Evaluation of Regional Lung Mass and Growth in Neonates with Bronchopulmonary Dysplasia using Ultrashort Echo Time MRI

Abstract

Rationale:

Bronchopulmonary dysplasia (BPD) is the most common long term pulmonary morbidity in premature infants and is characterized by impaired lung growth and development. We hypothesized that lung mass growth is a critical factor in determining outcomes in infants with BPD.

Objectives:

To measure regional lung density and mass in infants with BPD and compare to clinical variables.

Methods:

We conducted a retrospective cohort study of neonates (n=5 controls, n=46 with BPD). Lung mass and lung density were calculated using ultrashort echo time (UTE) MRI.

Measurements and Main Results:

Lung mass increased with increasing corrected gestational age at the time of MRI in all patients. Total, right and left lung mass in infants with BPD trended higher than control infants (65.7g vs 49.9g, 36.2g vs 26.8g, 29.5g vs 23.1g respectively). Babies with BPD who survived to discharge had higher relative lung mass than control infants and infants with BPD that did not survive to discharge (21.6 g/kg vs 15.7 g/kg, p=0.01). There was a significant association between the rate of lung mass growth and linear growth at the time of MRI (p=0.034).

Conclusions:

Infants with BPD are capable of building lung mass over time. While this lung mass growth in infants with BPD may not represent fully functional lung tissue, higher lung mass growth is associated with increased linear growth.

Introduction

Bronchopulmonary dysplasia (BPD) is the most common pulmonary complication in premature infants.^1–5 It is characterized by chronic inflammation, fibrosis, and scarring leading to impaired lung development.² The cause of injury observed in the lungs of infants with BPD is multifactorial.

The premature infant’s lung development is not complete having entered the saccular phase of lung development during week 24 of gestation.⁶ During this phase, alveolar sacs are not yet fully formed and the pneumocytes responsible for surfactant production have not proliferated. As a result, there is decreased functional lung parenchyma available in premature infants and they are forced to utilize external respiratory support to survive. These external mechanical forces combined with other complicating factors, such as infection and altered infant nutrition, cause an arrest in normal lung growth that impacts lung function throughout infancy and into adulthood.^7–11

Despite increasing knowledge of the pathological mechanisms of BPD, the relationship between BPD and parenchymal lung growth remains poorly understood. It is known that larger lung volume (measured as functional residual capacity) is an indicator of worse outcomes in infants with BPD.¹² Other volume measurements, including tidal volume and minute ventilation, are also increased in the most severe cases of BPD.^13,14 However, it is unknown if the increased lung volume correlates with an increase in lung mass growth or if the lungs are suffering from hyperinflation, resulting in pathologic low-density lung tissue.

We previously established that 3D ultrashort echo time (UTE) magnetic resonance imaging (MRI) is a safe and reliable option for measuring lung density, volume, and mass in infants due to its proton-density weighting.¹⁵ Using this approach, we demonstrated that nutrition and body growth correlates strongly with lung volume and mass in infants with congenital diaphragmatic hernia.^16,17

We hypothesize that lung mass growth will be abnormal in BPD and that MRI based measurements of lung density, volume, and mass will correlate with infant nutrition and clinical outcomes in infants with BPD. Therefore, the goal of the current study is to evaluate regional lung density and mass and their relation to growth and clinical outcomes of infants with BPD.

Methods

Study Subjects

Infants born before 32 weeks of gestational age (GA) and diagnosed with BPD of any severity who underwent research lung UTE MRI were included in this retrospective cohort study. The severity of BPD (mild, moderate, severe) was determined based on the amount of respiratory support at 36 weeks corrected GA using the NIH consensus definition.^18,19 Infants with major congenital abnormalities or genetic disorders that may affect lung growth, , or evidence of any respiratory infection at the time of MRI were excluded. MR images from a given patient are evaluated by a trained reader for motion artifact. Patients that are determined to have more than 50% of the MR images corrupted by significant motion artifact are removed from the study data set. Using this approach, less than 10% of patients are removed from the data set. Five late preterm and term neonates who did not have signs of respiratory distress or history of lung disease and did not require any respiratory support at the time of MRI were included as control subjects. All patients had UTE MRI imaging taken between 0 and 21 weeks from birth based on clinical indication. The study was performed with Institutional Review Board approval (IRB 2013-6101) and parental consent.

MRIs

All MRIs were performed on a 1.5T neonatal MRI system in the neonatal intensive care unit at Cincinnati Children’s Hospital as previously described.^20,21 Patients were free breathing on various levels of respiratory support (room air, nasal canula, continuous positive airway pressure, and mechanical ventilation) and scanned in supine position.

Lung Analysis

Segmentations of whole lung parenchyma were generated manually using Amira 6.3.0 (FEI Visualization Sciences Group). Identities and clinical status of the study subjects were blinded to the authors during all MRI analyses, including whether the subjects belonged to the control or BPD group. Care was taken to remove major blood vessels from the parenchymal material. Lung intensities of the UTE MRI images were normalized between noise and chest wall muscle to yield lung densities (~g/mL), as previously published.¹⁵ Mean lung densities and volumes were measured from these lung segmentations, and then used to calculate lung mass (g).

Whole lung was further segmented into regions based on anatomy. Each right and left lung was manually divided into superior/inferior segments and ventral/dorsal segments resulting in eight lung segments per patient. The tracheal carina served as the dividing line in both the sagittal and axial planes. For superior/inferior segments, the axial slice that most closely approximated the position of the carina was used as the partition between superior lung and inferior lung. Similarly, for the dorsal/ventral segments, the sagittal slice that most closely approximated the carina was used as a partition between ventral and dorsal lung. Although all neonates were scanned in a generally supine position, not all subjects were scanned with their back completely flat against the table. To correct for this, patients who were rotated ≥15° from the horizontal (measured in axial view) were manipulated to eliminate this tilting effect. Nine of the 51 study subjects required manual segmentation using this method. Each lung segment was then analyzed for volume, density, and mass as described above.

Calculation of Rate of Lung Mass Growth since Birth

The rate of lung mass growth since birth was approximated in the BPD population using normal reference lung masses from autopsy of fetuses with normal in utero lung development who died for non-lung related issues at various stages of gestation (15). For each subject, an estimated lung mass at birth was generated based on the gestational week that the baby was born. The difference between the expected lung mass at birth and lung mass at MRI was calculated to generate a change in lung mass from birth to MRI. This value was then divided by chronological age at MRI in weeks to calculate a rate of lung mass growth in the post gestational period.

Statistical Analysis

Data were summarized using means, standard deviations (SD) and range for numerical variables and frequency and percent (%) for categorical variables. Comparisons were made using Student’s t-test or Pearson’s correlation where appropriate. For all tests, a p value of 0.05 was considered statistically significant.

Results

Clinical characteristics of the patient population are summarized in Table 1. Forty six infants with BPD and 5 patients without BPD as healthy controls were included in the study. Of these 5 control patients, 2 were full-term births and 3 were mildly pre-term. Of the 46 patients diagnosed with BPD, 7 were of mild severity, 7 were moderate severity, and 32 were severe cases. The average gestational age at birth was 26.3 weeks for the BPD population and 36.7 weeks for the controls. The average corrected gestational age (CGA) at the time of MRI scan was 40.3 weeks.

Table 1.

Clinical Characteristics of the BPD Population

Characteristics (n=46)	Values
GAa at MRI (weeks)b	40.3 ± 3.5
GAa at Delivery (weeks)b	26.4 ± 2.0
Weight at birth (kg)b	0.8 ± 0.3
Weight at MRI (kg)b	3.2 ± 0.5
Height at MRI (m)b	0.46 ± 0.03
% Male	25 (54%)
Severity of BPD--Mild	7 (15%)
Severity of BPD--Moderate	7 (15%)
Severity of BPD--Severe	32 (70%)
Respiratory Support at time of discharge-- Room Air	12 (57%)
Respiratory Support at time of discharge-- Nasal Canula	11 (26%)
Respiratory Support at time of discharge-- Tracheostomy+Ventilation	19 (45%)
Deceased	4 (10%)

^aGestational Age (GA)

^bmean ± standard deviation

Lung Characteristics of BPD Population Compared to Healthy Controls

The average total, right, and left lung volume in infants with BPD trended higher than the control population, although this did not reach significance (97.6 mL vs 118.9 mL (p=0.22), 52.3 mL vs 64.5 mL (p=0.18), and 45.3 mL vs 53.4 mL (p=0.34) for total, right and left lung volume in the control vs BPD groups). While lung volumes were higher in the BPD group, this did not associate with a lower lung density since the average lung density of the control population was 0.50 g/mL, while the average lung density of the BPD population was 0.56 g/mL (p=0.11).

Density and mass of regions of lung parenchyma divided into quadrants based on lung anatomy were investigated (Table 2). In both the BPD and control groups in the supine position, the dorsal region of the lung was denser than the ventral region in the right and left lungs. Within the BPD population, asymmetric density measurements was observed between the upper and lower lungs such that the right upper lung was denser than the right lower lung (p<0.002) while the left upper lung was less dense than the left lower (p=0.007). Within the control population, there was no significant difference between the densities of the upper and lower segments in either the right (p= 0.26) or left (p=0.97) lungs.

Table 2.

Mass and Density of Lung Regions

	BPD (avg)	p value	Control (avg)	p value
Right Lung Mass (g)	36.1	0.10a	26.8
Left Lung Mass (g)	29.5	0.19a	23.1
Total Lung Mass (g)	65.7	0.13a	49.9
Density (g/mL)
Right Sagittal (upper)	0.59	<0.002b	0.54	0.25b
Right Sagittal (lower)	0.55		0.50
Left Sagittal (upper)	0.54	0.006b	0.50	0.97b
Left Sagittal (lower)	0.56		0.51
Right Axial (ventral)	0.48	<0.002c	0.43	0.01c
Right Axial (dorsal)	0.61		0.56
Left Axial (ventral)	0.49	<0.002c	0.45	0.05c
Left Axial (dorsal)	0.59		0.54

^aBPD vs. Control group p value

^bupper vs. lower p value

^cventral vs. dorsal p value

The mean lung mass of the total, right, and left lung of the control population was 49.9 g, 26.8 g and 23.1 g respectively. The mean lung mass of the total, right, and left lung of the BPD population was 65.7g, 36.2 g and 29.5 g, respectively (Table 2).

Comparison of Lung Mass and Corrected Gestational Age

To evaluate the potential of lung growth in premature infants with BPD, lung mass was compared to CGA of the subject at the time of the MRI (Figure 1). Although we observed considerable variability within our cohort, lung mass increased significantly with increasing gestational age at the time of MRI (average increase of 3.1 g/wk, p=0.005). Mass of the left lung and right lung were separately compared with CGA in the same manner and both comparisons displayed a significant positive increase in mass with increasing CGA at the time of MRI (right lung 1.8 g/wk, p=0.002; left lung 1.3 g/wk, p=0.015).

Figure 1.

Lung mass and corrected gestational age at the time of MRI are for total and right versus left lung are shown for infants in the BPD group (n=46). A: Total lung mass increased significantly with increasing gestational age at the time of MRI (average increase of 3.1 g/wk, p=0.005). Dashed lines represent upper and lower boundaries of normal range. Solid line represents trend line of our population. B: Mass of the left lung and right lung increase with increasing CGA at the time of MRI (Right lung, squares, 1.8 g/wk, p=0.002; Left Lung, triangles, 1.3 g/wk, p=0.015). Lines represent trend lines for left (dashed) and right (solid) lung.

Comparing Clinical Respiratory Outcome and Lung Mass

Infants with BPD were categorized based on the level of respiratory support at the time of discharge 1) no respiratory support required, 2) nasal cannula oxygen, 3) tracheostomy and long-term mechanical ventilation, and 4) deceased prior to discharge (Figure 2). Total lung mass was normalized to overall body weight of the infant. Compared to control patients without BPD and the BPD subjects deceased prior to discharge, surviving BPD subjects had a significantly greater lung mass/kg body weight (15.7 g/kg vs 21.6 g/kg, p=0.01). However, there was no statistical difference in lung mass between room air, nasal cannula, or mechanical ventilation.

Figure 2.

The BPD cohort was divided into three subgroups based on respiratory support at discharge and a fourth subgroup of infants that did not survive. The lung mass was normalized to infant body weight and the average for each category including the control group is shown. The three surviving BPD groups had significantly larger normalized lung mass when compared to the control and death groups (21.6g/kg vs 15.7g/kg for the surviving BPD vs control/death groups respectively, p=0.01). Error bars represent standard deviation.

Lung Mass Growth

To evaluate factors that might influence lung mass growth in premature infants, the rate of lung growth was determined by comparing the lung mass obtained at MRI to the expected lung mass at the birth gestational age. There was no significant association between the rate of lung mass growth and gestational age at birth (p=0.44, Figure 3a) or corrected gestational age at the time of MRI (p=0.79, Figure 3b). There was a non-significant positive correlation between increasing lung growth rate and higher birth weight (p=0.08, Figure 3c). There was also a non-significant trend between the static measurement of weight at the time of MRI and rate of lung growth (p=0.103, Figure 4a). However, the correlation between lung growth rate and infant growth was significant when we used the static measurement of height at the time of MRI as a marker of linear infant growth (p=0.034, Figure 4b). When lung mass was normalized to body weight and compared to the rate of weight gain and rate of linear growth again (rather than a static measurement of body growth), the association between weight and lung mass reversed such that there was a trend towards decreasing normalized lung mass growth with increasing body weight gain since birth (p=0.152, Figure 4c). In contrast, the association between linear growth and lung mass was more consistent with a trend towards increasing normalized lung mass growth and increasing linear growth since birth (p=0.143, Figure 4d).

Figure 3.

Lung growth rate from birth to MRI was calculated as described in methods and compared to clinical factors. There was no association between the gestational age at birth (p=0.44)(A) or corrected gestational age at the time of MRI (p=0.79)(B). There was a non-significant association between higher birth weight and increased lung growth rates (p=0.08)(C).

Figure 4.

The rate of lung mass growth from birth to MRI was calculated as described in methods and compared to the body weight and length at the time of MRI as static markers of nutritional status. There was a non-significant association between weight at the time of MRI and increased rate of lung mass growth (p=0.103)(A). The association between body length at the time of MRI and increased lung mass growth rate was significant (p=0.034)(B). When lung mass was normalized to body weight, there was a non-significant association between lower normalized lung mass and increased rate of weight gain per week (p=0.152)(C). In contrast, there was a non-significant association between higher normalized lung mass and increased rate of linear growth (D)(p=0.143).

Although only 4 infants in our study cohort of 46 infants were deceased, there was a trend towards decreased rate of lung mass growth normalized for body weight in the deceased group when compared to the surviving group at the time of MRI (0.33 vs 0.71 g/kg/week, p=0.13)

Discussion

Bronchopulmonary dysplasia is a lung disease in premature infants that causes significant morbidity and mortality.^1–5 Previous research has shown lung inadequacy in BPD is characterized by chronic tissue fibrosis and is associated with increased lung volumes.¹² However, our understanding of regional lung density, mass and the factors that influence lung growth in infants with BPD is still limited.

Our results support previous studies that demonstrated total lung volumes are increased in BPD, with an average total lung volume of 118.9 mL in infants with BPD compared to 97.6 mL in the control group, and we further demonstrate that right and left lung volume in infants with BPD tended to be higher than the control population.¹² Most have assumed that increased lung volumes seen in the BPD population are attributed to pathologic air trapping and overinflation of lung parenchyma.²² Surprisingly, although some infants did have areas of hypodensity in the BPD group, our density data did not demonstrate a uniform hypodensity in BPD, since the average lung density of the BPD population was 0.56 g/mL and the average lung density of the control population was 0.50 g/mL (p=0.1128). Although not statistically significant, these data certainly do not support the idea that large lung volumes in BPD are universally due to a hypodense and overinflated state. Interestingly, the theory of air trapping and overinflation is also contradicted by our recent study that demonstrated large cystic areas in BPD lungs move more air and have higher relative tidal volumes when compared to surrounding normal density lung tissue.²³

When the density of specific regions of the lung tissue in our BPD and control groups was examined, we observed the expected higher density of the dorsal regions in our supine infants. However, on the sagittal plane there was a significant increase in lung density in the right upper and left lower lung regions in the BPD group that was not observed in the control group. This may just be a statistical artifact of the smaller number of infants in the control group so a larger sample size will ultimately be needed to determine if the right left asymmetry of density in upper versus lower lung regions observed specifically in the infants with BPD persists.

Since our data suggested the BPD lung is more complex than just a low density, hyperinflated balloon, the factors that might influence lung mass were examined. We cross-sectionally compared the right, left and total lung mass to corrected gestational age at the time of MRI and found that lung mass steadily increased with increasing gestational age suggesting the lung in premature infants continues to grow in mass. When compared to previously reported normal ranges, the lung mass in our BPD cohort was often higher than the normal range.²⁴ This is consistent with the lung mass comparison we observed between our BPD and control group and suggests that the difference is due to a higher lung mass in the BPD group and not a lower lung mass in the control group.

Using previously published normal values for fetal lung mass, the rate of lung mass growth from birth to MRI was calculated for each patient.²⁴ No clear associations were observed between the rate of lung mass growth and the gestational age at birth but there was a potential association between birth weight and the rate of lung mass growth. Our previous analysis of lung volume and mass in infants with congenital diaphragmatic hernia suggested a role for infant total body growth and nutrition in promoting lung growth in that population.^16,17 Therefore, we utilized body weight and length at the time of MRI as static markers of infant growth and nutritional status and compared them to the lung growth rate. There was a non-significant trend suggesting an association between body weight at the time of MRI and increased lung growth rate. Since premature infants with BPD often have excessive body weight gain associated with poor linear growth, we questioned if the slight increase in lung growth rate seen in heavier infants was keeping up with the observed body growth.^7,25–27 Therefore, lung mass was normalized to body weight and compared to the rate of weight gain and now the trend reversed to potential association between excessive weight gain and lower normalized lung mass. In contrast to body weight, when body length was used as a static marker of infant growth and nutritional status, there was a significant association between body length at the time of MRI and the rate of lung growth. Moreover, although this association lost statistical significance, a positive trend was still observed when the normalized lung mass was compared to the rate of linear growth since birth. Therefore, while excessive weight gain may not associate with accelerated lung mass growth, a nutritional status that leads to good linear growth may associate with increased lung mass.

Beyond the potential association with infant nutrition, the reason for general elevated lung mass in BPD is uncertain. One potential explanation is that the increased lung density is due to tissue that is not normal lung parenchyma such as fibrosis and atelectasis. Another possibility is that the lungs in BPD are growing at an accelerated rate to compensate for reduced gas exchange and overall lung function. Although certainly not conclusive, one could interpret this latter theory to be consistent with the results in figure 2, where the surviving BPD groups had larger lung mass than the control and deceased BPD groups. Furthermore, it is possible that the lack of compensatory accelerated lung mass growth in the non-surviving BPD group contributed to their ultimate poor outcome. Although not statistically significant, we found it interesting that the rate of lung mass growth normalized for body weight in the deceased group was less than half of the rate of lung mass growth in the surviving groups at the time of MRI (0.71 vs 0.33 g/kg/week, p=0.13). This finding may also emphasize the importance of understanding the factors that influence good lung mass growth in premature infants.

Since this is the first reported investigation of lung mass in surviving infants with BPD, our results are limited by the relatively small size of our study population and the observational nature of the data. Our analysis of lung mass growth rates were also limited by our weight based estimate of lung mass at the time of delivery. Clearly serial measurements of lung mass will be needed before postnatal lung mass growth rates can be firmly determined in preterm infants. Some of the results reach statistical significance, while others are only trends that may be used for hypothesis formation and future studies, but certainly still need further investigation before we can accept them as valid. These limitations notwithstanding, the current study suggests that the lungs of infants with BPD grow in volume and mass. While it is uncertain if this is growth of functional lung tissue, our data demonstrates that more lung mass growth occurs in infants with BPD that survive which would suggest lung mass growth is desirable in infants with BPD. Of the many factors that might influence lung growth in BPD, one that clearly needs further investigation is the role of infant nutrition.

Abbreviations:

BPD

Bronchopulmonary dysplasia

UTE

Ultrashort echo time

NICU

neonatal intensive care unit

GA

gestational age

CGA

Controlled Gestational Age

PMA

Post-menstrual Age

RA

Room Air

MRI

Magnetic Resonance Imaging

Data Availability Statement

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