Introduction

Airways disease often shares features of both asthma and COPD¹. ¹²⁹Xe MRI is a highly sensitive imaging technique^{2, 3} which has shown utility in distinguishing these diagnostic subgroups⁴. This work aimed to assess whether lung T₁ and the associated proton spin density (M₀), and measures of heterogeneity could distinguish between patients with asthma and/or COPD, and whether they correlated with pulmonary function test (PFT) and ¹²⁹Xe MRI metrics⁴.

Methods

165 patients (aged 28-82 years) with asthma and/or COPD taking part in the NOVELTY study⁵ (NCT02760329), were recruited from 2 primary care centres. Patients were divided into three groups (asthma, COPD and asthma+COPD) based on physician-assigned diagnosis.

Imaging: Patients were scanned using a 1.5T whole body MRI system (GE HDx) and 8-element cardiac array coil. Spoiled gradient echo (SPGR) variable flip angle (VFA) images were acquired during four separate expiratory breath holds with flip angles (FAs) of 2°, 4°, 10° and 30°. HP ¹²⁹Xe MRI was acquired and analysed as previously described^{6, 7} to generate: ventilation defect percentage (VDP), ventilation coefficient of variation (Xe-CV), mean apparent diffusion coefficient (ADC) and mean acinar dimension (Lm_D).

PFTs: Spirometry, gas transfer and multiple breath washout were performed on the same day as MRI. Forced expiratory volume in 1 second (FEV₁), forced mid-expiratory flow (FEF_25-75%), lung clearance index (LCI) and the transfer capacity of the lung for the uptake of carbon monoxide (TLCO) are reported here.

T₁ and M₀ image analysis: The VFA image with FA=4° was the target image to which all other VFA images were co-registered using ANTs⁸. The same image was segmented semi-automatically using ITK-SNAP⁹ to exclude major vessels. T₁ and M₀ were calculated by fitting the signal to the SPGR equation¹⁰. Mean and standard deviation (SD) of T₁ and M₀ and the heterogeneity (median CV and interquartile range CV (IQR CV))¹¹ were calculated.

Statistical analysis: One-way ANOVA or Kruskal-Wallis test with multiple comparison correction assessed differences between groups and Spearman’s correlations¹² were performed between proton MRI and ¹²⁹Xe MRI / PFT output, with the Bonferroni-Dunn method applied for multiple comparisons.

Results

165 patients were scanned. 16 patients were excluded for a lack of analysable VFA imaging (n=4) or having a whole lung mean T₁>1500ms¹³ (n=12). Of the remaining 149 patients, 76 patients had asthma, 50 had asthma+COPD and 23 had COPD.

Mean T₁ and M₀ were higher in patients with asthma when compared to patients with asthma+COPD (p<0.0001 for both) and COPD (p=0.0191 and p=0.0107, respectively) (Figure 1). M₀ SD was higher in patients with asthma when compared to patients with asthma+COPD (p<0.0001) and COPD (p=0.0235).

Patients with asthma had lower median CV T₁ than patients with asthma+COPD (p=0.0011) and COPD (p=0.0003) (Figure 2) and lower T₁ IQR CV than patients with asthma +COPD (p=0.0110) and COPD (p=0.0034). Patients with asthma had lower median CV M₀ than patients with asthma+COPD (p<0.0001) and COPD (p<0.0001) and lower M₀ IQR CV than patients with asthma+COPD (p=0.0056) and COPD (p=0.0003). Figure 3 shows representative images from each patient cohort, where patients with asthma+COPD and COPD have increased heterogeneity when compared to the patient with asthma.

All T₁ and M₀ metrics (except SD T₁) showed moderate-to-strong correlations with Xe-MRI ventilation, ventilation heterogeneity, ADC and Lm_D (Table 1). M₀ median CV and M₀ IQR CV had the strongest correlations with Xe-MRI metrics, with the highest correlation between M₀ median CV and ADC. T₁ and M₀ metrics showed weak-to-moderate correlations with LCI, FEV₁, FEF_25-75% and TLCO (Table 2).

Discussion

Patients with asthma+COPD and COPD had reduced T₁ and M₀ when compared to patients with asthma. T₁ and M₀ displayed increased heterogeneity in patients with asthma+COPD and COPD¹⁴ when compared to those with asthma, with clearer trends between groups seen when considering the heterogeneity measures compared to mean and SD values. T₁ and M₀ decreased as ventilation heterogeneity and acinar dimensions increased, and FEV₁ and TLCO worsened.

The strong correlations between the M₀ metrics and ADC are logical given M₀ is representative of the lung tissue density and ADC represents an increase in the available air space for diffusion.

Neither T₁ nor M₀ metrics could distinguish patients with asthma+COPD from those with COPD, suggesting more sensitive biomarkers may be required⁴. The small number of patients with COPD when compared to asthma/asthma+COPD may contribute to this.

Whilst the variable flip angle method is a relatively quick (four ~14 second breatholds) acquisition strategy for T₁ mapping of the whole lung, there was a moderate failure rate of 9.7% in this cohort, with a number of these failures resulting in physiologically unreasonable values of T₁¹³ and/or M₀.

In conclusion, measures of heterogeneity of T₁ and M₀ correlate with PFTs and ¹²⁹Xe MRI and have the ability to separate patients with asthma+COPD and COPD from patients with asthma.