Myocardial blood flow reserve assessed by positron emission tomography myocardial perfusion imaging identifies patients with a survival benefit from early revascularization

We would like to thank Drs. Murthy and Hachamovitch for their interest in our manuscript. Two of their comments relate to different ways of analyzing and presenting our data. The additional information, data and graphs based on alternative methods of analysis and presentation that are already present in the Online Data Supplement should address these concerns.

Figures 3 and 5 in the paper show the instantaneous hazards of death for patients undergoing early revascularization (red lines) and for those who underwent medical therapy (blue lines), derived from the adjusted Cox proportional hazards model. The hazards with their 95% confidence intervals are plotted across different values of myocardial blood flow reserve (MBFR), and the MBFR value where the observed hazards with early revascularization and medical therapy are similar was chosen as the threshold for equipoise. As suggested by the authors, another way to present the data is to plot the hazard ratios of death (or cardiac death) for early revascularization compared to medical therapy across different levels of MBFR. This alternative method of displaying result is shown in Figure 2 in the Online Data Supplement. In this figure, hazard ratios < 1 suggest a survival benefit with early revascularization compared to medical therapy. A cutoff value of MBFR=1.8 was also demonstrated in this figure, with hazard ratios < 1 for MBFR values below 1.8 and hazard ratios > 1 for MBFR values > 1.8.

The choice of covariate adjustment as the primary analysis strategy was made prior to the analysis, as we believed it would show similar results to propensity score matching. Given the high number of events on follow-up, we had adequate power to adjust for most confounders. We did conduct a sensitivity analysis using an alternative strategy of propensity scoring which showed similar results to the primary analysis (as seen in the Supplemental Methods in the Online Data Supplement). In this sensitivity analysis, we generated a propensity score model to adjust for the non-randomization of treatment allocation of undergoing early revascularization in 90 days. This included a total of 27 covariates which could potentially affect the decision to revascularize, including patient demographics, clinical risk factors, prospectively elicited symptoms, medications, stress data, gated data and perfusion data. The interactions were then tested in a Cox proportional hazards regression model for mortality, which adjusted for propensity score for early revascularization as a covariate along with adjusting for early revascularization and MBFR.

We agree with the authors that our finding that the extent of ischemia did not modify the interaction between MBFR and revascularization is both unexpected and interesting. Spatially-relative perfusion defects have proved useful in decisions about referral to coronary angiography, but often they under-estimate or overestimate disease extent and severity,(1 )potentially explaining this finding. Myocardial blood flow reserve as measured by PET captures the combined hemodynamic effects of epicardial disease, diffuse disease, and the microcirculation, and as such is likely better suited as a measure of physiologically significant ischemic heart disease. Measures that lower the amount of myocardium with low flow reserve, including revascularization,(2) may help improve survival by this mechanism. It is also possible and likely that some of the survival benefit noted in the study was driven by benefit in patients with significant epicardial obstructive disease with or without concomitant microcirculatory disease, who have a reduction in the amount of myocardium with low MBFR following revascularization. We do acknowledge that we need better methods to non-invasively prospectively identify patients with purely microvascular disease from others, as they would not be expected to benefit from revascularization.

References:

1. Beller GA. Underestimation of coronary artery disease with SPECT perfusion imaging. Journal of Nuclear Cardiology. 2008;15(2):151-153.
2. Driessen RS, Danad I, Stuijfzand WJ, et al. Impact of Revascularization on Absolute Myocardial Blood Flow as Assessed by Serial [(15)O]H2O Positron Emission Tomography Imaging: A Comparison With Fractional Flow Reserve. Circ Cardiovasc Imaging. 2018;11(5):e007417.

We read with great interest the study by Patel, K.K., et al1, which confirmed the prognostic value of myocardial flow reserve (MFR) measured by positron emission tomography (PET) and proposed an interaction with early revascularization such that patients with MFR≤1.8 who underwent early revascularization compared to those who were managed medically.
We find Figures 3 and 5 to be confusing. The y axis, labelled as representing “hazards”, is labelled 0 to 0.7 in the former and 0 to 0.6 in the latter. What is the reference group for computing this? We would have expected a value of 1 somewhere in the range of observed MFR and are concerned this represents either an error in plotting or modelling. Alternatively, is this an expected event rate and if so over what duration?
Second, the finding that extent of ischemia did not modify the interaction between MFR and revascularization likely reflects clinically appropriate bias in referral to coronary angiography. Prior studies have shown that the majority of patients even with very low MFR do not have coronary anatomy with proven survival advantage from revascularization.2,3 This is due to the fact that MFR can be reduced by focal epicardial stenosis, diffuse disease or microvascular dysfunction. Furthermore, the subgroup normal perfusion or trivial relative perfusion defects is enriched for diffuse disease and microvascular dysfunction and would not be expected to have survival benefit from revascularization.
Finally, previous studies examining the question of treatment benefit utilized propensity scores to reduce bias from the lack of randomization. Recent work4 reports that routine covariate adjustment may be equivalent to propensity scores in certain settings and suggest a prespecified primary analysis strategy, secondary sensitivity analyses using alternative approaches, and that post-hoc methodology selection results in bias. Examination of the cohort absent outcomes to determine the optimal method used is suggested. The current study’s comparison of MFR and myocardial perfusion without a propensity score is problematic as a prognostic comparison between metrics when one is the basis of post-imaging catheterization and the other not likely introduces errors in observed relative event rates.5

Venkatesh L. Murthy, MD, PhD
Division of Cardiovascular Medicine
University of Michigan, Ann Arbor

Rory Hachamovitch, MD, MSc
Cardiovascular Imaging Section, Department of Cardiovascular Medicine
Cleveland Clinic


1. Patel KK, Spertus JA, Chan PS, Sperry BW, Al Badarin F, Kennedy KF, Thompson RC, Case JA, McGhie AI, Bateman TM. Myocardial blood flow reserve assessed by positron emission tomography myocardial perfusion imaging identifies patients with a survival benefit from early revascularization. Eur Heart J 2019;
2. Naya M, Murthy VL, Taqueti VR, Foster CR, Klein J, Garber M, Dorbala S, Hainer J, Blankstein R, Resnic F, Di Carli MF. Preserved coronary flow reserve effectively excludes high-risk coronary artery disease on angiography. J Nucl Med 2014;55:248–255.
3. Ziadi MC, Dekemp RA, Williams K, Guo A, Renaud JM, Chow BJW, Klein R, Ruddy TD, Aung M, Garrard L, Beanlands RSB. Does quantification of myocardial flow reserve using rubidium-82 positron emission tomography facilitate detection of multivessel coronary artery disease? J Nucl Cardiol 2012;19:670–680.
4. Elze MC, Gregson J, Baber U, Williamson E, Sartori S, Mehran R, Nichols M, Stone GW, Pocock SJ. Comparison of Propensity Score Methods and Covariate Adjustment: Evaluation in 4 Cardiovascular Studies. J Am Coll Cardiol 2017;69:345–357.
5. Hachamovitch R, Di Carli MF. Methods and Limitations of Assessing New Noninvasive Tests: Part II: Outcomes-Based Validation and Reliability Assessment of Noninvasive Testing. Circulation 2008;117:2793–2801.