Prognostic Value of Quantitative Stress Perfusion Cardiac Magnetic Resonance

Objectives This study sought to evaluate the prognostic usefulness of visual and quantitative perfusion cardiac magnetic resonance (CMR) ischemic burden in an unselected group of patients and to assess the validity of consensus-based ischemic burden thresholds extrapolated from nuclear studies. Background There are limited data on the prognostic value of assessing myocardial ischemic burden by CMR, and there are none using quantitative perfusion analysis. Methods Patients with suspected coronary artery disease referred for adenosine-stress perfusion CMR were included (n = 395; 70% male; age 58 ± 13 years). The primary endpoint was a composite of cardiovascular death, nonfatal myocardial infarction, aborted sudden death, and revascularization after 90 days. Perfusion scans were assessed visually and with quantitative analysis. Cross-validated Cox regression analysis and net reclassification improvement were used to assess the incremental prognostic value of visual or quantitative perfusion analysis over a baseline clinical model, initially as continuous covariates, then using accepted thresholds of ≥2 segments or ≥10% myocardium. Results After a median 460 days (interquartile range: 190 to 869 days) follow-up, 52 patients reached the primary endpoint. At 2 years, the addition of ischemic burden was found to increase prognostic value over a baseline model of age, sex, and late gadolinium enhancement (baseline model area under the curve [AUC]: 0.75; visual AUC: 0.84; quantitative AUC: 0.85). Dichotomized quantitative ischemic burden performed better than visual assessment (net reclassification improvement 0.043 vs. 0.003 against baseline model). Conclusions This study was the first to address the prognostic benefit of quantitative analysis of perfusion CMR and to support the use of consensus-based ischemic burden thresholds by perfusion CMR for prognostic evaluation of patients with suspected coronary artery disease. Quantitative analysis provided incremental prognostic value to visual assessment and established risk factors, potentially representing an important step forward in the translation of quantitative CMR perfusion analysis to the clinical setting.

I n recent years, stress perfusion cardiac magnetic resonance (CMR) has become one of the methods of choice for the diagnosis of coronary artery disease based on high diagnostic accuracy, lack of ionizing radiation, and the ability to simultaneously assess cardiac function, myocardial perfusion, and viability (1)(2)(3). In the clinical setting, perfusion CMR is assessed qualitatively with visual analysis (4).
There are data that show that ischemia detection by visual assessment performs similarly or is at least noninferior to other noninvasive modalities such as positron emission tomography (PET) or single-photon emission computed tomography (SPECT) (5-7).
There is a growing body of evidence that shows that the mere presence of ischemia is of prognostic value (8)(9)(10)(11)(12); however, in recent years, the burden of ischemia has become an important focus (13, 14).
SPECT allows quantification of relative ischemic burden in terms of percentage of ischemic myocardium, and accepted thresholds have been shown to be of prognostic value in large nuclear data sets (13,15). The thresholds of ischemic burden used in SPECT have been extrapolated to CMR (16,17); however, evidence regarding the validity of these thresholds for CMR is lacking. CMR perfusion quantification has been shown to be feasible and has been validated against invasive and noninvasive modalities, including fractional flow reserve (18) and PET (19). Although both visual and quantitative analysis can provide data on ischemic burden, there are data that suggest the diagnostic superiority of quantitative CMR perfusion analysis, specifically in the setting of multivessel coronary disease (20). A direct comparison between visual and quantitative perfusion CMR in terms of prognostic benefit has yet to be performed.
The main aim of this study was to assess the prognostic value of quantitative CMR perfusion analysis. Furthermore, we aimed to investigate the validity of accepted consensus-based thresholds for ischemic burden in the setting of adenosine stress perfusion CMR.  Following 4 min of adenosine (140 mg/kg/min, increased to 210 mg/kg/min if there was an inadequate response, which was considered an increase in heart rate at peak stress of <10% above baseline [25]), stress perfusion data were acquired in 3 short-axis slices with a saturation-recovery k-t sensitivity encoding accelerated gradient-echo method (26), which covered 16 of the standard myocardial segments (segment 17 was excluded).

METHODS
A dual-bolus contrast agent scheme was used to correct for signal saturation of the arterial input function as previously described (27). In brief, 0.0075 mmol/kg gadobutrol (Gadovist, Bayer, Berlin, Germany) was administered as a pre-bolus at peak stress with imaging of the arterial input function. First-pass perfusion data were then acquired after the injection of 0.075 mmol/kg gadobutrol at 4 ml/s, followed by a 20-ml saline. Resting perfusion imaging was per- Surface coil intensity correction was performed before quantification using pre-contrast imaging data (29). Time signal-intensity curves were extracted using commercially available software (CVI42).
Quantitative analysis was performed blinded to baseline clinical data.

Quantitative perfusion analysis was performed by
Fermi-constrained deconvolution according to the previously described methods (30,31), in which timesignal intensity curves for the tissue impulse response function, h(t), were fitted to the Fermi function using a Marquardt-Levenberg nonlinear least-squares algorithm according to the following analytical expression:    Figure 1A illustrates the receiver-operating characteristic curves for 2-year outcome for these 3 models.
The addition to the baseline model of dichotomized visual ischemic burden using a threshold of $2 segments yielded a model of:  Figure 1B), which translated into a significant improvement in risk reclassification (Figure 2).

Kaplan-Meier curves illustrating survival in patients
stratified according to LGE, and dichotomized visual and quantitative perfusion findings are shown in Figure 3.
On further analysis of the concordance of results obtained from visual and quantitative analysis, the incidence of primary events was noted to be highest in the context of concordant positive tests (53.9% of the total events). The lowest event rate was observed in patients with negative results for both tests (9.6%).
In cases of disagreement, we observed a higher number of patients meeting the primary endpoint when classified as positive on quantitative analysis (23.1%) compared with those who were positive only on visual assessment (13.5%).

DISCUSSION
The present study represents the first to address the prognostic benefit of quantitative analysis of stress perfusion CMR. A quantitative approach was previously shown to significantly improve diagnostic performance over visual assessment specifically in the setting of multivessel coronary artery disease (20). Our data demonstrated that a quantitative approach was also superior to a visual assessment in an unselected It was encouraging that there were only a small number of nonanalyzable cases (3.5%), mainly due to respiratory motion, which is known to be a limitation of the high-resolution k-t accelerated techniques used in this study (39). The advent of novel motion correction techniques will likely ameliorate this further (40).
We found that the consensus-based threshold of $10% ischemic myocardium or $2 abnormal segments could be validly extrapolated from nuclear medicine to CMR (16). The use of these thresholds not only improved model predictive performance, but also translated into significant reclassification of patient risk using established risk categories. This was a reassuring finding because important studies such as MR-INFORM and the ISCHEMIA (International Study of Comparative Health Effectiveness with Medical and Invasive Approaches) trial used these criteria (16,17).
In the present study, abnormal perfusion was defined by an MPR <1.5. This threshold was previously validated against fractional flow reserve (18) and was similar to the optimal threshold found in some PET studies in patients with angina who require revascularization (41).     Figure 1.

Mirroring previous clinical studies, in which visual
CMR results could have influenced revascularization decisions, we excluded early revascularization events (within 90 days) from the primary composite endpoint (24). All scan reports were issued within 5 days, with a median time between scan to early revascularization of 36 days. As an additional precaution, to assess any potential bias this might have had on our results, we forced early revascularization as a covariate in our models and found that it had no impact on our findings. On this basis, we believed that the cutoff of 90 days appeared to be reasonable in our cohort for minimizing the influence of visual CMR results on revascularization decisions. Furthermore, we also proposed if there were any bias present, this would have favored a visual analysis approach because this was used for clinical decision making. We recognize that we could not fully exclude the possibility that the impact of ischemia might have been underestimated as a result of prompt revascularization.
The presence of severe ischemia, defined as vasodilator-induced systolic dysfunction by CMR, was previously shown to predict poor prognosis in a large series of patients and to identify a subgroup of patients who benefitted most from revascularization (43,44). A comprehensive evaluation of the ischemic cascade, including induced-systolic dysfunction, was not performed in this study.
Finally, this study focused solely on perfusion CMR and used visual assessment as the clinical reference standard. CMR perfusion was not directly compared with other noninvasive modalities. PET remains the noninvasive reference method for quantitative perfusion measurements. However, it is not widely available, is costly, and uses ionizing radiation. SPECT is more widely used, but it also uses ionizing radiation and has lower spatial resolution than CMR. Visual and quantitative CMR analysis were shown to perform similarly or better than these modalities in other studies, but a direct comparison was not possible in this study (7,19).

CONCLUSIONS
This study supported the use of the current consensus-