Myocardial viability testing and impact of revascularization on prognosis in patients with coronary artery disease and left ventricular dysfunction: a meta-analysis

Literature search

A MEDLINE database search for literature published in English since 1966 was performed in August 1999, using PubMed, (National Library of Medicine, National Institutes of Health, Bethesda, Maryland 20894) and BioMedNet (Evaluated Medline). The search algorithm was: “viability, heart, outcome.”

Exclusions

Twenty-eight citations were returned: (5,6,8–33)and the manuscripts scrutinized. Those not reporting deaths or where deaths could not be apportioned to patients with versus without viability were excluded (6,28,29). In cases of apparent serial reporting of a patient cohort only the most recent was included (5).

Dataset entered into the analysis

The remaining 24 papers are summarized in Table 1. In two studies reporting results using multiple imaging techniques, data from only one technique are included to avoid duplicate entering of events: Pasquet et al. (31)(scintigraphy/echocardiography where scintigraphic data are included) and Tamaki et al. (9)(thallium/F-18 fluorodeoxyglucose [FDG] with positron emission tomography [PET] where PET data are included). However, all data were used for comparison between testing modalities.

Meta-analysis

Pooled, averaged rates of cardiac death plus patient age, gender and left ventricular ejection fraction (LVEF) were extracted from each report. Numbers of patients with and without demonstrated viability (according to individual studies’ author-defined criteria) were extracted (Table 1). These two groups were subdivided into patients subsequently revascularized and those treated medically. Annual mortality rates for each of the resulting four subgroups were calculated as well as average follow-up time (months) and follow-up completeness.

A meta-analysis was performed using a random effects model (34)to compare mortality rates in patients with/without viability treated by either revascularization/medical therapy. This model calculates a weighted-average percent decrease in mortality rates with 95% confidence intervals. A chi-square test for homogeneity was calculated, and Fisher exact test was used for comparing event rates (p < 0.05 considered significant). For the overall meta-analysis, three papers were considered outliers (31–33)and rendered the primary chi-square test with a p < 0.05. When these were removed, the chi-square test had a p > 0.05.

Meta-regression

The impact of revascularization on survival after risk adjustment for confounding variables was determined by multiple linear regression (meta-regression) using the event rate as the end point. This was used to examine the relationship between the severity of LV dysfunction and the prognostic benefit of revascularization as a function of the presence of viability (STATA software, version 6.0, Stata Corporation, College Station, Texas). This model included risk adjustment for all variables listed in Table 2including year of publication and study sample size. A final multiple linear regression model was identified with variable inclusion at p < 0.05. Finally, to compare relative diagnostic performance of the imaging modalities, three individual meta-analyses (thallium, FDG, echocardiography) were performed.

Patient characteristics

There were 3,088 patients (2,228 men), mean age 61 years and LVEF 32 ± 8%. Mean New York Heart Association (NYHA) functional class (where specified) was 2.8. Follow-up was 87.7% complete over 25 ± 10 months. Overall, 35% of the group underwent revascularization, and 65% received medical therapy. A total of 42% of patients had imaging-based evidence for myocardial viability (as defined by individual study authors). During follow-up 375 patients died (12%). Patient outcome data from the individual studies is summarized in Table 3.

Influence of myocardial viability and revascularization on death

Mortality rates from the pooled data are depicted in Figure 1a. For patients with defined myocardial viability, annual mortality rate was 16% in medically treated patients but only 3.2% in revascularized patients (chi-square = 147, p < 0.0001). This represents a 79.6% relative reduction in risk of death for revascularized patients (Fig. 1). For patients without viability, annual mortality was not significantly different by treatment method: 7.7% with revascularization versus 6.2% for medical therapy (p = NS).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

(a)Death rates for patients with and without myocardial viability treated by revascularization or medical therapy. There is 79.6% reduction in mortality for patients with viability treated by revascularization (p < 0.0001). In patients without myocardial viability, there was no significant difference in mortality with revascularization versus medical therapy. (b)Same data as (a)with comparisons based on treatment strategy in patients with and without viability. Annual mortality was lower in revascularized patients when viability was present versus absent (3.2% vs. 7.7%, p < 0.0001). Annual mortality was significantly higher in medically treated patients when viability was present versus absent (16% vs. 6.2%, p = 0.001). Revasc. = revascularization.

Examining these data grouped by treatment strategy (Fig. 1b), annual mortality was lower in revascularized patients when viability was present versus those without viability (3.2% vs. 7.7%, p < 0.0001). When patients were treated medically, those with viability had a 158% higher mortality than those without viability (16% vs. 6.2%, p = 0.001).

The multiple linear regression model most predictive of death included LVEF, presence of viability and use of revascularization (chi-square = 15, p = 0.004, pseudo r²= 0.71). This indicates that, even after adjusting to the extent possible for differences between individual patient populations, revascularization was associated with an enhanced survival rate (β = 2.79, z = 22.3, p < 0.001). This model also indicated that survival benefit with revascularization was limited to patients with viability.

Influence of severity of LV dysfunction on the effect of revascularization

This is depicted in Figure 2. The meta-regression demonstrated an inverse relationship between EF and reduction in risk of death with revascularization for patients with viability, that is, as EF decreased, the prognostic benefit with revascularization increased. No benefit was associated with revascularization in patients without viability at any level of EF.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Relation between left ventricular ejection fraction (EF) and predicted change in mortality for patients with viable (circles)versus nonviable (triangles)myocardium based on the results of meta-regression. This demonstrates increasing potential for improved survival with lower left ventricular EF in patients with viable myocardium, p < 0.0001 (broken plot line), but not in those without viability, p = 0.11 (continuous line).

Influence of viability testing technique

The individual prognostic benefit (prediction of reduced mortality with revascularization of viable myocardium) for individual imaging techniques is plotted in Figure 3. Confidence limits for thallium-201, FDG imaging and dobutamine echocardiography are wide and overlapping. No statistically significant difference in prediction of survival benefit with revascularization was detected between testing methods.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Decrease in mortality with revascularization of viable myocardium for each testing technique shown as mean value with 95% confidence limits. Note wide confidence limits, especially for thallium and echocardiography. No measurable differences in test performance were observed. EF = ejection fraction; FDG = F-18 fluorodeoxyglucose.

Prior studies

Before the advent of imaging techniques for myocardial viability testing, there were reports of the prognostic benefit of revascularization for some subgroups of patients with CAD, such as those with multivessel CAD and mild LV dysfunction (35–37). However, patients with ischemic LV dysfunction have higher periprocedural risk with revascularization compared with similar patients with normal LV function (38). This risk increases as LV dysfunction worsens. The presence of angina in the setting of significant LV dysfunction has been reported as a marker of potential survival benefit with revascularization (39). However, angina is an insensitive marker for ischemic, but viable, myocardium (40), and the benefit of revascularization may extend beyond patients with angina.

Contemporary studies

Contemporary studies employing viability testing suggest that patients with ischemic LV dysfunction may undergo revascularization with acceptable periprocedural risk and subsequent improvement in regional and global cardiac function, as well as improved symptoms (5,17,30). However, individual studies examining long-term outcomes have shown variable results, related at least in part to differences in patient populations and the limited patient numbers studied.

The current analysis

This meta-analysis yields results supporting the prognostic value of demonstrating myocardial viability in patients with CAD and severe LV dysfunction. The patients in this analysis have relatively severe LV dysfunction: mean EF 32%, mean NYHA functional class 2.8.

The strong association demonstrated between decreased mortality and revascularization is seen only in patients with myocardial viability. There is no apparent outcome benefit of revascularization in the absence of demonstrated viability, and there is a trend toward higher mortality with revascularization. This could reflect higher procedural risk for patients with severe LV impairment associated with revascularization in the absence of a balancing clinical benefit.

Relationship to severity of LV dysfunction

Multivariate modeling and meta-regression demonstrate an inverse relationship between EF and prognostic benefit associated with revascularization in patients with viability. As severity of LV dysfunction increased, the potential survival benefit associated with revascularization of patients with viability also increased. This implies that, despite an increasing procedural risk of revascularization with worsening LV dysfunction, evidence of preserved viability may provide information on potential clinical benefit to balance against that risk.

Medical therapy

The annual mortality rate observed for patients with viability treated medically is similar to that seen in contemporary clinical trials in advanced heart failure. The 16% annual mortality rate in the current analysis is comparable with the placebo group annual mortality rate of 18% in the Randomized Spironolactone Evaluation Study (RALES) (41)(patients with advanced heart failure on angiotensin-converting enzyme inhibitor therapy). The 80% reduction in death rate associated with revascularization in the current analysis exceeds the benefit generally observed in clinical trials of new therapeutics in heart failure (e.g., 30% mortality reduction in the RALES trial with spironolactone). This comparison must be tempered by the nonrandomized nature of the present analysis, as well as inevitable selection biases in making decisions for revascularization in the observational studies. However, a substantial reduction in mortality associated with revascularization in the setting of LV dysfunction is in keeping with recent autopsy data from a large heart failure clinical trial (42), suggesting that a considerable proportion of fatal events in patients with severe heart failure are associated with evidence of acute ischemia or infarction, even when death has been considered primarily arrhythmic or from progressive heart failure.

Imaging techniques

The three noninvasive testing techniques reported here interrogate distinct features of viable myocardial cells. Thallium-201 reflects cell membrane integrity; FDG reflects myocyte glucose utilization, and dobutamine echocardiography tests contractile reserve. However, there was no measurable difference between techniques in predicting prognostic benefit with revascularization. Differences between techniques have been reported in some studies regarding prediction of recovery of regional contractile function after revascularization (4), but these differences generally involve relatively small regions of myocardium. This analysis suggests that such small differences impact little on late survival. This is supported by a recent prospective randomized trial in which patients with ischemic cardiomyopathy and questions of viability were randomized to clinical decisions for revascularization based on FDG PET or Tc-99m sestamibi SPECT (43). There was no difference between groups in the proportion of patients sent for revascularization nor in two-year event-free survival, suggesting that clinical decisions and outcomes driven by these two techniques to assess viability were equivalent.

Study limitations

The data reported here are subject to limitations. The individual studies are observational, nonrandomized, unblinded and subject to publication and other biases, including patient selection bias to enter the studies and to then proceed to either medical or revascularization therapy. Furthermore, the technical aspects and completeness of revascularization and individual patients’ medical therapy regimens may have varied widely. There was little information in the reports on background medical therapy, and whether these results would hold under the conditions of contemporary medical therapy with aggressive use of statins and beta-adrenergic blocking agents is not certain. For each imaging technique, there are substantial differences in methodology, protocols and criteria for definition of clinically significant viability (Table 1). In this meta-analysis, viability could only be interpreted as “present” or “absent” based on individual studies’ definitions. Therefore, the potential significance of the extent of demonstrated viability or the presence of inducible ischemia in relationship to the degree of subsequent prognostic benefit could not be examined. The individual studies did not report late EF, so the relationship between any improvement in LV function and potential prognostic benefit could not be explored. This may have been instructive because it has recently been reported that patients with CAD and LV dysfunction who are revascularized may have similar survival regardless of improvement/no improvement in late EF (44).

Recent technical innovations including gated SPECT, nitrate-enhanced SPECT and second harmonic echocardiography were not routine at the time these studies were published. Thus, the imaging techniques may not fully reflect current practice.

Ascertainment of events was not fully complete. Finally, despite the fact that the random effects model is conservative (allowing for factors operating beyond the reported data), this allowance may not necessarily be sufficient. Thus, these findings may not necessarily be applicable to all CAD patients with severe LV dysfunction being assessed for prognostic coronary revascularization. A limitation of the literature on viability in general (and, thus, any pooled analysis of the literature) is the question of applicability to patients with very advanced degrees of heart failure symptoms and more severe LV dysfunction. In this analysis, mean NYHA class was 2.8 when reported, reflecting a mild-to-moderate degree of symptoms.

Implications

The results of this meta-analysis suggest that a search for preserved myocardial viability in patients with CAD and significant LV dysfunction using noninvasive imaging techniques identifies patients at substantial risk of death, a risk which may be reduced by successful revascularization. The magnitude of the potential reduction in mortality increases as the severity of LV dysfunction increases. Hence, noninvasive imaging of myocardial viability can be used to inform the often difficult clinical decision regarding revascularization in such patients, providing data on the potential benefit to balance against the known risks.