Wall thickening assessment with tissue harmonic echocardiography results in improved risk stratification for patients with non-ST-segment elevation acute chest pain

Abstract

Objectives: To demonstrate whether the improved imaging quality gained by using tissue harmonic echocardiography in place of fundamental echocardiography results in the improved risk stratification of patients presenting with non-ST-elevation acute chest pain.

Methods and results: Eighty patients with over 30 min of non-ST-elevation chest pain that had lasted less than 6 h were recruited. All patients underwent resting tissue harmonic and fundamental echocardiographic scans. Diagnosis for acute myocardial infarction was made on a 24 h creatine kinase-MB sample. Echocardiographic images were reported by two experienced blinded observers. Patients were followed up at least 4 months after admission. Endpoints included all-cause mortality, non-fatal myocardial infarction and revascularisation procedures. Tissue harmonic echocardiography allowed assessment of all myocardial segments in all patients compared to 43/78 patients ( p <0.001) with fundamental echocardiography. A wall thickening abnormality demonstrated on tissue harmonic echocardiography and not fundamental echocardiography was a significant predictor of index myocardial infarction on admission ( p <0.007) and for an adverse cardiac event during follow up ( p = 0.002).

Conclusions: Tissue harmonic echocardiography is superior to fundamental echocardiography for accurate assessment of systolic wall thickening and hence risk stratification for patients presenting with acute chest pain and non-diagnostic electrocardiogram changes.

acute coronary syndrome, risk stratification, tissue harmonic echocardiography

1 Introduction

Diagnosis of acute myocardial infarction is simple in the presence of classical symptoms and diagnostic electrocardiogram changes. However, less than a third of patients presenting with chest pain have an acute coronary syndrome and only one-third of those will have electrocardiogram changes diagnostic of acute myocardial infarction. ^1,2 Current protocols for risk stratification rely upon clinical parameters and are targeted to identify those patients that are perceived as being at high risk. Cardiac biomarkers assist in formulating a diagnosis, however, the fundamental disadvantage of their usage is that an abnormal result depends upon myocardial necrosis, not ischaemia. There is a delay of up to several hours before a significant raise can be detected. ^3,4

Fundamental 2D echocardiography has been shown to be a useful tool to identify regional wall motion abnormalities during acute coronary syndrome ⁵ and superior to conventional methods for the detection of acute myocardial infarction in the emergency setting. ^6,7 However, the major limitation of fundamental echocardiography is that it is often unable to visualise all ventricular segments, this affects its accuracy in assessing systolic wall thickening. Furthermore, compromised image quality may also result in missing subtle systolic wall thickening abnormalities in patients with small sub-endocardial infarcts.

Tissue harmonic echocardiography uses a transducer that only receives harmonics of the emitted frequency. Tissue harmonic echocardiography reduces artefact and improves endocardial border delineation in comparison to fundamental imaging ⁸ and is therefore likely to improve systolic wall thickening assessment.

The aim of this prospective study was to demonstrate that the superior imaging quality of tissue harmonic echocardiography results in improved risk assessment of patients presenting with non-ST-elevation acute chest pain.

2 Methods

2.1 Patient selection

Eighty consenting patients who presented to our A&E department with suspected cardiac chest pain that had lasted for greater than 30 min within the preceding 6 h were recruited. Patients with persistent ST-elevation in two or more leads were excluded, however, those with transient ST-elevation (ST-elevation that had spontaneously returned to baseline prior to planned thrombolysis) were included.

2.2 Study design

All patients were assessed clinically and underwent an electrocardiogram followed by echocardiography. Patients were managed on the basis of their clinical and electrocardiogram findings in line with our usual hospital protocols. An index myocardial infarction was diagnosed on the basis of a CK-MB sample reaching twice the reference limit 24 h after the onset of pain. Patients were followed up by means of a letter, telephone call and notes review for subsequent adverse cardiac events (non-fatal myocardial infarction, revascularisation and cardiac death). Cardiac death was defined as death due to known cardiac cause or sudden unexplained death in the community.

2.3 Electrocardiogram

The electrocardiogram was recorded as abnormal if there was ST-segment elevation, ST-segment depression, significant T-wave inversion, pathological Q-waves or bundle-branch block.

2.4 Echocardiography

Echocardiography was performed by experienced operators using a broad band transducer (transmitting frequency 2–3 MHz) on commercially available equipment (HDI 3000, ATL, Bothell, Washington, USA). Images were recorded in all five standard views (parasternal long and short axis, apical four, two and three chamber views) in the fundamental mode and then the tissue harmonic mode. Gain and depth were adjusted to optimise image quality. Images were recorded onto Super VHS videotape for subsequent analysis. Images were reported by two experienced blinded observers using a 14 segment model. The fundamental and tissue harmonic scans were blindly reported separately. Each segment was scored according to clarity of visualisation: 0—not seen, 1—poorly visualised, 2—clearly seen. Regional wall thickening was assessed as follows: 0—not interpretable, 1—normal, 2—reduced, 3—absent. A study was only defined as normal if all 14 segments could be visualised and were thickening normally.

2.5 Statistical analysis

Categorical variables are shown as proportions and continuous variables as means (standard deviation). A series of two-by-two contingency tables for each test category and outcome variable were constructed and the sensitivity, specificity, positive and negative predictive values for index AMI and subsequent cardiac event were determined. Univariate logistic regression analysis was performed for each categorical and continuous variable to predict index myocardial infarction on admission and subsequent cardiac events. Multiple logistic regression analysis was then performed including all variables with a p value <0.05 using forwards elimination. The same process was then repeated using Cox regression analysis for time-dependent endpoints. A p value of <0.05 was considered significant. Kaplan–Meier curves were also constructed for independent predictors. All statistical analyses were performed using SPSS 9.0.

3 Results

3.1 Patient demographics

The clinical characteristics of the 80 patients recruited into the study are shown in Table 1 . The mean time from the end of pain to echocardiography was 170 ± 190. Eighteen patients still had pain at the time of imaging. The electrocardiogram was classed as abnormal in 55 (69%) of patients on inclusion. Thirty-six (45%) had T-wave abnormalities, 10 (5-2-142-segment changes, seven (9%) had conduction abnormalities and two (3%) had Q waves. The remaining electrocardiograms were classed as being normal.

Table 1

Patient demographics

Clinical factors
Age	64 (±13.5) years
Male	62 (78%)
Previous cardiac history
Myocardial infarction	31 (39%)
CABG	9 (11%)
PTCA	7 (9%)
Risk factors for coronary disease
Hypertension	33 (41%)
Diabetes mellitus	20 (25%)
Hyperlipidaemia	19 (24%)
Current or ex-smoker	25 (66%)

CABG, coronary artery bypass grafting; PTCA, percutaneous transluminal coronary angioplasty.

3.2 Echocardiography

Of the 78 patients who had fundamental echocardiograms available, all myocardial segments were visualised, either poorly or well in 43 (55%) patients. Good segmental visualisation was seen in 736 (68%) segments, 231 (21%) were partially visualised and 125 (11%) segments were not visualised at all. Forty-seven (60%) scans were reported as abnormal, either because there was a demonstrated wall thickening abnormality (34 patients), or because not all segments could be adequately seen (13 patients).

Of the 80 patients who underwent tissue harmonic echocardiography, all segments were adequately visualised in all patients ( p <0.001). Good segmental visualisation was seen in 1104 (99%) segments and 16 (1%) segments were only partially visualised. Forty-seven (59%) of patients had wall-thickening abnormalities ( Fig. 1 ). Despite the differences in full endocardial visualisation, both tissue harmonic echocardiography and fundamental echocardiography interpreted 47 scans as being abnormal. However, the correlation between the two techniques was only 66% ( κ = 0.30, 95% CI 0.09–0.52).

Figure 1

Bar graph comparing quality of segmental visualisation with echocardiography using fundamental (FE) and tissue harmonics (THE) imaging.

3.3 Prediction of index myocardial infarction

Of the 13 patients who presented with acute myocardial infarction, 10 had wall thickening abnormalities on fundamental echocardiography and 12 on tissue harmonic echocardiography. The sensitivity, specificity, positive and negative predictive values of the two techniques for prediction of an index myocardial infarction are summarised in Table 2 . Regression analysis shows that an abnormal tissue harmonic scan is a significant predictor of index myocardial infarction ( χ² =7.2, p = 0.007). An abnormal fundamental scan was not a significant predictor of index myocardial infarction ( χ² =1.8, p >0.1). An abnormal electrocardiogram on presentation was also a significant predictor of index myocardial infarction ( χ² =5.4, p = 0.02) ( Table 2 ).

Table 2

Accuracy of clinical, ECG and echocardiographic variables for prediction of myocardial infarction as the presenting event

	Specificity (%)	Sensitivity (%)	Positive predictive value (%)	Negative predictive value (%)	Univariate ( χ2 )	Significance ( p )
Previous MI	58	23	10	80	1.6	>0.1
Hypertension	57	31	12	81	0.0	>0.1
Smoker	48	77	23	91	2.8	>0.1
Diabetes	72	8	5	80	2.5	0.09
Abnormal ECG	42	92	24	97	5.4	>0.1
ST-deviation	88	15	30	84	0.1	0.02
T-wave inversion	64	62	25	90	3.0	>0.1
Abnormal FE	43	77	21	90	1.8	>0.1
Abnormal THE	48	92	26	97	7.2	0.007

ECG, electrocardiogram; MI, myocardial infarction; FE, fundamental echocardiogram; THE, tissue harmonic echocardiogram.

3.4 Prediction of further adverse cardiac events

Follow up lasted a mean of 224 (89) days, during which time there were 33 adverse cardiac events (seven cardiac deaths, four non-fatal myocardial infarctions, nine patients underwent percutaneous transluminal coronary angioplasty, and 13 coronary artery bypass grafting) affecting 29 (36%) patients. The sensitivity, specificity, positive and negative predictive values for prediction of further adverse cardiac events are shown in Table 3 . Regression analysis showed that an abnormal tissue harmonic scan but not an abnormal fundamental scan or an abnormal electrocardiogram, was a predictor for a cardiac event during follow up ( χ² =10.0, p = 0.002). The only multivariate predictor of an adverse cardiac event was an abnormal tissue harmonic scan (RR 4.3; 95% CI 1.6–11.5, p = 0.002). A Kaplan–Meier plot shows significant separation (log rank test p = 0.002) in event free survival between those with and without wall thickening abnormalities on tissue harmonic echocardiography ( Fig. 2 ).

Figure 2

Kaplan–Meier plot showing separation of event free survival by tissue harmonic echocardiography.

Table 3

Accuracy of clinical, ECG and echocardiographic variables for the prediction of any cardiac event during follow up

	Specificity (%)	Sensitivity (%)	Positive predictive value (%)	Negative predictive value (%)	Univariate ( χ2 )	Significance ( p )
Previous MI	71	55	52	73	3.9	<0.05
Hypertension	59	41	36	64	0.0	>0.1
Smoker	44	55	36	63	0.0	>0.1
Diabetes	75	24	35	63	0.1	>0.1
Abnormal ECG	41	72	41	72	1.6	>0.1
ST-deviation	88	14	40	64	0.0	>0.1
T-wave inversion	59	38	64	63	0.1	>0.1
Abnormal FE	46	71	43	74	2.1	>0.1
Abnormal THE	55	83	51	85	10.0	0.002

ECG, electrocardiogram; MI, myocardial infarction; FE, fundamental echocardiogram; THE, tissue harmonic echocardiogram.

3.5 Incremental value of tissue harmonic echocardiography

Tissue harmonic echocardiography provided incremental information over and above clinical history and electrocardiogram, for the prediction of subsequent cardiac events ( Fig. 3 ).

Figure 3

Bar graph showing incremental value of clinical, electrocardiographic (ECG) and echocardiographic (THE) variables for the prediction index acute myocardial infarction or a subsequent cardiac event.

4 Discussion

We undertook an evaluation of tissue harmonic echocardiography in patients presenting with chest pain and non-diagnostic electrocardiogram changes. The use of echocardiography to assess regional wall motion relies on adequate visualisation of the endocardial border. We demonstrated that tissue harmonic echocardiography visualised all segments in all patients adequately, this was significantly better than fundamental echocardiography. Tissue harmonics improve endocardial visualisation for a number of reasons. Fundamental imaging, particularly of the apex, is often poor due to near field haze. Artefact is created by the reflection of ultrasound off components of the chest wall. ⁹ However, this sound never reaches a sufficient amplitude such that it creates second harmonic energy. It is only after the transmitted signal penetrates beyond the noise producing organs that second harmonics are produced. With fundamental echocardiography artefacts can also be produced from beams that emanate from the sides of the transducer. This is termed side lobe artefact. Second harmonic beams have a narrower main lobe and lower side lobes than fundamental beams, again resulting in less artefact production on echo.

4.1 Improved endocardial definition resulting in improved risk stratification

4.1.1 Diagnosis of AMI

This is the first study to compare fundamental and tissue harmonic echocardiography in the patients presenting with non-ST-elevation acute chest pain. The prediction of an index myocardial infarction using tissue harmonic echocardiography was superior to both fundamental echocardiography and electrocardiogram. However, even with full endocardial visualisation the sensitivity was not 100% because small sub-endocardial infarctions may not give rise to wall motion abnormalities. ¹⁰ In this study the infarct that was missed was small with a creatine kinase rise of 524.

There have been a few published studies that have investigated the efficacy of echocardiography for diagnosis of acute myocardial infarction, however, none have used tissue harmonic echocardiography. In previous studies using fundamental echocardiography, acute myocardial infarction has been predicted with a negative predictive value ranging from 94 to 98% and a positive predictive value from 31 to 50% ^7,11–13 which is similar to our findings with fundamental imaging. However, this minor degree of variability may be due to the differences in selection criteria used between the studies as well as the protocols followed. In the studies undertaken by Kontos et al. ¹³ and Peels et al. ¹² all patients with evidence of previous myocardial infarction were excluded, this significantly improved their specificity. In our study, patients were not pre-screened with echocardiography and even those with a preceding myocardial infarction (39%) were included. This relatively high proportion of patients with previous myocardial infarction was similar to the group of patients studied by Sabia et al. ⁷ However, in their study patients with persistent ST-elevation were included.

4.1.2 Tissue harmonics for prediction of a future cardiac event

Prediction of future cardiac events with echocardiography relies not only on the accurate diagnosis of acute myocardial infarction but also of acute ischaemia. Tennant et al. ¹⁴ demonstrated 65 years ago that coronary occlusion causes wall motion abnormalities in the ischaemic bed within seconds. These wall motion abnormalities may persist for as long as 24 h, but only after prolonged ischaemia. ¹⁵ Thus, to maximise the sensitivity for the diagnosis of ischaemia it is essential to perform the echocardiogram early after the onset of pain. In this study the median time from onset of pain to imaging was 6 h and only 18 (23%) of patients had ongoing chest pain at the time of imaging. It is possible that tissue harmonic echocardiography, with its superior endocardial visualisation, may still detect minor and persistent small wall motion abnormalities. Tissue harmonic echocardiography demonstrated a sensitivity of 83% and a negative predictive value of 85% for the prediction of a subsequent cardiac event.

Previous studies using fundamental echocardiography have shown added incremental early and late prognostic information when performed in the acute setting. ^7,13,16 In this study echocardiography performed using tissue harmonics, despite the delay from the onset of chest pain gave significant incremental additional prognostic information over and above that was gained from the clinical history and electrocardiogram findings. This was not seen when fundamental imaging was used.

5 Limitations of this study

Patients were recruited between 9 a.m. and 5 p.m. for the sake of convenience as a result they were not recruited consecutively. The sensitivity of the study would have been improved if more patients had undergone echocardiography during or immediately after, rather than several hours after their episode of pain. This may also account for the disparity seen when comparing our results using fundamental imaging to those from previous studies. Although analysis of the echocardiographic studies was blinded, performed in a random order, and each patient's two studies interpreted separately, an experienced observer may be able to differentiate between fundamental and tissue harmonic images. This could leave the study open to a degree of bias.

6 Conclusion

Our study has shown that the use of tissue harmonic echocardiography results in full endocardial visualisation significantly more frequently than fundamental echocardiography. As a result of this tissue harmonic echocardiography is superior to fundamental echocardiography for the risk assessment of patients presenting with acute chest pain and non-diagnostic electrocardiogram changes and provides incremental prognostic information to the electrocardiogram and clinical history. We suggest that when an acute echocardiographic study is going to be performed, tissue harmonics should be used. This will maximise diagnostic yield and give significantly improved prognostic information to aid in patient management.