Dynamic ¹⁹F-MRI of pulmonary ventilation in lung transplant recipients with and without chronic lung allograft dysfunction

Abstract

Background

By the time chronic lung allograft dysfunction (CLAD), with its main phenotypes bronchiolitis obliterans syndrome (BOS) and restrictive allograft syndrome (RAS), is diagnosed by pulmonary function testing, irreversible damage to the lung allograft may already have occurred. Dynamic ¹⁹F-MRI of inhaled perfluoropropane may detect subtle changes in regional lung ventilation and provides a quantitative measure of regional lung function. We assessed feasibility of detecting regional ventilation dysfunction due to CLAD in lung transplant recipients.

Methods

Dynamic ¹⁹F-MRI was performed in ten lung transplant recipients, four without CLAD and six with CLAD (5 BOS, 1 RAS). Gas wash-in and washout dynamics were assessed and regional lung clearance index (RLCI) provided a quantitative metric of regional lung ventilation.

Results

BOS patients had substantially greater variation in regional ventilation compared with stable patients, with more regions of reduced ventilation, especially in the periphery. Tracer washout was homogeneous and rapid in stable patients but highly heterogeneous in CLAD. CLAD patients exhibited significant difference in RLCI between central and peripheral lung regions (p = 0.0016) and a wider interquartile range of RLCI for wash-in compared with stable patients (no CLAD 4.1, BOS 10.5, p = 0.036). FEV₁ (% of baseline) negatively correlated with ventilation during wash-in, most strongly for the periphery (r = −0.844, p = 0.0021).

Conclusions

Dynamic ¹⁹F-MRI identified quantifiable differences in regional ventilation in lung transplant recipients with and without CLAD and was well tolerated. Larger longitudinal studies using this approach will determine if early detection of changes in regional ventilation in lung transplant patients allows earlier CLAD detection.

Background

Chronic lung allograft dysfunction (CLAD) is an umbrella term for chronic rejection after lung transplantation and occurs in approximately 50% of lung transplant recipients within five years of transplantation.¹ CLAD has two main clinical phenotypes, bronchiolitis obliterans syndrome (BOS) and restrictive allograft syndrome (RAS), characterised by airway and parenchymal fibrosis, respectively.² To diagnose CLAD as soon as possible after its onset, frequent monitoring is currently performed with pulmonary function tests and chest x-rays. However, a significant amount of lung damage has already occurred by the time lung function begins to decline, and is often irreversible.² Furthermore, routine follow-up with standard chest x-ray is insensitive to small changes. Chest computed tomography (CT) provides improved sensitivity and an ability to detect additional diagnostically useful structural change, such as air trapping between inspiratory and expiratory views or pleuroparenchymal abnormalities.³ However, chest CT scans bring significantly increased exposure to ionising radiation and early CT findings of CLAD may be nonspecific.⁴ Because of these limitations, better techniques are needed to diagnose CLAD earlier. Current imaging techniques are under continuous development to offer more accurate and comprehensive information about lung function, including regional lung ventilation.

Fluorine (¹⁹F) MRI of inhaled perfluoropropane (PFP; C₃F₈) is a promising new technique for ventilation imaging and offers a scalable alternative to hyperpolarised ³He or ¹²⁹Xe gas MRI, which requires additional expertise and hardware for on-site preparation of the tracer gas.⁵ The majority of tracer gas ventilation MRI are performed as a static imaging technique, where scans are acquired during breath-hold after standardised tracer wash-in breathing manoeuvres,6, 7, 8, 9, 10 with image thresholding used to identify regions of low or no signal as ventilation defects. For ¹⁹F-MRI of inhaled PFP this approach has been shown to be reproducible¹¹ and sensitive to multiple disease states,7, 12 including asthma and chronic obstructive pulmonary disease. However, regions defined as unventilated or ventilated are dependent on the number and depth of wash-in breaths and the threshold chosen during image analysis.¹² This approach therefore presents challenges in differentiating subtle differences in pathophysiological ventilation states. Dynamic ¹⁹F-MRI multi-breath wash-in methods have been tested in animal models of disease,13, 14 and in human studies of patients with cystic fibrosis15, 16 and chronic obstructive pulmonary disease.¹⁷

The aim of this feasibility study was to demonstrate that ¹⁹F-MRI imaging can detect regional lung ventilation dysfunction in lung transplant recipients with CLAD compared to stable allograft recipients, using multi-breath wash-in and washout of inhaled PFP.

Materials and methods

Study design

This single-centre, prospective study was approved by the NHS Health Research Authority (ref 22/EM/0075). Lung transplant recipients attending routine appointments between May 2022 and July 2023 at the Freeman Hospital, Newcastle upon Tyne Hospitals NHS Foundation Trust, who met study inclusion criteria were invited to participate. Bilateral lung or heart-lung transplant recipients ≥18 years of age with a body weight within the minimum and maximum scanner requirements of 50–100 kg were eligible. Patients were excluded if they were pregnant or lactating, had contraindications to MRI (i.e., incompatible implanted medical device or incompatible metallic implants), or had a body habitus incompatible with positioning within the MRI scanner sensor used for ¹⁹F-MRI. CLAD stage 4 patients and patients with an acute respiratory infection were also excluded. All patients provided written informed consent.

CLAD diagnosis and staging

CLAD diagnosis was made according to the most recent consensus document of the International Society for Heart and Lung Transplantation.² In our centre, patients are primarily monitored using spirometry. Since monitoring of total lung capacity is not routinely performed, forced vital capacity (FVC) was used as a surrogate marker to monitor for restriction. CLAD diagnosis and staging (0−4) were determined by experienced transplant physicians based on a persistent decline in forced expiratory volume in one second (FEV₁) of ≥20% from baseline in the absence of an alternative cause. Chest CT scans with inspiratory and expiratory phases are usually performed at the time of CLAD diagnosis for CLAD phenotyping and exclusion of other possible causes. CLAD phenotypes were determined according to the consensus document and staged as follows: stage 1: 66–80%, stage 2: 51–65%, stage 3: 36–50%, and stage 4: ≤35% of FEV₁ baseline.²

Protocol

All participants attended for a single study session. All MR imaging was performed with a clinical 3.0T MRI scanner (Philips Achieva, Philips Healthcare, Guildford, UK) equipped with multinuclear scan capability. All images were acquired using a transmit/receive birdcage coil tuned to the resonant frequency of ¹⁹F at 3T, with off-resonance ¹H imaging capability (Rapid Biomedical GmbH, Germany). Parameters for each of the imaging acquisitions are summarised in Supplemental Materials: Table S1.

Subjects were positioned supine. A morphologic ¹H-MRI scan was acquired using a spoiled gradient-echo (SPGR) sequence during a breath-hold at full inspiration after inhalation of room air. ¹⁹F-MRI acquisitions were performed during breath-holds after deep inhalation of a normoxic mixture of 79% PFP and 21% oxygen (PFP/O₂) from a 25 L reservoir bag via tubing, an anti-bacterial/anti-viral filter, and a mouthpiece (Figure 1). A two-way valve permitted switching between supplying room air and PFP/O₂ to the participant, as required. Exhaled gas was collected in a 100 L reservoir bag via a non-rebreathing valve. A medically qualified doctor was present throughout the scan session. Participants breathed room air for five minutes between ¹⁹F-MRI acquisitions to facilitate complete PFP washout.

Figure 1.

MRI and breathing protocol design. (A) Set-up of MRI. (B) Graphical summary of the breathing protocol used during the dynamic multi-breath ¹⁹F-MRI acquisition. A 7.6-second 3D ¹⁹F-MRI acquisition was acquired during a breath-hold at maximum inhalation on alternate breaths while inhaling either a 79% perfluoropropane/21% oxygen gas mixture or room air. This resulted in up to 7 perfluoropropane wash-in acquisitions and up to 7 washout acquisitions. A total scan session takes approximately 20–30 min: 5–10 min for set-up, 5–6 min for unlocalised ¹⁹F spectroscopy scan to measure ¹⁹F resonance frequency, performed during a short breath-hold of PFP and immediately followed by a 5-minute delay to permit full gas washout, and 8–10 min for dynamic ¹⁹F wash-in and washout MRI. ¹⁹F-MRI: fluorine-19 magnetic resonance imaging, PFP/O2: 79% perfluoropropane/21% oxygen gas mixture.

An unlocalised ¹⁹F spectroscopy pulse-acquire scan was collected (TR = 200 ms, flip angle = 90°, 256 datapoints, 8 kHz acquisition bandwidth, 200 averages) to measure the in vivo ¹⁹F resonant frequency of PFP’s CF₃ moiety, which informed subsequent ¹⁹F acquisitions.

PFP wash-in and washout imaging was performed by repeated acquisition over multiple breath-holds of an accelerated ¹⁹F SPGR sequence with undersampled phase encoding and compressed sensing reconstruction.¹⁸ Scans duration was 7.6 seconds, with acquisition performed at breath-hold on alternate deep inhalations. After up to 14 breathing cycles (i.e., 7 breath-hold imaging acquisitions) with inhalation of PFP/O₂ gas mixture, the gas supply was switched to room air and ¹⁹F SPGR scan acquisitions continued on alternate breaths for a further 14 breathing cycles (Figure 1B).

Image analysis

All image processing was performed via semiautomated in-house scripts using MATLAB (version R2022b, MathWorks Inc., Natick MA, USA) and is detailed in Supplemental Methods. In brief, the static 3D ¹⁹F-MR images making up each 4D (multiple breath-hold) dataset were co-registered and separate mono-exponential growth and decay curves were fitted to the PFP signal amplitude within each voxel during wash-in and washout phases respectively.

From these, the Regional Lung Clearance Index (RLCI) was determined for each image voxel, where RLCI_in is defined as the number of breathing cycles required to increase the ¹⁹F-MRI signal to 39/40th (97.5%) of the theoretical maximum signal and RLCI_out is the number of breaths required to equivalently reduce the ¹⁹F-MRI signal. Regional analysis assessed median RLCI values measured in the lung apex and base, and lung periphery and centre (Fig. S1).

Statistical analysis

Statistical analyses were performed using SPSS (Version 29.0.1, IBM Corp., Armonk, NY). Results from continuous data are expressed as mean ( ± standard deviation) or median (25th percentile–75th percentile) of voxelwise values across the lung volume, where appropriate. The distribution of RLCI across the lung is depicted graphically using histograms and boxplots. Where sample size was sufficient, a two-tailed Wilcoxon rank-sum test was used to test for statistically significant difference in participant median and interquartile range between groups.

Correlation between spirometric tests (FEV₁, measured as a percentage of the patient’s baseline) and whole-lung and regional RLCI measurements were analysed using the Pearson correlation coefficient. Significance of correlations was tested using Fisher z transformation. All tests were two-sided, and p-values less than 0.05 were considered significant.

Findings on ¹⁹F-MR images were correlated with chest CT findings using a descriptive approach.

Results

Study population

Ten patients were included at a median time of 5.5 (2.1–9.7) years post-transplant. Median age at time of MRI was 54 (31−63) and 70% were male. Five participants (50%) had been diagnosed with CLAD with a BOS phenotype (three stage 1, one stage 2 and one stage 3) and one participant had CLAD with a RAS phenotype (stage 1). Median time since onset of CLAD was 2.4 (1.2–5.4) years. In addition, four lung transplant recipients with stable allograft function without a diagnosis of CLAD were included (Table 1). The study protocol was well tolerated by all participants. For one non-CLAD patient, PFP washout data were not acquired at the participant’s request due to some discomfort with breath-holding.

Table 1.

Patient Characteristics

Patient characteristics	All study patients (n = 10)
Sex (n, %)
Female	3 (30)
Male	7 (70)
Age (years)	54 [31–63]
Body mass index (kg/m²)	24 [23–26]
Indication for lung transplantation
Chronic obstructive pulmonary disease	2 (20)
Cystic fibrosis	3 (30)
Interstitial lung disease	4 (40)
Pulmonary hypertension	1 (10)
CLAD diagnosis (n, %)
No CLAD	4 (40)
BOS	5 (50)
Stage 1	3 (30)
Stage 2	1 (10)
Stage 3	1 (10)
RAS	1 (10)
Stage 1	1 (10)
Time since transplant (years)	5.5 [2.1−9.7]
CLAD onset after transplant (years)	3.5 [1.8−8.1]
Time from CLAD onset to MRI (years)	2.4 [1.2−5.4]

BOS: bronchiolitis obliterans syndrome, CLAD: chronic lung allograft dysfunction, MRI: magnetic resonance imaging, RAS: restrictive allograft syndrome.

Characteristics of all patients included in the study. Continuous data are presented as median [25th percentile – 75th percentile].

Wash-in dynamics

¹⁹F-MR images from five participants taken during PFP wash-in are shown in Figure 2, displayed adjacent to corresponding RLCI_in maps calculated from these MRI datasets. ¹⁹F-MRI wash-in datasets from all participants are presented in Supplemental Fig. S2.

Figure 2.

Wash-in dynamics. 2D coronal slices through each of the dynamic 3D ¹⁹F-MRI wash-in acquisitions collected from five representative participants (displayed in greyscale) adjacent to the calculated RLCI_in map of the same slice, where a darker colour represents a higher RLCI value (i.e., slower wash-in, requiring more breaths to reach full gas wash-in). ¹⁹F-MRI: fluorine-19 magnetic resonance imaging, BOS: bronchiolitis obliterans syndrome, CLAD: chronic lung allograft dysfunction, RAS: restrictive allograft syndrome, RLCI_in: wash-in regional lung clearance index.

Median voxelwise RLCI_in was 8.7 (6.0–14.1) in participants with no CLAD and 9.5 (3.6–24.2) in participants with CLAD (p = 0.914). Median RLCI_in in participants with BOS was 10.2 (4.2–25.5), which was not significantly different to that of non-CLAD patients (p = 0.556). Median RLCI_in in the single participant with RAS was 3.1 (1.9–8.8); this difference could not be tested for statistical significance. The interquartile range of RLCI_in was statistically significantly wider in participants with BOS (IQR = 21.3) compared to patients without CLAD (IQR = 8.1) (W = 11, p = 0.032), representing increased heterogeneity in wash-in rates in participants with CLAD.

Box and whisker diagrams displaying the median and IQR of voxel-wise RLCI_in values for each patient group are shown in Figures 3A and 3B.

Figure 3.

Wash-in and washout RLCI values. Group sizes were adequate for statistical testing only in the case of no CLAD vs. BOS, where no significant difference was found when comparing median values, but a significant change was detected during both wash-in and wash-out when comparing the interquartile range (* = p < 0.005). (A) Box plots displaying the range of wash-in RLCI values from the 3D maps from all ten participants: four participants without CLAD, three participants with BOS 1, one participant with BOS 2, one participant with BOS 3, and one participant with RAS 1. (B) Box plots displaying the same wash-in RLCI values, grouping the five participants with BOS. (C) Box plots displaying the range of washout RLCI measurements made in nine participants: three participants without CLAD (washout data were not available in one participant without CLAD), three participants with BOS 1, one participant with BOS 2, one participant with BOS 3, and one participant with RAS 1. (D) Box plots displaying the same washout RLCI values, grouping the five participants with BOS. BOS: bronchiolitis obliterans syndrome, CLAD: chronic lung allograft syndrome, RAS: restrictive allograft syndrome, RLCI: regional lung clearance index.

The ratio of RLCI_in measured in the centre of the lungs to RLCI_in measured in the periphery of the lung volume was 0.77 (± 0.07) in the participants without CLAD compared with 0.44 (± 0.16) in the CLAD group (W = 34, p = 0.010), and 0.43 (± 0.17) in the participants with BOS (W = 30, p = 0.016). This suggests slower relative gas wash-in rates to the periphery of the pulmonary volume in patients with CLAD relative to participants without CLAD. The participant with RAS had a centre-to-periphery RLCI_in ratio of 0.54, which is greater than three standard deviations below the value measured in participants without CLAD. However, due to the sample size of one, this difference could not be tested for significance.

The ratio of RLCI_in measured in the apex of the lungs to RLCI_in measured in the base of the lung volume was 0.82 (± 0.16) in the participants without CLAD, compared to 0.71 (± 0.23) in the CLAD group and 0.70 (± 0.26) in the BOS group. These changes were not significant (W = 25, p = 0.610; W = 21, p = 0.556, respectively). An apex-base RLCI_in of 0.80 was measured in the participant with RAS, within one standard deviation of the mean measured in the participants without CLAD.

Washout dynamics

¹⁹F-MR images displaying wash-out dynamics in five representative participants are displayed in Figure 4. ¹⁹F-MRI washout datasets from all participants are displayed in Supplemental Materials: S3. Representative examples of full lung washout RCLI maps for a patient without CLAD, a patient with BOS stage 1 and a patient with BOS stage 3 are presented in Figure 5. Increased heterogeneity in RLCI_out values is visible in patients with CLAD, with substantial variation in regional ventilation between participants.

Figure 4.

Representative washout ventilation images. RLCI maps depicting washout RLCI across the 3D lung volume (top: coronal plane, base: axial plane) in three participants: one representative participant without CLAD, one representative participant with BOS stage 1, and one participant with BOS stage 3. RLCI values (colour) are overlaid on conventional anatomical ¹H MRI acquisitions (greyscale). Higher RLCI values (corresponding to slower gas wash-out rates) are present in the patients with BOS, particularly around the lung periphery. Histograms plotting the RLCI measured in every voxel display the increasingly positive skew towards higher RLCI values with worsening BOS severity. Median RLCI_out (solid line) and 25th and 75th percentiles (dashed lines) are indicated in red. BOS: bronchiolitis obliterans syndrome, CLAD: chronic lung allograft dysfunction, RLCI: regional lung clearance index, ¹H-MRI: hydrogen-1 magnetic resonance imaging.

Figure 5.

Washout dynamics. Unprocessed images of all dynamic ¹⁹F-MRI washout acquisitions collected from five representative participants. A 2D slice through the original washout datasets from the participants is displayed in greyscale, adjacent to the calculated RLCI_out map of the same slice, where a darker colour represents a higher RLCI value (i.e., slower washout, requiring more breaths to reach full gas washout). ¹⁹F-MRI: fluorine-19 magnetic resonance imaging, BOS: bronchiolitis obliterans syndrome, CLAD: chronic lung allograft dysfunction, RAS: restrictive allograft syndrome, RLCI: regional lung clearance index.

Median RLCI_out was 4.8 (3.0–7.1) in the non-CLAD group versus 8.7 (5.0–15.5) in participants with CLAD (p = 0.095). Median RLCI_out was 9.2 (5.1–16.3) in participants with BOS, which was not statistically significantly different compared to the participants without CLAD (p = 0.071). Median RLCI_out was 6.0 (4.4–8.0) in the participant with RAS.

The interquartile range of RLCI_out was statistically significantly wider in participants with BOS (IQR = 10.5) compared to patients without CLAD (IQR = 4.1) (W = 6, p = 0.036), representing increased heterogeneity in wash-out rates in BOS patients.

Box and whisker diagrams displaying the spread of voxel-wise RLCI_out values for each patient group are displayed in Figures 3C and 3D. The range of RLCI_out was significantly higher in BOS patients.

Correlations with spirometric indices

A strong negative correlation was detected between the interquartile range of peripheral ventilation during wash-in and the spirometric measure FEV₁ (measured as % of baseline) (r = −0.844, p = 0.002), retaining statistical significance after Bonferroni correction (p < 0.05) (Table 2).

Table 2.

Correlations with FEV₁

Variable	Whole lung	Centre	Periphery	Apex	Base
RLCIin	−0.136	0.288	−0.618	−0.264	−0.128
RLCIout	−0.382	−0.224	−0.552	−0.436	−0.430
IQRin	−0.797†	−0.286	−0.844*	−0.794†	−0.653†
IQRout	−0.556	−0.417	−0.606	−0.770†	−0.369

FEV1: forced expiratory volume in one second, RLCIin: regional lung clearance index calculated during PFP wash-in, RLCIout: regional lung clearance index calculated during PFP wash-out, IQRin: interquartile range of the regional lung clearance index calculated during PFP wash-in, IQRout: interquartile range of the regional lung clearance index calculated during PFP wash-out. Correlation between FEV₁ (% baseline) and whole-lung and regional-specific averages of RLCI_in/out and the IQR of the RLCI_in/out. r values for Pearson correlations are displayed. ^†: statistically significant correlations prior to Bonferroni correction. * : statistically significant following Bonferroni correction. The strongest correlations were seen between FEV₁ (% baseline) and the IQR during wash-in (especially peripherally, but also for the whole lung, apex and base).

Corresponding CT images

Post-transplant CT images acquired around the time of MRI (67 ± 92 days) were available for three participants and are displayed adjacent to ¹⁹F-MRI RLCI maps in Figure 6. In the BOS 1 and BOS 3 patient, chest CT showed minor mosaic attenuation and bronchiolectasis, respectively, while ¹⁹F-MRI RLCI_out maps illustrate quite pronounced ventilation defects in the apex of the lungs and the periphery, respectively. In the RAS patient, the CT and ¹⁹F-MRI findings are more consistent, with particularly a volume loss of the right lung due to fibrotic remodelling.

Figure 6.

Corresponding CT images. Clinical coronal CT scans taken 127 days before (BOS stage 1), 39 days before (BOS stage 3), and 113 days after (RAS stage 1) the respective ¹⁹F-MRI acquisitions. BOS 1: chest CT shows some minor mosaic attenuation, probably related to air trapping. No significant bronchial dilation or wall thickening. ¹⁹F-MRI RLCI_out shows ventilation defects (darker colours) predominantly in the apex of the lungs. BOS 3: chest CT shows bronchial dilation with some fine peribronchial nodularity and bronchiolectasis. ¹⁹F-MRI RLCI_out shows ventilation defects (darker colours) predominantly in the periphery of the lungs (right > left), and to a lesser extent also in both bases. RAS 1: chest CT shows parenchymal and subpleural fibrosis predominantly in the right lung with lung volume loss. ¹⁹F-MRI RLCI_out shows a smaller ventilated right lung due to the volume loss with some less pronounced ventilation defects in the apex (orange colour). ¹⁹F-MRI: fluorine-19 magnetic resonance imaging, BOS: bronchiolitis obliterans syndrome, CT: computed tomography, RAS: restrictive allograft syndrome, RLCI_out: washout regional lung clearance index.

Discussion

CLAD is an overarching clinical diagnosis based on objective physiological parameters and its diagnosis is currently based on a persistent decline in FEV₁, with or without a decrease in total lung capacity or forced vital capacity, that is not attributable to other causes. Both CLAD phenotypes require a patient to have an irreversible loss of more than 20% of post-transplant baseline to meet the criteria for diagnosis.² However, at the time of CLAD diagnosis, chronic inflammation has often already led to irreversible lung damage and fibrosis; significant allograft damage has already occurred by the time lung function begins to decline. Current methods in routine clinical use, such as chest x-ray and pulmonary function testing, have low sensitivity to detect early changes due to CLAD.² Similarly, molecular markers that are gradually being implemented into clinical practice lack specificity for allograft damage.¹⁹ Early diagnostic strategies that increase the likelihood of preserving lung function with rapid intervention are an important focus of current research efforts.

Our study presents the first application of dynamic ¹⁹F-MRI in lung transplant recipients with and without CLAD. We demonstrate that ¹⁹F-MRI with a multiple breath-hold dynamic approach is feasible and well tolerated even in patients with more severe CLAD, and provides good visualisation and quantification of ventilation defects. More specifically, we showed that ventilation of PFP into and out of the lungs was significantly more heterogeneous in BOS patients compared with lung transplant recipients without CLAD, as assessed by the IQR of the RLCI. Although not significantly different, most likely due to the small sample size, both the wash-in (10.2 vs 8.7) and the washout (9.2 vs 4.8) rates were slower in BOS patients compared with stable lung transplant recipients. This illustrates the substantial variation in regional ventilation and generally delayed ventilation in this population, with areas that were poorly or even not ventilated. In addition, we confirmed that peripheral ventilation of the lungs was significantly delayed in BOS patients, representing impaired ventilation due to small airways disease. Importantly, this delay in peripheral ventilation strongly correlated with FEV₁ (% baseline). Indeed, BOS is primarily a disease of the small airways (obliterative bronchiolitis) characterised by airflow limitation,² and our findings are consistent with functional small airways disease. The fact that there was no difference between apical and basal ventilation can probably be explained by the diminished effect of gravity on lung ventilation in supine position.

RAS, on the other hand, is mainly characterised by pleuroparenchymal fibrosis.² Due to its lower prevalence, we were only able to recruit one patient with RAS, prohibiting statistical testing. However, we observed a substantially lower ratio of RLCI_centre/RLCI_periphery than measured in participants without CLAD, representing slower relative wash-in rates to the lung periphery, suggesting some degree of small airway disease, which has also been described to be present in the majority of RAS patients.²⁰

Interestingly, we were able to visualise ventilation defects in BOS stage 1 patients (see also representative images Figure 2, Figure 4, Figure 5, Figure 6), which appear different to gas wash-in and washout rates in lung transplant recipients without CLAD. This suggests that ¹⁹F-MRI could be a promising technique for the early detection of regional ventilation defects, supporting further longitudinal studies. Longitudinal data would confirm whether ¹⁹F-MRI is a useful tool for the early diagnosis of CLAD, especially BOS, before pulmonary function tests begin to decline. As ¹⁹F-MRI does not use ionising radiation, this technique offers an approach well suited to longitudinal imaging.

¹⁹F-MRI offers advantages over other pulmonary MRI techniques in active development. Gas hyperpolarisation, required for inhaled hyperpolarised ³He or ¹²⁹Xe-MRI requires substantial additional hardware and expertise. Tracer gas hyperpolarisation persists for only tens of seconds following inhalation due to T₁ relaxation and excitation (RF)-mediated polarisation loss. This limits collection of multi-breath ventilation datasets to a small number of breathing cycles,21, 22, 23 with substantial post-processing required to deconvolve ventilation-mediated dynamic change in signal amplitude from T₁- and RF-mediated polarisation loss. In contrast, thermally polarised PFP is used as supplied by a commercial medical gas vendor at patient consumption grade with normoxic oxygen content. The relationship between local PFP gas concentration and MR signal amplitude is linear and constant, and the favourable ¹⁹F-MR relaxation properties of PFP permit a short repetition time and a high degree of signal averaging that enables scan durations within tolerably short breath-holds. ¹⁹F-MRI directly measures ventilation via the evolving change in lung regional PFP content, unlike ¹H-MRI techniques that assess lung function that infer ventilation from change in regional tissue magnetic susceptibility properties and water content, secondary to ventilatory structural and microstructural change.24, 25, 26

Several limitations regarding ¹⁹F-MRI and our study need to be mentioned. First of all, the protocol is reliant on the participants’ ability to reproduce the breathing pattern and maintain similar inspiration levels between breaths. The ¹⁹F-MRI acquisition was halted early for one (non-CLAD) participant due to mild discomfort with the deep breathing protocol. Furthermore, the RLCI value varies with breathing depth, although standardisation of RLCI values between participants was achieved by implementing a breathing manoeuvre comprising maximum inhalations and maximum exhalations. Finally, our sample size was small and only one patient with RAS was recruited, limiting analysis and interpretation of findings for the RAS phenotype. In recognition of the limitations of statistical testing of small groups, graphical displays of the RLCI data are presented to aid visual interpretation of the spread of measured RLCI values between participant (sub-)groups. Our findings need to be verified in a larger cohort and longitudinal data are needed.

In conclusion, patients with CLAD, particularly BOS, showed significant ventilation defects contributing to delayed and reduced ventilation of the periphery of the lung and greater heterogeneity compared with lung transplant recipients without CLAD. ¹⁹F-MRI appears to be a promising tool for the early detection of regional ventilation defects in CLAD before lung function declines, which warrants further investigation in longitudinal studies.

Funding information

This study was supported by funding from the Rosetrees Trust (grant reference PGL22/100063).

Following authors are supported by a research fellowship outside of the submitted work:

• •SB is funded by the Paul Corris International Clinical Research Training Scholarship.
• •AJF is funded in part by the National Institute for Health Research Blood and Transplant Research Unit (NIHR BTRU) in Organ Donation and Transplantation at the University of Cambridge in collaboration with Newcastle University and in partnership with NHS Blood and Transplant (NHSBT). The views expressed are those of the authors and not necessarily those of the NIHR, NHS Blood and Transplant or the Department of Health and Social Care.

CRediT authorship contribution statement

MAN: Project conceptualisation, experimental design, data acquisition, data analysis and interpretation, manuscript writing, review of manuscript for important intellectual content and final approval of submitted version.

SB: Project conceptualisation, experimental design, data acquisition, manuscript writing, review of manuscript for important intellectual content and final approval of submitted version.

CWH: Project conceptualisation, experimental design, data acquisition, review of manuscript for important intellectual content and final approval of submitted version.

KGH: Experimental design, data analysis and interpretation, review of manuscript for important intellectual content and final approval of submitted version.

GM: Data acquisition, review of manuscript for important intellectual content and final approval of submitted version.

AN: Data acquisition, review of manuscript for important intellectual content and final approval of submitted version.

JLL: Data acquisition, review of manuscript for important intellectual content and final approval of submitted version.

AJF: Project conceptualisation, data acquisition, manuscript writing, review of manuscript for important intellectual content and final approval of submitted version.

PET: Project conceptualisation, experimental design, data interpretation, manuscript writing, review of manuscript for important intellectual content and final approval of submitted version.

Conflict of Interest

None of the authors of this manuscript have any conflicts of interest to disclose in relation to this manuscript. Preliminary data from the present work has been submitted in abstract form to the European Respiratory Society annual conference (Vienna, September 2024) and Lung Transplantation conference (Paris, September 2024).

Appendix A. Supplemental material

Supplemental material

.

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.