Lung Parenchymal Development in Premature Infants Without Bronchopulmonary Dysplasia

Summary

Rationale

While infants who are born extremely premature and develop bronchopulmonary dysplasia (BPD) have impaired alveolar development and decreased pulmonary diffusion (DL_CO), it remains unclear whether infants born less premature and do not develop BPD, healthy premature (HP), have impaired parenchymal development. In addition, there is increasing evidence that pro-angiogenic cells are important for vascular development; however, there is little information on the relationship of pro-angiogenic cells to lung growth and development in infants.

Objective

and Methods Determine among healthy premature (HP) and fullterm (FT) infants, whether DL_CO and alveolar volume (V_A) are related to gestational age at birth (GA), respiratory support during the neonatal period (mechanical ventilation [MV], supplemental oxygen [O₂], continuous positive airway pressure [CPAP]), and pro-angiogenic circulating hematopoietic stem/progenitor cells (CHSPCs). We measured DL_CO, V_A, and CHSPCs in infants between 3–33 months corrected-ages; HP (mean GA = 31.7wks; N = 48,) and FT (mean GA = 39.3wks; N = 88).

Result

DL_CO was significantly higher in HP than FT subjects, while there was no difference in V_A, after adjusting for body length, gender, and race. DL_CO and V_A were not associated with GA, MVandO₂; however, higher values were associated with higher CHSPCs, as well as treatment with CPAP

Conclusion

Our findings suggest that in the absence of extreme premature birth, as well as BPD, prematurity per se, does not impair lung parenchymal development.

INTRODUCTION

Gas exchange is the primary function of the lung. The neonatal lung undergoes rapid growth and development early in life to increase the surface area for diffusion. In addition, it is increasingly clear that development of alveolar-capillary units for gas exchange is critically dependent upon angiogenesis.¹ In animal models, angiogenesis, and alveolarization are suppressed with hyperoxia or other anti-angiogenic factors,^2–6 while impaired alveolarization can be enhanced with infusion of pro-angiogenic cells.^7–9 During human fetal development, the rudimentary components of the alveolar-capillary unit are present by 23 weeks gestational age (GA), which enables most very premature infants to survive with significant respiratory support of supplemental oxygen (O₂) and mechanical ventilation (MV). However, following extreme premature birth, the majority of these infants develop bronchopulmonary dysplasia (BPD), which is characterized by impaired alveolar development with fewer, but larger alveoli and fewer capillaries,¹⁰ and lower pulmonary diffusion capacity (DL_CO).¹¹ While extreme prematurity and BPD is one end of the premature birth spectrum, the majority of premature births occur later in gestation, and the neonates require only modest amounts of respiratory support and do not develop BPD. We have previously demonstrated that late preterm infants who required minimal respiratory support have lower airway function compared to infants born fullterm.¹² This finding suggests that premature birth might also affect parenchymal lung development; however, there have been no studies of gas exchange during infancy of premature infants without BPD.

Older children and adults that were born premature and did not develop BPD do not consistently display deficits in DL_CO;^13–20 therefore, it remains unclear if these subjects have impaired alveolar-capillary development. Inconsistencies among follow-up studies may relate to: (1) the age of follow-up, which can reflect differences in respiratory support strategies over several decades; (2) the prolonged time period between infancy and follow-up that may provide time for catch-up in alveolar growth and development; and (3) follow-up studies that combined results of all premature subjects without separating those with and without BPD.

In infants born full term, we recently reported that an increased ratio of circulating pro-angiogenic cells was associated with a higher DL_CO and higher pulmonary capillary blood volume.²¹ This finding suggests that the angiogenic environment, as reflected by circulating pro-angiogenic cells, may be an important novel indicator and/or determinant of lung growth and development. The purpose of our current study was to determine whether those infants who were born premature, without developing BPD, exhibit impaired alveolar development as assessed by DL_CO. In addition, we evaluated whether DL_CO was related to GA at birth, respiratory support received as a neonate, and circulating pro-angiogenic cells. We hypothesized that those with greater immaturity at birth, as well as those receiving respiratory support, would have lower DL_CO, while those with an increased ratio of circulating pro-angiogenic cells would have a higher DL_CO.

METHODS

Subjects

Healthy Preterm Infants without BPD (HP)

Subjects born between 27 and 36 weeks gestation who did not require supplemental oxygen at 36 weeks gestation²² were recruited from the NICU, Newborn Follow-up Clinic, and the Pediatric Pulmonary Clinics at Riley Children’s Hospital, as well as the Newborn Intensive Care Units (NICU) of Methodist and Community Hospitals in Indianapolis (2010–2013). All subjects were clinically stable outpatients at the time of testing, did not require supplemental oxygen, and had no acute respiratory symptoms for >3 weeks.

Full term (FT)

Subjects born ≥37 weeks gestation were recruited by advertisements in local publications in Indianapolis (2007–2013). Subjects had histories negative for cardiopulmonary malformations, pneumonia, wheezing, or hospitalization for a respiratory illness.

DL_CO and V_A

Subjects were evaluated at Riley Children’s Hospital while sleeping after administration of oral chloral hydrate (50–100 mg/kg) and oxygen saturation (SO₂) and heart rate were monitored during testing. The study was approved by the Institutional Review Board at Indiana University and informed parental consent was obtained.

DL_CO and V_A were assessed using a single breath technique with an induced respiratory pause at an inflation pressure of 30 cm H₂O, as we have previously described.^11,23 Gas concentrations of the test gas (0.3% C¹⁸O, 4%He, 21%O₂, and balance N₂) were continuously measured with a mass spectrometer (Perkin-Elmer). Results for DL_CO and V_A measurements were expressed as the average of 2–3 values within 10%. DL_CO was adjusted for hemoglobin.²⁴

Our previous study found a difference of DL_CO between FT control and BPD infants to be 0.35 and standard deviation to be 0.466.¹¹ If the difference for DL_CO between FT and HP subjects is 0.24 (60% of difference for FT and premature with BPD) then evaluating 85 FT and 45 HP would have 80% power to detect this difference with a 5% Type I error rate.

Pro-Angiogenic Circulating Progenitor Cells

A venous blood sample (1–3 ml) was obtained for quantification of CHSPCs using a 7-color multi-parametric flow cytometry method, as previously described, and expressed as a ratio of pro-angiogenic (pCHSPCs) to non-angiogenic CHSPCs (nCHSPCs).^21,25,26

Statistical Analysis

Subjects’ demographics and lung function variables of each group (HP and FT) were summarized and compared using two-sample t-tests for continuous variables and Chi-Square tests for categorical variables. If the continuous variables had non-normal distributions, the Wilcoxon non-parametric test was used.

Relationships between lung function measurements (DL_CO and V_A) and the following covariates were evaluated using multiple linear regression models: body length and corrected-Age at testing, race, gender, GA, pCHSPC:nCHSPC ratio, and whether patients received supplemental O₂, mechanical ventilation, or CPAP. The coefficients and standard error for continuous predictors and adjusted least squared means for each categorical variable are presented. All analytic assumptions were verified and all tests were performed using SAS v9.3 (SAS Institute, Cary, NC).

RESULTS

Subject Demographics, DL_CO, V_A, and pCHSPC:nCHSPC

We evaluated 136 infants and toddlers with corrected ages between 3 and 33 months; forty-eight subjects were HP and 88 subjects were FT. Blood samples for pCHSPC: nCHSPC were obtained from 43 HP and 30 FT subjects. The demographics for each group of subjects are summarized in Table 1. There were no significant differences between the HP and FT groups in correct-age, body length or weight at the time of testing, or gender. There was a higher proportion of Caucasians compared to non-Caucasians in the HP group than the FT group (P = 0.04). Some of the preterm subjects required supplemental oxygen, mechanical ventilation or continuous positive airway pressure with values that ranged from 0–12 days.

TABLE 1.

Group Demographics, DL_CO, V_A, and pCHSPC:nCHSPC

	Full term controls	Preterm without BPD	P-value
Gestational age at birth (wks)	39.3 (37–41)	31.7 (27–35)	n/a
Corrected-age at testing (mths)	14.4 (3.1–32.8)	13.7 (5.6–21.3)	0.499
Body length at testing (cm)	75.8 (60.0–97.5)	76.1 (63.2–96.1)	0.798
Body weight at testing (kg)	10.0 (5.9–15.7)	9.9 (6.1–17.8)	0.817
Gender (Female, %)	47 (53.4)	22 (45.8)	0.398
Race (Caucasian, %)	57 (64.8)	39 (81.3)	0.044
Supplemental O2 (days)
0 days	–	12 (25.0)	n/a
1–5 days	–	14 (29.2)
6–10 days	–	5 (10.4)
11+ days	–	17 (35.4)
Mechanical ventilation (days)
0 days	–	32 (66.7)	n/a
1–5 days	–	14 (29.2)
6–10 days	–	1 (2.1)
11+ days	–	1 (2.1)
CPAP (days)
0 days	–	29 (60.4)	n/a
1–5 days	–	17 (35.4)
6–10 days	–	1 (2.1)
11+days	–	1 (2.1)
Hgb (gm/dL)	12.1 (1.0)	12.2 (1.0)	0.681
DLCO (ml/min/mmHg)	3.76 (1.74–6.87)	4.11 (2.30–7.11)	0.100	0.0271
VA (ml)	612 (260–1154)	642 (351–1161)	0.406	0.6141
pCHSPC:nCHSPC	1.76 (0.79–2.94)	1.59 (0.70–2.86)	0.184	0.1011

Values are means (standard deviation) for all variables except time/day variables which are represented by frequency (percent) for given categories.

Values are means (range) for Gestational Age at Birth, Body Length at Testing, Body Weight at Testing, Corrected-Age at Testing, DL_CO, V_A and pCHSPC:nCHSPC.

¹Comparison adjusted for body length, gender, and race.

The unadjusted group mean values for DL_CO, V_A, and pCHSPC:nCHSPC ratio were not significantly different between the FT and HP groups. However, after adjusting for body length at time of testing, gender, and race, the HP group had significantly higher DL_CO compared to FT, but no significant differences for V_A, pCHSPC:nCHSPC ratio, or Hgb (Table 1).

Effect of Respiratory Support and Pro-Angiogenic Cells on Pulmonary Diffusion

The results of the analysis of DL_CO by gestational age at birth, pCHSPC:nCHSPC ratio and respiratory support (oxygen, MV, CPAP) adjusting for gender, race, body length, and corrected-age at testing are summarized in Table 2. Gestational age at birth was not significantly associated with DL_CO. However, pCHSPC: nCHSPC ratio was significantly associated with DL_CO: those infants with a higher pCHSPC:nCHSPC ratio had higher values of DL_CO. Neither use of supplemental oxygen or mechanical ventilation were significantly associated with DL_CO. There was a significant relationship between use of CPAP and DL_CO: treatment with CPAP was associated with a higher DL_CO. Among the premature infants, those treated with CPAP (N = 19) compared to those premature infants not treated with CPAP (N = 29) had a significantly lower gestational age (30.9 vs. 32.2 weeks; P < 0.037), were more frequently treated with supplemental oxygen (P < 0.011), but did not differ in treatment with mechanical ventilation (P=0.980). If CPAP is removed from the model, GA, MV, and O₂ are still not significantly associated with DL_CO. Similar results were obtained if MV, O₂, and CPAP were analyzed as days of treatment log transformed to obtain normal distributions for analysis, rather than using dichotomous variables (Yes/No).

TABLE 2.

Effect of Respiratory Support and Pro-Angiogenic Cells on Pulmonary Diffusion

	Slope (StErr) or adjusted mean1	P-value
Gestational age at birth (weeks)	0.00 (0.03)	0.895
Supplemental O2 (Yes/No)	−0.07 (0.30)	0.814
Mechanical ventilation (Yes/No)	0.24 (0.24)	0.320
CPAP (Yes/No)	0.66 (0.23)	0.007
pCHSPC:nCHSPC	0.36 (0.15)	0.019
Body length at testing (cm)	0.10 (0.02)	<0.000
Corrected-age at testing (months)	0.02 (0.02)	0.366
Race		0.086
Caucasian	4.29 (0.14)
Non-Caucasian	3.99 (0.17)
Gender		0.002
Male	4.38 (0.14)
Female	3.90 (0.16)

¹Adjusted mean for categorical predictors.

Among the adjusted co-variates, body length at time of testing was significantly associated with DL_CO: increasing body length was associated with increasing DL_CO. In addition, after adjusting for body length, boys had higher DL_CO compared to girls; however, race was not statistical significant.

Effect of Respiratory Support and Pro-Angiogenic Cells on Alveolar Volume

The results of the analysis of V_A by gestational age at birth, pCHSPC:nCHSPC ratio, and respiratory support (oxygen, MV, CPAP) were adjusted for gender, race, body length, and corrected-age at testing and are summarized in Table 3. Gestational age at birth was not significantly associated with V_A. However, an increasing ratio of pro-angiogenic cells was marginally associated with higher values of V_A (P=0.057). Treatment with CPAP was significantly associated with a higher V_A, while neither O₂ nor MV was associated with V_A.

TABLE 3.

Effect of Respiratory Support and Pro-Angiogenic Cells on Alveolar Volume

	Slope (StErr) or adjusted LSM1	P-value
Gestational age at birth (weeks)	−1.90 (5.13)	0.712
Supplemental O2 (Yes/No)	−7.00 (47.39)	0.883
Mechanical ventilation (Yes/No)	4.82 (37.26)	0.898
CPAP (Yes/No)	74.59 (36.66)	0.046
pCHSPC:nCHSPC	45.28 (23.31)	0.057
Body length at testing (cm)	17.14 (2.45)	<0.000
Corrected-age at testing (months)	4.99 (3.13)	0.117
Gender		0.600
Male	625.83 (22.37)
Female	613.29 (24.57)
Race		0.001
Caucasian	666.53 (22.10)
Non-Caucasian	572.60 (26.25)

¹Adjusted least squared mean for categorical predictors.

Among the adjusted co-variates, body length was significantly associated with V_A: increasing body length was associated with increasing V_A; however, there was no significant gender effect after adjusting for body length. Race was significantly associated with V_A, Caucasians had higher values for V_A compared to Non-Caucasians (Table 3).

Determinants of pCHSPC:nCHSPC

pCHSPC:nCHSPC ratio was not significantly associated with GA, gender, race, body length or corrected-age at testing.

DISCUSSION

Our study is the first to evaluate the pulmonary diffusion capacity, during infancy, of preterm infants without BPD. We found that, after adjusting for somatic size, gender, and race, DL_CO was not significantly lower in premature infants without BPD compared to infants born full term; it was actually higher. GA was not associated with DL_CO or V_A, when analyzed as a continuous variable rather than the dichotomous variable (HP vs. FT) and accounting for exposure to mechanical ventilation and supplemental oxygen in the neonatal period. We previously reported reduced DL_CO in extremely preterm infants with BPD. Among these BPD infants, who had a mean GA of 26 weeks, decreasing GA at birth was associated with lower DL_CO,¹¹ which suggested that the stage of lung development at birth may be an important determinant of subsequent alveolar development. Despite our hypothesis that preterm birth and subsequent exposure to O₂ and MV would contribute to impaired parenchymal development throughout the spectrum of prematurity, there was not an association of GA, O₂, or MV with DL_CO within the current group of less premature infants (mean GA of 32 weeks). This absence of lower DL_CO in HP compared to FT subjects contrasts with previous findings that healthy late preterm infants have lower airway function compared to infants born fullterm.¹² This finding may reflect that the airways and lung parenchyma undergo differing patterns of growth early in life, as well as differences in their potential for catch-up growth. Our findings suggest that in the absence of extreme preterm birth, prematurity per se, may not be sufficient to impair lung growth and development; however, other factors may be important indicators or determinants of lung growth and development.

We did find that treatment with CPAP, as well as, a higher ratio of circulating pro-angiogenic cells were associated with higher DLCO. Among our study population, a higher pCHSPC:nCHSPC ratio was associated with a higher DL_CO, which suggests that a pro-angiogenic environment is associated with greater lung development. We previously found in infants born full term that increasing pCHSPC:nCHSPC ratio was associated with increasing DL_CO, which was secondary to the increasing pulmonary capillary blood volume, which more directly associates these pro-angiogenic cells to vascular development.²¹ Our current study extends these findings from infants born full term to infants born prematurely. Previous studies have reported that the frequency or function of pro-angiogenic cells in cord blood is related to gestational age at birth;^27,28 however, we did not find that the pCHSPC:nCHSPC ratio measured from peripheral blood at the time of testing during infancy related to GA at birth. Additionally, the pCHSPC: nCHSPC ratio remained significantly associated with DL_CO when GA was included in the analysis.

Previous studies have demonstrated mixed results when attempting to correlate proportions of various subsets of circulating hematopoietic stem and progenitor cells to the risk or severity of BPD. Studies by Borghesi et al., and others failed to demonstrate a relationship between the risk of significant lung disease in preterm infants and subsets of circulating hematopoietic progenitors at birth or throughout the first 4–6 weeks of age, leading them to conclude that circulating hematopoietic stem and progenitor subsets cannot be used to assign future risk of lung disease.^29–31 Important differences in the flow cytometry technique used to identify CHSPC subsets may account for our seemingly contradictory data. Indeed, based on our recent work utilizing multi-parametric flow cytometery to isolate ECFCs,³² there has been a call to reexamine and re-define the last decade’s literature that misidentified populations of circulating endothelial progenitor cells enumerated by conventional flow cytometry.³³

Our group previously isolated pCHSPC from humans and demonstrated their pro-angiogenic potential in a mouse xenograft model.^26,34 In addition, we have found that adolescents with type I diabetes have reduced pCHSPC, which directly correlated with endothelial dysfunction,³⁴ and women with gestational diabetes mellitus (GDM), as well as their infants, have a reduced pCHSPC:nCHSPC ratio, as well as endothelial dysfunction.²⁵ A pro-angiogenic environment may have systemic effects in multiple organs, as evidenced during the menstrual cycle, when the dramatic vascular growth in the uterus is associated with increased circulating pro-angiogenic cells and increased pulmonary diffusing capacity secondary to increased microvessel density.³⁵ In support of the “vascular hypothesis” of BPD pathogenesis, a previous study of premature infants demonstrated that, compared to those that developed no or mild BPD, those that developed moderate-severe BPD had a significantly reduced ratio of pCHSPC:nCHSPCs in cord blood.²⁷ Our study extends these early clinical observations by demonstrating that the ratio of pro-angiogenic cells during infancy is directly correlated with pulmonary diffusion capacity, a measure of physiologic gas exchange that depends upon angiogenesis. Additionally, our findings suggest that the angiogenic profile remains relevant even in early infancy. Animal models of BPD have demonstrated that infusion of pro-angiogenic cells or secretory products from these cells promotes alveolar development.^8,9 Therefore, the angiogenic environment, whether in utero or extra-utero, may be a more important determinant of lung growth and development than gestational age. Promoting an angiogenic environment within the fetal and the neonatal lung may provide a means for novel therapeutic interventions to stimulate lung growth.³⁶ Alternatively, the pCHSPC: nCHSPC ratio may be a marker of improved alveolar capillary development.

We also found that the use of CPAP in our premature infants was associated with higher values of DL_CO and V_A. The infants that received CPAP treatment had a lower GA, were treated more frequently with oxygen, but did not differ in treatment with mechanical ventilation. Therefore, the benefits of CPAP may extend beyond avoidance of mechanical ventilation and reduced oxygen exposure. CPAP is increasingly used in NICUs to avoidor wean from mechanical ventilation to potentially reduce lung injury and the risk for developing BPD. While CPAP minimizes airway closure and improves oxygenation by increasing lung volume, animal studies suggest that continuous mechanical distention of the lung can stimulate extrauterine lung growth in immature ferrets.³⁷ In addition, dynamic tracheal occlusion can stimulate intra-uterine lung growth in a model of congenital diaphragmatic hernia.³⁸ Therefore, CPAP treatment in the neonatal period may not only minimize lung injury, but may provide a therapeutic intervention to stimulate lung growth.

We found that DL_CO was higher in male than female infants after adjusting for body length at the time of testing. This association persisted in our large group of infants over a wide range of gestational ages at birth and is consistent with our recent report that DL_CO, as well as the pulmonary capillary blood volume, is higher in male than female infants born full term.²¹ The greater pulmonary diffusion, but not greater alveolar volume in male compared to female infants would be consistent with males having a greater number, but smaller sized alveoli compared to females, which could yield more surface area for diffusion with the same lung volume. Our physiologic results are consistent with morphometric findings in rhesus monkeys.³⁹ Our findings highlight that gender differences in lung growth and development observed in older children and adults are also present very early in life.^40,41 In contrast, pCHSPC:nCHSPC ratio was not related to gender, suggesting that its association with DL_CO occurs through an independent mechanism.

There are several limitations to our study. Our data is cross-sectional and the assessment of lung growth would require a longitudinal assessment of subjects with initial measurements ideally obtained at birth to determine the effects of in utero and neonatal factors. Current methodology for assessment of DL_CO and V_A in infants requires sedation and cannot be obtained in the neonatal period; however, our study of infants dramatically decreases the time between birth and follow-up evaluation that occurs in studies of older children and adults who were born premature. Another limitation of our study is that subjects were not part of a cohort; therefore, we may have had selection bias relative to a generalized population of premature infants. While we found significant associations between CPAP and DL_CO, as well as V_A, the number of subjects treated with CPAP, as well as the length of CPAP treatment was relatively short. Therefore, future studies using a controlled intervention with CPAP will be required to further evaluate our observation. Lastly, circulating pro-angiogenic cells were evaluated when DL_CO measurements were obtained following discharge from NICU. Therefore, the predictive value of these pro-angiogenic cells in cord blood for subsequent lung development could not be evaluated.

In summary, we found that DL_CO was higher, not lower for infants born premature who did not develop BPD compared to infants born full term. Additionally, neither DL_CO nor V_A were associated with GA at birth, MV, or supplemental O₂, which suggests that later-premature birth in the absence of developing BPD may not significantly impair lung parenchymal development. However, higher values for DL_CO and V_A were associated with use of CPAP in the NICU, and infants with a greater ratio of circulating pro-angiogenic cells, which may provide a novel marker of alveolar-capillary development.