Detection of Obstructive Coronary Artery Disease Using Peak Systolic Global Longitudinal Strain Derived by Two-Dimensional Speckle-Tracking: A Systematic Review and Meta-Analysis


Global longitudinal strain (GLS) is well validated and has important applications in contemporary clinical practice. The aim of this analysis was to evaluate the accuracy of resting peak GLS in the diagnosis of obstructive coronary artery disease (CAD).


A systematic literature search was performed through July 2015 using four databases. Data were extracted independently by two authors and correlated before analyses. Using a random-effect model, the pooled sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, diagnostic odds ratio, and summary area under the curve for GLS were estimated with their respective 95% CIs.


Screening of 1,669 articles yielded 10 studies with 1,385 patients appropriate for inclusion in the analysis. The mean age and left ventricular ejection fraction were 59.9 years and 61.1%. On the whole, 54.9% and 20.9% of the patients had hypertension and diabetes, respectively. Overall, abnormal GLS detected moderate to severe CAD with a pooled sensitivity, specificity, positive likelihood ratio, and negative likelihood ratio of 74.4%, 72.1%, 2.9, and 0.35 respectively. The area under the curve and diagnostic odds ratio were 0.81 and 8.5. The mean values of GLS for those with and without CAD were −16.5% (95% CI, −15.8% to −17.3%) and −19.7% (95% CI, −18.8% to −20.7%), respectively. Subgroup analyses for patients with severe CAD and normal left ventricular ejection fractions yielded similar results.


Current evidence supports the use of GLS in the detection of moderate to severe obstructive CAD in symptomatic patients. GLS may complement existing diagnostic algorithms and act as an early adjunctive marker of cardiac ischemia.

Timely exclusion of coronary artery disease (CAD) in symptomatic patients is critical to their management. Current noninvasive risk stratification algorithms are labor intensive and not sufficiently accurate to prevent redundant invasive angiographies. Left ventricular (LV) longitudinal strain, derived using two-dimensional speckle-tracking echocardiography, has emerged as a noninvasive marker of both global and regional LV dysfunction in patients at risk for developing CAD. Global longitudinal strain (GLS) has been shown to be superior to conventional echocardiography in most cases, and small variations among vendors have been noted.

The reported diagnostic performance of GLS in the detection of CAD varies. The aim of this study was to determine whether resting peak GLS by speckle-tracking might improve the diagnosis of obstructive CAD in symptomatic patients with intermediate pretest probability and, if so, to estimate its effect size by performing a systematic review and meta-analysis of the diagnostic accuracy of GLS in the detection of CAD.


Search Strategy

A systematic literature search was performed independently by two coauthors (K.L. and S.H.) in conjunction with E.A.R., our research coordinator, trained and experienced in performing systematic reviews. The search was completed in July 2015 using terms that reflect the concept of two-dimensional echocardiographic strain imaging by speckle-tracking and its interaction with the presence of CAD. Search hedges created are listed in the Supplemental Material and were applied to Ovid MEDLINE, Embase, the Cochrane Register of Controlled Trials, and the Cochrane Database of Systematic Reviews, with no date restriction. Furthermore, the reference lists of relevant studies were manually searched for additional studies addressing this subject.

Eligibility Criteria

We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines when possible in performing our systematic review. Two coauthors (K.L. and S.H.) reviewed and chose the studies on the basis of the following inclusion criteria: (1) prospective or retrospective studies in which the presence of CAD as the primary outcome of interest was confirmed by anatomic modalities such as invasive or cardiac computed tomographic angiography, (2) studies in which GLS was derived from three apical views, (3) studies including ≥20 adults patients presenting to the hospital with clinical suspicion of CAD, (4) studies that reasonably fulfilled the prespecified criteria of quality assessment as outlined below, and (5) studies containing raw data for retrieving directly or permitting indirect derivation of outcome of interests and their respective 95% CIs. All publications were limited to those involving human subjects. Abstracts, case reports, review articles, editorials, and expert opinions were excluded. When multiple reports that used the same data set were identified, only the most complete was included for assessment.

Outcomes of Interest

The diagnostic accuracy of GLS for obstructive CAD was evaluated. We elected to use the pooled diagnostic odds ratio (DOR) as the primary end point. Other parameters analyzed included pooled sensitivity, specificities, positive likelihood ratio (PLR) and negative likelihood ratio (NLR), and the overall diagnostic accuracy as assessed by the area under the receiver operating characteristic (ROC) curve (AUC). The PLR and NLR were calculated as sensitivity/(1 − specificity) and (1 − sensitivity)/specificity, respectively. The DOR describes the odds of a positive test result in a patient with the disease compared with the odds of a patient’s having a positive test result without the disease. The DOR was calculated as PLR/NLR. Finally, we sought to determine the overall GLS values for patients with and without obstructive CAD.


For GLS, we used the definitions of the European Association of Cardiovascular Imaging, American Society of Echocardiography, and industry task force consensus document. Briefly, GLS was calculated from the segmental averaging of the three apical views (four, three, and two chamber). For the purpose of this study, GLS denotes peak global longitudinal strain values obtained from resting echocardiograms.

Of the studies included, four defined CAD as the presence of lesions ≥70% in the epicardial arteries or ≥50% in the left main coronary artery. The remaining studies defined CAD as the presence of ≥50% obstruction in the epicardial arteries. For the purpose of this review, patients with “moderate to severe” CAD are those with ≥50% stenosis in the coronary arteries, whereas patients with “severe” CAD are those with ≥70% stenosis in the epicardial arteries or ≥50% stenosis in the left main coronary artery only.

Data Extraction

All authors of eligible studies were contacted by e-mail to notify them of our intention to include their studies in our meta-analysis and to obtain missing information identified after quality evaluation. The data were extracted independently by two of the coauthors (K.L. and S.H.) and summarized into a standardized extraction sheet. Eligible studies were excluded if missing information could not be extracted from the report or the authors did not provide it. Further details can be found in the Supplemental Material under “Search Strategy.”

Methodologic Quality Evaluation

The quality of the retrieved citations was assessed by two coauthors (K.L. and E.A.R.) against the second Quality Assessment of Diagnostic Accuracy Studies checklist criteria as well as those set out by Downs and Black. Studies that violated >25% of the items on these respective checklist tools were excluded from analysis.

The study was prospectively registered with the PROSPERO database of systematic reviews (registration number CRD42014015498).

Statistical Analysis

Between-study statistical heterogeneity was assessed using the Cochran Q statistic and was quantified using the I 2 method by measuring inconsistency ( I 2 is the percentage of total variance across studies attributable to heterogeneity rather than chance). I 2 values of 25%, 50%, and 75% corresponded to low, moderate, and high degrees of heterogeneity, respectively. Publication bias was assessed on the basis of study distribution using a funnel plot, Egger’s regression, and Duval and Tweedie’s trim and fill methodology. Duval and Tweedie’s trim and fill methodology is a nonparametric method for estimating the number of missing studies that might exist in a meta-analysis and the effect that these studies might have had on its outcome, while the Egger test is a test for asymmetry of the funnel plot.

Because of the anticipated heterogeneous nature of the included studies, we decided a priori to analyze the data using a random-effect model. The likelihood ratio and DOR were pooled using the method of DerSimonian and Laird. The pooled AUCs generated were weighted derivations from the summary ROC curve. The summary ROC curve was derived using the Moses-Shapiro-Littenberg method, in which each study represented a point on the graph.

We planned to conduct subgroup analyses a priori to assess the diagnostic accuracy of GLS in patients with severe CAD, normal LV ejection fractions (LVEFs), and no histories of ischemic heart disease, myocardial infarction, or CAD. Sensitivity analyses were performed to examine the effect of methodologic diversity (number of segments used) on the overall pooled estimates. To account for interstudy heterogeneity, a meta-regression was performed to evaluate the impact of the imbalanced distributions of age, LVEF, strain cutoff values, versions of strain analysis software, and prevalence of hypertension and diabetes on the DOR. In brief, the mean age, LVEF, strain threshold, software versions, and prevalence of diabetes and hypertension from each study were used as moderators for the meta-regression analyses using the general multivariate linear model. The strain cutoff values were determined by the authors of each individual study and formed the basis on which the sensitivity and specificity of GLS were derived. Furthermore, the mean GLS values for those with and without CAD were derived, and contributors of the observed variability were explored with meta-regression. Finally, the clinical value of GLS in the diagnosis of CAD was illustrated with Fagan’s nomogram. Fagan’s nomogram is a graphical calculator that quantifies the posttest probability that an individual is affected by a condition given an observed test result and given the probability of the individual’s having the condition before the test was run (pretest probability).

Diagnostic accuracy meta-analysis and meta-regression were performed with Meta-DiSc version 1.4 (Ramón y Cajal Hospital, Madrid, Spain). Assessment of publication bias and overall mean GLS values was performed with Comprehensive Meta-Analysis version 2 (Biostat, Englewood, NJ). Values are expressed as pooled values with their respective 95% CIs. P values < .05 indicate statistical significance.


Study Selection

The systematic search yielded a total of 1,669 records from the four electronic databases and review of the study citations. After excluding duplicates and triplicates (371), 1,298 studies were screened for relevance. Review articles, abstracts without text, and articles unrelated to the topics (1,283) were then excluded. Fifteen published studies were subjected to detailed review of their methodology and the quality of their data, and five were excluded because of failure to fulfill the inclusion criteria ( Figure 1 ). Consequently, 10 peer-reviewed publications ( Tables 1 and 2 ) were included for quantitative analysis. The quality of the studies as assessed using the quality assessment tool of Downs and Black and the second Quality Assessment of Diagnostic Accuracy Studies criteria is outlined in Supplemental Tables S1 and S2 .

Figure 1

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flowchart of the study selection process. CCTR , Cochrane Register of Controlled Trials; CDSR , Cochrane Database of Systematic Reviews; QUDAS2 , second Quality Assessment of Diagnostic Accuracy Studies.

Table 1

Summary of included studies

Study Design Patients ( n ) Exclusion criteria Baseline troponin Definition by angiography CAD incidence, n (%) Assessors blinded to strain values
CAD positive CAD negative
Biering-Sorensen et al . (2014) Single center
296 with SA (293 included for analysis) Ischemic heart disease, congestive cardiac failure, valve disease, LVEF < 50%, intraventricular conduction disorder, q waves, arrhythmias Negative ≥70% <70% 107 (35.5) Yes
Chang et al . (2014) Single center
142 with SA Acute and prior MI, significant valvular heart disease (greater than moderate severity), moderate or severe pulmonary hypertension (>50 mm Hg), pulmonary thromboembolic disease, atrioventricular block, unstable hemodynamic status (systolic blood pressure < 90 mm Hg), and pericardial disease Negative ≥50% in pRCA <50% in pRCA 59 (41.5) Yes
Dahlslett et al . (2014) Single center
64 with suspected NSTEACS <18 y of age, QRS duration > 0.12 sec, severe valve disease, AF with rapid ventricular rate, other arrhythmia, known CAD, severe mental disorder, abnormal initial troponin, abnormal results on ECG Negative ≥50% <50% 29 (45.3) Yes
Montgomery et al . (2012) Single center
United States
123 with chest pain or clinical suspicion of CAD LVEF < 40%, severe wall motion abnormality, LV hypertrophy, premature ventricular ectopic beats, use of echocardiographic contrast, prosthetic aortic valve NR ≥50% <50% 56 (45.5) Yes
Smedsrud et al . (2012) Single center
86 with stable chest pain or increased risk profile following exercise ECG; 77% had objective signs of ischemia ACS, history of MI, scar by cardiac magnetic resonance imaging, severe valve disease, previous heart surgery, AF, bundle branch block with QRS > 120 msec Negative ≥50% <50% 43 (50) Yes
Shimoni et al . (2011) Single center
62 with SA LVEF < 50%, regional LV dysfunction, poor image quality, STEMI, history of MI, significant valve disease, perimyocarditis or cardiomyopathy Negative ≥50% <50% 69 (71.1) Yes
35 with suspected ACS Positive in 13 within 12 h
Eek et al . (2010) Single center
150 with suspected NSTEACS Prior MI, STEMI, bundle branch block with QRS > 120 msec, severe valvular disease, previous heart surgery, extensive comorbidity, AF with heart rate > 100 beats/min Positive in 124 patients Acute occlusive lesions (TIMI grade 0 or 1 flow) Nonocclusive lesions 33 (22) Yes
Nucifora et al . (2010) NR
The Netherlands, Italy
182 referred for multislice cardiac computed tomography for chest pain or increased risk profile LVEF < 50%, LV wall motion abnormality, history of CAD, cardiomyopathy, significant valve disease, congenital heart disease, arrhythmia Negative ≥50% <50% 63 (34.6) NR
Tsai et al . (2010) Single center
152 with chest pain or clinical suspicion of CAD LVEF < 50%, hypertrophic obstructive cardiomyopathy, significant valve disease, ACS, equivocal diagnosis of CAD NR ≥75% <75% 75 (49.3) NR
Choi et al . (2009) Single center
96 patients with SA Wall motion abnormality, LV impairment, valvular pathology, arrhythmia or bundle branch block Negative 50% LM; ≥70% epicardial vessels <50% LM; <70% epicardial vessels 66 (68.8) NR

Only gold members can continue reading. Log In or Register to continue

Apr 17, 2018 | Posted by in CARDIOLOGY | Comments Off on Detection of Obstructive Coronary Artery Disease Using Peak Systolic Global Longitudinal Strain Derived by Two-Dimensional Speckle-Tracking: A Systematic Review and Meta-Analysis

Full access? Get Clinical Tree

Get Clinical Tree app for offline access