Prevalence and outcome of patients with non-ST segment elevation myocardial infarction with occluded “culprit” artery – a systemic review and meta-analysis

Background The aim was to determine the prevalence and impact of an occluded “culprit” artery (OCA) in patients with non-ST segment elevation myocardial infarction (NSTEMI). Methods We searched PubMed, EMBASE, and Web of Science, with no language restrictions, up to 1 Jul. 2016. Observational cohorts or clinical trials of adult NSTEMI were eligible for inclusion to determine the prevalence if the proportion of OCA on coronary angiography was reported. Studies were further eligible for inclusion to determine the outcome if the association between OCA and clinical endpoints was reported. Results Among the 60,898 patients with NSTEMI enrolled in 25 studies, 17,212 were found to have OCA. The average proportion of OCA in NSTEMI was 34% (95% CI 30–37%). Patients with OCA were more likely to have left circumflex artery as their culprit artery (odds ratio (OR) 1.65, 95% CI 1.15–2.37, p = 0.007), and this was associated with lower left ventricular ejection fraction (standard mean difference -0.29, 95% CI -0.34 to -0.34, p < 0.001), higher peak enzyme level (standard mean difference 0.43, 95% CI 0.27–0.58, p < 0.001), and higher risk for cardiogenic shock (OR 1.66, 95% CI 1.35–2.04, p < 0.001), compared with patients with a non-occlusive culprit artery. Death rate (OR 1.72, 95% CI 1.49–1.98, p < 0.001) and recurrent myocardial infarction (OR 1.7, 95% CI 1.06–2.75, p = 0.029) were also higher in patients with OCA, compared with patients with a non-occlusive culprit artery. Conclusions Patients with OCA comprised a substantial portion of the NSTEMI population. These patients present with more severe symptoms and worse clinical outcome. Whether these patients should be treated with more aggressive strategy warrants further study. Electronic supplementary material The online version of this article (10.1186/s13054-018-1944-x) contains supplementary material, which is available to authorized users.


Background
Acute coronary syndrome (ACS) has been categorized into ST segment elevation myocardial infarction (STEMI) and non-ST segment elevation ACS (NSTEACS) based on the results of initial 12-lead electrocardiography (ECG). In patients with STEMI, early reperfusion therapy of the culprit artery is a class I indication in current guidelines [1]. In contrast, NSTEACS represent a wide spectrum of clinical syndromes, ranging from unstable angina (without cardiomyocyte loss) to NSTEMI (with cardiomyocyte necrosis). The management of NSTEACS is guided by risk stratification, with an early invasive strategy favoured in high-risk patients [2], especially for patients with positive cardiac necrosis biomarkers [3].
STEMI results from acute total or nearly total occlusion of a coronary artery [4]. However, totally occluded culprit artery (OCA) has also been observed among patients with NSTEACS. According to recent large retrospective studies, up to 30% of patients with NSTEACS had OCA [5,6]. One of the reasons for OCA to present as NSTEACS, instead of STEMI, is that ECG is not sensitive enough to detect acute ischaemia or infarction over posterior or lateral walls, when the left circumflex artery (LCx) is usually the culprit artery. Indeed, the inferolateral territory was more frequently involved among patients with NSTEACS with OCA, compared with those with a non-occlusive culprit artery [6].
The clinical implication of OCA in NSTE-acute coronary syndrome (ACS) is still controversial, and how patients with NSTEACS with OCA should be treated is also unknown. Some studies report worse outcome for patients with OCA [5,[7][8][9], while other studies report otherwise [6,[10][11][12]. As unstable angina without positive cardiac markers was also included in some of these studies [6,9], the observations on outcomes were inevitably variable. Given the high incidence of NSTEACS worldwide [13], and the potential impact of OCA on outcome, it is necessary to analyse in detail the currently available data. To decrease ambiguity, we limited our analysis to patients with positive cardiac necrosis biomarkers, i.e. NSTEMI. The goals of this meta-analysis were (1) to determine the proportion of patients with OCA in NSTEMI and (2) to determine the impact of OCA on the clinical severity and outcomes including death and recurrent myocardial infarction (MI) following NSTEMI.

Data sources and search strategies
This systemic review and meta-analysis was performed according to the preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines [14]. A literature search was performed using PubMed, EMBASE, and Web of Science without restriction on language and year of publication. The following search terms were used in PubMed: (ACS or acute coronary syndrome[MeSH Terms] or (acute AND coronary AND syndrome) or "acute coronary syndrome" or NON-ST or NSTEMI or N-STEMI or NON-STEMI or NON STEMI or NSTEACS or angina, unstable[MeSH Terms]) AND (("total occlusion" or "totally occluded" or (occluded AND culprit)) or "TIMI flow" or "TIMI-flow" or "(TIMI) flow"). Similar expressions were used in EMBASE and Web of Science.

Eligibility criteria Study design
Studies included in the analysis were prospective or retrospective cohorts or randomized controlled trials. The exclusion criteria were (1) abstracts, review articles, case-control studies and case series with the number of patients fewer than 20; (2) studies on myocardial infarction that only reported Q wave or non-Q wave MI; (3) studies on patients with unstable angina, stable coronary artery disease (CAD), or STEMI; (4) studies selecting patients by angiography results (for example, limited to single-vessel disease, left main disease, or chronic total occlusion); (5) studies that did not specify the result of cardiac necrosis biomarkers; (6) animal studies; and (7) duplicate reports.

Patients
The inclusion criteria were (1) adults aged 18 or above; (2) patients with chest pain; and (3) patients with elevated cardiac necrosis biomarkers (troponin or creatine kinase-MB isoenzyme (CK-MB)) according to the criteria specified in individual studies. Patients with STEMI or patients with NSTEACS and negative cardiac necrosis markers were excluded from analyses.

Proportion of totally occluded culprit artery
The determination of the culprit artery and the definition of OCA were based on the protocol of the individual study. Sensitivity analysis will be performed to include only studies defining OCA as Thrombolysis In Myocardial Infarction (TIMI)-flow grade 0 or 1 in the culprit artery (the most commonly used definition in these studies).

Outcome
Outcomes were assessed as all-cause mortality and recurrent MI. Endpoints were taken at 1 year or the longest follow up available (if the follow-up period was less than 1 year) in each study.

Data extraction
Extraction of data on study characteristics and outcomes was performed independently by two reviewers (CSH and YHC). Discrepancies between reviewers were resolved by consensus. The following data were extracted from each study enrolled in our analysis: the first author's last name, year of publication, country of study performed, biomarkers measured, total number of patients, numbers of patients with OCA, mean age, gender, conventional risk factors for CAD (diabetes mellitus, hypertension, hyperlipidaemia, smoking) and outcome (recurrent MI or death). The study quality was judged using the Newcast-Ottawa scale (NOS) for observational studies [15]. This scale assesses the quality of the study based on patient selection, comparability and outcome. Studies controlled for age and sex were given one star for comparability; studies controlled for conventional risk factors for CAD were given another star for comparability. Studies with follow up of 1 year or longer were considered as long enough for outcome to occur. Studies with an NOS scale > =7 were considered of high quality.

Data analysis
We use the random-effects method because of its conservative summary estimate and because it incorporates between-study and within-study variance. We used the I 2 statistic to assess heterogeneity of the event rates across studies [16]. A funnel plot with a linear regression approach to measure asymmetry was used to assess for the presence of publication bias [17]. Subgroup analyses were performed in different geographical regions (Asia-Pacific, European, North American, and multinational). Sensitivity analyses using the leave-one-out method were performed to identify key studies with major influence on betweenstudy heterogeneity. Meta-regression was used to assess the influence of study characteristics on outcome. All statistical analyses were performed using Stata Version 13 software (Stata Corporation, Collage Station, TX, USA). All p values were two-sided, and those below 0.05 were regarded as statistically significant.

Results of the search
The initial search identified 1432 citations. After critical assessment of these papers, 25 studies fulfilled the inclusion criteria and were used to estimate the pooled proportion of OCA among NSTEMI patients. Ten of these 25 papers fulfilled the criteria and were used to analyse the impact of OCA on outcome (Fig. 1).

Study characteristics
Characteristics of the 25 included studies are presented in Table 1. The studies were conducted in North American (n = 5), European (n = 8), Asia-Pacific (n = 10), or multinational regions (n = 2). The number of patients varied from 42 to 30,386, with mean age from 58 to 69 years, and 60 to 78.6% were male. Presence of hypertension was reported in 35 to 85.5% of patients, diabetes in 11.5 to 53.9%, hyperlipidaemia in 13 to 69.2%, and smoking in 21.3 to 80.3% of patients. Coronary artery bypass grafting (CABG) was excluded in five studies [5,8,[18][19][20]. The definitions of OCA and exclusion criteria of the studies are summarised in Table 1. Among these 25 studies, 10 [5,[7][8][9][10][11][12][21][22][23] reported distribution of the culprit artery, presence of cardiogenic shock, peak creatine kinase (CK) level, and outcome by the status of the culprit artery. These 10 studies were used for the analysis of the impact of OCA on presentation, death, and recurrent MI. The details of NOS for these 10 studies are presented in Table 2. The total scores of NOS for the other 15 studies (not included in outcome analysis) are presented in Fig. 2.

Prevalence of OCA in NSTEMI
All 25 studies provided data to be used for analysis of the proportion of OCA in patients with NSTEMI. A total of 17,212 patients had OCA among the 60,898 patients included. The average proportion of OCA in NSTEMI was therefore 0.34 (95% confidence interval (CI) 0.30-0.38) (Fig. 2). However, there was evidence of substantial heterogeneity in the proportion rate (X 2 = 1644.9, degrees of freedom (df) = 24, p < 0.001, I 2 = 98.54%). After excluding studies with OCA definition other than TIMI flow of 0 or 1, the average proportion of OCA in the remaining 20 studies was 0.35 (95% CI 0.30-0.41; X 2 = 1067.58, df = 19, I 2 = 98.22%, p < 0.001) (Additional file 1: Figure S1). After further excluding studies with time to diagnostic angiography more than 1 week or unknown, the average proportion of OCA in the remaining 14 studies was 0.35 (95% CI 0.28-0.41; X 2 = 797.04, df = 13, p < 0.001, I 2 = 98.37%) (Additional file 1: Figure S2).
The median proportion of patients with multivessel disease in all 25 studies was 57.6% (interquartile range 45.9-62.1%). Two studies reported a substantially lower proportion of patients with multivessel disease compared with other studies (13.6% in Pride 2010 [9] and 11.7% in Daly 2012 [23]). After excluding these two studies with lower proportions of patients with multivessel disease, the average proportion of OCA in the remaining 23 studies was 0.33 (95% CI 0.29-0.36; X 2 = 1245.98, df = 22, p < 0.001, I 2 = 98.23%) (Additional file 1: Figure S3). Analysis by different locations showed possible heterogeneity among studies (Additional file 1: Figure S4 Figure S5). The leave-one-out sensitivity analyses could not identify studies with major impact on the between-study heterogeneity. The univariate meta-regression with the covariate of NOS, year of publication, age, proportion of patients with diabetes mellitus, hypertension, hyperlipidaemia and single-vessel disease, showed no significant impact on between-study heterogeneity (all p > 0.05).
Among the 10 studies used for analysis of outcome, the information about revascularization was reported in 6 studies [5,7,8,11,21,22] (Additional file 1: Table S1). In the three studies that reported the rate of successful percutaneous coronary intervention (PCI), the pooled OR for PCI success was  Figure S11) for patients with OCA, compared with patients with a nonocclusive culprit artery. In the three studies that reported stent length, the standard mean difference in stent length was 0.16 mm (95% CI 0.10-0.21, p < 0.001; X 2 = 9.63, df = 2, p = 0.008) (Additional file 1: Figure S12) longer in patients with OCA, compared with patients with a non-occlusive culprit artery. In the three studies that reported stent length and the use of a drug-eluting stent, there was no significant difference between patients with OCA and patients with a non-occlusive culprit artery (Additional file 1: Figure S13 and S14).
Risk stratification scores had been reported in four studies, with TIMI score reported in three [7,10,21] and GRACE score reported in one study [11]. There was no significant difference between patients with and without OCA using these two risk scores.

Test for publication bias
The funnel plots for studies reporting the proportion of OCA (by sample size vs. proportion) were asymmetrical (Fig. 5a). The Egger test confirmed the presence of small-study effects (p < 0.001). The small-study effect persisted even when excluding studies with a less strict OCA flow definition and studies with time to angiography more than 1 week (Egger test p < 0.001). The funnel plot for studies reporting on outcome suggested publication bias (Egger p < 0.001) (Fig. 5b).

Discussion
This meta-analysis showed that (1) the overall estimated proportion of OCA in NSTEMI was 34%; (2) NSTEMI with OCA was associated with more LCx as the culprit artery, higher peak enzyme level, lower LVEF, and more cardiogenic shock; and (3) NSTEMI with OCA was associated with a higher rate of recurrent MI and death. These important findings remained the same if we limited our analysis to studies using TIMI flow 0-1 as the definition of OCA and timely diagnostic angiography after onset.
There is substantial heterogeneity in our analysis, with a wide range of proportion of OCA reported in the studies included (0.15-0.75). The explanations for this variation include differences in the study geographic region, study design and OCA definition, patient population, and timing of angiography. The standard initial ECG evaluation protocol is often unavailable in retrospective studies, and subtle ST change, especially in the posterior leads, may be missed. Some of the studies excluded patients with prior coronary artery bypass grafting (CABG) [5,8,[18][19][20]. This difference in enrolment criteria may also influence the proportion of patients with OCA. The angiographic finding may change with time after symptom onset [24], but the estimated OCA proportion and outcome did not change after excluding studies with angiography performed more than 7 days after onset. Meta-regression did not identify the conventional risk factors as the source of the heterogeneity in the pooled estimation of OCA proportion. We did find different proportions among studies in different geographic regions. Ethnic difference has been reported in the prevalence of coronary artery disease [25,26] and in the distribution of coronary lesions [27]. However, whether the difference between geographic regions reflected a real difference in the ethnicity or a combination of different design, inclusion criteria, and timing of angiography, is not clear.
Our analysis suggested that OCA represents a substantial portion of NSTEMI, and carries a worse clinical outcome. Some researchers proposed that this condition should be considered as "STEMI equivalent" and treated as such [6,7]. However, whether OCA mandates a more aggressive OCA occluded culprit artery, OR odds ratio treatment strategy or simply reflects worse disease severity cannot be determined from the current analysis. There are several possible causes of OCA in NSTEMI: (1) acute total occlusion of vessel suppling some part of myocardium (esp lateral wall) which does not consistently lead to ST elevation in conventional 12-lead ECG, probably due to the absence of corresponding leads; (2) acute total occlusion of vessels with good collaterals; (3) acute total occlusion in the territory with dual blood supply [28]; (4) acute total occlusion with a small infarction area; or (5) chronic total occlusion misclassified as acute occlusion. In the first condition, timely reperfusion by primary percutaneous coronary intervention may improve the outcome as in STEMI. Although our analysis cannot provide evidence for any of the aforementioned hypotheses, the higher percentage of LCx as the culprit artery suggests that acute total occlusion not detected by standard 12-lead ECG may be the most possible cause. Acute LCx occlusion can result in isolated posterior infarction with ST elevation only detected in leads V7-V9 [29]. The isolated posterior infarction due to LCx occlusion is not uncommonly missed by medical personnel [30], and is associated with longer time to reperfusion [31] and is less likely to be treated by primary percutaneous coronary intervention [32]. These factors are possibly associated with less favourable outcome, as in our analyses. Collateral artery status was described in only one study, reporting higher incidence of angiographic collaterals in NSTEMI with OCA [11]. The authors also identified better outcome in these patients with OCA with collaterals, compared with those without. Further study on the impact of collateral in patients with NSTEMI with OCA is needed. Finally, determination of the culprit artery is sometimes difficult in NSTEMI without diagnostic ECG. Hence, misclassifying a chronic total occlusion as the culprit is possible clinically, and the true incidence of misclassified chronic total occlusion cannot be determined by our analysis.
Considering the large number of patients with NSTEMI worldwide [13], it is imperative to explore methods to identify patients with NSTEMI with OCA. Three studies included in the present metaanalysis reported the TIMI or GRACE risk score, but these scores cannot differentiate the outcomes of patients with OCA or those with a non-occlusive culprit artery. Body surface potential mapping using 80 chest leads may improve detection of ST segment elevation [23], but its clinical practicality needs further investigation. The LVEF assessed by echocardiography was significantly lower in patients with NSTEMI with OCA than in patients with a non-occlusive culprit artery according to our analysis. Whether LVEF helps to identify patients with NSTEMI with OCA should be validated prospectively. Other practices incorporating coronary computed tomography angiography, echocardiography, or routine posterior-leads ECG may all be helpful, but again mandating future research to establish their clinical significance.
Our analysis has several strengths. First, we enrolled only patients with elevated cardiac necrosis markers. Focusing our biomarker-positive NSTEACS population helps us to decrease the ambiguity, targeting only patients with objective myocardial injury. More importantly, patients with biomarker-positive NSTEACS may benefit from an early invasive strategy, compared with biomarker-negative patients [3]. Second, the potential influence of OCA definition and angiography timing were well-considered and controlled. By conducting sensitivity analyses, we showed that there was no major impact of these two factors on the heterogeneity of OCA incidence.
Our analysis does have several limitations. First, the ECG protocol used was not reported in detail in most of the studies. Whether posterior leads have been applied in individual studies was unclear. Second, patients with contraindication for angiography or percutaneous coronary intervention were excluded by the analysis design. Therefore our results cannot be generalized in this population. Third, angiography in the included studies was done within one week after symptom onset. Hence the true OCA incidence at the time of onset could not be deducted from the present analysis. Fourth, a certain proportion of patients with posterior STEMI may be classified as having NSTEMI. This misclassification may be improved in the future by including routine posterior leads to detect posterior STEMI in patients with chest pain. Fifth, because the outcomes reported in the studies had not been stratified by some important moderators including the procedures and the number of diseased vessels, further analysis on the impact of OCA according to these moderators cannot be performed. Finally, given the asymmetry of funnel plot, we cannot exclude the potential publication bias against studies finding no significant difference in the outcomes of NSTEMI with or without OCA.

Conclusion
Approximately one third of patients with NSTEMI have OCA. These patients present with more severe clinical symptoms and have a worse outcome, compared with those with a non-occlusive culprit artery. There is as yet no reliable tool to identify this group of patients before performing angiography. Whether timely reperfusion will benefit this group of patients warrants further studies.