Prediction of Short-Term Cardiovascular Events Using Quantification of Global Myocardial Flow Reserve in Patients Referred for Clinical ⁸²Rb PET Perfusion Imaging

DISCUSSION

Our results confirm the usefulness of MFR quantification for the noninvasive work-up of CAD. Reduced global flow reserve was indicative of short-term adverse events in the entire population undergoing clinical myocardial perfusion PET, but also in a subgroup that showed no evidence of significant disease by other results of the comprehensive test.

In the absence of flow-limiting epicardial coronary artery stenoses and myocardial disease, impaired flow reserve is thought to reflect microvascular dysfunction as a consequence of atherosclerotic risk factors and thus an early stage of disease. This finding is supported by several reports of PET-defined microvascular disease in well-selected subjects with a single risk factor such as smoking, hyperlipidemia, and insulin resistance (18–22). Other studies showed that impaired flow dynamics in such subjects can be improved by various interventions (19,23–25), suggesting that it is a reversible condition. Most of our patients (i.e., those who did not show evidence of regional ischemia on qualitative PET images) seem to fall into this category, and our study shows that reduced flow reserve is predictive of an unfavorable outcome. In contrast to these studies, however, strong correlations of flow reserve with one or more risk factors were not observed in our study. This difference may be explained by the heterogeneous mix of individuals with varying risk profiles, different degrees of preexisting atherosclerosis, and different concomitant medications, which is commonly encountered in the clinical setting of CAD work-up. The lower extraction of ⁸²Rb at higher flow, when compared with alternative PET tracers, may be another factor that attenuates such correlations. And finally, microvascular angina (syndrome X) may have been another contributor to low flow reserve in this group. No component of the PET test other than global flow quantification would have been able to identify risk in subjects without regional perfusion abnormalities, suggesting an added clinical value. It is tempting to speculate that the addition of flow quantification to standard noninvasive testing may refine the detection of early disease stages and help to improve preventive measures to reverse risk. Our study provides a foundation for future prospective trials to test this hypothesis.

Although impaired flow reserve is a marker of microvascular dysfunction in the absence of flow-limiting epicardial disease, the situation is different and more complex in the presence of obstructive, flow-limiting CAD. In this setting, global flow reserve reflects a composite marker of the degree of flow limitation by macroscopic stenosis and the degree of microvascular dysfunction. These can be distinguished only if information about coronary anatomy is available. Unfortunately, because of the retrospective nature of our study, because invasive angiography is not indicated in cases with normal relative regional perfusion, and because combined assessment by CT angiography and perfusion imaging is not yet a clinical standard, morphologic studies were not consistently available in our study population. Other studies have shown that flow reserve declines with increasing degree of stenosis (26), but microvascular dysfunction may determine the threshold of ischemia in the territory of the stenotic artery (27). Furthermore, it has been shown in single-vessel disease that flow reserve can be abnormal in remote regions subtended by normal macroscopic arteries (28). In the extreme setting of balanced, flow-limiting triple-vessel disease, global flow reserve may be severely reduced in the absence of regional flow heterogeneity (29). This would not allow for conclusions about the microcirculation but would still identify a state of high risk. Given their clinical profile, which mostly indicated low risk, this situation is unlikely among our subjects with normal regional perfusion. But in the absence of angiographic results, balanced ischemia cannot be ruled out completely as a contributor to the association between low flow reserve and adverse outcome in subjects with absence of regional perfusion heterogeneity. Our results nevertheless suggest a prognostic value of global flow reserve in subjects referred for work-up of CAD, for which it may be another marker of risk in addition to the well-known prognostic value of regional perfusion defect size, which has been well established not only for SPECT (30) but also for ⁸²Rb PET (9,31,32).

Several prior studies have reported a prognostic value of PET-derived flow reserve in cardiomyopathies (33,34) and in subjects with and without known CAD (35–37). However, these studies had in common only that the cyclotron-produced tracers ¹³N-ammonia or ¹⁵O-water was used. Although those tracers are considered a gold standard, they will remain limited to few academic institutions, and they may not reach clinical acceptance because they cannot be broadly distributed. Our study is innovative in that it reports about the prognostic value of flow reserve in the general clinical setting of perfusion PET with ⁸²Rb, which is a widely commercially available generator product and increasingly used for noninvasive CAD assessment (7). Current systems routinely feature list-mode PET acquisition, which synchronously enables the sampling of dynamic datasets that are necessary for flow quantification, along with static data for standard regional perfusion analysis and electrocardiographically gated datasets for standard functional analysis (6). ⁸²Rb has potential disadvantages for absolute flow quantification because of physical properties and nonlinear extraction fraction in the high-flow range. But several algorithms for quantification have been validated and seem to work in the setting of CAD work-up, in which high flow is uncommon (16,17,38,39). Although software for flow quantification is not yet commercially available, our results support a move forward with the commercialization of such products so that the assessment of microvascular function can be more broadly implemented into clinical noninvasive imaging practice.

Another interesting aspect of our study is that global stress flow alone was predictive of outcome in our entire study group at univariate analysis. However, flow reserve, which integrates rest and stress flow, was superior at multivariate analysis and retained its prognostic value in the subgroup without regional perfusion defects, whereas stress flow alone was no longer predictive on univariate analysis (and therefore not used for a multivariate analysis). These data suggest that both rest and stress measurements are necessary, especially in the absence of obstructive epicardial disease, to obtain the most accurate prognostic information.

It should be recognized that our study has several limitations. First, because of the small sample size and short observation period, events are limited and the statistical power for identification of a truly incremental prognostic value is limited. For the same reason, and to avoid overfitting (40), we did not perform more complex multivariate outcome analyses, adjustments for multiple risk factors other than age, and analyses of interaction terms among imaging variables and risk factors.

Second, the retrospective nature and the use of a composite endpoint are further limitations. Because of the low frequency of death and myocardial infarction in our relatively healthy population, we included other events such as revascularization and hospital readmission to increase power. Flow reserve measurements were conducted offline and were not available to clinical physicians. Clinical decision making was thus not influenced by flow quantification. The assumption of risk homogeneity among the included endpoints in our study, however, may not hold true, because soft events such as angiography or readmission were used alongside hard events such as death or myocardial infarction (40). Nevertheless, all included endpoints represent significant events that resulted either in increased costs of care or reduced quality of life.

Third, our study is a single-center report, obtained from results of an unselected consecutive group of patients, as reflects local clinical practice at our institution. Clinical characteristics suggest a relatively low risk. The percentage of studies with abnormal regional perfusion is, at 35%, not high but within the range of what has been described by others (31). Only a fraction of subjects underwent early angiography after a positive PET study. Multiple factors may have played a role, such as socioeconomic issues in the local patient population (lack of insurance coverage, lack of consent to recommended procedure, expected lack of compliance with antiplatelet therapy after intervention) and an increasing awareness of referring physicians that conservative therapy may be beneficial even in mildly to moderately abnormal perfusion studies. This low rate of early angiography may even be beneficial in that it reduces bias—because of the low rate of early intervention, fewer subjects with positive scan results had to be removed from follow-up.

Fourth, because we intentionally aimed at analyzing the value of quantitative flow reserve in an unselected group of patients referred for myocardial perfusion PET, patients with a history of myocardial infarction were not excluded. Prior myocardial infarction is expected to contribute to reduced flow reserve. Results of our study confirm an independent prognostic value in all subjects and after exclusion of subjects with prior infarctions, suggesting that flow reserve is an integrated risk marker independent of prior cardiac events.

And finally, the retrospective nature, along with the profile of our patients, resulted in a relatively large fraction of subjects who were lost to follow-up. Although an effect on the results of the study cannot be completely excluded, the drop-out group did not differ from other subjects with regard to their scan results.

It is important to emphasize that this initial, preliminary report should be seen as a stimulus to include flow quantification in future large-scale prospective outcome trials. Such trials may then provide the evidence needed for a broad implementation of quantitative flow measurements in clinical perfusion imaging practice.