- Email: [email protected]
Outcome-based anatomic criteria for defining the hostile aortic neck
CLINICAL RESEARCH STUDIES From the Midwestern Vascular Surgical Society
Outcome-based anatomic criteria for defining the hostile aortic neck William D. Jordan Jr, MD,a Kenneth Ouriel, MD,b Manish Mehta, MD, MPH,c David Varnagy, MD,d William M. Moore Jr, MD,e Frank R. Arko, MD,f James Joye, DO,g and Jean-Paul P. M. de Vries, MD,h Birmingham, Ala; New York and Albany, NY; Orlando, Fla; West Columbia, SC; Charlotte, NC; Mountain View, Calif; and Nieuwegein, The Netherlands Objective: There is abundant evidence linking hostile proximal aortic neck anatomy to poor outcome after endovascular aortic aneurysm repair (EVAR), yet the definition of hostile anatomy varies from study to study. This current analysis was undertaken to identify anatomic criteria that are most predictive of success or failure at the aortic neck after EVAR. Methods: The study group comprised 221 patients in the Aneurysm Treatment using the Heli-FX Aortic Securement System Global Registry (ANCHOR) clinical trial, a population enriched with patients with challenging aortic neck anatomy and failure of sealing. Imaging protocols were not protocol specified but were performed according to the institution’s standard of care. Core laboratory analysis assessed the three-dimensional centerline-reformatted computed tomography scans. Failure at the aortic neck was defined by type Ia endoleak occurring at the time of the initial endograft implantation or during follow-up. Receiver operating characteristic curve analysis was used to assess the value of each anatomic measure in the classification of aortic neck success and failure and to identify optimal thresholds of discrimination. Binary logistic regression was performed after excluding highly intercorrelated variables, creating a final model with significant predictors of outcome after EVAR. Results: Among the 221 patients, 121 (54.8%) remained free of type Ia endoleak and 100 (45.2%) did not. Type Ia endoleaks presented immediately after endograft deployment in 58 (58.0%) or during follow-up in 42 (42.0%). Receiver operating characteristic curve analysis identified 12 variables where the classification of patients with type Ia endoleak was significantly more accurate than chance alone. Increased aortic neck diameter at the lowest renal artery (P [ .013) and at 5 mm (P [ .008), 10 mm (P [ .008), and 15 mm (P [ .010) distally; aneurysm sac diameter (P [ .001), common iliac artery diameters (right, P [ .012; left, P [ .032), and a conical (P [ .049) neck configuration were predictive of endoleak. By contrast, increased aortic neck length (P [ .050), a funnel-shaped aortic neck (P [ .036), and neck mural thrombus content, as measured by average thickness (P [ .044) or degrees of circumferential coverage (P [ .029), were protective against endoleak. Binary logistic regression identified three variables independently predictive of type Ia endoleak. Neck diameter at the lowest renal artery (P [ .002, cutpoint 26 mm) and neck length (P [ .017, cutpoint 17 mm) were associated with endoleak, whereas some mural neck thrombus content was protective (P [ .001, cutpoint 11 of circumferential coverage). Conclusions: A limited number of independent anatomic variables are predictive of type Ia endoleak after EVAR, including aortic neck diameter and aortic neck length, whereas mural thrombus in the neck is protective. This study suggests that anatomic measures with identifiable threshold cutpoints should be considered when defining the hostile aortic neck and assessing the risk of complications after EVAR. (J Vasc Surg 2015;61:1383-90.)
Endovascular aneurysm repair (EVAR) surpassed open surgery as the most frequently performed treatment option for patients with abdominal aortic aneurysms (AAAs) within
a decade after the first endografts were marketed.1 EVAR, as a less invasive technique, has many advantages over open surgical repair, principally related to early morbidity and
From the Division of Vascular Surgery and Endovascular Therapy, University of Alabama-Birmingham, Birminghama; Syntactx, New Yorkb; Vascular Surgery, Albany Vascular Group, The Institute for Vascular Health and Disease, Albanyc; Vascular Surgery, Florida Hospital, Orlandod; the Southern Surgical Group, Lexington Medical Center, West Columbiae; Vascular Surgery, Sanger Heart & Vascular Institute, Carolinas Health Care System, Charlottef; Cardiovascular Disease, Heart & Vascular Institute, El Camino Hospital, Mountain Viewg; and the Department of Vascular Surgery, St. Antonius Hospital, Nieuwegein.h Author conflict of interest: K.O. has equity ownership in Syntactx, a company that receives fees for contract research activities from Aptus, the study sponsor. Aptus, the sponsor of the study, was not involved in the data collection, data analysis, data interpretation, manuscript writing, or the decision to submit the manuscript. Aptus employees were, however, provided with a draft copy of the manuscript for comment.
Presented at the Thirty-eighth Annual Meeting of the Midwestern Vascular Surgical Society, Coralville, Iowa, September 4-6, 2014. Additional material for this article may be found online at www.jvascsurg.org. Reprint requests: William D. Jordan, MD, Division of Vascular Surgery and Endovascular Therapy, University of Alabama at Birmingham, 1808 7th Ave S, BDB 503, Birmingham, AL 35294 (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. All rights reserved. http://dx.doi.org/10.1016/j.jvs.2014.12.063
1383
1384 Jordan et al
mortality.2,3 The procedure, however, lacks the flexibility and durability of open surgical graft insertion, primarily a result of the interdependence of endograft sealing and the anatomic aspects of the proximal aortic neck.4,5 Outcome after EVAR is inferior when aortic neck anatomic irregularities are encountered, including a short neck length, large neck diameter, extreme neck angulation, and abundant mural thrombus or calcium within the neck.4,6,7 Although an open surgical anastomosis can be more difficult to perform in the presence of such anatomic issues, an open procedure depends less on neck anatomy than EVAR. The term “hostile neck” was first used by Dillavou7 in 2003 to characterize EVAR outcome in patients with unfavorable aortic neck anatomy. Currently, the term is often used when the aortic neck anatomy falls outside the eligibility criteria for a manufacturer’s regulatory clinical trial. Recognition of hostile neck characteristics creates some concern for endograft performance at the aortic neck. Endograft trials are limited to patients with anatomy well suited for a particular endograft, and regulatory approval is granted before the device is used in the general population. Once approved for marketing, however, a broad range of anatomies are encountered, anatomies that were not evaluated in regulatory trials and thus have undefined clinical outcomes. Despite widespread reliance on anatomic criteria for clinical decision making, there has been little work on characterizing the relative importance of different anatomic measures and on identifying optimal thresholds for each. In part, the paucity of data is a result of the infrequency of aortic neck complications. The reported frequency of type Ia endoleak is <3% in most series, precluding a rigorous multivariate analysis of the factors related to the event.8,9 By contrast, the Aneurysm Treatment using the Heli-FX Aortic Securement System Global Registry (ANCHOR) data set provides a study population enriched with challenging aortic anatomy and aortic neck complications.10 As such, the data set enables robust statistical comparisons of each anatomic measure across multiple endografts and operators. In tandem with receiver operating characteristic (ROC) analysis, optimal thresholds for classifying neck complications can be formulated from anatomic criteria. In this report we describe an analysis of hostile neck criteria and their comparative effect on outcome after EVAR. METHODS The ANCHOR study is a prospective, nonrandomized, multicenter, multinational study of the real-world use of the Heli-FX EndoAnchor System (Aptus Endosystems, Sunnyvale, Calif) in patients undergoing EVAR or who have undergone EVAR for AAA in the past. The ANCHOR study is registered on ClinicalTrials.gov (NCT01534819). Institutional Review Board or Ethics Committee approval was obtained at each site. Each patient provided written informed consent. Details of the study methodology and the device have been previously described.11 The investigators are listed in the Appendix (online only). Briefly, the study eligibility criteria included patients with infrarenal AAA who had adequate iliofemoral access
JOURNAL OF VASCULAR SURGERY June 2015
to accommodate a 16F sheath and a life expectancy of at least 1 year. Commercially available endografts that underwent successful testing for EndoAnchor compatibility included the Zenith (Cook, Bloomington, Ind), the Excluder (W. L. Gore and Associates, Flagstaff, Ariz), and the AneuRx, Talent, and Endurant devices (Medtronic Vascular, Santa Rosa, Calif). The study cohort comprised 221 patients enrolled in ANCHOR at 26 United States and 9 European centers between April 2012 and January 2014. The study population included 66.6% of 332 patients enrolled in the full study over the same timeframe, selected by availability of adequate baseline (before the EndoAnchor implantation procedure) and postoperative computed tomography (CT) imaging for core laboratory analysis. The 221 patients were subdivided into 121 EVAR patients without type Ia endoleak (54.3%) and 100 with aortic neck failure (54.8%) evidenced by type Ia endoleak alone in 86 patients (38.4%) or endoleak in conjunction with endograft migration in 14 (6.3%). Patients with immediate type Ia endoleaks evident on intraoperative angiography performed at the time of the initial EVAR (58 patients [26.2%]) were included in the failure group when EndoAnchors (with or without extension cuffs or bare metal stents) were implanted to address the endoleak, irrespective of whether the additional interventions remediated the endoleak (52 patients [23.5%]) or did not (six patients [2.7%]). The remaining 42 patients (19.0%) were treated with EndoAnchors for type Ia endoleaks a median of 35 months (range, 0.2-168 months) after the initial EVAR procedure. Once enrolled in ANCHOR, patients were monitored clinically for a median of 19 months (range, 0-30 months), with CT imaging studies performed through a median of 7 months (range, 0-23 months). Imaging studies and definitions. Imaging protocols were not protocol-specified but were performed according to the institution’s standard of care. Independent core laboratory analyses (Syntactx, New York, NY) were performed on noncontrast and contrast CT imaging studies. Centerline reformatting and segmentation of CT data sets was performed using iNtuition imaging software (TeraRecon, Foster City, Calif). Imaging end points were measured and reported using the methodology from the Society for Vascular Surgery reporting standards guideline documents where end point definitions were specificed.12-15 Aortic diameters were measured as the average diameter of the centerline reformatted aortic adventitia-toadventitia contour. Circularity was not assumed; rather, any deviation from perfect circularity was taken into account with an electronically traced aortic contour on a plane orthogonal to the aortic centerline. Aortic neck calcium and thrombus content was measured and expressed in degrees of circumference where thickness was $2 mm, as evaluated on the CT image 5 mm distal to the lowest main renal artery. The aortic neck length was calculated using two methods. The first method corresponds to what has been
JOURNAL OF VASCULAR SURGERY Volume 61, Number 6
defined as the “anatomic neck” and represented the centerline distance between the lowest main renal artery and the level where the diameter of the infrarenal aortic neck reached 10% more than the diameter at the lowest main renal ostium.16 The second method, dubbed the “visual neck” length, defined the distal margin as the visually ascertained inflection point where the neck transitioned into the aneurysm sac. Not being based upon strict geometric calculations, the visual neck was more subjectively determined than was the anatomic neck. Further, it can be longer than the anatomic neck in conically shaped necks with a gradual, tapered configuration along its length. Suprarenal and infrarenal aortic neck angulation were measured using three points (knots) along the centerline spline created from reformatted CT data sets: a proximal knot, an apex, and a distal knot. Suprarenal angulation was defined with a proximal knot 20 mm proximal to the lowest main renal artery, an apex at the ostium of the lowest main renal artery, and a distal knot at the distalmost aspect of the anatomic neck. Infrarenal angulation was measured using two methods that differed with respect to the distal knot. Both methods used a proximal knot at the level of the lowest main renal artery and an apex at the level of the distal-most aspect of the anatomic neck. However, the first method used a distal knot placed 40 mm distal to the apex, whereas the second used a knot located at the center-most point on the aortic bifurcation. In addition, noting that angulation defined by two vectors and an apex is a linear simplification of the true anatomic curvature of a neck, an “aortic tortuosity index” was expressed as the comparison between the curvilinear and straight line distance between the proximal (lowest renal artery) and distal (40 mm) knots. The aortic tortuosity index was calculated as the curvilinear distance divided by the straight-line distance. The degree to which the aortic neck was conical in configuration was assessed with neck “conicity,” defined as the percentage increase in aortic neck diameter between the lowest main renal artery and a specified length distally. Conicity was defined through four levels: 5, 10, and 15 mm beyond to the lowest main renal artery and at the distal margin of the visual neck. A positive value for conicity indicated reverse tapering of the aortic neck where the neck diameter increased with distal progression. A negative value implied a funnel-shaped neck that was larger proximally than distally. Data analysis. Statistical analyses were performed with SPSS 22 software (IBM Corp, Armonk, NY). Values are expressed as mean 6 standard deviation unless otherwise specified. P values were considered significant when the two-tailed a was <.05. ROC curve analysis was used to evaluate the diagnostic value of anatomic measures that have been considered to be related to complications at the aortic neck and to determine optimal thresholds or cutpoints for classifying aortic neck failure (type Ia endoleak) and success. The evaluated anatomic measures included aortic neck diameter, length, thrombus/calcium content,
Jordan et al 1385
degree of neck conicity, infrarenal angulation, suprarenal angulation, and maximum aneurysm diameter. ROC curves were plotted with sensitivity on the y axis and (1-specificity) on the x axis. Areas under the curve (AUCs) were calculated to assess discriminatory accuracy over the spectrum of values for a particular anatomic measure. AUCs are expressed as the estimate and its 95% confidence interval. The Wilcoxon-Mann-Whitney estimator was used to assess whether a variable was significantly better than chance alone in the classification of success and failure.17 Sensitivities and specificities for each anatomic measure to discriminate between aortic neck success and failure were determined at sequential cutpoints ranging between the minimum and maximum observed values for each measure. The optimum cutpoint was defined by a maximum balance between sensitivity and specificity. Binary logistic regression was performed with inclusion of anatomic covariates where the ROC AUC probability was <.10. Highly intercorrelated, redundant explanatory variables with r values >0.70 were eliminated to reduce the effects of multicollinearity.18 A backward stepwise process was used, sequentially removing variables with the least significance until the significance of all variables was <.05. RESULTS The 165 male (74.7%) and 56 female (25.3%) patients were a mean age of 74.1 6 8.2 years (range, 56-97 years). Endografts implanted included the Medtronic Endurant in 96 patients (43.4%), the Gore Excluder in 77 (34.8%), the Cook Zenith in 28 (12.7%), and other devices in 20 (9.0%). The baseline anatomic characteristics of the study population are reported in Table I. ROC curve analysis. Aortic neck diameter was a significant determinant of type Ia endoleak when measured at the lowest renal artery and at 5-mm increments distally to 15 mm (Table II and Fig 1). The diameter at the distal margin of the visual neck was not of significant discriminatory value nor was diameter at the suprarenal level. A diameter of 26 mm at the lowest renal artery was the optimal cutpoint for maximizing the trade-off between sensitivity and specificity. Corresponding cutpoints for the more distal levels within the neck increased 1 mm for every 5 mm of distal neck length. Of the two methods of measuring aortic neck length, anatomic neck length was not statistically significant in the classification of neck success or failure (P ¼ .069), and visual neck length was of marginal statistical significance (P ¼ .050). The optimal cutpoints for the two measures were 11 mm and 17 mm, respectively (Fig 2). No measure of neck angulation attained statistical significance in the discrimination of type Ia endoleaks. The aortic tortuosity index was the best but still did not attain significance (P ¼ .068, cutpoint 1.04). However, extreme angulation was relatively rare in the study cohort, with 23 patients (10.4%) manifesting infrarenal neck angulation >60 . Aortic neck conicity was a useful determinant by ROC curve analysis. A reverse-taper neck configuration
JOURNAL OF VASCULAR SURGERY June 2015
1386 Jordan et al
Table I. Baseline anatomic characteristics in 100 patients with and 121 patients without type Ia endoleak Anatomic measure
Type Ia endoleak
Mean
SD
P value
Suprarenal aortic diameter, mm
No Yes No Yes
27.3 28.1 25.9 27.8
3.2 4.9 4.2 6.5
.113
No Yes No Yes No Yes No Yes No Yes No Yes
26.6 29.0 27.8 30.3 28.9 31.0 29.2 30.2 18.0 14.8 22.8 18.2
4.7 7.7 4.9 8.2 6.1 8.8 4.9 6.3 13.8 11.2 16.8 14.8
No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes
2.9 4.0 7.0 8.4 11.7 8.8 13.5 9.4 1.06 1.05 55.0 59.8 15.9 15.8 25.9 23.2 35.4 35.5 1.0 0.6 65.8 29.9 1.0 0.9 20.7 17.7
12.0 6.8 9.6 16.6 17.5 21.9 14.7 10.6 0.06 0.06 10.5 14.3 12.5 12.2 16.3 14.9 19.1 15.7 1.4 1.4 93.4 59.3 1.2 1.2 31.0 36.2
Aortic diameter at lowest renal, mm Neck diameter 5 mm distal to lowest renal, mm 10 mm distal to lowest renal, mm 15 mm distal to lowest renal, mm At distal end of visual neck, mm Anatomic neck length (10% threshold, mm) Visual neck length, mm Conicity to 5 mm (% increase over distance) 10 mm (% increase over distance) to 15 mm (% increase over distance) Conicity through visual neck (% increase) Neck tortuosity index (curvilinear/straight length) Maximum aneurysm sac diameter, mm Suprarenal angulation, Infrarenal angulation, Infrarenal angulation to bifurcation, Neck thrombus average thickness, mm
Neck thrombus circumference,
Neck calcium average thickness, mm
Neck calcium circumference,
.009 .006 .006 .054 .192 .065 .034 .422 .446 .316 .017 .176 .006 .972 .205 .961 .038 .001 .383 .509
SD, Standard deviation.
(distal diameter greater than immediate infrarenal diameter) through 5 mm increased the risk of type Ia endoleak (P ¼ .049, optimal cutpoint 2% grade). By contrast, protection against endoleak was conferred by a funnelshaped configuration of the visual neck, where the neck diameter at the visually determined start of the aneurysm was smaller than the diameter at the immediate infrarenal level (P ¼ .036, optimal cutpoint 9% grade). Neither aortic neck calcium thickness nor degrees of circumferential calcium content attained statistical significance as discriminators of type Ia endoleak (Fig 3). By contrast, a significant positive discriminatory value was observed for mural neck thrombus, expressed as average thickness (P ¼ .044) or as degrees of neck circumference covered with $2 mm of mural thrombus (P ¼ .029). The optimal thresholds for thrombus thickness and
circumference coverage were 0 mm and 11 , respectively, suggesting that virtually any mural neck thrombus positively affected outcome. However, noting only 14 patients (6.3%) had >180 of mural thrombus and only 12 patients (5.4%) had thrombus >3 mm in average thickness, the current analysis cannot exclude a negative effect of extreme levels of thrombus. Maximum aneurysm sac diameter was a good discriminator of type Ia endoleak (P ¼ .001). Right and left common iliac diameters were similarly associated with endoleak (P ¼ .012 and P ¼ .032, respectively). Optimal thresholds for discrimination were 55 mm for sac diameter, 19 mm for the right common iliac artery, and 18 mm for the left common iliac artery. Multivariate analysis. After excluding variables with significant intercorrelation or ROC AUC significance
JOURNAL OF VASCULAR SURGERY Volume 61, Number 6
Jordan et al 1387
Table II. Receiver operating characteristic (ROC) analyses of anatomic measures for type Ia endoleak after aneurysm repaira Anatomic measure Aortic neck diameter At lowest renal artery 5 mm distal to renal artery 10 mm distal to renal artery 15 mm distal to renal artery At end of visual neck Suprarenal aortic diameter Aortic neck length Anatomic neck Visual neck Neck angulation Suprarenal Infrarenal Infrarenal to bifurcation Aortic tortuosity index Aortic neck thrombus/calcium Thrombus thickness Thrombus circumference Calcium thickness Calcium circumference Aortic neck conicity Through 5 mm neck length Through 10 mm neck length Through 15 mm neck length Through length of visual neckc Vessel diameter Maximum sac diameter Right common iliac artery Left common iliac artery
AUC (95% CI) 0.597 0.604 0.604 0.601 0.532 0.459
(0.522-0.672) (0.528-0.679) (0.529-0.680) (0.526-0.677) (0.455-0.608) (0.382-0.536)
0.571 (0.495-0.647) 0.576 (0.501-0.652)
Optimal cutpoint <26 <27 <28 <29
mm mm mm mm
P value .013b .008b .008b .010b .415 .292
> 11 mm >17 mm
.069 .050b
<1.04
.715 .259 .527 .068
0.486 0.544 0.475 0.571
(0.409-0.562) (0.468-0.620) (0.399-0.551) (0.496-0.647)
0.579 0.585 0.530 0.548
(0.504-0.654) (0.510-0.660) (0.454-0.607) (0.472-0.624)
>0 mm >11
.044b .029b .439 .220
0.577 0.545 0.524 0.582
(0.500-0.653) (0.468-0.623) (0.447-0.601) (0.507-0.657)
<2% >9%
.049b .246 .541 .036b
<55 mm <19 mm <18 mm
.001b .012b .032b
0.621 (0.546-0.696) 0.600 (0.526-0.676) 0.586 (0.508-0.663)
AUC, Area under the curve; CI, confidence interval. a Optimal cutpoints balanced sensitivity and specificity; calculated for variables with significance <.10. Cutpoints were defined as the optimum thresholds above which type Ia endoleaks were more likely. b Significant at P < .05 a level. c Funnel shape (proximal larger than distal) through visual neck was protective against failure. Endoleak was less likely when the diameter at the distal end of the visual neck was less than the immediate infrarenal aortic diameter.
of <.10, 14 anatomic covariates were included in the binary logistic regression analysis. The backward stepwise model identified three variables that were significantly predictors of type Ia endoleak (Table III). These were a larger aortic neck diameter at the level of the lowest renal artery (P ¼ .002), a shorter anatomic neck length (P ¼ .017), and less mural thrombus (P ¼ .001). The final model with these covariates was of significant utility in predicting type Ia endoleaks (P < .001). DISCUSSION The link between aortic neck anatomy and the development of complications, such as type Ia endoleak and endograft migration, has been evident since the inception of EVAR in the early 1990s.19-21 The literature is replete with investigations linking the development of adverse clinical events and imaging findings with challenging neck anatomy.4,5,22 Further, the instructions for use of all currently marketed endografts contain anatomic indications that generally parallel the eligibility criteria of the investigational device exemption approval trials submitted for regulatory approval.
Despite the intuitive value and empiric association between various anatomic criteria and outcome after EVAR, there are sparse data in the literature to objectively compare the importance of the indices. As well, prior reports have not objectively evaluated optimal cutpoints to identify whether the anatomy of a particular patient places that patient at risk for aortic neck failure. The absence of such an analysis is understandable: few prior series have an adequate number of EVAR failures to adequately investigate predictive factors. The ANCHOR series is unique in this regard. Patients are specifically selected and enrolled based the presence of challenging neck anatomy or after the development of a type Ia endoleak.11 For this reason, the ANCHOR data set was used in conjunction with ROC curve analysis and logistic regression to identify those baseline anatomic measures associated with aortic neck complications during or after EVAR. The current study confirmed the importance of aortic neck length and diameter as the important determinants of type Ia endoleak after EVAR. These factors have been identified in prior reports as standard anatomic criteria that should be taken into account in the classification of
JOURNAL OF VASCULAR SURGERY June 2015
1388 Jordan et al
Fig 1. Receiver operating characteristic (ROC) curves for aortic neck diameter at the lowest renal artery and at 5-mm increments distally to 15 mm.
Fig 3. Receiver operating characteristic (ROC) curves for thrombus and calcium content within the proximal aortic neck, assessed 5 mm distal to the lowest renal artery. Each end point is measured as the average thickness over the area of involvement as well as the degrees of circumference covered with >1 mm of mural thrombus or calcium. Mural thrombus was protective against type Ia endoleak. The effect of calcium was opposite but did not attain statistical significance as a predictive variable.
Table III. Binary logistic regression: Final model Variable Aortic neck diameter at lowest renal artery (mm) Anatomic neck length (10% threshold, mm) Neck circumference with mural thrombus (mm) Intercept
Coefficient
Standard Odds P error ratioa value
0.102
0.033
1.11 .002
0.028
0.012
0.97 .017
0.007
0.002
0.99 .001
2.133
0.871
.014
a
Odds ratios indicate the increase (or reduction) in the risk of type Ia endoleak for every unit increase in the covariate. For instance, the odds ratio of 1.11 for aortic neck diameter indicates that the risk of type Ia endoleak increases 11% for each 1-mm increase in neck diameter. Similarly, the odds ratio of .99 for mural neck thrombus indicates the risk of type Ia endoleak decreases by 1% for each degree of circumferential mural thrombus.
Fig 2. Receiver operating characteristic (ROC) curves for anatomic neck length and visual neck length. Anatomic neck length is measured to the level at which the vessel diameter is 10% greater than its diameter at the lowest renal artery, whereas visual neck length is measured to the visually ascertained aortic wall inflection point that identifies the start of the aneurysm.
hostile vs friendly aortic neck anatomy.4-6 The current analysis did not identify other accepted criteriadmost notably aortic neck angulation and neck calcium contentdas independent predictors of type Ia endoleak. The failure to identify neck angulation as an important determinant may partly relate to a low prevalence of patients with extreme degrees of
angulation in the series. Similar limitations may exist for mural thrombus and calcium content; extreme amounts of each were uncommon in this series. Of importance, the multivariable analysis suggested that the most important determinants of aortic neck complications were neck diameter, neck length, and thrombus content. Other variables fell out as important predictors in the final binary logistic regression, supplanted by presumably more important variables in the reduced model. Aortic neck thrombus, intuitively, should be a risk factor for aortic neck failure.23 In fact, this is not the case in
JOURNAL OF VASCULAR SURGERY Volume 61, Number 6
prior reports. The study of Bastos Gonçalves et al24 documented a higher rate of 5-year clinical success in patients with thrombus >2 mm in thickness covering >25% of neck circumference, 74% vs 62%, but this difference did not attain statistical significance. Our data corroborate these observations; some amount of neck thrombus appears to be protective against type Ia endoleak. The current data cannot exclude a reversal of this relationship when the amount of mural thrombus is extreme. Because the clinical follow-up is relatively short, additional anatomic criteria other than those identified may also be predictive over longer-term follow-up. The ROC curve analysis used in the current analysis provides some potential advantages over other statistical methods. The technique is independent of arbitrarily chosen cutpoints or the prevalence of the event, in this case type Ia endoleak, in the population under study. The current analytic methodology, however, has some limitations: First, preoperative CT images from the time of the initial EVAR were unavailable in patients who presented with late type Ia endoleaks. Postimplant imaging studies were used in these cases, potentially underestimating preoperative baseline neck angulation or neck thrombus content and overestimating baseline neck diameter in the endoleak subgroup, to the extent and direction that these measures are known to change after endograft implantation.24-26 This limitation may underlie the failure to identify certain variables as risk factors for type Ia endoleak. A second limitation relates to the use of a data set comprising patients selected for EndoAnchor use. Patients with findings that preclude use of the device would not have been included. For instance, patients with extreme amounts of neck thrombus or calcium or those with neck dilatation where a large gap developed between the endograft and the aortic wall would not have been included. Third, the importance of neck and aneurysm sac diameter may be overestimated in the model because each may have increased to some extent compared with the preoperative baseline values. Fourth, the study design dichotomized the data set into just two groupsdsuccess and failuredwhere the failures include patients with type Ia endoleaks detected at the time of endograft implantation as well as those identified on postoperative imaging studies. Lastly, the relationship between anatomic factors and type Ia endoleak may be endograft dependent, and the current study pooled the results of all devices into a single analysis. CONCLUSIONS The current analysis suggests that just a few anatomic measures are associated with type Ia endoleak after EVAR, and ROC analysis offers a useful method to assign logical predictors and their cutpoints for defining the hostile aortic neck. Confirmation of these findings in a larger data set with complete assessment of all preoperative and longer-term postoperative imaging studies will be necessary to validate these findings. Until then, the observations of
Jordan et al 1389
this study provide the groundwork for a developing a uniform definition of the hostile neck when evaluating outcome after EVAR. The findings may provide some basis for the application of intraoperative adjunctive techniques to EVAR, even in the absence of an angiographically identified endoleak, and adjustments to postoperative surveillance imaging regimens when challenging aortic neck anatomy is encountered. AUTHOR CONTRIBUTIONS Conception and design: JD, KO, WJ Analysis and interpretation: JD, KO, MM, DV, WM, FA, JJ, WJ Data collection: KO Writing the article: JD, KO, WJ Critical revision of the article: JD, KO, MM, DV, WM, FA, JJ, WJ Final approval of the article: JD, KO, MM, DV, WM, FA, JJ, WJ Statistical analysis: KO Obtained funding: JD, KO, WJ Overall responsibility: KO REFERENCES 1. Schwarze ML, Shen Y, Hemmerich J, Dale W. Age-related trends in utilization and outcome of open and endovascular repair for abdominal aortic aneurysm in the United States, 2001-2006. J Vasc Surg 2009;50: 722-9. 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. Anderson PL, Arons RR, Moskowitz AJ, Gelijns A, Magnell C, Faries PL, et al. A statewide experience with endovascular abdominal aortic aneurysm repair: rapid diffusion with excellent early results. J Vasc Surg 2004;39:10-9. 4. 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. 5. Stather PW, Wild JB, Sayers RD, Bown MJ, Choke E. Endovascular aortic aneurysm repair in patients with hostile neck anatomy. J Endovasc Ther 2013;20:623-37. 6. Aburahma AF, Campbell JE, Mousa AY, Hass SM, Stone PA, Jain A, et al. Clinical outcomes for hostile versus favorable aortic neck anatomy in endovascular aortic aneurysm repair using modular devices. J Vasc Surg 2011;54:13-21. 7. Dillavou ED, Muluk SC, Rhee RY, Tzeng E, Woody JD, Gupta N, et al. Does hostile neck anatomy preclude successful endovascular aortic aneurysm repair? J Vasc Surg 2003;38:657-63. 8. Stokmans RA, Teijink JA, Forbes TL, Bockler D, Peeters PJ, Riambau V, et al. Early results from the ENGAGE registry: real-world performance of the Endurant Stent Graft for endovascular AAA repair in 1262 patients. Eur J Vasc Endovasc Surg 2012;44:369-75. 9. Cao P, De RP, Parlani G, Verzini F. Durability of abdominal aortic endograft with the Talent Unidoc stent graft in common practice: core lab reanalysis from the TAURIS multicenter study. J Vasc Surg 2009;49:859-65. 10. De Vries JP, Van De Pavoordt HD, Jordan WD Jr. Rationale of EndoAnchors in abdominal aortic aneurysms with short or angulated necks. J Cardiovasc Surg (Torino) 2014;55:103-7. 11. Jordan WD Jr, Mehta M, Varnagy D, Moore WM Jr, Arko FR, Joye J, et al. Results of the ANCHOR prospective, multicenter registry of EndoAnchors for type Ia endoleaks and endograft migration in patients with challenging anatomy. J Vasc Surg 2014;60:885-92.e2.
1390 Jordan et al
12. 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. 13. Chaikof EL, Fillinger MF, Matsumura JS, Rutherford RB, White GH, Blankensteijn JD, et al. Identifying and grading factors that modify the outcome of endovascular aortic aneurysm repair. J Vasc Surg 2002;35: 1061-6. 14. Ouriel K, Fowl RJ, Davies MG, Forbes TL, Gambhir RP, Morales JP, et al. Reporting standards for adverse events after medical device use in the peripheral vascular system. J Vasc Surg 2013;58:776-86. 15. Ouriel K, Fowl RJ, Davies MG, Forbes TL, Gambhir RP, Ricci MA, et al. Disease-specific guidelines for reporting adverse events for peripheral vascular medical devices. J Vasc Surg 2014;60:212-25. 16. Welborn MB 3rd, McDaniel HB, Johnson RC, Kennedy RE, Knott A, Mundinger GH, et al. Clinical outcome of an extended proximal seal zone with the AFX endovascular aortic aneurysm system. J Vasc Surg 2014;60:876-83; discussion: 883-4. 17. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988;44:837-45. 18. Tabachnick BG, Fidell LS. Using multivariate statistics. 6th edition. Boston: Pearson; 2013. 19. Fox AD, Whiteley MS, Murphy P, Budd JS, Horrocks M. Comparison of magnetic resonance imaging measurements of abdominal aortic aneurysms with measurements obtained by other imaging techniques and intraoperative measurements: possible implications for endovascular grafting. J Vasc Surg 1996;24:632-8. 20. Moore WS, Vescera CL. Repair of abdominal aortic aneurysm by transfemoral endovascular graft placement. Ann Surg 1994;220:331-9; discussion: 339-41.
JOURNAL OF VASCULAR SURGERY June 2015
21. Parodi JC. Endovascular repair of abdominal aortic aneurysms and other arterial lesions. J Vasc Surg 1995;21:549-55. 22. Rockman CB, Lamparello PJ, Adelman MA, Jacobowitz GR, Therff S, Gagne PJ, et al. Aneurysm morphology as a predictor of endoleak following endovascular aortic aneurysm repair: do smaller aneurysm have better outcomes? Ann Vasc Surg 2002;16:644-51. 23. Walker TG, Kalva SP, Yeddula K, Wicky S, Kundu S, Drescher P, et al. Clinical practice guidelines for endovascular abdominal aortic aneurysm repair: written by the Standards of Practice Committee for the Society of Interventional Radiology and endorsed by the Cardiovascular and Interventional Radiological Society of Europe and the Canadian Interventional Radiology Association. J Vasc Interv Radiol 2010;21: 1632-55. 24. Bastos Gonçalves GF, Verhagen HJ, Chinsakchai K, van Keulen JW, Voute MT, Zandvoort HJ, et al. The influence of neck thrombus on clinical outcome and aneurysm morphology after endovascular aneurysm repair. J Vasc Surg 2012;56:36-44. 25. Cao P, Verzini F, Parlani G, Rango PD, Parente B, Giordano G, et al. Predictive factors and clinical consequences of proximal aortic neck dilatation in 230 patients undergoing abdominal aorta aneurysm repair with self-expandable stent-grafts. J Vasc Surg 2003;37:1200-5. 26. Ishibashi H, Ishiguchi T, Ohta T, Sugimoto I, Yamada T, Tadakoshi M, et al. Remodeling of proximal neck angulation after endovascular aneurysm repair. J Vasc Surg 2012;56:1201-5.
Submitted Nov 24, 2014; accepted Dec 21, 2014.
Additional material for this article may be found online at www.jvascsurg.org.
JOURNAL OF VASCULAR SURGERY Volume 61, Number 6
APPENDIX (online only). Collaborators and institutions involved in the Aneurysm Treatment using the Heli-FX Aortic Securement System Global Registry (ANCHOR) William D. Jordan Jr., Vascular Surgery and Endovascular Therapy, University of Alabama, Birmingham, Ala, United States Jean Paul de Vries, Department of Vascular Surgery, St. Antonius Hospital, Nieuwegein, Netherlands James Joye, Cardiovascular Disease, Heart & Vascular Institute, El Camino Hospital, Mountain View, Calif, United States H.H. Eckstein, Department of Vascular Surgery, Clinic for Vascularand Endovascular Surgery, Klinikum Rechts der Isar der Technischen, Technical University Munich, Munich, Germany Joost van Herwaarden, Department of Vascular Surgery, University Medical Center Utrecht, Utrecht, Netherlands Frank R. Arko, Vascular Surgery, Sanger Heart & Vascular Institute, Carolinas Health Care System, Charlotte, NC, United States Paul Bove, Vascular and Endovascular Surgery, Peripheral Vascular Diagnostic Center, William Beaumont, Royal Oak, Mich, United States William Bohannon, Vascular Surgery, Scott & White Medical Center, Temple, Tex, United States Bram Fioole, Department of Vascular Surgery, Maasstad Medical Hospital, Rotterdam, The Netherlands Carlo Setacci, Vascularand Endovascular Surgery UniteUniversity of Siena, Siena, Italy Timothy Resch, Department of Vascular Diseases, Malmö University Hospital, Malmö, Sweden Vicente Riambau, Angiology and Vascular Surgery, Thorax Institute Hospital Clinic, Barcelona, Spain Dierk Scheinert and Andrej Schmidt, Angiology and Vascular Surgery, Park Hospital, Leipzig, Germany Daniel Clair, Vascular Surgery, Cleveland Clinic, Cleveland, Ohio, United States Mohammed Moursi, Vascular Surgery, Central Arkansas Veterans Health System, University of Arkansas for Medical Sciences, Little Rock, Ark, United States Mark Farber, Vascular Surgery, School of Medicine, University of North Carolina, Chapel Hill, NC, United States Joerg Tessarek, Department of Vascular Surgery, Vascular Center, St. Bonifatius Hospital, Lingen, Germany Giovanni Torsello, Vascular Surgery, St. Franziskus, Hospital GmbH, Munster, Germany Mark Fillinger, Vascular Surgery, Geisel School of Medicine, Dartmouth, Hitchcock Medical Center, Lebanon, NH, United States Marc Glickman, Vascular Surgery, Sentara Heart Hospital, Norfolk, Va, United States John Henretta, Vascular Surgery, Carolina Vascular, Mission Hospital, Asheville, NC, United States Kim Hodgson, Vascular Surgery Division, School of Medicine, Southern Illinois University, Springfield, Ill, United States
Jordan et al 1390.e1
Jeffrey Jim, Section of Vascular Surgery, Washington University School of Medicine, St. Louis, Mo, United States Barry Katzen, Vascular and Interventional Radiology, Baptist Cardiac & Vascular Institute, Baptist Hospital, Miami, Fla, United States Evan Lipsitz, Cardiovascular and Thoracic Surgery, Montefiore Medical Center, Bronx, NY, United States Mitchell Cox, Vascular Surgery, Duke University Medical Center, Durham, SC, United States Thomas Naslund, Division of Vascular Surgery, Vanderbilt University Medical Center, Nashville, Tenn, United States Venkatesh Ramaiah, Vascular Surgery, Arizona Heart Institute, Abrazo Health Care, Phoenix, Ariz, United States Marc Schermerhorn, Vascular and Endovascular Surgery, Beth Israel Deaconess Medical Center, Boston, Mass, United States Peter Schneider, Vascular and Endovascular Surgery, Hawaii Permanente Medical Group, Honolulu, Hawaii, United States Benjamin Ware Starnes, Vascular Surgery Division, Harborview Medical Center, University of Washington, Seattle, Wash, United States Carlos Donayre, Vascular and Endovascular Surgery, Harbor-UCLA Medical Center, Torrance, Calif, United States Manish Mehta, Vascular Surgery, Albany Vascular Group, The Institute for Vascular Health and Disease, Albany, NY, United States Burkhart Zipfel, Department of Cardiothoracic and Vascular Surgery, Deutsches Herzzentrum, Berlin, Germany Nitin Malhotra, Vascular Surgery, Michigan Vascular Center, Flint, Mich, United States David Varnagy, Vascular Surgery, Florida Hospital, Orlando, Fla, United States William Moore Jr, Vascular Disease, Lexington Medical Center, West Columbia, SC, United States Nick Cheshire and Colin Bicknell, Vascular Surgery, Imperial College, London, United Kingdom Martin Back, Vascular Surgery, Florida Hospital, University of South Florida, Tampa, Fla, United States Bart Muhs, Vascular and Endovascular Surgery, Yale School of Medicine, New Haven, Conn, United States Mahmoud B. Malas, Endovascular Surgery, Heart and Vascular Institute, Johns Hopkins, Baltimore, Md, United States Syed Hussain, Vascular Surgery, OSF Saint Francis Medical Center, HeartCare Peoria, Peoria, Ill, United States NavYash Gupta, Vascular Surgery, NorthShore University, Skokie, Ill, United States Dittmar Bockler, Department of Vascular Surgery, University of Heidelberg, Heidelberg, Germany Eric Verhoeven, Vascular and Endovascular Surgery, Klinikum Nuremberg, Nuremberg, Germany Michel Reijnen, Vascular Surgery, Rijnstate Hospital, Arnhem, Netherlands
Outcome-based anatomic criteria for defining the hostile aortic neck William D. Jordan Jr, MD,a Kenneth Ouriel, MD,b Manish Mehta, MD, MPH,c David Varnagy, MD,d William M. Moore Jr, MD,e Frank R. Arko, MD,f James Joye, DO,g and Jean-Paul P. M. de Vries, MD,h Birmingham, Ala; New York and Albany, NY; Orlando, Fla; West Columbia, SC; Charlotte, NC; Mountain View, Calif; and Nieuwegein, The Netherlands Objective: There is abundant evidence linking hostile proximal aortic neck anatomy to poor outcome after endovascular aortic aneurysm repair (EVAR), yet the definition of hostile anatomy varies from study to study. This current analysis was undertaken to identify anatomic criteria that are most predictive of success or failure at the aortic neck after EVAR. Methods: The study group comprised 221 patients in the Aneurysm Treatment using the Heli-FX Aortic Securement System Global Registry (ANCHOR) clinical trial, a population enriched with patients with challenging aortic neck anatomy and failure of sealing. Imaging protocols were not protocol specified but were performed according to the institution’s standard of care. Core laboratory analysis assessed the three-dimensional centerline-reformatted computed tomography scans. Failure at the aortic neck was defined by type Ia endoleak occurring at the time of the initial endograft implantation or during follow-up. Receiver operating characteristic curve analysis was used to assess the value of each anatomic measure in the classification of aortic neck success and failure and to identify optimal thresholds of discrimination. Binary logistic regression was performed after excluding highly intercorrelated variables, creating a final model with significant predictors of outcome after EVAR. Results: Among the 221 patients, 121 (54.8%) remained free of type Ia endoleak and 100 (45.2%) did not. Type Ia endoleaks presented immediately after endograft deployment in 58 (58.0%) or during follow-up in 42 (42.0%). Receiver operating characteristic curve analysis identified 12 variables where the classification of patients with type Ia endoleak was significantly more accurate than chance alone. Increased aortic neck diameter at the lowest renal artery (P [ .013) and at 5 mm (P [ .008), 10 mm (P [ .008), and 15 mm (P [ .010) distally; aneurysm sac diameter (P [ .001), common iliac artery diameters (right, P [ .012; left, P [ .032), and a conical (P [ .049) neck configuration were predictive of endoleak. By contrast, increased aortic neck length (P [ .050), a funnel-shaped aortic neck (P [ .036), and neck mural thrombus content, as measured by average thickness (P [ .044) or degrees of circumferential coverage (P [ .029), were protective against endoleak. Binary logistic regression identified three variables independently predictive of type Ia endoleak. Neck diameter at the lowest renal artery (P [ .002, cutpoint 26 mm) and neck length (P [ .017, cutpoint 17 mm) were associated with endoleak, whereas some mural neck thrombus content was protective (P [ .001, cutpoint 11 of circumferential coverage). Conclusions: A limited number of independent anatomic variables are predictive of type Ia endoleak after EVAR, including aortic neck diameter and aortic neck length, whereas mural thrombus in the neck is protective. This study suggests that anatomic measures with identifiable threshold cutpoints should be considered when defining the hostile aortic neck and assessing the risk of complications after EVAR. (J Vasc Surg 2015;61:1383-90.)
Endovascular aneurysm repair (EVAR) surpassed open surgery as the most frequently performed treatment option for patients with abdominal aortic aneurysms (AAAs) within
a decade after the first endografts were marketed.1 EVAR, as a less invasive technique, has many advantages over open surgical repair, principally related to early morbidity and
From the Division of Vascular Surgery and Endovascular Therapy, University of Alabama-Birmingham, Birminghama; Syntactx, New Yorkb; Vascular Surgery, Albany Vascular Group, The Institute for Vascular Health and Disease, Albanyc; Vascular Surgery, Florida Hospital, Orlandod; the Southern Surgical Group, Lexington Medical Center, West Columbiae; Vascular Surgery, Sanger Heart & Vascular Institute, Carolinas Health Care System, Charlottef; Cardiovascular Disease, Heart & Vascular Institute, El Camino Hospital, Mountain Viewg; and the Department of Vascular Surgery, St. Antonius Hospital, Nieuwegein.h Author conflict of interest: K.O. has equity ownership in Syntactx, a company that receives fees for contract research activities from Aptus, the study sponsor. Aptus, the sponsor of the study, was not involved in the data collection, data analysis, data interpretation, manuscript writing, or the decision to submit the manuscript. Aptus employees were, however, provided with a draft copy of the manuscript for comment.
Presented at the Thirty-eighth Annual Meeting of the Midwestern Vascular Surgical Society, Coralville, Iowa, September 4-6, 2014. Additional material for this article may be found online at www.jvascsurg.org. Reprint requests: William D. Jordan, MD, Division of Vascular Surgery and Endovascular Therapy, University of Alabama at Birmingham, 1808 7th Ave S, BDB 503, Birmingham, AL 35294 (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. All rights reserved. http://dx.doi.org/10.1016/j.jvs.2014.12.063
1383
1384 Jordan et al
mortality.2,3 The procedure, however, lacks the flexibility and durability of open surgical graft insertion, primarily a result of the interdependence of endograft sealing and the anatomic aspects of the proximal aortic neck.4,5 Outcome after EVAR is inferior when aortic neck anatomic irregularities are encountered, including a short neck length, large neck diameter, extreme neck angulation, and abundant mural thrombus or calcium within the neck.4,6,7 Although an open surgical anastomosis can be more difficult to perform in the presence of such anatomic issues, an open procedure depends less on neck anatomy than EVAR. The term “hostile neck” was first used by Dillavou7 in 2003 to characterize EVAR outcome in patients with unfavorable aortic neck anatomy. Currently, the term is often used when the aortic neck anatomy falls outside the eligibility criteria for a manufacturer’s regulatory clinical trial. Recognition of hostile neck characteristics creates some concern for endograft performance at the aortic neck. Endograft trials are limited to patients with anatomy well suited for a particular endograft, and regulatory approval is granted before the device is used in the general population. Once approved for marketing, however, a broad range of anatomies are encountered, anatomies that were not evaluated in regulatory trials and thus have undefined clinical outcomes. Despite widespread reliance on anatomic criteria for clinical decision making, there has been little work on characterizing the relative importance of different anatomic measures and on identifying optimal thresholds for each. In part, the paucity of data is a result of the infrequency of aortic neck complications. The reported frequency of type Ia endoleak is <3% in most series, precluding a rigorous multivariate analysis of the factors related to the event.8,9 By contrast, the Aneurysm Treatment using the Heli-FX Aortic Securement System Global Registry (ANCHOR) data set provides a study population enriched with challenging aortic anatomy and aortic neck complications.10 As such, the data set enables robust statistical comparisons of each anatomic measure across multiple endografts and operators. In tandem with receiver operating characteristic (ROC) analysis, optimal thresholds for classifying neck complications can be formulated from anatomic criteria. In this report we describe an analysis of hostile neck criteria and their comparative effect on outcome after EVAR. METHODS The ANCHOR study is a prospective, nonrandomized, multicenter, multinational study of the real-world use of the Heli-FX EndoAnchor System (Aptus Endosystems, Sunnyvale, Calif) in patients undergoing EVAR or who have undergone EVAR for AAA in the past. The ANCHOR study is registered on ClinicalTrials.gov (NCT01534819). Institutional Review Board or Ethics Committee approval was obtained at each site. Each patient provided written informed consent. Details of the study methodology and the device have been previously described.11 The investigators are listed in the Appendix (online only). Briefly, the study eligibility criteria included patients with infrarenal AAA who had adequate iliofemoral access
JOURNAL OF VASCULAR SURGERY June 2015
to accommodate a 16F sheath and a life expectancy of at least 1 year. Commercially available endografts that underwent successful testing for EndoAnchor compatibility included the Zenith (Cook, Bloomington, Ind), the Excluder (W. L. Gore and Associates, Flagstaff, Ariz), and the AneuRx, Talent, and Endurant devices (Medtronic Vascular, Santa Rosa, Calif). The study cohort comprised 221 patients enrolled in ANCHOR at 26 United States and 9 European centers between April 2012 and January 2014. The study population included 66.6% of 332 patients enrolled in the full study over the same timeframe, selected by availability of adequate baseline (before the EndoAnchor implantation procedure) and postoperative computed tomography (CT) imaging for core laboratory analysis. The 221 patients were subdivided into 121 EVAR patients without type Ia endoleak (54.3%) and 100 with aortic neck failure (54.8%) evidenced by type Ia endoleak alone in 86 patients (38.4%) or endoleak in conjunction with endograft migration in 14 (6.3%). Patients with immediate type Ia endoleaks evident on intraoperative angiography performed at the time of the initial EVAR (58 patients [26.2%]) were included in the failure group when EndoAnchors (with or without extension cuffs or bare metal stents) were implanted to address the endoleak, irrespective of whether the additional interventions remediated the endoleak (52 patients [23.5%]) or did not (six patients [2.7%]). The remaining 42 patients (19.0%) were treated with EndoAnchors for type Ia endoleaks a median of 35 months (range, 0.2-168 months) after the initial EVAR procedure. Once enrolled in ANCHOR, patients were monitored clinically for a median of 19 months (range, 0-30 months), with CT imaging studies performed through a median of 7 months (range, 0-23 months). Imaging studies and definitions. Imaging protocols were not protocol-specified but were performed according to the institution’s standard of care. Independent core laboratory analyses (Syntactx, New York, NY) were performed on noncontrast and contrast CT imaging studies. Centerline reformatting and segmentation of CT data sets was performed using iNtuition imaging software (TeraRecon, Foster City, Calif). Imaging end points were measured and reported using the methodology from the Society for Vascular Surgery reporting standards guideline documents where end point definitions were specificed.12-15 Aortic diameters were measured as the average diameter of the centerline reformatted aortic adventitia-toadventitia contour. Circularity was not assumed; rather, any deviation from perfect circularity was taken into account with an electronically traced aortic contour on a plane orthogonal to the aortic centerline. Aortic neck calcium and thrombus content was measured and expressed in degrees of circumference where thickness was $2 mm, as evaluated on the CT image 5 mm distal to the lowest main renal artery. The aortic neck length was calculated using two methods. The first method corresponds to what has been
JOURNAL OF VASCULAR SURGERY Volume 61, Number 6
defined as the “anatomic neck” and represented the centerline distance between the lowest main renal artery and the level where the diameter of the infrarenal aortic neck reached 10% more than the diameter at the lowest main renal ostium.16 The second method, dubbed the “visual neck” length, defined the distal margin as the visually ascertained inflection point where the neck transitioned into the aneurysm sac. Not being based upon strict geometric calculations, the visual neck was more subjectively determined than was the anatomic neck. Further, it can be longer than the anatomic neck in conically shaped necks with a gradual, tapered configuration along its length. Suprarenal and infrarenal aortic neck angulation were measured using three points (knots) along the centerline spline created from reformatted CT data sets: a proximal knot, an apex, and a distal knot. Suprarenal angulation was defined with a proximal knot 20 mm proximal to the lowest main renal artery, an apex at the ostium of the lowest main renal artery, and a distal knot at the distalmost aspect of the anatomic neck. Infrarenal angulation was measured using two methods that differed with respect to the distal knot. Both methods used a proximal knot at the level of the lowest main renal artery and an apex at the level of the distal-most aspect of the anatomic neck. However, the first method used a distal knot placed 40 mm distal to the apex, whereas the second used a knot located at the center-most point on the aortic bifurcation. In addition, noting that angulation defined by two vectors and an apex is a linear simplification of the true anatomic curvature of a neck, an “aortic tortuosity index” was expressed as the comparison between the curvilinear and straight line distance between the proximal (lowest renal artery) and distal (40 mm) knots. The aortic tortuosity index was calculated as the curvilinear distance divided by the straight-line distance. The degree to which the aortic neck was conical in configuration was assessed with neck “conicity,” defined as the percentage increase in aortic neck diameter between the lowest main renal artery and a specified length distally. Conicity was defined through four levels: 5, 10, and 15 mm beyond to the lowest main renal artery and at the distal margin of the visual neck. A positive value for conicity indicated reverse tapering of the aortic neck where the neck diameter increased with distal progression. A negative value implied a funnel-shaped neck that was larger proximally than distally. Data analysis. Statistical analyses were performed with SPSS 22 software (IBM Corp, Armonk, NY). Values are expressed as mean 6 standard deviation unless otherwise specified. P values were considered significant when the two-tailed a was <.05. ROC curve analysis was used to evaluate the diagnostic value of anatomic measures that have been considered to be related to complications at the aortic neck and to determine optimal thresholds or cutpoints for classifying aortic neck failure (type Ia endoleak) and success. The evaluated anatomic measures included aortic neck diameter, length, thrombus/calcium content,
Jordan et al 1385
degree of neck conicity, infrarenal angulation, suprarenal angulation, and maximum aneurysm diameter. ROC curves were plotted with sensitivity on the y axis and (1-specificity) on the x axis. Areas under the curve (AUCs) were calculated to assess discriminatory accuracy over the spectrum of values for a particular anatomic measure. AUCs are expressed as the estimate and its 95% confidence interval. The Wilcoxon-Mann-Whitney estimator was used to assess whether a variable was significantly better than chance alone in the classification of success and failure.17 Sensitivities and specificities for each anatomic measure to discriminate between aortic neck success and failure were determined at sequential cutpoints ranging between the minimum and maximum observed values for each measure. The optimum cutpoint was defined by a maximum balance between sensitivity and specificity. Binary logistic regression was performed with inclusion of anatomic covariates where the ROC AUC probability was <.10. Highly intercorrelated, redundant explanatory variables with r values >0.70 were eliminated to reduce the effects of multicollinearity.18 A backward stepwise process was used, sequentially removing variables with the least significance until the significance of all variables was <.05. RESULTS The 165 male (74.7%) and 56 female (25.3%) patients were a mean age of 74.1 6 8.2 years (range, 56-97 years). Endografts implanted included the Medtronic Endurant in 96 patients (43.4%), the Gore Excluder in 77 (34.8%), the Cook Zenith in 28 (12.7%), and other devices in 20 (9.0%). The baseline anatomic characteristics of the study population are reported in Table I. ROC curve analysis. Aortic neck diameter was a significant determinant of type Ia endoleak when measured at the lowest renal artery and at 5-mm increments distally to 15 mm (Table II and Fig 1). The diameter at the distal margin of the visual neck was not of significant discriminatory value nor was diameter at the suprarenal level. A diameter of 26 mm at the lowest renal artery was the optimal cutpoint for maximizing the trade-off between sensitivity and specificity. Corresponding cutpoints for the more distal levels within the neck increased 1 mm for every 5 mm of distal neck length. Of the two methods of measuring aortic neck length, anatomic neck length was not statistically significant in the classification of neck success or failure (P ¼ .069), and visual neck length was of marginal statistical significance (P ¼ .050). The optimal cutpoints for the two measures were 11 mm and 17 mm, respectively (Fig 2). No measure of neck angulation attained statistical significance in the discrimination of type Ia endoleaks. The aortic tortuosity index was the best but still did not attain significance (P ¼ .068, cutpoint 1.04). However, extreme angulation was relatively rare in the study cohort, with 23 patients (10.4%) manifesting infrarenal neck angulation >60 . Aortic neck conicity was a useful determinant by ROC curve analysis. A reverse-taper neck configuration
JOURNAL OF VASCULAR SURGERY June 2015
1386 Jordan et al
Table I. Baseline anatomic characteristics in 100 patients with and 121 patients without type Ia endoleak Anatomic measure
Type Ia endoleak
Mean
SD
P value
Suprarenal aortic diameter, mm
No Yes No Yes
27.3 28.1 25.9 27.8
3.2 4.9 4.2 6.5
.113
No Yes No Yes No Yes No Yes No Yes No Yes
26.6 29.0 27.8 30.3 28.9 31.0 29.2 30.2 18.0 14.8 22.8 18.2
4.7 7.7 4.9 8.2 6.1 8.8 4.9 6.3 13.8 11.2 16.8 14.8
No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes
2.9 4.0 7.0 8.4 11.7 8.8 13.5 9.4 1.06 1.05 55.0 59.8 15.9 15.8 25.9 23.2 35.4 35.5 1.0 0.6 65.8 29.9 1.0 0.9 20.7 17.7
12.0 6.8 9.6 16.6 17.5 21.9 14.7 10.6 0.06 0.06 10.5 14.3 12.5 12.2 16.3 14.9 19.1 15.7 1.4 1.4 93.4 59.3 1.2 1.2 31.0 36.2
Aortic diameter at lowest renal, mm Neck diameter 5 mm distal to lowest renal, mm 10 mm distal to lowest renal, mm 15 mm distal to lowest renal, mm At distal end of visual neck, mm Anatomic neck length (10% threshold, mm) Visual neck length, mm Conicity to 5 mm (% increase over distance) 10 mm (% increase over distance) to 15 mm (% increase over distance) Conicity through visual neck (% increase) Neck tortuosity index (curvilinear/straight length) Maximum aneurysm sac diameter, mm Suprarenal angulation, Infrarenal angulation, Infrarenal angulation to bifurcation, Neck thrombus average thickness, mm
Neck thrombus circumference,
Neck calcium average thickness, mm
Neck calcium circumference,
.009 .006 .006 .054 .192 .065 .034 .422 .446 .316 .017 .176 .006 .972 .205 .961 .038 .001 .383 .509
SD, Standard deviation.
(distal diameter greater than immediate infrarenal diameter) through 5 mm increased the risk of type Ia endoleak (P ¼ .049, optimal cutpoint 2% grade). By contrast, protection against endoleak was conferred by a funnelshaped configuration of the visual neck, where the neck diameter at the visually determined start of the aneurysm was smaller than the diameter at the immediate infrarenal level (P ¼ .036, optimal cutpoint 9% grade). Neither aortic neck calcium thickness nor degrees of circumferential calcium content attained statistical significance as discriminators of type Ia endoleak (Fig 3). By contrast, a significant positive discriminatory value was observed for mural neck thrombus, expressed as average thickness (P ¼ .044) or as degrees of neck circumference covered with $2 mm of mural thrombus (P ¼ .029). The optimal thresholds for thrombus thickness and
circumference coverage were 0 mm and 11 , respectively, suggesting that virtually any mural neck thrombus positively affected outcome. However, noting only 14 patients (6.3%) had >180 of mural thrombus and only 12 patients (5.4%) had thrombus >3 mm in average thickness, the current analysis cannot exclude a negative effect of extreme levels of thrombus. Maximum aneurysm sac diameter was a good discriminator of type Ia endoleak (P ¼ .001). Right and left common iliac diameters were similarly associated with endoleak (P ¼ .012 and P ¼ .032, respectively). Optimal thresholds for discrimination were 55 mm for sac diameter, 19 mm for the right common iliac artery, and 18 mm for the left common iliac artery. Multivariate analysis. After excluding variables with significant intercorrelation or ROC AUC significance
JOURNAL OF VASCULAR SURGERY Volume 61, Number 6
Jordan et al 1387
Table II. Receiver operating characteristic (ROC) analyses of anatomic measures for type Ia endoleak after aneurysm repaira Anatomic measure Aortic neck diameter At lowest renal artery 5 mm distal to renal artery 10 mm distal to renal artery 15 mm distal to renal artery At end of visual neck Suprarenal aortic diameter Aortic neck length Anatomic neck Visual neck Neck angulation Suprarenal Infrarenal Infrarenal to bifurcation Aortic tortuosity index Aortic neck thrombus/calcium Thrombus thickness Thrombus circumference Calcium thickness Calcium circumference Aortic neck conicity Through 5 mm neck length Through 10 mm neck length Through 15 mm neck length Through length of visual neckc Vessel diameter Maximum sac diameter Right common iliac artery Left common iliac artery
AUC (95% CI) 0.597 0.604 0.604 0.601 0.532 0.459
(0.522-0.672) (0.528-0.679) (0.529-0.680) (0.526-0.677) (0.455-0.608) (0.382-0.536)
0.571 (0.495-0.647) 0.576 (0.501-0.652)
Optimal cutpoint <26 <27 <28 <29
mm mm mm mm
P value .013b .008b .008b .010b .415 .292
> 11 mm >17 mm
.069 .050b
<1.04
.715 .259 .527 .068
0.486 0.544 0.475 0.571
(0.409-0.562) (0.468-0.620) (0.399-0.551) (0.496-0.647)
0.579 0.585 0.530 0.548
(0.504-0.654) (0.510-0.660) (0.454-0.607) (0.472-0.624)
>0 mm >11
.044b .029b .439 .220
0.577 0.545 0.524 0.582
(0.500-0.653) (0.468-0.623) (0.447-0.601) (0.507-0.657)
<2% >9%
.049b .246 .541 .036b
<55 mm <19 mm <18 mm
.001b .012b .032b
0.621 (0.546-0.696) 0.600 (0.526-0.676) 0.586 (0.508-0.663)
AUC, Area under the curve; CI, confidence interval. a Optimal cutpoints balanced sensitivity and specificity; calculated for variables with significance <.10. Cutpoints were defined as the optimum thresholds above which type Ia endoleaks were more likely. b Significant at P < .05 a level. c Funnel shape (proximal larger than distal) through visual neck was protective against failure. Endoleak was less likely when the diameter at the distal end of the visual neck was less than the immediate infrarenal aortic diameter.
of <.10, 14 anatomic covariates were included in the binary logistic regression analysis. The backward stepwise model identified three variables that were significantly predictors of type Ia endoleak (Table III). These were a larger aortic neck diameter at the level of the lowest renal artery (P ¼ .002), a shorter anatomic neck length (P ¼ .017), and less mural thrombus (P ¼ .001). The final model with these covariates was of significant utility in predicting type Ia endoleaks (P < .001). DISCUSSION The link between aortic neck anatomy and the development of complications, such as type Ia endoleak and endograft migration, has been evident since the inception of EVAR in the early 1990s.19-21 The literature is replete with investigations linking the development of adverse clinical events and imaging findings with challenging neck anatomy.4,5,22 Further, the instructions for use of all currently marketed endografts contain anatomic indications that generally parallel the eligibility criteria of the investigational device exemption approval trials submitted for regulatory approval.
Despite the intuitive value and empiric association between various anatomic criteria and outcome after EVAR, there are sparse data in the literature to objectively compare the importance of the indices. As well, prior reports have not objectively evaluated optimal cutpoints to identify whether the anatomy of a particular patient places that patient at risk for aortic neck failure. The absence of such an analysis is understandable: few prior series have an adequate number of EVAR failures to adequately investigate predictive factors. The ANCHOR series is unique in this regard. Patients are specifically selected and enrolled based the presence of challenging neck anatomy or after the development of a type Ia endoleak.11 For this reason, the ANCHOR data set was used in conjunction with ROC curve analysis and logistic regression to identify those baseline anatomic measures associated with aortic neck complications during or after EVAR. The current study confirmed the importance of aortic neck length and diameter as the important determinants of type Ia endoleak after EVAR. These factors have been identified in prior reports as standard anatomic criteria that should be taken into account in the classification of
JOURNAL OF VASCULAR SURGERY June 2015
1388 Jordan et al
Fig 1. Receiver operating characteristic (ROC) curves for aortic neck diameter at the lowest renal artery and at 5-mm increments distally to 15 mm.
Fig 3. Receiver operating characteristic (ROC) curves for thrombus and calcium content within the proximal aortic neck, assessed 5 mm distal to the lowest renal artery. Each end point is measured as the average thickness over the area of involvement as well as the degrees of circumference covered with >1 mm of mural thrombus or calcium. Mural thrombus was protective against type Ia endoleak. The effect of calcium was opposite but did not attain statistical significance as a predictive variable.
Table III. Binary logistic regression: Final model Variable Aortic neck diameter at lowest renal artery (mm) Anatomic neck length (10% threshold, mm) Neck circumference with mural thrombus (mm) Intercept
Coefficient
Standard Odds P error ratioa value
0.102
0.033
1.11 .002
0.028
0.012
0.97 .017
0.007
0.002
0.99 .001
2.133
0.871
.014
a
Odds ratios indicate the increase (or reduction) in the risk of type Ia endoleak for every unit increase in the covariate. For instance, the odds ratio of 1.11 for aortic neck diameter indicates that the risk of type Ia endoleak increases 11% for each 1-mm increase in neck diameter. Similarly, the odds ratio of .99 for mural neck thrombus indicates the risk of type Ia endoleak decreases by 1% for each degree of circumferential mural thrombus.
Fig 2. Receiver operating characteristic (ROC) curves for anatomic neck length and visual neck length. Anatomic neck length is measured to the level at which the vessel diameter is 10% greater than its diameter at the lowest renal artery, whereas visual neck length is measured to the visually ascertained aortic wall inflection point that identifies the start of the aneurysm.
hostile vs friendly aortic neck anatomy.4-6 The current analysis did not identify other accepted criteriadmost notably aortic neck angulation and neck calcium contentdas independent predictors of type Ia endoleak. The failure to identify neck angulation as an important determinant may partly relate to a low prevalence of patients with extreme degrees of
angulation in the series. Similar limitations may exist for mural thrombus and calcium content; extreme amounts of each were uncommon in this series. Of importance, the multivariable analysis suggested that the most important determinants of aortic neck complications were neck diameter, neck length, and thrombus content. Other variables fell out as important predictors in the final binary logistic regression, supplanted by presumably more important variables in the reduced model. Aortic neck thrombus, intuitively, should be a risk factor for aortic neck failure.23 In fact, this is not the case in
JOURNAL OF VASCULAR SURGERY Volume 61, Number 6
prior reports. The study of Bastos Gonçalves et al24 documented a higher rate of 5-year clinical success in patients with thrombus >2 mm in thickness covering >25% of neck circumference, 74% vs 62%, but this difference did not attain statistical significance. Our data corroborate these observations; some amount of neck thrombus appears to be protective against type Ia endoleak. The current data cannot exclude a reversal of this relationship when the amount of mural thrombus is extreme. Because the clinical follow-up is relatively short, additional anatomic criteria other than those identified may also be predictive over longer-term follow-up. The ROC curve analysis used in the current analysis provides some potential advantages over other statistical methods. The technique is independent of arbitrarily chosen cutpoints or the prevalence of the event, in this case type Ia endoleak, in the population under study. The current analytic methodology, however, has some limitations: First, preoperative CT images from the time of the initial EVAR were unavailable in patients who presented with late type Ia endoleaks. Postimplant imaging studies were used in these cases, potentially underestimating preoperative baseline neck angulation or neck thrombus content and overestimating baseline neck diameter in the endoleak subgroup, to the extent and direction that these measures are known to change after endograft implantation.24-26 This limitation may underlie the failure to identify certain variables as risk factors for type Ia endoleak. A second limitation relates to the use of a data set comprising patients selected for EndoAnchor use. Patients with findings that preclude use of the device would not have been included. For instance, patients with extreme amounts of neck thrombus or calcium or those with neck dilatation where a large gap developed between the endograft and the aortic wall would not have been included. Third, the importance of neck and aneurysm sac diameter may be overestimated in the model because each may have increased to some extent compared with the preoperative baseline values. Fourth, the study design dichotomized the data set into just two groupsdsuccess and failuredwhere the failures include patients with type Ia endoleaks detected at the time of endograft implantation as well as those identified on postoperative imaging studies. Lastly, the relationship between anatomic factors and type Ia endoleak may be endograft dependent, and the current study pooled the results of all devices into a single analysis. CONCLUSIONS The current analysis suggests that just a few anatomic measures are associated with type Ia endoleak after EVAR, and ROC analysis offers a useful method to assign logical predictors and their cutpoints for defining the hostile aortic neck. Confirmation of these findings in a larger data set with complete assessment of all preoperative and longer-term postoperative imaging studies will be necessary to validate these findings. Until then, the observations of
Jordan et al 1389
this study provide the groundwork for a developing a uniform definition of the hostile neck when evaluating outcome after EVAR. The findings may provide some basis for the application of intraoperative adjunctive techniques to EVAR, even in the absence of an angiographically identified endoleak, and adjustments to postoperative surveillance imaging regimens when challenging aortic neck anatomy is encountered. AUTHOR CONTRIBUTIONS Conception and design: JD, KO, WJ Analysis and interpretation: JD, KO, MM, DV, WM, FA, JJ, WJ Data collection: KO Writing the article: JD, KO, WJ Critical revision of the article: JD, KO, MM, DV, WM, FA, JJ, WJ Final approval of the article: JD, KO, MM, DV, WM, FA, JJ, WJ Statistical analysis: KO Obtained funding: JD, KO, WJ Overall responsibility: KO REFERENCES 1. Schwarze ML, Shen Y, Hemmerich J, Dale W. Age-related trends in utilization and outcome of open and endovascular repair for abdominal aortic aneurysm in the United States, 2001-2006. J Vasc Surg 2009;50: 722-9. 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. Anderson PL, Arons RR, Moskowitz AJ, Gelijns A, Magnell C, Faries PL, et al. A statewide experience with endovascular abdominal aortic aneurysm repair: rapid diffusion with excellent early results. J Vasc Surg 2004;39:10-9. 4. 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. 5. Stather PW, Wild JB, Sayers RD, Bown MJ, Choke E. Endovascular aortic aneurysm repair in patients with hostile neck anatomy. J Endovasc Ther 2013;20:623-37. 6. Aburahma AF, Campbell JE, Mousa AY, Hass SM, Stone PA, Jain A, et al. Clinical outcomes for hostile versus favorable aortic neck anatomy in endovascular aortic aneurysm repair using modular devices. J Vasc Surg 2011;54:13-21. 7. Dillavou ED, Muluk SC, Rhee RY, Tzeng E, Woody JD, Gupta N, et al. Does hostile neck anatomy preclude successful endovascular aortic aneurysm repair? J Vasc Surg 2003;38:657-63. 8. Stokmans RA, Teijink JA, Forbes TL, Bockler D, Peeters PJ, Riambau V, et al. Early results from the ENGAGE registry: real-world performance of the Endurant Stent Graft for endovascular AAA repair in 1262 patients. Eur J Vasc Endovasc Surg 2012;44:369-75. 9. Cao P, De RP, Parlani G, Verzini F. Durability of abdominal aortic endograft with the Talent Unidoc stent graft in common practice: core lab reanalysis from the TAURIS multicenter study. J Vasc Surg 2009;49:859-65. 10. De Vries JP, Van De Pavoordt HD, Jordan WD Jr. Rationale of EndoAnchors in abdominal aortic aneurysms with short or angulated necks. J Cardiovasc Surg (Torino) 2014;55:103-7. 11. Jordan WD Jr, Mehta M, Varnagy D, Moore WM Jr, Arko FR, Joye J, et al. Results of the ANCHOR prospective, multicenter registry of EndoAnchors for type Ia endoleaks and endograft migration in patients with challenging anatomy. J Vasc Surg 2014;60:885-92.e2.
1390 Jordan et al
12. 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. 13. Chaikof EL, Fillinger MF, Matsumura JS, Rutherford RB, White GH, Blankensteijn JD, et al. Identifying and grading factors that modify the outcome of endovascular aortic aneurysm repair. J Vasc Surg 2002;35: 1061-6. 14. Ouriel K, Fowl RJ, Davies MG, Forbes TL, Gambhir RP, Morales JP, et al. Reporting standards for adverse events after medical device use in the peripheral vascular system. J Vasc Surg 2013;58:776-86. 15. Ouriel K, Fowl RJ, Davies MG, Forbes TL, Gambhir RP, Ricci MA, et al. Disease-specific guidelines for reporting adverse events for peripheral vascular medical devices. J Vasc Surg 2014;60:212-25. 16. Welborn MB 3rd, McDaniel HB, Johnson RC, Kennedy RE, Knott A, Mundinger GH, et al. Clinical outcome of an extended proximal seal zone with the AFX endovascular aortic aneurysm system. J Vasc Surg 2014;60:876-83; discussion: 883-4. 17. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988;44:837-45. 18. Tabachnick BG, Fidell LS. Using multivariate statistics. 6th edition. Boston: Pearson; 2013. 19. Fox AD, Whiteley MS, Murphy P, Budd JS, Horrocks M. Comparison of magnetic resonance imaging measurements of abdominal aortic aneurysms with measurements obtained by other imaging techniques and intraoperative measurements: possible implications for endovascular grafting. J Vasc Surg 1996;24:632-8. 20. Moore WS, Vescera CL. Repair of abdominal aortic aneurysm by transfemoral endovascular graft placement. Ann Surg 1994;220:331-9; discussion: 339-41.
JOURNAL OF VASCULAR SURGERY June 2015
21. Parodi JC. Endovascular repair of abdominal aortic aneurysms and other arterial lesions. J Vasc Surg 1995;21:549-55. 22. Rockman CB, Lamparello PJ, Adelman MA, Jacobowitz GR, Therff S, Gagne PJ, et al. Aneurysm morphology as a predictor of endoleak following endovascular aortic aneurysm repair: do smaller aneurysm have better outcomes? Ann Vasc Surg 2002;16:644-51. 23. Walker TG, Kalva SP, Yeddula K, Wicky S, Kundu S, Drescher P, et al. Clinical practice guidelines for endovascular abdominal aortic aneurysm repair: written by the Standards of Practice Committee for the Society of Interventional Radiology and endorsed by the Cardiovascular and Interventional Radiological Society of Europe and the Canadian Interventional Radiology Association. J Vasc Interv Radiol 2010;21: 1632-55. 24. Bastos Gonçalves GF, Verhagen HJ, Chinsakchai K, van Keulen JW, Voute MT, Zandvoort HJ, et al. The influence of neck thrombus on clinical outcome and aneurysm morphology after endovascular aneurysm repair. J Vasc Surg 2012;56:36-44. 25. Cao P, Verzini F, Parlani G, Rango PD, Parente B, Giordano G, et al. Predictive factors and clinical consequences of proximal aortic neck dilatation in 230 patients undergoing abdominal aorta aneurysm repair with self-expandable stent-grafts. J Vasc Surg 2003;37:1200-5. 26. Ishibashi H, Ishiguchi T, Ohta T, Sugimoto I, Yamada T, Tadakoshi M, et al. Remodeling of proximal neck angulation after endovascular aneurysm repair. J Vasc Surg 2012;56:1201-5.
Submitted Nov 24, 2014; accepted Dec 21, 2014.
Additional material for this article may be found online at www.jvascsurg.org.
JOURNAL OF VASCULAR SURGERY Volume 61, Number 6
APPENDIX (online only). Collaborators and institutions involved in the Aneurysm Treatment using the Heli-FX Aortic Securement System Global Registry (ANCHOR) William D. Jordan Jr., Vascular Surgery and Endovascular Therapy, University of Alabama, Birmingham, Ala, United States Jean Paul de Vries, Department of Vascular Surgery, St. Antonius Hospital, Nieuwegein, Netherlands James Joye, Cardiovascular Disease, Heart & Vascular Institute, El Camino Hospital, Mountain View, Calif, United States H.H. Eckstein, Department of Vascular Surgery, Clinic for Vascularand Endovascular Surgery, Klinikum Rechts der Isar der Technischen, Technical University Munich, Munich, Germany Joost van Herwaarden, Department of Vascular Surgery, University Medical Center Utrecht, Utrecht, Netherlands Frank R. Arko, Vascular Surgery, Sanger Heart & Vascular Institute, Carolinas Health Care System, Charlotte, NC, United States Paul Bove, Vascular and Endovascular Surgery, Peripheral Vascular Diagnostic Center, William Beaumont, Royal Oak, Mich, United States William Bohannon, Vascular Surgery, Scott & White Medical Center, Temple, Tex, United States Bram Fioole, Department of Vascular Surgery, Maasstad Medical Hospital, Rotterdam, The Netherlands Carlo Setacci, Vascularand Endovascular Surgery UniteUniversity of Siena, Siena, Italy Timothy Resch, Department of Vascular Diseases, Malmö University Hospital, Malmö, Sweden Vicente Riambau, Angiology and Vascular Surgery, Thorax Institute Hospital Clinic, Barcelona, Spain Dierk Scheinert and Andrej Schmidt, Angiology and Vascular Surgery, Park Hospital, Leipzig, Germany Daniel Clair, Vascular Surgery, Cleveland Clinic, Cleveland, Ohio, United States Mohammed Moursi, Vascular Surgery, Central Arkansas Veterans Health System, University of Arkansas for Medical Sciences, Little Rock, Ark, United States Mark Farber, Vascular Surgery, School of Medicine, University of North Carolina, Chapel Hill, NC, United States Joerg Tessarek, Department of Vascular Surgery, Vascular Center, St. Bonifatius Hospital, Lingen, Germany Giovanni Torsello, Vascular Surgery, St. Franziskus, Hospital GmbH, Munster, Germany Mark Fillinger, Vascular Surgery, Geisel School of Medicine, Dartmouth, Hitchcock Medical Center, Lebanon, NH, United States Marc Glickman, Vascular Surgery, Sentara Heart Hospital, Norfolk, Va, United States John Henretta, Vascular Surgery, Carolina Vascular, Mission Hospital, Asheville, NC, United States Kim Hodgson, Vascular Surgery Division, School of Medicine, Southern Illinois University, Springfield, Ill, United States
Jordan et al 1390.e1
Jeffrey Jim, Section of Vascular Surgery, Washington University School of Medicine, St. Louis, Mo, United States Barry Katzen, Vascular and Interventional Radiology, Baptist Cardiac & Vascular Institute, Baptist Hospital, Miami, Fla, United States Evan Lipsitz, Cardiovascular and Thoracic Surgery, Montefiore Medical Center, Bronx, NY, United States Mitchell Cox, Vascular Surgery, Duke University Medical Center, Durham, SC, United States Thomas Naslund, Division of Vascular Surgery, Vanderbilt University Medical Center, Nashville, Tenn, United States Venkatesh Ramaiah, Vascular Surgery, Arizona Heart Institute, Abrazo Health Care, Phoenix, Ariz, United States Marc Schermerhorn, Vascular and Endovascular Surgery, Beth Israel Deaconess Medical Center, Boston, Mass, United States Peter Schneider, Vascular and Endovascular Surgery, Hawaii Permanente Medical Group, Honolulu, Hawaii, United States Benjamin Ware Starnes, Vascular Surgery Division, Harborview Medical Center, University of Washington, Seattle, Wash, United States Carlos Donayre, Vascular and Endovascular Surgery, Harbor-UCLA Medical Center, Torrance, Calif, United States Manish Mehta, Vascular Surgery, Albany Vascular Group, The Institute for Vascular Health and Disease, Albany, NY, United States Burkhart Zipfel, Department of Cardiothoracic and Vascular Surgery, Deutsches Herzzentrum, Berlin, Germany Nitin Malhotra, Vascular Surgery, Michigan Vascular Center, Flint, Mich, United States David Varnagy, Vascular Surgery, Florida Hospital, Orlando, Fla, United States William Moore Jr, Vascular Disease, Lexington Medical Center, West Columbia, SC, United States Nick Cheshire and Colin Bicknell, Vascular Surgery, Imperial College, London, United Kingdom Martin Back, Vascular Surgery, Florida Hospital, University of South Florida, Tampa, Fla, United States Bart Muhs, Vascular and Endovascular Surgery, Yale School of Medicine, New Haven, Conn, United States Mahmoud B. Malas, Endovascular Surgery, Heart and Vascular Institute, Johns Hopkins, Baltimore, Md, United States Syed Hussain, Vascular Surgery, OSF Saint Francis Medical Center, HeartCare Peoria, Peoria, Ill, United States NavYash Gupta, Vascular Surgery, NorthShore University, Skokie, Ill, United States Dittmar Bockler, Department of Vascular Surgery, University of Heidelberg, Heidelberg, Germany Eric Verhoeven, Vascular and Endovascular Surgery, Klinikum Nuremberg, Nuremberg, Germany Michel Reijnen, Vascular Surgery, Rijnstate Hospital, Arnhem, Netherlands