- Email: [email protected]
The Clinical Profile and Outcomes Associated With Coronary Collaterals in Patients With Coronary Artery Disease
Canadian Journal of Cardiology 27 (2011) 581–588
Clinical Research
The Clinical Profile and Outcomes Associated With Coronary Collaterals in Patients With Coronary Artery Disease M. Sean McMurtry, MD, PhD,a Adriane M. Lewin, BA, MSc,b Merril L. Knudtson, MD,b William A. Ghali, MD, MPH,b P. Diane Galbraith, BN,b Felix Schulte, MD,a,b Colleen M. Norris, PhD,a and Michelle M. Graham, MDa a
University of Alberta, Edmonton, Alberta, Canada b
University of Calgary, Calgary, Alberta, Canada
ABSTRACT
RÉSUMÉ
Background: The clinical correlates of coronary collaterals and the effects of coronary collaterals on prognosis are incompletely understood. Methods: We performed a study of 55,751 patients undergoing coronary angiography to evaluate the correlates of angiographically apparent coronary collaterals, and to evaluate their association with survival. Results: The characteristic most strongly associated with the presence of collaterals was a coronary occlusion (odds ratio [OR], 28.9; 95% confidence interval [CI], 27.1-30.6). Collaterals were associated with improved adjusted survival overall (hazard ratio [HR] 0.89; 95% CI, 0.85-0.95), and in both acute coronary syndrome (ACS) (HR 0.90; 95% CI, 0.84-0.96) and non-ACS (HR 0.84; 95% CI, 0.77-0.92) patients. Collaterals were associated with improved survival in those receiving angioplasty (HR 0.78; 95% CI, 0.71-0.85) and those with low risk anatomy treated medically (HR 0.84; 95% CI, 0.72-0.98), but not for those treated with coronary bypass graft surgery or those with high-risk anatomy treated without revascularization. Conclusions: The major correlate of coronary collaterals is the presence/extent of obstructive coronary artery disease. Collaterals are associated with better survival overall and in both ACS and non-ACS presentations, but not for those treated with coronary artery bypass graft (CABG) or those with high-risk anatomy who are not revascularized.
Introduction : Les corrélats cliniques des collatérales coronaires et les effets des collatérales coronaires sur le pronostic sont partiellement compris. Méthodes : Nous avons mené une étude auprès de 55 751 patients ayant subi une coronarographie pour évaluer les corrélats des collatérales coronaires apparentes à l’angiographie, ainsi que pour évaluer leur lien avec la survie. Résultats : Les caractéristiques les plus fortement associées à la présence des collatérales étaient l’occlusion coronarienne (ratio d’incidence approché [RIA], 28,9; intervalle de confiance [IC] de 95 %, 27,1-30,6). Les collatérales étaient associées à l’amélioration de la survie globale ajustée (rapport de risque [RR] 0,89; IC de 95 %, 0,85-0,95), et aux patients avec le syndrome coronarien aigu (SCA) (RR de 0,90; IC de 95 %, 0,84-0,96) et aux patients avec le non SCA (RR de 0,84; IC de 95 %, 0,77-0,92). Les collatérales étaient associées à l’amélioration de la survie chez ceux qui avaient subi une angioplastie (RR de 0,78; IC de 95 %, 0,71-0,85) et chez ceux qui avaient eu un risque anatomique traité médicalement (RR de 0,84; IC de 95 %, 0,72-0,98), mais non chez ceux qui avaient été traités par un pontage coronarien ou chez ceux avec une anatomie à haut risque qui avaient été traités sans revascularisation. Conclusions : Le plus grand corrélat de collatérales coronaires est la présence/l’étendue de la maladie coronarienne obstructive. Les collatérales sont associées à une meilleure survie globale et aux tableaux cliniques du SCA et du non SCA, mais ne sont pas associées aux patients traités par un pontage aortocoronarien (PAC) ou à ceux ayant une anatomie à haut risque qui ne sont pas revascularisés.
Coronary collaterals develop from pre-existing arterioles1 that expand in response to a stenosis or occlusion.2 They are thought protective in coronary artery disease (CAD), being
associated with smaller myocardial infarct size,3 fewer ischemic events,4,5 and better survival in patients with stable CAD.6 It is unknown why some have large coronary collaterals while others do not. For most variables that have been evaluated, conflicting reports exist (Table 1). Most studies are limited by small sample size or by confounding due to lack of adjustment for CAD burden.12,22,23 Though better coronary collateral function, measured invasively, is associated with fewer ischemic events5 and better sur-
Received for publication January 13, 2011. Accepted February 24, 2011. Corresponding author: Dr M.S. McMurtry, University of Alberta, 8440 112th Street, Edmonton, Alberta T6G 2B7, Canada. Tel.: ⫹1-780-407-2799. E-mail: [email protected] See page 586 for disclosure information.
0828-282X/$ – see front matter © 2011 Canadian Cardiovascular Society. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.cjca.2011.02.011
582
Canadian Journal of Cardiology Volume 27 2011
Table 1. Literature reports of clinical correlates of coronary collaterals are inconsistent Variable Age Diabetes Hypertension Tobacco smoke Dyslipidemia Sex Metabolic syndrome Obesity Impaired fasting glucose Stenosis severity
Positive association
20 21 9,11
No association 7-11 8,12,15-17 7-12,15,17 7-12,15,17 7,8,10,12,15,17 7-12,17
Negative association 12-15 9-11,18,19 45
7,10 7 19 12,17
vival in chronic stable CAD,6 the importance of angiographically measured coronary collaterals is less certain. Composite endpoints predominate in the literature, reflecting sample size constraints. Discrepancies exist: angiographically apparent coronary collaterals are associated with better outcomes in some studies,4,12,24,25 but not all,26-30 and some suggested that collaterals imply worse outcomes.27-29 Despite angiography’s limitations,31 the prognostic implications of angiographically apparent coronary collaterals remains an important question. To identify influences on coronary collateral formation and to evaluate whether collaterals influence survival, we measured the clinical correlates of collaterals and the association of collaterals with survival in an unselected cohort. Methods Study sample We used the Alberta Provincial Project for Outcomes Assessment in Coronary Heart Disease (APPROACH) database, an ongoing initiative in Alberta, Canada, since 1995.32 The cohort consisted of all Alberta residents undergoing catheterization between April 1, 1995, and March 31, 2007. Staff enter data and comorbidities are verified and data are enhanced to ensure comorbidity data are accurate and there are no empty fields.33,34 Mortality is ascertained through linkage to the Alberta Bureau of Vital Statistics. The Health Research Ethics Boards of the Universities of Calgary and Alberta approved the protocol. Variables APPROACH includes age, sex, left ventricular ejection fraction, myocardial infarction, congestive heart failure, diabetes, hypertension, hyperlipidemia, current and prior smoking, cerebrovascular disease, peripheral arterial disease, chronic pulmonary disease, chronic kidney disease, gastrointestinal disease, malignancy, and cardiovascular medication use. It records coronary anatomy, including angiographically apparent collaterals (present or absent) and therapeutic interventions. Inclusion criteria We included consecutive patients undergoing coronary angiography for stable angina, unstable angina, ST-elevation myocardial infarction (STEMI) and non-ST– elevation myocardial infarction (NSTEMI). We excluded patients with prior coronary revascularization or stenosis of less than 50%, nonAlberta residents, and patients with incomplete comorbidity
data. For the first analysis (correlates of collaterals), patients with STEMI were excluded to limit confounding due to time variation. Outcome variable For the first analysis, the main outcome was presence of angiographically apparent coronary collaterals. To verify the validity of APPROACH collateral data, the presence of angiographically apparent collaterals for 100 randomly selected subjects was determined by consensus agreement of 2 expert angiographers who independently reviewed the films and were blinded to the APPROACH data. The results of the consensus agreement were compared with APPROACH, and a statistic was calculated.35 For the second analysis, the main outcome measure was all cause mortality. Duke Jeopardy Score This score encompasses both the severity of a coronary lesion and the volume of myocardium subtended by the stenosis and provides prognostic information.36 Statistical analysis The baseline characteristics were compared with 2 or t tests. Crude and adjusted odds ratios for correlates were calculated by risk status, with high risk anatomy defined as 3-vessel disease, left main disease, or 2-vessel disease involving the proximal left anterior descending coronary artery per the Duke Coronary Index.37 Adjustment variables included demographic information, comorbidities, indication, left ventricular ejection fraction, Duke Jeopardy Score, and cardiac history. To account for hyperlipidemia as a marker for statin use, we included statin use as a binary variable. Odds ratios (ORs) for correlates were calculated using a backward stepwise multivariable logistic regression model.38 We performed a sensitivity analysis for those with at least 1 total occlusion of a major epicardial coronary artery, the anatomy for which angiographic assessment of collaterals is most accurate.39 After ensuring the proportional hazards assumption was met, we evaluated survival using the corrected groups prognosis method.40 We adjusted for clinical variables, including revascularization. We also stratified our results by presentation, treatment type in the following year, and further stratified the group treated with medical therapy by anatomy. SAS software version 9 (SAS Institute Inc, Cary, NC, USA) was used. Results Cohort description Of the 95,646 coronary angiograms between April 1, 1995, and March 31, 2007, a total of 55,751 met our inclusion criteria (Fig. 1). The clinical characteristics are displayed in Table 2. Most subjects with visible collaterals had a coronary occlusion. A total of 10,331 patients with STEMI were excluded from the cohort for the analysis of clinical correlates. Assessment of accuracy collateral data One hundred coronary angiograms were selected randomly from the cohort and independently reviewed by expert angiographers. A consensus opinion regarding the presence or absence
McMurtry et al. Coronary Collaterals and Survival
583
All paents in Alberta between April 1, 1995 and March 31, 2007 N = 95,646
Excluded: Non-Alberta residents (7655) Incomplete comorbidies (1697) Total: 9352
Alberta paents, comorbidity data complete N = 86,294
Excluded: Prior PTCA (5405) Prior CABG (3467) Normal coronaries or lesions < 50% (21,612) Duke Jeopardy Score missing (59) Total: 30,543
Cohort for Analysis of Survival N = 55,751
Collaterals Present N = 21,893 (39.3%)
At least one occlusion N = 15,347 (70.1%)
No occlusions N = 6546 (29.9%)
Collaterals Absent N = 33,858 (60.7%)
At least one occlusion N = 3420 (10.1 %)
No occlusions N = 30,438 (89.9%)
Figure 1. The cohort for analysis of coronary collaterals and survival. After exclusions, a total of 55,751 patients were included. CABG, coronary artery bypass graft; PTCA, percutaneous transluminal coronary angioplasty.
of angiographically apparent coronary collaterals was obtained. This consensus opinion, or gold standard, was compared with the APPROACH data. Of the 100 cases, 88 cases matched perfectly, rivaling the interobserver variability for angiographically apparent collaterals within other carefully done studies in the literature.39 In 2 cases, APPROACH data suggested that collaterals were present when they were not. In the remaining 10 cases, expert consensus identified subtle (Rentrop 141) collaterals that were not coded in APPROACH, suggesting a small degree of systematic underreporting of Rentrop 1 collaterals in APPROACH. The statistic was 0.74, just under the threshold of 0.75 for excellent agreement beyond chance.42
ence of an occlusion (OR 28.9; 95% CI, 27.1-30.6). Because the presence of collaterals was so overwhelmingly associated with occlusions, a sensitivity analysis restricted to patients with at least 1 occlusion was performed. As before, those with highrisk anatomy were more likely to have collaterals (OR 1.44; 95% CI, 1.25-1.67). The odds ratios for correlates of coronary collaterals (Table 3) demonstrate small positive associations for pulmonary disease, hyperlipidemia, peripheral arterial disease, statin therapy, previous smoking, and malignancy. Prior thrombolytic therapy and indications other than stable angina were negatively associated with collaterals. No sex-based differences were observed.
Correlates of coronary collaterals
Coronary collaterals and survival
Compared with lower risk patients, those with 3-vessel disease, left main disease, and 2-vessel disease involving the proximal left anterior descending coronary artery were more likely to have angiographically apparent coronary collaterals (OR 2.74; 95% confidence interval [CI], 2.64-2.85). The characteristic most strongly associated with collaterals was the pres-
Patients with visible coronary collaterals at the time of coronary angiography had worse long-term unadjusted survival those without coronary collaterals (Fig. 2), with a crude hazard ratio of 1.32 (95% CI, 1.27-1.37; Table 4). After adjustment, coronary collaterals were associated with better survival overall (hazard ratio [HR] 0.89; 95% CI, 0.85-0.94). Coronary col-
584
Canadian Journal of Cardiology Volume 27 2011
Table 2. Patient characteristics of the included cohort (N ⫽ 55,751)
Demographics Mean age (SD) Aged ⬎75 years Male Deceased (date of death in dataset) Comorbidities Cerebrovascular disease Pulmonary disease Congestive heart failure Liver/gastrointestinal disease Peripheral vascular disease Renal disease (creatinine ⬎200 mmol/L) Malignancy Hypertension Hyperlipidemia Dialysis Diabetes Current smoker Previous smoker Prior lytic therapy Cardiac history Prior myocardial infarction Indication for coronary angiogram Stable angina Myocardial infarction Unstable angina Other Ejection fraction ⬎50% 35%-50% 20%-34% ⬍20% Could not be performed Unknown Duke Jeopardy Score 0/12 2/12 4/12 6/12 8/12 10/12 12/12 Coronary anatomy* 1 VD 50%-75% 1 VD 95% 2 VD 2 VD PLAD 95% 1 VD 95% PLAD 2 VD 95% LAD 2 VD 95% PLAD 3 VD 3 VD 1%-95% 3 VD PLAD 3 VD 95% PLAD Left main Severe left main Revascularization within 1 year of catheterization Treatment within 1 year of catheterization Medical CABG within 1 year PCI within 1 year
Collaterals absent (n ⫽ 33,858; 60.7%)
Collaterals present (n ⫽ 21,893; 39.3%)
2 P value
63.2 (11.7) 6012 (18.4) 24,449 (72.2) 5301 (15.7)
64.3 (11.3) 4217 (19.3) 16,971 (77.5) 4562 (20.8)
⬍ 0.0001 0.006 ⬍ 0.0001 ⬍ 0.001
2184 (6.5) 4189 (12.4) 4185 (12.4) 1879 (5.6) 2326 (6.9) 1210 (3.6) 1243 (3.7) 20,231 (59.8) 21,456 (63.4) 539 (1.6) 7519 (22.2) 10,455 (30.9) 13,118 (38.7) 3397 (10.0)
1749 (8.0) 2987 (13.6) 3602 (16.5) 1235 (5.6) 2167 (9.9) 935 (4.3) 924 (4.2) 13,604 (62.1) 14,501 (66.2) 354 (1.6) 5658 (25.8) 6368 (29.1) 9111 (41.6) 1143 (5.2)
⬍ 0.0001 ⬍ 0.0001 ⬍ 0.0001 0.65 ⬍ 0.0001 ⬍ 0.0001 0.001 ⬍ 0.0001 ⬍ 0.0001 0.82 ⬍ 0.0001 ⬍ 0.0001 ⬍ 0.0001 ⬍ 0.0001
13,920 (41.1)
10,944 (50.0)
⬍ 0.0001 ⬍ 0.0001
7952 (23.5) 14,970 (44.2) 8050 (23.8) 2886 ( 8.5)
7630 (34.9) 7941 (36.3) 4631 (21.2) 1691 (7.7)
21,392 (63.2) 6701 (19.8) 1406 (4.2) 207 (0.6) 2776 (8.2) 1376 (4.1)
11,431 (52.2) 6142 (28.1) 1896 (8.7) 331 (1.5) 1474 (6.7) 619 (2.8)
4892 (14.5) 11,710 (34.6) 4958 (14.6) 5616 (16.6) 2428 (7.2) 2099 (6.2) 2155 (6.4)
195 (0.9) 5242 (23.9) 3906 (17.8) 3779 (17.3) 3168 (14.5) 2243 (10.3) 3360 (15.4)
2852 (8.4) 8617 (25.5) 3238 (9.6) 2256 (6.7) 2216 (6.5) 2429 (7.2) 1699 (5.0) 160 (0.5) 4317 (12.8) 1181 (3.5) 2373 (7.0) 1136 (3.4) 1384 (4.1) 24,190 (71.5)
20 (0.1) 3344 (15.3) 1208 (5.5) 1835 (8.4) 614 (2.8) 1619 (7.4) 1123 (5.1) 23 (0.1) 5039 (23.0) 1359 (6.2) 3200 (14.6) 1243 (5.7) 1266 (5.8) 15,452 (70.6)
9668 (28.6) 5915 (17.5) 18,275 (54.0)
6441 (29.4) 7358 (33.6) 8094 (37.0)
⬍ 0.0001
⬍ 0.0001
⬍ 0.0001
0.03 ⬍ 0.0001
Data are presented as n (%). CABG, coronary artery bypass graft; LAD, left anterior descending coronary artery; PCI, percutaneous coronary intervention; PLAD, proximal LAD; VD, vessel disease. * Percentage stenosis.
McMurtry et al. Coronary Collaterals and Survival
585
Table 3. Clinical correlates of angiographically apparent coronary collaterals* N ⫽ 14,731 Variable Comorbidities Pulmonary disease Peripheral vascular disease Malignancy Hyperlipidemia Previous smoker Prior lytic therapy Statin Indication Stable angina MI Unstable angina Other Ejection fraction, % ⬎50 35-50 20-34 ⬍20 Coronary anatomy Low risk High risk/left main
OR
95% CI
P value
1.44 1.26 1.40 1.22 1.12 0.79 1.53
1.22-1.69 1.05-1.51 1.05-1.87 1.08-1.38 1.01-1.24 0.64-0.97 1.34-1.75
⬍ 0.001 0.015 0.022 0.001 0.040 0.026 ⬍ 0.001
Referent 0.36 0.57 0.69
0.32-0.41 0.49-0.66 0.56-0.85
⬍ 0.001 ⬍ 0.001 ⬍ 0.001
Referent 0.84 0.80 0.75
0.74-0.95 0.66-0.96 0.52-1.09
0.007 0.017 0.137
Referent 1.44
1.25-1.67
⬍ 0.001
CI, confidence interval; MI, myocardial infarction; OR, odds ratio. * Sensitivity analysis, restricted to subjects with an occlusion.
laterals were also associated with improved adjusted survival for both acute coronary syndrome (ACS) (HR 0.90; 95% CI, 0.84-0.96) and non-ACS (HR 0.84; 95% CI, 0.77-0.92) groups. For those treated with coronary artery bypass graft (CABG) within 1 year, coronary collaterals did not influence survival (adjusted HR 1.00; 95% CI, 0.89-1.11). For those treated with percutaneous coronary intervention (PCI), however, adjusted survival was better for those with collaterals (adjusted HR 0.78; 95% CI, 0.71-0.85). In the medical therapy group, there was no difference in survival with and without collaterals (adjusted HR 1.00; 95% CI, 0.92-1.09). When those subjects treated medically were subdivided into those with low risk CAD and those with high risk or left main CAD, survival was better for those with low risk CAD who had col-
laterals (adjusted HR 0.84; 95% CI, 0.72-98) but not for those with high risk or left main CAD (adjusted HR 1.08; 95% CI, 0.97-1.20). These results did not change after adjustment for medications. Discussion A key strength of this work is our large data set, which is larger than most similar studies, includes validated comorbidity and survival data for up to 12 years, and covers the range of cases observed in routine practice. The rich anatomy data within APPROACH also facilitated adjustment for the influence of the extent of CAD on coronary collaterals.36 Our data show this is critical to reduce confounding by CAD severity. We found that coronary collateral formation is strongly associated with coronary occlusions. Some correlates with small positive associations, including pulmonary disease, hyperlipidemia, and tobacco smoke, may simply be markers of more advanced CAD. That collateral formation is associated with CAD severity has been corroborated by others12 and supports the hypothesis that collaterals develop from pre-existing bridging arterioles that expand in response to increased shear stress caused by stenosis or occlusion.43 Prior thrombolytic therapy and acute indications for angiography were negatively associated with collaterals. Because coronary collaterals develop over time,44 patients with recent thrombolytic therapy or acute presentations perhaps had less time to develop collaterals. We did not observe negative associations with collaterals for atherosclerotic risk factors that have previously been reported. Possible explanations include study design considerations for prior studies, such as the use of case-control designs with potential for selection bias,7,8,13,16,18 relatively small sample sizes,7-13,16-20,22,45-48 and lack of adjustment of the results for the coronary anatomy and myocardium at risk.7-9,11,13,16,17,19,20,45,46 Our data suggest a robust but small absolute increase in survival that is associated with angiographic coronary collaterals. Our data agree with the findings of Meier et al., who evaluated 845 stable angina patients and found that better recruitable collateral flow measured invasively with pressure wires was associated with better survival over 10 years.6 Because
Figure 2. Crude and adjusted survival curves for patients with and without collaterals. After adjustment, coronary collaterals were associated with a small but significant survival benefit.
586
Canadian Journal of Cardiology Volume 27 2011
Table 4. The association between angiographically apparent collaterals and survival stratified by presentation and treatment Cohort Full cohort Presentation ACS Non-ACS Treatment† CABG PCI Medical Medical treatment Low risk High risk or left main
n
Crude HR
95% CI
Adjusted HR*
95% CI
55,751
1.32
1.27-1.37
0.89
0.85-0.94
35,592 20,159
1.50 1.08
1.42-1.57 1.01-1.15
0.90 0.84
0.84-0.96 0.77-0.92
13,273 26,369 16,109
1.07 1.12 1.50
0.99-1.16 1.04-1.21 1.42-1.60
1.00 0.78 1.00
0.89-1.11 0.71-0.85 0.92-1.09
10,113 5996
1.05 1.33
0.96-1.16 1.23-1.45
0.84 1.08
0.72-0.98 0.97-1.20
ACS, acute coronary syndrome; CABG, coronary artery bypass graft; CI, confidence interval; HR, hazard ratio; PCI, percutaneous coronary intervention. * Adjusted for demographic, comorbidities, ejection fraction, Duke Jeopardy Score, and revascularization (for full cohort and stratified by ACS). † Treatment here describes treatment within 1 calendar year from the baseline coronary angiogram.
most other prognosis studies involve composite endpoints, our findings are a significant contribution to the literature. The fact that our crude hazard ratios were above unity may explain why some prior studies suggest that collaterals may be associated with adverse prognosis,27,29,49 and illustrate why adjustment for clinical variables is important. We found that collaterals were associated with better survival in patients treated with PCI, in accordance with other investigators.12,24,50 Our finding of lack of association between coronary collaterals and survival for those undergoing CABG agree with the results of Caputo et al.,49 but not Regieli et al.24 This might be explained by the use of a composite endpoint (myocardial infarction, coronary heart disease death, and percutaneous or surgical revascularization) by Regieli et al., and shorter follow-up period, though their analysis did not find a statistically significant difference for the CABG subgroup.24 Our findings of an association between collaterals and better survival in low risk patients treated medically is also in agreement with the findings of Meier et al.6 and Regieli et al.24 We speculate that our observation for the high risk group treated medically, that coronary collaterals are not associated with better survival, was not observed by Regieli et al. due to the small size of their medical therapy cohort (N ⫽ 372). Why might coronary collaterals not be protective for those undergoing CABG or those with high risk anatomy treated medically? Perhaps surgical revascularization leads to regression in coronary collaterals,51 and these “recruitable” channels cannot contribute to better outcomes. For those with high risk anatomy treated medically, perhaps ischemic myocardium exceeds the capacity of collaterals,52 or perhaps there is more coronary steal.53 An alternative explanation may be that our study design, which grouped all patients with any visible collaterals, might have led the high risk group to be dominated with visible but small and insufficient collaterals. Our study has limitations. We rely on the binary description of absence or presence of angiographically apparent coronary collaterals by angiographers. However, APPROACH data have previously been shown to accurately predict outcome in patients with CAD.36 We did not discriminate between ipsilateral and contralateral collaterals, and collaterals might require an unobstructed source to be effective. We did not quantify coronary collaterals with Rentrop et al.41 or continuous connection scores,39 because review of all films was not possible. Reassuringly, our statistic for agreement between expert consensus review and APPROACH indicates accurate collat-
eral data within APPROACH. Most disagreements were cases where the reviewers identified collaterals not coded in APPROACH, suggesting a small degree of underreporting. Repeating our analysis for those with occlusions, for which angiography is accurate,39 directly addressed this limitation, and our robust findings of improved survival in those with coronary collaterals suggests that our assessment was valid. Conclusion The major determinant of coronary collaterals is the presence of obstructive CAD. Angiographically apparent coronary collaterals are associated with better survival in most subjects, though coronary collaterals at the time of coronary angiogram do not influence survival in those treated with coronary bypass graft surgery, or those with high-risk or left main disease treated without revascularization. Funding Sources Funding sources include the Libin Trust Fund and the Alberta Heritage Foundation for Medical Research. APPROACH was originally funded by the Weston Foundation, and now operates with support from the Canadian Cardiovascular Outcomes Research Team and the Province Wide Services Committee of Alberta Health and Wellness. Disclosures The authors have no conflicts of interest to disclose. References 1. Wustmann K, Zbinden S, Windecker S, Meier B, Seiler C. Is there functional collateral flow during vascular occlusion in angiographically normal coronary arteries? Circulation 2003;107:2213-20. 2. Heil M, Schaper W. Pathophysiology of collateral development. Coron Artery Dis 2004;15:373-8. 3. Habib GB, Heibig J, Forman SA, et al. Influence of coronary collateral vessels on myocardial infarct size in humans. Results of phase I thrombolysis in myocardial infarction (TIMI) trial. The TIMI Investigators. Circulation 1991;83:739-46. 4. Billinger M, Kloos P, Eberli FR, et al. Physiologically assessed coronary collateral flow and adverse cardiac ischemic events: a follow-up study in 403 patients with coronary artery disease. J Am Coll Cardiol 2002;40:1545-50.
McMurtry et al. Coronary Collaterals and Survival
587
5. Pijls NH, Bech GJ, el Gamal MI, et al. Quantification of recruitable coronary collateral blood flow in conscious humans and its potential to predict future ischemic events. J Am Coll Cardiol 1995;25:1522-8.
24. Regieli JJ, Jukema JW, Nathoe HM, et al. Coronary collaterals improve prognosis in patients with ischemic heart disease. Int J Cardiol 2009;132: 257-62.
6. Meier P, Gloekler S, Zbinden R, et al. Beneficial effect of recruitable collaterals: a 10-year follow-up study in patients with stable coronary artery disease undergoing quantitative collateral measurements. Circulation 2007;116:975-83.
25. Monteiro P, Antunes A, Goncalves LM, Providencia LA. Long-term clinical impact of coronary-collateral vessels after acute myocardial infarction. Rev Port Cardiol 2003;22:1051-61.
7. Turhan H, Yasar AS, Erbay AR, et al. Impaired coronary collateral vessel development in patients with metabolic syndrome. Coron Artery Dis 2005;16:281-5. 8. Kocaman SA, Arslan U, Tavil Y, et al. Increased circulating monocyte count is related to good collateral development in coronary artery disease. Atherosclerosis 2008;197:753-6. 9. Kornowski R. Collateral formation and clinical variables in obstructive coronary artery disease: the influence of hypercholesterolemia and diabetes mellitus. Coron Artery Dis 2003;14:61-4. 10. Yilmaz MB, Caldir V, Guray Y, et al. Relation of coronary collateral vessel development in patients with a totally occluded right coronary artery to the metabolic syndrome. Am J Cardiol 2006;97:636-9. 11. Kilian JG, Keech A, Adams MR, Celermajer DS. Coronary collateralization: determinants of adequate distal vessel filling after arterial occlusion. Coron Artery Dis 2002;13:155-9. 12. Nathoe HM, Koerselman J, Buskens E, et al. Determinants and prognostic significance of collaterals in patients undergoing coronary revascularization. Am J Cardiol 2006;98:31-5. 13. Nakae I, Fujita M, Miwa K, et al. Age-dependent impairment of coronary collateral development in humans. Heart Vessels 2000;15:176-80. 14. Balci B, Yilmaz O. Extent of coronary collateral vessel decrease with advanced age. Acta Cardiol 2004;59:431-4. 15. Kurotobi T, Sato H, Kinjo K, et al. Reduced collateral circulation to the infarct-related artery in elderly patients with acute myocardial infarction. J Am Coll Cardiol 2004;44:28-34. 16. Zbinden R, Zbinden S, Billinger M, et al. Influence of diabetes mellitus on coronary collateral flow: an answer to an old controversy. Heart 2005;91: 1289-93. 17. Mason MJ, Walker SK, Patel DJ, Paul VE, Ilsley CD. Influence of clinical and angiographic factors on development of collateral channels. Coron Artery Dis 2000;11:573-8. 18. Abaci A, Oguzhan A, Kahraman S, et al. Effect of diabetes mellitus on formation of coronary collateral vessels. Circulation 1999;99:2239-42. 19. Nurkalem Z, Gorgulu S, Uslu N, et al. Effect of glucometabolic disorders on the formation of coronary collaterals in occlusive coronary artery disease. Acta Cardiol 2008;63:141-6.
26. Boehrer JD, Lange RA, Willard JE, Hillis LD. Influence of collateral filling of the occluded infarct-related coronary artery on prognosis after acute myocardial infarction. Am J Cardiol 1992;69:10-2. 27. Gohlke H, Heim E, Roskamm H. Prognostic importance of collateral flow and residual coronary stenosis of the myocardial infarct artery after anterior wall Q-wave acute myocardial infarction. Am J Cardiol 1991;67: 1165-9. 28. Koerselman J, de Jaegere PP, Verhaar MC, Grobbee DE, van der Graaf Y. Prognostic significance of coronary collaterals in patients with coronary heart disease having percutaneous transluminal coronary angioplasty. Am J Cardiol 2005;96:390-4. 29. Nicolau JC, Nogueira PR, Pinto MA, Serrano CV Jr, Garzon SA. Early infarct artery collateral flow does not improve long-term survival following thrombolytic therapy for acute myocardial infarction. Am J Cardiol 1999; 83:21-6. 30. Steg PG, Kerner A, Mancini GB, et al. Impact of collateral flow to the occluded infarct-related artery on clinical outcomes in patients with recent myocardial infarction: a report from the randomized occluded artery trial. Circulation 2010;121:2724-30. 31. van Liebergen RA, Piek JJ, Koch KT, et al. Quantification of collateral flow in humans: a comparison of angiographic, electrocardiographic and hemodynamic variables. J Am Coll Cardiol 1999;33:670-7. 32. Ghali WA, Knudtson ML. Overview of the Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease. On behalf of the APPROACH investigators. Can J Cardiol 2000;16:1225-30. 33. Norris CM, Ghali WA, Knudtson ML, Naylor CD, Saunders LD. Dealing with missing data in observational health care outcome analyses. J Clin Epidemiol 2000;53:377-83. 34. Faris PD, Ghali WA, Brant R, et al. Multiple imputation versus data enhancement for dealing with missing data in observational health care outcome analyses. J Clin Epidemiol 2002;55:184-91. 35. McGinn T, Wyer PC, Newman TB, et al. Tips for learners of evidencebased medicine: 3. Measures of observer variability (kappa statistic). CMAJ 2004;171:1369-73. 36. Graham MM, Faris PD, Ghali WA, et al. Validation of three myocardial jeopardy scores in a population-based cardiac catheterization cohort. Am Heart J 2001;142:254-61. 37. McAlister FA, Oreopoulos A, Norris CM, et al. Exploring the treatmentrisk paradox in coronary disease. Arch Intern Med 2007;167:1019-25.
20. Kyriakides ZS, Kremastinos DT, Michelakakis NA, et al. Coronary collateral circulation in coronary artery disease and systemic hypertension. Am J Cardiol 1991;67:687-90.
38. Sun GW, Shook TL, Kay GL. Inappropriate use of bivariable analysis to screen risk factors for use in multivariable analysis. J Clin Epidemiol 1996; 49:907-16.
21. Koerselman J, de Jaegere PP, Verhaar MC, Grobbee DE, van der Graaf Y. Coronary collateral circulation: the effects of smoking and alcohol. Atherosclerosis 2007;191:191-8.
39. Werner GS, Ferrari M, Heinke S, et al. Angiographic assessment of collateral connections in comparison with invasively determined collateral function in chronic coronary occlusions. Circulation 2003;107:1972-7.
22. Dincer I, Ongun A, Turhan S, et al. Association between the dosage and duration of statin treatment with coronary collateral development. Coron Artery Dis 2006;17:561-5.
40. Ghali WA, Quan H, Brant R, et al. Comparison of 2 methods for calculating adjusted survival curves from proportional hazards models. JAMA 2001;286:1494-7.
23. Koerselman J, de Jaegere PP, Verhaar MC, Grobbee DE, der Graaf Y. Cardiac ischemic score determines the presence of coronary collateral circulation. Cardiovasc Drugs Ther 2005;19:283-9.
41. Rentrop KP, Cohen M, Blanke H, Phillips RA. Changes in collateral channel filling immediately after controlled coronary artery occlusion by an angioplasty balloon in human subjects. J Am Coll Cardiol 1985;5:587-92.
588
Canadian Journal of Cardiology Volume 27 2011
42. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159-74.
adverse events in patients with diabetes mellitus. Eur Heart J 2003;24:1134-42.
43. Schaper W. Collateral circulation: past and present. Basic Res Cardiol 2009;104:5-21.
49. Caputo M, Anis RR, Rogers CA, et al. Coronary collateral circulation: effect on early and midterm outcomes after off-pump coronary artery bypass surgery. Ann Thorac Surg 2008;85:71-9.
44. Werner GS, Ferrari M, Betge S, et al. Collateral function in chronic total coronary occlusions is related to regional myocardial function and duration of occlusion. Circulation 2001;104:2784-90. 45. Koerselman J, de Jaegere PP, Verhaar MC, van der Graaf Y, Grobbee DE. High blood pressure is inversely related with the presence and extent of coronary collaterals. J Hum Hypertens 2005;19:809-17. 46. Pourati I, Kimmelstiel C, Rand W, Karas RH. Statin use is associated with enhanced collateralization of severely diseased coronary arteries. Am Heart J 2003;146:876-81. 47. Olijhoek JK, Koerselman J, de Jaegere PP, et al. Presence of the metabolic syndrome does not impair coronary collateral vessel formation in patients with documented coronary artery disease. Diabetes Care 2005;28:683-9. 48. Werner GS, Richartz BM, Heinke S, Ferrari M, Figulla HR. Impaired acute collateral recruitment as a possible mechanism for increased cardiac
50. Abbott JD, Choi EJ, Selzer F, Srinivas VS, Williams DO. Impact of coronary collaterals on outcome following percutaneous coronary intervention (from the National Heart, Lung, and Blood Institute Dynamic Registry). Am J Cardiol 2005;96:676-80. 51. Werner GS, Emig U, Mutschke O, et al. Regression of collateral function after recanalization of chronic total coronary occlusions: a serial assessment by intracoronary pressure and Doppler recordings. Circulation 2003;108:2877-82. 52. Werner GS, Surber R, Ferrari M, Fritzenwanger M, Figulla HR. The functional reserve of collaterals supplying long-term chronic total coronary occlusions in patients without prior myocardial infarction. Eur Heart J 2006;27:2406-12. 53. Werner GS, Fritzenwanger M, Prochnau D, et al. Determinants of coronary steal in chronic total coronary occlusions donor artery, collateral, and microvascular resistance. J Am Coll Cardiol 2006;48:51-8.
Clinical Research
The Clinical Profile and Outcomes Associated With Coronary Collaterals in Patients With Coronary Artery Disease M. Sean McMurtry, MD, PhD,a Adriane M. Lewin, BA, MSc,b Merril L. Knudtson, MD,b William A. Ghali, MD, MPH,b P. Diane Galbraith, BN,b Felix Schulte, MD,a,b Colleen M. Norris, PhD,a and Michelle M. Graham, MDa a
University of Alberta, Edmonton, Alberta, Canada b
University of Calgary, Calgary, Alberta, Canada
ABSTRACT
RÉSUMÉ
Background: The clinical correlates of coronary collaterals and the effects of coronary collaterals on prognosis are incompletely understood. Methods: We performed a study of 55,751 patients undergoing coronary angiography to evaluate the correlates of angiographically apparent coronary collaterals, and to evaluate their association with survival. Results: The characteristic most strongly associated with the presence of collaterals was a coronary occlusion (odds ratio [OR], 28.9; 95% confidence interval [CI], 27.1-30.6). Collaterals were associated with improved adjusted survival overall (hazard ratio [HR] 0.89; 95% CI, 0.85-0.95), and in both acute coronary syndrome (ACS) (HR 0.90; 95% CI, 0.84-0.96) and non-ACS (HR 0.84; 95% CI, 0.77-0.92) patients. Collaterals were associated with improved survival in those receiving angioplasty (HR 0.78; 95% CI, 0.71-0.85) and those with low risk anatomy treated medically (HR 0.84; 95% CI, 0.72-0.98), but not for those treated with coronary bypass graft surgery or those with high-risk anatomy treated without revascularization. Conclusions: The major correlate of coronary collaterals is the presence/extent of obstructive coronary artery disease. Collaterals are associated with better survival overall and in both ACS and non-ACS presentations, but not for those treated with coronary artery bypass graft (CABG) or those with high-risk anatomy who are not revascularized.
Introduction : Les corrélats cliniques des collatérales coronaires et les effets des collatérales coronaires sur le pronostic sont partiellement compris. Méthodes : Nous avons mené une étude auprès de 55 751 patients ayant subi une coronarographie pour évaluer les corrélats des collatérales coronaires apparentes à l’angiographie, ainsi que pour évaluer leur lien avec la survie. Résultats : Les caractéristiques les plus fortement associées à la présence des collatérales étaient l’occlusion coronarienne (ratio d’incidence approché [RIA], 28,9; intervalle de confiance [IC] de 95 %, 27,1-30,6). Les collatérales étaient associées à l’amélioration de la survie globale ajustée (rapport de risque [RR] 0,89; IC de 95 %, 0,85-0,95), et aux patients avec le syndrome coronarien aigu (SCA) (RR de 0,90; IC de 95 %, 0,84-0,96) et aux patients avec le non SCA (RR de 0,84; IC de 95 %, 0,77-0,92). Les collatérales étaient associées à l’amélioration de la survie chez ceux qui avaient subi une angioplastie (RR de 0,78; IC de 95 %, 0,71-0,85) et chez ceux qui avaient eu un risque anatomique traité médicalement (RR de 0,84; IC de 95 %, 0,72-0,98), mais non chez ceux qui avaient été traités par un pontage coronarien ou chez ceux avec une anatomie à haut risque qui avaient été traités sans revascularisation. Conclusions : Le plus grand corrélat de collatérales coronaires est la présence/l’étendue de la maladie coronarienne obstructive. Les collatérales sont associées à une meilleure survie globale et aux tableaux cliniques du SCA et du non SCA, mais ne sont pas associées aux patients traités par un pontage aortocoronarien (PAC) ou à ceux ayant une anatomie à haut risque qui ne sont pas revascularisés.
Coronary collaterals develop from pre-existing arterioles1 that expand in response to a stenosis or occlusion.2 They are thought protective in coronary artery disease (CAD), being
associated with smaller myocardial infarct size,3 fewer ischemic events,4,5 and better survival in patients with stable CAD.6 It is unknown why some have large coronary collaterals while others do not. For most variables that have been evaluated, conflicting reports exist (Table 1). Most studies are limited by small sample size or by confounding due to lack of adjustment for CAD burden.12,22,23 Though better coronary collateral function, measured invasively, is associated with fewer ischemic events5 and better sur-
Received for publication January 13, 2011. Accepted February 24, 2011. Corresponding author: Dr M.S. McMurtry, University of Alberta, 8440 112th Street, Edmonton, Alberta T6G 2B7, Canada. Tel.: ⫹1-780-407-2799. E-mail: [email protected] See page 586 for disclosure information.
0828-282X/$ – see front matter © 2011 Canadian Cardiovascular Society. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.cjca.2011.02.011
582
Canadian Journal of Cardiology Volume 27 2011
Table 1. Literature reports of clinical correlates of coronary collaterals are inconsistent Variable Age Diabetes Hypertension Tobacco smoke Dyslipidemia Sex Metabolic syndrome Obesity Impaired fasting glucose Stenosis severity
Positive association
20 21 9,11
No association 7-11 8,12,15-17 7-12,15,17 7-12,15,17 7,8,10,12,15,17 7-12,17
Negative association 12-15 9-11,18,19 45
7,10 7 19 12,17
vival in chronic stable CAD,6 the importance of angiographically measured coronary collaterals is less certain. Composite endpoints predominate in the literature, reflecting sample size constraints. Discrepancies exist: angiographically apparent coronary collaterals are associated with better outcomes in some studies,4,12,24,25 but not all,26-30 and some suggested that collaterals imply worse outcomes.27-29 Despite angiography’s limitations,31 the prognostic implications of angiographically apparent coronary collaterals remains an important question. To identify influences on coronary collateral formation and to evaluate whether collaterals influence survival, we measured the clinical correlates of collaterals and the association of collaterals with survival in an unselected cohort. Methods Study sample We used the Alberta Provincial Project for Outcomes Assessment in Coronary Heart Disease (APPROACH) database, an ongoing initiative in Alberta, Canada, since 1995.32 The cohort consisted of all Alberta residents undergoing catheterization between April 1, 1995, and March 31, 2007. Staff enter data and comorbidities are verified and data are enhanced to ensure comorbidity data are accurate and there are no empty fields.33,34 Mortality is ascertained through linkage to the Alberta Bureau of Vital Statistics. The Health Research Ethics Boards of the Universities of Calgary and Alberta approved the protocol. Variables APPROACH includes age, sex, left ventricular ejection fraction, myocardial infarction, congestive heart failure, diabetes, hypertension, hyperlipidemia, current and prior smoking, cerebrovascular disease, peripheral arterial disease, chronic pulmonary disease, chronic kidney disease, gastrointestinal disease, malignancy, and cardiovascular medication use. It records coronary anatomy, including angiographically apparent collaterals (present or absent) and therapeutic interventions. Inclusion criteria We included consecutive patients undergoing coronary angiography for stable angina, unstable angina, ST-elevation myocardial infarction (STEMI) and non-ST– elevation myocardial infarction (NSTEMI). We excluded patients with prior coronary revascularization or stenosis of less than 50%, nonAlberta residents, and patients with incomplete comorbidity
data. For the first analysis (correlates of collaterals), patients with STEMI were excluded to limit confounding due to time variation. Outcome variable For the first analysis, the main outcome was presence of angiographically apparent coronary collaterals. To verify the validity of APPROACH collateral data, the presence of angiographically apparent collaterals for 100 randomly selected subjects was determined by consensus agreement of 2 expert angiographers who independently reviewed the films and were blinded to the APPROACH data. The results of the consensus agreement were compared with APPROACH, and a statistic was calculated.35 For the second analysis, the main outcome measure was all cause mortality. Duke Jeopardy Score This score encompasses both the severity of a coronary lesion and the volume of myocardium subtended by the stenosis and provides prognostic information.36 Statistical analysis The baseline characteristics were compared with 2 or t tests. Crude and adjusted odds ratios for correlates were calculated by risk status, with high risk anatomy defined as 3-vessel disease, left main disease, or 2-vessel disease involving the proximal left anterior descending coronary artery per the Duke Coronary Index.37 Adjustment variables included demographic information, comorbidities, indication, left ventricular ejection fraction, Duke Jeopardy Score, and cardiac history. To account for hyperlipidemia as a marker for statin use, we included statin use as a binary variable. Odds ratios (ORs) for correlates were calculated using a backward stepwise multivariable logistic regression model.38 We performed a sensitivity analysis for those with at least 1 total occlusion of a major epicardial coronary artery, the anatomy for which angiographic assessment of collaterals is most accurate.39 After ensuring the proportional hazards assumption was met, we evaluated survival using the corrected groups prognosis method.40 We adjusted for clinical variables, including revascularization. We also stratified our results by presentation, treatment type in the following year, and further stratified the group treated with medical therapy by anatomy. SAS software version 9 (SAS Institute Inc, Cary, NC, USA) was used. Results Cohort description Of the 95,646 coronary angiograms between April 1, 1995, and March 31, 2007, a total of 55,751 met our inclusion criteria (Fig. 1). The clinical characteristics are displayed in Table 2. Most subjects with visible collaterals had a coronary occlusion. A total of 10,331 patients with STEMI were excluded from the cohort for the analysis of clinical correlates. Assessment of accuracy collateral data One hundred coronary angiograms were selected randomly from the cohort and independently reviewed by expert angiographers. A consensus opinion regarding the presence or absence
McMurtry et al. Coronary Collaterals and Survival
583
All paents in Alberta between April 1, 1995 and March 31, 2007 N = 95,646
Excluded: Non-Alberta residents (7655) Incomplete comorbidies (1697) Total: 9352
Alberta paents, comorbidity data complete N = 86,294
Excluded: Prior PTCA (5405) Prior CABG (3467) Normal coronaries or lesions < 50% (21,612) Duke Jeopardy Score missing (59) Total: 30,543
Cohort for Analysis of Survival N = 55,751
Collaterals Present N = 21,893 (39.3%)
At least one occlusion N = 15,347 (70.1%)
No occlusions N = 6546 (29.9%)
Collaterals Absent N = 33,858 (60.7%)
At least one occlusion N = 3420 (10.1 %)
No occlusions N = 30,438 (89.9%)
Figure 1. The cohort for analysis of coronary collaterals and survival. After exclusions, a total of 55,751 patients were included. CABG, coronary artery bypass graft; PTCA, percutaneous transluminal coronary angioplasty.
of angiographically apparent coronary collaterals was obtained. This consensus opinion, or gold standard, was compared with the APPROACH data. Of the 100 cases, 88 cases matched perfectly, rivaling the interobserver variability for angiographically apparent collaterals within other carefully done studies in the literature.39 In 2 cases, APPROACH data suggested that collaterals were present when they were not. In the remaining 10 cases, expert consensus identified subtle (Rentrop 141) collaterals that were not coded in APPROACH, suggesting a small degree of systematic underreporting of Rentrop 1 collaterals in APPROACH. The statistic was 0.74, just under the threshold of 0.75 for excellent agreement beyond chance.42
ence of an occlusion (OR 28.9; 95% CI, 27.1-30.6). Because the presence of collaterals was so overwhelmingly associated with occlusions, a sensitivity analysis restricted to patients with at least 1 occlusion was performed. As before, those with highrisk anatomy were more likely to have collaterals (OR 1.44; 95% CI, 1.25-1.67). The odds ratios for correlates of coronary collaterals (Table 3) demonstrate small positive associations for pulmonary disease, hyperlipidemia, peripheral arterial disease, statin therapy, previous smoking, and malignancy. Prior thrombolytic therapy and indications other than stable angina were negatively associated with collaterals. No sex-based differences were observed.
Correlates of coronary collaterals
Coronary collaterals and survival
Compared with lower risk patients, those with 3-vessel disease, left main disease, and 2-vessel disease involving the proximal left anterior descending coronary artery were more likely to have angiographically apparent coronary collaterals (OR 2.74; 95% confidence interval [CI], 2.64-2.85). The characteristic most strongly associated with collaterals was the pres-
Patients with visible coronary collaterals at the time of coronary angiography had worse long-term unadjusted survival those without coronary collaterals (Fig. 2), with a crude hazard ratio of 1.32 (95% CI, 1.27-1.37; Table 4). After adjustment, coronary collaterals were associated with better survival overall (hazard ratio [HR] 0.89; 95% CI, 0.85-0.94). Coronary col-
584
Canadian Journal of Cardiology Volume 27 2011
Table 2. Patient characteristics of the included cohort (N ⫽ 55,751)
Demographics Mean age (SD) Aged ⬎75 years Male Deceased (date of death in dataset) Comorbidities Cerebrovascular disease Pulmonary disease Congestive heart failure Liver/gastrointestinal disease Peripheral vascular disease Renal disease (creatinine ⬎200 mmol/L) Malignancy Hypertension Hyperlipidemia Dialysis Diabetes Current smoker Previous smoker Prior lytic therapy Cardiac history Prior myocardial infarction Indication for coronary angiogram Stable angina Myocardial infarction Unstable angina Other Ejection fraction ⬎50% 35%-50% 20%-34% ⬍20% Could not be performed Unknown Duke Jeopardy Score 0/12 2/12 4/12 6/12 8/12 10/12 12/12 Coronary anatomy* 1 VD 50%-75% 1 VD 95% 2 VD 2 VD PLAD 95% 1 VD 95% PLAD 2 VD 95% LAD 2 VD 95% PLAD 3 VD 3 VD 1%-95% 3 VD PLAD 3 VD 95% PLAD Left main Severe left main Revascularization within 1 year of catheterization Treatment within 1 year of catheterization Medical CABG within 1 year PCI within 1 year
Collaterals absent (n ⫽ 33,858; 60.7%)
Collaterals present (n ⫽ 21,893; 39.3%)
2 P value
63.2 (11.7) 6012 (18.4) 24,449 (72.2) 5301 (15.7)
64.3 (11.3) 4217 (19.3) 16,971 (77.5) 4562 (20.8)
⬍ 0.0001 0.006 ⬍ 0.0001 ⬍ 0.001
2184 (6.5) 4189 (12.4) 4185 (12.4) 1879 (5.6) 2326 (6.9) 1210 (3.6) 1243 (3.7) 20,231 (59.8) 21,456 (63.4) 539 (1.6) 7519 (22.2) 10,455 (30.9) 13,118 (38.7) 3397 (10.0)
1749 (8.0) 2987 (13.6) 3602 (16.5) 1235 (5.6) 2167 (9.9) 935 (4.3) 924 (4.2) 13,604 (62.1) 14,501 (66.2) 354 (1.6) 5658 (25.8) 6368 (29.1) 9111 (41.6) 1143 (5.2)
⬍ 0.0001 ⬍ 0.0001 ⬍ 0.0001 0.65 ⬍ 0.0001 ⬍ 0.0001 0.001 ⬍ 0.0001 ⬍ 0.0001 0.82 ⬍ 0.0001 ⬍ 0.0001 ⬍ 0.0001 ⬍ 0.0001
13,920 (41.1)
10,944 (50.0)
⬍ 0.0001 ⬍ 0.0001
7952 (23.5) 14,970 (44.2) 8050 (23.8) 2886 ( 8.5)
7630 (34.9) 7941 (36.3) 4631 (21.2) 1691 (7.7)
21,392 (63.2) 6701 (19.8) 1406 (4.2) 207 (0.6) 2776 (8.2) 1376 (4.1)
11,431 (52.2) 6142 (28.1) 1896 (8.7) 331 (1.5) 1474 (6.7) 619 (2.8)
4892 (14.5) 11,710 (34.6) 4958 (14.6) 5616 (16.6) 2428 (7.2) 2099 (6.2) 2155 (6.4)
195 (0.9) 5242 (23.9) 3906 (17.8) 3779 (17.3) 3168 (14.5) 2243 (10.3) 3360 (15.4)
2852 (8.4) 8617 (25.5) 3238 (9.6) 2256 (6.7) 2216 (6.5) 2429 (7.2) 1699 (5.0) 160 (0.5) 4317 (12.8) 1181 (3.5) 2373 (7.0) 1136 (3.4) 1384 (4.1) 24,190 (71.5)
20 (0.1) 3344 (15.3) 1208 (5.5) 1835 (8.4) 614 (2.8) 1619 (7.4) 1123 (5.1) 23 (0.1) 5039 (23.0) 1359 (6.2) 3200 (14.6) 1243 (5.7) 1266 (5.8) 15,452 (70.6)
9668 (28.6) 5915 (17.5) 18,275 (54.0)
6441 (29.4) 7358 (33.6) 8094 (37.0)
⬍ 0.0001
⬍ 0.0001
⬍ 0.0001
0.03 ⬍ 0.0001
Data are presented as n (%). CABG, coronary artery bypass graft; LAD, left anterior descending coronary artery; PCI, percutaneous coronary intervention; PLAD, proximal LAD; VD, vessel disease. * Percentage stenosis.
McMurtry et al. Coronary Collaterals and Survival
585
Table 3. Clinical correlates of angiographically apparent coronary collaterals* N ⫽ 14,731 Variable Comorbidities Pulmonary disease Peripheral vascular disease Malignancy Hyperlipidemia Previous smoker Prior lytic therapy Statin Indication Stable angina MI Unstable angina Other Ejection fraction, % ⬎50 35-50 20-34 ⬍20 Coronary anatomy Low risk High risk/left main
OR
95% CI
P value
1.44 1.26 1.40 1.22 1.12 0.79 1.53
1.22-1.69 1.05-1.51 1.05-1.87 1.08-1.38 1.01-1.24 0.64-0.97 1.34-1.75
⬍ 0.001 0.015 0.022 0.001 0.040 0.026 ⬍ 0.001
Referent 0.36 0.57 0.69
0.32-0.41 0.49-0.66 0.56-0.85
⬍ 0.001 ⬍ 0.001 ⬍ 0.001
Referent 0.84 0.80 0.75
0.74-0.95 0.66-0.96 0.52-1.09
0.007 0.017 0.137
Referent 1.44
1.25-1.67
⬍ 0.001
CI, confidence interval; MI, myocardial infarction; OR, odds ratio. * Sensitivity analysis, restricted to subjects with an occlusion.
laterals were also associated with improved adjusted survival for both acute coronary syndrome (ACS) (HR 0.90; 95% CI, 0.84-0.96) and non-ACS (HR 0.84; 95% CI, 0.77-0.92) groups. For those treated with coronary artery bypass graft (CABG) within 1 year, coronary collaterals did not influence survival (adjusted HR 1.00; 95% CI, 0.89-1.11). For those treated with percutaneous coronary intervention (PCI), however, adjusted survival was better for those with collaterals (adjusted HR 0.78; 95% CI, 0.71-0.85). In the medical therapy group, there was no difference in survival with and without collaterals (adjusted HR 1.00; 95% CI, 0.92-1.09). When those subjects treated medically were subdivided into those with low risk CAD and those with high risk or left main CAD, survival was better for those with low risk CAD who had col-
laterals (adjusted HR 0.84; 95% CI, 0.72-98) but not for those with high risk or left main CAD (adjusted HR 1.08; 95% CI, 0.97-1.20). These results did not change after adjustment for medications. Discussion A key strength of this work is our large data set, which is larger than most similar studies, includes validated comorbidity and survival data for up to 12 years, and covers the range of cases observed in routine practice. The rich anatomy data within APPROACH also facilitated adjustment for the influence of the extent of CAD on coronary collaterals.36 Our data show this is critical to reduce confounding by CAD severity. We found that coronary collateral formation is strongly associated with coronary occlusions. Some correlates with small positive associations, including pulmonary disease, hyperlipidemia, and tobacco smoke, may simply be markers of more advanced CAD. That collateral formation is associated with CAD severity has been corroborated by others12 and supports the hypothesis that collaterals develop from pre-existing bridging arterioles that expand in response to increased shear stress caused by stenosis or occlusion.43 Prior thrombolytic therapy and acute indications for angiography were negatively associated with collaterals. Because coronary collaterals develop over time,44 patients with recent thrombolytic therapy or acute presentations perhaps had less time to develop collaterals. We did not observe negative associations with collaterals for atherosclerotic risk factors that have previously been reported. Possible explanations include study design considerations for prior studies, such as the use of case-control designs with potential for selection bias,7,8,13,16,18 relatively small sample sizes,7-13,16-20,22,45-48 and lack of adjustment of the results for the coronary anatomy and myocardium at risk.7-9,11,13,16,17,19,20,45,46 Our data suggest a robust but small absolute increase in survival that is associated with angiographic coronary collaterals. Our data agree with the findings of Meier et al., who evaluated 845 stable angina patients and found that better recruitable collateral flow measured invasively with pressure wires was associated with better survival over 10 years.6 Because
Figure 2. Crude and adjusted survival curves for patients with and without collaterals. After adjustment, coronary collaterals were associated with a small but significant survival benefit.
586
Canadian Journal of Cardiology Volume 27 2011
Table 4. The association between angiographically apparent collaterals and survival stratified by presentation and treatment Cohort Full cohort Presentation ACS Non-ACS Treatment† CABG PCI Medical Medical treatment Low risk High risk or left main
n
Crude HR
95% CI
Adjusted HR*
95% CI
55,751
1.32
1.27-1.37
0.89
0.85-0.94
35,592 20,159
1.50 1.08
1.42-1.57 1.01-1.15
0.90 0.84
0.84-0.96 0.77-0.92
13,273 26,369 16,109
1.07 1.12 1.50
0.99-1.16 1.04-1.21 1.42-1.60
1.00 0.78 1.00
0.89-1.11 0.71-0.85 0.92-1.09
10,113 5996
1.05 1.33
0.96-1.16 1.23-1.45
0.84 1.08
0.72-0.98 0.97-1.20
ACS, acute coronary syndrome; CABG, coronary artery bypass graft; CI, confidence interval; HR, hazard ratio; PCI, percutaneous coronary intervention. * Adjusted for demographic, comorbidities, ejection fraction, Duke Jeopardy Score, and revascularization (for full cohort and stratified by ACS). † Treatment here describes treatment within 1 calendar year from the baseline coronary angiogram.
most other prognosis studies involve composite endpoints, our findings are a significant contribution to the literature. The fact that our crude hazard ratios were above unity may explain why some prior studies suggest that collaterals may be associated with adverse prognosis,27,29,49 and illustrate why adjustment for clinical variables is important. We found that collaterals were associated with better survival in patients treated with PCI, in accordance with other investigators.12,24,50 Our finding of lack of association between coronary collaterals and survival for those undergoing CABG agree with the results of Caputo et al.,49 but not Regieli et al.24 This might be explained by the use of a composite endpoint (myocardial infarction, coronary heart disease death, and percutaneous or surgical revascularization) by Regieli et al., and shorter follow-up period, though their analysis did not find a statistically significant difference for the CABG subgroup.24 Our findings of an association between collaterals and better survival in low risk patients treated medically is also in agreement with the findings of Meier et al.6 and Regieli et al.24 We speculate that our observation for the high risk group treated medically, that coronary collaterals are not associated with better survival, was not observed by Regieli et al. due to the small size of their medical therapy cohort (N ⫽ 372). Why might coronary collaterals not be protective for those undergoing CABG or those with high risk anatomy treated medically? Perhaps surgical revascularization leads to regression in coronary collaterals,51 and these “recruitable” channels cannot contribute to better outcomes. For those with high risk anatomy treated medically, perhaps ischemic myocardium exceeds the capacity of collaterals,52 or perhaps there is more coronary steal.53 An alternative explanation may be that our study design, which grouped all patients with any visible collaterals, might have led the high risk group to be dominated with visible but small and insufficient collaterals. Our study has limitations. We rely on the binary description of absence or presence of angiographically apparent coronary collaterals by angiographers. However, APPROACH data have previously been shown to accurately predict outcome in patients with CAD.36 We did not discriminate between ipsilateral and contralateral collaterals, and collaterals might require an unobstructed source to be effective. We did not quantify coronary collaterals with Rentrop et al.41 or continuous connection scores,39 because review of all films was not possible. Reassuringly, our statistic for agreement between expert consensus review and APPROACH indicates accurate collat-
eral data within APPROACH. Most disagreements were cases where the reviewers identified collaterals not coded in APPROACH, suggesting a small degree of underreporting. Repeating our analysis for those with occlusions, for which angiography is accurate,39 directly addressed this limitation, and our robust findings of improved survival in those with coronary collaterals suggests that our assessment was valid. Conclusion The major determinant of coronary collaterals is the presence of obstructive CAD. Angiographically apparent coronary collaterals are associated with better survival in most subjects, though coronary collaterals at the time of coronary angiogram do not influence survival in those treated with coronary bypass graft surgery, or those with high-risk or left main disease treated without revascularization. Funding Sources Funding sources include the Libin Trust Fund and the Alberta Heritage Foundation for Medical Research. APPROACH was originally funded by the Weston Foundation, and now operates with support from the Canadian Cardiovascular Outcomes Research Team and the Province Wide Services Committee of Alberta Health and Wellness. Disclosures The authors have no conflicts of interest to disclose. References 1. Wustmann K, Zbinden S, Windecker S, Meier B, Seiler C. Is there functional collateral flow during vascular occlusion in angiographically normal coronary arteries? Circulation 2003;107:2213-20. 2. Heil M, Schaper W. Pathophysiology of collateral development. Coron Artery Dis 2004;15:373-8. 3. Habib GB, Heibig J, Forman SA, et al. Influence of coronary collateral vessels on myocardial infarct size in humans. Results of phase I thrombolysis in myocardial infarction (TIMI) trial. The TIMI Investigators. Circulation 1991;83:739-46. 4. Billinger M, Kloos P, Eberli FR, et al. Physiologically assessed coronary collateral flow and adverse cardiac ischemic events: a follow-up study in 403 patients with coronary artery disease. J Am Coll Cardiol 2002;40:1545-50.
McMurtry et al. Coronary Collaterals and Survival
587
5. Pijls NH, Bech GJ, el Gamal MI, et al. Quantification of recruitable coronary collateral blood flow in conscious humans and its potential to predict future ischemic events. J Am Coll Cardiol 1995;25:1522-8.
24. Regieli JJ, Jukema JW, Nathoe HM, et al. Coronary collaterals improve prognosis in patients with ischemic heart disease. Int J Cardiol 2009;132: 257-62.
6. Meier P, Gloekler S, Zbinden R, et al. Beneficial effect of recruitable collaterals: a 10-year follow-up study in patients with stable coronary artery disease undergoing quantitative collateral measurements. Circulation 2007;116:975-83.
25. Monteiro P, Antunes A, Goncalves LM, Providencia LA. Long-term clinical impact of coronary-collateral vessels after acute myocardial infarction. Rev Port Cardiol 2003;22:1051-61.
7. Turhan H, Yasar AS, Erbay AR, et al. Impaired coronary collateral vessel development in patients with metabolic syndrome. Coron Artery Dis 2005;16:281-5. 8. Kocaman SA, Arslan U, Tavil Y, et al. Increased circulating monocyte count is related to good collateral development in coronary artery disease. Atherosclerosis 2008;197:753-6. 9. Kornowski R. Collateral formation and clinical variables in obstructive coronary artery disease: the influence of hypercholesterolemia and diabetes mellitus. Coron Artery Dis 2003;14:61-4. 10. Yilmaz MB, Caldir V, Guray Y, et al. Relation of coronary collateral vessel development in patients with a totally occluded right coronary artery to the metabolic syndrome. Am J Cardiol 2006;97:636-9. 11. Kilian JG, Keech A, Adams MR, Celermajer DS. Coronary collateralization: determinants of adequate distal vessel filling after arterial occlusion. Coron Artery Dis 2002;13:155-9. 12. Nathoe HM, Koerselman J, Buskens E, et al. Determinants and prognostic significance of collaterals in patients undergoing coronary revascularization. Am J Cardiol 2006;98:31-5. 13. Nakae I, Fujita M, Miwa K, et al. Age-dependent impairment of coronary collateral development in humans. Heart Vessels 2000;15:176-80. 14. Balci B, Yilmaz O. Extent of coronary collateral vessel decrease with advanced age. Acta Cardiol 2004;59:431-4. 15. Kurotobi T, Sato H, Kinjo K, et al. Reduced collateral circulation to the infarct-related artery in elderly patients with acute myocardial infarction. J Am Coll Cardiol 2004;44:28-34. 16. Zbinden R, Zbinden S, Billinger M, et al. Influence of diabetes mellitus on coronary collateral flow: an answer to an old controversy. Heart 2005;91: 1289-93. 17. Mason MJ, Walker SK, Patel DJ, Paul VE, Ilsley CD. Influence of clinical and angiographic factors on development of collateral channels. Coron Artery Dis 2000;11:573-8. 18. Abaci A, Oguzhan A, Kahraman S, et al. Effect of diabetes mellitus on formation of coronary collateral vessels. Circulation 1999;99:2239-42. 19. Nurkalem Z, Gorgulu S, Uslu N, et al. Effect of glucometabolic disorders on the formation of coronary collaterals in occlusive coronary artery disease. Acta Cardiol 2008;63:141-6.
26. Boehrer JD, Lange RA, Willard JE, Hillis LD. Influence of collateral filling of the occluded infarct-related coronary artery on prognosis after acute myocardial infarction. Am J Cardiol 1992;69:10-2. 27. Gohlke H, Heim E, Roskamm H. Prognostic importance of collateral flow and residual coronary stenosis of the myocardial infarct artery after anterior wall Q-wave acute myocardial infarction. Am J Cardiol 1991;67: 1165-9. 28. Koerselman J, de Jaegere PP, Verhaar MC, Grobbee DE, van der Graaf Y. Prognostic significance of coronary collaterals in patients with coronary heart disease having percutaneous transluminal coronary angioplasty. Am J Cardiol 2005;96:390-4. 29. Nicolau JC, Nogueira PR, Pinto MA, Serrano CV Jr, Garzon SA. Early infarct artery collateral flow does not improve long-term survival following thrombolytic therapy for acute myocardial infarction. Am J Cardiol 1999; 83:21-6. 30. Steg PG, Kerner A, Mancini GB, et al. Impact of collateral flow to the occluded infarct-related artery on clinical outcomes in patients with recent myocardial infarction: a report from the randomized occluded artery trial. Circulation 2010;121:2724-30. 31. van Liebergen RA, Piek JJ, Koch KT, et al. Quantification of collateral flow in humans: a comparison of angiographic, electrocardiographic and hemodynamic variables. J Am Coll Cardiol 1999;33:670-7. 32. Ghali WA, Knudtson ML. Overview of the Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease. On behalf of the APPROACH investigators. Can J Cardiol 2000;16:1225-30. 33. Norris CM, Ghali WA, Knudtson ML, Naylor CD, Saunders LD. Dealing with missing data in observational health care outcome analyses. J Clin Epidemiol 2000;53:377-83. 34. Faris PD, Ghali WA, Brant R, et al. Multiple imputation versus data enhancement for dealing with missing data in observational health care outcome analyses. J Clin Epidemiol 2002;55:184-91. 35. McGinn T, Wyer PC, Newman TB, et al. Tips for learners of evidencebased medicine: 3. Measures of observer variability (kappa statistic). CMAJ 2004;171:1369-73. 36. Graham MM, Faris PD, Ghali WA, et al. Validation of three myocardial jeopardy scores in a population-based cardiac catheterization cohort. Am Heart J 2001;142:254-61. 37. McAlister FA, Oreopoulos A, Norris CM, et al. Exploring the treatmentrisk paradox in coronary disease. Arch Intern Med 2007;167:1019-25.
20. Kyriakides ZS, Kremastinos DT, Michelakakis NA, et al. Coronary collateral circulation in coronary artery disease and systemic hypertension. Am J Cardiol 1991;67:687-90.
38. Sun GW, Shook TL, Kay GL. Inappropriate use of bivariable analysis to screen risk factors for use in multivariable analysis. J Clin Epidemiol 1996; 49:907-16.
21. Koerselman J, de Jaegere PP, Verhaar MC, Grobbee DE, van der Graaf Y. Coronary collateral circulation: the effects of smoking and alcohol. Atherosclerosis 2007;191:191-8.
39. Werner GS, Ferrari M, Heinke S, et al. Angiographic assessment of collateral connections in comparison with invasively determined collateral function in chronic coronary occlusions. Circulation 2003;107:1972-7.
22. Dincer I, Ongun A, Turhan S, et al. Association between the dosage and duration of statin treatment with coronary collateral development. Coron Artery Dis 2006;17:561-5.
40. Ghali WA, Quan H, Brant R, et al. Comparison of 2 methods for calculating adjusted survival curves from proportional hazards models. JAMA 2001;286:1494-7.
23. Koerselman J, de Jaegere PP, Verhaar MC, Grobbee DE, der Graaf Y. Cardiac ischemic score determines the presence of coronary collateral circulation. Cardiovasc Drugs Ther 2005;19:283-9.
41. Rentrop KP, Cohen M, Blanke H, Phillips RA. Changes in collateral channel filling immediately after controlled coronary artery occlusion by an angioplasty balloon in human subjects. J Am Coll Cardiol 1985;5:587-92.
588
Canadian Journal of Cardiology Volume 27 2011
42. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159-74.
adverse events in patients with diabetes mellitus. Eur Heart J 2003;24:1134-42.
43. Schaper W. Collateral circulation: past and present. Basic Res Cardiol 2009;104:5-21.
49. Caputo M, Anis RR, Rogers CA, et al. Coronary collateral circulation: effect on early and midterm outcomes after off-pump coronary artery bypass surgery. Ann Thorac Surg 2008;85:71-9.
44. Werner GS, Ferrari M, Betge S, et al. Collateral function in chronic total coronary occlusions is related to regional myocardial function and duration of occlusion. Circulation 2001;104:2784-90. 45. Koerselman J, de Jaegere PP, Verhaar MC, van der Graaf Y, Grobbee DE. High blood pressure is inversely related with the presence and extent of coronary collaterals. J Hum Hypertens 2005;19:809-17. 46. Pourati I, Kimmelstiel C, Rand W, Karas RH. Statin use is associated with enhanced collateralization of severely diseased coronary arteries. Am Heart J 2003;146:876-81. 47. Olijhoek JK, Koerselman J, de Jaegere PP, et al. Presence of the metabolic syndrome does not impair coronary collateral vessel formation in patients with documented coronary artery disease. Diabetes Care 2005;28:683-9. 48. Werner GS, Richartz BM, Heinke S, Ferrari M, Figulla HR. Impaired acute collateral recruitment as a possible mechanism for increased cardiac
50. Abbott JD, Choi EJ, Selzer F, Srinivas VS, Williams DO. Impact of coronary collaterals on outcome following percutaneous coronary intervention (from the National Heart, Lung, and Blood Institute Dynamic Registry). Am J Cardiol 2005;96:676-80. 51. Werner GS, Emig U, Mutschke O, et al. Regression of collateral function after recanalization of chronic total coronary occlusions: a serial assessment by intracoronary pressure and Doppler recordings. Circulation 2003;108:2877-82. 52. Werner GS, Surber R, Ferrari M, Fritzenwanger M, Figulla HR. The functional reserve of collaterals supplying long-term chronic total coronary occlusions in patients without prior myocardial infarction. Eur Heart J 2006;27:2406-12. 53. Werner GS, Fritzenwanger M, Prochnau D, et al. Determinants of coronary steal in chronic total coronary occlusions donor artery, collateral, and microvascular resistance. J Am Coll Cardiol 2006;48:51-8.