Introduction

Bronchopulmonary dysplasia (BPD) is a chronic lung disease and major cause of morbidity/mortality in pre-term infants but remains poorly characterised. Free-breathing, retrospectively-gated 3D radial ¹H ultra-short echo time (UTE) MRI has been demonstrated as a safe, non-ionizing alternative to CT for visualizing structural lung disease in neonates with BPD, with UTE-derived biomarkers correlating to clinical outcomes¹. However, UTE provides limited information on lung function, and conventional pulmonary function testing poses technical and safety challenges in infants.
Hyperpolarized (HP) ¹²⁹Xe MRI is a safe, sensitive means to assess lung function, including ventilation, microstructure and gas exchange, in adult and pediatric populations². One proof-of-concept study reported preliminary ³He gas findings³, but HP ¹²⁹Xe gas MRI in infants has not been explored, and maintaining a coordinated breath-hold apnea is challenging. Recent work has shown that HP ¹²⁹Xe MRI can detect long-term functional lung changes in children born pre-term assessed at school age⁴. It is therefore vital to be able to detect such changes in early life to guide patient management. Here, we report our initial experience in performing HP ¹²⁹Xe MRI in neonates with BPD.

Methods

Prior to MRI, the feasibility of a 3-second induced breath-hold apnea in nine infants with and without lung disease aged 0—5 months was assessed using inhaled nitrogen gas at bedside. There were no adverse events or significant changes in heart rate; in some infants a ~5% drop in SpO₂ ~20s post-breath-hold was observed, which recovered within a few seconds.
Separately, two preterm male neonates with severe BPD on nasal cannula (born a) 26 and b) 24 weeks post-menstrual age, imaged at 38 and 41 weeks, respectively) were recruited for MRI, with parental consent and IRB approval from Cincinnati Children’s. MRI was performed on a multi-nuclear 1.5T neonatal-sized MRI system⁵ using a ¹²⁹Xe/¹H switched frequency high-pass birdcage coil⁶ (Figure 1). ¹²⁹Xe ventilation images were acquired using 2D slice-selective spoiled gradient echo sequences, with either a sequential Cartesian readout (FOV:24cm, resolution:3.75x3.75x9.0mm, FA:11°, ~0.5s per slice) or spiral readout (FOV:24cm, resolution:3x3x7.5mm, 12 arms, FA:20°, ~0.2s per slice). In subject b) dynamic dissolved-phase ¹²⁹Xe spectroscopy was performed⁷ (BW:31.25kHz, resolution:61Hz/pt, FA:20° frequency-selective RF pulse, TR:25.7ms).
For all acquisitions, pure ¹²⁹Xe polarized to ~40% was delivered using a facemask via Tedlar bag compression at 15cm H₂O by a neonatologist. Breath-hold apnea of ≤3 s was induced. Data were acquired repeatedly during and after the breath-hold (total period of ≥10s). To provide complementary high-quality structural images, 3D radial ¹H UTE data were acquired during tidal breathing⁸ (resolution:0.7mm³).

Results

Heart-rate and SpO₂ were within 40 bpm and 8% of baseline 2 minutes post-inhalation, with no safety concerns at follow-up. Image quality obtained from conventional Cartesian SPGR acquisitions was poor, with inconsistent signal appearance across slices (Figure 2). Spiral acquisitions showed good image quality across several temporal phases (Figure 3); SNR of the averaged image derived from the first 5 dynamics was 24.8±2.4. Qualitatively, a region of trapped gas on UTE MRI appears to spatially correspond to a ventilation defect on ¹²⁹Xe MRI (red arrows), and additional ventilation abnormalities with no apparent accompanying structural abnormality were observed (e.g. green arrow). Dynamic ¹²⁹Xe spectroscopy revealed the existence of resonances near the conventional tissue/plasma (TP) and red blood cells (RBC) resonances observed in pediatric and adult populations, which persisted throughout the breath-hold duration. The mean spectrum in this N=1 study indicates a low apparent RBC resonance at ~221ppm and possible sub-structure to the apparent TP peak at ~194 ppm. The mean RBC/TP signal ratio was ~0.24 (c.f. ~0.5 for healthy adults)).

Discussion

Whilst interpretation of the clinical significance of our findings is limited by small N, it is encouraging that HP ¹²⁹Xe MRI shows sensitivity to minor ventilation abnormalities in this population. The observation of a low RBC component relative to TP and Gas, and possible sub-structure / temporal variations in the TP peak in dissolved-phase ¹²⁹Xe spectroscopy are inconclusive, but the SNR is sufficient that the spectral appearance of fetal hemoglobin and its differences to adult hemoglobin could eventually be investigated.
Challenges: sequential Cartesian SPGR sequences appear unsuitable for imaging this population, where motion and incomplete apnea will be common; there is a need for high SNR and k-space encoding efficiency sequences (e.g. spiral), wherein the center of k-space is sampled regularly. Gas delivery is a major determinant of image quality, and whilst we have demonstrated that a short breath-hold apnea procedure is feasible in infants, maintaining the desired pressure through the facemask was challenging and often the images indicated gas not reaching the lung region. A tailored free-breathing approach with HP ¹²⁹Xe and O₂ mixtures may offer a more robust gas-delivery solution.

Conclusion

We have demonstrated that inhaled ¹²⁹Xe MRI is feasible in neonates using a short breath-hold apnea technique. It is early to draw clinically-relevant conclusions, but initial observations of minor ventilation abnormalities apparently unassociated with structural pathology, and potential differences in the dissolved-phase ¹²⁹Xe resonance structures compared with older children and adults, are promising. With further data in neonates with a range of BPD severities, we anticipate potential implications on the management of infant lung diseases and avenues for longitudinal monitoring in the neonatal period and beyond.