Introduction

Although several randomised clinical trials have shown that sirolimus-eluting stents (SES) substantially reduce angiographic restenosis and the need for repeat revascularisation [1, 2], concerns have been raised regarding late adverse events such as late target lesion revascularisation and stent thrombosis beyond 1 year [3, 4]. Recently, stent fractures have been suggested as a cause of restenosis within 1 year after successful SES implantation [5–8]. However, long-term outcomes of stent fractures without early clinical events have not been fully clarified. Stent fractures may be underdiagnosed because of the difficulty of conventional coronary angiography (CAG) in detecting its occurrence. Multislice computed tomography (MSCT) has been developed as a useful non-invasive imaging modality for the diagnosis of coronary artery disease [9–13] and can depict stent struts with inherent high-contrast resolution, thereby detecting stent fractures with high accuracy [14–17]. The aim of this study was to investigate the impact of SES fractures without early cardiac events on long-term clinical outcomes using 64-slice MSCT.

Materials and methods

Study population

Between September 2004 and September 2009, 2,956 patients and 3,932 native lesions were treated with SESs (Cypher, Cordis, Johnson & Johnson Company, Miami Lakes, FL) in our institute. Follow-up CAGs were recommended in all patients without renal insufficiency and allergies to contrast media. In patients reluctant to undergo CAG, MSCT was suggested as an alternative follow-up examination. A total of 591 patients opted to undergo MSCT between 6 and 18 months after stenting. Of these, we excluded 28 patients who underwent target lesion revascularisation (TLR) within 90 days after follow-up MSCT as early cardiac events. MSCTs were not evaluable in 30 patients because of insufficient image quality due to metal artefacts, severe calcification, or motion artefacts. Five patients were lost to follow-up. Finally, 644 lesions in 528 patients were retrospectively analysed in this study (Fig. 1). All patients were pretreated with aspirin plus either ticlopidine or clopidogrel before stenting and at least 1 year of dual antiplatelet therapy was recommended. The research protocol was approved by the Institutional Review Boards and all patients provided written consent for the use of their data.

Fig. 1
figure 1

Flow chart of patient enrolment

Full size image

Scan protocol, data acquisition, and image reconstruction of follow-up MSCT

MSCT angiography was performed with a 64-slice scanner (SOMATOM Sensation 64 Cardiac, Siemens Medical Solutions, Forchheim, Germany). When a patient’s heart rate was >60 bpm, beta-blockers (metoprolol 20 to 60 mg orally or 1 to 2 mg propranolol intravenously or both) were administered for heart rate control. A bolus of contrast media (Omnipaque, 350 mg iodine/ml, or iopamidol, 370 mg iodine/ml) was intravenously injected, followed by 30 ml saline. The injection rate was determined according to the patient’s body weight (2.5–5.0 ml/s). The examination was performed between the tracheal bifurcation and diaphragm with the following parameters: collimation width 64 × 0.6 mm, rotation time 330 ms, tube voltage 120 kV, maximal effective tube current 800 mA, table feed 11.5 mm/rotation, and pitch 0.2. Image reconstruction was retrospectively gated to an electrocardiogram, and the optimal cardiac phase showing the minimum motion artefact was individually determined. Two sets of images were reconstructed with different types of convolution kernels: one set was reconstructed with a smooth kernel (B25f), and the other set was reconstructed with a sharp (Heartview, Siemens Medical Solutions) kernel (B46f). For delineating low-contrast objects such as the coronary lumen or vessel wall, we used images generated with the smooth kernel, whereas to observe the stented segment, we used both of the images that were generated with smooth and sharp kernels. CT data sets were transferred to an offline workstation (Aquarius NetStation, Terarecon Inc., San Mateo, CA) for image analysis.

Definition of stent fracture

A CT-detected stent fracture was defined as a complete gap upon visual inspection with Hounsfield units (HU) <300 (the lowest HU in the stent area) at the site of separation (Fig. 2a–c) [15]. Distinction between stent fracture and lack of the overlap of multiple stents can be easily made by checking whether the number of implanted stents is identical to that observed by MSCT. Stent fractures were assessed by two independent observers each. Observers were blinded to the other test. In the case of observer readings differing, a consensus reading was performed and used in the final analysis.

Fig. 2
figure 2

Definition of stent fracture. a Two overlapped sirolimus-eluting stents (3.0 × 23 mm, 3.0 × 23 mm, lines) were implanted in the right coronary artery. b Fluoroscopy immediately after the procedure demonstrating no separation of stent struts. c Multislice computed tomography 10 months after the procedure showing complete separation of stent struts at the proximal stent. No stent struts could be depicted in the cross-sectional image (arrow)

Full size image

Clinical follow-up and definition of major adverse cardiac events

Clinical follow-up information was obtained from outside patient records or telephone interviews. Major adverse cardiac events (MACE) were defined as a composite of cardiac death, stent thrombosis, and TLR >90 days after follow-up MSCT. All deaths without unequivocal noncardiac causes were considered to be cardiac. Stent thrombosis was defined as acute coronary syndrome (ACS) with angiographic evidence of a thrombus within a stent (definite according to the Academic Research Consortium definition [18]). TLR was defined as a clinically driven repeat intervention (either PCI or coronary artery bypass graft surgery) to a > 50 % diameter stenosis within the stent or 5 mm proximal or distal segments adjacent to the stent for recurrent angina and/or signs of ischaemia.

Quantitative coronary angiographic analysis

CAG was performed within 30 days after MSCT in 210 patients (263 lesions) with significant stenosis based on MSCT findings and/or with segments that could not be evaluated due to some artefact. Off-line quantitative coronary angiography (QCA) was performed before and after the index procedure and at follow-up with the Cardiovascular Measurement System (CMS-MEDIS, Medical Imaging Systems, Leiden, The Netherlands). Reference diameters and minimal lumen diameters were measured on the view demonstrating the smallest lumen diameter at diastolic frames.

Statistical analysis

Quantitative variables are described as mean ± SD. Discrete variables are presented as numbers and percentages. T-tests were used to compare quantitative variables, while χ 2 tests and Fisher’s exact test were performed on discrete variables. Cox regression analysis was performed to identify the potential predictors of MACE after MSCT follow-up examinations. Variables of patient characteristics with a p value <0.10 in the univariate model were used in the multivariate model. Covariates in univariate analysis for the prediction of MACE were CT-detected stent fracture, stent length, diabetes mellitus, age, chronic total occlusion (CTO), male gender, stent diameter, hypertension, hyperlipidaemia, and ACS. Event survival curves were estimated using Kaplan-Meier methods. A log rank test was performed to evaluate differences between event-free survival curves. Significance was defined as a p value <0.05. SPSS 15.0 (SPSS Inc., Chicago, IL, USA) was used for data analysis.

Results

Baseline characteristics

Stent fractures were observed in 44 lesions in 39 patients on MSCT (7.4 %). Baseline patient demographics are shown in Table 1. There were no significant differences in patient characteristics. Beta-blockers were used prior to CT examination in 338 (64 %) patients (heart rate during scan, 63.4 ± 10.4 bpm). The mean estimated radiation exposure associated with MSCT was 11.5 mSv. Lesion characteristics are compared in Table 2. RCA lesions, ACC/AHA type C lesions, and CTO lesions were more common in CT-detected stent fracture lesions. In procedural characteristics, stent length was longer and stent diameter, the number of stents, and rate of overlapping stents were larger in the CT-detected stent fracture group. In QCA analysis, the baseline reference diameter was larger and lesion length was longer in the CT-detected stent fracture group. At follow-up, late loss and % diameter stenosis were larger in the CT-detected stent fracture group (Table 3).

Table 1 Baseline patient demographics
Full size table
Table 2 Lesion characteristics
Full size table
Table 3 Quantitative coronary angiography
Full size table

Clinical outcomes of patients with or without stent fracture

The median follow-up interval after MSCT was 4.6 years (interquartile range 3.1–5.6 months). Clinical outcomes were presented in Table 4. MACE was observed in 52 patients (9.8 %). There were no significant differences in cardiac death and stent thrombosis. TLR and total MACE were more frequently observed in the CT-fracture group than in the fracture-free group. In the stent fracture group, all TLR sites were visually the same or very close to the CT-detected fracture sites. Univariate Cox regression analysis indicated a significant relationship between MACE and a CT-detected stent fracture, age, stent length, diabetes mellitus, and CTO. In the multivariate model, a CT-detected stent fracture and age remained significant predictors of MACE (Table 5). MACE-free survival rates after follow-up MSCT were significantly lower in the CT-fracture group than in the fracture-free group (Fig. 3). Figure 4 shows a representative case of very late restenosis at the site of stent fracture, which could be detected by follow-up MSCT, but not by CAG.

Table 4 Clinical follow-up after follow-up MSCT
Full size table
Table 5 Predictors of MACE after follow-up MSCT
Full size table
Fig. 3
figure 3

Major adverse cardiac event-free survival after follow-up multislice computed tomography

Full size image
Fig. 4
figure 4

Representative case of late target lesion revascularisation at the site of a stent fracture detected by MSCT. a, b A sirolimus-eluting stent (3.5 × 23 mm, line) was implanted at significant stenosis (arrow) in the right coronary artery. c Follow-up coronary angiography 9 months after the procedure showing no restenosis. d Fluoroscopy at follow-up angiography could not detect a stent fracture. e Follow-up multislice computed tomography (MSCT) 9 months after the procedure could depict the stent fracture clearly. f Coronary angiography was performed 45 months after stenting following an episode of acute chest pain showing late restenosis (arrowhead) at the site of the stent fracture detected by MSCT. Intravascular ultrasound image shows the absence of the stent strut and neointimal hyperplasia

Full size image

Discussion

To our knowledge, this is the first observational study to report the clinical impact of SES fractures detected by MSCT on long-term outcomes. The main finding of our analysis is that an SES fracture detected by MSCT without early clinical events was associated with long-term clinical adverse events.

Detection of stent fractures by MSCT

In previous angiographic studies, the incidence of SES fractures was relatively low (0.8 % to 7.7 %) [5–8]. On the other hand, an autopsy study has shown a higher rate of drug-eluting stent fractures (29 %) [19]. It implies that stent fractures may be under-diagnosed by fluoroscopy because of the limited sensitivity to detect stent fractures. Therefore, a more sensitive method is needed to examine the clinical relevance of SES fractures. Intravascular ultrasound (IVUS), which provides an accurate cross-sectional image of stent strut distribution, can more reliably detect SES fractures [20]. However, IVUS only provides a single cross-sectional view, making it difficult to understand the whole stent structure [21]. In addition, IVUS is costly, invasive, and unsuitable for routine follow-up in patients with coronary stents in clinical practice. MSCT, which can depict the configuration of the whole stent three-dimensionally with inherent high-contrast resolution less invasively, can detect more stent fractures than angiography [14–17, 21]. MSCT may be the most appropriate modality to evaluate stent deformities.

Stent fractures and long-term prognosis

Although several case reports and observational studies have reported that SES fractures could be a cause of cardiac events within 1 year after successful SES implantation [5–8, 14, 15], the outcome of SES fractures without early clinical events is controversial. Umeda et al. showed that SES fractures were not associated with MACE within 1 to 4 years after stenting [22]. Ino et al. demonstrated that late restenosis was not observed in SES fracture sites without early restenosis during mid-term (a minimum of 16 months) follow-up [23]. On the other hand, 10.9 % of MACE was reported in patients with stent fractures without early restensosis [24] and an SES fracture without early restenosis was suggested to be a strong predictor of late in-stent restenosis [25]. This disagreement regarding the outcomes of SES fractures without early clinical events could be due to the fact that SES fractures were diagnosed with angiography, which could overlook fractures, as well as differences in the definition of stent fracture and study population. Thus, a study using a more reliable imaging modality such as MSCT to diagnose stent fractures should be conducted to investigate the relationship between SES fractures and outcomes. In our study, stent fractures detected by MSCT were associated with long-term MACE, indicating that MSCT can detect abnormalities associated with long-term outcomes possibly overlooked by angiography.

The exact mechanisms by which an SES fracture causes clinical events are still uncertain. We speculated several causal relationships. First, local drug delivery may be unavailable because of the absence of stent struts. Second, separated stent struts could induce mechanical stimulation to the vessel wall resulting in inflammation, which could cause neointimal hyperplasia or aneurysm formation. An optical coherence tomography study demonstrated that neointimal hyperplasia was enhanced at the site of an SES fracture [26]. Coronary aneurysm was frequently observed at the DES fracture site by IVUS [27], and we reported a case of stent thrombosis at the site of stent fracture and stent malaposition as detected by MSCT [17].

Potential of MSCT for the evaluation of patients with coronary stents

MSCT has been more widely available for the evaluation of patients with native coronary artery disease, including not only coronary artery stenosis [9], but also coronary plaque characteristics [10, 11] and outcomes [12, 13] as a less invasive imaging technique. However, assessment of the coronary artery with a stent has been challenging because the artefact of the stent strut interferes with clear visualisation of the in-stent lumen. According to appropriate use criteria for cardiac computed tomography, assessment of the coronary stent is considered uncertain or inappropriate with the exception of left main stents [28]. In our study, CAG was performed in 210 patients (40 %) after follow-up MSCT because of insufficient total coronary analysis, which means that routine use of 64-slice MSCT is not efficient for patients with coronary stents. However, at present, advances in the technology of MSCT have made it possible to assess in-stent lumens with high accuracy, particularly in the case of large stent diameters [29, 30]. Therefore, in appropriately selected patients with coronary stents, MSCT can simultaneously evaluate in-stent-lumens and stent fractures, which are strong predictors of long-term MACE, as shown in our study. In addition, by providing information on coronary stenosis and plaque characterisation in vessels or locations not touched by the intervention, MSCT could be an ideal imaging modality suitable for comprehensive assessment of the whole coronary artery in patients after stenting.

Study limitation

The limitations of this study are as follows. First, this was a retrospective, single-centre study and the sample size was small, yielding a small number of cases of stent fracture. Second, the study population consisting of patients undergoing follow-up MSCT suffers from selection bias. Third, TLR, which was defined as MACE in our study, is not a real hard endpoint. Fourth, it was difficult to confirm that the site of TLR was identical to that of CT-detected stent fracture and the restenotic lesion originated from the fracture site, especially in cases of stent thrombosis and diffuse restenosis. Fifth, we evaluated only SES and our results cannot be applied to other drug-eluting stents. Finally, stent fractures detected by MSCT could not be validated pathologically.

Conclusions

An SES fracture detected by MSCT without early clinical events was associated with long-term clinical adverse events.