Sensitive Structural and Functional Measurements and 1-Year Pulmonary Outcomes in Pediatric Cystic Fibrosis

Abstract

Background:

Two functional measurements (multiple breath washout [MBW] and hyperpolarized ¹²⁹Xe ventilation magnetic resonance imaging [¹²⁹Xe MRI]) have been shown to be more sensitive to cystic fibrosis (CF) lung obstruction than traditional spirometry. However, functional techniques may be sensitive to different underlying structural abnormalities. The purpose of this study was to determine relationships between these functional markers, their pathophysiology, and 1-year clinical outcomes.

Methods:

Spirometry, MBW, ¹²⁹Xe MRI, and ultrashort echo-time (UTE) MRI were obtained in a same-day assessment of 27 pediatric CF patients (ages 11.5±5.0) who had not begun CFTR modulator therapies. UTE MRI was scored for structural abnormalities and functional metrics obtained via spirometry, MBW and ¹²⁹Xe MRI. 1-year outcomes (ΔFEV₁ and pulmonary exacerbations), during which ≈50% initiated modulator therapy, were obtained from the electronic medical record.

Results:

MBW, ¹²⁹Xe MRI, and UTE MRI detected clinically significant disease in more subjects (>78%) compared to spirometry (<30%). UTE MRI suggests increased odds of bronchial changes when mucus plugging is present in the same lobe. MBW and ¹²⁹Xe MRI correlated best with mucus plugging, while spirometry correlated best with consolidations. Bronchial abnormalities were associated with future pulmonary exacerbations.

Conclusions:

MBW, ¹²⁹Xe MRI, and UTE MRI are more sensitive for detection of pediatric CF lung disease when compared to spirometry. MBW and ¹²⁹Xe MRI correlated with structural abnormalities which occur in early CF disease, suggesting MBW and ¹²⁹Xe MRI are valuable tools in mild CF lung disease that can guide clinical decision making.

Introduction:

Cystic Fibrosis (CF) results in multiple morphological lung abnormalities on computed tomography (CT) examination that are identified as mucus plugging, bronchial wall thickening, bronchiectasis, ground glass opacities, consolidations, and cysts. Significant intra- and inter-subject heterogeneities in abnormalities exist with variability in severity and lobar distribution [1]. The structural abnormalities eventually result in a loss of clinical lung function, which is periodically monitored to follow disease progression and treatment response.

Previous studies have shown the use of cystic fibrosis transmembrane conductance regulator (CFTR) modulator therapies increased lung function and reduced bronchial wall thickening and mucus plugging [2]. Recently, a triple-combination CFTR) modulator therapy has been shown to be a highly effective treatment, with a greater than 10% improvement in lung function for any patient with a single F508del mutation (~90% of all CF patients) [3]. With highly-effective therapies becoming more widely used, we anticipate a larger fraction of subjects will likely exhibit milder lung disease, both structurally and functionally. Thus, sensitive measures of lung disease progression and outcomes is a growing necessity.

Chest CT is the imaging gold standard in CF care. To provide quantification—primarily in the context of research studies—scoring systems have been developed and validated to quantify extent and severity of the structural abnormalities (e.g., Brody [4], Helbich [5], PRAGMA-CF [6], Eichinger [7]). Since then, ultrashort echo-time ¹H magnetic resonance imaging (UTE MRI) of pediatric CF patients [8,9] has been shown to be amenable to CF image scoring [7,9,10], eliminating the need for ionizing-radiation exposure. While intrinsically subjective to some degree, structural image scoring predicts future pulmonary exacerbations [11,12] and lung function loss [13].

Functionally, multiple breath washout (MBW) has been shown to be more sensitive than ubiquitously used spirometry. MBW measures the number of lung turnovers, as defined by the functional residual capacity, to reduce a tracer gas to a concentration of 2.5% of its initial value (lung clearance index; LCI_2.5). However, MBW lacks regional specificity needed to fully assess lung dysfunction [14]. To this end, hyperpolarized Xenon-129 ventilation MRI (¹²⁹Xe MRI) [15] shows promise in bridging the gaps between sensitivity, regionality, structure, and function.

¹²⁹Xe MRI determines the percent of lung volume which does not receive an adequate amount of gas following an inhalation (ventilation defect percentage, VDP). Both MBW [16,17] and ¹²⁹Xe MRI [18–21] have demonstrated sensitivity to early CF lung disease and treatment responses. However, each functional measurement may have a potential sensitivity bias towards certain structural abnormalities, which is important when considering clinical prognosis [22].

Thus, we compare functional measurements (spirometry, MBW, and ¹²⁹Xe MRI) to quantitative structural scores (UTE MRI) to determine the sensitivity of functional measures to structural defects in CF lung disease. We hypothesized both MBW and ¹²⁹Xe correlate more with mucus plugging and bronchial wall thickening compared to spirometry, as they are more common in mild disease and would lead to early ventilation heterogeneity. Additionally, we hypothesize these more sensitive functional/structural measurements would correlate better with 1-year outcomes (e.g., change in FEV₁ and number of pulmonary exacerbations requiring hospitalization) and would be dependent on modulator therapy usage.

Materials and Methods:

This prospective study was performed with approval from the IRB of Cincinnati Children’s Hospital Medical Center and with oversight from the US-FDA under an investigational new drug (IND 123,577) protocol for hyperpolarized ¹²⁹Xe MRI. The study incorporated same-day assessment using four techniques: spirometry, MBW, ¹²⁹Xe MRI, and UTE MRI. This was performed in 27 modulator-naïve pediatric CF patients (ages 11.5±5.0; at least one copy of the F508del mutation; 15/12 M/F). Only pediatric subjects were assessed to limit to mild CF lung disease. Following the same-day assessment, patients were followed for 1 year per standard-of-care clinical assessments at our institution, and of the 27 subjects, 26 continued to be seen at our institution. Measured FEV₁ (frequency = 5.8±3.5/year) and hospitalizations due to pulmonary exacerbations were recorded from the electronic medical record for short-term outcomes. Average yearly change in FEV₁ (ΔFEV₁) and number of inpatient hospitalizations for pulmonary exacerbations (determined by a scoring system accounting for symptoms, lung exam findings, and FEV₁) were calculated. For ΔFEV₁, all clinically-acquired measurements were fit to a linear regression with the slope (in ΔFEV₁/year) taken as ΔFEV₁. Of the 26 subjects, 14 of the subjects began ivacaftor or double-modulator therapy (lumacaftor/ivacaftor–11 subjects, ivacaftor–2 subjects, and tezacaftor/ivacaftor–1 subject) within the year (mean start of 54 days after same-day assessment).

Spirometry (Koko, nSpire, Longmont, CO) and MBW (Exhalyzer D, EcoMedics, Duernten, Switzerland), were performed according to American Thoracic Society and European Respiratory Society Guidelines [23,24]. MBW incorporated N₂ as the tracer gas, and the number of turnovers to reach a lung clearance index of 2.5% (LCI_2.5) was recorded.

Immediately following spirometry and MBW, the subjects underwent a single MRI session, during which ¹²⁹Xe and UTE MR images were acquired using a 3.0T Philips Achieva scanner and a homebuilt dual-loop coil for ¹²⁹Xe MRI [25] and a 32-channel cardiac coil for UTE MRI. ¹²⁹Xe images were acquired during a breath hold (~10 seconds, dependent on field of view, maximum of 16 seconds) of hyperpolarized ¹²⁹Xe gas equivalent to 1/6^th the subjects predicted total lung capacity [26]. ¹²⁹Xe images were acquired as axial slices with a gradient recalled echo MRI sequence (3mm × 3mm × 15mm resolution). RF bias correction was completed by extracting and removing linear inhomogeneity profiles within the lung for all three dimensions. Ventilation defect percentage (VDP) was calculated using the linear binning method [27], with manual lung segmentation (D.J.R.) and a control cohort of healthy age-matched controls (23 subjects; ages 12.1±4.1; 13/10 M/F; ventilation = 0.54 ± 0.15). Structural information was obtained via UTE MRI, acquired with a 3D-radial MRI sequence with rectilinear encoding in the head-to-foot direction (stack-of-stars). The axial images were gated to end-expiration using a pencil-beam navigator (1.5mm × 1.5mm × 4.0mm resolution ≈5 minutes). Additional MRI parameters can be found in Table S1 of the Supplementary Materials. Thus, to acquire both structural and functional MRI images, less than 10 minutes was necessary.

Quantitative structural analysis of the UTE MRI was completed using a simplified Brody score [8]. The reader (M.M.W.), trained and certified for Brody scoring with 3 years of experience, was blinded to the subject/measurements obtained. The lung was divided into 6 lobar-regions (the lingula separated from the left lower lobe, for parallelism), each scored for mucus plugging, bronchial wall thickening, bronchiectasis, ground glass opacities, consolidations, air trapping and cysts. For each abnormality and lobe, a score of 0 to 2 was determined where 0 indicated not present, 1 indicates presence in <1/2 the lobe, and 2 indicates presence in >1/2 the lobe. Thus, the total possible abnormality score is 6 (lobes) × 7 (abnormalities) × 2 (extent) = 84. Additionally, since the Brody score is calculated for each lobe and structure, the distribution of structural abnormalities can be assessed by summing applicable scores.

For sensitivity to mild lung disease, upper/lower limits of normal (normal for 95% of a healthy population: mean ± 1.645×standard deviation or a z-score of ±1.645) were used as clinically-significant abnormality cutoffs and were calculated from the healthy control cohort (spirometry %-predicted values and VDP) or from previously published data (UTE score and LCI_2.5), as shown in Table 1 [8,28].

Table 1:

Demographic, same-day assessment results, and short-term outcomes for the cohort.

	Mean	Standard Deviation	Range	Limit of Normal	Percent Abnormal
Demographics
Gender	15M/12F	-	-	-	-
Age	11.9	4.9	6.0–20.5	-	-
Same-day Assessment (N=27)
FEV1 %-predicted	98.7%	15.3%	66–131%	>70%	4%
FEV1 GLI z-score	−0.32	1.25	−3.07–2.60	>−1.645	15%
FEF25–75 %-predicted	92.9%	33.3%	47–147%	-	-
FEF25–75 GLI z-score	−0.55	1.41	−2.96–2.22	>−1.645	30%
FEV1/FVC	84.7%	8.1%	70–99%	>73%	7%
FEV1/FVC GLI z-score	−0.54	1.19	−2.52–1.71	>−1.645	22%
LCI2.5	10.0	2.1	6.6–14.4	<7.91	78%
VDP	18.5%	11.0%	2.6–38.9%	<6.0%	85%
Total Brody Score	9.7	7.9	0–32	<4.6	78%
Mucus Plugging	1.4	1.9	0–8	-	-
Bronchial Wall Thickening	3.0	2.7	0–10	-	-
Bronchiectasis	3.4	2.5	0–10	-	-
Ground Glass	0.4	0.9	0–3	-	-
Consolidation/Atelectasis	0.6	0.9	0–3	-	-
Air Trapping	0.9	1.0	0–4	-	-
Cysts/Bullae	0.0	0.2	0–1	-	-
1 Year Outcomes (N=26)
ΔFEV1 (%/year)	2.4	11.7	−27.6–19.2	-	-
# Pulmonary Exacerbations	0.27	0.52	0–2	-	-

Row headings in italics denote structural sub scores which make up the total Brody score. The limits of normal were derived as described in the text. Missing areas represent either not applicable or available value

All functional measures were normally distributed according to the Shapiro-Wilk normality test (P ≥ 0.05); however, structural abnormality scores and number of pulmonary exacerbations were not. Thus, spearman correlations were used to compare structural abnormalities and lung function measurements, as well as their associations with clinical outcomes. A logistic regression was completed to calculated odds ratios. Two-tailed T-tests with unequal variances were implemented for comparisons of outcomes between groups. Statistical significance was determined by a P-value <0.05. Multiple testing correction for significant correlations was completed using the Benjamini-Hochberg method with a Bonferonni α correction allowing for a 0.1 experimental false discovery rate to reduce Type I errors; P-values which are no longer significant are marked by an asterisk (*).

Results:

A summary of the data can be seen in Table 1 for all subjects and in Table S2 for modulator and modulator-naïve cohorts. Additionally, example results for 3 subjects can be seen in Figure 1. From spirometry, the average FEV₁ %-predicted of the cohort was 98.7±15.3% with only 3 subjects (11%) below 80%, the typical clinical abnormal function threshold, with one below our healthy cohort-derived lower limit of 70%. Similarly, the FEV₁/FVC ratio detected significant abnormality in 7%. Moving from percent predicted to GLI z-scores, spirometry results were more sensitive with 15%, 30%, and 22% of subjects having significantly abnormal lung function according to FEV₁, FEF_25–75, and FEV₁/FVC, respectively. Thus, GLI z-scores will be used for future analysis. MBW found a mean LCI_2.5 of 10.0±2.1 with an upper limit of normal (7.91) indicating abnormal lung function in 21 (78%) subjects. ¹²⁹Xe MRI analysis showed a VDP = 18.5±11.0% (ventilation = 0.42 ± 0.18) with an upper limit of normal (6.0%) indicating abnormal lung function in 23 (85%) subjects. The structural UTE MR images had a mean total score of 9.7±7.9 with the mean abnormality sub-scores ranging from 0.0 (cysts/bullae) up to 3.4 (bronchiectasis). The upper limit of 4.6 for the total score indicated 21 (78%) of the subjects had significant visible structural abnormalities.

Figure 1:

Example image slices for ¹²⁹Xe (a.) and UTE (b.) images in CF subjects. Subject 14 (left) is an example of a subject a relatively healthy subject with only LCI_2.5 being slightly elevated. Subject 18 (middle) is an example of a subject with elevated LCI_2.5, VDP and structural abnormalities despite normal FEV₁. Subject 26 (right) is an example subject who also had an abnormal FEV₁ %-predicted. In (a.), regions circled in red illustrate examples of ventilation defects and in (b.), example regions are circled in orange for combined bronchiectasis/wall thickening, yellow for only bronchiectasis and blue for mucus plugging. Each image provides the subject’s FEV₁ %-predicted, LCI_2.5, VDP, and Brody score.

Structurally, the right lung experienced more severe structural scores (average combined abnormality score of 5.9±4.4) than the left lung (3.9±3.8, P < 0.01). Additionally, the upper lobes displayed more severe structural disease (4.1±3.4) than the middle lobe/lingula (2.8±2.7, P < 0.01) and lower lobes (3.0±2.9, P < 0.01), in agreement with previous CT findings [4,29]. Notably, lobes with mucus plugs were strongly correlation with the occurrence of bronchiectasis and/or bronchial wall thickening. Eighty-nine percent of lobes with mucus plugging also exhibited more permanent structural airway changes, while only 37% of lobes lacking mucus plugging exhibited the bronchial abnormalities. This results in an odds ratio of 12.9 for more permanent airway structural changes given mucus plugging is visibly present in the same lobe (confidence interval = 4.3–39.0; P<0.0001).

Significant correlations between the functional measures (Table S3) were found (all P < 0.003). Variance between FEV₁ z-score, FEF_25–75 z-score, and FEV₁/FVC z-score was minimal (R² > 0.5) as was the same for LCI_2.5 and VDP. However, correlations between spirometry and the newer functional measures all had an R² below 0.5.

Correlations of structural changes/severity with functional measurements were investigated to investigate which structural abnormalities most-influenced each functional metric (Table 2 and Figure 2). FEV₁/FVC z-score, LCI_2.5 and VDP correlated with the total structural abnormality score while FEV₁ and FEF_25–75 z-scores did not. FEV₁ and FEF_25–75 z-scores only correlated with ground glass opacities and consolidation/atelectasis abnormalities. FEV₁/FVC z-score correlated with mucus plugging and consolidation/atelectasis. LCI_2.5 and VDP significantly correlated with mucus plugging while LCI_2.5 also correlated with bronchial wall thickening.

Table 2:

Spearman r and P-value for correlations between structural abnormality scores from UTE MRI and functional measures from spirometry, MBW, and ¹²⁹Xe MRI.

	FEV1 z-score	FEF25–75 z-score	FEV1/FVC z-score	LCI2.5	VDP
Total Brody Score	r = −0.31	r = −0.35	r = −0.44	r = 0.47	r = 0.41
	(P = 0.12)	(P = 0.07)	(P = 0.02)	(P = 0.01)	(P = 0.03*)
Mucus Plugging	r = −0.23	r = −0.33	r = −0.45	r = 0.51	r = 0.53
	(P = 0.23)	(P = 0.10)	(P = 0.02)	(P < 0.01)	(P <0.01)
Bronchial Wall Thickening	r = −0.16	r = −0.17	r = −0.27	r = 0.39	r = 0.32
	(P = 0.43)	(P = 0.40)	(P = 0.17)	(P =0.05*)	(P = 0.10)
Bronchiectasis	r = −0.14	r = −0.14	r = −0.21	r = 0.36	r = 0.33
	(P = 0.50)	(P = 0.49)	(P = 0.30)	(P = 0.07)	(P = 0.09)
Ground Glass	r = −0.44	r = −0.47	r = −0.30	r = 0.09	r = −0.00
	(P = 0.02)	(P = 0.01)	(P = 0.12)	(P = 0.66)	(P = 0.98)
Consolidation/Atelectasis	r = −0.46	r = −0.45	r = −0.46	r = 0.35	r = 0.34
	(P = 0.02)	(P = 0.02)	(P = 0.02)	(P = 0.07)	(P = 0.08)
Air Trapping	r = −0.10	r = −0.19	r = −0.27	r = 0.28	r = 0.12
	(P = 0.55)	(P = 0.33)	(P = 0.17)	(P = 0.16)	(P = 0.95)
Cysts/Bullae	r = 0.09	r = 0.16	r = 0.18	r = −0.03	r = 0.01
	(P = 0.66)	(P = 0.41)	(P = 0.38)	(P = 0.90)	(P = 0.95)

Statistically significant correlations are in bold.

^*indicates no longer significant after correction for multiple tests.

Figure 2:

Select correlations between structural scores and the functional measurements - FEV_1/FVC z-score (left column), LCI (middle column), and VDP (right column). Dots correspond to subjects and the dashed line shows the linear regression.

Correlations between the same-day assessment metrics and outcomes are shown in Table 3. Seven pulmonary exacerbations in 6 subjects were reported, with 4 occurring in subjects who started modulator therapy (P = 0.87). ΔFEV₁ was only associated with VDP (P = 0.05*). For pulmonary exacerbations, the only significant correlation was bronchial wall-thickening (P=0.05*). However, there is a difference when considering modulator usage. For the subjects who began modulator therapy, we found three significant correlations (bronchial wall thickening, P < 0.01; bronchiectasis, P=0.01; and total Brody score, P = 0.03*).

Table 3:

Spearman r and P-value for correlations between quantitative functional and structural scores with outcomes including the yearly change in FEV₁ and number of pulmonary exacerbations for the year following the same-day assessment.

	Cohort	ΔFEV1	# Pulmonary Exacerbations
FEV1 z-score	All	r = −0.18 (P = 0.38)	r = −0.23 (P = 0.25)
	Modulator	r = −0.36 (P = 0.20)	r = −0.16 (P = 0.59)
	No Modulator	r = 0.10 (P = 0.74)	r = −0.42 (P = 0.18)
FEF25–75 z-score	All	r = −0.21 (P = 0.30)	r = −0.21 (P = 0.30)
	Modulator	r = −0.36 (P = 0.20)	r = −0.26 (P = 0.38)
	No Modulator	r = 0.04 (P = 0.91)	r = −0.25 (P = 0.44)
FEV1/FVC z-score	All	r = −0.08 (P = 0.68)	r = −0.13 (P = 0.51)
	Modulator	r = −0.20 (P = 0.48)	r = −0.27 (P = 0.34)
	No Modulator	r = 0.12 (P = 0.71)	r = −0.02 (P = 0.96)
LCI2.5	All	r = 0.12 (P = 0.57)	r = 0.19 (P = 0.36)
	Modulator	r = 0.46 (P = 0.09)	r = 0.04 (P = 0.89)
	No Modulator	r = −0.34 (P = 0.29)	r = 0.35 (P = 0.26)
VDP	All	r = −0.39 (P = 0.05*)	r = 0.28 (P = 0.17)
	Modulator	r = −0.44 (P = 0.12)	r = 0.27 (P = 0.34)
	No Modulator	r = −0.44 (P = 0.15)	r = 0.35 (P = 0.26)
Total Brody Score	All	r = −0.03 (P = 0.90)	r = 0.30 (P = 0.13)
	Modulator	r = 0.25 (P = 0.39)	r = 0.57 (P = 0.03*)
	No Modulator	r = −0.33 (P = 0.30)	r = 0.06 (P = 0.85)
Mucus Plugging	All	r = −0.10 (P = 0.61)	r = 0.08 (P = 0.70)
	Modulator	r = 0.35 (P = 0.23)	r = 0.15 (P = 0.60)
	No Modulator	r = −0.42 (P = 0.17)	r = 0.15(P = 0.63)
Bronchial Wall Thickening	All	r = −0.12 (P = 0.56)	r = 0.39 (P = 0.05*)
	Modulator	r = 0.10 (P = 0.74)	r = 0.71 (P < 0.01)
	No Modulator	r = −0.31 (P = 0.33)	r = 0.08 (P = 0.79)
Bronchiectasis	All	r = −0.17 (P = 0.40)	r = 0.34 (P = 0.09)
	Modulator	r = 0.08 (P = 0.80)	r = 0.65 (P = 0.01)
	No Modulator	r = −0.44 (P = 0.15)	r = 0.06 (P = 0.85)
Ground Glass	All	r = 0.34 (P = 0.09)	r = 0.24 (P = 0.25)
	Modulator	r = 0.46 (P = 0.10)	r = 0.23 (P = 0.44)
	No Modulator	r = 0.32 (P = 0.32)	r = 0.32 (P = 0.31)
Consolidation/Atelectasis	All	r = 0.07 (P = 0.73)	r = 0.15 (P = 0.46)
	Modulator	r = 0.17 (P = 0.57)	r = 0.11 (P = 0.71)
	No Modulator	r = −0.18 (P = 0.57)	r = 0.20 (P = 0.52)
Air Trapping	All	r = 0.06 (P = 0.75)	r = −0.31 (P = 0.13)
	Modulator	r = 0.23 (P = 0.43)	r = −0.21 (P = 0.46)
	No Modulator	r = −0.29 (P = 0.36)	r = −0.42 (P = 0.17)
Cysts/Bullae	All	r = 0.01 (P = 0.95)	r = 0.35 (P = 0.08)
	Modulator	r = −0.03 (P = 0.91)	r = 0.44 (P = 0.12)
	No Modulator	N/A*	N/A*

Statistics are given for the entire cohort, those who started a modulator therapy the following year (n=14), and those who did not (n=12). Statistically significant correlations are in bold.

^*indicates no longer significant after correction for multiple tests.

^*N/A indicates inability to calculate due to all subjects having the same value (0, not present).

Discussion:

In this study, we assessed multiple functional measurements in a group of pediatric CF patients and compared to same-day structural MRI findings. Additionally, clinical outcomes were analyzed to determine the relationship between structural/functional measurements to future ΔFEV₁ and number of pulmonary exacerbations. This improves the ability for clinicians to select the appropriate test, interpret the results, and determine the best treatment/intervention in a personalized manner.

From the regional correlations between mucus plugging and bronchial changes, the odds ratio of 12.9 illustrates a likely causative relationship between mucus plugging and permanent airway structural changes and suggests that identifying and treating early mucus plugging with effective airway clearance and pharmaceutic interventions may yield long-term benefits. Additionally, while we used a simplified Brody scoring system, we suspect any validated CF scoring system [30] would show similar results. Moreover, while the structural scoring allowed for prognostic power with the ability to discriminate different abnormalities, it is currently a time-consuming process.

In agreement with previous publications, LCI_2.5 and VDP were more sensitive than FEV₁ at detecting mild disease [18–21]. This is demonstrated by LCI_2.5 and VDP detecting functional abnormalities in over 78% of the subjects despite mostly normal FEV₁ z-scores. This may be due to the sensitivity of LCI_2.5 and VDP to mucus plugging and bronchial abnormalities.

Additionally, an upper limit of normal for VDP is not well established and can vary depending on the cohort, acquisition methods, and precise quantitative analysis methods. Thus, the normal limit listed here is subject to change as this technique becomes more developed/standardized. Future refinements in analysis of ¹²⁹Xe MRI, will likely decrease variation within healthy subjects, lowering the limit of normal and increasing sensitivity. Thus, future studies should ensure consistent acquisition and analysis protocols. Lastly, previous studies showed moderate correlation between FEV₁ and total abnormality scores, but it was only significant when including subjects with abnormal spirometry [4]. Thus, the mild severity of disease in this cohort further emphasizes the sensitivity of LCI_2.5 and VDP compared to FEV₁.

There were some disagreements between the structural findings and the functional measurements. For example, subject 4 had the 7^th highest VDP and 2^nd highest LCI_2.5 but the structural score was average (16^th highest). This is potentially due to the sensitivity of these functional scores to certain structural abnormalities as shown here. Subject 4’s abnormalities were primarily present in mucus plugging and bronchial abnormalities with no ground glass, cysts, or air trapping observed which would explain the disagreement between VDP/LCI_2.5 and the structural scores. Additionally, the location of defects could lead to a disproportionate functional decline. For example, a mucus plug in a lobar bronchus would lead to a larger functional decline than one in a subsegmental bronchus.

While most of the variation in the functional measurements were not explained by structural measurements (max R² = 0.28), there was also noticeable variation in the correlations between functional measures (such as FEF_25–75 z-score and VDP with an R² of 0.30). Since the correlations were not perfect between the functional measures, a potential sensitivity bias for each functional measure towards different structural abnormalities is likely, as supported by the functional measures being correlated with different structural abnormalities.

Transitioning onto modulator therapy was an effect modifier when investigating the correlations with pulmonary outcomes. None of the measurements acquired during the initial assessment (functional/structural scores) were different (all P > 0.05) between the modulator (n=14) and non-modulator (n=12) groups. Thus, prior to the 1-year observation, all structural and functional metrics were similar for the two cohorts. There is an expected reversal of non-permanent structural findings (i.e., mucus plugging), leaving only more permanent structural changes (i.e., bronchial damage) once modulator therapy is initiated. Thus, outcome associations are likely dependent on the abnormalities each technique is most sensitive. This effect modification resulted in the functional measurements (spirometry, LCI_2.5, and VDP) being poorly associated with 1 year outcomes. Associations of outcomes with UTE MRI scoring was also affected by modulator therapy, but since different types of abnormalities could be discriminated, we could observe that the more-permanent abnormalities were correlated for those who did initiate modulator therapy.

Another limitation of this study is the small number of pulmonary exacerbations in this cohort, likely due to the mild lung disease present in the pediatric CF population. While this increases the type II errors, the goal was to identify the most strongly correlated relationships between these measures. Despite this, statistically significant correlations were found but many more correlations likely exist. Future studies should consider including larger cohorts for an increased number of exacerbations or more sensitive outcome measures for improved sensitivity to smaller effects missed in this lower-powered study. Lastly, one of the modulator subjects who experienced a pulmonary exacerbation in the year following did not start their modulator until 286 days post-assessment (the last subject to initiate modulator therapy) and the exacerbation occurred before the start of the modulator therapy. However, adjusting for this does not change any conclusions within this study.

Our study suggests that LCI_2.5 and VDP may provide clinicians with earlier indicators of lung abnormalities. Furthermore, the association of mucus plugging with more permanent bronchial abnormalities suggests clinical interventions to prevent mucus plugs may prevent progression of lung disease. As highly effective modulators become available to the broader CF population and approved for younger patients, we anticipate these sensitive tests will be important clinical tools to screen for pulmonary abnormalities and/or monitor lung function.

Conclusions:

LCI_2.5 from MBW and VDP from ¹²⁹Xe MRI were more sensitive to mild CF lung disease than spirometry—similar to that shown by others using ³He MRI [20]. Reader scores of structural UTE MRI demonstrated the presence of mucus plugs in mild disease that correlated with structural bronchial changes (and irreversible lung-function declines). The significant correlations for MBW and ¹²⁹Xe MRI with mucus plugging—more common in earlier CF lung disease—are probable reasons for increased sensitivity to mild lung function decline. More permanent abnormalities (bronchiectasis and bronchial wall thickening) are associated with exacerbations for those who began modulator therapy. These findings suggest functional imaging could be used for screening, due to its sensitivity, and structural imaging can also be used in determination of interventions/treatments. Thus, the use of tomographic imaging to visualize both structural (UTE MRI or CT) and functional (¹²⁹Xe MRI) abnormalities will likely benefit future studies/clinical decisions in the coming age of milder lung disease in cystic fibrosis.

Highlights:

• LCI_2.5, ¹²⁹Xe VDP, and UTE more sensitive to pediatric CF disease than spirometry

• ¹H MRI suggests increased odds for airway damage in lobes with mucus plugging

• Most spirometry measures correlated with consolidations and ground glass

• LCI_2.5 and ¹²⁹Xe VDP correlated with mucus plugging

• Increased exacerbations most strongly correlated with mucus plugs and airway damage