Same-Day Repeatability and 28-Day Reproducibility of Xenon MRI Ventilation in Children With Cystic Fibrosis in a Multi-Site Trial

Abstract

Background: MRI with xenon-129 gas (Xe MRI) can assess airflow obstruction and heterogeneity in lung diseases. Specifically, Xe MRI may represent a sensitive modality for future therapeutic trials of cystic fibrosis (CF) therapies. The reproducibility of Xe MRI has not yet been assessed in the context of a multi-site study.

Purpose: To determine the same-day repeatability and 28-day reproducibility of Xe MRI in children with CF.

Study type: Four-center prospective, longitudinal.

Population: Thirty-eight children (18 females, 47%), median interquartile range (IQR) age 12 (9-14) years old, with mild CF (forced expiratory volume in 1 second (FEV₁) ≥85% predicted).

Field strength/sequence: 3-T, two-dimensional (2D) gradient-echo (GRE) sequence.

Assessment: Xe MRI, FEV₁, and nitrogen multiple-breath wash-out for lung-clearance index (LCI_2.5) were performed. To assess same-day reproducibility, Xe MRI was performed twice within the first visit, and procedures were repeated at 28 days. Xe hypoventilation was quantified using ventilation-defect percentage (VDP) and reader-defect volume (RDV). For VDP, hypoventilated voxels from segmented images were identified using a threshold of <60% mean whole-lung signal and expressed as a percentage of the lung volume. For RDV, hypoventilation was identified by two trained readers and expressed as a percentage.

Statistical tests: Inter-site comparisons were conducted using Kruskal-Wallis nonparametric tests with Dunn's multiple-comparisons tests. Differences for individuals were assessed using Wilcoxon matched-pairs tests. Bland-Altman tests were used to evaluate same-day repeatability, 28-day reproducibility, and inter-reader agreement. A P-value ≤0.05 was considered significant.

Results: Median FEV₁ %-predicted was 96.8% (86%-106%), and median LCI_2.5 was 6.6 (6.3-7.4). Xe MRI had high same-day reproducibility (mean VDP difference 0.12%, 95% limits of agreement [-3.2, 3.4]; mean RDV difference 0.42% [-2.5, 3.3]). At 28 days, 26/31 participants (84%) fell within the same-day 95% limits of agreement.

Data conclusion: Xe MRI may offer excellent same-day and short-term reproducibility.

Evidence level: 2 TECHNICAL EFFICACY: Stage 2.

Keywords: Xe MRI; cystic fibrosis; pediatrics; ventilation.

FIGURE 1

Inter‐site comparison of Xe dose equivalent (DE) (a) and Xe MRI signal‐to‐noise ratio (SNR) (b). Box‐and‐whisker plots, denoting the minimum, 25th percentile, median, 75th percentile, and maximum are shown, with inter‐site comparisons conducted using Kruskal–Wallis nonparametric tests with Dunn's multiple‐comparisons tests. Panel (c) is a scatter plot of Xe MRI SNR vs. Xe dose‐equivalent volume with markers differentiating the study sites (64 instances from Visit 1 data).

FIGURE 2

Inter‐site comparison of Xe image quality. Four representative cases with demographics, FEV₁, and LCI_2.5 are shown, one from each site. These were all data from Visit 1, Scan 1 of the study. Xe MRI SNR, VDP, and consensus RDV measurements are given.

FIGURE 3

Same‐day repeatability of Xe MRI VDP and consensus RDV measurements. (a) Individual VDP changes are shown between Scan 1 and Scan 2, and in the Bland–Altman plot in (b), the mean difference in VDP measurements was 0.12% with 95% limits of agreement −3.2% to 3.4%. For the consensus RDV measurement, individual changes between Scan 1 and Scan 2 are shown in (c) and in (d) the mean RDV difference was 0.42% with 95% limits of agreement of −2.5% to 3.3%.

FIGURE 4

Same‐day repeatability of Xe MRI vs. PFTs. In (a) and (b), the same‐day difference in VDP is compared to FEV₁ %‐predicted and LCI_2.5, respectively. In (c) and (d), the same comparisons are made for RDV. The dashed lines on the panels are the same‐day limits of agreement for VDP (−3.2% to 3.4%) and RDV (−2.5% to 3.3%). There was no relationship between pulmonary function and variability in the same‐day Xe MRI measurements.

FIGURE 5

The 28‐day reproducibility of Xe MRI. In (a) and (d), individual trajectories for VDP and RDV are shown, respectfully, with no statistically significant change. In (b) and (c), the 28‐day change in VDP is plotted against the relative 28‐day changes in FEV₁ and LCI_2.5, respectively, with the dashed lines representing the same‐day limits of agreement for VDP (−3.2% to 3.4%). In (e) and (f), the 28‐day change in RDV is compared to relative 28‐day changes in FEV₁ and LCI_2.5 with the dashed lines representing the same‐day limits of agreement for RDV (−2.2% to 3.3%). The gray boxes are bound by the same‐day Xe MRI limits of agreement and by ±10% relative change in PFTs.

FIGURE 6

(a) Bland–Altman analysis to assess inter‐rater agreement in RDV measurements. The mean difference in RDV between the two readers was 0.2% with 95% limits of agreement −2.9% to 3.2%. (b) Bland–Altman analysis to compare VDP and consensus RDV measurements. The mean difference was 3.2% with 95% limits of agreement −0.7% to 7.0%, with VDP systematically producing a higher measurement of ventilation heterogeneity compared to the consensus RDV. For both these panels, Xe scans from Visit 1 (two scans per participant) and Visit 2 were combined.

FIGURE 7

Representative single coronal Xe ventilation MRI images from Visit 1 (top two rows for Scan 1 and Scan 2) and from Visit 2 (bottom row) for three participants. VDP and RDV measurements are shown below the images. CF1 (left column) showed stable ventilation with conserved ventilation deficits appearing in the same location across all three scans, and similar VDP and RDV measurements. This stability of ventilation pattern was typical of most participants. CF2 (middle column) had large same‐day ventilation changes during Visit 1, where the two scans were separated by 19 minutes; the arrows highlight ventilation deficits that were present at the first scan but not in the second scan. Notice the ventilation pattern at Visit 2 looks similar to that of Visit 1, Scan 1. For CF3 (right column), same‐day ventilation in Visit 1 was consistent, but there was an improvement in ventilation at Visit 2 with a reduction in both VDP and RDV and improvement in the right‐lung ventilation deficit marked by arrows. This participant had consistent 82% FEV₁ at both visits but improvement in LCI2.5 (11.573 at Visit 1, 10.542 at Visit 2).