200 - Lung MRI findings in preterm infants: role of a new score for severity stratification of BPD (BPDMRI) and correlation with respiratory status at discharge

Poster Number: 200
Publication Number: 200.3234

Background: Diagnosis and staging of BPD may improve the treatment, the family counseling and follow-up of the affected infants. Lung CT scans offer high sensitivity and correlation with severity and outcomes, but are not suitable for routine imaging and longitudinal monitoring due to cumulative radiation exposure. Lung MRI score (modified Ochiai) still lacks of clinical validation.

Objective: To evaluate the accuracy of a new lung MRI score (BPDMRI) performed with commercially available sequences in assessing and stratifying BPD and to correlate it with respiratory status.

Design/Methods: Lung MRI was performed in preterm infants with or without a BPD diagnosis (Jobe Bancalari definition) while receiving a brain MRI at term. Our new score, the BPD magnetic resonance index (BPDMRI), sum of the scores of lung abnormalities based on four hyperaeration (hyperexpansion, emphysema, cysts, mosaic pattern) and four parenchymal variables (triangular subpleural opacities, fibrous stripes and bands, edema and atelectasis, distortion of bronchovascular bundles) and the modified Ochiai scoring system were used for MRI evaluation (Tab 1). The scores were retrospectively correlated with respiratory status at discharge.
The BPDMRI was also compared with the distribution of duration of MV between severity groups. We studied the intra- and inter-reader reproducibility of the BPDMRI.

Results: 25 infants were included (Tab 2). MRI showed a wide range of lung parenchyma appearances among BPD and no BPD patients and across MRI score ranges. Both the modified Ochiai score and the BPDMRI positively correlated to MV duration (r=0.828, p<.001; r=0.900, p<.001) and to any respiratory support duration until discharge (r=0,897, p<.001; r=0.953, p<.001). In the severe disease group, the length of respiratory support showed a smaller IQR using the BPDMRI rather than the clinical definition (Tab 3). Intra-reader reproducibility: bias of 0 and a CoR of 1 for each observer. Inter-reader reproducibility: bias of 1 and a CoR of 2 between reader 1 and reader 2 and reader 1 and reader 3 and bias of 0 and a CoR of 1 between reader 2 and reader 3.

Conclusion(s): Pulmonary MRI quantified lung abnormalities associated with BPD with a strong correlation with the duration and the need for MV until discharge. BPDMRI demonstrated good intra-reader and inter-reader reproducibility. BPDMRI identified infants with severe disease and greater respiratory morbidity at discharge better than clinical severity stratification; BPDMRI may contribute to a new definition of BPD based on underlying lung pathology and to predict short-term respiratory outcomes.