Regional Structure-Function in Cystic Fibrosis Lung Disease Using Hyperpolarized ¹²⁹Xe and Ultrashort Echo Magnetic Resonance Imaging

To the Editor:

Cystic fibrosis (CF) is a genetic disorder that exhibits a number of different structural pulmonary abnormalities such as mucus plugs (MP), bronchiectasis (BR), bronchial wall (BW) thickening, and consolidations (CNs), each of which contribute to abnormal ventilation via regional obstruction. Although structural imaging methods (generally X-ray computed tomography [CT]) can depict regional structural pathologies, the precise extent to which these structural abnormalities contribute to lung function decline is not well understood. Here, we demonstrate that hyperpolarized (HP) gas magnetic resonance imaging (MRI) can be combined with ultrashort echo (UTE) MRI (a radiation-free alternative to CT [1–3]) to quantify the relationships between individual regional pathologies and regional ventilation (4–6). The aim of this work was to quantify the size and extent of regional ventilation defects in CF using HP ¹²⁹Xe MRI and to associate these with the presence of specific structural abnormalities identified by UTE MRI: BR, BW thickening, MP, ground-glass opacities, and CN. We hypothesized that low-ventilation regions could be attributed to spatially matched pathologies seen in UTE images.

Methods

A total of five healthy control subjects (mean ± SD age, 10.8 ± 3.9 yr) and 22 clinically stable patients with CF (age, 14.5 ± 10.6 yr) were imaged under an Institutional Review Board–approved protocol, with a Food and Drug Administration Investigational New Drug (123577) for ¹²⁹Xe; informed consent was obtained from adult subjects or parents, and age-appropriate assent from pediatric subjects (demographic data in Table 1). Spirometry was obtained in each subject before imaging according to American Thoracic Society and European Respiratory Society guidelines. Subjects were imaged on a Philips 3T Achieva (Philips Healthcare) by coaching subjects to FRC before inhalation of a breath of HP ¹²⁹Xe (one-sixth TLC, based on height and sex [7]) for ventilation imaging via a gradient-echo sequence (two-dimensional multislice, TR/TE = 8/4 ms; voxel size = 3 × 3 × 15 mm³; readouts = 96–196; phase encodes = 50–64; FA = 9–12°; and slices = 9–15). UTE magnetic resonance images were acquired during resting breathing, gated to FRC, with a 32-channel proton cardiac phased-array coil (radial three-dimensional stack-of-stars sequence, 40,000 projections, TE = 0.2 ms, TR = 4.8–6.2 ms, FA = 5°, voxel size = 1.19–1.45 mm² in-plane resolution, slice thickness = 4 mm, and bandwidth ≈ 1.8 kHz). ¹²⁹Xe signal intensity was normalized to the whole-lung HP gas signal mean and lung voxels with ¹²⁹Xe signal <60% of the whole-lung signal were identified as defects (8, 9). The percentage of an individual’s lung volume identified as defect is quantified as the subject’s ventilation defect percentage (VDP). UTE magnetic resonance images were visually analyzed independently by two radiologists (A.S. and A.B.) for regions of structural abnormality: BR, BW thickening, MP, CN, and ground-glass opacities. The identified abnormalities were then visually matched to corresponding regions in the HP ¹²⁹Xe MRI to associate the identified structural abnormalities with corresponding ventilation defects (Figure 1). Radiologists also assigned an integer quality score between 1 and 5 (low and high, respectively) to each subject’s UTE dataset, which were averaged to provide an overall image quality score.

Table 1.

Individual Subject Demographic Data and Spirometry/VDP Results

Subject Number		Age (yr)	Sex	Height (cm)	CF Genotype	ppFEV1 (%)	VDP (%)
1		16	F	169	Control	110	0.8
2		12	M	160	Control	92	1.0
3		12	M	141	Control	103	1.5
4		6	F	116	Control	95	2.0
5		6	M	116.4	ΔΔF508	100	2.3
6		6	M	112	ΔΔF508	99	2.7
7		8	M	136	Control	91	2.8
8		7	M	109.9	ΔΔF508	116	3.9
9		6	F	105.2	ΔΔF508	97	5.4
10		11	F	149.9	ΔΔF508	119	6.1
11		6	M	121	ΔF508/R560T	81	8.6
12		11	F	147	ΔΔF508	86	8.8
13		11	F	152	ΔΔF508	89	9.0
14		13	M	146	3120+1G>A/3120+1G>A	78	9.2
15		12	M	158.6	ΔF508/W1282X	89	9.4
16		46	M	165	ΔΔF508	83	15.9
17		6	M	119	ΔΔF508	74	16.5
18		11	F	144.1	ΔΔF508	87	18.1
19		9	F	132.1	ΔΔF508	101	19.8
20		8	M	120.7	ΔΔF508	88	19.9
21		16	F	153	3849/849+10kbC>T	61	21.4
22		11	M	142	F508del/R1066H	102	22.8
23		16	F	159	ΔΔF508	72	23.9
24		19	F	163	ΔΔF508	66	25.6
25		26	F	157	ΔΔF508	78	26.8
26		26	M	175	F508del/G551D	39	33.8
27		37	F	151.5	F508del/S945L	38	44.0

Definition of abbreviations: CF = cystic fibrosis; ppFEV₁ = percent predicted FEV₁; VDP = ventilation defect percentage.

Figure 1.

Example of image analysis pipeline shown in a patient with cystic fibrosis (subject 22). (A and C) ¹²⁹Xe and ultrashort echo (UTE) magnetic resonance images are collected in the same imaging session, respectively. (B and D) Ventilation defects are quantified in the ¹²⁹Xe magnetic resonance imaging (B, defects colored cyan) and abnormalities are independently identified within the UTE by trained readers (D). (E) ¹²⁹Xe defects are separately associated with proximal or adjacent structural abnormalities for quantitative analysis.

Results

Table 1 shows individual results for FEV₁ and VDP. Average FEV₁ was 98.2% ± 8.1% for controls and 83.8% ± 20.6% for subjects with CF (P = 0.02). Average VDP was 1.6% ± 0.8% for control subjects and 16.1% ± 10.8% for patients with CF (P < 10⁻⁵). Of the total calculated ¹²⁹Xe defect volume across all subjects, only 50.6% was associated with a spatially matched structural abnormality from UTE MRI. Of the calculated ¹²⁹Xe defect volumes that were associated to structural abnormalities, 76.8% were associated with BR and/or BW thickening and 76.6% were associated with MP, with significant overlap in attribution (only 16.2% and 17.9% exclusively associated with BR/BW and MP, respectively). Individual subject counts of structural abnormalities correlated strongly with VDPs (Pearson r = 0.81; P < 10⁻⁶) but only moderately with FEV₁ (r = −0.68; P = 0.0001), as expected. Pearson correlation was r = −0.78 (P < 10⁻⁵) between subject VDP and FEV₁.

Discussion

This work presents the first direct quantification of ¹²⁹Xe ventilation impairment associated with specific regional structural abnormalities in CF lung disease. BR and MP were responsible for the vast majority of structurally attributed defects, but the large quantity of defects that could not be associated with structural abnormalities underscores the sensitivity of ¹²⁹Xe MRI to mild obstruction. The lower resolution inherent to UTE MRI (here, ∼1.2 mm) may have resulted in undetectable small structural abnormalities that would have been detected with high-resolution CT (0.4 mm). X-ray CT remains the gold standard for structural imaging, especially with respect to quantitative assessment of airways and reduced lung parenchymal density on expiration, associated with air trapping. However, this further highlights ¹²⁹Xe MRI as a sensitive technique for evaluating and quantifying regional lung function, particularly in cases in which structural imaging alone yields inconclusive results and in mild disease, in which functional declines are subclinical. This combination of MRI techniques may be beneficial in the future to aid clinical decision-making, particularly in evaluation of individual patient outcomes, in which repeated testing is useful, and as a biomarker for upcoming clinical trials. We find that ¹²⁹Xe MRI often shows ventilation impairment even in the absence of identifiable structural abnormalities, and that within defect regions, ¹²⁹Xe signal is greater if no structural abnormality could be regionally matched from structural imaging. This highlights ¹²⁹Xe imaging as a sensitive tool for pulmonary research and may be clinically useful in the evaluation and/or management of patients with CF in early stages of lung function decline.