4140

The Nature of Early Airway Obstruction in Cystic Fibrosis Lung Disease using UTE MRI and 129Xe MRI

Robert P Thomen¹, Laura L Walkup², Nara S Higano², David J Roach², Zackary I Cleveland², Andrew Schapiro³, Alan Brody³, John Paul Clandy⁴, and Jason C Woods²

¹University of Missouri, Columbia, MO, United States, ²Center for Pulmonary Imaging Research, Cincinnati Children's Hospital, Cincinnati, OH, United States, ³Radiology, Cincinnati Children's Hospital, Cincinnati, OH, United States, ⁴Pulmonary Medicine, Cincinnati Children's Hospital, Cincinnati, OH, United States

Synopsis

Cystic Fibrosis (CF) lung disease is characterized by identifiable structural abnormalities on CT or UTE MRI, such as mucus plugs and bronchiectasis. Similarly, hyperpolarized ¹²⁹Xe MRI can provide functional information related to regional pulmonary ventilation, with high sensitivity to early obstruction. Here we combined UTE and ¹²⁹Xe MRI to quantify the extent to which specific structural abnormalities contribute to regional ventilation defects. Mucus plugs and bronchiectasis account for the majority of ventilation defects that can be attributed to structural abnormalities. However, structurally-unattributable ventilation defects are more common, implying higher sensitivity than structural imaging to lung dysfunction in milder disease.

Introduction

Cystic fibrosis (CF) lung disease is characterized a number of different structural abnormalities that traditionally are identified as bronchiectasis, bronchial wall thickening, mucus plugs, and regions of consolidation or atelectasis, but the precise extent to which these structural abnormalities determine lung function decline is poorly understood. This is especially true in pediatric patients, where lung function decline is most challenging to characterize by repeatable spirometry. Many of these structural abnormalities are easily visualized by x-ray CT, but research efforts have shifted toward emerging techniques in MRI (namely Ultra-Short Echo [UTE] MRI) as a radiation-free surrogate that can be used for more frequent screening¹. Further, hyperpolarized ¹²⁹Xe MRI has shown exceptional sensitivity to early-stage lung function decline in CF, with regional specificity². By combining the structural abnormalities identified by UTE MRI with the functional abnormalities identified by ¹²⁹Xe MRI, it is possible to directly quantify the extent to which individual abnormalities affect regional ventilation. In this study, we quantified ¹²⁹Xe defect volumes attributable (and unattributable) to specific structural abnormalities in order to characterize regional and global structure-function correlates. We hypothesized that mucus plugs would be responsible for the majority of ventilation defects in earlier disease and that this would be accompanied by increased bronchiectasis/bronchial wall thickening with increasing ventilation defects and decreased pulmonary function.

Methods

5 control subjects (mean age 10.8 years) and 22 CF patients (mean age 14.7, 5 adults) were imaged via UTE MRI (TR/TE=5.78ms/0.2ms, Flip Angle=5°, Voxel Size=1.39x1.39x4mm3) and ¹²⁹Xe MRI (FA=10°-12°, TR/TE=8ms/4ms, Voxel size=3x3x15mm³) and underwent standard spirometry. Regions of poor signal in ¹²⁹Xe images (<60% of the whole-lung signal mean) were identified as defects and globally quantified as a subject’s ventilation defect percentage [VDP]². UTE images were independently scored by 2 expert radiologists who regionally identified 5 structural abnormalities via the Brody scoring method (bronchiectasis, bronchial wall thickening, mucus plugs, consolidations, and atelectasis)³. Radiologists also provided a nominal score between 0 and 5 indicating the quality of each subject’s UTE image data. UTE and ¹²⁹Xe images were then spatially matched and individual regional defects were attributed to an associated structural abnormality where possible. Because of the potential ambiguity between bronchiectasis and wall thickening, and the small number of consolidations and atelectasis, defects were ultimately attributed into 4 classes: bronchiectasis/bronchial wall thickening [BR], mucus plugs [MP], consolidations/atelectasis [CA], and unattributed (Fig 1).

Results

Pearson correlation coefficients were r=-0.78 (p<10^-5) between subject FEV₁ (Forced Expiratory Volume in 1 second) and VDP, and r=0.69 (P<10^-4) between FEV₁ and unattributed VDP (Fig 2). Correlation between subject VDP and total number of identified pathologies via Brody scoring was r=0.85 (P<10^-8). For subjects with mild disease (FEV₁>80%), an average 65% of defects were unattributed and 10% were due to BR/MP. Conversely, in subjects with FEV₁<80%, 86% of defects were due to BR/MP and 27% were unattributed. A Pearson r=-0.84 (p<10^-6) was found between unattributed VDP and absolute VDP. Radiologists’ quality scores averaged 3.3 ± 0.7 for all UTE datasets. An insignificant correlation was found between percentage of unattributed defect percentage and radiologists’ average image quality score (r = -0.23, p=0.24). Average ¹²⁹Xe Signal-to-Noise ratio (SNR) was 11.4 ± 7.0 (mean whole-lung signal divided by standard deviation of all signal in voxels outside the lung), and an insignificant correlation was found between subject ¹²⁹Xe SNR and VDP (r =-0.23, p=0.23). Figure 3 presents a bar plot of individual subject VDP’s color-coded by attributed pathology, ordered by increasing VDP, which correlated with subject age (r = 0.49, p = 0.01).

Discussion

This work complements previous studies highlighting ¹²⁹Xe MRI as a highly sensitive measure of regional lung function, particularly in mild disease. The large percentage of unattributed defects in those with mild disease indicates that ¹²⁹Xe is likely much more sensitive to lung function decline than structural imaging alone. The weak correlation between UTE image quality and percentage of unattributed defects indicates that lack of abnormality attribution was not due to inadequate image quality in this reasonably high-quality set of images. Likewise, SNR was found to be adequate for proper defect characterization and did not impact VDP.

Conclusions

Hyperpolarized ¹²⁹Xe reveals significant functional deficits even in the absence of visible structural abnormalities on UTE MRI; attribution of defects to structural abnormalities increases with increasing VDP (and decreasing FEV₁). Mucus plugs and bronchiectasis/bronchial wall thickening account for the vast majority of attributable defects in pediatric CF lung disease.

Acknowledgements

No acknowledgement found.

References

1. Roach DJ, Crémillieux Y, Fleck RJ, Brody AS, Serai SD, Szczesniak RD, et al. Ultrashort Echo-Time Magnetic Resonance Imaging Is a Sensitive Method for the Evaluation of Early Cystic Fibrosis Lung Disease. Annals of the American Thoracic Society. 2016;13(11):1923-31.

2. Thomen RP, Walkup LL, Roach DJ, Cleveland ZI, Clancy JP, Woods JC. Hyperpolarized 129Xe for investigation of mild cystic fibrosis lung disease in pediatric patients. J Cyst Fibros. 2017;16(2):275-82.

3. Brody AS, Kosorok MR, Broderick LS, Foster JL, Laxova A, Bandla H, Farrell PM. Reproducibility of a scoring system for computed tomography scanning in cystic fibrosis. J Thorac Imaging. 2006;21(1):14-21

Figures

Figure 1. Example UTE and ¹²⁹Xe axial image slices with and without pathologies/defects identified in a male, 8 year old, CF patient with FEV₁ = 102% (subject 23 in Fig 3). Color legend is given below.

Figure 2. (a) Ventilation Defect Percentage vs FEV₁ percent of predicted. (b) Percentage of subject VDP which was not attributed to pathology vs FEV₁ percent of predicted. (c) Subject FEV₁ vs Age. (d) Subject VDP vs Age. Blue points indicate healthy controls and r values are Pearson correlation coefficients with respective p values.

Figure 3. Bar plot of ventilation defect percentages for each subject color-coded by pathology contribution: Red indicates defects caused by Bronchiectasis/Bronchial Wall Thickening [BR] but not mucus plugs [MP], yellow indicates defects caused by MP but not BR, orange indicates both BR and MP, blue indicates defects which were attributed to consolidation/atelectasis [CA] pathologies, green indicates defects which were not attributed to pathology [unattributed]. Subject numbers for controls are colored blue; numbers above bars are subject FEV₁ percent of predicted.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)

4140