Impact of Smoking on Clinical and Angiographic Restenosis After Percutaneous Coronary Intervention
Another Smoker’s Paradox?
Background Recent studies have suggested that smokers may require less frequent repeated revascularization after percutaneous coronary intervention (PCI) compared with nonsmokers. However, the mechanism of this phenomenon is unknown.
Methods and Results We examined the association between smoking and restenosis using pooled data from 8671 patients treated with PCI in 9 multicenter clinical trials. Clinical restenosis was examined in the cohort of 5682 patients who were assigned to clinical follow-up only. Angiographic restenosis was evaluated in the subset of 2989 patients who were assigned to mandatory angiographic restudy. Among those patients assigned to clinical follow-up only, target lesion revascularization (TLR) occurred in 6.6% of smokers and 10.1% of nonsmokers (P<0.001). After adjustment for baseline clinical and angiographic differences, the rate of TLR remained significantly lower in smokers with an adjusted relative risk of 0.69 (95% CI, 0.54 to 0.88). Among the angiographic cohort, there were no differences in the rates of angiographic restenosis or follow-up diameter stenosis in either univariate or multivariate analyses. This dissociation between clinical and angiographic restenosis was explained in part by reduced sensitivity to restenosis on the part of smokers and by the greater reluctance of smokers to seek medical attention despite recurrent angina.
Conclusions In patients undergoing contemporary PCI, cigarette smoking is associated with a lower rate of subsequent TLR without affecting angiographic restenosis. These findings have important implications for the follow-up of smokers after PCI and suggest that cross-study comparisons of rates of clinical restenosis must account for the potential confounding effect of smoking.
Received February 12, 2001; revision received June 1, 2001; accepted June 7, 2001.
In the United States, ≈25% of patients who undergo percutaneous coronary intervention (PCI) are active smokers.1 Although it is well recognized that continued smoking confers an adverse long-term prognosis after PCI,2 the impact of smoking on other outcomes such as coronary restenosis is less clear. Recently, one large-scale study reported that smokers actually required less frequent repeated revascularization after PCI than nonsmokers.2 Other studies have reported either no significant relationship between smoking and restenosis3,4 or even increased rates of repeated revascularization among smokers.5 Most of these studies have involved routine angiographic follow-up, however, which may have impaired the ability to reliably assess the need for clinically indicated repeated revascularization.6 Moreover, all of these studies were performed in the balloon angioplasty era and therefore provide little insight into the impact of smoking on restenosis among patients undergoing contemporary PCI.
In this study, we sought to analyze the relationship between smoking and the rates of angiographic and clinical restenosis in a large population of patients undergoing contemporary PCI. By examining angiographic restenosis in a subgroup of patients who underwent routine angiographic follow-up and clinical restenosis in a separate subgroup of patients who underwent only clinical follow-up, we attempted to resolve the conflicting results that have been suggested by previous studies.
The study population was derived from 8671 patients enrolled in 9 recent multicenter clinical trials of percutaneous coronary devices. The individual studies included 1 trial comparing directional atherectomy with balloon angioplasty (Balloon vs Optimal Atherectomy Trial [BOAT]),7 4 randomized trials and their associated registries comparing new stent designs (MultiLink, NIR, Bard XT, and AVE Micro II) with the Palmaz-Schatz stent,8–10 2 additional stent registries (Crossflex and GFX),10 and the Stent Antithrombotic Regimen Study (STARS), which compared 3 antithrombotic regimens after successful Palmaz-Schatz coronary stenting.11 Each of these trials had similar inclusion and exclusion criteria. In general, patients were eligible if they had documented coronary ischemia attributable to ≥1 coronary stenoses amenable to PCI. Patients with documented myocardial infarction (MI) within 3 days of the planned revascularization procedure or a left ventricular ejection fraction <30% were excluded from the trials. Informed consent was obtained from each patient according to a protocol approved by the institutional review board at each participating center. All trials were coordinated by the Cardiovascular Data Analysis Center, Harvard Clinical Research Institute (Boston, Mass).
All interventional procedures were performed in standard fashion.7,11 Aspirin and 4 weeks of ticlopidine made up the standard poststent antithrombotic regimen for all studies except STARS, in which 557 patients were randomly assigned to aspirin only and 550 patients were assigned to an aspirin plus warfarin regimen. Glycoprotein IIb/IIIa inhibitors were used at the discretion of the investigator but accounted for only 2.2% of patients overall (range, 0% to 3.9% for the individual studies).
Data Collection and Clinical Follow-Up
Detailed case report forms concerning baseline demographic and clinical data, procedural details, and clinical events during the initial hospitalization and follow-up were completed by a research coordinator at each site and submitted to the data coordinating center. All procedural angiograms were submitted to the Core Angiographic Laboratory (Washington Hospital Center, Washington, DC), where quantitative coronary angiography was performed with an automated edge-detection algorithm.12 All patients underwent clinical follow-up at 6 and 12 months after the index procedure to determine their symptomatic and clinical status. All end points (death, MI, repeated revascularization) were reviewed by an independent clinical events committee.
Routine angiographic follow-up was mandated for all patients in BOAT (n=989) and the MultiLink RX stent registry (n=208) but was not required for any patients in STARS (n=1965). For the remaining 6 studies, a varying percentage of patients (range, 23% to 55%) were randomly assigned to undergo routine angiographic follow-up for the evaluation of restenosis. Angiographic follow-up was performed 6 to 9 months after the index PCI unless earlier angiography was required clinically. Angiographic restenosis was defined as late diameter stenosis ≥50% by core laboratory assessment. Angiographic measures included acute gain, late loss, and loss index as previously defined.13
Smoking status was ascertained at the time of the initial revascularization procedure as previously described.14 Any patient who had smoked within the year preceding the index procedure was considered a smoker. All other patients were considered nonsmokers. Nonsmokers thus included those patients who had never smoked and former smokers who had quit >1 year before the index procedure. Smoking status was not reassessed routinely during follow-up. However, in a subset of our patients (n=1432) in whom smoking status was assessed at 1 year, no nonsmokers began smoking after the index procedure.14
The 2 main end points of our analysis were angiographic restenosis and target lesion revascularization (TLR; ie, clinical restenosis). To avoid confounding of clinical outcomes by the performance of mandatory angiographic follow-up (the “oculostenotic reflex”), we examined the relationship between smoking and TLR in the subgroup of patients assigned to clinical follow-up only (the clinical cohort, n=5682). Similarly, to avoid confounding of angiographic outcomes by symptom levels or referral bias,15 we examined the relationship between smoking status and angiographic restenosis in the patient subgroup assigned to mandatory angiographic follow-up (the angiographic cohort, n=2989).
Continuous variables were summarized as mean±SD and compared by use of Student’s t tests. Categorical variables were displayed as percentages and compared by use of Fisher’s exact and χ2 tests. We used logistic regression to assess the independent effect of smoking status on angiographic restenosis while adjusting for potential confounders. First, stepwise logistic regression was used to identify independent predictors of restenosis excluding smoking status. Candidate explanatory variables for this model included demographic characteristics (age, sex), clinical characteristics (diabetes mellitus, hypertension, hyperlipidemia, prior MI, multivessel disease, prior bypass surgery, unstable angina), and angiographic characteristics (lesion location, reference vessel diameter, lesion length, restenotic lesion, calcification, eccentricity, thrombus). Once an optimal model was identified, smoking status was added to this full model to assess the independent effect of smoking status on angiographic restenosis while adjusting for potential confounders. A similar process was used to assess the independent effect of smoking on death, nonfatal MI, and TLR. Finally, the impact of smoking on follow-up diameter stenosis was assessed by use of linear regression with an analogous modeling strategy.
For models involving clinical outcomes (eg, death, MI), the patient was the unit of analysis. For analyses that involved angiographic outcomes (eg, restenosis, TLR), the lesion served as the unit of analysis. A 2-sided value of P<0.05 was considered statistically significant. All statistical analyses were performed with SAS for Windows version 6.12 (SAS Institute).
Pooled data from the 9 clinical trials yielded 8671 patients who underwent PCI on 8762 coronary lesions. At time of their initial procedures, 6285 patients (72.5%) were nonsmokers and 2386 patients (27.5%) were smokers.
Baseline Clinical and Angiographic Characteristics
Smokers and nonsmokers undergoing PCI had marked differences in baseline clinical characteristics (Table 1). In particular, smokers were ≈9 years younger than nonsmokers and were less likely to have other predisposing risk factors such as hypertension, diabetes, and hyperlipidemia. Smokers were less likely to have a history of bypass surgery but more likely to have had a prior MI. Consistent with their earlier presentation, smokers were less likely to have multivessel coronary disease.
There also were modest differences in angiographic features between smokers and nonsmokers (Table 2). Smokers were less likely to have significant lesion calcification and to undergo treatment for a stenosis in the left anterior descending artery. Quantitative angiography demonstrated that lesions in smokers had a slightly larger reference vessel diameter and minimal lumen diameter before treatment. Lesion length and complexity were similar in both groups, however. Postprocedural quantitative angiographic measures did not differ significantly between the 2 groups.
In the overall population (n= 8671), there were no significant differences between smokers and nonsmokers in the rates of in-hospital death (0.13% versus 0.12%, P=0.72), Q-wave MI (0.8% versus 0.9%, P=0.45), bypass surgery (1.0% versus 0.8%, P=0.34), or repeated PCI (0.6% versus 0.7%, P=0.90). Among the patients assigned to clinical follow-up only (n=5682), there was no difference in crude 1-year mortality between smokers and nonsmokers (1.6% for both groups; Table 3). After adjustment for differences in baseline clinical and angiographic characteristics, however, smoking was associated with an increased risk of death at 1 year (relative risk, 1.8; 95% CI, 1.1 to 3.0; P=0.02). In contrast, smokers were significantly less likely to undergo TLR during follow-up than nonsmokers (6.6% versus 10.1%, P<0.001). This lower risk of TLR for smokers persisted after adjustment for baseline clinical factors (adjusted relative risk, 0.69; 95% CI, 0.54 to 0.88; P=0.003). Among nonsmokers at the time of PCI, the risk of TLR was similar for former smokers (ie, those who quit >1 year before PCI) and for those who never smoked (adjusted relative risk, 0.89; 95% CI, 0.66 to 1.19; P=0.17).
Repeated angiography was performed in 2217 of 2989 patients (74.2%) assigned to mandatory angiographic follow-up a median of 7.4 months after the index procedure. There was no significant difference in the proportion of smokers and nonsmokers who underwent follow-up angiography (74.5% versus 74.0%, P=0.80). Follow-up minimum lumen diameter, diameter stenosis, late loss, and loss index did not differ between the 2 groups (Table 4). Angiographic restenosis (≥50% diameter stenosis) was present in 29.6% of smokers and in 31.3% of nonsmokers (P=0.45). After adjustment for baseline clinical and angiographic differences, there was no independent effect of smoking on angiographic restenosis (relative risk, 0.91; 95% CI, 0.75 to 1.11; P=0.35) or follow-up diameter stenosis (P=0.45).
To identify potential mechanisms for the dissociation of angiographic and clinical restenosis according to smoking status, we performed several additional analyses. To evaluate the possibility that smokers have a higher threshold for symptomatic restenosis than nonsmokers, we examined the relationship between angina (Canadian Cardiovascular Society [CCS] class 1 or greater) during follow-up and the degree of restenosis among the patients who underwent mandatory angiographic follow-up (Figure 1). Among patients with a diameter stenosis <50% or ≥70% at follow-up, there were no significant differences in the prevalence of angina according to smoking status. However, among patients with an intermediate degree of stenosis (50% to 69%) at follow-up angiography, smokers were significantly less likely to report angina (40.2% versus 54.7%, P=0.015). These findings were unchanged in the subgroup of patients with single-vessel disease and persisted after adjustment for baseline clinical and angiographic differences (adjusted relative risk, 0.49; 95% CI, 0.26 to 0.92).
To determine whether a bias on the part of physicians could account for our findings, we examined the relationship between follow-up angiographic stenosis and performance of repeated revascularization. Among patients in the angiographic cohort, overall TLR rates did not differ significantly between smokers and nonsmokers (14.8% versus 16.8%, P=0.24). When stratified according to the severity of stenosis at follow-up, there also were no significant differences in TLR rates between smokers and nonsmokers (Figure 2).
Finally, we examined the relationship between recurrent ischemia during follow-up and the performance of repeated angiography. In the clinical cohort, 24.1% of smokers and 23.0% of nonsmokers reported angina at ≥1 follow-up interviews. Among those patients who reported angina during follow-up, fewer smokers underwent repeated angiography than nonsmokers (50.5% versus 58.7%, P=0.015); this difference was most pronounced among patients with CCS class 1 or 2 angina (Figure 3). After adjustment for baseline clinical differences and the degree of recurrent angina, smoking was associated with a 33% lower rate of repeated angiography (adjusted relative risk, 0.67; 95% CI, 0.48 to 0.93; P=0.017).
In this study, we examined the impact of smoking on both clinical and angiographic outcomes of contemporary PCI, including balloon angioplasty, directional atherectomy, and coronary stenting. Our major finding was that although smokers experienced a 31% reduction in clinical restenosis (ie, TLR), paradoxically there were no differences in angiographic restenosis or other angiographic parameters at follow-up according to smoking status. Although there were important differences in baseline clinical and angiographic characteristics between smokers and nonsmokers, our findings persisted in multivariable analyses that controlled for these differences.
Comparison With Previous Studies
To date, Hasdai and colleagues2 have performed the largest study of the impact of smoking on the outcomes of PCI. In their cohort of >5400 patients, they found that compared with nonsmokers, former smokers and persistent smokers had a 20% and 33% reduction, respectively, in the need for repeated revascularization during the first year of follow-up—results remarkably similar to our experience. However, their study lacked systematic angiographic follow-up in a subset of patients and was thus unable to elucidate the precise mechanism for this apparent “benefit” of smoking. Other studies that have examined the impact of smoking on angiographic restenosis have generally found no differences in either angiographic or clinical outcomes between smokers and nonsmokers.3,4 However, the performance of angiographic follow-up in the full population of these studies limited their ability to reliably assess clinical outcomes because of the well-recognized oculostenotic reflex.6 Until now, it has been unclear whether these differences between angiographic and clinical outcomes reflect underlying differences in the patient populations of the various studies or a true case of “angiographic-clinical dissociation.”16 By virtue of its split-sample design and large sample size, our study is the first to definitively demonstrate that the reduction in clinical restenosis associated with smoking is not due to a reduction in angiographic restenosis. Moreover, our study extends the results of previous studies by demonstrating these findings in a group of patients undergoing contemporary PCI (mainly coronary stenting).
We considered several possible mechanisms to explain this angiographic-clinical dissociation. First, it is possible that patients who smoke are biologically different from nonsmokers so that they are “less sensitive” to restenosis. Indeed, our study supports this hypothesis. Among patients with a 50% to 69% diameter stenosis (by qualitative coronary angiography) at follow-up angiography, smokers were 51% less likely to report angina. One potential mechanism for this finding is the greater prevalence of prior MI among smokers. As a result, the coronary artery treated during the index PCI would be more likely to supply an infarcted, denervated territory. Because our findings persisted in analyses that controlled for prior MI (and other potential confounders), however, it is likely that other factors also are involved in this desensitization.
A second possible explanation would be that physicians are less likely to revascularize patients with restenosis if they continue to smoke. We did not find any evidence of such bias in our study population, however. Among those patients who underwent protocol-mandated follow-up angiography, there were no differences in TLR rates according to the degree of restenosis at follow-up.
Finally, it is possible that smokers are less likely to undergo repeated revascularization because of a workup bias, resulting in a lower referral rate to repeated coronary angiography.17 Our exploratory analyses suggest that this form of bias does exist and may have contributed to the angiographic-clinical dissociation. In the subgroup of patients who underwent standard clinical follow-up, smokers with recurrent angina were 33% less likely to undergo repeated angiography compared with nonsmokers. Whether this finding reflects a bias on the part of physicians not to refer smokers for repeated angiography despite recurrent symptoms or whether smokers are simply less likely to report mild symptoms to their physicians or to comply with medical follow-up cannot be distinguished from our study design.
Our study has several important limitations. First, it is an observational study. Thus, although we controlled for most factors that have been associated with angiographic and clinical restenosis in previous studies, we cannot exclude the possibility that unmeasured confounding may explain some of our findings. Second, it is possible that our angiographic analysis was underpowered to detect a meaningful reduction in angiographic restenosis. Type II error is unlikely to explain our results, however, because our study had >90% power to detect even a 25% reduction in angiographic restenosis in smokers, an effect size smaller than the reduction in clinical restenosis. Finally, because smoking status was not routinely assessed during follow-up, it is possible that some patients who were classified as smokers quit during follow-up. However, any random misclassification of smoking status would only tend to dilute the observed association between smoking and reduced TLR. Thus, our results may actually be viewed as defining the lower limit of this effect.
This study has several clinical implications. First, it demonstrates that smoking continues to be associated with increased late mortality after PCI, even in the current era of coronary stenting and relatively aggressive secondary coronary prevention. Thus, efforts to encourage smoking cessation in this high-risk population remain essential. In addition, the novel observation of a true dissociation between angiographic and clinical restenosis among smokers may have important implications for follow-up after PCI in smokers. Although restenosis does not usually have an adverse impact on prognosis,18 there are certain patient subsets (eg, unprotected left main intervention and PCI in the setting of significant left ventricular dysfunction) for whom coronary restenosis might be a fatal event.19 Because smokers appear to be less sensitive to restenosis (either biologically or behaviorally), very close follow-up of such high-risk subsets among smokers with either routine functional testing or even mandatory angiography might be warranted.
Finally, these findings have important implications for the interpretation of clinical research on coronary restenosis. In recent years, there has been increasing interest in the interventional cardiology community in performing simple studies to assess the impact of new treatments on restenosis. Although the randomized clinical trial with angiographic follow-up remains the gold standard for such evaluations, there is increasing interest on the part of industry sponsors and regulatory agencies in performing simple studies that use clinical follow-up. Although comparison of clinical restenosis outcomes (eg, TLR) in the context of a randomized trial would be valid, our results suggest that any attempt to examine clinical restenosis in a registry (with an external reference population) must control for the potential confounding effect of smoking status.
Dr Cohen was supported in part by a Clinician-Scientist Award from the American Heart Association. Dr Doucet was supported by the Fondation de Cardiologie de l’Hopital Saint-Luc.
Violaris AG, Thury A, Regar E, et al. Influence of a history of smoking on short term (six month) clinical and angiographic outcome after successful coronary angioplasty. Heart. 2000; 84: 299–306.
Baim DS, Cutlip DE, Sharma SK, et al. Final results of the Balloon vs Optimal Atherectomy Trial (BOAT). Circulation. 1998; 97: 322–331.
Kuntz RE, Baim DS. Defining coronary restenosis: newer clinical and angiographic paradigms. Circulation. 1993; 88: 1310–1323.
Taira DA, Seto TB, Ho KK, et al. Impact of smoking on health-related quality of life after percutaneous coronary revascularization. Circulation. 2000; 102: 1369–1374.
Kuntz RE, Keaney KM, Senerchia C, et al. A predictive method for estimating the late angiographic results of coronary intervention despite incomplete ascertainment. Circulation. 1993; 87: 815–830.
Topol EJ, Nissen SE. Our preoccupation with coronary luminology: the dissociation between clinical and angiographic findings in ischemic heart disease. Circulation. 1995; 92: 2333–2342.
Weintraub WS, Ghazzal ZM, Douglas JS Jr, et al. Long-term clinical follow-up in patients with angiographic restudy after successful angioplasty. Circulation. 1993; 87: 831–840.
Ellis SG, Tamai H, Nobuyoshi M, et al. Contemporary percutaneous treatment of unprotected left main coronary stenoses: initial results from a multicenter registry analysis, 1994–1996. Circulation. 1997; 96: 3867–3872.