¹²⁹Xenon Gas Exchange Magnetic Resonance Imaging as a Potential Prognostic Marker for Progression of Idiopathic Pulmonary Fibrosis

To the Editor:

Idiopathic pulmonary fibrosis (IPF) is a chronic and progressive pulmonary scarring disorder in which individuals exhibit distinct clinical trajectories (1, 2). Currently, these trajectories are defined retrospectively, with disease progression being assessed by clinical symptoms and pulmonary function testing (PFT) (3–5). It is desirable to identify robust, specific, and sensitive methods to assess IPF prognosis. This has led to development of clinical, physiological, genetic, and molecular predictors (3, 4). However, these fail to capture regional disease variation, a task better suited to three-dimensional (3D) imaging. To this end, a computed tomography (CT)-based structural analysis of several radiological factors, including pulmonary vessel volume, has been associated with IPF prognosis (6).

Alternatively, 3D measurement of pulmonary function is now possible using hyperpolarized xenon-129 gas (¹²⁹Xe) magnetic resonance imaging (MRI) (7). This enables regional quantitative mapping of gas distribution in the alveolar space, its uptake in interstitial barrier tissues (“barrier uptake”), and its transfer to red blood cells (“RBC transfer”) (8). Application of ¹²⁹Xe MRI to subjects with IPF identified regional enhancement of barrier uptake and focal defects in RBC transfer (9). To further these observations, in the present hypothesis-generating pilot study, we used the heterogeneity of ¹²⁹Xe imaging features from this previously described cohort to investigate how they relate to patient-specific clinical outcomes.

The 12 subjects with IPF, previously described (9) and diagnosed by multidisciplinary consensus criteria (2), together with 13 healthy control subjects, underwent ¹²⁹Xe MRI (Table 1). All subjects with IPF underwent baseline PFT and high-resolution CT scanning scored by a chest radiologist (10). After ¹²⁹Xe MRI, subjects received clinical care as directed by their pulmonary provider. Prospective clinical data were collected for up to 36 months, including outcomes such as lung transplant or death.

Table 1.

Demographics, pulmonary function test results, and xenon-129 gas metrics for healthy subjects (n = 13) and subjects with idiopathic pulmonary fibrosis (n = 12)

	Healthy	Group 1	Group 2	Group 3
Subject demographics
n	13	2	5	5
Males/females, n (%)	9/4 (69%/31%)	1/1 (50%/50%)	5/0 (100%/0%)	4/1 (80%/20%)
Age, yr	33.6 ± 15.7	64.5 ± 4.5	67 ± 3	67 ± 5
Diagnosis confirmed via CT/via biopsy	N/A	1/1 (50%/50%)	2/3 (40%/60%)	1/4 (20%/80%)
Pulmonary function tests
FVC, % predicted	94 ± 15	86 ± 6	62 ± 12	59 ± 18
DlCO, % predicted	93 ± 17	61 ± 9	53 ± 6	38 ± 5
Xenon-129 gas MRI metrics
BarrierHigh, %	4 ± 9	6 ± 5	49 ± 15	76 ± 47
RBCLow, %	24 ± 9	62 ± 20	35 ± 11	61 ± 10
RBC-to-barrier ratio	0.56 ± 0.13	0.18 ± 0.03	0.20 ± 0.03	0.13 ± 0.003
CT metrics
Radiographic pattern	—	U (1), A (1)	P (4), U (1)	P (3), U (2)
Subjects with emphysema	—	1	0	1
CT fibrosis score	—	12 ± 3	17 ± 15	15 ± 15
Combined metrics
GAP score	—	2 ± 1	4 ± 1	5 ± 2
Outcomes of patients with IPF
Median/mean follow-up, mo	—	36/36	36/28	33/33
Transplants/deaths	—	0/0	0/0	2/2 (80%)
Median transplant-free survival, mo	—	N/A	N/A	21.2

Definition of abbreviations: Barrier_High = functionally high barrier uptake; CT = computed tomography; Dl_CO = diffusing capacity of the lung for carbon monoxide; FVC = forced vital capacity; GAP = gender, age, and physiology; IPF = idiopathic pulmonary fibrosis; MRI = magnetic resonance imaging; RBC = red blood cell; RBC_Low = functionally low red blood cell transfer.

Values listed for the healthy group are in the format: mean ± standard deviation. Values listed for IPF groups are in the format: median ± interquartile range. For radiographic pattern, we used the American Thoracic Society criteria for usual interstitial pneumonia (UIP): P = probable UIP; U = UIP pattern, A = alternative diagnosis (diagnosed as IPF by surgical lung biopsy with UIP pattern and multidisciplinary consensus). The subjects with IPF are separated into xenon-129 groupings and include CT fibrosis scores, GAP scores, and clinical outcomes.

Anatomical ¹H images of the thoracic cavity and functional ¹²⁹Xe images of ventilation, barrier uptake, and RBC transfer were acquired using a 1.5-T scanner with multinuclear capabilities (GE EXCITE 15M4; GE Healthcare). Functional image voxels were binned according to thresholds derived from a healthy reference cohort (8). Our investigational biomarkers, Barrier_High and RBC_Low, were defined as the percentages of lung volume in which barrier uptake and RBC transfer were in the highest and lowest two bins, respectively. The heterogeneity of Barrier_High and RBC_Low across the IPF cohort enabled the classification of subjects into distinct groups. If Barrier_High or RBC_Low values exceeded the mean values of the healthy reference cohort by more than 2 standard deviations, they were categorized as abnormal. Using this definition, we identified three groups: group 1 with normal Barrier_High but abnormal RBC_Low, group 2 with abnormal Barrier_High and normal RBC_Low, and group 3 with combined abnormal Barrier_High and RBC_Low (Figure 1).

Figure 1.

Representative barrier uptake and red blood cell (RBC) transfer measured by xenon-129 gas magnetic resonance imaging (ventilation images not shown). (A) Healthy volunteer demonstrating normal amounts of barrier uptake and RBC transfer. (B) Group 1 subject with idiopathic pulmonary fibrosis (IPF) showing normal barrier but an abnormally large relative volume of lung with functionally low RBC transfer (RBC_Low). (C) Group 2 subject with IPF demonstrating an abnormally large relative volume of lung with functionally high barrier uptake (Barrier_High) but normal RBC transfer. (D) Group 3 subject with IPF with larger-than-normal volumes of both Barrier_High and RBC_Low.

Differences in baseline clinical metrics among groups were tested for significance using analysis of variance (ANOVA), followed by Tukey-Kramer honestly significant difference post hoc tests (applied when P < 0.1) to evaluate differences between groups. Transplant-free survival analysis was performed between subject groups by using the Kaplan-Meier method and Wilcoxon test. Reported statistical confidence intervals (CIs) are at 95%.

Interestingly, cumulative prospective data revealed different patterns of disease progression between ¹²⁹Xe MRI–defined groups. Subjects with both abnormal barrier uptake and abnormal RBC transfer (group 3) were the only group to progress to death (n = 2) or require lung transplant (n = 2). Median transplant-free survival in group 3 (21.2 mo) was less than in group 2 (no deaths/transplants) (Wilcoxon; P = 0.05).

We compared ¹²⁹Xe MRI groupings with baseline clinical, physiological, and radiographic scoring systems (Table 1). The average subject age was similar between groups (Table 1). We did not detect differences in forced vital capacity (FVC) across all groups (ANOVA; P = 0.06), and post hoc testing between each pair did not reveal differences in mean FVC between group 1 (mean, 86%; 95% CI, 68–104%), group 2 (61%; 49–72%), and group 3 (62%; 51–73%). The diffusing capacity of the lung for carbon monoxide (Dl_CO) values were different across IPF groups (ANOVA; P = 0.03). Post hoc testing revealed differences between the mean Dl_CO in group 3 (40%; 32–48%) and group 1 (61%; 49–73%) (P = 0.02; 95% CI for difference, 3–37%), but group 2 (51%; 44–59%) was not distinguishable from either of the other two groups. Notably, the average Dl_CO of group 3 is the lowest of all three groups (Table 1, Figure 2). CT fibrosis (ANOVA; P = 0.84) or GAP (gender, age, and physiology) scores (ANOVA; P = 0.06) were not different across groups (Table 1 and Figure 2). Subdivision and post hoc analysis of GAP scores did not identify differences between group 1 (2.0; 0.1–3.9), group 2 (4.6; 3.4–5.8), and group 3 (4.4; 3.2–5.6).

Figure 2.

Subjects with idiopathic pulmonary fibrosis (IPF) in each group at the time of their initial scan. (A–C) Forced vital capacity (FVC), diffusing capacity of the lung for carbon monoxide (Dl_CO), and computed tomography (CT) fibrosis score. (D) Xenon-129 gas magnetic resonance imaging groupings by classification of normal/abnormal functionally high barrier uptake (Barrier_High) and functionally low red blood cell transfer (RBC_Low). (E) RBC-to-barrier ratio.

FVC decline over 12 months could be assessed for 10 of 12 subjects with IPF (two of the group 2 subjects have longitudinal outcome data but insufficient follow-up PFT). No group 1 subjects exhibited greater than 10% FVC decline. For the three group 2 subjects with FVC data available, all had FVC declines greater than 10%, but all were transplant-free survivors. Three of the five group 3 subjects had FVC declines greater than 10%, with only one of these remaining classified as a transplant-free survivor after 3 years; of the four group 3 subjects with IPF who progressed to transplant or death, two had FVC decline greater than 10% and two had FVC decline less than 10%.

Our hypothesis-generating pilot study presents preliminary evidence that ¹²⁹Xe MRI features may prospectively classify patients with IPF. Most notably, we identified that subjects exhibiting both abnormal barrier uptake and abnormal RBC transfer (group 3) had a higher likelihood of death or lung transplant than subjects with abnormal barrier uptake alone. Neither death nor transplant was observed in the abnormal barrier (group 2) or abnormal RBC transfer (group 1) groupings. Notably, these MRI-derived groupings appeared distinct from FVC, Dl_CO, CT fibrosis scores, or GAP index; these traditional metrics, with the exception of group 1 FVC, all demonstrated overlapping values across the three groups (Figure 2). Furthermore, differences between groups were not identified in terms of radiographic usual interstitial pneumonia pattern or emphysema (Table 1). Although this may be attributable to the small sample size, it suggests that visualizing the spatial extent of functional abnormalities present at the alveolar–capillary interface could provide greater predictive power than traditional metrics.

By imaging more than just tissue density, ¹²⁹Xe MRI adds functional information over standard CT scans (11). To this end, Weatherley and colleagues recently demonstrated, in a small IPF cohort, that whole-lung ¹²⁹Xe MR spectroscopic indices (the RBC-to-barrier signal ratio) worsened over time in patients with IPF, despite unchanged FVC or Dl_CO measures (12). Although this ¹²⁹Xe spectroscopic metric is not spatially resolved, it benefits from little sensitivity to the patient breathing maneuver and ability to repeat measures every 10–20 milliseconds (13). Our present study builds on this work by separating and spatially resolving the ¹²⁹Xe uptake in barrier tissues and transfer to RBCs and by associating these measures with clinical outcomes. Imaging identifies distinct regions of abnormally high uptake of ¹²⁹Xe in barrier tissues and regions of reduced RBC transfer, which we hypothesize to indicate areas of active tissue fibrosis (see Figure 5 in Wang and colleagues [9]). The ability to quantify each compartment by its percentage of abnormal volume provides a potential marker of regional disease burden and/or activity.

We acknowledge limitations of this hypothesis-generating study, including a small IPF cohort, nonstandardized clinical follow-up, and age differences between the IPF and healthy cohorts. Our observations require replication in larger, controlled, prospective studies using rigorous outcome intervals and measures. Further study of healthy subjects will also determine whether the thresholds of Barrier_High and RBC_Low should be adjusted for age and sex. Despite these limitations, our findings warrant larger prospective studies of these ¹²⁹Xe biomarkers while seeking to better define their relationship with disease pathology/activity.