Abstract
Background
Drug-eluting stents (DES) reduce restenosis but require prolonged antiplatelet therapy, when compared with bare metal stents. Ideally, the patient should be involved in this risk:benefit assessment prior to selecting DES, to maximize the benefits and cost-effectiveness of care, and to improve medication adherence. However, accurate estimation of restenosis risk may require angiographic factors identified at cardiac catheterization.
Methods
In a large PCI registry, we used logistic regression to identify clinical and angiographic predictors of clinically-driven target lesion revascularization (TLR) over the first year after stent placement. Discrimination c -statistic and integrated discrimination improvement (IDI) were used to calculate the incremental utility of angiographic variables when added to clinical predictors.
Results
Of 8501 PCI patients, TLR occurred in 4.5%. After adjusting for DES use, clinical TLR predictors were younger age, female sex, diabetes, prior PCI, and prior bypass surgery (model c -statistic 0.630). Angiographic predictors were vein graft PCI, in-stent restenosis lesion, longer stent length, and smaller stent diameter ( c -statistic 0.650). After adding angiographic factors to the clinical model, c -statistic improved to 0.680 and the average separation in TLR risk among patients with and without TLR improved by 1% (IDI = 0.010, 95% CI 0.009–0.014), primarily driven by those experiencing TLR (from 5.9% to 6.9% absolute risk).
Conclusions
Among unselected PCI patients, the incidence of clinically-indicated TLR is <5% at 1-year, and standard clinical variables only moderately discriminate who will and will not experience TLR. Angiographic variables significantly improve TLR risk assessment, suggesting that stent selection may be best performed after coronary anatomy has been delineated.
Short summary (for annotated table of contents)
Although several recent studies have challenged traditional expectations regarding the duration of dual antiplatelet therapy, current guidelines recommend at least 6 to 12 months of treatment after implantation of a drug eluting stent, with a shorter course for bare metal stents. Stent selection ideally should involve input from the patient receiving these stents, but multiple studies have suggested that angiographic factors – obtained after the patient has received sedation during the diagnostic catheterization – are important predictors of repeat revascularization. In this analysis from a large registry of patients receiving coronary stents, angiographic characteristics were found to significantly improve risk assessment for target lesion revascularization, when added to clinical variables alone.
Highlights
- •
Risk of renarrowing often drives selection of drug eluting vs. bare metal stent.
- •
However, relative importance of angiographic data vs. clinical data remains unclear.
- •
In a large PCI registry, we found 4.5% rate of target lesion revascularization.
- •
Angiographic characteristics improved risk assessment by >20% over clinical data.
- •
Thus, stent selection may be best performed after coronary anatomy is known.
1
Introduction
Drug-eluting stents (DES) reduce restenosis when compared with traditional bare metal stents (BMS), but patients receiving DES require prolonged antiplatelet therapy to reduce the risk of stent thrombosis and other ischemic events . With the cost and higher bleeding rates associated with DES and antiplatelet medications, several recent studies have renewed the debate regarding the magnitude of restenosis reduction provided by DES over BMS . These studies are particularly relevant given the lower rates of clinically-driven target lesion revascularization (TLR) demonstrated in real-world practice , when compared with the landmark clinical trials of DES in more highly selected lesions and patient populations .
Physicians therefore have a need to identify patients at high risk of restenosis who would benefit most from DES, and conversely, those with relatively low restenosis risk (or high risk of medication nonadherence) who may be best served by using BMS. In order to personalize the benefits of TLR reduction using DES in individual patients, risk models have been constructed to help estimate the risk of repeat revascularization after PCI . Ideally, it would be desirable to involve the patient in this risk:benefit assessment, to maximize the benefits and cost-effectiveness of care, and to improve adherence to dual antiplatelet therapy after PCI. However, accurate estimation of restenosis risk is believed to require angiographic factors , implying that stent selection for a given patient may be best performed after diagnostic catheterization has been performed. As a result, the informed consent process (where a detailed discussion of DES versus BMS ought to occur) may not be comprehensive until after the diagnostic catheterization, unless pre-procedural characteristics can provide sufficient estimates of DES benefits to discuss stent selection with the patient beforehand.
To evaluate the utility of clinical versus angiographic factors for predicting TLR, we used data from a large multicenter PCI registry to create a TLR risk prediction model for clinical variables available before angiography has been performed. We then quantified the incremental utility of adding angiographic variables for estimating TLR risk, both in the overall patient population and among several pre-specified subgroups.
2
Methods
2.1
Patient population
The Evaluation of Drug Eluting Stents and Ischemic Events (EVENT) was a multicenter registry designed to evaluate interventional practice in the DES era . Approximately 50 centers in the United States enrolled unselected patients age 18 and older expected to undergo placement of an intracoronary stent between 2004 and 2007 for any clinical indication. Broadly-inclusive enrollment strategies were employed to minimize selection bias, with PCI or bypass surgery within the past 4 weeks or prior participation in EVENT as the only exclusion criteria. Standard demographic, clinical, and treatment variables were prospectively collected as well as detailed descriptions of medications and cardiac biomarkers. Angiographic characteristics such as lesion complexity and location were determined by the operators at each site. In-hospital clinical events were recorded, and site coordinators contacted patients and/or referring physicians by telephone at 6 and 12 months after the index PCI to identify significant clinical events including hospitalization, myocardial infarction, repeat revascularization, and death. The human studies committee at each site approved the study protocol, and each subject provided written informed consent.
Within EVENT, all patients undergoing PCI with at least one stent (DES or BMS) were eligible for this study. Patients with missing predictor variable data were excluded from the present analysis. Because the informed consent process and clinical decision-making (including selection of DES versus BMS) are approached differently during ST-elevation myocardial infarction, we also excluded patients undergoing primary PCI for this diagnosis. During the time this study was performed, only sirolimus-eluting and paclitaxel-eluting DES were available.
2.2
Data definitions
All follow-up events were reviewed by members of the research team (who were experienced clinical cardiologists), and each repeat revascularization was adjudicated by reviewing the discharge summaries and angiogram reports submitted by each enrolling site. Additional data were obtained from the enrolling hospital when necessary. Patients with missing follow-up data or those who died during follow-up (without first experiencing TLR) were censored at the time of last known event-free contact.
The primary endpoint was defined as TLR occurring during the 12 months after index PCI. TLR included repeat PCI or bypass graft placement for a stenosis in the lesion stented at index PCI, or occurring within 5 mm of the stent (“edge restenosis”), as determined by the investigator at each enrolling site, and then confirmed during the adjudication process.
2.3
Risk model construction
Characteristics of patients experiencing TLR were compared with those not experiencing TLR using chi-square for categorical variables (reported as proportions) and t-tests for continuous variables (reported as mean ± standard deviation). Potential clinical predictors of 12-month TLR were selected from candidate variables from prior restenosis literature, and from variables with nominal statistical significance (at p < 0.1 level) in the bivariate comparisons from the present study. Candidate variables were sociodemographic factors (age, gender, body-mass index, tobacco use), medical comorbidities (hypertension, diabetes, hyperlipidemia, prior myocardial infarction, prior PCI, prior coronary bypass surgery, heart failure, peripheral arterial disease, glomerular filtration rate), and indication for PCI. We then used logistic regression with backward stepwise elimination (stay criterion p ≤ 0.05) to identify clinical predictors of TLR with associated hazard ratios (HR) and 95% confidence intervals (CI).
Using the same methodology, we performed a separate analysis to identify the best angiographic risk model based on variables obtained from diagnostic coronary angiography. Variables considered were patient-level angiographic characteristics (number of diseased vessels, PCI vessel location, number of lesions and vessels undergoing PCI, total stent number, total stent length) and lesion-specific factors at the stented segment (bifurcation location, in-stent restenosis, TIMI flow grade prior to PCI, lesion severity classification, presence of thrombus prior to PCI, maximal lesion stenosis, minimum stent diameter). In order to account for changing practice patterns during patient enrollment in the EVENT registry, we also adjusted for the date of index PCI in both models. In addition, due to the anticipated reduction in TLR when using DES (versus BMS), we adjusted for DES placement at the index PCI.
2.4
Incremental utility analysis
To evaluate the relative importance of clinical versus angiographic variables for predicting TLR, we first calculated the c -statistics separately for the clinical model and for the angiographic model . We then added the angiographic model predictors to the clinical model predictors and calculated the c -statistic for the combined TLR risk model, plus the improvement in c -statistics and likelihood ratios after adding the new variables . Incremental value was calculated using the integrated discriminatory improvement (IDI) statistic—a measure of change in the separation of predicted probabilities of an event between those with and without events, after adding the second set of variables . Stated another way, the IDI estimates the increase in TLR probability for the subset experiencing TLR, minus the decrease in TLR probability for the subset not experiencing TLR, after adding the angiographic TLR predictors. In previous studies of cardiovascular disease risk prediction, IDI values above 0.010 usually were found to be statistically and clinically significant , whereas values below 0.010 in general have not been associated with incremental utility when adding new data to an existing risk model .
2.5
Secondary analyses
We considered the potential for specific patient subgroups to need more complete TLR risk stratification using angiographic variables, and conversely, other subgroups of patients who may be at high enough restenosis risk to adequately undergo risk assessment before diagnostic catheterization (and thus allow more comprehensive informed consent discussions based on clinical variables alone). As such, we categorized patients according to the presence versus absence of each clinical risk predictor and then repeated the IDI analysis within each of these subgroups, after adding the angiographic variables to each subgroup’s clinical risk model. For example, given the long-established association between diabetes and restenosis, the incremental utility of angiographic variables was calculated separately among PCI patients with and without diabetes.
As an additional secondary analysis, we created a logistic regression model using the 3 “traditional” TLR predictors from the initial clinical trials of BMS and DES (total stent length, minimum stent diameter, diabetes) and calculated the c -statistic for the model. We then added all additional variables from the comprehensive risk model of clinical and angiographic factors associated with TLR, and we assessed the incremental value of these other predictor variables using IDI analysis.
Finally, given the potential for differential patient characteristics among those with missing datapoints, we performed a post hoc sensitivity analysis in which multiple imputation was performed to estimate the missing variables for this subset of patients in the EVENT registry. We then compared characteristics between the original patient population of our study against those with missing/imputed data, and we recalculated c -statistics and incremental value analytics after re-introducing these individuals to the expanded final dataset.
All statistical analyses were performed with SAS software (version 9.2, SAS Institute, Cary, NC). Unless otherwise stated, statistical significance was determined by a 2-sided p -value of 0.05.
2
Methods
2.1
Patient population
The Evaluation of Drug Eluting Stents and Ischemic Events (EVENT) was a multicenter registry designed to evaluate interventional practice in the DES era . Approximately 50 centers in the United States enrolled unselected patients age 18 and older expected to undergo placement of an intracoronary stent between 2004 and 2007 for any clinical indication. Broadly-inclusive enrollment strategies were employed to minimize selection bias, with PCI or bypass surgery within the past 4 weeks or prior participation in EVENT as the only exclusion criteria. Standard demographic, clinical, and treatment variables were prospectively collected as well as detailed descriptions of medications and cardiac biomarkers. Angiographic characteristics such as lesion complexity and location were determined by the operators at each site. In-hospital clinical events were recorded, and site coordinators contacted patients and/or referring physicians by telephone at 6 and 12 months after the index PCI to identify significant clinical events including hospitalization, myocardial infarction, repeat revascularization, and death. The human studies committee at each site approved the study protocol, and each subject provided written informed consent.
Within EVENT, all patients undergoing PCI with at least one stent (DES or BMS) were eligible for this study. Patients with missing predictor variable data were excluded from the present analysis. Because the informed consent process and clinical decision-making (including selection of DES versus BMS) are approached differently during ST-elevation myocardial infarction, we also excluded patients undergoing primary PCI for this diagnosis. During the time this study was performed, only sirolimus-eluting and paclitaxel-eluting DES were available.
2.2
Data definitions
All follow-up events were reviewed by members of the research team (who were experienced clinical cardiologists), and each repeat revascularization was adjudicated by reviewing the discharge summaries and angiogram reports submitted by each enrolling site. Additional data were obtained from the enrolling hospital when necessary. Patients with missing follow-up data or those who died during follow-up (without first experiencing TLR) were censored at the time of last known event-free contact.
The primary endpoint was defined as TLR occurring during the 12 months after index PCI. TLR included repeat PCI or bypass graft placement for a stenosis in the lesion stented at index PCI, or occurring within 5 mm of the stent (“edge restenosis”), as determined by the investigator at each enrolling site, and then confirmed during the adjudication process.
2.3
Risk model construction
Characteristics of patients experiencing TLR were compared with those not experiencing TLR using chi-square for categorical variables (reported as proportions) and t-tests for continuous variables (reported as mean ± standard deviation). Potential clinical predictors of 12-month TLR were selected from candidate variables from prior restenosis literature, and from variables with nominal statistical significance (at p < 0.1 level) in the bivariate comparisons from the present study. Candidate variables were sociodemographic factors (age, gender, body-mass index, tobacco use), medical comorbidities (hypertension, diabetes, hyperlipidemia, prior myocardial infarction, prior PCI, prior coronary bypass surgery, heart failure, peripheral arterial disease, glomerular filtration rate), and indication for PCI. We then used logistic regression with backward stepwise elimination (stay criterion p ≤ 0.05) to identify clinical predictors of TLR with associated hazard ratios (HR) and 95% confidence intervals (CI).
Using the same methodology, we performed a separate analysis to identify the best angiographic risk model based on variables obtained from diagnostic coronary angiography. Variables considered were patient-level angiographic characteristics (number of diseased vessels, PCI vessel location, number of lesions and vessels undergoing PCI, total stent number, total stent length) and lesion-specific factors at the stented segment (bifurcation location, in-stent restenosis, TIMI flow grade prior to PCI, lesion severity classification, presence of thrombus prior to PCI, maximal lesion stenosis, minimum stent diameter). In order to account for changing practice patterns during patient enrollment in the EVENT registry, we also adjusted for the date of index PCI in both models. In addition, due to the anticipated reduction in TLR when using DES (versus BMS), we adjusted for DES placement at the index PCI.
2.4
Incremental utility analysis
To evaluate the relative importance of clinical versus angiographic variables for predicting TLR, we first calculated the c -statistics separately for the clinical model and for the angiographic model . We then added the angiographic model predictors to the clinical model predictors and calculated the c -statistic for the combined TLR risk model, plus the improvement in c -statistics and likelihood ratios after adding the new variables . Incremental value was calculated using the integrated discriminatory improvement (IDI) statistic—a measure of change in the separation of predicted probabilities of an event between those with and without events, after adding the second set of variables . Stated another way, the IDI estimates the increase in TLR probability for the subset experiencing TLR, minus the decrease in TLR probability for the subset not experiencing TLR, after adding the angiographic TLR predictors. In previous studies of cardiovascular disease risk prediction, IDI values above 0.010 usually were found to be statistically and clinically significant , whereas values below 0.010 in general have not been associated with incremental utility when adding new data to an existing risk model .
2.5
Secondary analyses
We considered the potential for specific patient subgroups to need more complete TLR risk stratification using angiographic variables, and conversely, other subgroups of patients who may be at high enough restenosis risk to adequately undergo risk assessment before diagnostic catheterization (and thus allow more comprehensive informed consent discussions based on clinical variables alone). As such, we categorized patients according to the presence versus absence of each clinical risk predictor and then repeated the IDI analysis within each of these subgroups, after adding the angiographic variables to each subgroup’s clinical risk model. For example, given the long-established association between diabetes and restenosis, the incremental utility of angiographic variables was calculated separately among PCI patients with and without diabetes.
As an additional secondary analysis, we created a logistic regression model using the 3 “traditional” TLR predictors from the initial clinical trials of BMS and DES (total stent length, minimum stent diameter, diabetes) and calculated the c -statistic for the model. We then added all additional variables from the comprehensive risk model of clinical and angiographic factors associated with TLR, and we assessed the incremental value of these other predictor variables using IDI analysis.
Finally, given the potential for differential patient characteristics among those with missing datapoints, we performed a post hoc sensitivity analysis in which multiple imputation was performed to estimate the missing variables for this subset of patients in the EVENT registry. We then compared characteristics between the original patient population of our study against those with missing/imputed data, and we recalculated c -statistics and incremental value analytics after re-introducing these individuals to the expanded final dataset.
All statistical analyses were performed with SAS software (version 9.2, SAS Institute, Cary, NC). Unless otherwise stated, statistical significance was determined by a 2-sided p -value of 0.05.
3
Results
3.1
Patients enrolled
Between July 2004 and June 2007, EVENT enrolled 10,144 subjects (13,928 lesions treated) who received an average of 1.6 stents in 1.4 lesions per patient. Of note, 7531 patients (89%) were treated with at least one DES. After excluding patients treated for ST-elevation myocardial infarction and those with missing predictor variable data, the final analytic cohort for this analysis consisted of 8501 patients receiving any kind of stent at index PCI ( Fig. 1 ). During the 12 months after stent placement, clinically-driven TLR occurred in 382 of these 8501 patients (4.5%). When compared with no TLR, patients experiencing TLR were slightly younger, more commonly female, and more likely to have diabetes and other cardiovascular risk factors, prior coronary revascularization, and more severe baseline angina ( Table 1 ). There were no differences in PCI indication, vital signs, or baseline laboratory studies between the 2 groups. Angiographic characteristics associated with TLR included higher numbers of lesions treated at index PCI, more severe lesion stenosis and complexity, the presence of an in-stent restenosis lesion treated at index PCI, longer stent length, smaller minimum stent diameter, stent placement in a saphenous vein graft (SVG), and absence of DES use ( Table 2 ). The type of DES placed (sirolimus-eluting versus paclitaxel-eluting) was not related to subsequent TLR.
| Characteristic | TLR (n = 382) | No TLR (n = 8119) | p -Value |
|---|---|---|---|
| Demographics | |||
| Age (y) | 63 ± 12 | 65 ± 11 | 0.003 |
| Male gender | 64% | 69% | 0.038 |
| Uninsured at enrollment | 3% | 3% | 0.82 |
| Medical history | |||
| Diabetes | 44% | 34% | <0.001 |
| Hypertension | 84% | 79% | 0.022 |
| Hyperlipidemia | 79% | 75% | 0.046 |
| Smoker within the past year | 23% | 24% | 0.66 |
| Peripheral arterial disease | 13% | 11% | 0.14 |
| Cardiac history | |||
| Prior MI | 36% | 34% | 0.60 |
| Prior PCI | 50% | 37% | <0.001 |
| Prior coronary bypass surgery | 29% | 20% | <0.001 |
| Heart failure | 9% | 10% | 0.75 |
| CCS angina class | <0.001 | ||
| No angina | 9% | 16% | |
| I | 11% | 14% | |
| II | 32% | 29% | |
| III | 24% | 24% | |
| IV | 24% | 17% | |
| Indication for PCI | 0.28 | ||
| Acute coronary syndrome | 41% | 38% | |
| Chronic stable angina or positive stress test | 54% | 55% | |
| Other | 5% | 7% | |
| Enrollment vital signs | |||
| Body-mass index (kg/m 2 ) | 31 ± 12 | 30 ± 7 | 0.08 |
| Systolic blood pressure (mm Hg) | 143 ± 24 | 142 ± 24 | 0.24 |
| Diastolic blood pressure (mm Hg) | 75 ± 13 | 75 ± 13 | 0.27 |
| Heart rate (beats per minute) | 70 ± 13 | 69 ± 13 | 0.36 |
| Laboratory studies prior to PCI | |||
| White blood cell count (10 −3 per μL) | 8 ± 2 | 8 ± 3 | 0.96 |
| Hemoglobin (g/dL) | 14 ± 2 | 14 ± 2 | 0.18 |
| GFR by Cockcroft–Gault (mL/min) | 90 ± 43 | 89 ± 54 | 0.85 |
| Last documented LVEF | 0.07 | ||
| <25% | 4% | 3% | |
| 25 to 35% | 10% | 7% | |
| 36 to 50% | 22% | 25% | |
| >50% | 64% | 65% |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
