Author + information
- Received November 5, 1999
- Revision received November 10, 2000
- Accepted December 13, 2000
- Published online March 15, 2001.
- Steven L Goldberg, MD, FACC∗,* (, )
- Arthur Loussararian, MD, FACC∗,
- Joseph De Gregorio, MD, FACC†,
- Carlo Di Mario, MD†,
- Remo Albiero, MD† and
- Antonio Colombo, MD, FACC†
- ↵*Reprint requests and correspondence:
Dr. Steven L. Goldberg, Division of Cardiology, University of Washington School of Medicine, Box 356115, 1959 NE Pacific Street, Seattle, Washington 98195-6115
This study was performed to investigate the causes of diffuse and aggressive intra-stent restenosis.
Although restenosis is usually considered to be a dichotomous variable, there is clinical relevance to the severity of restenosis. It is not known which variables are predictive of diffuse or aggressive intra-stent restenosis.
A consecutive series of 456 coronary lesions with in-stent restenosis was evaluated for the type of restenosis using quantitative coronary angiography. Restenosis was defined as ≥50% diameter stenosis at follow-up angiography, diffuse restenosis as a follow-up lesion length ≥10 mm and aggressive restenosis as either an increase in lesion length from the original lesion or a restenotic narrowing tighter than the original. Clinical, anatomic and procedural characteristics were evaluated for lesions associated with these types of restenosis.
Diffuse restenosis was associated with a smaller reference artery diameter, longer lesion length, female gender, longer stent length and the use of coil stents. Aggressive restenosis was more common in women, with the use of Wallstents and with long stent to lesion length ratios. Aggressive restenosis occurred earlier and was more closely associated with symptoms and myocardial infarctions than nonaggressive restenotic lesions.
Markers for diffuse restenosis were also important markers for the presence of any restenosis. A long stent to lesion length ratio is an important marker for aggressive restenosis. When severe forms of in-stent restenosis occur, they tend to present earlier and with more symptoms, including myocardial infarction. More careful consideration of the type of in-stent restenosis may aid in identifying when alternative strategies may be useful.
Although stents reduce the incidence of angiographic and clinical restenosis of percutaneous coronary interventions, they do not abolish restenosis altogether. Recently, it has become apparent that restenosis after stenting may at times be refractory to repeat balloon angioplasty (1,2). It has been observed that when intra-stent restenosis presents focally, balloon angioplasty is usually successful; however, when a diffuse restenosis occurs, recurrent restenosis is likely (1,3). This study investigated the causes of diffuse and other patterns of restenosis after stenting.
All lesions undergoing stent implantation at Centro Cuore Columbus in Milan, Italy, are entered into a dedicated database (Filemaker Pro, Claris) that includes baseline clinical features, lesion characteristics, procedural variables and outcomes. All patients with successful procedures are asked to return in four to six months for a follow-up angiogram, at which time they are evaluated for clinical events and symptoms. Quantitative coronary angiography was performed on baseline and follow-up angiograms. The definitions related to restenosis are shown in Table 1. Angiographic restenosis was defined as ≥50% diameter stenosis on the follow-up angiogram. A diffuse restenosis was defined as a follow-up lesion length ≥10 mm or a total occlusion. A proliferative restenosis was defined as an increase in lesion length from the time of the original intervention to the follow-up angiogram. An aggressive restenosis was defined as either a proliferative restenosis or a restenotic narrowing that was tighter than the original narrowing. A previous report from this institution identified the characteristics predictive of angiographic restenosis (4). For this analysis, characteristics were evaluated for predictors of diffuse restenosis and aggressive restenosis. The evaluated characteristics are listed in Table 2and can be organized as baseline clinical and angiographic features as well as procedural characteristics.
Quantitative coronary angiography
Quantitative coronary angiography was performed using computerized software (Cardiovascular Measurement System, CMS, MEDIS, The Netherlands) with a contrast-filled catheter of known diameter as a scaling device. The mean reference artery diameter was interpolated from the proximal and distal reference segments. The lesion length was measured from “shoulder to shoulder.” When multiple narrowings were present in a single artery, the lesions were considered to be separate when occurring in different segments of the artery (proximal, mid or distal) with at least 10 mm of normal artery interposed between them. Otherwise the lengths of the segments were added together and the lesion was considered single. When a dissection occurred, or a new narrowing appeared to develop after the initial balloon inflation, only the original lesion length was used. The minimum lumen diameters (MLDs) before intervention, after final balloon expansion of the stent and at follow-up were measured. Acute gain, late loss and net gain were calculated (5). When a lesion was totally occluded, the lesion length was measured after opening the occlusion. When this was not possible, a length was ascribed to the narrowing that reflected the mean measured length of totally occluded stenoses for the population overall (20 mm).
Stent length was calculated as the sum of the nominal lengths of each implanted stent. Stents were characterized as being: 1) tubular (Palmaz-Schatz, Cordis/Johnson & Johnson Interventional, Warren, New Jersey; AVE GFX, Arterial Vascular Engineering, Inc., Santa Rosa, California; Multilink, Guidant/Advanced Cardiovascular Systems, Santa Clara, California; NIR, Medinol Ltd., Tel Aviv, Israel; BeStent, Medtronic Interventional, San Diego, California; Pura and Pura-Vario, Devon Medical, Hamburg, Germany); 2) coil (Gianturco-Roubin Flexstent, Cook Interventional, Bloomington, Indiana; Gianturco-Roubin 2, Cook Interventional; Wiktor, Medtronic Interventional Inc., San Diego, California; Cordis stent, Cordis/Johnson & Johnson, Warren, New Jersey); or 3) mesh (Wallstent or Magic Wallstent, Schneider Europe, Lausanne, Switzerland).
Categorical variables were assessed by chi-square analysis. Normally distributed continuous variables are reported as mean ± standard deviation and compared by analysis of variance followed by the Scheffe F test. Variables without normal distributions are presented as median values and were evaluated with the Mann-Whitney Uor the Mantel-Cox log-rank test. A p value <0.05 was considered statistically significant.
The association between the restenosis outcomes (diffuse restenosis or aggressive restenosis) and the potential predictors was assessed via logistic regression methods. To allow for correlation of type of restenosis within the same patient, the random effect of patient on lesions was controlled for in the logistic models using the generalized estimating equation methods of Zeger and Liang (6).
Factors that were biologically relevant and statistically significant in a univariate analysis were included in the multivariate analysis. Backward elimination was used to eliminate potential predictors. The univariate and multivariate odds ratios (ORs) and their corresponding 95% confidence intervals (CIs) are given. The unit measurements used to calculate the ORs were 1 mm for MLD and reference lumen diameter, 10 mm for lesion length and stent length, 1 mm2for intra-stent cross-sectional area and 10 years for age. Outcomes specific to lesions, such as the length of restenosis, were evaluated on a per-lesion basis. Outcomes relating to patients, such as death or myocardial infarction (MI), were evaluated on a per-patient basis.
From 1992 to 1997, 1,816 or 70% of stented lesions were evaluated with follow-up angiography. Of these, 456 had a follow-up stenosis diameter ≥50%, providing an overall angiographic restenosis rate of 25%. A previous report from this laboratory revealed the following characteristics as predictive of intra-stent restenosis: 1) a smaller final intra-stent cross-sectional area; 2) a longer total stent length; 3) a smaller reference artery lumen diameter; 4) the presence of a dissection; 5) older age and 6) prior history of coronary artery bypass graft surgery (4).
Diffuse in-stent restenosis
Diffuse restenosis was seen in 286 of the 456 restenotic narrowings (63%), with 84 of these presenting at follow-up as total occlusions (18% of all in-stent restenotic narrowings or 4.6% of all stented lesions). Univariate and multivariate predictors of diffuse restenosis are presented in Tables 2 and 3⇓respectively. Diffuse restenosis was associated with a smaller reference artery diameter, smaller baseline MLD, longer lesion length, female gender and a smaller final minimal lumen diameter. The stent type used, in particular coil design or Wallstent, was also associated with diffuse restenosis. By multivariate logistic regression, a longer baseline lesion length, a smaller final result achieved and the use of coil stents remained independent predictors of diffuse in-stent restenosis.
When restenosis occurred inside Wiktor stents it was diffuse 53% of the time, similar to the 58% incidence of noncoil stents. Gianturco-Roubin II stents, however, were associated with diffuse restenosis 91% of the time (p = 0.003 compared with Wiktor stents).
Technical features associated with achieving larger intra-stent luminal results were not associated with diffuse restenosis. Final balloon size used to expand the stent, balloon-to-artery ratio or inflation pressures used were not significantly different between diffuse and focal in-stent restenosis. Diffuse restenosis was associated with smaller final luminal dimensions compared with focally restenotic lesions. There were 358 patients with more than one lesion treated. Similar to other centers’ findings (7), we found that when restenosis occurred in one of a patient’s lesions, other lesions within that individual were more likely to have restenosis (54% vs. 18%) when the first lesion did not demonstrate restenosis (p < 0.001, OR 5.3). In the 118 instances when two or more lesions were restenotic in the same patient, there was a strong correlation with the type of restenosis (diffuse or focal) between the two lesions, with an OR of 3.50 (CI 1.45 to 8.45, p = 0.005) that if one lesion was diffuse the other was more likely to be diffuse.
Predictors of aggressive restenosis
An aggressive restenosis process was defined as either: 1) an increase in lesion length; or 2) a decrease in MLD at the time of in-stent restenosis compared with baseline. Lesions with aggressive restenosis had greater late loss (2.2 ± 0.7 vs. 1.9 ± 0.6, p < 0.0001), despite lesser acute gains during the intervention (2.1 ± 0.7 vs. 2.4 ± 0.6, p < 0.0001). Predictors of an aggressive restenosis process are shown in Tables 2 and 4. ⇓Aggressive in-stent restenosis was more common in women, in shorter lesions and with larger baseline MLDs. The use of Wallstents, and long stent to lesion length ratios, were associated with aggressive restenosis as were less optimal results identified by intravascular ultrasound. By multivariate logistic regression analysis controlling for person effect, we found that female gender, a large baseline MLD, a shorter baseline lesion length and a greater stent to lesion length ratio were predictors of aggressive restenosis.
Increase in lesion length or decrease in lumen diameter
There was a correlation between the components of aggressive restenosis, with proliferative lesions generally tighter than nonproliferative lesions (57% vs. 34%, p < 0.0001). Aggressive restenosis occurred earlier, and was earlier still when the lesion was tighter and longer (Fig. 1). Patients with aggressive restenosis were more likely to be symptomatic than those without an aggressive restenosis (47% vs. 35%, p = 0.05). Myocardial infarctions before follow-up angiography were more common with aggressive restenosis (6% vs. 2%, p = 0.03) and were particularly prevalent when the restenotic narrowing was both tighter and longer (Fig. 2).
A report from this laboratory identified several characteristics that predicted intra-stent restenosis (4). However, not only the presence but also the severity of in-stent restenosis carries clinical significance. For example, the length of the in-stent restenosis lesion has been identified as a marker for the development of recurrent restenosis after repeat percutaneous intervention (1,2,8,9). A long in-stent restenotic lesion is most likely to occur when the baseline lesion length is long. However, a severe restenosis may occur even in short original lesions, so alternative types of severe restenosis less influenced by long baseline lesion lengths were also evaluated in this study. This study evaluated the baseline and procedural characteristics predictive of more clinically severe restenoses. In this study an aggressive restenosis was increasingly likely to present earlier and as a MI, compared with nonaggressive restenotic lesions, supporting a clinical relevance to this definition.
Diffuse and aggressive restenosis predictors
In addition to the baseline lesion length, the presentation of a diffuse restenosis was predicted by a smaller final luminal result achieved inside the stent and by the use of specific coil stents.
An aggressive restenosis, defined as the development of a longer or tighter lesion at follow-up compared with baseline, was predicted using multivariate logistic regression analysis by the ratio of the stent length to the baseline lesion length. The strong relationships between a large MLD at baseline, or a short baseline lesion length, and the development of an aggressive restenosis are artifacts of the definition of aggressive restenosis. A very small lumen diameter to start with (such as a baseline total occlusion) has very little room to become tighter, whereas a large baseline lumen is more likely to be at least slightly smaller when restenosis occurs. Similarly, a short lesion length was more likely to be longer on follow-up. Women were more likely to develop an aggressive restenosis, suggesting important person-related effects.
Some have proposed that aggressive stent implantation techniques using oversized balloons and/or high-pressure inflations may contribute to a more aggressive restenotic process (10–12). In this study there was little evidence that inflation pressures or balloon sizes used to expand the stents were related to worsened restenosis. Instead, the final result obtained, particularly as identified by intravascular ultrasound, was consistently related to the restenotic process, with larger final results associated with more favorable angiographic outcomes (Fig. 3)(4).
An unusual relationship between acute gain and late loss is present for lesions developing an aggressive restenosis. Typically there is a direct relationship between acute gain and late loss, as a larger tax is imposed with greater degrees of lesion expansion (5). However, aggressive restenotic lesions demonstrated an inverse relationship, with less acute gain but greater late loss. This conflicts with the theory that aggressive stent expansion leads to aggressive restenosis.
The prevalent finding that coil stents and Wallstents had higher associations with the presence and severity of restenosis suggests there may be important relationships between certain stent types and the restenotic process. In randomized studies, a higher angiographic restenosis rate was seen with the Gianturco-Roubin II stent compared with the Palmaz-Schatz stent (13). In this study Gianturco-Roubin II coil stents were strongly associated with diffuse restenosis. These studies together suggest there is an important interaction of stent design with the restenotic process.
Previous analysis has supported a relationship between the amount of metal implanted and the development of in-stent restenosis (4). The current analysis demonstrates a strong association with the amount of metal implanted and the development of a proliferative restenosis. A significant predictor of an increase in lesion length was the stent to lesion length ratio. This ratio was not predictive of any angiographic restenosis, however. It appears that the interaction of the metal with the arterial wall promotes intimal proliferation in a subset of individuals. When this predilection is not present, additional stent length may help in achieving an optimal luminal result along the length of the artery. However, when this predilection is present, the resultant restenosis tends to extend further along the length of the stented segment.
Restenosis as a nondichotomous variable
The current study showed that the most important characteristics predicting the presence or absence of restenosis were also important in predicting the severity of restenosis (Fig. 3). This suggests that restenosis is a graded, not merely a dichotomous, process. Large arteries are more commonly free of restenosis after stenting, whereas small arteries develop not only restenosis but diffuse and/or aggressive restenosis, with intermediate arteries presenting with less severe forms.
The tradition of considering the restenotic process as strictly dichotomous may have directed attention away from clinically relevant differences. For example, two different stent types may have similar rates of restenosis, but if one presents in a simple, easily treated fashion and the other presents in a complex, more refractory manner, clinical differences may be significant. None of the randomized trials of stenting to date have reported on the characteristics of restenosis.
This analysis is retrospective and uncontrolled. There were wide variations in lesion characteristics, stent types and strategies employed. Multiple statistical analyses were performed, making it possible that some of the associations were due to type I errors. Consistency of relationships across different analyses should be considered more important than any single finding alone.
The end points of this analysis depended upon the quantitative coronary angiographic determinations of the measured lesion length; however, this measurement carries greater subjectivity than other quantitative angiographic determinants. In-stent restenosis often presents as closely grouped sequential narrowings as opposed to single, specific lesions. Precise definitions of single versus multiple lesions within an artery were prospectively created and adhered to when the quantitative angiographic measurements were performed to minimize subjectivity.
Follow-up after in-stent restenosis was not available for this study. However, more MIs and earlier presentations occurred as the restenosis severity increased, supporting an important clinical significance of the degree of restenosis severity. Other investigators have found correlations with the severity of restenosis presented and an increased likelihood of further events (14,15).
Markers for diffuse restenosis were also important markers for the presence of any restenosis and included the baseline lesion length, the length of stent implanted, the final result achieved inside the stented segment and the stent type used. An important marker for a proliferative restenosis was the relative length of stent implanted to the original lesion length. When aggressive forms of in-stent restenosis occur, they tend to present earlier and are more likely to present as a MI. More careful consideration of the type of in-stent restenosis may aid in identifying when alternative strategies may be useful.
The authors acknowledge the assistance of Jeffery Gornbein, SBCC, from the Department of Biostatistics at the UCLA School of Medicine in the performance of the statistical analyses.
Presented in part at the 1997 American Heart Association meetings in Orlando, Florida, the 1998 American Heart Association meetings in Dallas, Texas, and the 1998 Transcatheter Cardiovascular Therapeutics Conference in Washington, D.C.
- confidence interval
- myocardial infarction
- minimum lumen diameter
- odds ratio
- Received November 5, 1999.
- Revision received November 10, 2000.
- Accepted December 13, 2000.
- American College of Cardiology
- Eltchaninoff H.,
- Koning R.,
- Tron C.,
- Gupta V.,
- Cribier A.
- Bauters C.,
- Banos J.L.,
- Van Belle E.,
- McFadden E.,
- Lablanche J.M.,
- Bertrand M.E.
- Mintz GS, Mehran R, Waksman R, et al. Treatment of in-stent restenosis. Semin Interv Cardiol 98;3:117–21.
- Kasaoka S.,
- Tobis J.,
- Akiyama T.,
- et al.
- Kuntz R.E.,
- Safian R.D.,
- Carrozza J.P.,
- Fishman R.F.,
- Mansour M.,
- Baim D.S.
- Kastrati A.,
- Schomig A.,
- Elezi S.,
- Schuhlen H.,
- Wilhelm M.,
- Dirschinger J.
- Mehran R.,
- Ito S.,
- Abizaid A.,
- et al.
- Dirschinger J.,
- Kastrati A.,
- Neumann F.J.,
- et al.
- Kornowski R.,
- Hong M.K.,
- Tio F.O.,
- Branwell O.,
- Wu H.,
- Leon M.B.
- Lansky A.J.,
- Roubin G.S.,
- O’Shaughnessy C.D.,
- et al.
- Mehran R.,
- Dangas G.,
- Abizaid A.S.,
- et al.
- Bossi I.,
- Klersy C.,
- Cortina R.,
- et al.