Abstract
Introduction Xenon-129 (129Xe) ventilation magnetic resonance imaging (MRI) is sensitive to detect early cystic fibrosis (CF) lung disease and response to treatment. 129Xe-MRI could play a significant role in clinical trials and patient management. Here we present data on the repeatability of imaging measurements and their sensitivity to longitudinal change.
Methods 29 children and adults with CF and a range of disease severity were assessed twice, a median (interquartile range (IQR)) of 16.0 (14.4–19.5) months apart. Patients underwent 129Xe-MRI, lung clearance index (LCI), body plethysmography and spirometry at both visits. 11 patients repeated 129Xe-MRI in the same session to assess the within-visit repeatability. The ventilation defect percentage (VDP) was the primary metric calculated from 129Xe-MRI.
Results At baseline, mean±sd age was 23.0±11.1 years and forced expiratory volume in 1 s (FEV1) z-score was −2.2±2.0. Median (IQR) VDP was 9.5 (3.4–31.6)% and LCI was 9.0 (7.7–13.7). Within- and inter-visit repeatability of VDP was high. At 16 months there was no single trend of 129Xe-MRI disease progression. Visible 129Xe-MRI ventilation changes were common, which reflected changes in VDP. Based on the within-visit repeatability, a significant short-term change in VDP is >±1.6%. For longer-term follow-up, changes in VDP of up to ±7.7% can be expected, or ±4.1% for patients with normal FEV1. No patient had a significant change in FEV1; however, 59% had change in VDP >±1.6%. In patients with normal FEV1, there were significant changes in ventilation and in VDP.
Conclusions 129Xe-MRI is a highly effective method for assessing longitudinal lung disease in patients with CF. VDP has great potential as a sensitive clinical outcome measure of lung function and end-point for clinical trials.
Abstract
129Xe-MRI in CF is highly repeatable. In patients with normal FEV1, 129Xe-MRI is also sensitive to detect changes in longitudinal lung function and should be highly informative in an era of CFTR modulators and increasingly preserved FEV1 https://bit.ly/2C0D8Np
Introduction
For people with cystic fibrosis (CF), advances in treatments and patient management have significantly increased expected survival. These advances have greatly improved the lung health of patients, and now the median forced expiratory volume in 1 s (FEV1) for UK patients aged <18 years is well preserved at 88% predicted [1]. However, it is also well accepted that a value for FEV1 within the range of normal does not necessarily mean that the patient's lung function is truly normal [2, 3]. With highly effective cystic fibrosis transmembrane conductance regulator (CFTR) modulator therapies being increasingly administered to patients with an FEV1 within the normal range, more sensitive outcome methods for assessing lung function are required.
Hyperpolarised gas ventilation magnetic resonance imaging (MRI) using either helium-3 (3He) or xenon-129 (129Xe) provides a direct visual and quantitative assessment of the distribution of ventilation within the lung in three dimensions [4]. Ventilation abnormalities are clearly identified as areas of signal deficit and are termed ventilation defects. The ventilation heterogeneity seen on MRI can be quantified using different approaches to assess the degree of abnormality. The metric most widely used is the ventilation defect percentage (VDP), which quantifies the proportion of the image without any ventilation present. A major strength of ventilation MRI is the ability to measure individual ventilation defects, which allows for small regional changes in lung function to be assessed [5].
Previous studies in CF populations using 3He showed that ventilation MRI is highly sensitive to detect early lung disease in subjects with normal values for FEV1 [6–12], lung clearance index (LCI) [5, 6] and computed tomography imaging [6]. The latter two methods are already recognised as sensitive methods for the detection of early lung disease [3, 13]. A previous study of young patients with mild CF lung disease showed that 3He MRI was sensitive to longitudinal changes in lung function that were largely undetected by FEV1 and LCI [14]. In more recent years, ventilation MRI research has moved from 3He towards 129Xe due to the lower cost and greater availability of 129Xe. Recent studies utilising 129Xe MRI in CF have found that it is well tolerated [15] and is sensitive to detect early lung disease [16, 17]. In addition, 129Xe MRI has been shown to be sensitive to treatment response to pulmonary exacerbation in children with CF, with 129Xe VDP showing a larger treatment response than both LCI and FEV1 [18].
Hyperpolarised gas ventilation MRI is therefore an attractive method for assessing CF lung disease, but more data are required on the background longitudinal changes seen in stable CF lung disease. This includes a systematic assessment of the intrinsic technical repeatability of the measurement in this population, which preliminary data suggest is promising [19], as well as the pathophysiological variability seen in stable disease. Therefore, in this study we aimed to assess the potential of 129Xe MRI as a quantitative outcome measure of lung health and a possible candidate end-point for clinical trials and patient management. In order to do this, we assessed the short- and long-term repeatability of imaging (VDP) in a cohort of children and adults with CF and a range of lung disease. In addition, we aimed to better understand the longitudinal changes in lung function on 129Xe MRI in comparison to LCI and FEV1 in those patients with an FEV1 within the normal range.
Methods
This prospective study recruited adults and children (aged >5 years) with CF from three specialist CF centres in the UK (Sheffield Children's Hospital and Northern General Hospital, Sheffield, UK and Manchester Adult CF Centre, Manchester, UK). For inclusion, patients had to be clinically stable for 4 weeks prior to assessment (defined as free from intravenous antibiotics or any hospital stay within 4 weeks) and have an FEV1 >30% pred within the previous 6 months. This study was approved by the Yorkshire and Humber – Leeds West research ethics committee (16/YH/0339). Parents/guardians of children and all adult patients provided written informed consent.
Lung imaging
Hyperpolarised 129Xe ventilation MRI of the lungs was performed at a single site (Sheffield) on a 1.5 T GE HDx scanner (GE, Milwaukee, WI, USA) using previously described protocols [20]. 129Xe was polarised using a bespoke spin exchange optical pumping polariser [21], under a UK Medicines and Healthcare Products Regulatory Agency manufacturing specials regulatory licence (MS-18739). Ventilation imaging was acquired at a lung volume of end-inspiratory tidal volume, by inhaling a volume of 129Xe titrated with a balance of medical-grade nitrogen, from a lung volume of functional residual capacity. The total inhaled gas volume ranged from 0.4 to 1.0 L and was calculated based on the subject's height. For 129Xe analysis a hydrogen-1 (1H) anatomical image was performed in a separate breath-hold (immediately prior to the 129Xe image) in order to calculate the thoracic cavity volume. For quantitative analysis, the 1H and ventilation images were segmented using a semi-automated method [22], from which the VDP and the ventilation heterogeneity index (VHI), which reflects the heterogeneity of ventilated voxels within ventilated lung regions, were calculated as described previously [5]. Further details of image acquisition and processing methods can be found in the supplementary material.
In order to assess the within-visit technical repeatability of 129Xe-MRI, 129Xe imaging was repeated in a subgroup of patients within 15 min of the initial baseline measurement and without the subject leaving the scanner. The same imaging protocol, respiratory manoeuvres and volume of gas were used for both scans.
Lung physiology
Patients performed multiple-breath washout on the same day and at the same centre as imaging using an open-circuit Innocor gas analyser (Pulmotrace, Glamsberg, Denmark) using 0.2% sulfur hexafluoride [23]. LCI was calculated from the average of three trials as recommended [24]. Body plethysmography was performed using a PFT Pro (Vyaire, Basingoke, UK) according to guidelines [25], in order to calculate the ratio of residual volume (RV) to total lung capacity (TLC). Finally, spirometry was performed according to guidelines [26] and expressed as z-scores [27]. Either MRI or LCI was performed first, followed by the other. Spirometry was always performed last.
The assessments of 129Xe ventilation MRI, multiple-breath washout and spirometry were then repeated at a second stable visit using the methods described.
Statistical analysis
Data were analysed using Prism version 8.0 (GraphPad, San Diego, CA, USA) and SPSS statistics version 26.0 (IBM, Armonk, NY, USA). Normal distribution was assessed using the Shapiro–Wilk test. Data are expressed as mean±sd for normally distributed data, and median (interquartile range (IQR)) for nonparametric data. Within-visit repeatability of 129Xe-MRI was assessed from the 95% limits of agreement (LoA) of a Bland–Altman analysis of the repeat measurements. Wilcoxon signed-rank test and Bland–Altman analysis were used to compare MRI and lung function metrics between the baseline and follow-up study visits. Within- and inter-visit repeatability was calculated using the intraclass correlation coefficient (ICC). Spearman correlation analysis was used to compare the change in different metrics. Sample size power calculations were calculated for different effect sizes based on the longitudinal data [28]. Statistical significance was set at p<0.05.
In order to assess whether a clinically significant change in VDP had occurred over time, firstly the Bland–Altman LoA from the within-visit repeatability measurements was used as a minimal threshold. Secondly, a further threshold to represent a clinically significant change in absolute VDP was set at ±3%; this threshold represents the mean absolute change in 129Xe VDP in response to treatment of an exacerbation of CF lung disease [18]. In order to compare the significance of change in VDP, similar thresholds for short- and long-term repeatability were applied for LCI and FEV1. Repeatability of ±10% in LCI has been shown for healthy volunteers [23], while longitudinal changes of ±20% have been reported in clinically stable CF patients [29]. For FEV1, a within-patient longitudinal change of >±10% is deemed significant [30], while the short-term change of FEV1 in patients with CF is approximately ±5% [31, 32].
Results
29 children and adults with CF were assessed on two occasions, a median of 16 months apart. At baseline, patients were aged between 6 and 47 years. Baseline demographics, MRI metrics and lung function are detailed in table 1. All but one patient had visible ventilation defects present at both study visits. 14 (48%) patients had a normal FEV1 value (>−1.64 z-score) at both baseline and follow-up. Three (10%) patients at baseline, and seven (24%) at follow-up had normal LCI values.
Within-visit repeatability of 129Xe MRI
11 (35%) patients performed repeat 129Xe-MRI within 15 min of the baseline scan. Median (IQR) age 23.7 (17.7–33.2) years, baseline VDP 7.3 (2.5–30.8)%, LCI 8.3 (7.3–14.0), FEV1 −2.4 (−2.8–−0.5) z-score. There was no significant difference in VDP between scans and good repeatability with a bias of 0.2% and 95% LoA −1.4–1.8%. This represents the intrinsic technical repeatability of the measurement in vivo assuming no true change in underlying lung ventilation. Based on this analysis a threshold of absolute change in VDP of ±1.6% was used in part to assess 129Xe VDP longitudinal change. (Repeatability for VHI can be found in figure 5, alongside the Bland–Altman plots for VDP.) For VHI, there was again minimal bias (−0.6), with 95% LoA −2.5–1.2%. The within-visit ICC for VDP was excellent at 0.99 (95% CI 0.99–1.0), and was 0.96 (95% CI 0.84–0.99) for VHI.
Longitudinal change in 129Xe MRI
All 29 patients successfully repeated 129Xe MRI and lung function testing at a second visit, after a median (IQR) interval of 16.0 (14.4–19.5) months. There was no single pattern of disease progression in the cohort and no lung function or MRI metric demonstrated a statistically significant group change between visits. Instead, significant inter-subject variation was seen in the degree and direction of change in ventilation distribution on 129Xe MRI. For many patients there were clear and often large visible changes in the distribution of ventilation, independent of underlying disease severity (figures 1–3). Figure 1 shows eight example images from patients, all of whom had FEV1 in the normal range, where there was a change in the distribution of ventilation and in VDP, but without significant change in FEV1 or LCI (see also figures 2 and 3).
Longitudinal change relative to baseline for VDP, LCI and FEV1 are shown in figure 4. Overall, 17 (59%) patients had an increase (worsening) in VDP at follow-up, which correlated with the visual image analysis (figures 4 and 5). 17 (59%) patients had a change in 129Xe VDP of >±1.6%, while nine (31%) also had an absolute change in VDP >±3%. In comparison, 13 (45%) had a relative change in LCI >10% from baseline, but only two (7%) had a relative change >±20%. For FEV1, 10 (34%) had an absolute change in FEV1 of >±5% pred and no patients had an absolute change >±10%. Of the nine patients with a change in VDP >±3%, no patient had a corresponding significant change in LCI or FEV1. Of all the metrics, 129Xe VDP had the highest median relative change over time (8.2%).
Inter-visit repeatability
The inter-visit ICC for 129Xe VDP was excellent at 0.97, which was similar to FEV1 (0.98) and higher than LCI (0.95), RV/TLC (0.96) and 129Xe VHI (0.89) (table 1). The change in VDP tended to be larger for those with higher baseline VDP. When only patients with normal FEV1 were considered (and therefore with lower values for VDP) the 95% LoA fell from −6.9% to 8.5% for the whole cohort to −4.3% to 4.0% (n=14) (figure 5).
Correlation of the changes in metrics over time
The absolute or relative change in VDP was not correlated with absolute or relative change in FEV1, LCI or RV/TLC. In contrast, the absolute and the relative changes in VHI and LCI were significantly correlated with each other (r=0.68, p<0.001 and r=0.73, p<0.001, respectively). There was no correlation in the change in either FEV1 or RV/TLC with the other metrics. In addition, there was no relationship between the magnitude of change in VDP with age or underlying lung disease as measured at baseline.
Sample size power calculations
With a view to using VDP from 129Xe MRI as an intervention outcome marker, sample size calculations for four different effect sizes and three populations were derived. Effect sizes include the minimal change of 1.6% in VDP, the mean change of 3% seen with i.v. antibiotics and 5% and 10% change. Population mean±sd are taken from the whole-cohort data (representing a mixed CF population with a wide range of disease severity) and, separately, only those with normal-range FEV1. The results are presented in table 2. This emphasises the importance of the baseline variability and appropriate population selection, but also shows the low numbers that are potentially required to detect significant change. For example, a 3% change could be detected with a power of 90% in a study population of 11 patients with CF and normal FEV1.
Discussion
The data reported in this study are the first longitudinal assessment of patients with CF using 129Xe lung ventilation MRI. In this study we demonstrate that 1) 129Xe MRI VDP has high within-visit repeatability; 2) a qualitative and quantitative approach to image analysis is complementary in assessing CF lung disease; and 3) in patients with a preserved FEV1, as well as those with more advanced disease, VDP demonstrates changes in ventilation distribution in patients where FEV1 and LCI do not show significant change.
There is a growing body of evidence that ventilation MRI provides valuable and detailed insights into the underlying lung function of patients with CF that is not detected by other methods [5–12, 14, 16–18, 33]. Previous studies have shown that ventilation MRI is highly sensitive to early lung disease and, in the case of 3He, is repeatable [34, 35] and sensitive to disease progression [14]. The data presented in this study add to this evidence base by demonstrating that 129Xe MRI is also highly sensitive to detect longitudinal changes in CF lung disease. Unlike the previous study from our group [14], performed in children with mild CF lung disease using 3He MRI, here we did not see one single pattern of disease progression. Of the cohort reported here, 59% of patients had evidence of increased (worsening) VDP, while the remaining patients showed improvements in ventilation. This is not surprising given that this is a broad cohort of patients, in terms of age and disease severity, although there was no relationship between the magnitude of change in VDP with either age or disease severity. Ventilation defects caused by mucus obstruction will not necessarily remain stable. Thus, some visible defects will represent short-term reversible obstructions while others may be caused by underlying disease progression and airway narrowing. Figure 1 shows how in patients with normal FEV1, 129Xe MRI is able to detect early disease-related changes, as we previously reported with 3He MRI. Furthermore, figure 1 shows that changes in lung ventilation can be measured on 129Xe MRI that are not necessarily detected by LCI. This is a particularly important finding in the era of new and expensive CFTR-modulating therapies, where there is a need to be able to measure clinical response to therapies even in those with apparently normal lung function. 129Xe MRI VDP is a metric that may provide this detail, as has been shown previously for 3He MRI in the assessment of ivacaftor [33]. In addition, there is also increasing evidence for the application of 1H structural MRI in the clinical assessment of CF lung disease [36–39]. 1H MRI can be performed at the same visit as 129Xe MRI, allowing for the combined assessment of lung structure and function when measured together.
Inter-visit repeatability is affected by sources of both intrinsic (technical) and physiological variability, as well as true disease progression. For VDP, a change of >±1.6% is greater than the inter-subject intrinsic repeatability of the measurement, and potentially represents a lower threshold for significant change. For comparison, a median change of >±3% in VDP in response to i.v. antibiotics should represent a clinically significant degree of change [18]. The true threshold for clinically relevant change in VDP therefore probably lies between these limits, but cannot be determined more precisely from the data available. However, over longer time courses, CF patients have natural fluctuations in mucus plugging and symptoms, which are separate from underlying disease progression. We have shown that over 16 months, a change in VDP of up to ±7.7% is seen in CF patients considered to be clinically stable and without obvious disease progression by other lung function metrics; the change is less (at ±4.1%) in those with a preserved FEV1.
The findings that lung function metrics on average are unchanged with time are consistent with longitudinal analyses of patients with CF using LCI, where minimal longitudinal change was reported [40–43]. Our findings, in addition, highlight that patients with CF are as likely to improve clinically as they are to have deteriorating ventilation heterogeneity during observational follow-up. Despite this, figures 1 and 3–5 show that patients often had subclinical changes in VDP without significant change in FEV1 or LCI. It is likely that some of these changes we have seen in ventilation are transient and some are the precursor to exacerbation and potentially irreversible ventilation changes.
In this study we highlight how VDP can sensitively track changes in underlying lung function in patients with preserved FEV1, which correspond to visual changes on the ventilation images (figures 1 and 2). In order to assess this specific patient population in clinical trials, relatively large sample sizes are required to measure modest treatment effects when FEV1 is the primary outcome [44]. LCI has been used as an alternative outcome in more recent studies [45], which allows for smaller sample sizes in patients with normal FEV1. However, 129Xe VDP has high repeatability and low standard deviation and large effect sizes can be measured, which is reflected in the relatively small sample sizes required to measure the different reported effect sizes.
129Xe VDP is an attractive potential outcome measure/end-point in both clinical trials and clinical management. A strength of 129Xe MRI is not only that can we produce summary whole-lungs metrics like VDP that are more sensitive than LCI and FEV1 at detecting early lung disease, but the images themselves also contain more detailed regional functional information [5]. Therefore, it is possible to detect clinically relevant regional change even in the face of apparently unchanged lung physiology tests [5, 46]. This applies both to detecting disease progression in clinical practice over time courses like the one described in this study, as well as detecting much shorter term improvements due to therapeutic interventions [18, 33]. In order to generate quantitative regional metrics of lung physiology from ventilation MRI, future work should focus on reliable parameterisation of regional ventilation heterogeneity to further improve the clinical utility of ventilation MRI.
We recognise that hyperpolarised gas ventilation MRI is not currently available to many CF centres. Estimates of regional lung function may be acquired indirectly, without the use of inhaled contrast agents using time-resolved 1H MRI techniques [47, 48] or by using contrast enhanced perfusion MRI [38]. These techniques are promising and may provide an alternative, more widely accessible method to the wider CF community. A further limitation of this study is the lack of detailed clinical data to cover the period between visits, which may have helped explain some of the changes seen. We also acknowledge that this is a single-centre analysis, which may have an impact on the data; however, a recent study reported the high repeatability of 129Xe VDP in a multicentre setting [49], which highlights the potential of VDP as an end-point in multicentre studies.
In conclusion, 129Xe MRI is a highly effective method for the assessment of CF lung disease. In this study, 129Xe VDP has high within-visit repeatability and measures underlying changes in lung function that are not necessarily detected by other methods. Measuring small changes in lung function in a patient population with increasingly normal, preserved spirometry values is challenging but highly relevant. 129Xe ventilation MRI can both qualitatively and quantitatively meet this requirement and should therefore be considered as a future end-point for clinical trials and patient management.
Supplementary material
Supplementary Material
Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.
Supplementary material ERJ-00441-2020.Supplement
Shareable PDF
Supplementary Material
This one-page PDF can be shared freely online.
Shareable PDF ERJ-00441-2020.Shareable
Acknowledgements
The authors would like to acknowledge all members of the POLARIS research group at the University of Sheffield (Sheffield, UK) for their support. In particular, we would like to thank Oliver Rodgers, Guilhem Collier, Madhwesha Rao and Leanne Armstrong. We would also like to thank the Cystic Fibrosis clinical teams at Sheffield Children's Hospital, Sheffield Teaching Hospital and Manchester CF Centre for their support. Finally, we would like to thank all of the participants for their time in taking part in this research.
Footnotes
This article has supplementary material available from erj.ersjournals.com
Conflict of interest: L.J. Smith reports grants from National Institute for Health Research, during the conduct of the study.
Conflict of interest: A. Horsley reports grants from National Institute of Health Research, during the conduct of the study; grants from Cystic Fibrosis Trust and Cystic Fibrosis Foundation, personal fees from Mylan Pharmaceuticals and Vertex Pharmaceuticals, outside the submitted work.
Conflict of interest: J. Bray has nothing to disclose.
Conflict of interest: P.J.C. Hughes reports grants from GlaxoSmithKline, outside the submitted work.
Conflict of interest: A. Biancardi has nothing to disclose.
Conflict of interest: G. Norquay has nothing to disclose.
Conflict of interest: M. Wildman has nothing to disclose.
Conflict of interest: N. West has nothing to disclose.
Conflict of interest: H. Marshall has nothing to disclose.
Conflict of interest: J.M. Wild has nothing to disclose.
Support statement: This report is independent research supported by the National Institute for Health Research (NIHR) and Health Education England and also the Medical Research Council (MRC). This work was supported by the NIHR grants ICA-CDRF-2015-01-027 (awarded to L.J. Smith) and NIHR-RP-R3-12-027 (awarded to J.M. Wild) and MRC grant MR/M008894/1 (awarded to J.M. Wild). A. Horsley was supported by an NIHR Clinician Scientist award (NIHR-CS012-13) and by the NIHR Manchester Biomedical Research Centre. The views expressed in this publication are those of the author(s) and not necessarily those of the NHS, the National Institute for Health Research, Health Education England or the Department of Health. Funding information for this article has been deposited with the Crossref Funder Registry.
- Received February 26, 2020.
- Accepted June 29, 2020.
- Copyright ©ERS 2020