Interventional Cardiology

Mechanisms of Very Late Drug-Eluting Stent Thrombosis

Assessed by Optical Coherence Tomography
Masanori Taniwaki, MD; Maria D. Radu, MD, PhD; Serge Zaugg, MSc;
Nicolas Amabile, MD, PhD; Hector M. Garcia-Garcia, MD, PhD; Kyohei Yamaji, MD, PhD;
Erik Jørgensen, MD, DMSc; Henning Kelbæk, MD, DMSc; Thomas Pilgrim, MD;
Christophe Caussin, MD; Thomas Zanchin, MD; Aurelie Veugeois, MD;
Ulrik Abildgaard, MD, DMSc; Peter Jüni, MD; Stephane Cook, MD;
Konstantinos C. Koskinas, MD, MSC; Stephan Windecker, MD; Lorenz Räber, MD, PhD

Background—The pathomechanisms underlying very late stent thrombosis (VLST) after implantation of drug-eluting
stents (DES) are incompletely understood. Using optical coherence tomography, we investigated potential causes of this
adverse event.
Methods and Results—Between August 2010 and December 2014, 64 patients were investigated at the time point of VLST as
part of an international optical coherence tomography registry. Optical coherence tomography pullbacks were performed
after restoration of flow and analyzed at 0.4 mm. A total of 38 early- and 20 newer-generation drug-eluting stents were
suitable for analysis. VLST occurred at a median of 4.7 years (interquartile range, 3.1–7.5 years). An underlying putative
cause by optical coherence tomography was identified in 98% of cases. The most frequent findings were strut malapposition
(34.5%), neoatherosclerosis (27.6%), uncovered struts (12.1%), and stent underexpansion (6.9%). Uncovered and
malapposed struts were more frequent in thrombosed compared with nonthrombosed regions (ratio of percentages, 8.26;
95% confidence interval, 6.82–10.04; P<0.001 and 13.03; 95% confidence interval, 10.13–16.93; P<0.001, respectively).
The maximal length of malapposed or uncovered struts (3.40 mm; 95% confidence interval, 2.55–4.25; versus 1.29 mm;
95% confidence interval, 0.81–1.77; P<0.001), but not the maximal or average axial malapposition distance, was greater
in thrombosed compared with nonthrombosed segments. The associations of both uncovered and malapposed struts with
thrombus were consistent among early- and newer-generation drug-eluting stents.
Downloaded from http://ahajournals.org by on January 30, 2023

Conclusions—The leading associated findings in VLST patients in descending order were malapposition, neoatherosclerosis,
uncovered struts, and stent underexpansion without differences between patients treated with early- and new-generation
drug-eluting stents. The longitudinal extension of malapposed and uncovered stent was the most important correlate of
thrombus formation in VLST.   (Circulation. 2016;133:650-660. DOI: 10.1161/CIRCULATIONAHA.115.019071.)
Key Words: drug-eluting stents ◼ stents ◼ thrombosis ◼ tomography, optical coherence

T he advent of coronary stents in conjunction with potent

adjunctive medical treatment has resulted in substantial
improvements in the efficacy and safety of percutaneous coro-
nary interventions (PCIs).1 Although infrequent, stent throm-
bosis remains an important concern because of its sequelae,
including myocardial infarction and death in up to 80%.2,3
Early stent thrombosis is largely independent of stent type
and mainly related to procedural variables, including major
edge dissections and stent underexpansion.4 Conversely, the
mechanisms underlying very late stent thrombosis (VLST),
occurring beyond 1 year after drug-eluting stent (DES)
implantation, remain poorly defined, and the translation of
mechanistic insights into therapeutic approaches is unsatis-
factory. Despite an attenuation of the risk for VLST with the
use of newer-generation DES, which is similar to that of bare
metal stents, the accumulated long-term risk is still notable.
Clinical Perspective on p 660
Because of its high resolution (10–20 μm), optical coher-
ence tomography (OCT) has become the imaging modality
of choice for the in vivo assessment of stent failures, includ-
ing VLST.5 As a result of the infrequent encounter of this

Received August 17, 2015; accepted January 5, 2016.

From Department of Cardiology, Bern University Hospital, Switzerland (M.T., K.Y., T.P., T.Z., K.C.K., S.W., L.R.); Department of Cardiology, The Heart
Center, Rigshospitalet, Copenhagen, Denmark (M.D.R., E..J.); Clinical Trials Unit, Bern University, Bern, Switzerland (S.Z.) Applied Health Research
Centre (AHRC), Li Ka Shing Knowledge Institute of St. Michael’s Hospital, and Department of Medicine, University of Toronto, Canada (P.J.); Department
of Cardiology, Institut Mutualiste Montsouris, Paris, France (N.A., C.C., A.V.); Interventional Cardilogy, Washington Hospital Center, Washington, DC
(H.M.G.-G.); Department of Cardiology, Roskilde Hospital, Denmark (H.K.); Department of Cardiology, Copenhagen University Hospital Gentofte,
Gentofte, Denmark (U.A.); and Department of Cardiology, Fribourg University and Hospital, Switzerland (S.C.).
The online-only Data Supplement is available with this article at http://circ.ahajournals.org/lookup/suppl/doi:10.1161/CIRCULATIONAHA.
115.019071/-/DC1.
Correspondence to Lorenz Räber, MD, PhD, Department of Cardiology, Bern University Hospital, 3010 Bern, Switzerland. E-mail lorenz.raeber@insel.ch
© 2016 American Heart Association, Inc.
Circulation is available at http://circ.ahajournals.org DOI: 10.1161/CIRCULATIONAHA.115.019071

650
 Taniwaki et al   OCT in Very Late DES Thrombosis   651
complication and the technical challenges of performing intra-
coronary imaging in the acute course of stent thrombosis, pre-
vious OCT studies were limited to single-center reports with
small patient numbers and inclusion of a substantial propor-
tion of bare metal stents, resulting in discordant reports on the
relative contribution of potential mechanisms.6–10 Although
autopsy studies suggest substantial differences in the vascular
healing response between early-generation and new-genera-
tion DES,11 in vivo investigation of mechanisms underlying
VLST according to DES type is limited. Although histopa-
thology and OCT have identified mechanical factors and focal
stent abnormalities at the strut level,6,8 the spatial pattern of
these abnormalities in stented segments with compared with
those without thrombosis has not been reported; such evi-
dence could be meaningful for the purpose of optimizing stent
deployment to prevent stent thrombosis and to guide therapeu-
tic approaches at the time of VLST.
Against this background, the purpose of this study was
to investigate the mechanisms underlying VLST as assessed
by OCT in the largest cohort of patients treated with early-
generation and newer-generation DES in the framework of an
international OCT registry.
Methods
Patient Population
Between August 2010 and December 2014, 85 patients presenting
with definite VLST according to the Academic Research Consortium
criteria12 underwent OCT at the time of VLST at 4 centers (Bern,
Switzerland; Fribourg, Switzerland; Copenhagen, Denmark; and
Downloaded from http://ahajournals.org by on January 30, 2023
Paris, France). A total of 7 participating operators of each center
(T.P., S.W., and L.R. in Bern; E.J. and H.K. in Copenhagen; S.C. in
Fribourg; NA and C.C. in Paris) attempted to include consecutive
VLST patients referred for urgent revascularization. The investiga-
tors in Paris contributed to this study before their participation in
the French Morphological Parameters Explaining Stent Thrombosis
assessed by OCT registry. Exclusion criteria were hemodynamic
instability precluding OCT imaging and the inability of the OCT
catheter to cross the region of interest. Patients with bare metal stents
(n=24) were excluded from the analysis. Figure 1 shows the patient
flow. Among the studied stents, VLST included early-generation DES
(sirolimus- and paclitaxel-eluting stents) or new-generation DES
(zotarolimus-, everolimus-, and biolimus-eluting stents). Participants
provided written informed consent for the purpose of this analysis,
and the protocol was approved by the local ethics committees.
OCT Acquisition and Analyses
OCT images were acquired with commercially available time-domain
M2 and M3 systems and the frequency-domain C7 system from
LightLab/St. Jude (Westford, MA) and Lunawave (Terumo Corp,
Tokyo, Japan). Frame rates were 15.6, 100, 160, and 180 frames per
second, and pullback speed ranged from 2 to 40 mm/s, depending on
the console used. To minimize the possibility of iatrogenic disruption
of in-stent tissue, OCT was performed after wire crossing in cases
of Thrombolysis in Myocardial Infarction (TIMI) grade 2 or greater
flow; thrombus aspiration and, if necessary, balloon angioplasty were
performed before OCT only in cases with flow less than TIMI grade
2 after wire crossing, as previously described. OCT pullbacks were
assessed offline with a proprietary software (QCU-CMS Medis,
Leiden, the Netherlands) at the Bern imaging core laboratory. OCT
analyses were performed as previously described.13–15 Stented seg-
ments were analyzed at 0.4-mm intervals by an independent analyst
(M.T.) blinded to time point after DES implantation. All pullbacks
were validated by a second assessor (L.R.). To evaluate the relation-
ship between focal structural abnormalities and VLST, we divided the
Figure 1. Study flowchart. BMS indicates bare metal stent; DES,
drug-eluting stent; and VLST, very late stent thrombosis.
stented portion of the region of interest according to the presence and
absence of intracoronary thrombus. The latter was identified as any
irregular mass inside the lumen or attached to the neointima with a
sharp border and various degrees of attenuation. When this extended
>3 consecutive frames, it was called a thrombus region, and the end of
the region was demarcated by >3 consecutive frames without throm-
bus. The thrombus-free regions were used as control. Lumen and stent
area definitions are provided in the online-only Data Supplement.
Neointimal thickness was measured from the midportion of the
endoluminal surface of the struts to the lumen border, and struts were
defined in the absence of a homogeneous coverage by neointima. If
the apposed struts were covered by irregular tissue (eg, thrombus)
with no neointimal layer in between, struts were categorized as
uncovered (Figure I in the online-only Data Supplement). If a neo-
intimal layer underneath the thrombus could not be excluded, struts
were considered covered. In the presence of a highly attenuating layer
of thrombus, struts were categorized as not analyzable.
As described previously,15 malapposition (or incomplete stent
apposition [ISA]) was identified in struts when the distance between
the midpoint of the endoluminal strut surface and the lumen contour
exceeded the actual strut thickness of the individual DES plus 10 μm
(to account for the minimal axial resolution). If the lumen contour
could not be detected because of attenuating thrombus but stent struts
were visible, malapposition was not assessed, and the strut was con-
sidered unclassifiable. The area between the endoluminal midpoint of
the struts and the vessel wall was measured as ISA area, and the neo-
intimal area was measured as follows: stent area−(lumen area−ISA
area). Stent strut maps were computed with the x axis representing
the length of the stent (millimeters) and the y axis representing the
circumference (0°–360°).
Neoatherosclerosis was defined as the presence of a fibroath-
eroma (either thin-cap fibroatheroma with a cap thickness ≤65μm
or thick-cap fibroatheroma with a cap thickness >65 μm) or fibro-
calcific plaque within the neointima, as specified in the online-only
Data Supplement. To determine the intraobserver and interobserver
reproducibility of the stent strut assessment (coverage), 10 pullbacks
were randomly chosen and analyzed by 2 assessors at 2 time points (1
month apart), and then the Cohen κ was calculated.
Three-dimensional reconstruction of all cases was systematically
performed to visualize eventual stent fractures.
Assessment of the Primary Cause of VLST
The primary cause of the VLST was independently assessed by 5
investigators (M.T., M.R., H.M.G.G., N.A., and L.R.). In case of >1
plausible mechanism, investigators ranked the cause from 1 (most
plausible) to 3 (least plausible). Selectable causes included malap-
position (stent struts visually malapposed in relation to the lumen),
uncovered struts (not covered by neointima), bifurcation stenting
(thrombus at the bifurcation site in case of a stent in a main and side
branch), stent fracture (discontinuity of the stent contour), evagina-
tion (outward bulging of the vessel wall between stent struts more
than one fourth of lumen diameter), neoatherosclerosis (lipid or
 652  Circulation  February 16, 2016
calcific pool in the neointima), restenosis without neoatherosclerosis
(neointimal hyperplasia >70% by visual estimate), stent underexpan-
sion (minimal stent area <50% of the average stent area by visual
estimate), edge dissection (flap in the context of thrombus at the stent
edge), and plaque progression at the stent edge segment (ruptured
fibroatheroma at the stent edge). The predominant feature (ranking 1)
was defined as the most prevalent in direct contact with any thrombus.
A final consensus meeting was held to determine the final plausible
cause of VLST, as reported in Figure 2.
Statistical Analyses
We report frame-, segment-, and stent-level analyses. In frame-level
analyses, data from 1 stent per patient with OCT outcomes from
multiple frames were analyzed. Thrombus and control regions were
defined in each stent. Associations of region-type with OCT out-
comes were estimated with bayesian linear or logistic mixed models
based on vague priors. Random intercepts and slopes were included
at the patient level to account for the nonindependence of measure-
ments from the same patient, implicitly assigning analytical weights
to patients according to their contribution in terms of statistical preci-
sion. These models allow for between-patient variability in overall
outcome value and in strength of association. Stratified analyses used
pooled estimates of random effects and residual variances as in the
main analyses (details are given in the online-only Data Supplement).
Bayesian mixed models were based on vague priors and posterior
distributions estimated with Markov Chain Monte Carlo simulations
(online-only Data Supplement). We reported medians of marginal
posterior distributions with 95% credibility intervals and Bayesian
analogs to P values calculated from posterior distributions, which can
be interpreted similarly to conventional confidence intervals and P
values when vague priors were used, as is the case here. The segment-
and stent-level analyses, convergence diagnostics, and software used
are explained in the online-only Data Supplement. Two-sided P val-
ues are reported throughout.
Downloaded from http://ahajournals.org by on January 30, 2023
Results
Baseline Characteristics
Patient characteristics at the time of VLST are summarized
in Table 1. Among 58 patients included in this analysis, 38
patients (66%) had received an early-generation DES, and
20 patients (34%) had received new-generation DES (Table
I in the online-only Data Supplement). The median and mean
Figure 2. Frequency of the single or highest-ranked (most
plausible) mechanism of very late stent thrombosis in the 58
analyzed stents. TCFA indicates thin-cap fibroatheroma.
time from the index procedure to VLST was 4.7 years (inter-
quartile range, 3.1–7.5 years) and 5.2±2.8 years, respectively.
Clinical presentation of the VLST was most commonly ST-­
segment–elevation myocardial infarction (n=45, 77.6%) with
TIMI grade 0 flow (n=39, 67.2%).
Causes of VLST at the Lesion Level
The underlying mechanism in each individual case as iden-
tified by a consensus among 5 investigators is presented in
Figure 2. The majority of patients (n=32, 55.2%) exhibited
≥2 coexisting putative causes of the VLST, whereas a single
cause was identified in 43.1% of patients (n=25), and in only
1 patient, the cause of VLST could not be identified despite
sufficient image quality. When the patients with multiple
underlying findings were considered, the causes of VLST in
descending order were the combination of malapposition and
uncovered struts (29.3%); neoatherosclerosis alone (25.9%);
the combination of malapposition, uncovered struts, and stent
underexpansion (10.3%); and the combination of malapposi-
tion with stent underexpansion (5.2%; Table II in the online-
only Data Supplement). Considering only the single or (in case
of multiple causes) the highest-ranked mechanism for VLST,
the leading cause remained malapposition (34.5%), followed
by neoatherosclerosis (27.6%), uncovered struts (12.1%),
stent underexpansion (6.9%), and other less frequent causes
(Figure 2). In patients without antiplatelet therapy, malapposi-
tion remained the leading primary mechanism (Table III in the
online-only Data Supplement).
The κ value for the agreement among the 5 assessors
(before the consensus meeting) with respect to the highest-
ranked reason was moderate (κ=0.59) and for the assessment
of various causes independently of the ranking was good
(κ=0.72).
OCT Findings in Thrombotic Versus Control
Regions
OCT findings from the detailed analysis of the regions with
thrombus compared with those without (controls) are summa-
rized in Tables 2 through 4. A total of 11 742 of 12 661 struts in
the thrombotic region were analyzable (92.7%), as were a total
of 21 152 of 21 315 (98.7%) in the nonthrombotic region. This
includes 43 and 121 struts located over side branches that were
not analyzed in Table 3. The intraobserver variability for the
assessment of stent strut coverage was κ=0.883 and κ=0.886
and the interobserver variability was κ=0.758 and κ=0.857,
respectively, for neoatherosclerosis. In frame-level analyses
(Table 2), regions with versus without thrombus had higher
lumen eccentricity and a smaller stent area and average and
minimum stent diameter. Although the neointimal area tended
to be smaller in thrombotic regions, there was no difference in
terms of mean and maximal neointimal thickness. The mean
and maximal ISA distance and area were similar between
thrombus-containing and control regions when frames with
malapposed struts were compared. Notably, regions with
thrombus were characterized by a 8-times-higher proportion of
apposed uncovered struts and an 13-fold higher proportion of
malapposed struts (Table 3). In stents with neoatherosclerosis
(n=19), fibroatheromas were twice as frequent in the thrombus-
containing region compared with the nonthrombotic reference
 Taniwaki et al   OCT in Very Late DES Thrombosis   653

Table 1.  Patient Characteristics Table 1.  Continued

Variable All Patients Variable All Patients
Index PCI patient characteristics Treatment of VLST, n (%)
 Age at index PCI, y 59.45±11.06  No further treatment 3 (5.17)
 Male sex, n (%) 50 (86.21)  Balloon dilatation 28 (48.28)
 Body mass index, kg/m2 27.18±4.18  Bare metal stent 3 (5.17)
 Arterial hypertension, n (%) 34 (58.62)  Drug-eluting stent 23 (39.66)
 Family history of CAD, n (%) 18 (31.03)  CABG 1 (1.72)
 Current smoker, n (%) 23 (39.66) Values are reported as mean±SD when appropriate. CAD indicates coronary
 Dyslipidemia, n (%) 40 (68.97) artery disease; NSTEMI, non–ST-segment–elevation myocardial infarction; OCT,
 Diabetes mellitus, n (%) 12 (20.69) optical coherence tomography; PCI, percutaneous coronary intervention; STEMI,
ST-segment–elevation myocardial infarction; TIMI, Thrombosis in Myocardial
 Left ventricular ejection fraction, % 52.82±10.33 Infarction; and VLST, very late stent thrombosis.
Clinical presentation at index PCI, n (%)
 Stable angina 23 (39.66) region (Table 3). The 2-dimensional stent strut maps in Figure
 Unstable angina 6 (10.34) 3 provide a visual overview of the correlation between throm-
 NSTEMI 10 (17.24) bus and accumulated adverse stent strut characteristics (malap-
 STEMI 19 (32.76) position and uncovered struts or neoatherosclerosis).
Index PCI procedural characteristics In a stent-level analysis comparing the entire thrombotic
 Stent in VLST lesion, n (%)
and nonthrombotic regions, areas with thrombus had a lower
minimal stent area, lower stent expansion, and similar ISA
  Paclitaxel-eluting stent 10 (17.24)
area and ISA distance but a greater maximal length of con-
  Sirolimus-eluting stent 28 (48.28)
secutive uncovered struts or consecutive malapposition (Table
  Zotarolimus-eluting stent 3 (5.17) 4). The confidence intervals of the maximal length of malap-
  Everolimus-eluting stent 11 (18.97) position and uncoverage indicate that a longitudinal extension
  Biolimus-eluting stent 6 (10.34) of malapposed struts of >0.99 mm and for uncovered struts
 Stent overlap, n (%) 14 (24.14) of >1.63 mm was associated with very late thrombus forma-
tion. Figure 4 shows an analysis of 1.2-mm-long segments
Downloaded from http://ahajournals.org by on January 30, 2023

 Mean stent diameter, mm 3.02±0.44

 Total stent length, mm 28.05±16.84
indicating the percentage of segments with thrombus in rela-
tion to the proportion of uncovered or malapposed struts per
Characteristics at time of stent thrombosis
segment, suggesting a close association of thrombosis with
 Age at stent thrombosis, y 64.67±10.84
increasing extent of malapposed or uncovered stent struts.
 Time from index PCI to stent thrombosis, d 1906.36±1005.3 Figure 5 shows representative OCT cross sections or 3-dimen-
 Antiplatelet medications at time of VLST, n (%) sional reconstructions illustrating the most frequent causes of
  Aspirin only 34 (58.6) VLST and the corresponding stent strut maps.
  Clopidogrel only 3 (5.2)
  Aspirin and clopidogrel 14 (24.1) VLST in Early- Versus New-Generation DES
To account for the substantial difference in time from the index
  No antiplatelet therapy 5 (8.6)
PCI to the occurrence of VLST in early-generation compared
Clinical presentation at time of VLST, n (%)
with new-generation DES (6.5±2.6 versus 3.1±1.5 years;
 Stable angina 1 (1.72) P<0.001), we stratified early-generation DES according to the
 Unstable angina 3 (5.17) time interval since the index PCI above or below the respective
 NSTEMI 9 (15.52) median duration for all early-generation DES (6 years). Figure 6
 STEMI 45 (77.59) shows OCT findings in regions with and without thrombus in
TIMI flow grade at time of VLST, n (%) relation to DES generation. The association of malapposition
 0 39 (67.24) with stent thrombosis appeared similar among the 3 groups. The
association of uncovered struts with the occurrence of thrombo-
 1 2 (3.45)
sis was significant in all 3 DES groups but less pronounced in
 2 3 (5.17)
early-generation DES with “later VLST” and most pronounced
 3 14 (24.14) in new-generation DES. Of note, no difference in the cause of
Treatment before OCT, n (%) VLST at lesion level was noted between early- and new-gener-
 None 14 (24.14) ation DES (Table II in the online-only Data Supplement).
 Thrombus aspiration 20 (34.48)
 Balloon dilatation 15 (25.86) Discussion
 Thrombus aspiration and balloon dilatation 9 (15.52) The principal findings of this study, the largest cohort and first
multicenter analysis to date to assess the mechanisms of VLST
(Continued )
after DES implantation using OCT, can be summarized as
 654  Circulation  February 16, 2016

Table 2.  Continuous, Frame-Level OCT Findings in Regions With Versus Without Thrombus
Full Stent Region, Thrombus Region, Control Region, Difference or Ratio,*
Median (95% CI) Median (95% CI) Median (95% CI) Median (95% CI) P Value†
Patients (frames) analyzed, n 58 (1428) 58 (2338) Difference
 Lumen area, mm2 5.78 (5.11 to 6.5) 5.79 (4.98 to 6.6) 5.64 (5.04 to 6.24) 0.15 (−0.43 to 0.72) 0.602
 Average lumen diameter, mm 2.62 (2.45 to 2.77) 2.58 (2.4 to 2.76) 2.6 (2.46 to 2.74) −0.02 (−0.14 to 0.1) 0.718
 Minimal lumen diameter, mm 2.37 (2.22 to 2.51) 2.29 (2.12 to 2.47) 2.37 (2.24 to 2.5) −0.08 (−0.19 to 0.04) 0.190
 Maximal lumen diameter, mm 2.87 (2.7 to 3.05) 2.87 (2.67 to 3.07) 2.85 (2.7 to 3.01) 0.01 (−0.11 to 0.14) 0.826
 Lumen eccentricity index,‡ % 82.22 (80.97 to 79.55 (77.24 to 81.79) 83.2 (81.95 to 84.44) −3.65 (−5.68 to −1.67) <0.001
83.44)
Patients (frames) analyzed, n 55 (1291) 58 (2203) Difference
 Stent area, mm2 6.86 (6.27 to 7.49) 6.70 (6.02 to 7.39) 6.96 (6.32 to 7.6) −0.26 (−0.51 to −0.01) 0.040
 Average stent diameter, mm 2.91 (2.77 to 3.04) 2.86 (2.72 to 3.01) 2.93 (2.79 to 3.06) −0.07 (−0.12 to −0.01) 0.018
 Minimal stent diameter, mm 2.75 (2.63 to 2.88) 2.70 (2.56 to 2.82) 2.76 (2.63 to 2.88) −0.06 (−0.12 to −0.01) 0.024
 Maximal stent diameter, mm 3.07 (2.93 to 3.21) 3.02 (2.87 to 3.17) 3.09 (2.95 to 3.23) −0.07 (−0.13 to −0.01) 0.024
 Stent eccentricity index,‡ % 89.4 (88.72 to 89.35 (88.52 to 90.17) 89.44 (88.78 to 90.11) −0.09 (−0.66 to 0.48) 0.746
90.06)
Patients (frames) analyzed, n 55 (1241) 58 (2203) Ratio*
 Neointimal area, mm
2
0.83 (0.66 to 1.04) 0.75 (0.56 to 1) 0.87 (0.7 to 1.09) 0.86 (0.72 to 1.04) 0.106
 Average neointimal thickness, mm 0.1 (0.08 to 0.12) 0.09 (0.07 to 0.12) 0.10 (0.08 to 0.12) 0.87 (0.72 to 1.06) 0.156
 Maximal neointimal thickness, mm 0.29 (0.25 to 0.34) 0.29 (0.24 to 0.35) 0.29 (0.25 to 0.34) 1.01 (0.89 to 1.15) 0.898
Patients (frames) analyzed, n 36 (330) 29 (116) Ratio*
 ISA area, mm2 0.53 (0.35 to 0.78) 0.57 (0.36 to 0.89) 0.43 (0.28 to 0.67) 1.31 (0.89 to 1.93) 0.156
 Average ISA distance, mm 0.2 (0.16 to 0.25) 0.20 (0.15 to 0.26) 0.20 (0.15 to 0.25) 1.02 (0.81 to 1.26) 0.856
 Maximal ISA distance, mm 0.35 (0.28 to 0.43) 0.35 (0.27 to 0.45) 0.33 (0.26 to 0.42) 1.06 (0.85 to 1.32) 0.578
Reported values are median (95% CI) from posterior distribution. CI indicates credibility interval; ISA, incomplete stent apposition; and OCT, optical coherence
Downloaded from http://ahajournals.org by on January 30, 2023

tomography.
*Outcomes with skewed distribution were log-transformed for analysis, and the ratio of mean outcome in the thrombus region divided by mean outcome in the control
region is reported. Bayesian analogs to P values were calculated from posterior distributions.
†P values for the null hypothesis that the difference is equal to 0 or the ratio is equal to 1.
‡Eccentricity index=minimum diameter/maximum diameter.

follows. First, OCT is able to identify the underlying putative
cause of VLST in the majority of cases (>98%). The combina-
tion of multiple mechanisms of VLST within the same lesion
was observed more frequently (55%) than the presence of a
single cause (43%). Malapposition combined with uncovered
struts was the most common substrate of VLST (29%), fol-
lowed by neoatherosclerosis (26%). Second, the adverse stent
strut characteristics of malapposition and uncovered struts
were strongly associated with the presence of thrombus. Third,
the longitudinal extension of malapposed and uncovered struts,
but not the incomplete stent malapposition distance or area,
was related to the occurrence of VSLT. Fourth, the association
of malapposition and uncovered stent struts with VLST was
consistent in early- and new-generation DES, and no differ-
ence in the frequency of neoatherosclerosis was observed.
VLST is a clinical manifestation that is of specific interest
because it represents a low but continued risk during long-term
follow-up.16,17 The present study advances our understand-
ing of the mechanisms responsible for VLST in several ways.
Compared with previous studies, this is the first to focus exclu-
sively on the entity of VLST in DES in the largest population
to date. Moreover, this study uniquely provides a direct com-
parison of detailed OCT findings in stent regions with versus
without thrombus, thereby serving as an internal control of
the thrombosed stent region for each given patient. This novel
approach adds new insights into the underlying mechanisms
of the complication by demonstrating for the first time that in-
stent regions with versus adjacent regions without thrombus
more frequently contained malapposed struts, showed greater
longitudinal extension of consecutive malapposed struts, more
commonly included uncovered struts, and more frequently con-
tained fibroatheromas when neoatherosclerosis was present.
A control group of patients without VLST would have been
highly desirable, but a reasonably good match is not feasible,
considering the late presentation of VLST patients. By map-
ping all imaged stents in detail, we provide for the first time a
visual estimation of the colocalization of abnormal stent- and
neointima-related findings that are suggestive of a direct caus-
ative effect. By providing analyses of the local accumulation
(length, percentage per 1.2-mm subsegments) of the most com-
mon abnormalities related to VLST, that is, malapposition and
uncoverage, we provide novel evidence that may be useful for
clinical translation, for example, as thresholds for interventions
during PCI or during follow-up investigations.
Malapposed Stent Struts
Previous studies of late stent thrombosis have focused primar-
ily on the lack of strut coverage as a primary mechanism. Our
data suggest that malapposition (ie, ISA) is an even stron-
ger correlate in the context of VLST.18 When malapposition
 Taniwaki et al   OCT in Very Late DES Thrombosis   655

Table 3.  Categorical, Frame-Level OCT Findings in Regions With Versus Without Thrombus
Ratio of Percentages
Full Stent Region, Thrombus Region, Control Region, (Thrombus/Control),
Median (95% CI) Median (95% CI) Median (95% CI) Median (95% CI) P Value*
Descriptive information
 Patients, n 58 58 58
 Frames, n 3840 1502 2338
 Struts, n 32 730 11 699 21 031
 Average frames per stent, n 66.21±35.06 25.90±18.72 40.31±26.68
 Average struts per stent, n 564.31±385.03 201.71±190.69 362.6±275.42
 Average region length, mm 25.42±14.24 9.72±7.4 15.7±11.29
Percentage of struts
 Uncovered, apposed, %† 4.24 (2.75–6.43) 10.59 (6.67–16.21) 1.28 (0.78–2.03) 8.26 (6.82–10.04) <0.001
 Malapposed, %‡ 1.28 (0.72–2.13) 3.73 (2.24–5.81) 0.29 (0.17–0.46) 13.03 (10.13–16.93) <0.001
 Uncovered or malapposed, %‡ 7.3 (4.94–10.3) 17.73 (12.3–25.3) 1.92 (1.27–2.94) 9.22 (7.76–11.01) <0.001
Percentage of frames with
 ≥30% Uncovered, apposed struts, %§ 6.45 (4–9.6) 16.15 (10.4–23.16) 1.83 (1.07–2.92) 8.82 (6.43–12.31) <0.001
 ≥30% Malapposed strut, % 1.59 (0.67–3.07) 5.51 (3.04–9.41) 0.2 (0.09–0.42) 27.06 (16.23–46.88) <0.001
 ≥30% Uncovered or malapposed strut, % 10.59 (7.09–15.01) 28.21 (19.64–37.81) 2.36 (1.42–3.7) 11.9 (8.88–16.21) <0.001
Percentage of frames with
 TCFA or ThCFA, % 0.12 (0.01–0.73) 1.89 (0.96–3.57) 0.82 (0.42–1.57) 2.3 (1.53–3.49) <0.001
 TCFA or ThCFA, %¶ 23.2 (14.22–34.6) 33.39 (20.29–49.17) 16.72 (9.31–26.91) 1.99 (1.44–2.82) <0.001
Bayesian analogs to P values were calculated from posterior distributions. Reported values are median (95% CI) from posterior distribution. CI indicates credibility
interval; OCT, optical coherence tomography; TCFA, thin-cap fibroatheroma; and ThCFA, thick-cap fibroatheroma.
*P values for the null hypothesis that the ratio is equal to 1.
†Denominator: all apposed struts.
‡Denominator: all measured struts.
Downloaded from http://ahajournals.org by on January 30, 2023

§Denominator: all frames with apposed struts.

¶Subset of 19 patients with any neoatherosclerosis in whom model convergence was excellent.

was considered in conjunction with other causes of VLST
(eg, underexpansion, side branch stenting), it was indeed the
foremost finding (61.8%). The pathomechanistic association
of malapposition leading to VLST has previously been high-
lighted in pathology,19 intravascular ultrasound,20,21 and OCT
studies.6,8 As for the mechanistic cause of malapposition, the
present observational study cannot distinguish whether this
could have been late acquired malapposition (as a result of
positive remodeling and vessel wall hypersensitivity reaction)
or persistent malapposition in relation to suboptimal stent
deployment at the index procedure resulting from the absence
of serial imaging.
We found that the presence of malapposition per se, but
not the maximal distance of ISA, was associated with the
occurrence of thrombosis. Although ISA distance did not dif-
fer between in-stent regions with or without thrombus, this
finding needs to be interpreted in light of the principal find-
ing that malapposition per se was substantially more com-
mon in regions with thrombus. Moreover, although there was
substantial overlap of ISA distance in regions with or without
thrombosis, we found that 95% of regions with thrombus had
a maximal ISA distance >300 μm (lower confidence interval).
Furthermore, we found a graded risk for VLST across higher
proportions of malapposed struts (Figure 2). Along these
lines, a novel finding of this study is that in-stent regions with
versus without thrombus were characterized by significantly
greater length of consecutive frames with malapposed struts;
the respective length was >0.9 mm in 95% of regions with
thrombus. Overall, our observations have clinically relevant
implications in ongoing efforts to optimize PCI guided by
intracoronary imaging in 2 important ways. First, the cutoffs
suggested by this study (ie, ISA >300 μm) are above those
suggested by previous investigations.22 Second, our present
findings highlight the previously unappreciated importance of
the longitudinal extension of malapposition, with a respective
cutoff >1 mm appearing to be more relevant for the risk and
localization of VLST compared with the perpendicular ISA
distance.
The absence of correlation between the perpendicular ISA
distance and in vivo thrombus formation in this study may be
perceived as contradictory to previous flow simulation studies
in which shear rate increased with growing perpendicular ISA
distance.23 Of note, in vitro investigations directly assessing
thrombogenicity could not confirm a correlation between the
perpendicular ISA distance and clot mass volume.24 In addi-
tion, the impact of ISA length or volume was not assessed in
previous studies and should be assessed in future flow models
or in vitro investigations to corroborate our in vivo findings.
Stent Underexpansion
Postprocedural stent expansion and lumen size represent
important risk factors of early stent thrombosis. These param-
eters were not associated with VLST in previous investiga-
tions. Therefore, the observation that stent underexpansion
 656  Circulation  February 16, 2016

Table 4.  Region-Level OCT Findings in Regions With Versus Without Thrombus
Thrombus Region, Mean Control Region, Difference,
(95% CI) Mean (95% CI) Mean (95% CI) P Value*
Patients, n 58 58
Minimal lumen area, mm2 3.61 (3 to 4.22) 3.79 (3.23 to 4.35) −0.18 (−0.58 to 0.22) 0.370
Minimal stent area, mm 2
5.18 (4.63 to 5.73) 5.55 (4.97 to 6.13) −0.37 (−0.71 to −0.04) 0.030
Stent expansion (minimum stent area/ 0.97 (0.84 to 1.09) 1.01 (0.91 to 1.12) −0.05 (−0.1 to 0) 0.065
reference lumen area)
Maximal ISA distance, mm 0.44 (0.3 to 0.58) 0.42 (0.29 to 0.55) 0.02 (−0.08 to 0.13) 0.696
Maximal ISA area, mm2 1.62 (0.82 to 2.43) 1.38 (0.83 to 1.93) 0.24 (−0.35 to 0.83) 0.431
Maximal neointimal thickness, mm 0.69 (0.59 to 0.78) 0.74 (0.65 to 0.84) −0.06 (−0.17 to 0.05) 0.291
Maximal length with consecutive
 Uncoverage, mm 2.24 (1.63 to 2.85) 0.99 (0.64 to 1.34) 1.25 (0.58 to 1.92) <0.001
 Malapposition (ISA), mm 1.56 (0.99 to 2.13) 0.51 (0.26 to 0.76) 1.05 (0.51 to 1.59) <0.001
 Uncoverage or malapposition, mm 3.4 (2.55 to 4.25) 1.29 (0.81 to 1.77) 2.11 (1.22 to 3) <0.001
Reported values are mean (95% CI). Outcomes were derived separately from 2 regions per patient (thrombus versus control) via mean, minimum,
or maximum over several frames and analyzed with generalized estimating equations (gaussian error, patient as clustering variable, exchangeable
correlation). Note that maximal ISA distance pertains to the perpendicular distance between the midpoint of the endoluminal strut surface and the lumen
contour at a cross-sectional level, whereas maximal length with consecutive malapposition refers to the longitudinal extension of consecutive malapposed
struts along the length of the stent. CI indicates confidence interval; ISA, incomplete stent apposition; and OCT, optical coherence tomography.
*P values for the null hypothesis that the difference is equal to 0.

emerged as the most obvious cause for VLST in 6.9% (by
eyeballing) of cases in this study is noteworthy. These findings
were corroborated by the systematic frame-by-frame analysis:
Stent area (P=0.040) and minimal stent diameter (P=0.024)
were significantly lower and stent expansion tended to be
Downloaded from http://ahajournals.org by on January 30, 2023
segment with a similar number of uncovered struts. This is
pathomechanistically plausible because malapposed struts are
more exposed to disturbed blood flow26 and are more likely
to be associated with underlying vessel inflammation (in
case of late acquired malapposition). Our exploratory anal-

smaller (P=0.065) in thrombotic compared with nonthrom-
botic region. Therefore, stent underexpansion appears to affect
the risk of stent thrombosis not only during the early period
but for an extended time, which further advocates the achieve-
ment of an optimal stent expansion after the procedure.
Uncovered Stent Struts
We found that uncovered stent struts remain an important
variable involved in VLST mainly in conjunction with malap-
position. Uncovered struts were 8 times more frequent in
the thrombosed segments compared with control regions.
Pathological studies have shown that delayed arterial heal-
ing with uncovered stent struts represents a major correlate
of VLST in DES19,25 and that delayed healing is more pro-
nounced in early-generation compared with new-generation
DES.11 Guagliumi et al6 previously identified uncovered struts
as the most common OCT correlate of stent thrombosis.
In that study, the vast majority of analyzed DES (16 of 18)
were early-generation sirolimus- or paclitaxel-eluting stents,
and stent thrombosis occurred at a median of 1.7 years after
implantation. The present investigation, a large cohort focus-
ing on VLST in DES, does not grant uncovered struts the same
importance in the very long term. Interestingly, we observed
a direct association of thrombosed segments with the propor-
tion of struts with malapposition or uncovered struts with a
steeper slope (Figure 4) for malapposition, with a plateau
phase observed at lower proportions of affected struts com-
pared with the respective curve for uncoverage. This finding
suggests that the likelihood of thrombosis could be greater
for any given number of malapposed struts compared with a
ysis according to DES type and implantation-to-thrombosis
interval provides novel in vivo insights by demonstrating that
even >6 years after implantation of paclitaxel- or sirolimus-
eluting stents, uncovered struts were still associated with the
occurrence of stent thrombosis. These findings suggest that
although the proportion of uncovered stent struts diminished
over time, uncovered stent struts remained a substantial cor-
relate of VLST consistently for different DES types and dif-
fering durations after the index PCI.
Neoatherosclerosis
Neoatherosclerosis has been suggested to be the lead-
ing cause of VLST in a smaller OCT registry. In the pres-
ent study, this was the case for one fourth of the patients.
Although previous autopsy studies reported an increasing
incidence of neoatherosclerosis over time, no correlation
between neoatherosclerosis-related VLST and implantation-
to-thrombosis time could be observed. The frequency of
neoatherosclerosis-related VLST was similar between early-
and newer-generation DES despite a considerably shorter
implantation-to-thrombosis interval in the latter group. The
similar frequency is in line with a recent human autopsy
suggesting that early- and newer-generation DES are associ-
ated with a similar risk of neoatherosclerosis formation.11,27
Mitigating the risk of neoatherosclerosis formation remains
an important target in the era of new-generation DES. We
recently showed that neoatherosclerosis is correlated with
the natural progression of coronary artery disease, suggest-
ing that the occurrence of VLST may be attenuated by opti-
mal secondary prevention measures.28
 Taniwaki et al   OCT in Very Late DES Thrombosis   657
Downloaded from http://ahajournals.org by on January 30, 2023

Figure 3. Strut maps of the stented regions from the post–stent thrombosis optical coherence tomography OCT analysis showing the
spatial distribution of uncoverage and malapposition for the 55 patients. The longitudinal extension of neoatherosclerosis, stent overlap,
and thrombus regions is shown as bars. Unwrapped stent circumference is plotted along the x axis, and longitudinal stent extension is
plotted along the y axis. Each dot represents a strut. BES indicates biolimus-eluting stent; EES, everolimus-eluting stent; PES, paclitaxel-
eluting stent; SES, sirolimus-eluting stent; and ZES, zotarolimus-eluting stent.

DES Type and Cause of VLST
By assessing a wide spectrum of DES of different genera-
tions, this study indicates in principle similar underlying pri-
mary mechanisms of VLST in early- and new-generation
DES. Although the occurrence of VLST has been shown to
decline with the use of new-generation DES compared with
early-generation DES in large clinical studies and meta-anal-
yses,16,17 the occurrence of the complication per se appears to
be triggered by largely the same stent-related abnormalities as
identified by OCT.
Limitations
This study has several limitations. First, in this observa-
tional, cross-sectional study, the absence of a control group
of patients without stent thrombosis is a notable limitation.
However, the inclusion of an adequate number of well-
matched control patients was not feasible in this registry
and is unlikely to be feasible in future relevant investiga-
tions focusing on VLST, considering also the substantially
delayed time frame from stent implantation to thrombosis.
Second, serial OCT imaging at the time of stent implan-
tation or at an earlier time point before the occurrence of
thrombosis was not available. Third, intravascular ultra-
sound assessment was not available, precluding information
on vessel wall remodeling and therefore additional informa-
tion on the cause of malapposition. Fourth, the number of
patients analyzed is relatively small. However, to the best of
our knowledge, this is the largest cohort to date assessing
VLST in DES by OCT. Fifth, the timing of stent throm-
bosis occurrence differed substantially between early- and
 658  Circulation  February 16, 2016
Figure 4. Continuous assessment of the association between
the proportion of segments with thrombus and the proportion of
malapposition (solid line) or uncoverage (dotted line) in 1.2-mm-long
segments. The posterior median is represented with a thick line; the
95% credibility intervals are represented with thin lines.
new-generation DES. Although we attempted to adjust for
this difference by dividing early-generation DES accord-
ing to earlier versus more delayed VLST occurrence, this
Downloaded from http://ahajournals.org by on January 30, 2023
adjustment is imperfect and limited by the small numbers
in each subgroup. Mechanisms are likely to be different for
earlier compared with more delayed VLST events among
early-generation DES. Sixth, in 2 patients (3.4%), time-
domain OCT rather than frequency-domain OCT was used.
Related to the challenging clinical situation of performing
OCT in the setting of VLST and the participation of a lim-
ited number of interventional cardiologists in the enroll-
ment at each center (see Methods), 21.8% of all patients
with VLST presenting for PCI were enrolled in this registry
(data available only for Bern and Copenhagen), limiting the
external validity of this study.
Conclusions
OCT was able to identify the putative cause of VLST in
the majority of cases. The leading associated findings in
descending order were malapposition, neoatherosclerosis,
uncovered struts, and stent underexpansion. The longitudi-
nal extension of malapposed and uncovered stents was the
most important correlate of thrombus formation in VLST,
whereas no such correlation could be observed for the axial
malapposition distance. The association of malapposition
and uncovered stent struts with thrombus formation was
consistent among early- and new-generation DES, and no
differences in the major causes of VLST were observed
between the 2 groups, suggesting that the most relevant
Figure 5. Representative optical coherence
tomography cross sections, spread-out plots
of struts, and 3-dimensional reconstructions
indicating different underlying causes of very late
stent thrombosis. A, Malapposed struts surrounded
by thrombus (arrowheads) and uncovered
struts (arrows). B, Uncovered struts (arrows)
and struts covered by thrombus (arrowheads).
C, Neoatherosclerosis (asterisk) and thrombus
(arrows); D, significant underexpansion (arrow).
E, Thrombus (arrow) and restenosis (asterisk)
at the edge without malapposition or significant
uncoverage. F, Malapposed struts at the carina of
the bifurcation surrounded by thrombus (arrow).
In spread-out plots of struts (left), the red bar
indicates the thrombosis region; green bar, the
neoatherosclerosis region; and horizontal light gray
bar, the stent overlap region. Yellow dots indicate
uncovered struts; red, malapposed struts; and gray,
struts covered by neointima without malapposition.
In the 3-dimensional reconstructions (D and F,
right), yellow indicates lumen contour; blue, struts;
and red, thrombus. EES indicates everolimus-
eluting stent; PES, paclitaxel-eluting stent; and
SES, sirolimus-eluting stent.
 Taniwaki et al   OCT in Very Late DES Thrombosis   659
Figure 6. Comparison of categorical optical coherence tomography outcomes in thrombus and control regions stratified for early- vs
new-generation drug-eluting stents and for timing of stent thrombosis after implantation. BES indicates biolimus-eluting stent; EES,
everolimus-eluting stent; PES, paclitaxel-eluting stent; SES, sirolimus-eluting stent; and ZES, zotarolimus-eluting stent.
pathomechanisms are comparable between different genera-
tions of DES devices.
Acknowledgments
We acknowledge the excellent support of the catheterization labora-
Downloaded from http://ahajournals.org by on January 30, 2023
tory staff at all participating sites; Sebastian Wiberg, MD, for assis-
tance in data collection; Jouke Dijkstra, PhD, for his support with
imaging analysis software; Dik Heg, PhD, for statistical advice; and
Soheila Aghlmandi and Marcel Zwahlen for their helpful advice with
bayesian analyses.
Disclosures
Drs Räber and Windecker received speaker fees and research support
from St. Jude Medical, Zurich, Switzerland. The other authors report
no conflicts.
References
1. Stefanini GG, Holmes DR Jr. Drug-eluting coronary-artery stents. N Engl
J Med. 2013;368:254–265. doi: 10.1056/NEJMra1210816.
2. Kimura T, Morimoto T, Kozuma K, Honda Y, Kume T, Aizawa T,
Mitsudo K, Miyazaki S, Yamaguchi T, Hiyoshi E, Nishimura E, Isshiki
T; RESTART Investigators. Comparisons of baseline demograph-
ics, clinical presentation, and long-term outcome among patients with
early, late, and very late stent thrombosis of sirolimus-eluting stents:
observations from the Registry of Stent Thrombosis for Review and
Reevaluation (RESTART). Circulation. 2010;122:52–61. doi: 10.1161/
CIRCULATIONAHA.109.903955.
3. van Werkum JW, Heestermans AA, de Korte FI, Kelder JC, Suttorp
MJ, Rensing BJ, Zwart B, Brueren BR, Koolen JJ, Dambrink JH,
van’t Hof AW, Verheugt FW, ten Berg JM. Long-term clinical outcome
after a first angiographically confirmed coronary stent thrombosis: an
analysis of 431 cases. Circulation. 2009;119:828–834. doi: 10.1161/
CIRCULATIONAHA.108.799403.
4. Holmes DR Jr, Kereiakes DJ, Garg S, Serruys PW, Dehmer GJ, Ellis
SG, Williams DO, Kimura T, Moliterno DJ. Stent thrombosis. J Am Coll
Cardiol. 2010;56:1357–1365. doi: 10.1016/j.jacc.2010.07.016.
5. Windecker S, Kolh P, Alfonso F, Collet JP, Cremer J, Falk V, Filippatos G,
Hamm C, Head SJ, Jüni P, Kappetein AP, Kastrati A, Knuuti J, Landmesser
U, Laufer G, Neumann FJ, Richter DJ, Schauerte P, Sousa Uva M,
Stefanini GG, Taggart DP, Torracca L, Valgimigli M, Wijns W, Witkowski
A. 2014 ESC/EACTS guidelines on myocardial revascularization: the
Task Force on Myocardial Revascularization of the European Society of
Cardiology and the European Association for Cardio-Thoracic Surgery.
Eur Heart J. 2014;35:2541–619.
6. Guagliumi G, Sirbu V, Musumeci G, Gerber R, Biondi-Zoccai G, Ikejima
H, Ladich E, Lortkipanidze N, Matiashvili A, Valsecchi O, Virmani R,
Stone GW. Examination of the in vivo mechanisms of late drug-eluting
stent thrombosis: findings from optical coherence tomography and intra-
vascular ultrasound imaging. JACC Cardiovasc Interv. 2012;5:12–20. doi:
10.1016/j.jcin.2011.09.018.
7. Amabile N, Souteyrand G, Ghostine S, Combaret N, Slama MS, Barber-
Chamoux N, Motreff P, Caussin C. Very late stent thrombosis related to
incomplete neointimal coverage or neoatherosclerotic plaque rupture iden-
tified by optical coherence tomography imaging. Eur Heart J Cardiovasc
Imaging. 2014;15:24–31. doi: 10.1093/ehjci/jet052.
8. Parodi G, La Manna A, Di Vito L, Valgimigli M, Fineschi M, Bellandi
B, Niccoli G, Giusti B, Valenti R, Cremonesi A, Biondi-Zoccai G,
Prati F. Stent-related defects in patients presenting with stent throm-
bosis: differences at optical coherence tomography between subacute
and late/very late thrombosis in the Mechanism Of Stent Thrombosis
(MOST) study. EuroIntervention. 2013;9:936–944. doi: 10.4244/
EIJV9I8A157.
9. Kang SJ, Lee CW, Song H, Ahn JM, Kim WJ, Lee JY, Park DW, Lee SW,
Kim YH, Mintz GS, Park SW, Park SJ. OCT analysis in patients with very
late stent thrombosis. JACC Cardiovasc Imaging. 2013;6:695–703. doi:
10.1016/j.jcmg.2013.02.006.
10. Alfonso F, Dutary J, Paulo M, Gonzalo N, Pérez-Vizcayno MJ, Jiménez-
Quevedo P, Escaned J, Bañuelos C, Hernández R, Macaya C. Combined
use of optical coherence tomography and intravascular ultrasound imaging
in patients undergoing coronary interventions for stent thrombosis. Heart.
2012;98:1213–1220. doi: 10.1136/heartjnl-2012-302183.
11. Otsuka F, Vorpahl M, Nakano M, Foerst J, Newell JB, Sakakura K, Kutys
R, Ladich E, Finn AV, Kolodgie FD, Virmani R. Pathology of second-
generation everolimus-eluting stents versus first-generation sirolimus- and
paclitaxel-eluting stents in humans. Circulation. 2014;129:211–223. doi:
10.1161/CIRCULATIONAHA.113.001790.
12. Cutlip DE, Windecker S, Mehran R, Boam A, Cohen DJ, van Es GA,
Steg PG, Morel MA, Mauri L, Vranckx P, McFadden E, Lansky A,
Hamon M, Krucoff MW, Serruys PW; Academic Research Consortium.
Clinical end points in coronary stent trials: a case for standard-
ized definitions. Circulation. 2007;115:2344–2351. doi: 10.1161/
CIRCULATIONAHA.106.685313.
13. Tearney GJ, Regar E, Akasaka T, Adriaenssens T, Barlis P, Bezerra HG,
Bouma B, Bruining N, Cho JM, Chowdhary S, Costa MA, de Silva R,
Dijkstra J, Di Mario C, Dudek D, Dudeck D, Falk E, Falk E, Feldman
MD, Fitzgerald P, Garcia-Garcia HM, Garcia H, Gonzalo N, Granada JF,
 660  Circulation  February 16, 2016

Guagliumi G, Holm NR, Honda Y, Ikeno F, Kawasaki M, Kochman J,
Koltowski L, Kubo T, Kume T, Kyono H, Lam CC, Lamouche G, Lee
DP, Leon MB, Maehara A, Manfrini O, Mintz GS, Mizuno K, Morel
MA, Nadkarni S, Okura H, Otake H, Pietrasik A, Prati F, Räber L, Radu
MD, Rieber J, Riga M, Rollins A, Rosenberg M, Sirbu V, Serruys PW,
Shimada K, Shinke T, Shite J, Siegel E, Sonoda S, Sonada S, Suter M,
Takarada S, Tanaka A, Terashima M, Thim T, Troels T, Uemura S, Ughi
GJ, van Beusekom HM, van der Steen AF, van Es GA, van Es GA, van
Soest G, Virmani R, Waxman S, Weissman NJ, Weisz G; International
Working Group for Intravascular Optical Coherence Tomography (IWG-
IVOCT). Consensus standards for acquisition, measurement, and report-
ing of intravascular optical coherence tomography studies: a report from
the International Working Group for Intravascular Optical Coherence
Tomography Standardization and Validation. J Am Coll Cardiol.
2012;59:1058–1072. doi: 10.1016/j.jacc.2011.09.079.
14. Räber L, Zanchin T, Baumgartner S, Taniwaki M, Kalesan B,
Moschovitis A, Garcia-Garcia HM, Justiz J, Pilgrim T, Wenaweser P,
Meier B, Jüni P, Windecker S. Differential healing response attributed
to culprit lesions of patients with acute coronary syndromes and stable
coronary artery after implantation of drug-eluting stents: an optical
coherence tomography study. Int J Cardiol. 2014;173:259–267. doi:
10.1016/j.ijcard.2014.02.036.
15. Radu MD, Räber L, Kalesan B, Muramatsu T, Kelbæk H, Heo J,
Jørgensen E, Helqvist S, Farooq V, Brugaletta S, Garcia-Garcia HM, Jüni
P, Saunamäki K, Windecker S, Serruys PW. Coronary evaginations are
associated with positive vessel remodelling and are nearly absent fol-
lowing implantation of newer-generation drug-eluting stents: an optical
coherence tomography and intravascular ultrasound study. Eur Heart J.
2014;35:795–807. doi: 10.1093/eurheartj/eht344.
16. Bangalore S, Kumar S, Fusaro M, Amoroso N, Attubato MJ, Feit F,
Bhatt DL, Slater J. Short- and long-term outcomes with drug-eluting and
bare-metal coronary stents: a mixed-treatment comparison analysis of
117 762 patient-years of follow-up from randomized trials. Circulation.
2012;125:2873–2891. doi: 10.1161/CIRCULATIONAHA.112.097014.
17. Palmerini T, Biondi-Zoccai G, Della Riva D, Stettler C, Sangiorgi D,
D’Ascenzo F, Kimura T, Briguori C, Sabatè M, Kim HS, De Waha A,
Kedhi E, Smits PC, Kaiser C, Sardella G, Marullo A, Kirtane AJ, Leon
Downloaded from http://ahajournals.org by on January 30, 2023
attributed to first-generation sirolimus- and paclitaxel-eluting stents. J Am
Coll Cardiol. 2011;57:390–398. doi: 10.1016/j.jacc.2010.05.066.
20. Cook S, Ladich E, Nakazawa G, Eshtehardi P, Neidhart M, Vogel R, Togni
M, Wenaweser P, Billinger M, Seiler C, Gay S, Meier B, Pichler WJ, Jüni
P, Virmani R, Windecker S. Correlation of intravascular ultrasound find-
ings with histopathological analysis of thrombus aspirates in patients with
very late drug-eluting stent thrombosis. Circulation. 2009;120:391–399.
doi: 10.1161/CIRCULATIONAHA.109.854398.
21. Lee CW, Kang SJ, Park DW, Lee SH, Kim YH, Kim JJ, Park SW, Mintz
GS, Park SJ. Intravascular ultrasound findings in patients with very late
stent thrombosis after either drug-eluting or bare-metal stent implantation.
J Am Coll Cardiol. 2010;55:1936–1942. doi: 10.1016/j.jacc.2009.10.077.
22. Attizzani GF, Capodanno D, Ohno Y, Tamburino C. Mechanisms, patho-
physiology, and clinical aspects of incomplete stent apposition. J Am Coll
Cardiol. 2014;63:1355–1367. doi: 10.1016/j.jacc.2014.01.019.
23. Foin N, Gutiérrez-Chico JL, Nakatani S, Torii R, Bourantas CV, Sen S,
Nijjer S, Petraco R, Kousera C, Ghione M, Onuma Y, Garcia-Garcia HM,
Francis DP, Wong P, Di Mario C, Davies JE, Serruys PW. Incomplete stent
apposition causes high shear flow disturbances and delay in neointimal
coverage as a function of strut to wall detachment distance: implications for
the management of incomplete stent apposition. Circ Cardiovasc Interv.
2014;7:180–189. doi: 10.1161/CIRCINTERVENTIONS.113.000931.
24. Kolandaivelu K, Swaminathan R, Gibson WJ, Kolachalama VB,
Nguyen-Ehrenreich KL, Giddings VL, Coleman L, Wong GK, Edelman
ER. Stent thrombogenicity early in high-risk interventional set-
tings is driven by stent design and deployment and protected by poly-
mer-drug coatings. Circulation. 2011;123:1400–1409. doi: 10.1161/
CIRCULATIONAHA.110.003210.
25. Joner M, Finn AV, Farb A, Mont EK, Kolodgie FD, Ladich E, Kutys
R, Skorija K, Gold HK, Virmani R. Pathology of drug-eluting stents in
humans: delayed healing and late thrombotic risk. J Am Coll Cardiol.
2006;48:193–202. doi: 10.1016/j.jacc.2006.03.042.
26. Koskinas KC, Chatzizisis YS, Antoniadis AP, Giannoglou GD. Role of
endothelial shear stress in stent restenosis and thrombosis: pathophysi-
ologic mechanisms and implications for clinical translation. J Am Coll
Cardiol. 2012;59:1337–1349. doi: 10.1016/j.jacc.2011.10.903.
27. Nakazawa G, Otsuka F, Nakano M, Vorpahl M, Yazdani SK, Ladich E,

MB, Stone GW. Stent thrombosis with drug-eluting and bare-metal
stents: evidence from a comprehensive network meta-analysis. Lancet.
2012;379:1393–1402. doi: 10.1016/S0140-6736(12)60324-9.
18. Lüscher TF, Steffel J, Eberli FR, Joner M, Nakazawa G, Tanner FC,
Virmani R. Drug-eluting stent and coronary thrombosis: biological mech-
anisms and clinical implications. Circulation. 2007;115:1051–1058. doi:
10.1161/CIRCULATIONAHA.106.675934.
19. Nakazawa G, Finn AV, Vorpahl M, Ladich ER, Kolodgie FD, Virmani R.
Coronary responses and differential mechanisms of late stent thrombosis
Kolodgie FD, Finn AV, Virmani R. The pathology of neoatherosclerosis
in human coronary implants bare-metal and drug-eluting stents. J Am Coll
Cardiol. 2011;57:1314–1322. doi: 10.1016/j.jacc.2011.01.011.
28. Taniwaki M, Windecker S, Zaugg S, Stefanini GG, Baumgartner S,
Zanchin T, Wenaweser P, Meier B, Jüni P, Räber L. The association
between in-stent neoatherosclerosis and native coronary artery disease
progression: a long-term angiographic and optical coherence tomography
cohort study. Eur Heart J. 2015;36:2167–2176. doi: 10.1093/eurheartj/
ehv227.

Clinical Perspective
Very late stent thrombosis (VLST) is a devastating event that by definition occurs beyond 1 year after stent implantation.
Although newer-generation drug-eluting stents led to a considerable reduction VLST continues to occur up to 20 years after
the index procedure. Pathomechanisms underlying VLST are poorly understood. The use of intracoronary optical coherence
tomography, a light-based, high-resolution imaging technique, offers unique opportunities to study the potential causes of
VLST. In this registry, a total of 58 patients with available optical coherence tomography of the thrombotic lesion were
assessed. The leading causes of VLST in descending order were malapposition, neoatherosclerosis, stent strut uncoverage,
and underexpansion. These findings were consistent among early- and newer-generation drug-eluting stents, suggesting that
the causes of VLST may not differ among different generations of metallic drug-eluting stents despite a lower frequency
with newer-generation drug-eluting stents. This study further suggests that in-stent regions with versus without thrombus
more frequently contained malapposed struts, showed a greater longitudinal extension of malapposed struts in the absence
of differences in the axial malapposition distance, and more commonly included uncovered struts. These findings may have
implications for ongoing efforts to improve imaging-guided percutaneous coronary intervention procedures by suggest-
ing thresholds for corrective measures. The cutoffs associated with thrombus formation in this study (ie, incomplete stent
apposition >300 µm) were above those suggested by previous investigations. The findings further highlight the previously
unappreciated importance of the longitudinal extension of malapposition, with a respective cutoff >1 mm.