Author + information
- Received May 27, 2015
- Revision received October 6, 2015
- Accepted October 8, 2015
- Published online February 8, 2016.
- Kristina Hambraeus, MD, PhDa,b,∗ (, )
- Karin Jensevik, MScc,
- Bo Lagerqvist, MD, PhDb,c,
- Bertil Lindahl, MD, PhDb,c,
- Roland Carlsson, MD, PhDd,
- Ramin Farzaneh-Far, MDe,
- Thomas Kellerth, MDf,
- Elmir Omerovic, MD, PhDg,
- Gregg Stone, MD, PhDh,
- Christoph Varenhorst, MD, PhDb,c and
- Stefan James, MD, PhDb,c
- aDepartment of Cardiology, Falun Hospital, Falun, Sweden
- bDepartment of Medical Sciences, Cardiology, Uppsala University, Uppsala, Sweden
- cUppsala Clinical Research Center, Uppsala University, Uppsala, Sweden
- dPCI Unit, Department of Cardiology, Central Hospital, Karlstad, Sweden
- eGilead Sciences, Inc., Foster City, California
- fDepartment of Cardiology, University Hospital, Örebro, Sweden
- gDepartment of Cardiology, Sahlgrenska University Hospital, Gothenburg, Sweden
- hNew York Presbyterian Hospital, Columbia University Medical Center, and the Cardiovascular Research Foundation, New York, New York
- ↵∗Reprint requests and correspondence:
Dr. Kristina Hambraeus, Department of Cardiology, Falun Hospital, S-791-82 Falun, Sweden.
Objectives The aim of this study was to describe current practice regarding completeness of revascularization in patients with multivessel disease undergoing percutaneous coronary intervention (PCI) and to investigate the association of incomplete revascularization (IR) with death, repeat revascularization, and myocardial infarction (MI) in a large nationwide registry.
Background The benefits of multivessel PCI are controversial.
Methods Between 2006 and 2010 we identified 23,342 patients with multivessel disease in the SCAAR (Swedish Coronary Angiography and Angioplasty Registry) and merged data with official Swedish health data registries. IR was defined as any nontreated significant (60%) stenosis in a coronary artery supplying >10% of the myocardium.
Results Patients with IR (n = 15,165) were older, had more extensive coronary disease, and more often had ST-segment elevation MI at presentation than those with complete revascularization (CR) (n = 8,177). All-cause 1-year mortality, MI, and repeat revascularization were higher in IR than CR: 7.1% versus 3.8%, 10.4% versus 6.0%, and 20.5% versus 8.5%, respectively. Propensity score methodology was used in the adjusted analyses. Adjusted hazard ratio (HR) for the composite of death, MI, or repeat revascularization at 1 year was higher in IR than CR: 2.12 (95% confidence interval [CI]: 1.98 to 2.28; p < 0.0001). Adjusted HR for death and the combination of death/MI were 1.29 (95% CI: 1.12 to 1.49; p = 0.0005) and 1.42 (95% CI: 1.30 to 1.56; p < 0.0001), respectively.
Conclusions Incomplete revascularization at the time of hospital discharge in patients with multivessel disease undergoing PCI is associated with a high risk of recurrent 1-year adverse cardiac events.
Multivessel disease, that is, more than 1 significant coronary artery stenosis, is frequently identified in patients with suspected or confirmed ischemic heart disease. In fact around one-half of all patients presenting with ST-segment elevation myocardial infarction (STEMI) (1,2), and up to two-thirds of patients with non–ST-segment elevation acute coronary syndrome (NSTE-ACS) have significant stenoses in more than 1 vessel (3,4).
Incomplete revascularization (IR), i.e., performing percutaneous coronary interventions (PCI) of some but not all significant lesions, has been associated with an increased risk for future cardiovascular events in observational studies and subgroup analyses of randomized trials (5–8). Complete revascularization (CR) of all significant stenoses, however, necessitates a more complex intervention with a potentially higher risk for complications, increases radiation exposure and use of contrast agent, and has been associated with worse prognosis for patients with STEMI in some observational studies (9,10), whereas others have suggested the opposite (11). A recent randomized trial demonstrated a substantially reduced risk of adverse cardiovascular events with multivessel PCI (12).
For patients presenting with NSTE-ACS, the European Society of Cardiology guidelines for revascularization recommends to base the revascularization strategy (ad hoc culprit-lesion PCI/multivessel PCI/coronary artery bypass graft [CABG]) on the clinical status and comorbidities as well as the disease severity (13). For patients with stable coronary artery disease, professional practice guidelines do not formally address the issue of performing CR versus IR with PCI; however, assessment of lesions with functional methods such as fractional flow reserve (FFR) is recommended to avoid unnecessary treatment of nonsignificant stenoses (14–16).
For patients presenting with STEMI, PCI of nonculprit vessels in the acute setting is generally discouraged in guidelines, except for patients presenting with cardiogenic shock, persistent ischemia, or in patients where electrocardiogram localization of the infarction is ambiguous and leads to difficulties in identifying the culprit lesion (13,17,18).
Because previous observational studies and small randomized trials have yielded conflicting evidence as to which strategy is superior, the evidence base for these recommendations is rather weak.
Therefore, the aim of this study was to describe current practice regarding completeness of revascularization in a large and unselected national cohort of patients with multivessel disease undergoing PCI, and to investigate the association of IR with death, repeat revascularization, and myocardial infarction (MI).
This was an observational study of prospectively collected data from the SWEDEHEART (Swedish Web-system for Enhancement and Development of Evidence-based care in Heart disease Evaluated According to Recommended Therapies) and the Swedish national health data registries. The study was approved by the Regional Ethical Review Board in Uppsala.
The SWEDEHEART registry contains the SCAAR (Swedish Coronary Angiography and Angioplasty Registry) together with registries for acute cardiac care, thoracic surgery, and secondary prevention, forming a national quality registry prospectively collecting demographic, clinical, and angiographic data for all patients treated in every Swedish coronary care unit and/or revascularized with PCI or CABG at every Swedish center performing these procedures (19).
Consecutive patients with PCI performed from January 1, 2006, to July 1, 2010, were identified in the SWEDEHEART/SCAAR registry. Vital status, date of death, medical and drug history, occurrence of new myocardial infarction, and prescribed and dispensed medications were collected from the mandatory Swedish national health data registries: National Cause of Death Register, National Patient Register, and the Prescribed Drug Register. Data from SWEDEHEART/SCAAR was merged with the Swedish national health data registries to form a study database.
Patients with single-vessel disease or previous coronary bypass surgery were excluded, as well as patients with missing data for revascularization status and missing data for time of day (on call versus within office hours) for the procedure. No multiple imputations or any other imputation method was used, but a sensitivity analysis was done regarding the variable for time of day (Figure 1).
The Charlson Index was used to measure comorbidities (20). This index includes 19 diseases that have been selected and weighted on the basis of the strength of their association with in-hospital and 1-year mortality (21).
Interventional strategy, lesion selection, and completeness of revascularization were at the operators’ discretion. IR was defined as any nontreated significant (at least 60%) stenosis in a coronary artery supplying >10% of the myocardium (22), on the basis of the operator’s assessment and entered as a variable into the registry by the operator at the end of the procedure. Additional methods to determine the significance of lesions, such as FFR measurements, were used at the operators’ discretion and noted in the registry. Data regarding the procedure such as indication, angiographic finding, type of procedure, stent type, stent size, and pharmacological treatment were entered into the registry by the operator. All Swedish PCI centers are treating chronic total occlusions (CTOs); however, a more dedicated and advanced CTO treatment program that is used in some centers today was not introduced in any of the centers during the study period.
Patient follow-up commenced after the last PCI that was performed during the index hospitalization. The registry did not allow for identification of planned staging of procedures. Therefore, we determined revascularization status (IR or CR) after the last procedure that was performed before discharge from hospital. Because some of the early revascularizations after discharge may have been part of a planned strategy of staged procedures, we performed sensitivity analyses extending the time interval by 2, 4, and 8 weeks before revascularization status was determined.
If the patient had any drug-eluting stent (DES) implanted during the index hospitalization they were allocated to the DES group, whereas stents implanted before the index hospitalization were disregarded in the determination of DES status.
The primary objective was to describe the association of IR versus CR with the pre-defined composite endpoint of death (all cause), MI, and repeat revascularization (PCI or CABG) up to 12 months after the initial PCI. Secondary pre-defined outcomes were the individual components of the composite endpoint and the combination of death and MI during the first 12 months.
MI was diagnosed using standard clinical criteria for International Statistical Classification of Diseases, Tenth Revision codes I21-I22 collected from the National Patient Register. Repeat PCI data were collected from SWEDEHEART and included regardless of whether the procedure was clinically or angiographically driven. Vital status including date of death was collected from the National Cause of Death Registry. The Swedish personal identification number enables collection of outcome data for each patient; thus, no patients were lost to follow-up.
The unadjusted and adjusted associations of IR versus CR with the composite endpoint of death, MI, and repeat revascularization; the secondary composite endpoint of death/MI; and the individual endpoint of death at 1 year were estimated using Cox proportional hazard regression models. For the survival analyses of the individual endpoint of repeat revascularization, patients who died during follow-up were censored. For the analyses of death and of the composite endpoints, patients were only censored for end of follow-up.
Because the treatment groups differed in observed pre-treatment variables at baseline, propensity score methodology was used in the adjusted analyses. A multivariable logistic regression model was used to estimate the propensity score (23). The model included: age, sex, number of PCI procedures during hospitalization, smoking status, indication for index PCI (noncoronary disease, stable angina, NSTE-ACS, STEMI) cardiogenic shock at presentation, medical history of hypertension, Charlson comorbidity index, previous use of statins, warfarin and long-acting nitroglycerin, number of stents placed during index hospitalization, sum of stent length, the smallest stent diameter, number of diseased vessels at first PCI, year of inclusion in registry, treating center, and time of day (on call versus within office hours) at index PCI. The C-statistics for the propensity score-model was 0.81. Proportional hazards assumption was assessed by inspecting Schoenfeld residual plots. To check if the propensity score assessed balance between patients with IR and CR status for the observed covariates, we used the Student t test and Wald test for continuous variables and chi-square test for proportions. The following variables were included along with the propensity score in the adjusted Cox regression model: mean stent diameter (index PCI), findings (first PCI), angiotensin-converting enzyme inhibitors, beta-blockers, long-acting nitroglycerin and statins at discharge, oral antidiabetics at discharge and in medical history, heart failure and chronic obstructive pulmonary disease in medical history, and finally, previous PCI. As a sensitivity analysis, the models for the primary endpoint were reanalyzed using propensity score matching 1:1 with an 8-digit match without replacement (24,25). A match was found for 5,924 (39.1%) of the patients with IR status. Student paired t test for continuous variables and McNemar’s or Bowker’s test for proportions was used to check if the ps-matching assessed balance between patients with IR and CR status. Cox proportional hazards models stratified on the match pairs were used to estimate the unadjusted and adjusted associations of IR versus CR for the primary endpoint.
For the primary endpoint, analyses were also performed at 30, 90, and 180 days from the index PCI date, as well as landmark analyses from these time points up to day 365 after the index PCI date. These intervals were chosen on the basis of clinically used treatment periods for dual antiplatelet treatment (26).
Pre-specified stratified unadjusted and adjusted analyses were also performed for ACS versus non-ACS status at index PCI, STEMI versus NSTE-ACS, STEMI versus stable coronary artery disease, diabetes mellitus versus no diabetes, treatment versus no treatment with long-acting nitroglycerin, history versus no history of MI, and/or previous PCI and any DES versus only bare-metal stent placement at the index hospitalization.
Because new PCI procedures during follow-up might change the revascularization status, Cox proportional hazards models with revascularization status as a time-dependent covariate were used to estimate the associations of IR with the secondary endpoints of death and the composite of death and MI.
Secondary analyses restricted to the patients who survived to discharge from hospital were also performed. Statistical analyses were performed using SAS version 9.3 (SAS Institute, Cary, North Carolina) and R version 3.0.2 (R Foundation for Statistical Computing, Vienna, Austria).
Baseline characteristics, angiographic findings, and medication at discharge
Of the 23,342 patients included in the study, 15,165 (65%) were classified as IR at the index procedure. Patients in the IR group were older, more likely to be female, more likely to have previous cardiovascular disease, and less likely to have undergone previous PCI compared with the CR group. The overall comorbidity measured with Charlson comorbidity index was higher in the IR group than in the CR group (Table 1).
More than 1 procedure was performed in 711 cases before discharge from hospital.
In-hospital death occurred in 571 patients: 475 in the IR group and 96 in the CR group.
IR was performed in the majority of patients with 3-vessel disease (6,302 of 7,524; 84%), presenting with STEMI (6,143 of 7,673; 80%), and presenting after office hours (4,961 of 6,570; 76%). IR patients were more often treated with DES. FFR measurements were used in 3.0% of the IR group and 6.1% of the CR group (Table 2).
Adherence to current secondary prevention guidelines regarding medication was high in both IR and CR groups. Prescription of long-acting nitroglycerin at discharge was more common in the IR group (Table 3).
Frequencies of the individual endpoints are shown in Table 4. A total of 5,071 events of the primary composite endpoint (death, MI, or repeat revascularization) occurred up to 1 year in the IR group compared with 1,254 events in the CR group (unadjusted HR: 2.48; 95% CI: 2.33 to 2.64; p < 0.0001) (Figure 2). After multivariable adjustment including the propensity score, the difference in risk of death, myocardial infarction, or repeat revascularization for the IR and CR groups remained significant: adjusted hazard ratio (HR): 2.12 (95% confidence interval [CI]: 1.98 to 2.28; p < 0.0001).
The event rate 30 days after PCI was 1,702 in the IR group and 347 in the CR group: unadjusted HR: 2.73 (95% CI: 2.43 to 3.06; p < 0.0001), and after 90 days it was 3,351 versus 611: unadjusted HR: 3.19 (95% CI: 2.93 to 3.48; p < 0.0001).
Landmark analyses from 30 and 90 days showed similar differences between groups: unadjusted HR for the combined endpoint was 2.38 (95% CI: 2.21 to 2.56; p < 0.0001) and 1.78 (95% CI: 1.63 to 1.95; p < 0.0001) for IR versus CR at 30 and 90 days.
When data was reanalyzed using propensity score matching, consistent results were found with a similar difference in risk of death, MI, or repeat revascularization between the IR and CR groups: adjusted HR 2.30 (95% CI: 2.10 to 2.52; p < 0.0001) (Online Table 1S).
When we analyzed death, with repeat revascularization as a time dependent covariate, the unadjusted HR for IR compared with CR was 2.05 (95% CI: 1.80 to 2.32; p < 0.0001). For the combined endpoint of death/MI, HR was 1.92 (95% CI: 1.77 to 2.09; p < 0.0001) for IR compared with CR. These differences remained statistically significant after adjustment: HR: 1.29 (95% CI: 1.12 to 1.49; p = 0.0005) for death and HR: 1.42 (95% CI: 1.30 to 1.56; p < 0.0001) for the combination of death/MI.
A significantly higher hazard for IR compared with CR patients was consistent across all subgroups (Table 5).
When we prolonged the time interval to 2, 4, or 8 weeks to determine revascularization status of the study patients, the magnitude of effect size for the combined endpoint was similar (Online Tables 2S to 4S).
However, when we analyzed death up to 1 year only in patients surviving to discharge, the differences in HR between the IR and CR groups did not remain in all subgroups (Table 6).
In this nationwide, observational, registry-based study of all Swedish PCI patients with multivessel disease treated in 2006 to 2010, we observed a high event rate associated with IR after PCI. Death or MI occurred in 17% of patients with IR and an additional 20% required repeat revascularization. Compared with CR, IR was associated with substantially higher event rates after discharge.
After multivariable adjustment including propensity score analysis, patients IR experienced a 2-fold risk for the composite of death, MI, or repeat revascularization at 1 year compared with patients with CR. Because the decision to perform repeat revascularization could be biased due to the knowledge of remaining significant stenoses, we also investigated differences in outcome leaving out repeat revascularization. The differences persisted with approximately the same magnitude regarding the individual components of death and MI. To reduce the effect of possible planned staged procedures, we also performed landmark studies at different time points with consistent findings.
The anatomical disease burden, that is, number of diseased vessels, is of prognostic value in patients with stable angina (27) and in NSTE-ACS patients treated with early invasive strategy and contemporary adjunctive antithrombotic and anticoagulant therapy (28). The negative effect of multivessel disease on prognosis compared with single-vessel disease was, however, not demonstrated in STEMI patients in a recent observational study (29).
The effect of performing CR on mortality, repeat revascularization, and MI has been investigated only in a few small prospective randomized trials. In a single-center study, 219 patients without ongoing MI were randomized to PCI of the culprit vessel only versus CR, resulting in lower strategy success rate and higher procedural costs for the CR group, but similar in-hospital and 1-year major adverse cardiac event rates (30). The PRAMI (Preventive Angioplasty In Myocardial Infarction) trial, which randomized 465 patients with STEMI to PCI of infarct-related coronary artery only or CR in the same setting, demonstrated reduced risk of adverse cardiovascular events with multivessel PCI after a median follow-up of 23 months (12). The open-label design, small number of events, and large number of patients lost to follow-up limited the interpretability of this trial. Nonetheless, the results of the PRAMI trial are in line with the conclusions in a recent meta-analysis of 3 randomized and 23 nonrandomized studies investigating culprit-only versus multivessel PCI in STEMI patients, where multivessel PCI was associated with improved long-term survival and reduced repeat PCI. There was, however, an increase in in-hospital mortality when multivessel PCI was performed during the index primary PCI procedure, which was not seen with a staged procedure (31).
When we analyzed the subgroup of STEMI patients in our study, we did not find a significant difference in risk of death after hospital discharge up to 12 months between the IR and CR groups. However, for the composite of death, MI, or repeat revascularization, STEMI patients reflected the results of the overall study with an adjusted HR of 2 for the IR group compared with the CR group at 12 months.
The results in our study were also consistent with another recent meta-analysis excluding STEMI studies but including studies where revascularization was performed with CABG or PCI, in which the authors found an association between CR and lower risk for death, MI, and repeat revascularization (32).
In addition to survival and repeat procedures, health-related quality of life is an important outcome. For elderly patients and patients with comorbidities and a shorter expected remaining life span, this may be perceived as even more important than a reduction in event rates. In this study, we did not specifically capture measures of quality of life or disease burden. However, treatment with long-acting nitroglycerin likely reflects symptoms of angina pectoris, and we did observe that long-acting nitroglycerin use was more common in the IR group. In contrast, a previous study suggested that selective revascularization can lead to acceptable results for patients above the age of 75 years (33).
Despite its size and national scope, this observational study has inherent limitations. First, the issue of unmeasured confounders must always be considered when interpreting results of nonrandomized studies. The basic assumption for the propensity score methodology is that there are no unmeasured confounders that influence treatment assignment and outcome, and this can never be achieved in any observational study. In the absence of clear guideline recommendations, the decision to perform complete rather than more limited revascularization is likely on the basis of the patient’s characteristics and preferences together with the knowledge and experience of the operator, the appearance of the coronary vessels, and the technical complexity of the lesions. The granular details of these factors cannot be completely captured in this registry despite taking many other potentially confounding variables into account.
Second, some of the lesions not being revascularized may have been CTOs that may or may not have been technically addressable by PCI. Presence of CTO in a noninfarct-related artery has been demonstrated to be an indicator of mortality (1), whereas successful CTO PCI has been associated with lower mortality in patients in stable condition and a small study with STEMI patients (34,35).
Third, the use of FFR measurement as an additional method to evaluate which stenoses were functionally significant, as opposed to relying on anatomical estimation, was used in only a minority of patients. Increased use of FFR measurement in non-STEMI ACS patients and patients in stable condition has the potential to reclassify patients regarding number of diseased vessels and has been shown to improve outcome (36).
Because our study is observational, the question of whether the higher event rates in incompletely revascularized patients can be lowered with a treatment strategy aiming at full revascularization still remains unclear and cannot be answered in a registry study.
For patients with multivessel disease undergoing PCI, IR achieved at the time of hospital discharge was associated with a high risk of death, repeat revascularization, or MI in the first year. Whether this risk can be mitigated by efforts to achieve CR, or by the use of novel adjunctive pharmacological strategies, remains to be further investigated in large randomized trials.
WHAT IS KNOWN? There is conflicting evidence regarding CR versus IR in patients with multivessel disease undergoing PCI.
WHAT IS NEW? Our observational study showed that IR at the time of hospital discharge was associated with a high risk of 1-year mortality, MI, or repeat revascularization.
WHAT IS NEXT? Whether this high risk can be reduced by a strategy of CR needs to be further investigated in a randomized trial.
This study was supported by an unrestricted grant from Gilead Sciences, Inc. Dr. Farzaneh-Far is an employee of Gilead Sciences, Inc. Dr. Omerovic has received lecture fees from Medtronic; research grants from Abbott Vascular and AstraZeneca; and has served on advisory boards for Novartis and AstraZeneca. Dr. Varenhorst has received a research grant from AstraZeneca; has received lecture fees from/served on the advisory board of AstraZeneca, Eli Lilly and Company, The Medicines Company, and St. Jude; and has received an institutional research grant from The Medicines Company. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- coronary artery bypass grafting
- complete revascularization
- drug-eluting stent(s)
- fractional flow rate
- incomplete revascularization
- non–ST-segment elevation acute coronary syndrome
- percutaneous coronary intervention
- ST-segment elevation myocardial infarction
- Received May 27, 2015.
- Revision received October 6, 2015.
- Accepted October 8, 2015.
- American College of Cardiology Foundation
- Claessen B.E.,
- Dangas G.D.,
- Weisz G.,
- et al.
- Wallentin L.,
- Lagerqvist B.,
- Husted S.,
- Kontny F.,
- Stahle E.,
- Swahn E.,
- for the FRISC II Investigators
- The TIMI IIIB Investigators
- Hannan E.L.,
- Racz M.,
- Holmes D.R.,
- et al.
- Wu C.,
- Dyer A.,
- King S.B. III.,
- et al.
- Farooq V.,
- Serruys P.W.,
- Garcia-Garcia H.M.,
- et al.
- Rosner G.F.,
- Kirtane A.J.,
- Genereux P.,
- et al.
- Cavender M.A.,
- Milford-Beland S.,
- Roe M.T.,
- Peterson E.D.,
- Weintraub W.S.,
- Rao S.V.
- Windecker S.,
- Kolh P.,
- Alfonso F.,
- et al.
- Levine G.N.,
- Bates E.R.,
- Blankenship J.C.,
- et al.
- Montalescot G.,
- Sechtem U.,
- Achenbach S.,
- et al.
- Hamm C.W.,
- Bassand J.P.,
- Agewall S.,
- et al.
- Steg P.G.,
- James S.K.,
- Atar D.,
- et al.
- O'Gara P.T.,
- Kushner F.G.,
- Ascheim D.D.,
- et al.
- Jernberg T.,
- Attebring M.F.,
- Hambraeus K.,
- et al.
- Radovanovic D.,
- Seifert B.,
- Urban P.,
- et al.
- Cowley M.J.,
- for the Multivessel Angioplasty Prognosis Study (MAPS) Group
- ↵Parsons LS. Performing a 1:N case-control match on propensity score. 2004. Available at: http://www2.sas.com/proceedings/sugi29/165-29.pdf. Accessed September 13, 2014.
- Varenhorst C.,
- Jensevik K.,
- Jernberg T.,
- et al.
- Mancini G.B.,
- Hartigan P.M.,
- Bates E.R.,
- et al.
- Lansky A.J.,
- Goto K.,
- Cristea E.,
- et al.
- Bainey K.R.,
- Mehta S.R.,
- Lai T.,
- Welsh R.C.
- Garcia S.,
- Sandoval Y.,
- Roukoz H.,
- et al.
- Jones D.A.,
- Weerackody R.,
- Rathod K.,
- et al.
- Yang Z.K.,
- Zhang R.Y.,
- Hu J.,
- Zhang Q.,
- Ding F.H.,
- Shen W.F.