Ventilation heterogeneity is a hallmark finding in obstructive lung disease. In patients with asthma, magnetic resonance imaging (MRI) has provided evidence of ventilation defects and heterogeneity; the etiology of ventilation heterogeneity is not well-understood, nor is its relationship with clinically-relevant measurements of lung mechanics. In addition, in asthmatics, respiratory system impedance values of resistance (R_rs) and reactance (X_rs) are often abnormal, and the frequency dependence of respiratory resistance is thought to reflect ventilation heterogeneity. To better understand the physiological meaning of ventilation defects quantified using MRI, we evaluated the potential relationships between ventilation defects and respiratory system mechanics measured in vivo using forced oscillation technique and using a computational airway-tree model in silico, in subjects with a clinical diagnosis of asthma. We hypothesized that the in vivo results would support our in silico findings, supporting the utility of our computational model designed to better understand MRI ventilation defects observed in asthma.

Subjects: Seventeen poorly-controlled and twenty-five well-controlled asthmatics provided written informed consent and were evaluated using noble gas MRI. Prior to imaging, all poorly-controlled asthmatics performed the forced oscillation technique. For the well-controlled asthmatics, predictions of lung impedance were derived using a computational airway tree model.¹

Image Acquisition & Analysis: Imaging was performed on a whole body 3.0 Tesla Discovery MR750 (General Electric Health Care, Milwaukee, WI) with broadband imaging capability. Hyperpolarized noble gas static ventilation and conventional ¹H MRI were acquired as previously described.^{2 3}He MRI static ventilation semi-automated segmentation was performed to generate VDP, as previously described.²

In silico Lung Impedance Predictions: MRI-derived ventilation defect maps were co-registered to an anatomically-correct airway-tree model.¹ A computational model was applied to simulate airway constriction proximal to ventilation defects, simulated measurements of R_rs and X_rs at 5 Hz were then derived from the model.

In vivo Lung Impedance Measurements: R_rs and X_rs were measured experimentally using a tremoFlo C100 (THORASYS Inc., Halifax, Canada) airflow oscillation device which employed a multi-frequency waveform with frequencies ranging from 5 to 37 Hz. Oscillometry was performed during tidal breathing while sitting upright and while wearing nose-clips.

Statistical Analysis: Data were tested for normality using the Shapiro-Wilk normality test and when data were not normal, non-parametric tests were performed. Univariate relationships were evaluated using linear regressions (r²), Pearson correlations (r) and when the data were not normal, Spearman correlations (ρ) were generated using GraphPad Prism version 6.02 (GraphPad Software Inc.; La Jolla, California, USA).

VDP was significantly worse for the 17 poorly-controlled asthmatics (12±11%) as compared to the 25 well-controlled asthmatics (4±4%, p<0.05). For the group of well-controlled asthmatics, airways proximal to MRI ventilation defects were narrowed in the computational model and respiratory system mechanics measurements were computationally derived. As shown in Figure 1, for these simulations, the relationship for VDP with model-predictions of R_rs (r=0.92, p<0.0001) and X_rs (r=-0.96, p<0.0001) at 5 Hz were statistically significant. For the group of poorly-controlled asthmatics, experimental oscillometry respiratory system mechanics measurements were obtained and VDP was significantly correlated with X_rs at 5 Hz (r=-0.53, p=0.030), but not R_rs at 5 Hz (r=0.25, p>0.05), as shown in Figure 1.In well-controlled asthmatics, impedance predictions of R_rs and X_rs at 5 Hz derived using a computation model were strongly correlated with MRI VDP. In poorly-controlled asthmatics, experimental X_rs was related to MRI VDP, but R_rs was not. These finding suggest that X_rs may be more sensitive to or a better predictor of MRI ventilation defects than R_rs. Importantly, the in vivo experimental data presented here reinforces the validity of our computational airway-tree model. In conclusion, MRI-derived ventilation defects in asthmatics can be explained in part by lung impedance, measured experimentally and using a computational model. Taken together, the combined use of experimental and simulated impedance measurements of R_rs and X_rs helps provide a better understanding of the physiological relevance of ventilation defects in asthma.