Note: This copy is for your personal non-commercial use only.

To order presentation-ready
and Interventional Radiology copies for distribution to your colleagues or clients, contact us at www.rsna.org/rsnarights.

CT Features of Early Type II

Endoleaks after Endovascular
Repair of Abdominal Aortic
Aneurysms Help Predict
Aneurysm Sac Enlargement1
Original Research  n  Vascular

René Müller-Wille, MD
Purpose: To determine computed tomographic (CT) features of
Sophie Schötz
early type II endoleaks associated with aneurysm sac
Florian Zeman, MMath enlargement after endovascular aortic aneurysm repair
Wibke Uller, MD (EVAR) of abdominal aortic aneurysm.
Oliver Güntner, MD
Karin Pfister, MD Materials and Institutional review board approval was not required for
Piotr Kasprzak, MD Methods: this retrospective study. The authors reviewed imaging
Christian Stroszczynski, MD and clinical data from 56 patients (seven women, 49 men;
Walter A. Wohlgemuth, MD mean age 6 standard deviation, 71 years 6 7.9; age range,
52–85 years) with early type II endoleak who had under-
gone EVAR between December 2002 and December 2011
and who had been followed up with imaging and clinical
evaluation for at least 6 months. The number and diameter
of all feeding and/or draining arteries were measured, and
endoleaks were classified according to their sources into
simple inferior mesenteric artery (IMA), simple lumbar
artery (LA), complex LA, and complex IMA-LA type II en-
doleaks. Volume and attenuation of the nidus were mea-
sured. Aneurysm enlargement was defined as an increase
in the aneurysm volume of more than 5% during follow-up.
Simple and multivariate logistic regression analyses were
performed to identify independent clinical and imaging
variables associated with aneurysm enlargement.

Results: Twenty-three of the 56 patients (41%) showed aneurysm

1
 From the Department of Radiology (R.M.W., S.S., W.U.,
O.G., C.S., W.A.W.), Center for Clinical Studies (F.Z.), and
Department of Surgery (K.P., P.K.), University Medical
Center Regensburg, Franz-Josef-Strauss-Allee 11,
93053 Regensburg, Germany. Received February 14,
2014; revision requested March 31; revision received
May 29; accepted July 10; final version accepted August
sac enlargement during follow-up (mean follow-up, 3.0
years 6 2.0). With the multivariate model, the variables
that showed the strongest indicators for aneurysm sac en-
largement were complex IMA-LA type II endoleak (odds
ratio [OR] = 10.29, P = .004) and the diameter of the
largest feeding and/or draining artery (OR = 4.55, P =
.013). Patients without complex IMA-LA type II endoleak
in whom the largest feeding and/or draining artery was
larger than 3.8 mm and patients with a complex IMA-LA
type II endoleak in whom the largest feeding and/or drain-
ing artery was larger than 2.2 mm were at high risk for
aneurysm sac enlargement.
Conclusion: The strongest indicators for aneurysm sac enlargement
are complex IMA-LA type II endoleak and the diameter of
the largest feeding and/or draining artery.
 RSNA, 2014
q

26. Address correspondence to R.M.W. (e-mail: rene.

mueller-wille@ukr.de).

q
 RSNA, 2014

906 radiology.rsna.org  n  Radiology: Volume 274: Number 3—March 2015

VASCULAR AND INTERVENTIONAL RADIOLOGY: CT Features of Early Type II Endoleaks after Aneurysm Repair Müller-Wille et al
E
ndovascular aortic aneurysm re-
pair (EVAR) plays an important
role in the treatment of abdom-
inal aortic aneurysm because it is as-
sociated with lower rates of surgical
mortality and complications compared
with open repair (1). A disadvantage
of EVAR is the development of type II
endoleaks, which occur in 17%–19%
of patients (2,3). This type of en-
doleak is related to retrograde filling
Advances in Knowledge
nn Patients with aneurysm sac en-
largement had significantly more
feeding and/or draining arteries
(mean 6 standard deviation, 3.5
6 1.3 vs 2.2 6 1.1; odds ratio
[OR] = 2.43; P = .002) and a sig-
nificantly larger nidus volume in
the arterial phase (median, 10.2
cm3 [interquartile range, 4.1–
19.2 cm3] vs 4.1 cm3 [interquar-
tile range, 2.1–8.6 cm3]; OR =
1.07; P = .04).
nn The CT variables with the highest
association with aneurysm sac
enlargement were complex infe-
rior mesenteric artery (IMA)–
lumbar artery (LA) type II
endoleak (OR = 10.29, P = .004)
and the diameter of the largest
feeding and/or draining artery
(OR = 4.55, P = .013).
nn The combination of complex
IMA-LA type II endoleak and the
diameter of the largest feeding
and/or draining artery gives an
indication of patient risk for
aneurysm sac enlargement
during follow-up (sensitivity, 87%
[20 of 23 patients]; specificity,
82% [27 of 33 patients]).
nn Patients without complex IMA-LA
type II endoleak in whom the
largest feeding and/or draining
artery is larger than 3.8 mm in
diameter and patients with a
complex IMA-LA type II endoleak
in whom the largest feeding and/
or draining artery is greater than
2.2 mm in diameter are at high
risk for aneurysm sac
enlargement.
Radiology: Volume 274: Number 3—March 2015  n  radiology.rsna.org 907
of the aneurysm sac from aortic side
branches (4). Typical sources of type
II endoleaks are the inferior mesen-
teric artery (IMA), one or more lum-
bar arteries (LAs), the median sacral
artery, or even accessory renal arteries
(5). A simple type II endoleak is filled
and drained via a single artery, whereas
a complex type II endoleak has multiple
inflow and/or outflow vessels (5). An
endoleak detected within 90 days after
EVAR is defined as an early endoleak,
and one detected after 90 days is de-
fined as a late endoleak (4).
The clinical importance of early type
II endoleaks, however, is still controver-
sial (6,7), and interventional therapy
of type II endoleaks remains challeng-
ing (8–10). Nearly half of them resolve
spontaneously within 6 months (tran-
sient), whereas a persistent endoleak
may be associated with further substan-
tial aneurysm sac enlargement (2,3,11).
Little is known about the factors that
help predict the natural course of an
early type II endoleak (12,13). It would
be beneficial to understand which fea-
tures on the first postoperative contrast
material–enhanced computed tomo-
graphic (CT) scan are associated with
significant aneurysm sac enlargement
during follow-up. This might be helpful
for selecting patients for endoleak em-
bolization, potentially even before en-
largement of the aneurysm.
We performed this study to deter-
mine CT features of early type II en-
doleaks associated with aneurysm sac
enlargement after EVAR of abdominal
aortic aneurysm.
Materials and Methods
Patients
Institutional review board approval
was not required by our hospital for
this retrospective study. To determine
Implication for Patient Care
nn In patients with CT features in-
dicative of a high risk for aneu-
rysm enlargement, a more inten-
sive surveillance and an earlier
intervention may be justified.
CT features of early type II endoleak
that were associated with substantial
aneurysm sac enlargement during fol-
low-up, radiology reports from 384
consecutive patients (Fig 1) who un-
derwent EVAR for infrarenal abdomi-
nal aortic aneurysm between Decem-
ber 2002 and December 2011 were
retrospectively reviewed (R.M.W. and
O.G., with 9 and 2 years of experience
in the interpretation of vascular CT
studies, respectively). Twenty-two of
the 384 patients (5.7%) had early type
I endoleak, 89 (23.2%) had early type
II endoleak, eight (2.1%) had early
type III endoleak, and 265 (69%) did
not have endoleak. Patients with type
I or type III endoleak and those with
fewer than 6 months of follow-up (33
of 384 patients, 8.6%) were excluded
(Fig 1). For further analyses, we iden-
tified 56 of the 384 patients (14.6%)
with early type II endoleak who had
undergone CT follow-up of at least 6
months (mean age 6 standard devia-
tion, 71 years 6 7.9; age range, 52–85
years). Seven of the 56 patients (12%)
were women (mean age, 79 years 6
3.7; age range, 72–83 years) and 49
(88%) were men (mean age, 70 years
6 7.8; age range, 52–85 years) (P ,
.001). EVAR was performed with the
Zenith device (Cook, Bloomington,
Ind) in 16 of the 56 patients (29%),
Published online before print
10.1148/radiol.14140284  Content codes:
Radiology 2015; 274:906–916
Abbreviations:
EVAR = endovascular aortic aneurysm repair
IMA = inferior mesenteric artery
IQR = interquartile range
LA = lumbar artery
OR = odds ratio
Author contributions:
Guarantors of integrity of entire study, R.M.W., S.S., O.G.,
K.P., C.S., W.A.W.; study concepts/study design or data
acquisition or data analysis/interpretation, all authors;
manuscript drafting or manuscript revision for important
intellectual content, all authors; manuscript final version
approval, all authors; literature research, R.M.W., S.S., W.U.;
clinical studies, R.M.W., S.S., O.G., P.K., W.A.W.; statistical
analysis, R.M.W., F.Z.; and manuscript editing, R.M.W., S.S.,
F.Z., W.U., K.P., C.S., W.A.W.
Conflicts of interest are listed at the end of this article.
VASCULAR AND INTERVENTIONAL RADIOLOGY: CT Features of Early Type II Endoleaks after Aneurysm Repair Müller-Wille et al

Figure 1 Patients with early type II endoleak

were commonly followed up at 6 and
12 months and then yearly thereafter.
All CT scans obtained during follow-up
were multiphase scans as described
earlier. Overall, images from 221 con-
trast-enhanced CT examinations were
available for further analysis. Diagnos-
tic angiography was usually performed
in patients undergoing endoleak embo-
lization (Fig 2).

Endoleak Features (Risk Factors)

All measurements for risk factor
analysis were performed with a sep-
arate graphical workstation by using
the digitally stored CT datasets (syngo.
via; Siemens, Forchheim, Germany).
Patients were evaluated in a random
order. Readers were blinded to the
results of aneurysm volume measure-
ments during follow-up. The following
potential risk factors were measured in
separate reading sessions.
Endoleak sources.—Evaluation of
endoleak sources was performed by
R.M.W. and W.A.W. (with 20 years
of experience in the interpretation of
vascular CT studies) in consensus by
using the arterial and delayed phase
CT images. The total number of feed-
Figure 1:  Flowchart of study population. A total of 384 patients underwent EVAR for infrarenal
ing and/or draining arteries of the ni-
abdominal aortic aneurysm. Fifty-six of 384 patients (14.6%) with early type II endoleak had
dus was identified, including the IMA,
data available from at least 6 months of follow-up CT (mean, 3.0 years). Twenty-three of 56
LAs, median sacral artery, and acces-
endoleaks (41%) were transient and 33 (59%) were persistent. Overall, 23 of 56 patients (41%)
showed aneurysm sac enlargement during follow-up. In two patients, early type II endoleak sory renal artery (Fig 3). A common
resolved spontaneously within 6 months after EVAR (transient type II endoleak); however, slight stem of LAs and a common stem of
aneurysm sac enlargement was detected during this time. the fourth LAs and the median sacral
artery were counted as one feeding
and/or draining artery. The diameter
the Anaconda device (Vascutek, Inch- 30 days after EVAR (median, 3.5 days; of each feeding and/or draining artery
innan, Scotland) in 15 (27%), the interquartile range [IQR], 2.0, 5.0 days; was measured on axial sections by
Excluder device (W. L. Gore & Asso- range, 1–23 days). All CT scans were ob- each of the readers, independently,
ciates, Flagstaff, Ariz) in 14 (25%), the tained by using helical CT scanners (So- with use of electronic calipers. For
Endologix device (Endologix, Irvine, matom Sensation 16 or Somatom Flash further analyses, we used the mean
Calif) in seven (12%), and the Endu- Dual Source; Siemens, Forchheim, Ger- value of the diameter measurements.
rant device (Medtronic, Minneapolis, many). Contrast-enhanced arterial phase In addition, interobserver variability
Minn) in four (7.1%). Patients’ med- images were generated during injection was calculated. The communicating
ical records were reviewed for cardio- of 90–120 mL of nonionic contrast ma- artery with the largest diameter was
vascular risk factors and the use of terial at a flow rate of 4 mL/sec by using defined as the largest feeding and/or
anticoagulant and/or antiplatelet drugs bolus tracking with a threshold of 100 draining artery. All endoleaks were
(S.S., final year medical student). HU. Delayed phase images were obtained classified into simple and complex type
70 seconds after the arterial phase scan. II endoleaks by consensus (Table 1). In
Imaging Studies For image analysis, we reconstructed addition, we measured the diameter of
For the evaluation of potential predictors, transverse and coronal plane images with the left and right internal iliac arteries
we used the digitally stored CT data- a section thickness of 3.0 mm and a sec- as a potential source of collateral flow
sets that were obtained within the first tion increment of 3.0 mm. to the LAs.

908 radiology.rsna.org  n Radiology: Volume 274: Number 3—March 2015

VASCULAR AND INTERVENTIONAL RADIOLOGY: CT Features of Early Type II Endoleaks after Aneurysm Repair Müller-Wille et al

Figure 2 cm2 6 0.22; range, 0.02–0.71 cm2).

Furthermore, we measured the atten-
uation of the contrast-enhanced aortic
lumen 1 cm above the proximal end of
the stent-graft (uniform circular region
of interest, 1.0 cm2). The relative nidus
attenuation was calculated by dividing
the nidus attenuation by the aortic lu-
men attenuation. Measurements were
obtained by R.M.W.

Aneurysm Volume Enlargement during

Follow-up (Endpoint)
The aneurysm volume was measured
on the first postoperative CT scan and
on every available follow-up CT scan by
O.G., who was blinded to the results
of the endoleak measurements. The
aneurysm sac volume was measured
by means of manual segmentation be-
tween the lowest renal artery and the
aortic bifurcation. All volumetric mea-
surements of the aneurysm sac were
performed in a random order. Accord-
ing to the reporting standards, a change
in the volume of the aneurysm sac of
5% or more during follow-up was con-
sidered relevant (14). The relative ni-
dus volume was calculated by dividing
the nidus volume by the aneurysm sac
Figure 2:  Images obtained before, during, and after transarterial nidus embolization in 76-year-old man. volume.
A, Axial contrast-enhanced CT scan (delayed phase) shows complex IMA-LA type II endoleak after em-
bolization. B, Diagnostic angiogram (lateral view) of nidus obtained after catheterization of IMA via Riolan Statistical Analysis
arcade. IMA, nidus, and lumbar outflow (LA) were visualized. C, Microcatheter with detachable tip (Apollo;
Normality was verified according to
Covidien/ev3, Irvine, Calif) was used to embolize nidus with 15 mL of ethylene vinyl alcohol copolymer
statistical parameters (mean, median,
(Onyx-18, Covidien/ev3). The detachable tip of the microcatheter was left in the Onyx cast. The trunk
of the IMA was finally closed with multiple coils (Concerto, Covidien/ev3). D, Control angiogram reveals skewness, and kurtosis) and visually by
occlusion of endoleak after embolization. means of Q-Q plots. Paired nonnormal-
ly distributed data were compared with
the Mann-Whitney U test. Interrater
Endoleak volume.—We measured follow-up examinations were available agreement was assessed with the intra-
the volume of the nidus on arterial and who had a persistent type II endoleak class correlation coefficient (intraclass
delayed phase images. A line was drawn that had not been treated during the 1st correlation coefficient [3,1], two-way
manually around the nidus to define year after EVAR. Volume measurements mixed, single measures). Aneurysm
the nidus outline (Fig 3, E). This out- of the nidus were obtained by R.M.W. sac enlargement in relation to a po-
line was drawn for each section, and Data from diameter measurements are tential predictor variable was analyzed
the software computed the volume of not reported because we found that, in by means of simple logistic regression
the nidus in cubic centimeters. To eval- all cases with multiple feeding and/or models that calculated odds ratios
uate endoleak volume changes during draining arteries, diameter measure- (ORs) and corresponding 95% confi-
follow-up, we looked for patients with ments were not reproducible owing to dence intervals. Then, we developed a
a persistent type II endoleak who did the complex three-dimensional shape of multivariate logistic regression model
not undergo endoleak treatment within the nidus. by adding all significant predictors of
the 1st year after EVAR and for whom Endoleak attenuation.—The atten- the simple logistic regression models to
contrast-enhanced CT scans from 6- uation of the nidus was measured on a multivariate model followed by a man-
and 12-month follow-up were available. transverse sections by using a circular ually backward elimination according
Therefore, we identified 16 patients in region of interest (greatest possible to the likelihood ratio test (cutoff, P ,
whom CT scans from 6- and 12-month circular region of interest; mean, 0.57 .05). A receiver operating characteristic

Radiology: Volume 274: Number 3—March 2015  n  radiology.rsna.org 909

VASCULAR AND INTERVENTIONAL RADIOLOGY: CT Features of Early Type II Endoleaks after Aneurysm Repair Müller-Wille et al

Figure 3

Figure 3:  Evaluation of morphologic characteristics of endoleak on first postoperative contrast-enhanced CT scan in 59-year-old man 4
days after EVAR. A–D, Four feeding and/or draining arteries were identified: IMA, right third LA, right fourth LA, and left fourth LA. E, F, Axial
CT scans show nidus in arterial phase. A line was drawn manually around nidus (arrow). This outline was drawn for each section, and the
nidus volume was computed with software. G, Volume-rendered images of endoleak (anterior view). Nidus volume was 5.72 cm3.

software (SPSS 21.0 [IBM, Chicago,

Table 1
Ill], SAS 9.3 [SAS Institute, Cary, NC],
Classification of Simple and Complex Type II Endoleaks and R 3.02 [www.r-project.org]).
Endoleak Type Definition Subtypes

Simple Single artery (to-and-fro) Simple LA,* simple IMA Results

Complex Two or more arteries Complex LA,*† complex IMA-LA*‡
  creating a circuit (flow-through) Endoleaks and Aneurysm Sac
Enlargement
* A common stem of LAs and a common stem of the fourth LAs and median sacral artery were counted as one artery.
†
Complex type II endoleak predominantly supplied by two or more LAs and not supplied by IMA, possibly supplied by additional
The mean follow-up time of the 56 pa-
median sacral artery or accessory renal artery. tients who met the inclusion criteria
‡
Complex type II endoleak predominantly supplied by the IMA and one or more LAs, possibly supplied by additional median was 3.0 years 6 2.0 (range, 0.6–7.7
sacral artery or accessory renal artery. years). The initial median postopera-
tive aneurysm volume in all patients
was 177 cm3 (IQR: 136–270 cm3; range:
79–474 cm3). In 23 of the 56 patients
curve of the model’s predicted proba- time was specified as autoregressive. (41%), the median aneurysm volume
bilities was plotted and the area under All reported P values are two sided, and increased from 162 cm3 (IQR: 137–275
the curve used to assess overall differ- P = .05 was considered the threshold cm3; range: 116–474 cm3) to 226 cm3
entiation of patients with and patients of statistical significance. The threshold (IQR: 162–343 cm3; range: 132–777
without enlargement of the aneurysm of significance of the posthoc pairwise cm3). Thirty-three of the 56 patients
volume. Longitudinal data were ana- comparisons of the rank-based linear (59%) showed no aneurysm sac en-
lyzed by means of rank-based linear mixed models was adjusted by using largement. The aneurysm volume was
mixed models. The correlation struc- Bonferroni adjustment. P , .006 was stable in 13 of the 56 patients (23%).
ture between the arterial and delayed considered indicative of a statistically In 20 patients (36%), the median aneu-
phase was specified as unstructured, significant difference. Each test had 1 rysm volume decreased from 208 cm3
whereas the correlation structure over df. Calculations were performed with (IQR: 169–329 cm3; range: 121–374

910 radiology.rsna.org  n Radiology: Volume 274: Number 3—March 2015

VASCULAR AND INTERVENTIONAL RADIOLOGY: CT Features of Early Type II Endoleaks after Aneurysm Repair Müller-Wille et al

cm3) to 152 cm3 (IQR: 115–192 cm3; Table 2

range: 101–354 cm3).
The early type II endoleak resolved Patient Characteristics
spontaneously within the first 6 months Aneurysm Volume during Follow-up*
in 23 of the 56 patients (41%). Thir-
No Enlargement Enlargement
ty-three of the 56 patients (59%) had
Variable (n = 33) (n = 23) OR† P Value
a persistent type II endoleak (Fig 1). A
persistent type II endoleak was signif- Mean age (y) 71.2 6 7.9 71.7 6 8.1 1.01 (0.94, 1.08) .80
icantly associated with aneurysm sac Sex 4.31 (0.76, 24.52)‡ .10‡
enlargement (21 of 33 patients with  M 31 (94) 18 (78)
persistent endoleak vs two of 23 pa-  F 2 (6) 5 (22)
tients with transient endoleak; OR = Cardiovascular risk factors
18.4; P , .001).   Current smoker 9 (27) 2 (9) 0.25 (0.05, 1.31) .10
Twenty of the 56 patients (36%) un-  Hypertension 24 (73) 16 (70) 0.86 (0.27, 2.77) .80
derwent treatment of their early type  Diabetes 9 (27) 4 (17) 0.56 (0.15, 2.11) .39
 Hyperlipidemia 11 (33) 12 (52) 2.18 (0.73, 6.50) .16
II endoleak, generally as a consequence
Anticoagulant and/or antiplatelet
of a persistent endoleak and aneurysm
  drugs
sac enlargement (median: 16.3 months;
  No therapy 3 (9) 2 (9) 0.95 (0.15, 6.21) .96
IQR: 6.8–31.7 months; range: 1–63
  Antiplatelet (monotherapy§) 18 (55) 13 (57) 1.08 (0.37, 3.17) .88
months after EVAR). The endoleak was   Antiplatelet (dual therapy||) 7 (21) 3 (13) 0.56 (0.13, 2.43) .44
initially treated with transarterial or   Antiplatelet and coumarin# 2 (6) 4 (17) 3.26 (0.54, 19.56) .20
translumbar endoleak embolization in  Coumarin# 3 (9) 1 (4) 0.46 (0.04, 4.67) .51
16 of the 56 patients (29%), with surgi- Aneurysm volume (cm3) 187 (136, 267)# 162 (127, 275)** 1.00 (1.00, 1.01) .88
cal ligation of the IMA in three patients Stent-graft
(5.4%), and with open surgical conver-  Manufacturer
sion in one patient (1.8%). Diagnostic   Zenith 9 (27) 7 (30) 1.17 (0.36, 3.77) .80
angiography before embolization re-   Anaconda 6 (18) 9 (39) 2.89 (0.86, 9.78) .09
vealed the sources of the endoleak that   Excluder 10 (30) 4 (17) 0.48 (0.13, 1.79) .28
had been identified on postoperative   Endologix 5 (15) 2 (9) 0.53 (0.09, 3.02) .48
CT scans in all cases.   Endurant 3 (9) 1 (4) 0.46 (0.04, 4.67) .51
 Configuration
Factors Associated with Aneurysm Sac   Bifurcated 32 (97) 23 (100) … …
Enlargement   Aortouni-iliac 1 (3) 0 (0) … …
There were no differences regarding * Except where indicated, continuous data are presented as means 6 standard deviations. Categoric data are given as numbers
sex, age, cardiovascular risk factors, of patients, with percentages in parentheses.
use of anticoagulant and/or antiplate- †
Numbers in parentheses are 95% confidence intervals.
let drugs, initial aneurysm volume, and ‡
Data indicate likelihood of aneurysm enlargement in women relative to that in men.
type of stent-graft devices between pa- §
Monotherapy with aspirin (100 mg).
tients with and patients without aneu- ||
Dual therapy with aspirin (100 mg) and clopidogrel (75 mg).
rysm sac enlargement (Table 2). #
Phenprocoumon (Marcumar; MEDA, Bad Homburg, Germany).
We identified 154 feeding and/ ** Numbers in parentheses are the IQR.
or draining arteries in the 56 patients
(mean, 2.8 arteries 6 1.3; range,
1–6) (Table 3). In 41 of the 56 pa-
tients (73%), the fourth LAs were in without aneurysm enlargement (mean, patients] vs 36% [12 of 33 patients];
connection to the nidus, followed by 3.5 arteries 6 1.3 [range, 1–6] vs 2.2 OR = 18.38; P , .001).
the IMA (33 of 56 patients, 59%), the arteries 6 1.1 [range, 1–5]); OR = The mean diameter of all feed-
third LAs (29 of 56 patients, 52%), 2.43; P = .002) (Table 3). Analyses of ing and/or draining arteries per pa-
and the second LAs (11 of 56 patients, the endoleak sources showed that the tient was 2.3 mm 6 0.7 (range, 1.1–
20%). Two of the 56 patients (3.6%) mean number of LAs was not differ- 4.0 mm) (Table 3). The interrater
with complex IMA-LA type II endole- ent between patients with and patients agreement was high, with an interclass
aks had an accessory renal artery in without an increase in aneurysm vol- correlation coefficient of 0.90. The
communication with the nidus. Patients ume (Table 3). The IMA was more of- mean diameter was significantly larger
with an increase in aneurysm volume ten a source of the type II endoleak in in patients with an increase in aneu-
had significantly more feeding and/ patients with than in patients without rysm volume than in those without an
or draining arteries than did patients aneurysm enlargement (91% [21 of 23 increase (mean, 2.5 mm 6 0.7 [range,

Radiology: Volume 274: Number 3—March 2015  n  radiology.rsna.org 911

VASCULAR AND INTERVENTIONAL RADIOLOGY: CT Features of Early Type II Endoleaks after Aneurysm Repair Müller-Wille et al

Table 3 mean diameter of LAs (Table 3). The

diameter of the largest feeding and/
CT Features of Early Type II Endoleak or draining vessels was, on average,
Aneurysm Volume during Follow-up* significantly larger in patients with
aneurysm enlargement than in those
No Enlargement Enlargement
without enlargement (mean, 3.4 mm
Variable (n = 33) (n = 23) OR† P Value
6 0.9 [range, 2.0–5.0 mm] vs 2.4 mm
Feeding and/or draining 6 0.7 [range, 1.1–3.8 mm], respec-
   arteries of the nidus tively; OR = 6.47; P , .001). In most
  All arteries patients with aneurysm enlargement
   Mean no. of arteries 2.2 6 1.1 3.5 6 1.3 2.43 (1.39, 4.24) .002 (19 of 23 patients, 83%), the IMA was
   Mean diameter (mm) 2.1 6 0.6 2.5 6 0.7 2.93 (1.16, 7.39) .023 the artery with the largest diameter of
   Diameter of the largest 2.4 6 0.7 3.4 6 0.9 6.47 (2.32, 18.06) ,.001 all feeding and/or draining arteries.
  artery (mm) The frequency of simple LA, simple
 IMA IMA, complex LA, and complex IMA-
   No. of arteries 12 (36) 21 (91) 18.38 (3.66, 92.34) ,.001
LA type II endoleak was 8.9% (five of
   Mean diameter (mm) 2.7 6 0.6 3.5 6 0.9 4.24 (1.26, 14.31) .020
56 patients), 8.9% (five of 56 patients),
 LAs
32% (18 of 56 patients), and 50% (28
   Mean no. of arteries 1.9 6 1.1 2.5 6 1.3 1.54 (0.96, 2.47) .076
of 56 patients), respectively. A complex
   Mean diameter (mm) 2.0 6 0.5 2.1 6 0.6 1.56 (0.57, 4.30) .39
IMA-LA type II endoleak was strongly
  Accessory renal arteries
   No. of arteries 0 (0) 2 (9) … …
associated with aneurysm sac enlarge-
   Mean diameter (mm) … 3.5 6 0.6 … … ment (OR = 20.8, P , .001) (Table 3).
  Combinations (subtypes) Patients with a complex IMA-LA type
  Simple LA 5 (15) 0 (0) … … II endoleak had the highest rate of an-
  Simple IMA 4 (12) 1 (4) 0.33 (0.03, 3.16) .34 eurysm volume increase (20 of 28 pa-
  Complex LA 16 (49) 2 (9) 0.10 (0.02, 0.50) .005 tients, 71%) (Fig 4). Only one of five
  Complex IMA-LA 8 (24) 20 (87) 20.8 (4.9, 88.9) ,.001 patients with a simple IMA type II en-
Volume doleak (20%) and two of 18 patients
 Nidus with a complex LA type II endoleak
   Arterial phase (cm3) 4.1 (2.1, 8.6)‡ 10.2 (4.1, 19.2)‡ 1.07 (1.01, 1.15) .040 (11%) had an aneurysm volume in-
   Delayed phase (cm3) 5.6 (2.3, 9.9)‡ 8.7 (3.4, 18.7)‡ 1.06 (0.99, 1.13) .077 crease. All five patients with a simple
  Nidus-to-aneurysm ratio LA type II endoleak (100%) and 11 of
   Arterial phase (%) 1.9 (0.8, 6.1)‡ 5.4 (2.1, 7.5)‡ 1.07 (0.97, 1.18) .20 the 18 patients with a complex LA type
   Delayed phase (%) 2.1 (1.1, 6.5)‡ 4.9 (2.0, 9.7)‡ 1.07 (0.97, 1.19) .17 II endoleak (61%) showed a decrease in
Attenuation the aneurysm sac volume.
 Nidus There was no significant difference
   Arterial phase (HU) 204 6 45 206 6 53 1.00 (0.99, 1.01) .89
between patients with and patients
   Delayed phase (HU) 110 6 27 110 6 14 1.00 (0.98, 1.03) .99
without aneurysm sac enlargement
  Aortic lumen
with regard to the patency and diam-
   Arterial phase (HU) 284 6 77 266 6 43 1.00 (0.99, 1.00) .29
eter of the right internal iliac artery
   Delayed phase (HU) 118 6 29 119 6 18 1.00 (0.98, 1.02) .96
(mean, 8.8 mm 6 1.6 [range, 0.5–1.2
  Nidus-to–aortic lumen ratio
   Arterial phase (%) 75.3 6 20.9 78.2 6 18.1 1.01 (0.98, 1.04) .59
mm] vs 8.7 mm 6 1.7 [range, 0.6–1.3
   Delayed phase (%) 95.1 6 18.2 93.7 6 8.9 0.99 (0.96, 1.03) .74 mm], respectively; OR = 0.96; P = .82)
and the left internal iliac artery (mean,
* Except where indicated, continuous data are presented as means 6 standard deviations. Categoric data are given as numbers 8.4 mm 6 1.7 [range, 0.5–1.3 mm] vs
of patients or arteries, with percentages in parentheses.
8.7 mm 6 1.8 [range, 0.6–1.3 mm];
†
Numbers in parentheses are 95% confidence intervals.
‡
OR = 1.10; P = .55).
Data are medians. Numbers in parentheses are the IQR (lower quartile [Q1], upper quartile [Q3]).
In all patients, the nidus was visi-
ble on arterial and delayed phase CT
images. The median nidus volume was
1.4–4.0 mm] vs 2.1 mm 6 0.6 [range, aneurysm enlargement than in those 6.4 cm3 (IQR: 2.4–12.8 cm3; range:
1.1–3.4 mm], respectively; OR = 2.93; without enlargement (mean, 3.5 mm 0.3–51.5 cm3) in the arterial phase and
P = .023). Separate analysis of the 33 6 0.9 [range, 2.2–5.0 mm] vs 2.7 mm 6.5 cm3 (IQR: 2.9–13.2 cm3; range: 0.1–
patients with a patent IMA showed 6 0.6 [range, 1.7–3.8 mm], respec- 56.8 cm3) in the delayed phase (P = .10).
that the diameter of the IMA was tively; OR = 4.24; P = .02). There was There was a significant difference in ni-
significantly greater in patients with no significant difference regarding the dus volume between patients with and

912 radiology.rsna.org  n Radiology: Volume 274: Number 3—March 2015

VASCULAR AND INTERVENTIONAL RADIOLOGY: CT Features of Early Type II Endoleaks after Aneurysm Repair Müller-Wille et al

patients without aneurysm enlargement Figure 4

in the arterial phase (median, 10.2 cm3
[IQR: 4.1–19.2 cm3; range: 0.5–51.5 cm]
vs 4.1 cm3 [IQR: 2.1–8.6 cm3; range:
0.3–39.3 cm3], respectively; OR = 1.07;
df = 1; P = .04). However, the results of
the delayed phase did not reach statisti-
cal significance (median, 8.7 [IQR: 3.4–
18.7 cm3; range, 1.7–56.8 cm3] vs 5.6
cm3 [IQR: 2.3–9.9 cm3; range, 0.1–45.6
cm3] for patients with and patients with-
out aneurysm enlargement, respectively;
OR = 1.06; P = .077). Patients with
complex IMA-LA type II endoleak had
the largest nidus volumes (Figs 5, 6).
The relative nidus volume was not sig-
nificantly different between patients
with and patients without aneurysm sac
enlargement (Table 3). In 16 of the 56
patients (29%) with an untreated per-
sistent type II endoleak during the 1st
year after EVAR (endoleak visible on the
first postoperative scan and on 6-month
and 12-month CT scans), the median ni-
dus volume decreased on arterial phase
Figure 4:  Bar charts show frequency of aneurysm sac enlargement in rela-
images but increased on delayed phase
tion to endoleak subtype. Patients with complex IMA-LA type II endoleak had
images during follow-up (Table 4).
the highest frequency of aneurysm sac enlargement during follow-up.
The mean nidus attenuation was
204.7 HU 6 47.8 (range, 109.0–319.0
HU) in the arterial phase and 110.2 HU Figure 5
6 22.6 (range, 66.0–180.0 HU) in the
delayed phase. There was no significant
difference between patients with and
patients without aneurysm enlargement
(Table 3). Compared with the attenua-
tion of the aortic lumen (nidus-to–aor-
tic lumen ratio), there was no signifi-
cant difference between patients with
and patients without aneurysm sac en-
largement (Table 3).
According to the multivariate logis-
tic regression model, the variables with
the strongest significance for aneurysm
sac enlargement were complex IMA-LA
type II endoleak (OR = 10.29; 95% con-
fidence interval: 2.15, 49.23; P = .004)
and the diameter of the largest feed-
ing and/or draining artery (OR = 4.55;
95% confidence interval: 1.38, 14.99; P
= .013). The logistic regression equation
logit (p) = 26.087 + 1.514 3 diameter
of the largest artery (in millimeters) +
2.331 3 complex IMA-LA (no = 0, yes =
1), where p is the probability of the pres- Figure 5:  Boxplots of nidus volume in relation to endoleak subtype. Patients with complex IMA-LA type II
ence of aneurysm sac enlargement, gave endoleak had, on average, the largest nidus volume. Horizontal line in box is median, top and bottom of box
an indication of patient risk for aneurysm are the IQR, whiskers are first and fourth quartiles.  = mild outliers,  = extreme outliers.

Radiology: Volume 274: Number 3—March 2015  n  radiology.rsna.org 913

VASCULAR AND INTERVENTIONAL RADIOLOGY: CT Features of Early Type II Endoleaks after Aneurysm Repair Müller-Wille et al

Figure 6 arteries and a significantly larger vol-

ume of the nidus in the arterial phase.
Complex IMA-LA type II endoleak and
the diameter of the largest feeding and/
or draining artery remained in the fi-
nal model as associated with a high risk
for aneurysm enlargement. According
to our multivariate logistic regression
model, we were now able to identify
patients at high risk and low risk for
aneurysm sac enlargement on the first
postoperative CT scan with a sensitivity
of 87% (20 of 23 patients) and spec-
ificity of 82% (27 of 33 patients). On
the basis of our data, a patient with no
complex IMA-LA type II endoleak (eg,
complex LA type II endoleak) in whom
the largest feeding and/or draining
artery was 2.0 mm in diameter might
have a risk of 5% for aneurysm sac en-
largement (low risk), and a patient with
a complex IMA-LA endoleak in whom
the largest feeding and/or draining
artery was 4.0 mm in diameter might
have a risk of 91% for aneurysm sac
enlargement (high risk).
This study is not directly compara-
ble to the investigations of Keedy et al
(12) and Timaran et al (13). The main
difference is the choice of study end-
points. Keedy et al (12) used interven-
tion of type II endoleak and Timaran et
al (13) the maximum aneurysm diame-
Figure 6:  Volume-rendered images in patients with different endoleak subtypes. A, Images in 61-year-old ter as study endpoints. In contrast, we
man with early type II endoleak from a single LA (common stem of fourth LAs and median sacral artery; used the aneurysm volume. It has been
nidus volume, 4.1 cm3 ). Endoleak was transient, and aneurysm volume decreased during the 1st year from clearly demonstrated that the aneurysm
332.7 cm3 to 251.2 cm3 (224.5%). B, Images in 62-year-old man with a complex LA type II endoleak volume is more reliable and useful in the
supplied by third and fourth LAs (nidus volume, 6.58 cm3 ). Endoleak was transient, and aneurysm volume evaluation of aneurysm sac changes af-
decreased during 3 years from 344.6 cm3 to 236.5 cm3 (231.4%). C, Images in 76-year-old man with a ter EVAR than the more frequently used
simple IMA type II endoleak (nidus volume, 1.46 cm3 ). Endoleak was transient, and the aneurysm sac shrank maximum aneurysm diameter (15–18).
during the 1st year from 193.6 cm3 to 110.4 cm3 (243.0%). D, Images in 79-year-old man with a complex Wever et al (15) demonstrated that the
IMA-LA type II endoleak supplied by the IMA and three LAs (nidus volume, 21.6 cm3 ). Endoleak persisted,
maximum aneurysm diameter failed to
and aneurysm volume increased from 348.4 cm3 to 393.8 cm3 (13.0%).
help detect an increase or decrease in
one-third of the cases compared with
sac enlargement during follow-up (Fig 3.8 mm in diameter and patients with volume measurements. Furthermore,
7). The area under the receiver operat- a complex IMA-LA type II endoleak in van Keulen et al (18) showed that 63%
ing characteristic curve for the predicted whom the largest feeding and/or drain- of the volume increase in patients with
probabilities was 0.90 (Fig 8). An optimal ing artery was larger than 2.2 mm in di- type II endoleaks was missed with
cutoff of the predicted probability was ameter are at high risk for aneurysm sac transverse diameter measurements and
found at P = .40, providing a sensitivity enlargement. 50% was missed with orthogonal diam-
of 87% (20 of 23 patients) and a speci- eter measurements.
ficity of 82% (27 of 33 patients) for the However, Timaran et al (13) showed
prediction of aneurysm sac enlargement. Discussion that the median number of feeding and/
Patients without complex IMA-LA type We found that patients with aneurysm or draining arteries was greater in the
II endoleak in whom the largest feeding enlargement had a significantly higher enlarging aneurysm group (three vs two
and/or draining artery was larger than number of feeding and/or draining arteries; P = .02). This is comparable to

914 radiology.rsna.org  n Radiology: Volume 274: Number 3—March 2015

VASCULAR AND INTERVENTIONAL RADIOLOGY: CT Features of Early Type II Endoleaks after Aneurysm Repair Müller-Wille et al

Table 4
Changes of Endoleak Volume during the 1st Year after EVAR in 16 Patients without Treatment
Endoleak Volume (cm3)* P Value
First Postoperative Scan 6-month Scan First Postoperative Scan
Phase First Postoperative Scan 6-month Scan 12-month Scan vs 6-month Scan vs 12-month Scan vs 12-month Scan

Arterial phase 10.8 (1.5, 17.0) 7.8 (1.9, 13.6) 7.0 (2.8, 13.3) .15 .35 .70
Delayed phase 9.5 (2.7, 18.9) 12.2 (4.5, 24.5) 14.3 (6.3, 31.0) .24 .15 .056

Note.—Data were analyzed with a rank-based linear mixed model. The results for fixed effects are as follows: type of phase, F1,15 = 22.89, P , .001; time, F2,30 = 0.77, P = .47; and interaction phase
time, F2,30 = 24.27, P , .001. Unadjusted P values for pairwise comparison (arterial phase vs delayed phase) are as follows: P = .25 for first postoperative scan, P , .001 for 6-month scan, and
P , .001 for 12-month scan; P , .006 was considered indicative of a statistically significant difference for all post hoc comparisons according to Bonferroni adjustment.
* Data are medians, with IQR (lower quartile [Q1], upper quartile [Q3]) in parentheses.

Figure 7

Figure 7:  A, Graph shows risk of aneurysm sac enlargement during follow-up as a function of whether the endoleak is a complex IMA-LA type II endoleak and ac-
cording to the diameter of the largest feeding and/or draining artery. Horizontal line shows optimal cutoff obtained at receiver operating characteristic curve analysis
for predicting aneurysm sac enlargement (P = .40). B, Graph shows distribution of patients with and patients without aneurysm sac enlargement.

our results. Keedy et al (12) reported
that the maximum diameter of the larg-
est communicating vessel was signifi-
cantly greater in the intervention group
(mean, 0.40 cm 6 0.11 vs 0.34 cm 6
0.11; P = .045). In our study, the diam-
eter of the largest feeding and/or drain-
ing artery was also a strong indicator
for aneurysm sac enlargement.
In contrast to other studies, we cate-
gorized the type II endoleaks into differ-
ent subtypes. Complex IMA-LA type II
endoleaks were strongly associated with
aneurysm volume increase, whereas the
aneurysm volume in patients with even
large complex LA endoleaks tends to
decrease. Accordingly, Lee et al (19)
observed that complex IMA-LA type II
endoleaks in particular had to be treated
during follow-up, whereas single-chan-
nel and complex LA type II endoleaks
tend to resolve spontaneously.
In contrast to others, we were not
able to accurately and reliably measure
the diameters of the nidus on axial CT
sections as described by Keedy et al
(12) and Timaran et al (13) in all cases
owing to the complex three-dimensional
shape of the nidus. However, Keedy et al
(12) identified a significant difference in
the transverse diameter of the endoleak
cavity between patients with interven-
tion and patients without intervention
(mean, 1.85 cm 6 1.01 vs 1.13 cm 6
0.83, respectively; P = .007). Timaran
et al (13) reported that a maximum ni-
dus diameter greater than 15 mm was
associated with a more than 10-fold
increased risk of aneurysm expansion.
Volumetric measurements of the nidus
in the arterial phase were performed by
Keedy and colleagues (12). The nidus
volume was greater in the intervention
group, but the difference lacked statisti-
cal significance (mean, 7.38 cm3 6 8.30

Radiology: Volume 274: Number 3—March 2015  n  radiology.rsna.org 915

VASCULAR AND INTERVENTIONAL RADIOLOGY: CT Features of Early Type II Endoleaks after Aneurysm Repair Müller-Wille et al

Figure 8 In conclusion, the strongest indi-
cators for aneurysm sac enlargement
during follow-up are a complex IMA-LA
type II endoleak and the diameter of
largest feeding and/or draining artery.
Disclosures of Conflicts of Interest: R.M.W.
disclosed no relevant relationships. S.S. dis-
closed no relevant relationships. F.Z. disclosed
no relevant relationships. W.U. disclosed no rel-
evant relationships. O.G. disclosed no relevant
relationships. K.P. disclosed no relevant rela-
tionships. P.K. disclosed no relevant relation-
ships. C.S. disclosed no relevant relationships.
W.A.W. disclosed no relevant relationships.
ment of persistent type 2 endoleak after
endovascular aneurysm repair. J Vasc Surg
2012;56(3):630–636.
10. Müller-Wille R, Wohlgemuth WA, Heiss P,
et al. Transarterial embolization of type II
endoleaks after EVAR: the role of ethylene
vinyl alcohol copolymer (Onyx). Cardiovasc
Intervent Radiol 2013;36(5):1288–1295.
11. Nolz R, Teufelsbauer H, Asenbaum U, et al.
Type II endoleaks after endovascular repair
of abdominal aortic aneurysms: fate of the
aneurysm sac and neck changes during long-
term follow-up. J Endovasc Ther 2012;19(2):
193–199.

References
Figure 8:  Graph shows receiver operating charac-
teristic curve of predicted probabilities according to
multivariate logistic regression model on aneu-
rysm sac enlargement. Point estimates at curve
indicate cutoff (sensitivity [sens] of 0.87 [20 of
23 patients], specificity [spec] of 0.82 [27 of 33
patients], P = .40). AUC = area under the receiver
operating characteristic curve.
vs 3.80 cm3 6 5.17; P = .08). In con-
trast to these findings, we found signif-
icant differences of nidus volume in the
arterial phase. It should be mentioned
that there was a similar effect in the de-
layed phase that did not reach statistical
significance. In an experimental study,
Timaran et al (20) showed that the an-
eurysm wall pressure increases propor-
tionally with increasing endoleak vol-
umes up to 3 cm3. For larger endoleak
volumes, the pressure at the aneurysm
wall was dependent only on the mean
arterial pressure.
The major limitation of our study is
its retrospective design. There might be
a selection bias because 33 patients were
excluded owing to a lack of CT follow-up
for more than 6 months. Furthermore,
we only included patients with early type
II endoleak; thus, our findings cannot be
applied to patients with late-onset type
II endoleaks. The sensitivity and spec-
ificity of our final model were based
on thresholds developed from our own
population and therefore may overesti-
mate performance. Subcategorization of
endoleaks was performed by consensus
review without documentation of inter-
observer variability.
1. Prinssen M, Verhoeven EL, Buth J, et al.
A randomized trial comparing conventional
and endovascular repair of abdominal aor-
tic aneurysms. N Engl J Med 2004;351(16):
1607–1618.
2. Higashiura W, Greenberg RK, Katz E, Gei-
ger L, Bathurst S. Predictive factors, mor-
phologic effects, and proposed treatment
paradigm for type II endoleaks after repair
of infrarenal abdominal aortic aneurysms. J
Vasc Interv Radiol 2007;18(8):975–981.
3. Jones JE, Atkins MD, Brewster DC, et al.
Persistent type 2 endoleak after endovas-
cular repair of abdominal aortic aneurysm
is associated with adverse late outcomes. J
Vasc Surg 2007;46(1):1–8.
4. Veith FJ, Baum RA, Ohki T, et al. Nature and
significance of endoleaks and endotension:
summary of opinions expressed at an inter-
national conference. J Vasc Surg 2002;35(5):
1029–1035.
5. Golzarian J, Valenti D. Endoleakage after
endovascular treatment of abdominal aortic
aneurysms: diagnosis, significance and treat-
ment. Eur Radiol 2006;16(12):2849–2857.
6. Karthikesalingam A, Thrumurthy SG, Jack-
son D, et al. Current evidence is insufficient
to define an optimal threshold for interven-
tion in isolated type II endoleak after endo-
vascular aneurysm repair. J Endovasc Ther
2012;19(2):200–208.
7. van Marrewijk CJ, Fransen G, Laheij RJ,
Harris PL, Buth J; EUROSTAR Collabo-
rators. Is a type II endoleak after EVAR
a harbinger of risk? Causes and outcome
of open conversion and aneurysm rupture
during follow-up. Eur J Vasc Endovasc Surg
2004;27(2):128–137.
8. Aziz A, Menias CO, Sanchez LA, et al. Out-
comes of percutaneous endovascular interven-
tion for type II endoleak with aneurysm ex-
pansion. J Vasc Surg 2012;55(5):1263–1267.
9. Abularrage CJ, Patel VI, Conrad MF,
Schneider EB, Cambria RP, Kwolek CJ. Im-
proved results using Onyx glue for the treat-
12. Keedy AW, Yeh BM, Kohr JR, Hiramoto JS,
Schneider DB, Breiman RS. Evaluation of
potential outcome predictors in type II en-
doleak: a retrospective study with CT angi-
ography feature analysis. AJR Am J Roent-
genol 2011;197(1):234–240.
13. Timaran CH, Ohki T, Rhee SJ, et al. Pre-
dicting aneurysm enlargement in patients
with persistent type II endoleaks. J Vasc
Surg 2004;39(6):1157–1162.
14. Chaikof EL, Blankensteijn JD, Harris PL, et
al. Reporting standards for endovascular aor-
tic aneurysm repair. J Vasc Surg 2002;35(5):
1048–1060.
15. Wever JJ, Blankensteijn JD, Th M Mali WP,
Eikelboom BC. Maximal aneurysm diameter
follow-up is inadequate after endovascular
abdominal aortic aneurysm repair. Eur J
Vasc Endovasc Surg 2000;20(2):177–182.
16. Prinssen M, Verhoeven EL, Verhagen HJ,
Blankensteijn JD. Decision-making in fol-
low-up after endovascular aneurysm repair
based on diameter and volume measure-
ments: a blinded comparison. Eur J Vasc
Endovasc Surg 2003;26(2):184–187.
17. Bargellini I, Cioni R, Petruzzi P, et al. En-
dovascular repair of abdominal aortic an-
eurysms: analysis of aneurysm volumetric
changes at mid-term follow-up. Cardiovasc
Intervent Radiol 2005;28(4):426–433.
18. van Keulen JW, van Prehn J, Prokop M,
Moll FL, van Herwaarden JA. Potential value
of aneurysm sac volume measurements in
addition to diameter measurements after
endovascular aneurysm repair. J Endovasc
Ther 2009;16(4):506–513.
19. Lee JT, Aziz IN, Lee JT, et al. Volume regres-
sion of abdominal aortic aneurysms and its
relation to successful endoluminal exclusion.
J Vasc Surg 2003;38(6):1254–1263.
20. Timaran CH, Ohki T, Veith FJ, et al. Influence
of type II endoleak volume on aneurysm wall
pressure and distribution in an experimental
model. J Vasc Surg 2005;41(4):657–663.

916 radiology.rsna.org  n Radiology: Volume 274: Number 3—March 2015