Multireader Determination of Clinically Significant Obstruction Using Hyperpolarized ¹²⁹Xe-Ventilation MRI.

Abstract

The objective of our study was to identify the magnitude and distribution of ventilation defect scores (VDSs) derived from hyperpolarized (HP) ¹²⁹Xe-MRI associated with clinically relevant airway obstruction. From 2012 to 2015, 76 subjects underwent HP ¹²⁹Xe-MRI (48 healthy volunteers [mean age ± SD, 54 ± 17 years]; 20 patients with asthma [mean age, 44 ± 20 years]; eight patients with chronic obstructive pulmonary disease [mean age, 67 ± 5 years]). All subjects underwent spirometry 1 day before MRI to establish the presence of airway obstruction (forced expiratory volume in 1 second-to-forced vital capacity ratio [FEV₁/FVC] < 70%). Five blinded readers assessed the degree of ventilation impairment and assigned a VDS (range, 0-100%). Interreader agreement was assessed using the Fleiss kappa statistic. Using FEV₁/FVC as the reference standard, the optimum VDS threshold for the detection of airway obstruction was estimated using ROC curve analysis with 10-fold cross-validation. Compared with the VDSs in healthy subjects, VDSs in patients with airway obstruction were significantly higher (p < 0.0001) and significantly correlated with disease severity (r = 0.66, p < 0.0001). Ventilation defects in subjects with airway obstruction did not show a location-specific pattern (p = 0.158); however, defects in healthy control subjects were more prevalent in the upper lungs (p = 0.014). ROC curve analysis yielded an optimal threshold of 12.4% ± 6.1% (mean ± SD) for clinically significant VDS. Interreader agreement for ¹²⁹Xe-MRI was substantial (κ = 0.71). This multireader study of a diverse cohort of patients and control subjects suggests a ¹²⁹Xe-ventilation MRI VDS of 12.4% or greater represents clinically significant obstruction.