Image Phenotyping of Preterm-Born Children Using Hyperpolarized ¹²⁹Xe Lung Magnetic Resonance Imaging and Multiple-Breath Washout

Abstract

Rationale

Preterm birth is associated with low lung function in childhood, but little is known about the lung microstructure in childhood.

Objectives

We assessed the differential associations between the historical diagnosis of bronchopulmonary dysplasia (BPD) and current lung function phenotypes on lung ventilation and microstructure in preterm-born children using hyperpolarized ¹²⁹Xe ventilation and diffusion-weighted magnetic resonance imaging (MRI) and multiple-breath washout (MBW).

Methods

Data were available from 63 children (aged 9–13 yr), including 44 born preterm (⩽34 weeks’ gestation) and 19 term-born control subjects (⩾37 weeks’ gestation). Preterm-born children were classified, using spirometry, as prematurity-associated obstructive lung disease (POLD; FEV₁ < lower limit of normal [LLN] and FEV₁/FVC < LLN), prematurity-associated preserved ratio of impaired spirometry (FEV₁ < LLN and FEV₁/FVC ⩾ LLN), preterm-(FEV₁ ⩾ LLN) and term-born control subjects, and those with and without BPD. Ventilation heterogeneity metrics were derived from ¹²⁹Xe ventilation MRI and SF₆ MBW. Alveolar microstructural dimensions were derived from ¹²⁹Xe diffusion-weighted MRI.

Measurements and Main Results

¹²⁹Xe ventilation defect percentage and ventilation heterogeneity index were significantly increased in preterm-born children with POLD. In contrast, mean ¹²⁹Xe apparent diffusion coefficient, ¹²⁹Xe apparent diffusion coefficient interquartile range, and ¹²⁹Xe mean alveolar dimension interquartile range were significantly increased in preterm-born children with BPD, suggesting changes of alveolar dimensions. MBW metrics were all significantly increased in the POLD group compared with preterm- and term-born control subjects. Linear regression confirmed the differential effects of obstructive disease on ventilation defects and BPD on lung microstructure.

Conclusion

We show that ventilation abnormalities are associated with POLD, and BPD in infancy is associated with abnormal lung microstructure.

At a Glance Commentary

Scientific Knowledge on the Subject

A common consequence after preterm birth is decreased lung function or prematurity-associated lung disease (PLD). PLD is commonly reported in preterm-born children who developed bronchopulmonary dysplasia (BPD) in infancy. Different phenotypes of PLD, such as prematurity-associated obstructive lung disease and prematurity-associated preserved ratio of impaired spirometry, may be associated with different early-life factors and endotypes.

What This Study Adds to the Field

Hyperpolarized ¹²⁹Xe ventilation and diffusion-weighted magnetic resonance imaging (MRI) and multiple-breath washout were used to assess the differential associations between the historical diagnosis of BPD and current lung function phenotypes on lung ventilation and microstructure in preterm-born children and term-born control subjects. Ventilation abnormalities from ¹²⁹Xe ventilation MRI and multiple-breath washout were observed in the lungs of preterm-born children who have prematurity-associated obstructive lung disease, and abnormal alveolar dimensions from ¹²⁹Xe diffusion-weighted MRI were associated with preterm-born children who had BPD in infancy. ¹²⁹Xe MRI can be used to assess and phenotype functional and microstructural abnormalities in the lungs of preterm-born children.

Decreased lung function (FEV₁ < lower limit of normal [LLN]) or prematurity-associated lung disease (PLD) is a common consequence after preterm birth. PLD is commonly reported in preterm-born children who develop bronchopulmonary dysplasia (BPD; also called chronic lung disease of prematurity) in infancy (1–3). Children born late preterm at 33–36 weeks’ gestation are also now recognized to be at risk of PLD in childhood and beyond (4, 5). Indeed, we recently showed that gestation and intrauterine growth restriction (IUGR) but not BPD were significantly associated with PLD in multivariable regression models (6). Furthermore, it is likely that different phenotypes of PLD may be associated with different early-life factors, such as BPD, IUGR, and gestation, and may result in different endotypes (6). Most focus has been on those who develop prematurity-associated obstructive lung disease (POLD; FEV₁ < LLN and FEV₁/FVC < LLN) (7), but preserved ratio of impaired spirometry (FEV₁ < LLN and FEV₁/FVC ⩾ LLN) has been recently shown to be associated with the development of chronic obstructive pulmonary disease, cardiovascular disease, and increased all-cause mortality (8, 9). Although PLD has been shown to be associated with decreased lung function, the different phenotypes of PLD, including POLD and prematurity-associated preserved ratio of impaired spirometry (pPRISm), have been less well reported (6).

Magnetic resonance imaging (MRI) has emerged as a powerful tool for functional and structural assessment of pediatric lung diseases, including the use of advanced methods such as ultrashort echo time structural MRI and inhaled hyperpolarized gas functional MRI (10, 11). Hyperpolarized helium-3 (³He) or xenon-129 (¹²⁹Xe) MRI provides three-dimensional (3D) in vivo measurements of lung ventilation and microstructure through ventilation and diffusion-weighted imaging, respectively (12, 13). Hyperpolarized gas MRI is safe and well tolerated in children (14) and has demonstrated subclinical sensitivity with the detection of lung ventilation abnormalities in diseases such as cystic fibrosis (15–17), asthma (18), and primary ciliary dyskinesia (19). However, there is a paucity of MRI studies of lung disease in preterm-born children. To date, MRI studies have included structural proton ultrashort echo time imaging (20–25), ³He diffusion-weighted MRI (26, 27), and dissolved ¹²⁹Xe gas transfer MRI (28) in preterm-born neonates and children with BPD, and dynamic contrast-enhanced lung perfusion MRI in adult preterm-born survivors (29). To our knowledge, to date, no studies with ¹²⁹Xe ventilation or diffusion-weighted MRI have been reported in the lungs of preterm-born children.

Multiple-breath washout (MBW) of inhaled inert gases (SF₆ or N₂) provides global measures of ventilation heterogeneity, which are sensitive to early lung disease in children with cystic fibrosis (30). Metrics derived from MBW, such as lung clearance index (LCI) and phase III slope in the conducting airways (S_cond) and acinar lung regions (S_acin), have demonstrated significant differences between preterm-born children, including those who had BPD in infancy, and term-born children (31–33), but this has not been consistently reported (34–36), most likely because of heterogeneous lung disease in preterm-born subjects. LCI measurements are likely to complement those findings from ¹²⁹Xe ventilation and diffusion-weighted MRI to assess the function and microstructure of the preterm lung.

In this study, we assessed the differential associations between the historical diagnosis of BPD and current lung function with lung ventilation and microstructure in preterm-born children using ¹²⁹Xe ventilation and diffusion-weighted MRI and MBW, comparing results from term-born children. Some of the results of this study have been previously reported in the form of abstracts (37, 38).

Methods

Study Subjects

The RHiNO (Respiratory Health Outcomes in Neonates) study (European Union Drug Regulating Authorities Clinical Trials Database 2015-03712-20) is a comprehensive study of respiratory disease of preterm-born children (⩽34 weeks’ gestation) and term-born control subjects (⩾37 weeks’ gestation) from South Wales aged 7–12 years at enrollment, evaluating mechanisms (6), a randomized controlled trial (39), and hyperpolarized ¹²⁹Xe MRI. Prebronchodilator spirometry was performed 2–3 months before the MRI scans, quality controlled as per guidelines (40), and reported against Global Lung Function Initiative reference equations (41). Details on study recruitment and spirometry are described in the online supplement.

Sixty-five children, aged 9–13 years at the time of MRI, comprising 24 who entered the randomized controlled trial, 20 born preterm with FEV₁ > 85%, and 21 term-born control subjects, underwent MRI scanning at the University of Sheffield (see Figure E1 in the online supplement for a Consolidated Standards of Reporting Trials diagram). The children were classified according to findings on spirometry: POLD (n = 13; FEV₁ < LLN and FEV₁/FVC < LLN), pPRISm (n = 4; FEV₁ < LLN and FEV₁/FVC ⩾ LLN), preterm control (PT_C) subjects (n = 27; FEV₁ ⩾ LLN), and term-born control subjects (n = 21) regardless of findings on spirometry. Information on early-life factors, including gestational age, birth weight, diagnosis of BPD (according to National Institute of Child Health and Human Development criteria of oxygen supplementation at 28 days of age [42]), and IUGR, were obtained from the neonatal medical notes. IUGR was defined as <10th percentile for birth weight adjusted for sex and gestation using the LMS Growth program (Medical Research Council) (43). All children were free of respiratory infections for at least 3 weeks before the MRI visit. Ethical approval was obtained from the South-West Bristol Research Ethics Committee (15/SW/0289), and written informed consent and assent was obtained from the parents and children.

Hyperpolarized ¹²⁹Xe MRI

Hyperpolarized ¹²⁹Xe MRI was performed on a 1.5-T (HDx; GE) scanner using a flexible transmit/receive quadrature vest coil (Clinical MR Solutions). 3D ¹²⁹Xe lung ventilation and diffusion-weighted MRI were performed in separate breath-holds ranging from 10–16 seconds after inhalation of a gas mixture of ¹²⁹Xe and N₂ from FRC. Hyperpolarized ¹²⁹Xe was produced with an in-house (POLARIS) regulatory licensed polarizer (∼25% polarization [44]), and gas mixture volumes and ¹²⁹Xe doses were titrated according to the subjects’ heights to account for differences in lung volume (see Table E1). ¹²⁹Xe ventilation imaging was performed using a 3D balanced steady-state free precession sequence as described previously (45). ¹H images of the thorax were acquired with a spoiled gradient echo sequence for anatomical reference. ¹²⁹Xe diffusion-weighted MRI was acquired with a 3D multiple–b value spoiled gradient echo sequence with compressed sensing undersampling, as described previously (46), with a ¹²⁹Xe diffusion time of 8.5 ms and b values of 0, 12, 20, and 30 s/cm². Further details on ¹²⁹Xe diffusion-weighted parameters are reported in the online supplement.

¹²⁹Xe ventilation and ¹H image pairs were segmented using a semiautomated method (47) to calculate the ventilated lung and thoracic cavity masks, respectively. These two segmentations were used to derive the ventilation defect percentage (VDP) and ventilation heterogeneity index (VHI). VDP is the percentage of unventilated lung volume, and VHI is a measure of ventilation heterogeneity based on the coefficient of variation of surrounding pixels (17). Undersampled ¹²⁹Xe diffusion-weighted images were reconstructed, and voxelwise maps of apparent diffusion coefficient (ADC) and mean alveolar dimension (Lm_D) were derived from the stretched exponential model of hyperpolarized gas diffusion in the lungs (48). ADC was calculated from a monoexponential fit of the first two diffusion b values (0 and 12 s/cm²), while Lm_D was calculated from a stretched exponential model fit of all four b values. The global mean and interquartile range (IQR) of all values were calculated from the respective ¹²⁹Xe ADC and Lm_D maps.

MBW

MBW was performed a minimum of three times on the same day as ¹²⁹Xe MRI with a modified open-circuit Innocor (Innovision) and 0.2% SF₆ (49). LCI, S_cond, and S_acin were calculated from the average of at least two technically acceptable trials.

Statistical Analyses

The LLN of spirometry was defined as z-score < −1.64 (41). For each ¹²⁹Xe MRI and MBW metric, the 95% upper limit value (mean + 1.64 SD) was calculated from the respective term-born control group as a threshold to evaluate abnormal metrics in the preterm-born children. One-way ANOVA with post hoc Tukey’s test for multiple comparisons was performed (Prism 7.04; GraphPad). Univariable and multivariable linear regression modeling was performed (SPSS version 23.0; IBM) to identify associations between early-life factors and ¹²⁹Xe MRI and MBW metrics in the preterm-born children only. P values <0.05 were considered to indicate statistical significance.

Results

From the 65 children assessed, two term-born children were excluded, one because of an upper respiratory tract infection at the time of MRI scanning and the other because of a technical issue with the ¹²⁹Xe MRI coil. Thus, data were available from 19 term-born children and 44 preterm-born children (13 with POLD, 4 with pPRISm, and 27 PT_C subjects). In addition, 11 preterm-born children had BPD in infancy. Participant demographics are shown in Table E2, and Table 1 summarizes the ¹²⁹Xe MRI and MBW metrics for the lung function phenotype and BPD groups separately.

Table 1.

Summary of Subject Demographics and Metrics from Spirometry, ¹²⁹Xe Magnetic Resonance Imaging, and Multiple-Breath Washout for Each Grouping of Preterm-Born Children on the Basis of Either Current Lung Function or Historical Diagnosis of Bronchopulmonary Dysplasia and Term-Born Control Subjects

	Preterm Lung Function Grouping				Preterm BPD Grouping
	POLD (FEV1 < LLN, FEV1/FVC < LLN)	pPRISm (FEV1 < LLN, FEV1/FVC ⩾ LLN)	PTC (FEV1 ⩾ LLN)	Term Control Subjects	BPD	No BPD
Subjects, n (% total)	13 (21)	4 (6)	27 (43)	19 (30)	11 (17)	33 (52)
Sex (M:F), n	4:9	1:3	11:16	9:10	4:7	12:21
Age, yr	11.7 (1.0)	11.4 (1.6)	11.7 (1.0)	10.6 (1.1)	11.6 (0.7)	11.6 (1.1)
Height, cm	144.7 (9.5)	150.8 (18.6)	150.2 (10.7)	143.5 (9.8)	146.0 (11.8)	149.5 (11.0)
Weight, kg	37.7 (12.4)	44.9 (16.9)	40.6 (11.4)	36.9 (12.6)	39.5 (15.1)	40.4 (11.2)
Gestational age, wk	29 (2)	30 (3)	31 (3)	40 (1)	27 (2)	31 (2)
BPD, n (%)	5/13 (38)	1/4 (25)	5/27 (19)	N/A	11/11 (100)	0/33 (0)
Mild BPD, n (%)	1/13 (8)	1/4 (25)	3/27 (11)	N/A	5/11 (45)	—
Moderate/severe BPD, n (%)	4/13 (31)	0/4 (0)	2/27 (7)	N/A	6/11 (55)	—
FEV1 (z-score)	−2.97 (0.66)	−1.99 (0.26)	−0.38 (0.94)	0.32 (0.46)	−1.61 (1.75)	−1.19 (1.34)
FEV1 (% predicted)	64.5 (8.2)	76.5 (3.1)	95.3 (10.6)	103.9 (6.5)	80.7 (20.5)	85.7 (15.7)
FVC (z-score)	−0.54 (0.78)	−1.77 (0.30)	0.10 (0.99)	0.50 (0.59)	−0.27 (1.20)	−0.25 (1.00)
FVC (% predicted)	93.7 (9.0)	79.5 (3.5)	100.7 (11.1)	106.1 (6.8)	96.8 (13.9)	96.7 (11.2)
FEV1/FVC (z-score)	−3.14 (0.56)	−0.63 (0.46)	−0.78 (1.01)	−0.34 (0.86)	−1.83 (1.41)	−1.35 (1.38)
FEV1/FVC (% predicted)	68.6 (8.8)	95.7 (3.2)	94.1 (7.7)	97.3 (5.4)	82.7 (15.4)	88.0 (13.6)
129Xe VDP, %	2.51 (3.56)	0.99 (1.04)	0.57 (0.75)	0.41 (0.46)	1.13 (1.64)	1.22 (2.38)
129Xe VHI, %	9.92 (2.31)	8.37 (0.80)	7.92 (1.35)	7.81 (1.05)	8.40 (2.08)	8.63 (1.82)
Mean 129Xe ADC, cm2/s	0.0288 (0.0046)	0.0276 (0.0029)	0.0271 (0.0025)	0.0269 (0.0029)	0.0298 (0.0044)	0.0270 (0.0026)
129Xe ADC IQR, cm2/s	0.0101 (0.0022)	0.0095 (0.0004)	0.0095 (0.0018)	0.0086 (0.0009)	0.0111 (0.0024)	0.0092 (0.0014)
Mean 129Xe LmD, μm	252 (23)	246 (15)	244 (13)	244 (16)	257 (22)	243 (14)
129Xe LmD IQR, μm	66 (8)	67 (4)	63 (9)	59 (5)	72 (9)	62 (7)
LCI	7.09 (1.55)	6.09 (0.11)	6.03 (0.42)	6.07 (0.28)	6.48 (0.98)	6.30 (1.03)
Scond, L−1	0.046 (0.025)	0.018 (0.007)	0.023 (0.012)	0.022 (0.014)	0.030 (0.022)	0.030 (0.020)
Sacin, L−1	0.119 (0.049)	0.069 (0.044)	0.083 (0.028)	0.092 (0.034)	0.110 (0.047)	0.087 (0.037)

Definition of abbreviations: ADC = apparent diffusion coefficient; BPD = bronchopulmonary dysplasia; F = female; IQR = interquartile range; LCI = lung clearance index; LLN = lower limit of normal; Lm_D = mean alveolar dimension; M = male; POLD = prematurity-associated obstructive lung disease; pPRISm = prematurity-associated preserved ratio of impaired spirometry; PT_C = preterm with normal lung function; S_acin = phase III slope in the lung acinar regions; S_cond = phase III slope in the conducting airways; VDP = ventilation defect percentage; VHI = ventilation heterogeneity index.

Data are expressed as mean (SD) except as indicated.

Significantly higher ¹²⁹Xe VDP (P = 0.009 and P = 0.007), ¹²⁹Xe VHI (P = 0.001 and P = 0.001), S_cond (P < 0.001 and P < 0.001), and LCI (P < 0.001 and P = 0.003) were observed in the POLD group compared with the PT_C and term control groups, respectively (Figure 1). S_cond (P = 0.015) was also significantly higher in the POLD group compared with the pPRISm group, and S_acin (P = 0.024) was significantly higher in the POLD group compared with the PT_C group only. There were no significant differences among the pPRISm, PT_C, and term-born control groups for any ¹²⁹Xe ventilation or MBW metrics. No significant differences among the lung function groups were observed for any ¹²⁹Xe diffusion-weighted MRI metrics (see Figure E2).

Figure 1.

(A–E) Plots of ¹²⁹Xe VDP (A), ¹²⁹Xe VHI (B), S_cond (C), S_acin (D), and LCI (E) for preterm-born children lung function phenotypes and term-born children. All plots have mean and SD bars and P values from ANOVA (P values in brackets are for Tukey’s multiple-comparison tests between groups). Dotted lines represent 95% upper limit of each ¹²⁹Xe magnetic resonance imaging or multiple-breath washout metric calculated from the term-born children group. LCI = lung clearance index; LLN = lower limit of normal; POLD = prematurity-associated obstructive lung disease; pPRISm = prematurity-associated preserved ratio of impaired spirometry; PT_C = preterm control; S_acin = phase III slope in the acinar lung regions; S_cond = phase III slope in the conducting airways; VDP = ventilation defect percentage; VHI = ventilation heterogeneity index.

Representative ¹²⁹Xe ventilation images and 3D rendered ventilation videos from each group are shown in Figure 2 and Videos E1–E4, respectively. Among the 13 children with POLD, 6 (46%) had ¹²⁹Xe VDP greater than the 95% upper term-born control value (VDP > 1.16%) (Figure 1A, dotted line). For these six subjects with POLD, ¹²⁹Xe VHI, S_cond, S_acin, and LCI were also elevated, with respect to the 95% upper term-born control values, in five, six, two, and five of the subjects, respectively. The elevated ¹²⁹Xe VDP and VHI metrics in this subset of the POLD group were reflected in heterogeneous patterns in the ¹²⁹Xe ventilation images (Figure 2A). Conversely, the remaining seven subjects in the POLD group did not exhibit patterns of ¹²⁹Xe ventilation heterogeneity despite having significant obstructive patterns on spirometry (Figure 2B). ¹²⁹Xe ventilation images in the pPRISm, PT_C, and term-born control groups all had a homogeneous pattern, with low VDP and VHI values (Figures 2C–2E).

Figure 2.

Representative single-slice ¹²⁹Xe ventilation magnetic resonance images from each preterm lung function phenotype group and term-born children. (A) Subjects with prematurity-associated obstructive lung disease (POLD) with elevated ventilation MRI metrics. (B) Subjects with POLD with ventilation MRI metrics within the normal range. (C–E) Subjects with pPRISm (C), PT_C subjects (D), and term-born control subjects (E) with ventilation MRI metrics within the normal ranges. MRI = magnetic resonance imaging; pPRISm = prematurity-associated preserved ratio of impaired spirometry; PT_C = preterm control; VDP = ventilation defect percentage; VHI = ventilation heterogeneity index.

Global mean ¹²⁹Xe ADC was significantly increased in the BPD group compared with the no-BPD (P = 0.034) and term-born control (P = 0.049) groups (Figure 3A). Similar trends were observed for global mean ¹²⁹Xe Lm_D, just failing to reach statistical significance on ANOVA (P = 0.055). However, only 3 of 11 children with BPD had global mean ¹²⁹Xe ADC and Lm_D values greater than the 95% upper term-born control value of ADC (>0.032 cm²/s) and Lm_D (>270 μm) (Figures 3A and 3B, dotted lines). Significantly increased ¹²⁹Xe ADC and Lm_D IQR were observed in the BPD group compared with both the no-BPD (P = 0.003 and P < 0.001) and term-born control (P < 0.001 and P < 0.001) groups (Figures 3C and 3D). Within the BPD group, 6 of 11 and 7 of 11 children had ¹²⁹Xe ADC and Lm_D IQR, respectively, greater than the 95% upper term-born control values (ADC IQR > 0.010 cm²/s and Lm_D IQR > 67 μm) (Figures 3C and 3D, dotted lines). There were no significant differences between the no-BPD and term-control groups for any ¹²⁹Xe diffusion-weighted metrics.

Figure 3.

(A–D) Plots of ¹²⁹Xe global mean apparent diffusion coefficient (ADC) (A), ¹²⁹Xe global mean alveolar dimension (Lm_D) (B), ¹²⁹Xe ADC interquartile range (IQR) (C), and ¹²⁹Xe Lm_D IQR (D) for preterm-born children grouped by BPD diagnosis and term-born children. All plots have mean and SD bars and P values from ANOVA (P values in brackets are for Tukey’s multiple-comparison tests between groups). Dotted lines represent 95% upper limit of each ¹²⁹Xe diffusion-weighted magnetic resonance imaging metric calculated from the term-born children group. BPD = bronchopulmonary dysplasia.

Representative ¹²⁹Xe ADC and Lm_D maps for each of the BPD groupings are shown in Figure 4. More elevated ADC or Lm_D regions were observed in the maps of the preterm-born subjects with BPD associated with increased alveolar airspace heterogeneity and elevated ¹²⁹Xe ADC or Lm_D IQR (Figure 4A). In contrast, representative ¹²⁹Xe ADC and Lm_D maps in the no-BPD and term-born control groups were less heterogeneous compared with the BPD maps (Figures 4B and 4C). No significant differences were observed for any of the ¹²⁹Xe ventilation or MBW metrics between the BPD groupings (see Figure E3).

Figure 4.

Representative single-slice ¹²⁹Xe diffusion-weighted magnetic resonance imaging–derived maps of ADC and Lm_D of three different subjects from each BPD diagnosis grouping (A–B) and term-born children (C). ADC = apparent diffusion coefficient; BPD = bronchopulmonary dysplasia; IQR = interquartile range; Lm_D = mean alveolar dimension.

As both POLD and BPD groupings showed differential effects on the ventilation and microstructure metrics of the lung, we explored the associations of these two groupings as well as sex and IUGR with the various ¹²⁹Xe MRI and MBW metrics in a subgroup analysis of the preterm-born children only. Univariable analyses showed that sex was not associated with any MRI or MBW metrics, but IUGR was associated with ¹²⁹Xe ADC IQR, Lm_D IQR, and S_acin. POLD was significantly associated with ¹²⁹Xe VDP, ¹²⁹Xe VHI, and all MBW metrics. In contrast, BPD was associated with all of the ¹²⁹Xe diffusion-weighted metrics (Table 2). These differential associations of POLD with ventilation parameters and BPD with microstructure remained unchanged when multiple variables with P values <0.10, namely, S_acin, ¹²⁹Xe ADC IQR, and ¹²⁹Xe Lm_D IQR, were included in multivariable models (Table 3).

Table 2.

Summary of Univariable Linear Regression Analyses of ¹²⁹Xe Magnetic Resonance Imaging and Multiple-Breath Washout Metrics in the Preterm-Born Children Only

Factor			129Xe VDP	129Xe VHI	Scond	Sacin	LCI	129Xe ADC Mean	129Xe ADC IQR	129Xe LmD Mean	129Xe LmD IQR
Sex (ref: male)		β	−0.321	−0.098	−0.006	−0.012	−0.336	0.002	0	7.412	1.317
		SE	0.683	0.582	0.006	0.308	0.308	0.001	0.001	5.17	2.577
		95% CI	−1.66 to 1.02	−1.24 to 1.04	−0.02 to 0.01	−0.62 to 0.59	−0.94 to 0.27	0 to 0.004	−0.002 to 0.002	−2.72 to 17.55	−3.73 to 6.37
		P value	0.638	0.866	0.348	0.317	0.275	0.121	0.692	0.152	0.609
IUGR (ref: no IUGR)		β	0.111	0.136	0.013	0.042	0.548	0.001	0.002	6.927	7.638
		SE	0.955	0.811	0.008	0.016	0.429	0.001	0.001	7.202	3.359
		95% CI	−1.76 to 1.98	−1.45 to 1.73	0 to 0.03	0.01 to 0.07	−0.29 to 1.39	−0.001 to 0.003	0 to 0.004	−7.19 to 21.04	1.05 to 14.22
		P value	0.908	0.867	0.108	0.010*	0.202	0.312	0.002*	0.336	0.023*
Preterm lung function (ref: PTC)	POLD	β	1.936	1.997	0.023	0.036	1.064	0.002	8.094	0.001	2.454
		SE	0.675	0.547	0.006	0.012	0.294	0.001	5.604	0.001	2.767
		95% CI	0.61 to 3.26	0.92 to 3.07	0.01 to 0.03	0.01 to 0.06	0.49 to 1.64	0 to 0.004	−2.89 to 19.08	−0.001 to 0.003	−2.97 to 7.88
		P value	0.004*	0.000*	0.000*	0.003*	0.000*	0.126	0.149	0.27	0.375
	pPRISm	β	0.413	0.445	−0.006	−0.014	0.063	0.001	2.061	0	3.187
		SE	1.067	0.865	0.009	0.019	0.467	0.002	8.653	0.001	4.272
		95% CI	−1.68 to 2.5	−1.25 to 2.14	−0.02 to 0.01	−0.05 to 0.02	−0.85 to 0.98	−0.003 to 0.005	−14.9 to 19.02	−0.002 to 0.002	−5.19 to 11.56
		P value	0.699	0.607	0.514	0.451	0.892	0.755	0.812	0.998	0.456
BPD (ref: no BPD)		β	−0.09	−0.227	0.001	0.023	0.182	0.003	0.002	13.735	9.65
		SE	0.758	0.644	0.007	0.013	0.346	0.001	0.001	5.591	2.52
		95% CI	−1.58 to 1.40	−1.49 to 1.04	−0.01 to 0.01	0 to 0.05	−0.5 to 0.86	0.001 to 0.005	0 to 0.004	2.78 to 24.69	4.71 to 14.59
		P value	0.905	0.724	0.921	0.091	0.598	0.010*	0.002*	0.014*	0.000*

Definition of abbreviations: ADC = apparent diffusion coefficient; BPD = bronchopulmonary dysplasia; CI = confidence interval; IQR = interquartile range; IUGR = intrauterine growth restriction; LCI = lung clearance index; Lm_D = mean alveolar dimension; POLD = prematurity-associated obstructive lung disease; pPRISm = prematurity-associated preserved ratio of impaired spirometry; PT_C = preterm with normal lung function; ref = reference; S_acin = phase III slope in the lung acinar regions; S_cond = phase III slope in the conducting airways; VDP = ventilation defect percentage; VHI = ventilation heterogeneity index.

Statistically significant factors related to early life (sex, BPD diagnosis, and IUGR) and reduced lung function on spirometry are indicated with an asterisk.

Table 3.

Summary of Multivariable Linear Regression Analyses of ¹²⁹Xe Magnetic Resonance Imaging and Multiple-Breath Washout Metrics in the Preterm-Born Children Only

Factors			Sacin	129Xe ADC IQR	129Xe LmD IQR
IUGR (ref: no IUGR)		β	0.028	0.002	4.961
		SE	0.016	0.001	3.082
		95% CI	−0.003 to 0.06	0 to 0.004	−1.08 to 11
		P value	0.079	0.010*	0.107
Preterm lung function (ref: PTC)	POLD	β	0.029	—	—
		SE	0.012	—	—
		95% CI	0.01 to 0.05	—	—
		P value	0.016*	—	—
	pPRISm	β	−0.012	—	—
		SE	0.018	—	—
		95% CI	−0.05 to 0.02	—	—
		P value	0.508	—	—
BPD (ref: no BPD)		β	—	0.002	8.613
		SE	—	0.001	2.528
		95% CI	—	0 to 0.004	3.66 to 13.57
		P value	—	0.008*	0.001*

Definition of abbreviations: ADC = apparent diffusion coefficient; BPD = bronchopulmonary dysplasia; CI = confidence interval; IQR = interquartile range; IUGR = intrauterine growth restriction; Lm_D = mean alveolar dimension; POLD = prematurity-associated obstructive lung disease; pPRISm = prematurity-asociated preserved ratio of impaired spirometry; PT_C = preterm with normal lung function; ref = reference; S_acin = phase III slope in the lung acinar regions.

Statistically significant factors related to BPD diagnosis, IUGR, and reduced lung function on spirometry are indicated with an asterisk.

Discussion

By using hyperpolarized ¹²⁹Xe ventilation and diffusion-weighted MRI, alongside MBW, we report the differential effects of obstructive lung disease and BPD on functional and microstructural changes in the lungs of preterm-born children.

Increased ¹²⁹Xe ventilation metrics (VDP and VHI) were observed in some preterm-born children with reduced FEV₁ and obstructive pattern on spirometry (POLD) compared with preterm-born children with preserved FEV₁ (PT_C subjects) and term-born children. Although increased ¹²⁹Xe VDP and VHI have been reported in children with other obstructive lung diseases, such as cystic fibrosis (17) and asthma (18), ¹²⁹Xe ventilation heterogeneity has not previously been reported in the preterm-born population. These data also demonstrate that most preterm-born children with preserved FEV₁ did not have ventilation abnormalities on MRI, with ¹²⁹Xe ventilation images generally resembling those observed in the term-born control children. The preterm-born children with pPRISm had findings similar to those of the two control groups, but because only four subjects were available for study, the findings should be interpreted with caution.

Furthermore, within the POLD group, despite their FEV₁ < LLN, 7 of the 13 children (54%) did not have significant ventilation abnormalities on ¹²⁹Xe ventilation MRI (VDP < 1.16%, the 95% upper term-born control value). This finding highlights significant discordance between the degree of impairment measured by FEV₁ and the ventilation images in preterm-born children with reduced lung function compared with other pediatric airway disease. Studies of children with cystic fibrosis, primary ciliary dyskinesia, and asthma, for example, have all shown a strong correlation between low FEV₁ values and significant ventilation abnormalities (15–19). We have demonstrated here, however, that in preterm-born children with FEV₁ values as low as −3 z-scores, their ventilation images often appeared normal (Figure 2B). This suggests that the airflow obstruction has a different phenotype from those seen in subjects with ongoing lung disease progression, such as in cystic fibrosis, in which active inflammation, infection, and mucus plugging are ongoing. In contrast, the remaining six children with POLD demonstrated significant ventilation abnormalities that may highlight a phenotype in which there is an ongoing active pathophysiological component to their airflow obstruction. Whether both or either group would benefit from inhaler treatment is speculative (39).

MBW metrics were consistent with the ¹²⁹Xe ventilation MRI measures, with trends toward increased S_cond, LCI, and S_acin observed in the POLD group compared with the pPRISm, PT_C, and term-born control groups. In addition, similar to ¹²⁹Xe VDP and VHI, only a small subset of children in the POLD group had MBW metrics that were greater than those observed in the term-born control group. Significantly increased S_cond in the POLD group is in agreement with MBW measured with N₂ reported by Yammine and colleagues (32); however, the significantly elevated S_acin and LCI in the POLD group are contrary to what was observed in that study.

Increased LCI and S_acin in preterm-born children with or without BPD have previously been reported (31, 33), although other studies have not demonstrated any differences (34–36). In contrast, compared with previous studies (31, 33), in our study we did not find any differences for ¹²⁹Xe ventilation or MBW metrics between the BPD and no-BPD groups compared with term-born control subjects. Taken together, these observations suggest that POLD can result in ventilation defects, and BPD affects lung microstructure.

The ¹²⁹Xe diffusion-weighted MRI measurements of acinar dimensions demonstrated increased global mean ¹²⁹Xe ADC alongside significantly elevated ¹²⁹Xe ADC and Lm_D IQR in the BPD group compared with the no-BPD and term-born control groups. Increased ¹²⁹Xe ADC and Lm_D IQR in the BPD group indicates regional heterogeneity in alveolar dimensions, which may result from preterm birth at an early stage of lung development, especially in those born extremely preterm and/or because of neonatal interventions, including mechanical respiratory support and increased supplemental oxygen. Visible regions of elevated values in ¹²⁹Xe ADC and Lm_D maps in some preterm-born subjects with BPD (Figure 4A) illustrate this increased heterogeneity of alveolar dimensions. Furthermore, these subtle regional changes in alveolar dimensions go toward explaining why the IQR is a more sensitive metric than global mean in detecting alveolar dimension changes as a consequence of BPD in infancy.

The trends of increased and more heterogeneous ¹²⁹Xe diffusion-weighted MRI metrics in BPD from our study are in agreement with a previous ³He diffusion-weighted MRI study by Flors and colleagues (26) in which significantly increased ³He ADC was observed in preterm-born children with BPD. The differences in diffusion metrics between term-born control subjects and children with BPD in our ¹²⁹Xe study were less striking than those observed in study of Flors and colleagues (26). This discrepancy could possibly be related to the fewer children with moderate or severe BPD in our ¹²⁹Xe study. Flors and colleagues (26) studied 16 subjects with moderate or severe BPD, compared with 5 with mild BPD and 6 with moderate or severe BPD in our study. Furthermore, subjects with BPD in our study were likely less severe, as evidenced by only three subjects with global ¹²⁹Xe ADC and Lm_D greater than the 95% upper term-born control values.

Contrary to the trends observed in this work and the study by Flors colleagues (26), the ³He diffusion-weighted MRI study by Narayanan and colleagues (27) revealed no significant differences for ³He ADC between children with BPD and term-born control subjects. However, the longer diffusion times of the diffusion sequence used in that study may inherently sensitize those measurements to longer interacinar diffusion length scales beyond alveolar dimensions (50). In addition, the spectroscopic acquisition used by Narayanan and colleagues derived only a global metric of alveolar dimensions and therefore was unable to measure more subtle regional heterogeneity differences in alveolar dimensions that were observed in our study with ¹²⁹Xe diffusion-weighted MRI.

In survivors of BPD, it has been demonstrated that lung function decline is present throughout childhood (51) and into adulthood (4); therefore, we speculate that preterm-born subjects with BPD, including those with normal alveolar dimensions as measured on ¹²⁹Xe MRI, may progress and exhibit more severe or elevated diffusion-weighted MRI metrics as the lungs continue to develop into adulthood. This is supported by a small study of two adult survivors of BPD who demonstrated increased global mean ³He ADC compared with healthy adults of similar ages (52).

To further confirm that preterm-born children with low lung function were associated with ventilation heterogeneity, and that the BPD group was associated with lung microstructure, we used univariable and multivariable linear regression analyses to investigate which early-life factors, among sex, IUGR, BPD, and spirometry, were most closely associated with ¹²⁹Xe MRI and MBW metrics. The multivariable regression results corroborated the findings observed after stratifying into groups with low lung function and with BPD by reconfirming that all ¹²⁹Xe ventilation MRI and MBW metrics were significantly associated with a current obstructive pattern of lung disease, whereas ¹²⁹Xe diffusion-weighted metrics were significantly associated with a historical BPD diagnosis and the presence of IUGR. The ¹²⁹Xe diffusion regression results also suggest that alterations in lung microstructure are associated with both antenatal (IUGR) and postnatal (supplemental oxygen, respiratory support) factors.

This study has several strengths and weaknesses. The strengths of the study are that we used two different methods, ¹²⁹Xe MRI and MBW, which both showed similar findings regarding ventilation heterogeneity. Furthermore, ¹²⁹Xe MRI was able to demonstrate changes in regional ventilation and microstructural heterogeneity in the lungs of preterm-born children, which is not possible with conventional lung function tests. The sensitivity of ¹²⁹Xe diffusion-weighted MRI to lung microstructural changes is a particular strength, especially with the paucity of histological or quantitative computed tomography data available in the lungs of preterm-born children. We also assessed the role of important early-life factors to demonstrate the differential effects of obstructive lung disease and BPD on ventilation and microstructure of the lung, respectively.

The weaknesses of the study are that we studied a relatively small number of subjects, especially in some of the stratified groups, particularly the pPRISm group, and findings should be interpreted with caution. Selection bias and recall bias given the small numbers may also be relevant. In addition, a wider range of gestation would have identified if the findings were associated with decreasing gestation, and including greater numbers, especially those with moderate or severe BPD, would have strengthened our findings. A possible limitation for the acquisition of ¹²⁹Xe MRI is the absence of precise lung volume control, which could affect derived ¹²⁹Xe MRI metrics. However, this was mitigated by coaching and practice bag inhalations before MRI scanning.

Conclusions

¹²⁹Xe ventilation and MBW metrics were noted to be significantly increased in the lungs of preterm-born children with POLD. In contrast, ¹²⁹Xe diffusion-weighted metrics were increased in preterm-born children who had BPD in infancy, indicating alterations of alveolar dimensions. ¹²⁹Xe MRI can therefore be used to assess and phenotype functional and microstructural abnormalities in the lungs of preterm-born children.