Primary Stenting Versus Balloon Angioplasty in Occluded Coronary Arteries
The Total Occlusion Study of Canada (TOSCA)
Background—Balloon angioplasty (PTCA) of occluded coronary arteries is limited by high rates of restenosis and reocclusion. Although stenting improves results in anatomically simple occlusions, its effect on patency and clinical outcome in a broadly selected population with occluded coronary arteries is unknown.
Methods and Results—Eighteen centers randomized 410 patients with nonacute native coronary occlusions to PTCA or primary stenting with the heparin-coated Palmaz-Schatz stent. The primary end point, failure of sustained patency, was determined at 6-month angiography. Repeat target-vessel revascularization, adverse cardiovascular events, and angiographic restenosis (>50% diameter stenosis) constituted secondary end points. Sixty percent of patients had occlusions of >6 weeks’ duration, baseline flow was TIMI grade 0 in 64%, and median treated segment length was 30.5 mm. With 95.6% angiographic follow-up, primary stenting resulted in a 44% reduction in failed patency (10.9% versus 19.5%, P=0.024) and a 45% reduction in clinically driven target-vessel revascularization at 6 months (15.4% versus 8.4%, P=0.03). The incidence of adverse cardiovascular events was similar for both strategies (PTCA, 23.6%; stent, 23.3%; P=NS). Stenting resulted in a larger mean 6-month minimum lumen dimension (1.48 versus 1.23 mm, P<0.01) and a reduced binary restenosis rate (55% versus 70%, P<0.01).
Conclusions—Primary stenting of broadly selected nonacute coronary occlusions is superior to PTCA alone, improving late patency and reducing restenosis and target-vessel revascularization.
Primary use of coronary stents to reduce restenosis after coronary angioplasty in simple lesions is well supported by major clinical trials.1 2 Use of stents in complex lesions, although now routine, has not been subjected to equal scrutiny.
Nonacute occlusions are complex not only from the occlusion itself but also from the frequent coexistence of extended length, contiguous diffuse disease, and thrombus. The late results of balloon angioplasty (PTCA) for coronary occlusions are poor.3 Their tendency to reocclude may prompt selection of surgical revascularization or medical therapy as initial treatment. Importantly, the failure of angioplasty to provide a high rate of sustained patency has hindered trials testing routine late recanalization of infarct-related arteries (the “open-artery” hypothesis).4
Although stenting of highly selected simple coronary occlusions can improve results, it is untested in larger, broadly selected cohorts.5 6 We report final angiographic and 6-month clinical outcomes from a controlled randomized trial testing the heparin-coated Palmaz-Schatz stent (Cordis, Johnson & Johnson) in a diverse spectrum of patients with nonacute coronary occlusions.
Participating Centers and Sponsorship
The Total Occlusion Study of Canada (TOSCA) was a multicenter prospective randomized controlled trial of primary stenting in native coronary artery occlusions conducted at 18 centers in accordance with local institutional guidelines after review board approval (Appendix 1). Study design and rationale have been reported previously.7 The primary purpose was to determine whether routine stent placement improves late complete patency (TIMI flow grade 3) of recanalized coronary artery occlusions. The protocol was developed and implemented by the investigators and supported by a grant from Cordis, Johnson & Johnson.
Patients 18 to 80 years old undergoing clinically indicated coronary interventions were eligible provided that 1 target segment met the study definition of occlusion: a high-grade native coronary stenosis accompanied by TIMI grade 0 or 1 antegrade flow. The target segment was required to be able to accommodate a balloon of ≥3.0-mm diameter and to be suitable for stent placement. Exclusion criteria were (1) <72 hours since onset of ST-segment elevation, (2) extensive lesion-related thrombus (TIMI thrombus grade 3 or 4, Appendix 2), (3) occlusions previously revascularized by patent bypass grafts, (4) uncontrolled heart failure or shock, (5) patient unwilling or unsuitable for protocol-required 6-month angiography, (6) patient of child-bearing potential, and (7) inability to cross occlusion with guidewire. All patients provided written informed consent.
Age of occlusion was determined by the interval from the last episode of acute coronary syndrome or infarction consistent with the location of the occlusion. Randomization, stratified by center and age of occlusion (≤6 weeks or >6 weeks/uncertain) occurred after the occlusion was crossed with a guidewire but before initial dilation. Patients undergoing attempted recanalization of coronary occlusions but not enrolled were entered into a previously reported registry.7
Clinical Care and Monitoring
Operators were instructed to optimize PTCA results using a balloon of ≥3.0-mm nominal diameter and a balloon-to-artery ratio of ≥1. Prolonged inflations of ≥15 minutes’ duration were required before crossover to stent. However, because significant residual stenosis and dissection frequently characterize initial PTCA results in occlusions but seldom result in acute ischemic events, crossover was recommended only if a ≥70% visual stenosis or major dissection causing impaired flow or ischemia (chest pain or ECG changes) was present.7 8
The stent arm used the Carmeda process heparin-coated 15-mm-long PS-153 Palmaz-Schatz coronary stent premounted on a sheathed delivery system.9 Complete coverage of the occluded segment and adjacent lesion was required, using as many contiguous stents as necessary. Stent expansion at ≥14 bar was recommended.
Treatment of lesions in major segments proximal or distal to the occlusion with >50% visual stenosis was required in both arms. Use of stents in noncontiguous lesions or other vessels was not determined by protocol. All patients received aspirin 325 mg/d beginning before the procedure. Ticlopidine administration was left to operator discretion.
Clinical events and ECGs were routinely recorded at 24 hours, hospital discharge, 1 month, and final angiography. Creatine kinase (CK) and CK-MB were obtained at baseline, 8 hours, and 18 to 24 hours after the procedure.
After intracoronary nitroglycerine, ≥2 orthogonal views were reproduced at baseline, after the initial procedure, and at final 6-month angiography. Angiograms performed before 6 months qualified as final if (1) failure of patency was present, (2) the interval from enrollment to angiography was >16 weeks, or (3) the findings led to target-vessel bypass grafting.
Analysis of lumen dimensions and flow was performed independently in a dedicated core laboratory using the ImageCom system (Quinton). Target lesion work length, defined as the length of contiguous target segment exposed to balloon inflation, was measured by use of images recorded throughout the procedure.7 Minimum lumen diameter (MLD) was derived from analysis of the entire treated segment. With rigorously selected disease-free proximal-only reference segments used, restenosis was defined as >50% diameter stenosis at the point of MLD. Reference values were deemed unmeasurable in ostial and near-ostial occlusions.
Failure of sustained complete target-vessel patency at final angiography, defined as TIMI flow <3 as determined by the Core Angiographic Laboratory, was the primary end point. To prevent counting of nonepicardial causes of reduced flow, an accompanying target-vessel stenosis of >70% was also required. Target segment MLDs and restenosis rates were secondary end points.
Clinical end points were (1) repeat target-vessel revascularization (TVR, defined as any attempted percutaneous or surgical revascularization of the target vessel after the initial procedure) and (2) a composite end point composed of any revascularization (of any coronary vessel), myocardial infarction (MI), or death. MI was defined as any CK-MB elevation above local laboratory normal range and was classified according to the magnitude of peak CK-MB elevation (≤5 times or >5 times normal). A 3-member committee adjudicated all events.
All end-point analyses were performed on an intention-to-treat basis. Prespecified subgroups included patients with age of occlusion ≤6 weeks versus >6 weeks (or uncertain) and baseline TIMI flow grade 0 versus 1. Discrete variables are provided as counts and percentages and compared by 2-tailed Fisher’s exact test. Normally distributed continuous variables are expressed as mean±SD, and abnormally distributed continuous variables as medians with interquartile ranges.
From March 1996 through May 1997, 1118 patients were screened for inclusion.7 Of these, 380 (34%) did not meet eligibility criteria (including 101 excluded for inability to cross with a guidewire), 328 (29%) were eligible but not enrolled, and 410 were enrolled. Baseline characteristics of the treatment groups were well matched (Tables 1⇓ and 2⇓).
Median target lesion work length was 30.5 mm (range, 20.9 to 43.7 mm) (Table 3⇓). Crossover during the initial procedure occurred in 20 patients (9.6%) assigned to PTCA and 8 (4.0%) assigned to stent. On average, 2.0 contiguous target lesion stents were deployed per stent-assigned patient. Both operator-estimated reference vessel diameter and mean nominal balloon size were similar in the 2 groups, although maximum inflation pressures were higher in stent patients. Use of abciximab was infrequent (3%). Apart from ticlopidine, which was administered more frequently to stent patients (93% versus 57%, P<0.01), discharge medications were similar.
By final angiography, TVR had occurred in 15.4% of patients assigned to PTCA and 8.4% assigned to stent, a relative reduction of 45% (P=0.03, Figure 1⇓, Tables 4⇓ and 5⇓). TVR rates diverged early after the index procedures. No instances of acute or subacute stent thrombosis were apparent. Composite adverse cardiac event rates were similar (23.6% PTCA versus 23.3% stent, P=NS). Although a trend toward reduced revascularization of any vessel was seen in the stent group, more MIs occurred in this group. The majority of infarctions, however, were characterized by peak CK-MB elevations of ≤5 times normal, and none resulted in new Q waves. None of the 20 patients who experienced MI in the periprocedural period (4 PTCA, 16 stent) suffered recurrent MI or death during follow-up. No cardiac deaths occurred.
Patency and Quantitative Angiography
Angiographic follow-up was obtained in 392 patients (95.6%), refused in 15 (6 PTCA, 9 stent), and not possible in 3 (1 PTCA, 2 stent). At final angiography, each reduced TIMI flow grade (0, 1, or 2) was observed less frequently in the stent group than in the PTCA group (Table 6⇓). Therefore, failure to sustain patency was present in 21 of 191 patients assigned to stent (10.9%) versus 39 of 201 patients assigned to PTCA (19.5%), a 44% relative reduction (Figure 2⇓). Although operator-estimated mean vessel size, target-segment location, and final mean balloon size were very similar in PTCA and stent patients, core laboratory–determined mean baseline proximal reference diameters were 0.16 mm larger in stent patients. However, no relationship between this variable and failed patency was present on univariate or multivariate logistic modeling.
Initial postprocedural and final follow-up MLDs were significantly larger in the stent group, despite greater late loss (Table 7⇓, Figure 3⇓). Angiographic restenosis (>50% diameter stenosis) derived by use of proximal reference segments was present in 55% of patients assigned to stent and 70% assigned to PTCA (P<0.01).
Reocclusion after PTCA of symptomatic coronary occlusions is reported in 16% to 34% and restenosis in 45% to 72%.3 5 6 10 11 Results in asymptomatic patients from preliminary open-artery trials have been equally discouraging. In the Total Occlusion Post-MI Intervention Study (TOMIIS), for example, sustained patency was achieved in just 43% of the intervention arm.12
In the wake of nonrandomized reports,13 14 15 2 randomized trials of stents in coronary occlusions with ≥100 patients have been published. Both the Stents in Chronic Coronary Occlusion (SICCO) and Gruppo Italiano di Studio sullo Stent nelle Occlusioni Coronariche (GISSOC) trials evaluated stents in highly selected populations excluding long lesions, other complex features, and patients with recent infarction.5 6 Both demonstrated that stents reduce restenosis (32% versus 72% and 32% versus 68%, respectively). GISSOC also demonstrated reduced reocclusion with stents (8% versus 34%), although only stent patients received routine warfarin therapy.
Enrollment criteria for TOSCA were much broader. Patients with long occlusions, occlusions contained within diffuse segments of disease, moderate lesion-related thrombus, recent infarction, and other complex features that might diminish the effectiveness of stents were not excluded. Unlike SICCO and GISSOC, randomization occurred immediately after the guidewire was advanced across the occlusion. Thus, operators were committed to enrollment and treatment assignment before initial dilation permitted detailed assessment of target segment length, distal vessel quality or dissection, or unrecognized thrombus.
The choice of failed patency as the primary end point was based on several considerations. First, because failed patency occurs less frequently than restenosis, only a study powered to detect improved patency would be adequately powered to examine both. Second, the significance of angiographic restenosis in vessels previously occluded is unclear, given the inevitability of previous downstream infarction, developed collaterals, or both. Third, because strong circumstantial evidence links patency (but not restenosis) to improved post-MI outcome, prospective trials testing the open-artery hypothesis await therapies proven to enhance sustained patency.4 16
Our primary end-point results are clear. Compared with PTCA, stenting of occluded coronary arteries results in a marked improvement in sustained patency. Primary end-point results in the 4 prespecified subgroups, although they demonstrate heterogeneity of the point estimate, are compatible with general benefit. CIs cross unity only in subgroups with <150 patients.
Stenting, however, did not eliminate patency failures. Nearly 11% of patients assigned to stenting demonstrated failed patency, and 8.3% (versus 13.5% of PTCA patients, P=0.15) developed complete reocclusion (TIMI 0 or 1 flow). New stent designs and periprocedural or long-term glycoprotein IIb/IIIa antagonists may have roles yet to be defined. Interestingly, patients assigned to PTCA had lower rates of failed patency than previously reported, despite inclusion of complex lesions, rigorous crossover rules, and low crossover rates.
Our convention of using disease-free proximal reference segments as the sole denominator for all quantitative analyses was driven by the inability to obtain distal reference segments in occluded vessels. Given the natural taper of coronary arteries, this approach would be expected to increase percent stenosis values compared with analyses using interpolated references. Nevertheless, we observed large values of absolute late loss in both arms attributable to the additive effects of large initial gain, extended lesion length, thrombus, and frequent use of multiple stents.17 18 19 Furthermore, although PTCA restenosis rates in TOSCA, SICCO, and GISSOC were remarkably similar, we found less difference in stent and PTCA restenosis rates in our complex population, despite the possible advantage conveyed by slightly larger proximal reference segments in the stent arm. As with patency, stents with greater length and improved uniformity of radial support may prove particularly effective in this challenging setting.17 20
In contrast to previous studies, stenting reduced TVR before the confounding effect of protocol-specified angiography.1 2 6 The immediate divergence of the curves is consistent with previous observations that early reocclusion is common.8
Overall, TVR was infrequent (11.9%), despite frequent angiographic restenosis. However, some stable patients may have had repeat revascularization deliberately delayed pending protocol angiography. Continued follow-up is therefore essential to determine the durability of our 6-month findings.
Adverse Cardiovascular Events
Although composite adverse cardiac event rates were similar in the stent and PTCA groups, certain patterns emerged. Revascularization procedures in any vessel tended to be more frequent in the PTCA group. This trend, however, was driven entirely by procedures involving the target segment (Table 5⇑).
As defined in the protocol, MI was more frequent in the stent group (11.9% versus 3.8%, P=0.01). We screened for periprocedural MI with routine sequential enzyme samples and used a very inclusive definition (any elevation of CK-MB, regardless of total CK). Most MIs were indeed periprocedural and small. The mechanisms for periprocedural infarction after coronary occlusion angioplasty are enigmatic. Early closure of the target lesion is probably not responsible; both early TVR and late failed patency were more common in the treatment group with fewer MIs. More likely, distal thromboembolism or obstruction of side branches arising near target occlusions accounts for this phenomenon. The significance of small periprocedural CK-MB elevations after stent placement is controversial.21 22 23 Regardless, recent data suggest that routine use of abciximab would have largely abolished group differences in rates of periprocedural infarction.24
Although our angiographic follow-up rate was high, 4.4% of the study population was not reevaluated angiographically. A conservative remedy would be to impute the overall combined rate of failed patency (15.3%) to all patients without final angiography. This approach does not, however, materially affect our results (P=0.03).
As in other device trials, treatment assignment could not readily be blinded. Thus, a bias influencing subsequent patient management could have existed. The study design does not permit conclusions regarding the importance of the heparin coating of the investigational stents. Similarly, our study did not address occlusions that cannot be crossed with a guidewire, a difficulty encountered in 9% of patients screened for TOSCA.
We have demonstrated that primary stent placement improves late patency and reduces restenosis in a broadly selected population with occluded coronary arteries. Stenting resulted in a reduced need for repeat TVR but no net difference in overall adverse cardiac events. Although continued clinical follow-up is necessary, these data support routine stenting when occluded coronary arteries are reopened. In contrast to previous stent trials, TOSCA had few criteria to exclude lesions perceived as unfavorable. These data, therefore, provide new insights regarding the utility of stents in the treatment of complex lesions frequently encountered in current practice.
Study chair: Dr Christopher Buller. Executive committee: Dr Christopher Buller, Dr Vladimir Dzavik (cochair), Dr Merrill Knudtson, Dr Jean-François Marquis. Coordinating center: Interventional Cardiology Research, Vancouver General Hospital, Rebecca Fox, MSc. Data center: EPICORE Center, University of Alberta: Dr Koon Teo, director; Paula Priest, data coordinator; Diane Catellier, PhD, biostatistician. Angiographic core laboratory: Dr G.B. John Mancini, director; Joseph Li, senior technician. Events committee: Dr Ronald Carere, chair; Dr Koon Teo; Dr Anoop Chauhan. Data Safety and Monitoring Committee: Ruth Milner, chair; Dr Christopher Granger; Dr Paul Armstrong.
Participating centers: Vancouver General Hospital: Dr Christopher Buller, Rebecca Fox. Toronto Hospital: Dr Charles Lazzam, Arlene Carter. Hôpital Laval, Ste-Foy: Dr Gerald Barbeau, Marie-Mai Lariviere. St Paul’s Hospital, Vancouver: Dr Ronald Carere, Margot Wilson. Kokuryo Toyohashi Higashi Hospital, Japan: Dr Takahiko Suzuki, Dr Hiroaki Hosokawa. University of Alberta Hospital, Edmonton: Dr Vladimir Dzavik, Cheryl Kee. Foothills Hospital, Calgary: Dr Todd Anderson, Lana Shewchuk. Royal Columbian Hospital, New Westminster: Dr Robert I.G. Brown, Marjorie Colclough. Royal Alexandra Hospital, Edmonton: Dr Neil Brass, Donna Daniec. Green Lane Hospital, New Zealand: Dr James Stewart, Sarah Heath. Queen Elizabeth II Health Sciences Center, Halifax: Dr Lawrence Title, Kim Foshay. University of Ottawa Heart Institute, Ottawa: Dr Jean-François Marquis, Sharon Kearns. Sunnybrook Health Science Center, Toronto: Dr Eric Cohen, Lynn Balleza. Winnipeg Health Sciences Center: Dr John Ducas, Usha Schick. New Brunswick Heart Center, Saint John: Dr Robert MacDonald, Hallie Bass. London Health Sciences Center, London: Dr David Almond, Joanne White. Duke Medical Center, Durham, NC: Dr James Zidar, Steve Sawchak. Mayo Clinic, Rochester, Minn: Dr Peter Berger, Jeanette Ramaker.
TIMI Thrombus Grades
Grade 0: No cineangiographic characteristics suggestive of thrombus.
Grade 1: Angiography demonstrates characteristics such as reduced contrast density, haziness, irregular lesion contour, or a smooth convex meniscus at the site of total occlusion suggestive but not diagnostic of thrombus.
Grade 2: Definite thrombus with greatest dimensions ≤1/2 vessel diameter.
Grade 3: Definite thrombus with greatest linear dimension >1/2 but <2 vessel diameters.
Grade 4: As in grade 3 but with largest dimension ≥2 vessel diameters.
Funding was provided through an unrestricted grant from Cordis, Johnson & Johnson, Miami Lakes, Fla.
Reprint requests to Christopher E. Buller, MD, FRCPC, FACC, 865 W 10th Ave, Vancouver, BC, Canada V5Z 1L7.
- Received December 31, 1998.
- Revision received April 19, 1999.
- Accepted April 22, 1999.
- Copyright © 1999 by American Heart Association
Fischman D, Leon M, Baim D, Schatz R, Savage M, Penn I, Detre K, Veltri L, Ricci D, Nobuyoshi M, Cleman M, Heuser R, Almond D, Teirstein P, Fish D, Columbo A, Brinker J, Moses J, Shaknovich A, Hirshfeld J, Bailey S, Ellis S, Rake R, Goldberg S. A randomized comparison of coronary-stent placement and balloon angioplasty in the treatment of coronary artery disease. N Engl J Med. 1994;331:496–501.
Serruys P, deJaegere P, Kiemeneij F, Macaya C, Rutsch W, Heyndrickx G, Emanuelsson H, Marco J, Legrand V, Materne P, Belardi J, Sigwart U, Colombo A, Goy J-J, VandenHeuvel P, Delcan J, Morel M-A. A comparison of balloon-expandable-stent implantation with balloon angioplasty in patients with coronary artery disease. N Engl J Med. 1994;331:489–495.
Ivanhoe R, Weintraub W, Douglas J, Lembo N, Furman M, Gershony G, Cohen C, King S III. Percutaneous transluminal coronary angioplasty of chronic total occlusions. Circulation. 1992;85:106–115.
Sirnes A, Golf S, Myreng Y, Mostad, Emanuelsson H, Albertsson, Brekke M, Mangschau A, Endresen K, Kjekshus J. Stenting in Chronic Coronary Occlusion (SICCO): a randomized, controlled trial of adding stent implantation after successful angioplasty. J Am Coll Cardiol. 1996;28:1444–1451.
Dzavik V, Carere R, Teo K, Knudtson M, Marquis J-F, Buller CE, for the TOSCA Investigators. An open design, multicentre, randomized trial of percutaneous transluminal coronary angioplasty versus stenting, with a heparin-coated stent, of totally occluded coronary arteries: rationale, trial design and baseline patient characteristics. Can J Cardiol. 1998;14:825–832.
Buller C, Penn I, Ricci D, Ray S, Fox R, Mancini G. Recoil, dissection, and reocclusion early following PTCA of total coronary occlusions. Can J Cardiol. 1994;10:135C. Abstract.
Serruys P, Emanuelsson H, van der Giessen W, Lunn A, Kiemeney F, Macaya C, Rutsch W, Heyndrickx G, Suryapranata H, Legrand V, Goy J, Materne P, Bonnier H, Morice M, Fajadet J, Belardi J, Colombo A, Garcia E, Ruygrok P, de Jaegere P, Morel M, for the BENESTENT-2 Study Group. Heparin-coated Palmaz Schatz stents in human coronary arteries: early outcome of the Benestent-II pilot study. Circulation. 1996;93:412–422.
Ishizaka N, Issiki T, Saeki F, Ishizaka J, Ikari Y, Abe J, Soumitsu Y, Hashimoto H, Masaki K, Yamaguchi T. Angiographic follow-up after successful percutaneous coronary angioplasty for chronic total coronary occlusion: experience in 110 consecutive patients. Am Heart J. 1994;127:8–12.
Violaris A, Melkert R, Serruys P. Long-term luminal renarrowing after successful elective coronary angioplasty of total occlusions. Circulation. 1995;91:2140–2150.
Dzavik V, Beanlands D, Davies R, Leddy D, Marquis J-F, Teo K, Ruddy T, Burton J, Humen D. Effects of late percutaneous transluminal coronary angioplasty of an occluded infarct-related coronary artery on left ventricular function in patients with a recent (<6 weeks) Q-wave acute myocardial infarction (Total Occlusion Post-Myocardial Infarction Intervention Study [TOMIIS]: a pilot study). Am J Cardiol. 1994;73:856–861.
Ozaki Y, Violaris A, Hamburger J, Melkert R, Foley D, Keane D, deFeyter P, Serruys P. Short-and long-term clinical and quantitative angiographic results with the new, less shortening Wallstent for vessel reconstruction in chronic total occlusion: a quantitative angiographic study. J Am Coll Cardiol. 1996;28:354–360.
Violaris AG, Melkert R, Herrman J-PR, Serruys PW. Role of angiographically identifiable thrombus on long-term luminal renarrowing after coronary angioplasty. Circulation. 1996;93:889–897.
Califf RM, Fortin DF, Frid DJ, Harlan WR, Ohman M, Bengtson JR, Nelson CL, Tcheng JE, Mark DB, Stack RS. Restenosis after coronary angioplasty: an overview. J Am Coll Cardiol. 1991;17:2B–13B.
Califf R, Abdelmeguid A, Kuntz R, Popma J, Davidson C, Cohen E, Kleiman N, Mahaffey K, Topol E, Pepine C, Lipicky R, Granger C, Harrington R, Tardiff B, Crenshaw B, Bauman R, Zuckerman B, Caitman B, Bittl J, Ohman E. Myonecrosis after revascularization procedures. J Am Coll Cardiol. 1998;31:241–251.
Kugelmass A, Cohen D, Moscucci M, Piana R, Senerchia C, Kuntz R, Baim D. Elevation of the creatine kinase myocardial isoform following otherwise successful directional coronary atherectomy and stenting. Am J Cardiol. 1994;15:748–754.
Saucedo J, Popma J, Mehran R, Boucher T, Abizaid A, Curran M, Mastoor M, Kent K, Pichard A, Satler L, Leon M. Lack of association of intermediate CPK-MB elevation and late mortality in patients treated with intracoronary stents. Am J Cardiol. 1998;31:215. Abstract.