OUP user menu

The potential clinical utility of intravascular ultrasound guidance in patients undergoing percutaneous coronary intervention with drug-eluting stents

Probal Roy , Daniel H. Steinberg , Steven J. Sushinsky , Teruo Okabe , Tina L. Pinto Slottow , Kimberly Kaneshige , Zhenyi Xue , Lowell F. Satler , Kenneth M. Kent , William O. Suddath , Augusto D. Pichard , Neil J. Weissman , Joseph Lindsay , Ron Waksman
DOI: http://dx.doi.org/10.1093/eurheartj/ehn249 1851-1857 First published online: 11 June 2008


Aims To assess the impact on clinical outcomes of intravascular ultrasound (IVUS) guidance during drug-eluting stent (DES) implantation. IVUS guidance during percutaneous coronary intervention (PCI) has been demonstrated to be useful in optimizing stent deployment. However, it is not proved that routine use of IVUS guidance with DES implantation can prevent stent thrombosis or restenosis.

Methods and results The clinical outcomes of 884 patients undergoing IVUS-guided intracoronary DES implantation from April 2003 to May 2006 were compared with the outcomes of a propensity-score matched population undergoing DES implantation with angiographic guidance alone. The primary endpoint of the study was definite stent thrombosis at 12 months. The secondary endpoint was major adverse cardiac events (MACE). After propensity-score matching, the two groups were well matched for clinical and angiographic characteristics. Patients undergoing IVUS-guided DES implantation underwent less direct stenting, more post-dilation, and had greater cutting balloon and rotational atherectomy use. At 30 days and at 12 months, a higher rate of definite stent thrombosis was seen in the No IVUS group (0.5 vs. 1.4%; P = 0.046) and (0.7 vs. 2.0%; P = 0.014), respectively. There were no major differences in late stent thrombosis and MACE (14.5 vs. 16.2%; P = 0.33) at 12 month follow-up between the groups. Rates of death and Q-wave myocardial infarction were similar, and there was no significant difference between groups in target vessel revascularization. However, a trend was seen in favour of the IVUS group in target lesion revascularization (5.1 vs. 7.2%; P = 0.07). IVUS guidance was an independent predictor of freedom from cumulative stent thrombosis at 12 months (adjusted HR 0.5, CI 0.1–0.8; P = 0.02).

Conclusion IVUS guidance during DES implantation has the potential to influence treatment strategy and reduce both DES thrombosis and the need for repeat revascularization.

  • Intravascular ultrasound
  • Drug-eluting stent
  • Percutaneous coronary intervention

Drug-eluting stents (DES) have proved to be highly efficacious in reducing restenosis compared with bare-metal stents (BMS).1,2 Despite this significant advance in the percutaneous treatment of coronary artery disease, these stents are not free of restenosis3 and are limited by late stent thrombosis.4,5 Intravascular ultrasound (IVUS) studies have suggested that suboptimal stent deployment is a major aetiology underlying both DES restenosis68 and thrombosis.9,10 In the BMS era, IVUS guidance during percutaneous coronary intervention (PCI) was shown to have a favourable impact on restenosis.11,12 Though the IVUS predictors of BMS subacute thrombosis have been reported,1315 it has not been shown that routine IVUS guidance is associated with reductions in stent thrombosis. Furthermore, there are no reports assessing the clinical utility of routine IVUS guidance in the current era of DES.

Given the importance of optimal stent deployment (stent expansion and apposition to the vessel wall) with DES, it is intuitive that IVUS guidance should yield clinical benefit in patients undergoing DES implantation. IVUS correlates of both DES restenosis68 and thromboses9,10 have been previously described. We examined the hypothesis that IVUS-guided PCI with DES yields clinical benefit in terms of stent thrombosis and major adverse cardiac events (MACE).


Study population and design

The study population was identified from a registry of consecutive patients who underwent DES implantation from April 2003 to May 2006 at Washington Hospital Center. Patients who underwent IVUS guidance to all treated lesions were considered. The control group was taken from a cohort of 1281 patients, who underwent DES implantation with angiographic guidance only during the same time period, enrolled in the registry. A complete data set to allow for propensity-score matching and 12 month follow-up was required for patients to be included in either group of the study. Propensity-score matching balanced the clinical and angiographic characteristics of the two groups (IVUS and No IVUS).

The outcomes in 884 patients (1296 lesions) who underwent IVUS-guided DES implantation to all treated lesions were compared with those in 884 propensity-score matched patients (1312 lesions) who underwent DES implantation with angiographic guidance alone. The in-hospital, 30-day, and 1-year clinical outcomes were compared between the IVUS and No IVUS groups. All patients underwent their procedure at Washington Hospital Center, a tertiary referral hospital with 11 catheterization laboratories serviced by 31 independent interventional cardiologists. All patients gave written consent for the PCI procedure and the study was conducted under local Institutional Review Board approval.

Procedures and adjunctive medical therapy

PCI was performed using standard technique via the femoral approach in most cases. Patients received either sirolimus-eluting stents (SES; Cypher, Cordis, Johnson and Johnson, Miami Lakes, FL, USA. Diameters 2.5–3.5 mm, lengths 8–33 mm) or paclitaxel-eluting stents (PES; Taxus, Boston Scientific Corp., Natick, MA, USA; diameters 2.5–3.5 mm, lengths 8–32 mm) or both. All patients were treated with aspirin 325 mg prior to PCI and loaded with clopidogrel 300–600 mg if not already on a maintenance dose. Dual antiplatelet therapy was recommended to all study patients for a minimum duration of 6 months. During PCI, patients were anticoagulated with either bivalirudin (a bolus of 0.75 mg/kg, followed by an intravenous infusion of 1.75 mg/kg/h) or unfractionated heparin (a bolus of 40 U/kg and additional heparin to achieve an activated clotting time of 250–300 s). Use of adjunctive devices, including IVUS, and platelet glycoprotein IIb/IIIa inhibitors were at the discretion of the operator. IVUS was performed using standard technique. IVUS was performed either pre-intervention, post-intervention or both at the discretion of the operator. One of two commercially available systems: Atlantis S (Boston Scientific Corp/SCIMED, Minneapolis, MN, USA) or Eagle Eye (Volcano Therapeutics, Inc., Rancho Cordova, CA, USA) were used. IVUS images were recorded after administration of 100–200 µg of nitroglycerin. The ultrasound catheter was advanced >5 mm beyond the lesion/stent and was pulled back to a point >5 mm proximal to the lesion/stent using motorized transducer pullback at 0.5 mm/s or 1.0 mm/s. IVUS was performed and interpreted by the treating physician and at least one experienced IVUS technician. Routine measurements were recorded pre- and post-stent implantation. The response to the IVUS findings was at the discretion of the treating physician.

Clinical endpoints and definitions

The primary endpoint of the study was definite stent thrombosis, defined as angiographic or autopsy-proven partial or complete stent occlusion. Acute stent thrombosis was defined as occurring within 24 h of the procedure, subacute stent thrombosis from 1 to 30 days, and late stent thrombosis beyond 30 days. Probable stent thrombosis included all definite stent thrombosis, death of cardiac cause within 30 days of the index procedure, and Q-wave myocardial infarction (MI) attributable to the target vessel. The secondary endpoint of the study was MACE, including death, Q-wave MI, and target vessel revascularization (TVR). Death was all cause mortality. Cardiac death included all deaths where a non-cardiac cause could not be demonstrated. Q-wave MI was defined as an elevation in creatine kinase-MB (CK-MB) ≥2 times the upper normal value (2.6 ng/mL) in the presence of new Q-waves on the electrocardiograph in ≥2 contiguous leads. CK-MB was routinely measured before and after the procedure. Measurements were repeated every 8 h until a peak value was reached and values began returning to normal. Peri-procedural MI was defined using a 2-fold definition. For patients with baseline CK-MB ≤2.6 ng/ml, post-procedural MI was defined as CK-MB >5.2 ng/ml. For patients with baseline CK-MB >2.6 ng/ml, post-procedural MI was defined as ≥1.5 times baseline CK-MB. Target lesion revascularization (TLR) was defined as revascularization, either percutaneous or surgical, for a stenosis within stent or in the 5 mm segments proximal or distal to the stent. TVR was defined as either percutaneous or surgical revascularization of the stented epicardial vessel. Angiographic success was defined as a residual stenosis <30% with thrombolysis In MI 3 flow.

Data collection and follow-up

Demographic, clinical, and procedural data, along with in-hospital outcomes, were collected and entered into a prospective database. These data were obtained from hospital chart review by independent research personnel blinded to the study objectives. All data management and analysis was performed by a dedicated data coordinating centre (Data Center, Cardiovascular Research Institute, Washington, DC, USA). Clinical follow-up was performed at 1, 6, and 12 months by trained quality assurance nurses, who worked exclusively with the database to determine post-PCI clinical events. Clinical follow-up was performed by telephone contact or office visit. A committee independently adjudicated all subsequent clinical events. Clinical follow-up was available on all patients in the study population.


The propensity score was estimated from a non-parsimonious logistic regression model for treatment with IVUS vs. No IVUS. To calculate the propensity score, the following variables were entered into the model: male, age; stable angina, unstable angina, acute MI, and cardiogenic shock; prior MI, coronary artery bypass surgery, and prior PCI; diabetes, insulin requiring diabetes, hypertension, hypercholesterolaemia, current smoking, family history, chronic renal insufficiency, and peripheral vascular disease; left main coronary, left anterior descending, left circumflex, and right coronary arteries; saphenous vein graft; ostial, proximal, mid-, and distal lesion locations; ACC/AHA classification Type A, B, and C lesions, restenotic lesions; number of lesions treated and glycoprotein IIb–IIIa use. No interactions were considered in this model. Patients receiving IVUS were then one-to-one matched to the patients receiving No IVUS on propensity score using the nearest available pair matching method by personnel blinded to patient outcomes. Subgroups comparing IVUS vs. No IVUS defined by propensity-score quintiles were well matched, particularly for those variables established to be predictive of stent thrombosis. Hosmer and Lemeshow Goodness-of-Fit test was used to assess the model fit to the data. The Chi-square test statistic was 8.02 (P = 0.43), which indicates a good fit of the model to the data. The c-statistic for the model was 0.7, indicating relatively good discrimination. All analysis was stratified by the matching pair. Statistical analysis was performed using SAS version 9.1 (SAS Institute, Cary, NC, USA). Continuous variables were expressed as mean ± SD. A generalized linear model was used to compare groups adjusting for the matched pair. Categorical variables were expressed as percentages and compared with Cochran–Mantel–Haenszel test adjusting for the matched pair. P < 0.05 was considered to indicate statistical significance. Patients were stratified by the propensity score and a proportional hazards model was used to calculate adjusted hazard ratios.


Patient characteristics

Clinical, angiographic, and procedural characteristics are presented in Tables 1 and 2. After propensity-score matching, the two groups were well matched for clinical and angiographic characteristics. Clinical presentation was similar among groups. There were a number of procedural differences between groups. Patients undergoing IVUS-guided PCI underwent less direct stenting, more post-dilation, and had greater use of rotational atherectomy and cutting balloons. There was a significant difference in stent type between groups.

View this table:
Table 1

Patient characteristics

IVUS (n = 884)No IVUS (n = 884)P-value
Demographics and clinical history, n (%)
 Male gender613 (69.3%)619 (70.0%)0.75
 Age, years (mean ± SD)66.0 ± 11.665.6 ± 11.80.14
 Diabetes317 (35.9%)304 (34.4%)0.52
 Insulin requiring diabetes111 (12.6%)110 (12.4%)0.94
 Current smoker186 (21.0%)181 (20.5%)0.77
 Hypertension723 (81.8%)721 (81.6%)0.90
 Dyslipidaemia762 (86.2%)770 (87.1%)0.57
 Familial history432 (48.9%)441 (49.9%)0.66
 Chronic renal insufficiency110 (12.4%)112 (12.7%)0.89
 Prior myocardial infarction380 (43.0%)365 (41.3%)0.39
 Prior coronary bypass surgery206 (23.3%)198 (22.4%)0.60
 Prior percutaneous coronary intervention240 (27.1%)216 (24.4%)0.18
 Peripheral vascular disease145 (16.4%)142 (16.1%)0.85
Clinical presentation
 Stable angina213 (24.1%)222 (25.1%)0.58
 Unstable angina397 (44.9%)381 (43.1%)0.43
 Acute myocardial infarction (ST segment elevation)152 (17.2%)157 (17.8%)0.71
 Cardiogenic shock28 (3.2%)27 (3.1%)0.88
 Left ventricular ejection fraction (% ±SD)47 ± 1548 ± 130.44
View this table:
Table 2

Angiographic and procedural details

IVUS (n = 884)No IVUS (n = 884)P-value
Target vessel, n (%)
 Left main coronary artery26 (2.0%)30 (2.3%)0.51
 Left anterior descending artery427 (32.9%)433 (33.0%)
 Left circumflex320 (24.7%)305 (23.2%)
 Right coronary artery446 (34.4%)450 (34.3%)
 Saphenous vein graft75 (5.8%)84 (6.4%)
Lesion location, n (%)
 Ostial50 (3.9%)48 (3.7%)0.48
 Proximal595 (45.9%)613 (46.9%)
 Mid496 (38.3%)496 (37.9%)
 Distal149 (11.5%)146 (11.2%)
Lesion type (ACC/AHA classification), n (%)
 Type A59 (4.6%)65 (5.0%)0.56
 Type B944 (73.1%)953 (73.1%)
 Type C289 (22.4%)286 (21.9%)
 In-stent restenosis70 (5.4%)57 (4.3%)0.26
Procedural details, n (%)
 Treated lesionsa
  Mean, n ± SD1.70 ± 0.851.70 ± 1.11.0
  Median, n, interquartile range2.0 (1.0–2.0)1.0 (1.0–2.0)0.1
 Stents implanteda
  Mean, n ± SD1.48 ± 0.81.5 ± 0.90.22
  Median, n, interquartile range1.0 (1.0–2.0)1.0 (1.0–2.0)0.4
 Sirolimus-eluting stent832 (64.2)779 (59.4)0.001
 Paclitaxel-eluting stent464 (35.8)533 (40.6)0.001
 Stent diameter (mm ± SD)3.05 ± 0.43.09 ± 1.80.12
 Stent length (mm ± SD)20.73 ± 6.420.1 ± 6.90.47
 Direct stenting370 (28.7%)537 (41.1%)<0.001
 Post-dilatation343 (31.0%)219 (17.7%)<0.001
 Emergent intra-aortic balloon pumpa22 (2.5%)32 (3.6%)0.17
 Intravascular ultrasound1296 (100.0%)0
 Rotational atherectomy46 (3.5%)20 (1.5%)0.002
 Cutting balloon95 (7.3%)37 (2.8%)<0.001
 Glycoprotein IIb–IIIa Inhibitor usea157 (17.8%)163 (18.4%)0.70
 Angiographic success1278 (99.2%)1289 (98.9%)0.68
 Dissection5 (0.4%)10 (0.8%)0.17
 Abrupt closure3 (0.2%)3 (0.2%)0.93
 No reflow2 (0.2%)8 (0.6%)0.11
  • aPatient-based variables.

In-hospital outcomes

Angiographic success was similar between IVUS and No IVUS groups (99.2 vs. 98.9%; P = 0.68). There was also no significant difference in the rate of in-hospital all-cause mortality (1.2 vs. 2.3%; P = 0.11) and the need for emergent coronary bypass surgery (0.5 vs.0.8%; P = 0.21). There was, however, a significantly higher rate of Q-wave MI in patients undergoing DES placement with angiographic guidance alone (0.1 vs. 0.9%; P = 0.02). There were no significant differences between groups in the rates of in-hospital renal insufficiency and neurological events.

Intermediate outcomes (30 days)

At 30-day follow-up, a higher rate of definite stent thrombosis was seen in the No IVUS group (0.5 vs. 1.4%; P = 0.046). Patients in the IVUS group had a lower rate of MACE (2.8 vs. 5.2%; P = 0.01) driven primarily by increased death in the No IVUS group (1.7 vs. 3.3%; P = 0.03). There was no significant difference in the rates of cardiac death and Q-wave MI between groups. There was no advantage seen with IVUS guidance in TVR, however, a trend for less TLR was seen in the IVUS group (2.3 vs. 4.4%; P = 0.05).

Longer-term outcomes (12 months)

There was significantly more definite stent thrombosis in the No IVUS group (0.7 vs. 2.0%; P = 0.014) at 12 months. Figure 1 illustrates a significant difference in the rate of freedom from stent thrombosis between groups over 12 months. This was primarily driven by increased subacute stent thrombosis in the No IVUS group. There was no significant difference in late stent thrombosis between groups. There was a non-significant reduction in the rate of probable stent thrombosis in favour of the IVUS group (4.0 vs. 5.8%; P = 0.08). Clopidogrel compliance rates were similar among patients presenting with definite stent thrombosis in the IVUS and No IVUS groups (66.7 vs. 55.6%; P = 1.0). The average duration of clopidogrel therapy was non-significantly longer in the No IVUS group (P = 0.06). No significant difference in MACE at 12 month follow-up between groups was seen (14.5 vs. 16.2%; P = 0.33). Similar rates of death, cardiac death, and Q-wave MI were seen. Again there was no significant difference between groups in TVR but a trend toward less TLR remained in favour of the IVUS group (5.1 vs.7.2%; P = 0.07) (Table 3). The numbers needed to treat were 74 and 48 to prevent one stent thrombosis and one TLR, respectively. IVUS guidance was an independent predictor of freedom from cumulative stent thrombosis at 12 months (adjusted HR 0.5, CI 0.1–0.8; P = 0.02). Stent type was not predictive of definite stent thrombosis (SES: adjusted HR 1.0, CI 0.4–18.1; P = 0.3).

Figure 1

Kaplan–Meier curves illustrating freedom from stent thrombosis in IVUS and no IVUS groups over 12 months (P = 0.013).

View this table:
Table 3

Clinical outcomes of patients in IVUS and No IVUS groups

IVUS (n = 884)No IVUS (n = 884)P-Value
In-hospital outcomes, n (%)
 Death11 (1.2%)20 (2.3%)0.11
 Cardiac death5 (0.6%)11 (1.2%)0.13
 Q-wave myocardial infarction1 (0.1%)8 (0.9%)0.02
 Peri-procedural myocardial infarction132 (14.9%)148 (16.7%)0.29
 Coronary bypass surgery4 (0.5%)7 (0.8%)0.21
 Neurological event1 (0.1%)6 (0.7%)0.06
 Renal insufficiency29 (3.7%)28 (4.0%)0.68
30 Day outcomes, n (%)
 Major adverse cardiac events25 (2.8%)46 (5.2%)0.01
 Death15 (1.7%)29 (3.3%)0.03
 Cardiac death6 (0.7%)14 (1.6%)0.07
 Q-wave myocardial infarction6 (0.7%)12 (1.4%)0.16
 Target vessel revascularization10 (1.1%)17 (2.0%)0.18
 Target lesion revascularization6 (0.7%)15 (1.7%)0.05
 Cumulative stent thrombosis4 (0.5%)12 (1.4%)0.046
12 Month outcomes, n (%)
 Major adverse cardiac events128 (14.5%)143 (16.2%)0.33
 Death50 (5.7%)62 (7.1%)0.24
 Cardiac death16 (1.9%)24 (2.8%)0.19
 Q-wave myocardial infarction18 (2.1%)26 (3.1%)0.12
 Target vessel revascularization73 (8.5%)77 (9.1%)0.67
 Target lesion revascularization43 (5.1%)61 (7.2%)0.07
 Definite stent thrombosis6 (0.7%)18 (2.0%)0.014
 Probable stent thrombosis35 (4.0%)51 (5.8%)0.08
 Late definite stent thrombosis2 (0.2%)6 (0.7%)0.16


The present study evaluates the impact of IVUS guidance on clinical outcomes in patients undergoing DES implantation. Of most interest is the significant reduction in cumulative stent thrombosis and a trend toward less TLR at 12 months. The impact on stent thrombosis was primarily driven by reductions in the rate of subacute stent thrombosis in patients undergoing IVUS-guided coronary intervention. The study did not demonstrate any benefit of IVUS guidance in overall MACE and death. Nevertheless, IVUS guidance in this analysis was associated with a clinical benefit. We propose the procedural differences seen between groups (direct stenting, post-dilation and use of rotational atherectomy and cutting balloons) are driven by IVUS utilization. Lesion information gained on IVUS led to a different treatment approach which in turn may have impacted clinical outcomes. IVUS parameters predictive of stent thrombosis have been reported from our institution in both the BMS15 and DES10 eras. We have not reported IVUS parameters in the present study but have focused on the impact of IVUS guidance during DES implantation on clinical outcomes in ‘real world’ practice. To our knowledge, this is the first report to assess the clinical utility of IVUS guidance during coronary intervention with DES.

Subacute stent thrombosis was an initial major limitation of BMS implantation. Combination antiplatelet therapy with aspirin and a thienopyridine,1619 along with high-pressure stent deployment,20 significantly reduced this problem. Subacute stent thrombosis was reported to have an incidence of 0.9% in the modern BMS era.21 Critical insights into the pathophysiology of subacute stent thrombosis in the BMS era were gained from IVUS studies. Features on IVUS found to be associated with subacute stent thrombosis included stent under-expansion, malapposition, inflow/outflow disease, dissection, thrombus, and tissue prolapse.1315 These IVUS findings confirmed that the mechanisms underlying subacute stent thrombosis were mechanical and potentially treatable when identified. Despite these findings, there was no study in the BMS era which conferred an advantage of routine IVUS-guided PCI in preventing stent thrombosis, possibly related to its low frequency.

Though the focus in the DES era has shifted to late stent thrombosis, most series reporting DES thrombosis have shown that subacute stent thrombosis still outnumbers late events and hence remains an important entity.5 Studies pertaining to IVUS findings in patients presenting with stent thrombosis in the DES era are limited. Reduced minimal stent area and stent expansion along with greater residual disease have been correlated with stent thrombosis post-DES implantation.9,10 It is reasonable to assume that the underlying mechanisms of subacute stent thrombosis after both implantation of BMS and DES remain the same, predominantly mechanical. IVUS may allow the identification of these anomalies and lead to their subsequent treatment. The finding of reduced subacute stent thrombosis in the IVUS group in the present study is in keeping with this. The failure of IVUS guidance to impact late stent thrombosis suggests the predominance of non-mechanical aetiologies such as delayed endothelialization, late malapposition, and hypersensitivity. However, the study is underpowered to detect significant relationships with late events.

DES restenosis is predominantly focal and hence mechanical causes are thought to be contributory in the majority of cases. Stent under-expansion is the most important mechanical cause leading to DES restenosis. IVUS remains the best modality to assess stent architecture post-procedure. Correlation between post-procedural minimal stent area on IVUS examination and restenosis in patients undergoing SES implantation has been made from a number of IVUS studies.68 Inadequate lesion coverage has also been identified to be associated with edge restenosis with DES.22 Though DES have reduced restenosis rates, IVUS guidance during DES implantation can further potentiate this advantage by eliminating stent under-expansion and providing assessment of lesion length, thus ensuring complete lesion coverage. There exist no randomized trials studying IVUS guidance in PCI with DES in regard to restenosis. Studies from the BMS era suggest a benefit of IVUS guidance in reducing restenosis.11,12 Though only a trend, our findings suggest a benefit of IVUS guidance in the prevention of DES restenosis. This may have important implications in complex patient and lesion subsets.

Despite the significant reductions in definite stent thrombosis and the trend toward reduced TLR in patients undergoing IVUS-guided DES implantation, no impact on mortality or MACE was seen. Thus, the application of IVUS guidance may be best reserved for patients at increased risk of stent thrombosis and restenosis. The findings in the present study need to be confirmed in prospective, randomized trials adequately powered to seek out differences in these events.

This was a non-randomized, single-centre, and observational study. Propensity-score matching was performed to reduce the potential for bias in treatment selection. Despite the populations studied being well matched for baseline characteristics, particularly for the established predictors of stent thrombosis, there remains the potential for unaccounted variables to influence treatment selection. In this observational study, the decision to use IVUS guidance was at the discretion of the treating physician and despite the large number of operators performing the study procedures, operator level bias remains a limitation. To be eligible for inclusion in the study population, patients were required to have had IVUS to all treated lesions either pre- or post-intervention or both. The majority of patients had IVUS performed at least post-procedure. This study reflects the utilization of IVUS in routine practice at a high-volume centre in the current era of DES. Critical information gained from IVUS pre- (lesion assessment) and post-intervention (stent architecture) can both contribute to facilitate an optimal result. The low frequency of the primary endpoint of the study, definite stent thrombosis, did not allow for further subgroup analysis.


IVUS guidance during DES implantation has the potential to influence treatment strategy and reduce both DES thrombosis and the need for repeat revascularization.

Conflict of interest: none declared.



The above article uses a new reference style being piloted by the EHJ that shall soon be used for all articles.

View Abstract