Parametric Response Mapping of Bronchiolitis Obliterans Syndrome progression after lung transplantation

Abstract

Bronchiolitis obliterans syndrome (BOS) remains a major complication after lung transplantation. Air trapping and mosaic attenuation are typical radiological features of BOS, however quantitative evaluation remains troublesome.

We evaluated parametric response mapping (PRM, voxel-to-voxel comparison of inspiratory and expiratory CT scans) in lung transplant recipients diagnosed with BOS (n=20) and time-matched stable lung transplant recipients (n=20). Serial PRM measurements were performed pre-diagnosis, at time of BOS diagnosis and post-diagnosis (T_pre, T₀ and T_post respectively); or at a post-operatively matched time in stable patients. PRM results were correlated with pulmonary function and confirmed by microCT analysis of end-stage explanted lung tissue.

We observed by PRM an increase in functional small airway disease (fSAD), from T_pre to T₀ (p=0.006) and a concurrent decrease in healthy parenchyma (p=0.02) in the BOS group. This change in PRM continued to T_post, which was significantly different compared to the stable patients (p=0.0002). At BOS diagnosis, the increase in fSAD was strongly associated with a decrease in FEV₁ (p=0.011). MicroCT confirmed the presence of airway obliteration in a sample of a BOS patient identified with 67% fSAD by PRM.

We demonstrated the use of PRM as an adequate readout to monitor BOS progression in lung transplant recipients.

Introduction

Long-term survival after lung transplantation (LTx) is hampered by chronic lung allograft dysfunction (CLAD) which is believed to reflect chronic rejection (1). With one of the highest rejection rates amongst all solid organ transplantations, approximately 50% of LTx recipients suffer from CLAD within five years after transplantation (2). Phenotypes of CLAD with different clinical characteristics and prognoses have been reported in the literature. The most common being bronchiolitis obliterans syndrome (BOS), characterized by an obstructive pulmonary function defect, obliterative bronchiolitis (OB) on histopathologic examination and air trapping and mosaic attenuation on imaging. As a result of the complex pathophysiology of these CLAD phenotypes, treating physicians often face difficulties to obtain an exact and early diagnosis

The gold standard for BOS diagnosis continues to be monitoring of lung function, which is easy to use, cost-effective and provides sufficient repeatability. Lung imaging, primarily via computed tomography (CT), is commonly used to complement spirometry for BOS diagnosis. In BOS, air trapping is visually identified on end-expiration CT scans as parenchymal areas with low attenuation and lack of volume reduction. De Jong and colleagues proposed a composite CT score for BOS diagnosis, comprised of bronchiectasis, mucus plugging, airway wall thickening, consolidation, mosaic pattern and air trapping (AT), which was associated with forced expiratory volume in 1 second (FEV₁) (3). Although this scoring system showed good reproducibility, this is a semi-quantitative technique that requires experienced readers for accurate and repeatable scoring.

Quantitative CT-based analytical measures have shown promise in improving disease diagnosis, phenotypes and prognosis, as well as providing 3D visualization of the disease extent. Parametric response mapping (PRM), a quantitative imaging processing technique applied to inspiration and expiration CT scans was shown to quantify the extent of parenchyma with emphysematous AT and non-emphysematous AT, which is referred to as functional small airways disease (fSAD) in a cohort of COPD patients (4). In a retrospective study of hematopoietic stem cell recipients, PRM was found to be a strong readout of BOS even in the presence of infection (5). Although serial PRM measurements were obtained in only one subject, this technique showed promise in monitoring BOS progression. Here we evaluated PRM as a readout of BOS diagnosis and progression in LTx recipients.

Material and methods

Patient population

Serial paired CT scans were obtained from LTx recipients diagnosed with BOS (n=20) and recipients without CLAD (n=20) as part of a single site retrospective study. Included patients were all double LTx recipients; and both groups were matched for post-operative day of CT, native lung disease, age and gender. All patients have reached a best post-operative FEV₁ >80% predicted and all BOS patients received azithromycin treatment for BOS but were found to be non-responsive. All BOS patients experienced a persistent FEV₁ decrease >20% compared to the mean of the two best post-operative FEV₁ values without a concomitant decrease in TLC≥10% (thus excluding rCLAD). Inspiratory and expiratory CT scans from BOS recipients were acquired 3 months to 1 year before BOS, T_pre, at time of BOS diagnosis, T₀, and at the last available CT, T_post, which may have occurred at end of follow-up, last CT prior to a second LTx or death. Time matched CT scans were obtained for the stable recipients. At the moment of each CT examination, there was no acute rejection, lymphocytic bronchiolitis or infection. This study was approved by the local hospital’s ethical committee (S57752).

Computed Tomography

CT data were obtained as whole lung volumetric CT scans at full inspiration (TLC) and incremental scans at relaxed expiration (functional residual capacity) on Siemens Somatom scanner and reconstructed using a b60 or b70 reconstruction kernel. Slice thicknesses were 1.25 mm for all scans, with slice numbers on average around 220 for inspiration scans and 15 for expiration scans. All CT scans were checked for Hounsfield unit (HU) drift and if necessary corrected based on aortic blood (50 HU) and central air (−1000 HU) as previously described (6).

Parametric Response Map

PRM was applied to all paired CT scans from both study groups. Briefly, lungs from both paired CT scans were segmented from the thoracic cavity using an in-house algorithm written in Matlab (The MathWorks, Inc. Natick, MA, USA). The whole-lung CT inspiration scan was spatially registered and aligned to the incremental CT scan obtained at expiration using Elastix, an open source image registration algorithm (7,8). Once complete, the images share the same geometric space where each voxel, the smallest unit of volume in a 3D image dataset, consists of HU values at inspiration and expiration. To minimize the effects of noise on the PRM analysis an adaptive noise-removal filter was applied to both spatially aligned CT scans. Individual voxels were then classified based on predetermined thresholds as normal (green), fSAD (yellow), emphysema (red) and parenchymal disease (purple). Voxels with values ≥-950 HU and <-810 HU at inspiration and ≥-856 HU at expiration were classified normal, ≥-950 HU and <-810 HU at inspiration and <-856 at expiration were fSAD, <-950 HU at inspiration and <-856 HU at expiration were emphysema and ≥-810 HU at inspiration were parenchymal disease, PD (5). The emphysema and air trapping thresholds (-950HU and -856HU, respectively) have been previously defined and are used for CT analysis in the COPDGene clinical trial (9–12). The upper limit on the inspiratory CT (-810HU) was determined using healthy smokers accrued as part of the NORM trial (5,13). Global PRM measures were presented as relative lung volumes. The relative volumes for each classification are defined as the sum of all voxels within a class normalized to the sum of all voxels within the volume of the expiratory lungs multiplied by 100. PRM^Normal, PRM^fSAD, PRM^Emph and PRM^PD denote the relative volume for each classification.

MicroCT investigation

Further evaluation of the PRM technique was performed by microCT analysis of an explanted lung from a single recipient that had undergone whole-lung inspiration and expiration CT scans immediately prior to a redo LTx. The lung specimen was collected and scanned using an adapted protocol of our previous work (14,15) (online supplement 1).

Statistical Analysis

All results are shown as median (interquartile range). Differences in continuous variables between BOS and stable patients were determined using a student’s T-test. A contingency test was used to assess group difference in discrete data. Paired t-test was used to assess differences between time points (i.e. T_pre, T₀ and T_post). In addition to the absolute values determined by PRM, the relative change in PRM^Normal and PRM^fSAD were evaluated between time points and groups. The relative change in PRM metric is defined as 100*[PRMⁱ(T_j)-PRMⁱ(T_pre)]/PRMⁱ(T_pre), where i is an index to denote normal and fSAD and j denotes time points (T₀, T_post). Spearman-rank test was performed for correlation analysis between the relative change in pulmonary function measurements and PRM^fSAD. ROC analysis was performed to estimate the specificity and sensitivity of rPRM^fSAD at T₀ to detect BOS using the entire study population (n=40). All tests were performed using graph pad prism 6.0 (GraphPad Software, Inc. La Jolla, CA, USA).

Results

Study population

Patient characteristics are presented in table 1. Patients were selected to minimize potentially confounding effects as well as to time-match post-operative CT acquisitions. As such, no significant differences in age, underlying disease, gender and post-operative day of CT scanning between the BOS and stable groups were observed. At T_pre, 9 patients from the BOS group had BOS0p versus 5 in the stable group. Four patients in the stable group had BOS0p at T₀ which increased to 9 patients at T_post. However none of the stable patients reached BOS1 at any timepoint.

Table 1.

patient characteristics of the 20 BOS and 20 stable patients included in the PRM study.

	BOS	Stable	p-value
Number of patients, n	20	20
Donor smoking, n (%)			0.56
yes	3 (15%)	5 (25%)
no	13 (65%)	13 (65%)
unknown	4 (20%)	2 (10%)
Donor age, yr	42±4	43±3	0.93
Recipient age, yr	46±3	49±3	0.54
Recepient gender, n(%)			0.75
Male	8 (40%)	9 (45%)
Female	12 (60%)	11 (55%)
Underlying disease, n(%)			0.75
COPD	9 (45%)	11 (55%)
CF+BRECT	3 (15%)	5 (25%)
ILD	2 (10%)	1 (5%)
Redo	4 (20%)	2 (10%)
Eisenmenger	1 (5%)	1 (5%)
Histiocytosis X	1 (5%)	0
POD of CT, yr
Timepoint 1, Tpre	2.7±0.4	2.6±0.4	0.90
Timepoint 2, T0	3.3±0.5	3.5±0.5	0.80
Timepoint 3, Tpost	4.9±0.4	5.0±0.6	0.82
FEV1 at CT, L
Timepoint 1, Tpre	2.65±0.16	3.01±0.17	0.14
Timepoint 2, T0	1.88±0.15	3.12±0.17	<0.0001
Timepoint 3, Tpost	1.11±0.15	3.02±0.19	<0.0001
FEF25-75 at CT, L
Timepoint 1, Tpre	2.54±0.26	3.05±0.29	0.19
Timepoint 2, T0	1.17±0.24	2.97±0.29	<0.0001
Timepoint 3, Tpost	0.41±0.07	2.79±0.24	<0.0001
BOS stage at diagnosis, n (%)
BOS1	13 (65%)	NA
BOS2	4 (20%)	NA
BOS3	3 (15%)	NA

Abbreviations: COPD: chronic obstructive pulmonary disease, CF: cystic fibrosis; BRECT: bronchiectasis, ILD: interstitial lung disease, Redo: redo lung transplantation; POD: post-operative day; CT: computed tomography; BOS: bronchiolitis obliterans syndrome. Contingency tables were used to compare group differences, unpaired T-test was used for continuous data.

Serial examination of PRM measurements

Representative PRM slices from two cases are shown in Figure 1. Population results for serial PRM^fSAD and PRM^Normal measurements are shown in Figure 2A and 2B. BOS patients demonstrated a significant increase in PRM^fSAD from T_pre (18±3%) to T₀ (33±4%; p=0.006). Although the moderate increase in PRM^fSAD from T₀ to T_post (45±6%) was not significant (p=0.18), PRM^fSAD at T_post was found to be significant higher compared to T_pre values (p<0.0001). In stable patients, a negligible variation in PRM^fSAD was observed from T_pre to T₀ and T_post (p=0.27 and p=0.28 respectively). PRM^Normal decreased from T_pre to T₀ (62±4% to 48±4%, p=0.02) in BOS patients with no further decline after diagnosis, T_post (38±6%, p=0.36). However, PRM^Normal at T_post demonstrated a significant decrease from T_pre values (p=0.0007). Negligible changes were observed in PRM^Normal for the control group (53±4% at T_pre, 61±5% at T₀ and 55±6% at T_post).

Figure 1.

Presented are representative axial slices of PRM from two cases demonstrating the evolution of fSAD in a BOS patient (upper panel) and a stable patient (lower panel) at T_pre, T₀ or T_post (left to right). The stable case is a male, 40 yrs of age that underwent LTx for COPD. The PRM values at 5 years post-transplantation (T_pre) for normal and fSAD were 71% and 12% of the lung volume, respectively. Negligible changes in these PRM values were observed 5.5 yrs (T₀; PRM^Normal=77% and PRM^fSAD=11%) and 7 yrs (T_post; PRM^Normal=78% and PRM^fSAD=6%) post-transplantation. The BOS case is a 36 year old male transplanted for histocytosis X. PRM^Normal and PRM^fSAD at 5 years post-transplantation were 76% and 13%, respectively. This patient was diagnosed with BOS 5.5 years post-transplantation (T₀). At T₀, PRM^Normal was 30% of the lung volume with PRM^fSAD accounting for 56%. Prior to redo-transplantation (6 years post-transplantation), PRM^fSAD changed slightly from the T₀ value (58%), yet PRM^Normal decreased to 18% of the lung volume. The resulting loss of PRM^Normal is attributed to an increase of low attenuation regions identified by PRM as emphysema (1% at T₀ to 12% at T_post). BOS, bronchiolitis obliterans syndrome; COPD, chronic obstructive pulmonary disease; fSAD, functional small airway disease; LTx, lung transplantation; PRM, parametric response mapping.

Figure 2.

Group comparisons of serial measurements of PRM. Line plots are used to present the group differences observed for the absolute and relative change in (A and C) PRM^fSAD and (B and D) PRM^Normal obtained at T_pre, T₀ and T_post. Significant differences between timepoints for the BOS groups is denoted as * for p<0.05, ** for p<0.01, *** for p<0.001. Significant differences between groups is denoted as # for p<0.05, ## for p<0.01, ### for p<0.001. Data is presented as the mean ± standard error of the mean. PRM, parametric response mapping.

Group Comparisons

Negligible group differences were observed in PRM values of fSAD and healthy parenchyma (p=0.17 and p=0.14, respectively; Figure 2A and B) and FEV₁ (p=0.14; Table 1) at T_pre. Interestingly, the mean values for PRM^fSAD at T_pre were slightly higher in both groups (BOS and stable had 18±3% and 24±4%, respectively) compared to a previously described healthy (non-transplant) control groups (<10% (4) and <15%, supplement material (5)). Although PRM^fSAD was higher at T₀ for BOS patients than stable patients, these values were not significant (p=0.09; Figure 2A). The steady increase in PRM^fSAD in BOS patients resulted in a significant difference between BOS and stable patients at T_post (p=0.0002). Similarly, PRM^Normal were significantly lower at T₀ (p=0.02) and T_post (p=0.04) in BOS patients when compared to stable patients (Figure 2B). To evaluate if PRM^fSAD can detect BOS severity, patients diagnosed with BOS were separated based on BOS grade. Results from this analysis found that patients with BOS 2–3 (n=7) at diagnosis had three times more PRM^fSAD (26±7%) than BOS 1 patients (n=13; 9±7%). Although this result was not found to be significant (p=0.10), the general trend suggests PRM^fSAD may provide clinical insight into BOS severity. At the end of current follow-up (T_post), 11 patients had graft loss (4 death, 7 redo LTx) with 9 patients alive (BOS1, n=4; BOS2, n=4, BOS3, n=1) compared to no graft loss in the stable group. Patients that succumbed to BOS (all BOS 3) had elevated PRM^fSAD (59±2%) compared to living patients (25±6%) (p=0.0007), while PRM^Normal was significantly lower (21±3% vs. 56±8% respectively, p=0.0007).

Relative Change in PRM

The relative change in PRM^fSAD (rPRM^fSAD) and PRM^Normal (rPRM^Normal) from T_pre are presented in Figure 2C–D. As shown in Figure 2C, patients diagnosed with BOS were found to have a relative increase in PRM^fSAD of 422±127% at T₀ (p=0.005) and 436±95% at T_post. These values were significantly higher than rPRM^fSAD observed in stable patients (rPRM^fSAD at T₀ of 92±16% (p=0.004) and at T_post 116±30% (p=0.001). Presented in Figure 2D, rPRM^Normal was found to be lower in BOS patients as compared to stable patients. These values were significantly different between groups at T₀ (p=0.004), but lost significance by T_post (p=0.07). The extent of change in rPRM^Normal was only significant at T_post for BOS patients (p=0.009). By ROC analysis, the relative increase in PRM^fSAD at T₀ was found to be significantly associated with BOS (p=0.005). An optimal cutoff of rPRM^fSAD at T₀ was found to be 173%, which had 62.5% sensitivity and 93.8% specificity to predict BOS development.

Correlation of PRM^fSAD with Spirometric Measures and Air Trapping

Correlations were performed in BOS patients to determine the relationship between PRM and pulmonary function (Figure 3). A relative decline in FEV₁ (%pred) from T_pre to T₀ was inversely associated with PRM^fSAD (p=0.006 and R=−0.64). Similarly, the decline in FEF25-75 was inversely correlated with PRM^fSAD (p=0.031 and R=−0.53). In contrast, no association was observed for the relative change in FVC (%pred) (p=0.43). The inverse correlation between PRM^fSAD to both relative changes in FEV₁ (%pred) and FVC (%pred) became stronger from T_pre to T_post (p=0.0002, R=−0.59 and p=0.0006 and R=−0.53, respectively). A correlation analysis using the absolute value of FEV₁ and FVC in liter yielded similar results (data not shown).

Figure 3.

Assessment of the relationship between PRM and pulmonary function. Correlation analyses are presented for the relative change in PRM^fSAD to the relative change in FEV₁ from (A) T_pre to T₀ and (B) T_pre to T_post. PRM, parametric response mapping.

Confirmation by microCT

Presented in Figure 4 are PRM and microCT results from a representative BOS explant specimen. Using CT scans acquired at T_post, this patient was found to have PRM measures over both lungs of 15% for healthy parenchyma, 56% for fSAD, 19% for PD and 4% for emphysema. As seen in Figure 4A and B, the PRM, with corresponding scatter plot, at the exact location of the selected core showed a high percentage of fSAD (PRM^fSAD=67%) with lower values in PD (PRM^PD=16%), normal parenchyma (PRM^Normal=9%) and emphysema (PRM_Emph=1%). MicroCT confirmed the presence of small airway obliteration on 2 separate locations within the single scanned specimen (Figure 4C and D) which is compatible with obliterative bronchiolitis, the pathological correlate of BOS.

Figure 4.

Confirmation of OB by microCT in a lung region of high PRM^fSAD. (A) Presented are serial in vivo expiration CT scans with PRM image or cored region from superior to inferior and the explant slice with the location of the scanned core indicated by a blue dashed circle. (B) Corresponding PRM scatter plot for the cored region. The PRM at the exact location of the selected core showed a high percentage of fSAD, yellow (PRM^fSAD=67%); while high attenuation regions, purple (PRM^PD=16%) and normal parenchyma, green (PRM^Norm=9%) were less frequent, while emphysema was almost absent, red (PRM_Emph=1%). (C) MicroCT images of the core showing initially a patent airway (yellow arrow), which completely obliterates (red arrow). (D) MicroCT images within the same core at a different location illustrating another airway lesion where a patent airway (yellow arrow) is found to completely obliterate (red arrow). PRM, parametric response mapping.

Discussion

In this manuscript, we demonstrated the potential value of serial radiological PRM measurements in diagnosing and monitoring BOS. We provided further support for our findings by demonstrating a correlation between PRM^fSAD and FEV₁. Lastly, microCT examination confirmed the presence of small airway obliteration, explaining the fSAD documented with the PRM measurement.

An interesting find in this study was the elevated PRM^fSAD observed in all LTx recipients prior to any pulmonary function defect. These values were higher than observed from non-transplant healthy smokers reported in the literature (4,5). Although all patients in this study had a FEV₁>80% predicted at T_pre, this find corroborates the observation that air trapping occurs early after LTx, even in stable patients. This is thought to be attributed by some degree of small airway disease secondary to the transplantation procedure (16). Also, more extended criteria donors, typically older with a longer smoking history, are more commonly used (17).

Recently, Solyanik et al compared 3 methods of quantifying air trapping in LTx patients (18). Their approach differed from the present study in that multiple measurements within the same patients were not obtained. Nevertheless, one method applied a density mapping approach with voxel-to-voxel comparison of inspiratory and expiratory CT analogous to PRM, with the difference being the use of a single threshold applied to the expiratory CT scan. The authors demonstrated a correlation with residual volume/total lung capacity (used to define pathologic air trapping) and the density mapping technique. In parallel, we also quantified air trapping by radiological scoring and compared this to PRM (online supplement 2). Although air trapping progressively increased in BOS patients, we were unable to demonstrate a correlation between FEV₁ decrease and increase in air trapping, demonstrating that PRM might be a more sensitive and objective measurement for estimating small airway obstruction compared to subjective scoring as PRM is probably better in detecting subtle differences in air trapping.

Interestingly, in end-stage BOS lungs 59% of the lung was found to be fSAD by PRM. This find corresponds well with the previously observed percentage of obstructed airways we calculated using our previously reported ex-vivo CT analysis of explanted lungs where we found that between 50 and 60% of the airways were obliterated from generation 7 on (14). PRM^fSAD generated from the core (67%) slightly differed from what was calculated over the both lungs (56%). This is most likely attributed to spatial heterogeneity of the disease within the lungs. Therefore, severe airway obstruction observed in the analyzed lung strengthens the utility of PRM as a viable tool for monitoring BOS progression.

The objective of this study was to evaluate PRM as an indicator of BOS progression. As such patients with rCLAD were deliberately excluded as they demonstrate elevated attenuation on inspiration CT scans as a result of central and peripheral ground glass opacities (19–21), since PRM has not been validated yet in restrictive lung disease, these patients were not further investigated.

Other possible limitations include the relative low number of patients included in this study, which complicates analyses that require further group stratification. PRM is a quantitative density-based CT technique that applies thresholds for sorting individual voxels into specific classifications. As such various sources of variability may alter the density measurements effecting PRM classifications. In our previous work, we evaluated the effect these sources have on altering the PRM measurements (22). The limitations and generalizability of PRM is further discussed in online supplement 3. However, all CT scans in this study design were quality controlled by assessing HU drift and corrected when needed. In an effort to reduce the effects of variation in CT acquisitions and reconstruction kernels, a noise-reducing filter was applied to all scans prior to PRM analysis. It is possible that the PRM thresholds used may not fully represent disease state of the studied patient population. PRM^fSAD was found to be elevated in pre-BOS recipients relative to healthy smokers, which was consistent to observations from a previous study (5). This may be a consequence of the expiratory threshold (i.e. -856HU) over-estimating the extent of air trapping in the lungs. Confirmation of PRM^fSAD by microCT analysis of explanted lungs alleviates some of these possible concerns as OB lesions were observed in regions with high PRM^fSAD. The big advantage of our study is the serial measurements from a stable to a diseased state (BOS), which is, to the best of our knowledge, not been performed in this context. It is well understood that spirometric decline is a hallmark of the onset of BOS. Yet clinical practice dictates that CT scans must be used to rule out other pulmonary complications that cause a decline in pulmonary function and to potentially differentiate with rCLAD. As CT scans are part of standard clinical management of these patients, PRM analysis of these scans may complement FEV₁ as an objective readout of the presence and extent of OB as indicated by PRM^fSAD and might be specifically useful in patients where spirometry is difficult to interpret such as suture problems, pneumonectomy or single lung transplanted patients (see online supplement 4 for a representative example of a patient with fluctuating FEV₁ but stable PRM^fSAD values).

In conclusion, we demonstrate that PRM served as an imaging readout of BOS. Further proof of its applicability is provided by the correlation with spirometry and the presence of OB with microCT. PRM investigation may serve as a complementary tool to aid treating physicians in diagnosing BOS. Further research in this field would include a large, multi-center prospective trial and inclusion of patients with restrictive CLAD. These steps could further confirm the potential use and benefit of PRM for monitoring CLAD progression in LTx recipients.

Supplementary Material

Supp Info

Figure S1: (A) deformable registration of the in vivo inspiration, in vivo expiration and explant lung CT. (B) Linear registration of the explant CT with the specific (drilled) lung section. (C) Final transformation of the in vivo and expiration CT.

Figure S2: Comparison of PRM_fSAD to CT scored airtrapping

Figure S3: Pulmonary function evolution of a 36 year old female patient transplanted for non-specific interstitial fibrosis with additional endobronchial stenting within 2 months after transplantation for collapse of the right bronchus. The FEV₁ fluctuates during her later follow-up declining below the BOS threshold (≥20 FEV₁ decline; red line) on several separate occasions. Despite this fluctuating pulmonary function, the patients remains BOS-free at the end of follow-up. Two pulmonary function decreases could be explained by overt infection which are indicated in the graph. The PRM measurements of the acquired CT’s during her follow-up show a relatively stable percentage of PRM^fSAD and PRM^Normal.

Abbreviations

AT

air trapping

BOS

bronchiolitis obliterans syndrome

CLAD

chronic lung allograft dysfunction

CT

computed tomography

FEV₁

forced expiratory volume in 1 second

FVC

forced vital capacity

HU

Hounsfield unit

LTx

lung transplantation

PD

parenchymal disease

PRM

parametric response mapping

fSAD

Functional small airway disease

rCLAD

restrictive CLAD