Interstitial lung diseases (ILD) are a group of complex diseases characterized by various patterns of inflammation and fibrosis affecting the alveolar, interstitial, and vascular compartments of the lung [1
]. ILD can be classified by its numerous causes, such as hypersensitivity pneumonitis, connective tissue disease, pneumoconiosis, drug toxicities or radiation exposure. ILD with no identifiable cause is commonly due to idiopathic pulmonary fibrosis (IPF), one of the most aggressive forms of ILD. Patients with IPF often have other comorbid conditions, including emphysema, lung cancer, and pulmonary hypertension [2
]. A recent systematic review investigating the incidence of IPF estimated it in Europe and North America to be 3 to 9 cases per 100,000 per year. In East Asia and South America, the incidence was lower [3
]. The mortality of IPF is high. There is a median survival of 2–3 years after diagnosis, and evidence shows survival has not been improving over the last several years. In fact, mortality rates seem to be rising, though this could be due to increased awareness and knowledge of IPF [4
Our current understanding of the pathogenesis of IPF posits that there is an abnormal repair process of injured alveoli in the lungs. Oxidative injury, mitochondrial dysfunction, and shortened telomeres can result in decreased proliferation of alveolar epithelial cells and release of profibrotic mediators. These mediators, such as platelet-derived growth factor (PDGF) and tumor necrosis factor-beta (TGF-β), can stimulate matrix deposition by myofibroblasts. Accordingly, IPF shows a characteristic pattern termed usual interstitial pneumonia (UIP) on lung biopsy. This pattern describes heterogenous paraseptal fibrosis with remodeling of the lung architecture. On high-resolution computed tomography (HRCT), the UIP pattern consists of reticulation and honeycombing in the peripheral regions of the lungs and in the lower lobes [5
No cure for IPF exists. The current management of IPF involves attempting to slow disease progression. Two medications that have shown some evidence in accomplishing this include the antifibrotic agents Nintedanib and Pirfenidone, both of which have adverse effects that must be considered when choosing the best option for a patient. On the other hand, anti-acid therapy has fewer adverse effects, and one study showed it may increase survival in IPF patients. However, it is important to note that anti-acid therapy has not been studied in randomized trials [6
A critical barrier to progress in the field is the unpredictability of the clinical course of interstitial lung diseases: for example, the rate of decline in lung function is highly heterogeneous in IPF [7
]. Furthermore, the progression IPF is often episodic, characterized by prolonged periods of relative stability, and abrupt and unpredictable acute exacerbations that are associated with markedly worsened prognosis [8
]. Developing tools for prior identification of patients at risk for disease progression would constitute a major advancement in the field by allowing targeting of such at-risk populations for currently approved antifibrotic therapies and for trials of novel agents. Achieving an earlier diagnosis using more sensitive technologies has the additional potential to allow for patients to receive earlier treatment [9
]. Finally, the stratification of ILD patients is a necessary step toward the development of more effective and personalized treatments [10
]. Currently, HRCT, lung biopsy, and spirometry tests are the main techniques used for the diagnosis of IPF, for classifying disease severity, and for assessing the efficacy of antifibrotic therapies [2
]. However, these techniques entail radiation exposure, are invasive, or can only examine global lung function.
Hyperpolarized Xenon-129 magnetic resonance imaging (HP Xe-129 MRI) is a promising candidate for fulfilling these unmet needs. It has enabled many approaches, such as ventilation and diffusing imaging, for the evaluation of lung structure and function [11
]. More specifically, xenon has several unique properties that are well suited for probing regional gas exchange impairment in IPF. Xenon is sensitive to its chemical environment, produces an enormous range of chemical shifts, and has a high solubility in biological tissues [14
]. Upon inhalation, approximately 2% of the gaseous xenon dissolves into the surrounding pulmonary tissue, plasma, and red blood cells (RBC). The gaseous and dissolved xenon produces a set of detectable MR spectral peaks at distinct frequencies that can be measured and quantified [16
]. Three peaks are typically observed in human Xe-129 lung spectra, corresponding to gaseous xenon in the airspaces (0 ppm), xenon dissolved in the interstitium (197 ppm), and xenon associated with hemoglobin in the blood (217 ppm).
To exploit these properties of xenon, we developed an MR spectroscopic imaging technique termed 3D Single-Breath Chemical Shift Imaging (3D-SBCSI). This technique is capable of non-invasively assessing regional lung ventilation and gas exchange within a single breath hold of less than 10 s [18
]. Previous studies in animal lung fibrosis models have demonstrated that tissue thickening can be detected using 3D-SBCSI [21
Because lung anatomy, ventilation, and gas exchange can be examined by 3D-SBCSI, radiation exposure that comes with HRCT can be avoided, and the assessment of regional lung function rather than global lung function can be achieved in a non-invasive manner. In this study, we aimed to determine how effective 3D-SBCSI is in detecting regional lung function and structural changes in patients with IPF by correlating those findings with spirometry tests and HRCT scores.
2. Materials and Methods
This study was performed under a protocol approved by the Institutional Review Board and an investigational new drug application for Xe-129 MRI. Written informed consent was obtained from all participants. A total of 20 subjects, including 9 healthy (29 yo ± 9.2; FEV1pred
= 99 ± 9.9%; FVCpred
= 104 ± 10.6%) and 11 who met diagnostic criteria for IPF (65 yo ± 12.5; FEV1pred
= 63 ± 13.9%; FVCpred
= 63 ± 15.2%) underwent HP Xe-129 ventilation MRI and 3D-SBCSI (Table 1
). Spirometry was performed on all subjects before imaging. Diffusion capacity for carbon monoxide (DLCO) and hematocrit were measured only in IPF subjects after imaging.
All imaging protocols were performed on a 1.5 T clinical MRI system (Avanto, Siemens Medical Solutions, USA) using a transmit/receive RF coil (Clinical MR Solutions, WI) tuned to the Xe-129 frequency. Subjects were placed supine on the MR table with the coil strapped to their chest. 3D proton localizer MR images were used to position the subject’s lungs at the isocenter. A test dose of xenon was well tolerated by all subjects, and there were no signs of any adverse reactions. Isotopically enriched (~83%) Xe-129 was polarized to ~35% using a commercial Xe-129 polarizer. For each Xe-129 MRI acquisition, the subjects inhaled a gaseous mixture of HP Xe-129 (maximum volume of 1.0 L) and nitrogen for a total volume equal to one-third of their measured FVC. Immediately after inhalation, the subjects were asked to hold their breath for the entire duration of the acquisition (<10 s). The 3D-SBCSI sequence parameters used were: TR/TE 13 ms/1.0 ms; FA 25° centered at 200 ppm of the gas frequency; vector size 512; weighted phase-encoding; BW 50 kHz; minimum voxel size 6.5 × 6.5 mm2; and 6–8 slices with 20–25 mm slice thickness.
H1-MRI images, covering the whole lungs with 17 slices and using a pulse sequence with a spiral K-space trajectory, were obtained during the same breath hold as HP Xe-129 Gradient Echo (GRE) images, from which we created a mask for quantifying relative ventilation in the GRE images. The intensity of the GRE images corresponds to the amount of HP Xe-129 detected in a particular voxel and can be used to measure relative ventilation in each voxel, segmenting the GRE images into four classifications: hyperventilation, normal ventilation, hypoventilation, and no ventilation. This analysis was completed using the Advanced Normalization Tools (ANTs) package, as described elsewhere [25
Following the segmentation of the GRE images, the volumes of three classifications were determined; the two lower ventilation classifications (no ventilation and hypoventilation) correspond to ventilation defects, while the higher ventilation classification indicates normal ventilation. These classification volumes were normalized using the sum of each of the classification volumes, the total measured volume of the lung, for each particular subject, calculating the percentage of ventilation defects and normal ventilation in the total volume of each subject’s lungs.
Post-processing of 3D-SBCSI data was completed with custom software written in MATLAB (MathWorks, Natick, MA, USA), which allowed the production of the desired spectrum for quantification. Traditionally, a peak in a spectrum is quantified by calculating the peak’s area under the curve by directly integrating the spectrum within specified boundaries. However, a spectrum with overlapping peaks such as the Xe-129 spectrum cannot be directly integrated and requires additional processing for accurate quantification [26
]. For this reason, the HP Xe-129 complex spectrum was fitted to a sum of complex Lorentzian functions. Each peak was individually modeled by a Lorentzian function that accounted for the peak’s amplitude, width, chemical shift, and phase [28
]. Subsequently, the area under the curve was calculated for each peak using its respective fitted parameters.
The quantification process was repeated for every acquired voxel to generate maps describing the relative amounts of xenon in the Gas, Tissue, and RBC compartments. To allow for more meaningful and normalized comparisons, Tissue/Gas, RBC/Gas, and Tissue/RBC ratio maps were calculated from the component maps. In addition to the ratio maps, the fitted parameters were also used to generate maps describing the chemical shift and T2* of the Tissue and RBC peaks.
HRCT evaluations were acquired only for IPF subjects because unnecessary exposure of healthy subjects to ionizing radiation was not ethically justifiable. HRCT images were evaluated for reticulation, traction bronchiectasis, ground-glass opacities, honeycombing, and emphysema at three different regions of each lung (above the aortic arch, between the aortic arch and pulmonary veins, and below the pulmonary veins), using a validated scoring system [30
]. A whole lung score for each radiographic feature was calculated by averaging of the six regional scores. Each HRCT was independently scored by two experienced radiologists, with a good agreement between the two (Krippendorf’s alpha = 0.80).
Statistical analysis was performed for each 3D-SBCSI parameter on a whole lung and on an individual slice level. The Mann–Whitney U-test was used to determine significance between healthy and IPF subjects. A two-tailed Pearson correlation was used to determine if there was a statistically significant correlation between 3D-SBCSI, pulmonary function tests (PFT), and HRCT.