Detection of Right Ventricular Dysfunction in Acute Pulmonary Embolism By CT Scan: A Systematic Review and Meta-Analysis

Background: Right ventricular (RV) dysfunction is associated with higher mortality in patients with acute pulmonary embolism (PE). While precise definitions of RV dysfunction vary, up to 40% of patients with acute PE have evidence of RV dysfunction on echocardiography. The incremental utility of echocardiography for identifying RV dysfunction in acute PE is uncertain particularly when evidence of RV dysfunction is present on computed tomography (CT) pulmonary angiogram. However, the diagnostic utility of CT for detecting RV dysfunction is not well established; if CT is shown to have good diagnostic utility, it may reduce the need for routine urgent echocardiography and associated healthcare costs and resource use. We conducted a systematic review and meta-analysis to assess the diagnostic utility of CT scan findings for detecting RV dysfunction compared to echocardiography.

Methods: Using a pre-specified protocol (Prospero ID: CRD42020187812) we systematically searched the MEDLINE and EMBASE databases from inception to April 2020. Studies of any design were included so long as they (i) included patients with acute PE diagnosed by CT scan and received an echocardiographic study within a reasonable timeframe and (ii) reported on test characteristics such that a 2x2 table could be constructed to compare the two diagnostic modalities with echocardiography considered the reference standard. Results were analyzed with the bivariate mixed-effects regression framework. Study quality was assessed with the QUADAS-2 risk of bias tool.

Results: A total of 631 unique studies were identified by the search. After screening titles and abstracts and reviewing full texts in duplicate, 26 studies (n=3,404 patients) were eligible for inclusion. In individual studies, the average age (mean or median) of patients ranged from 41 to 70 years and 20% to 64.3% were female. The prevalence of RV dysfunction on echocardiography ranged studies from 22.6% to 85.7% and was 63.7% overall. The most common measurements on CT scan were increased RV/LV ratio (21 studies), pulmonary artery measurements (6 studies), RV dilatation or increased size (5 studies), measurements of vena cava size (3 studies), and coronary sinus size (2 studies). Other parameters included reflux of contrast into the inferior vena cava (IVC reflux, 4 studies) and interventricular septal deviation (5 studies). Different combinations of measurements were evaluated in 5 studies. Pooled analysis was performed for septal deviation, IVC reflux, and RV/LV ratio. In pooled analysis septal deviation had a sensitivity of 0.31 (95% CI=0.25- 0.38; I²= 51.6%) and a specificity of 0.98 (95% CI=0.90 - 1.00; I²= 46.9%). IVC reflux had a sensitivity of 0.75 (95% CI = 0.40 – 0.93; I²=95.9%) and specificity 0.75 (95% CI=0.47 – 0.93; I²= 91.1%). Where multiple RV/LV ratios were reported, we utilized the one with the highest sensitivity. The pooled sensitivity of increased RV/LV ratio was 0.83 (95% CI=0.78 – 0.87; I²= 82.9) while the pooled specificity was 0.75 (95% CI=0.66- 0.82; I²= 94.6) (Figure 1). Considering all RV/LV ratio studies, the summary receiver operating characteristic (SROC) curve had an area under the curve (AUC) of 0.86 (0.83-0.89) for identifying the presence of RV dysfunction present on echocardiography. Overall, most studies were considered at high risk of bias. Subgroup analyses demonstrated that sensitivity was significantly affected by the risk of bias for RV/LV ratio for all QUADAS-2 domains. Specificity was not affected by the risk of bias for any of the QUADAS-2 domains.

Conclusions: Our results suggest CT imaging findings appear to have moderate to good diagnostic utility for detecting the presence of RV dysfunction in acute PE. With the exception of septal deviation there was significant heterogeneity amount included studies. Overall, increased RV/LV ratio had the best test characteristics (sensitivity 0.83, specificity 0.75 and AUC 0.86). However, the veracity of these findings is limited by the high risk of bias among included studies. Further analyses regarding the diagnostic utility of combinations of multiple CT findings are warranted.