The Variability in Prognostic Values of Right Ventricular-to-Left Ventricular Diameter Ratios Derived From Different Measurement Methods on Computed Tomography Pulmonary Angiography

Abstract

To evaluate variability in right ventricular-to-left ventricular (RV/LV) diameter ratios introduced by differences in measurement methods and the subsequent influence on the accuracy of predicting outcomes for patients with acute pulmonary embolism (PE). For 200 consecutive computed tomography pulmonary angiograms positive for acute PE, RV/LV diameter ratios were retrospectively measured using 3 different 4-chamber reformations and from axial images alone. The first 4-chamber reformation method (4ch-1) was a single oblique technique using LV morphology landmarks; the other 2 methods (4ch-2 and 4ch-3) were double oblique techniques that created an intermediate short-axis image to identify the maximum RV diameter but with different approaches to reach short-axis images. Interobserver variability was measured using 30 cases. Receiver-operating characteristic analysis compared the accuracy of predicting outcomes among the 4 measurements for PE-related death, and for death or the need for intensive therapies (composite outcome). The difference in median RV/LV diameter ratios was insignificant among 4ch-2 (1.01), 4ch-3 (1.02), and axial (1.03) datasets, whereas that from 4ch-1 (0.93) was significantly lower (P<0.001). Correlation between observers was excellent for all 4 datasets (r=0.881 to 0.925). Compared with 4ch-1, the other 3 datasets equally achieved higher accuracy in predicting PE-related 30-day mortality (area under curve: 0.55 vs. 0.69 to 0.73, P=0.007 to 0.019) and a composite outcome (area under curve: 0.65 vs. 0.77 to 0.78, P=0.003 to 0.010). Double oblique 4-chamber reformation methods that use intermediate short-axis images to optimize RV size predict outcomes better in patients with acute PE than do single oblique methods using only LV landmarks; however, their accuracy is not superior to that from measurements based on axial images.