A new CT scan pulmonary angiography model may help optimize the diagnostic work-up process for patients with suspected pulmonary hypertension (PH), according to a recent study.
The diagnostic and prognostic utility of the model was validated in a tertiary referral population of treatment-naive patients who had a high pretest probability of PH.
“The aim of this study was to (a) build a diagnostic CT model in patients with suspected PH using the current guideline definition of PH (mPAP [mean pulmonary arterial pressure] ≥25 mm Hg) and the recent proposed definition of >20 mm Hg and (b) test its prognostic significance,” wrote Andrew J. Swift, MBChB, PhD, of the University of Sheffield (England) and colleagues in European Radiology.
The study cohort included 491 patients with suspected PH who underwent routine CT pulmonary angiography and right-heart catheterization between April 2012 and March 2016. CT metrics for patients with PH were developed using axial and reconstructed images.
The researchers identified the derivation (n = 247) and validation (n = 244) cohorts using random patient selection. In the derivation cohort, multivariate regression analysis was conducted to develop a model with the ability to predict mPAP ≥25 mm Hg and >20 mm Hg.
In the validation cohort, receiver operating characteristic analysis was performed to establish compromise CT thresholds, as well as sensitivity and specificity. The prognostic utility of the model was evaluated using Kaplan-Meier analysis.
Derivation cohort
Among the 247 patients in the derivation cohort, a CT regression model was identified, which included right-ventricle outflow tract thickness, main pulmonary artery diameter, and left ventricular area and interventricular septal angle; the area under the curve (AUC) in this cohort was 0.92.
Validation cohort
Among the 244 patients in the validation cohort, the model demonstrated strong diagnostic utility for the detection of PH, with an AUC of 0.91 and 0.94 for mPAP >20 mm Hg and ≥25 mm Hg, respectively.
With respect to the prognostic utility of the model, the researchers found that the diagnostic thresholds were prognostic in the CT model (all P < .01).
“The diagnostic CT thresholds are also of prognostic value; patients found not to have PH on CT have an excellent outcome,” they explained.
Dr. Swift and colleagues acknowledged that positive and negative predictive values will change based on the diagnostic setting. As a result, the findings from the current study may only be applicable to tertiary referral patient populations.
“This data may be particularly helpful when triaging patients with suspected severe PH for consideration of targeted pulmonary vascular therapies,” they concluded.
The study was supported by Wellcome Trust, the National Institute for Health Research, MRC POLARIS, and Bayer. The authors reported having no conflicts of interest with any companies related to the publication.
SOURCE: Swift AJ et al. Eur Radiol. 2020 Apr 27. doi: 10.1007/s00330-020-06846-1.