The Age-dependent Upper Limit of Normal for Hyperpolarized 129Xe MRI Ventilation Defect Percent in Healthy Individuals Using a Multi-center Database

Rationale: Hyperpolarized 129Xe magnetic resonance imaging (129XeMRI) is an innovative, clinically approved imaging technique that provides novel insights into lung structure and function such as the ventilation defect percent (VDP). An upper limit of normal (ULN) for VDP has previously been proposed (McIntosh et al, 2023), but ULN accounting for variations in age as well as imaging sites, scanner platforms, and acquisition protocols remain undefined. We aimed to determine the ULN for VDP in healthy individuals across sites, scanners, acquisition methods, and VDP-quantification techniques, offering a reference for assessing age-related lung function and setting benchmarks for clinical interpretation and multisite comparisons. Methods: Hyperpolarized 129Xe and anatomical MRI were collected from 281 healthy participants (age 33±17yrs [5-84yrs]; 142F/139M) at 10 sites (Figure-A), all of which are part of the 129XeMRI Clinical Trials Consortium (https://www.129xectc.org/). Participants inhaled 129Xe dosed based on a percentage of TLC, FVC, or up to 1L, followed by a <20-second breath-hold. Images were acquired using Siemens, Philips, or GE MRI systems (1.5T/3T) with different acquisitions protocols (2D/3D, cartesian/spiral, axial/coronal). Anatomical images were registered to 129Xe ventilation images, and lung masks were generated using pretrained segmentation models with manual corrections. N4-bias correction was applied to correct 129Xe signal variations. VDP was calculated using six methods: 60% thresholding of the mean lung signal (TH60%), K-means, Adaptive-K-means (AKmeans), linear-binning normalized by mean-signal (LBmean), and generalized-linear-binning normalized by median (GLBmed) and 99-percentile (GLB99p). The ULN for VDP was calculated for the total group and for age subgroups as Mean(VDP)+1.96·(SD(VDP)/√N), where N is the number of participants in the respective group. Results: Representative 129Xe ventilation images with VDP for four healthy participants across age and VDP methods (Figure-B) demonstrated a clear trend of increased VDP with increasing age. Across all participants, VDP correlated significantly with age (P<0.001) for all methods (Figure-C). However, mean VDP and variability were numerically greater for TH60% (2.47±2.12%), GLBmed (1.73±1.62%), and GLB99p (1.64±2.19%) compared to Kmeans (0.34±0.53%), AKmeans (0.33±0.67%), and LBmean (0.33±0.83%). ULN values stratified by age group are shown for each method in Figure-D. Conclusions: The observed age-dependent increase in the ULN for 129XeMRI VDP across sites, platforms, acquisition protocols, and VDP quantification methods reflects “real-world” conditions and serves as a critical reference for clinical interpretation. Notably, significant differences in mean and variability of VDP were observed across methods, highlighting the need for method-specific benchmarks to ensure consistent interpretation and facilitate comparisons across studies and clinical settings.

Duke Scholars

Author Bastiaan Driehuys Radiology