Stress CMR With Vasodilators to Predict Long-Term Mortality

Background The prognostic implications of stress cardiac magnetic resonance (CMR) imaging with vasodilators are receiving increasing attention. However, current prognostic data from this modality are mostly derived from small cohorts or meta-analyses and tend to focus on short- to mid-term outcomes, often involving composite endpoints of minor events.

All-cause mortality is the most severe and objective clinical outcome across all diseases. To date, no large registries have demonstrated the value of stress CMR in predicting long-term all-cause mortality. Furthermore, the potential impact of vasodilator stress CMR-guided revascularization on all-cause mortality rates has not been systematically analyzed.

Stress CMR allows for a comprehensive evaluation of patients with suspected or known ischemic heart disease. Beyond well-established left ventricular (LV) parameters such as ejection fraction (EF), this technique enables the simultaneous assessment of ischemia-based on stress-induced perfusion deficits-and myocardial necrosis-based on late gadolinium enhancement (LGE) imaging.

The objective of this study is to assess the prognostic value of vasodilator stress CMR for predicting long-term all-cause mortality in patients with known or suspected ischemic heart disease, using a large registry of over 5,500 patients followed over 15 years. In addition, the study aims to explore therapeutic implications regarding mortality reduction based on CMR results and subsequent clinical management strategies.

Materials and Methods Study Population This is a retrospective registry including all consecutive patients referred by their attending cardiologists for clinically indicated vasodilator stress cardiac magnetic resonance (CMR) at the CMR Unit between January 2001 and December 2016. Of the 5,704 patients initially included in the database, 152 were excluded due to poor image quality (n = 75) or incomplete studies (n = 77). Thus, the final study population consisted of 5,552 patients.

Reasons for vasodilator stress CMR were as follows:

• Suspected ischemic heart disease (ambulatory patients: n = 397, 23%; hospitalized patients: n = 397, 23%)
• Assessment of residual ischemic burden in chronic ischemic heart disease (ambulatory: n = 397, 23%; hospitalized: n = 397, 23%)
• Assessment of residual ischemic burden in patients hospitalized with acute coronary syndromes, including:

Non-ST-segment elevation acute coronary syndrome (NSTE-ACS): n = 173, 10% ST-elevation myocardial infarction (STEMI): n = 173, 10% All clinical characteristics and CMR data were recorded prospectively and entered immediately into the database. Stress CMR results were made available to the attending cardiologists, who were also responsible for patient management and treatment decisions. Cardiac events and revascularization procedures will be collected retrospectively. All participants provided written informed consent before undergoing vasodilator stress CMR.

CMR Protocol All CMR studies were performed using a 1.5-T scanner (Sonata Magnetom, Siemens, Erlangen, Germany) following a standardized protocol. Imaging was acquired using a body surface coil during breath-holds with ECG gating.

Cine imaging was performed in long-axis and contiguous short-axis slices from the base to apex of the left ventricle (LV) using a steady-state free precession (SSFP) sequence (TR/TE: 2.8/1.2 ms; flip angle: 58°; matrix: 256 × 100%; slice thickness: 7 mm).

Vasodilation was induced in most patients (n = 4,700; 85%) using dipyridamole intravenously (0.56 mg/kg over 4 min, and if tolerated, 0.84 mg/kg over 6 min). In 852 patients (15%), vasodilation was achieved using:

• Adenosine (n = 300; 10%) at 140 μg/kg/min for 6 minutes
• Regadenoson (n = 106; 5%) as a single 0.4 mg intravenous bolus injected over 10 seconds Following vasodilator infusion, 0.1 mmol/kg of gadopentetate dimeglumine (Magnograf, Schering AG, Berlin, Germany) was injected IV at 5 mL/s.

First-pass perfusion imaging during hyperemia included at least four short-axis slices and two long-axis views (2- and 4-chamber), using a saturation-prepared SSFP sequence (TI: 125 ms; TR/TE: 202/1 ms; flip angle: 50°; matrix: 192 × 96; FOV: 350 × 220 mm; slice thickness: 8 mm).

Late gadolinium enhancement (LGE) images were acquired 10 minutes post-contrast in the same locations as cine images using an inversion-recovery segmented SSFP sequence (TR/TE: 700/1.26 ms; flip angle: 45°; matrix: 256 × 184; FOV: 340 × 235 mm; slice thickness: 8 mm). The inversion time was adjusted to null normal myocardium, as previously described.

Rest perfusion imaging was performed at the end of the study using the same first-pass protocol.

CMR Data Analysis Experienced operators conducted and interpreted CMR studies using validated software (Syngo, Siemens, Erlangen, Germany). Ambiguous findings were reviewed jointly, and final decisions were made by consensus.

Left ventricular ejection fraction (LVEF, %) and indexed end-diastolic and end-systolic volumes (mL/m²) were measured by manual planimetry of endocardial and epicardial borders across all short-axis cine images.

Three segmental CMR indices were defined using the 17-segment AHA model:

• Abnormal wall motion at rest: segments with hypokinesia, akinesia, or dyskinesia
• Ischemia: defined as a persistent perfusion delay in ≥3 consecutive frames in segments with normal resting perfusion; segments with resting perfusion defects (e.g., core infarct) were not considered ischemic
• Late gadolinium enhancement (LGE): defined as signal intensity >2 SD above remote myocardium in ≥50% of myocardial wall thickness To reduce artifact-related misclassification, an index was considered abnormal if findings were present in more than one segment and seen in both short- and long-axis views.

The inter- and intraobserver variability for CMR indices in this study was <5%, based on previously published reproducibility methods from our group.

For outcome prediction, CMR results were classified using a two-step system: ischemia was considered present if it involved more than one segment.

Definitions of Events

• Primary outcome: All-cause mortality during follow-up.
• Secondary outcome: Cardiac death, defined as death due to acute myocardial infarction, congestive heart failure, arrhythmias, or cardiac arrest.

CMR-related revascularization was defined as any procedure directly guided by CMR results or performed within 3 months (in the absence of events) following CMR. An exploratory analysis will evaluate the impact of CMR-guided revascularization on major events based on the presence or absence of ischemia.

Follow-up will be centralized and conducted by four cardiologists from the Cardiology Department at Hospital Clínico Universitario de Valencia. Event adjudication will require consensus between two cardiologists.

Statistical Analysis All variables will be tested for normal distribution using the Kolmogorov-Smirnov test.

• Normally distributed continuous variables will be expressed as mean ± SD and compared using Student's t-test
• Non-parametric data will be reported as median [interquartile range] and compared using the Mann-Whitney U test
• Group comparisons will be performed using one-way ANOVA with Bonferroni correction for multiple testing
• Categorical variables will be compared using Chi-square or Fisher's exact test, as appropriate Based on a pilot study (601 patients), we planned a case-control design with 15 controls (patients without inducible ischemia) per case (patients with inducible ischemia).
• Assuming an event rate of 0.0485 in controls and 0.136 in cases, we estimated a required sample of 106 cases and 1,590 controls to achieve 90% power (β = 0.1) with a significance level of 0.05 (α).
• An uncorrected chi-square test was used for initial hypothesis testing. The final study population exceeds these requirements, ensuring sufficient statistical power.

Time-to-event associations for CMR variables will be analyzed using Cox proportional hazards models, with stepwise multivariate adjustment for baseline and CMR variables with p < 0.2 in univariate analysis.

• Hazard ratios (HR) and 95% confidence intervals (CI) will be calculated
• Kaplan-Meier curves will estimate survival distributions To evaluate the effect of CMR-guided revascularization on outcomes, multivariate Cox models adjusted for propensity scores (derived from logistic regression including baseline and CMR variables) will be used to account for referral bias.
• The C-statistic will assess the accuracy of the propensity model
• A p-value < 0.05 will be considered statistically significant All analyses will be performed using SPSS version 13.0 (SPSS Inc., Chicago, IL, USA).