20.0) 2 (3.6)
24 (72.7) 7 (21.2) 2 (6.1)
41 (71.9) 14 (24.6) 3 (3.5)
29 (70.7) 10 (24.4) 2 (4.9)
12 (75.0) 4 (25.0) 0 (0.0)
P valuea
.48
.32
.94
>.99
.85
>.99
AAA, Abdominal aortic aneurysm; ARM, aneurysm-related mortality; IFU, instructions for use. Values are given as number (%). a For comparisons between the IFU-adherent and IFU-nonadherent groups. b AAA sac size change was defined as sac size at the time point of interest (eg, 1 year after the initial EVAR) minus the sac size before the initial EVAR. Therefore, this is limited to patients with AAA sac size measurement during follow-up imaging.
adverse events, or reintervention after the initial EVAR between the IFU-adherent and IFU-nonadherent groups (Table IV). The aneurysm sac size change over time appeared similar between the two groups (Fig 3), with no significant difference in sac size change at any time point during follow-up (Table IV). In a multivariable analysis, age $80 years was predictive of adverse events, all-cause mortality, and ARM (Table V). Other risk factors predictive of adverse events included adjunctive maneuver and endoleak during the initial EVAR procedure and type I or type III endoleak during follow-up. For all-cause mortality, additional predictors included preoperative aneurysm sac size $5.5 cm, diabetes, and peripheral vascular disease. Type I or type III endoleak after the initial EVAR was also predictive of ARM. Last, IFU nonadherence was not predictive of adverse events, all-cause mortality, or ARM (Table V). DISCUSSION As the utilization of EVAR relative to open surgery expands and the number of trained operators continues to
increase, so do the number of patients considered for EVAR who fall outside of the conventional IFU guidelines. It has been estimated that 20% of AAA patients have neck morphology that would exclude them from standard EVAR placement on the basis of IFU guidelines.5 Intuitively, one might believe that off-label use of EVAR devices would result in higher complication rates and worse outcomes. However, the importance and clinical impact of strict implantation within these guidelines have been debated in the literature. This study demonstrates that in a large population of EVAR patients undergoing repair with commercially available devices during a recent 10year period, adherence or nonadherence to IFU guidelines did not have a measurable impact on all-cause mortality, ARM, or adverse events. These findings differ from those of previous studies, in which devices placed outside of IFU guidelines were associated with worse outcomes. In their study of 565 EVAR patients, Abbruzzese et al6 showed that violation of at least one IFU parameter did not have an impact on perioperative mortality but did result in higher ARM at 1 year and
1156 Walker et al
JOURNAL OF VASCULAR SURGERY May 2015
Fig 1. Effect of instructions for use (IFU) guidelines on all-cause mortality. Kaplan-Meier survival curves for 489 endovascular aortic aneurysm repair (EVAR) patients with M2S anatomic data at the initial EVAR procedure stratified by IFU guideline status (P ¼ .82, log-rank test). All standard errors were <10%.
Fig 2. Effect of instructions for use (IFU) guidelines on aneurysm-related mortality (ARM). Kaplan-Meier survival curves for 489 endovascular aortic aneurysm repair (EVAR) patients with M2S anatomic data at the initial EVAR procedure stratified by IFU guideline status (P ¼ .17, log-rank test). All standard errors were <10%.
5 years, which was mainly attributed to graft thrombosis in those devices placed outside of IFU guidelines. There was no effect on the rate of aneurysm rupture, conversion to open repair, graft migration, or kinking. They found an
incremental effect with an increasing number of parameters outside of IFU guidelines leading to increased rates of reintervention. Similarly, Fulton et al7 showed that patients whose neck anatomy fell outside of IFU guidelines had
JOURNAL OF VASCULAR SURGERY Volume 61, Number 5
Walker et al 1157
Fig 3. Abdominal aortic aneurysm (AAA) sac size change in 489 endovascular aortic aneurysm repair (EVAR) patients with M2S anatomic data at the initial EVAR procedure: overall and stratified by instructions for use (IFU) guideline status. All standard errors were <10%.
Table V. Multivariable predictors of all-cause mortality and aneurysm-related mortality (ARM) in 489 patients with preoperative anatomic data at the initial endovascular aortic aneurysm repair (EVAR) procedure
Predictors IFU adherence Female Age $80 years Preoperative AAA sac size $5.5 cm Treated with statin Coronary artery disease Diabetes Hyperlipidemia Hypertension Peripheral vascular disease Preoperative embolization Smoking Neverb Current Former Operative adjunctive maneuver Bifurcated graft Any intraoperative endoleak Any postoperative type I or III endoleak Any reinterventionc
Adverse eventsa (n ¼ 48)
All-cause mortality (n ¼ 104)
HR (95% CI)
HR (95% CI)
P value
1.0 1.2 1.9 2.4 1.2 0.8 1.6 0.7 1.4 1.6 1.4
(0.7-1.5) (0.6-2.3) (1.2-2.8) (1.4-3.8) (0.8-2.1) (0.5-1.3) (1.04-2.6) (0.4-1.2) (0.8-2.6) (1.04-2.6) (0.7-3.0)
.91 .60 <.01 <.001 .39 .38 .03 .15 .27 .04 .33
0.7 1.1 0.8 1.3 1.0 1.3 0.9
1.0 (0.3-1.2) (0.6-1.8) (0.3-2.4) (0.7-2.3) (0.5-1.8) (0.6-3.0) (0.5-1.5)
.18 .85 .72 .36 .88 .52 .62
2.2 (1.2-4.0)
P value
<.01
1.0 4.2 (1.4-12.2)
<.01
0.1 (0.01-0.6) 8.8 (4.5-17.2) N/A
.02 <.001
ARMa (n ¼ 10) HR (95% CI)
P value
6.7 (1.7-25.9)
<.01
1.0
8.6 (2.2-33.6)
<.01
AAA, Abdominal aortic aneurysm; CI, confidence interval; HR, hazard ratio; IFU, instructions for use; N/A, not applicable. a Because of the small number of patients with adverse events and ARM, a “stepwise” method was used in the multivariable regression model; therefore, the results included in the table are limited to predictors with effects meeting the significance level of .05 to enter and to remain in the model. b Referent group. c Used in the multivariable regression models for all-cause mortality and ARM only.
higher rates of graft migration, device-related complications, and reintervention. Schanzer et al3 demonstrated in their series that nonadherence to IFU guidelines was associated with higher rate of sac enlargement, but this study was hampered by lack of information about device use, clinical complications, and outcomes as well as an unknown incidence of reintervention.
Several recent studies have shown that patients who undergo EVAR outside of IFU guidelines have favorable outcomes. Lee et al8 reported results of 218 patients, one third of whom fell outside of IFU criteria, and showed that these patients had no difference in morbidity, mortality, rate of proximal endoleak, or reintervention after a mean follow-up of 35 months (range,
JOURNAL OF VASCULAR SURGERY May 2015
1158 Walker et al
12-72 months). Similarly, the patients in the current study, 42% of whom fell outside of the IFU guidelines for the specific device they received, exhibited no significant difference in any of the primary or secondary outcomes. All of the criteria that determined IFU nonadherence were related to the morphology of the infrarenal neck, which has been considered to be the primary driver of EVAR durability (ie, no isolated iliac sizing or landing zone issues were observed). Thus, despite no increase in intraoperative adjunctive maneuvers, there was also no observed increase in type I or type III endoleaks due to these suboptimal infrarenal aortic necks. Moreover, the IFU-nonadherent group had significantly more women and larger preoperative aneurysm sac size, two modifiers that are known to increase morbidity.9 This is possibly a reflection of growing operator experience over time and careful patient selection, which are of paramount importance in this population. Given the differing results of these recent studies, it is not surprising that a recent meta-analysis by Antoniou et al10 found that there was insufficient clinical information to provide solid evidence for or against the safety of EVAR in patients with neck anatomy that did not fall within the parameters of the IFU. In their analysis of 1559 patients, neck anatomy did not significantly affect perioperative mortality, but those with non-IFU neck anatomy did experience higher perioperative morbidity. The authors speculated that inclusion of patients unable to tolerate an open operation may have influenced the outcomes in the nonIFU group. Our results may reflect selection bias in that not all clinicians in our health system use M2S for preoperative planning, thereby explaining the modest 28% proportion of the overall cohort available for this analysis. M2S utilization in this cohort may range from routine clinical use to selective case-by-case submission, but we do not have user-specific preferences on M2S or other three-dimensional image reconstruction modalities and thus cannot explore this area for further analysis. Furthermore, despite detailed aneurysm morphologic data, anatomic considerations such as number and size of lumbar vessels, inferior mesenteric artery patency and size, calcification, and thrombus burden were lacking in our data set, and we were therefore unable to analyze the effect of these variables. The decision-making of each clinician to determine the suitability of EVAR, the device and specific size to be used, and whether and when to reintervene represents an additional source of variation, although our findings reflect a “real-world” community-based practice pattern over a broad geographic region. In addition, although graft type was known, the data on graft sizing were incomplete, which limited our ability to specifically and fairly assess device performance. We also recognize it is possible that the differences observed in the proportions of patients with endoleak, adverse events, or reintervention may be due to type II error. A post hoc analysis of sample size showed that patient cohorts of 7573, 2535, and 1212 in each group, respectively, were needed to show a statistical
difference. Finally, it is possible that not all postoperative events pertaining to IFU adherence occurred within our follow-up period. We therefore might be underestimating the true impact of nonadherence to IFU guidelines. Our ongoing stent graft surveillance registry will allow us to capture events that occur with a longer delay. Although the 10% loss to follow-up rate was low, a worst-case analysis (high rate of events in the group lost to follow-up), although unlikely, leaves open the possibility of different findings. CONCLUSIONS In our cohort of EVAR patients with detailed preoperative anatomic information and long-term follow-up, all-cause mortality and ARM were unaffected by IFU adherence, despite a higher proportion of female patients and larger aneurysms in the nonadherent group. In addition, rates of late endoleak and reintervention were similar, suggesting that lack of IFU-based anatomic suitability was not a driver of outcomes. We thank Ann Rhoades, RN, for her assistance in all of the data collection and preparation of this manuscript and the Kaiser Permanente National Implant Registries Group for their assistance in data collection and analysis. AUTHOR CONTRIBUTIONS Conception and design: JW, RC Analysis and interpretation: JW, LT, PG, LC, HH, SO, BH, RC Data collection: JW, LT Writing the article: JW Critical revision of the article: JW, LT, PG, LC, HH, SO, BH, RC Final approval of the article: JW, LT, RC Statistical analysis: JW, LT Obtained funding: RC Overall responsibility: RC REFERENCES 1. Parodi JC, Palmaz JC, Barone HD. Transfemoral intraluminal graft implantation for abdominal aortic aneurysms. Ann Vasc Surg 1991;5: 491-9. 2. Schermerhorn ML, O’Malley AJ, Jhaveri A, Cotterill P, Pomposelli F, Landon BE. Endovascular vs. open repair of abdominal aortic aneurysms in the Medicare population. N Engl J Med 2008;358: 464-74. 3. Schanzer A, Greenberg RK, Hevelone N, Robinson WP, Eslami MH, Goldberg RJ, et al. Predictors of abdominal aortic aneurysm sac enlargement after endovascular repair. Circulation 2011;123:2848-55. 4. Chaikof EL, Blankensteijn JD, Harris PL, White GH, Zarins CK, Bernhard VM, et al. Reporting standards for endovascular aortic aneurysm repair. J Vasc Surg 2002;35:1048-60. 5. Malina M, Resch T, Sonesson B. EVAR and complex anatomy: an update on fenestrated and branched stent grafts. Scand J Surg 2008;97: 195-204. 6. Abbruzzese TA, Kwolek CJ, Brewster DC, Chung TK, Kang J, Conrad MF, et al. Outcomes following endovascular abdominal aortic aneurysm repair (EVAR): an anatomic and device-specific analysis. J Vasc Surg 2008;48:19-28.
JOURNAL OF VASCULAR SURGERY Volume 61, Number 5
7. Fulton JJ, Farber MA, Sanchez LA, Godshall CJ, Marston WA, Mendes R, et al. Effect of challenging neck anatomy on mid-term migration rates in AneuRx endografts. J Vasc Surg 2006;44:932-7; discussion: 937. 8. Lee JT, Ullery BW, Zarins CK, Olcott C, Harris EJ Jr, Dalman RL. EVAR deployment in anatomically challenging necks outside the IFU. Eur J Vasc Endovasc Surg 2013;46:65-73. 9. Chang RW, Goodney P, Tucker LY, Okuhn S, Hua H, Rhoades A, et al. Ten-year results of endovascular abdominal aortic aneurysm
Walker et al 1159
repair from a large multicenter registry. J Vasc Surg 2013;58: 324-32. 10. Antoniou GA, Georgiadis GS, Antoniou SA, Kuhan G, Murray D. A meta-analysis of outcomes of endovascular abdominal aortic aneurysm repair in patients with hostile and friendly neck anatomy. J Vasc Surg 2013;57:527-38.
Submitted Sep 10, 2014; accepted Dec 18, 2014.