Predictive accuracy of semi-quantitative scoring to screen for unfavorable ejection fraction and infarct size

Background

Manual planimetry for quantitative analysis of ejection fraction (EF) and infarct size (IS) on cardiac magnetic resonance (CMR) imaging are too time-consuming, and therefore, impractical in daily clinical practice. We sought to derive and examine the predictive accuracy of a semi-quantitative scoring technique to screen patients with an EF ≤35% or an IS ≥18.5%, known independent predictors of increased cardiac events and mortality.

Methods

The CMR derivation cohort consisted of 122 ST-segment elevation myocardial infarction patients. The validation cohort consisted of an additional 172 patients from the multi-center DEfibrillators To REduce Risk by MagnetIc ResoNance Imaging Evaluation (DETERMINE) trial. Cines were scored on a 17-segment model for wall motion and totaled for the Sum Motion Score (SMS): 0 = normal, 1 = mild hypokinesis, 2 = moderate to severe hypokinesis, 3 = akinesis, 4 = dyskinesis. Viability images were scored for infarct transmurality and totaled for the Sum Infarct Score (SIS): 0 = none, 1 = 1-25%, 2 = 26-50%, 3 = 51-75%, 4 = 76-100%. Quantitative EF and IS were manually planimetered using QMass (Medis).

Results

From the derivation cohort, the SMS correlated with EF (R=-0.91, p<0.001) and SIS correlated with IS (R=0.94, p<0.001). Linear regression equations between SMS vs. EF and SIS vs. IS were obtained to estimate EF (est-EF = 55 - SMS) and IS (est-IS = 1.383 * SIS).

In the validation cohort, the mean EF was 39.0 ± 11.7% (32% with EF ≤35%), and the mean IS was 17.3 ± 10.4% (40% with IS ≤18.5%). Using the derivation formula, the est-EF (36.2 ± 10.9%) correlated with EF (R = 0.9) with a slight underestimation of the mean difference by 2.8 ± 4.9%. In addition, the est-IS (21.8 ± 12.8%) correlated with IS (R = 0.9). The est-IS tended to overestimate IS by 4.5 ± 5.9%. The sensitivity for detecting an EF ≤35% using the est-EF was 94% with a negative predictive value of 97%. The sensitivity for detecting an IS ≥18.5% using the est-IS was 96% with a negative predictive value of 96%.

Conclusions

Semi-quantitative scoring is a sensitive screening tool that can be used to identify patients with an ejection fraction ≤35% or infarct size ≥18.5% to provide a rapid alternative method to manual planimetry.