Inflammatory imbalance in tracheal aspirate of very preterm newborns is associated with airway obstruction and lung function deficiencies at school age: a cohort study

Abstract

Objective

A low expression of club cell secretory protein (CC16) and high levels of proinflammatory cytokines at preterm birth are associated with airway inflammation and more severe neonatal lung disease. The present study aimed to investigate if low levels of CC16, proinflammatory cytokines and vascular endothelial growth factors (VEGF) in tracheal aspirate early after birth were associated with lung function impairment at school age.

Patients and methods

Participants were 20 children, born very preterm (median gestational age 25+3 weeks+days, IQR: 24+1–27+0 weeks+days), who had tracheal aspirates collected during mechanical ventilation in their first day of life. CC16, cytokines, VEGF and matrix metalloproteinase-9 were measured in the tracheal aspirate and later correlated to results from advanced lung function measurements at 12 years of age.

Results

Low levels of CC16 and high levels of the proinflammatory cytokines IL-1β and TNF-α in tracheal aspirate were associated with airway obstruction at school age but not with other lung function parameters. The correlation with airway obstruction was even stronger when the ratio between the respective proinflammatory cytokine and CC16 was used. In addition, low levels of VEGF and CC16 were associated with impaired diffusion capacity of the lung.

Conclusions

An imbalance in inflammatory mediators and growth factors in the lungs at birth may have consequences for airway function and vasculature at school age in preterm born children.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• A proinflammatory state at the time of very preterm birth is known to be associated with a more severe lung disease in the neonatal period, but the possible long-term consequences of such inflammation have been less well studied.

WHAT THIS STUDY ADDS

• Our findings suggest that there may be a direct association between an imbalance including proinflammatory mediators, deficient defence mechanisms, especially CC16, and low growth factors in the lung early after birth and worse lung function at school age.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Early anti-inflammatory interventions in very preterm infants may have more long-lasting consequences for lung function than previously assumed. Long-time follow-up of such interventions is warranted.

Introduction

An imbalance between proinflammatory and anti-inflammatory mediators during pregnancy may lead to aberrant inflammation and is associated with increased mortality and morbidity of the mother and offspring.^{1 2} The maternal environment during gestation influences the offspring health at birth and throughout the life course. A dysregulated inflammation can negatively affect birth outcomes, resulting in an increased risk for restricted foetal growth and preterm birth.³ There is evidence that bronchopulmonary dysplasia (BPD) is associated with an imbalance between proinflammatory and anti-inflammatory factors, with an overweight of proinflammatory mediators.^{4 5}

Club cell secretory protein (CC16) is abundant in normal airway secretions and is thought to be important in protecting the lung from inflammatory insults and oxidative stress. CC16 concentration in amniotic fluid increases during pregnancy and reaches a plateau at 30 weeks.⁶ Infants born at lower gestations may have a relative CC16 deficiency and thereby an insufficient defence against inflammation. In infants born before 30 weeks, those who developed BPD had lower levels of CC16 in tracheal aspirate on day 1 and did not show the expected increase of CC16 in the trachea during the first week of life.⁷

Vascular endothelial growth factor (VEGF) regulates vasculogenesis and angiogenesis, which are important steps during embryogenesis. The vascular growth has been shown to be impaired in preterm born infants with BPD,⁸ and low VEGF levels in tracheal aspirate during the first week of life is associated with BPD.⁹ Also, later in life, children born extremely preterm had significantly impaired exercise capacity at the age of 19,¹⁰ which might be due to a lower lung function and/or an impaired vascular function.

The proinflammatory cytokines interleukin-1β (IL-1β) and tumour necrosis factor-α (TNF-α) are important mediators during inflammation by recruiting and activating inflammatory cells and by inducing pathways of lung injury.¹¹ Both IL-1β and TNF-α have been implicated in the mechanisms of preterm parturition¹² and were increased in tracheal aspirate in infants who developed early lung disease.^{13 14} For example, the proinflammatory cytokine IL-1 is associated with chorioamnionitis, preterm prelabour rupture of membranes and preterm birth, possibly due to the increased levels of IL-1 in amniotic fluid and increased infiltration of leucocytes in the chorion and amnion.¹⁵

Gastric fluid aspirated early after birth is a source of lung fluid swallowed by the fetus. In very preterm infants, low CC16 in gastric fluid at birth was associated with higher concentrations of the proinflammatory cytokines IL-1β and TNF-α, and of matrix metalloproteinase-9 (MMP-9), in the trachea within 24 hours, and with a more severe respiratory disease in the neonatal period.¹⁶ In 102 very preterm infants with respiratory distress syndrome, those who developed BPD had higher levels of proinflammatory cytokines and a lower level of the anti-inflammatory cytokine IL-10 in tracheal fluid aspirated early after birth.¹⁷ However, possible long-term childhood effects of this proinflammatory state in the newborn lung need to be further investigated.

We hypothesised that an inflammatory imbalance in the lung early after birth, quantified as low CC16 and high proinflammatory cytokines in tracheal aspirate, would be associated with persistent airway symptoms and lung function abnormalities that could be detected by lung function testing at school age.

Participants and methods

Study population

The 20 participants in the study all come from a prospective cohort of 64 infants born before 31 weeks gestation¹⁸ Out of 55 survivors, 41 participated in a follow-up study and performed a comprehensive lung function testing at 12 years of age. 20 of these children had also been intubated before 24 hours of age and had tracheal aspirate collected; they became participants of the present study. They (11 boys and 9 girls) were born at a median gestational age of 25+3 weeks+days and IQR: 24+1–27+0 weeks+days (range 23+0–28+4 weeks+days), with a median birth weight of 760 g (IQR: 637–966 g). Characteristics of the children who had tracheal aspirate collected at birth and later did lung function testing are shown in table 1, as well as characteristics of all 41 children from the cohort who were followed up at 12 years.

Table 1.

Characteristics of the children who did lung function testing at 12 years of age and the subgroup of children from whom tracheal aspirate had been collected at birth

	Children who performed lung function testing n=41	Children with tracheal aspirate and lung function testing n=20
Neonatal characteristics
Gestational age, weeks+days	26+4 (24+4–28+1)	25+3 (24+1–27+0)
Birth weight, g	854 (686–1143)	760 (637–966)
Male/female*	20/21 (49%/51%)	11/9 (55%/45%)
Surfactant at birth*	24 (62%)†	17 (85%)
Mechanical ventilation*	29 (71%)	19 (95%)
Ventilator days	3 (0–18)	13 (2–24)
BPD*	30 (73%)	17 (85%)
Anthropometrics at the time of lung function testing
Age, years	12.0 (11.9–12.1)	12.0 (11.9–12.1)
Height, cm	150 (146–157)	151 (145–158)
Weight, kg	40.3 (36.2–51.1)	43.2 (36.3–54.0)

BPD is defined as supplemental oxygen at 36 weeks postmenstrual age.

*Data are presented as median (IQR) or a number of individuals (in %).

†Data are missing from two individuals.

BPD, bronchopulmonary dysplasia.

Sampling and analyses of tracheal aspirate

Using a tracheal suction set attached to a sampling tube (Unomedical A/S, Lejre, Denmark), tracheal aspirate was collected in mechanically ventilated infants during a routine tracheal suctioning procedure within the first 24 hours of age but not earlier than 6 hour after surfactant instillation. After instillation of 0.3 mL of sodium chloride (9 mg/mL) in the tracheal tube, continuous suctioning at a standardised pressure was applied and the suction catheter was slowly retracted. Thereafter, an additional 1 mL of sodium chloride was flushed through the catheter into the sampling tube. The sample was thereafter centrifuged at 1200 rpm for 10 min and the supernatant was stored at −80°C until later analyses.¹⁶

Tracheal aspirates were analysed for CC16, proinflammatory and modulatory cytokines (IL-1β, IL-6, IL-10, monocyte chemoattractant protein (MCP)-1 and TNF-α), VEGF, the protease MMP-9 and total protein, as further described in supplement and previously.¹⁶

Lung function testing at 12 years of age

A history of respiratory symptoms, medication and allergy was obtained from a questionnaire completed by the accompanying parent¹⁹.

A static and dynamic spirometry manoeuvre was performed using a Jaeger MasterScope (Erich Jaeger, Würzburg, Germany). Forced expiratory volume in 1 s (FEV₁), forced vital capacity (FVC), mean forced expiratory flow between 25% and 75% of FVC (FEF_25–75) and slow vital capacity (VC) were measured and the ratio FEV₁/FVC was calculated.²⁰

Impulse oscillometry (IOS) was performed using the Jaeger MasterScreen IOS (Erich Jaeger). The differences in resistance between 5 Hz and 20 Hz (R₅–R₂₀), reactance at 5 Hz (X₅), resonant frequency (F_res) and reactance area (AX) were investigated. These parameters all reflect the degree of obstruction in small, peripheral airways²¹ and except for X₅, where the opposite is true, higher values indicate a more severe obstruction.

Static lung volumes, that is, total lung capacity (TLC) and residual volume (RV), as well as inspiratory, expiratory and total airways resistance were measured by body plethysmography using a MasterScreen Body, Erich Jaeger). The ratio RV/TLC was calculated as an estimate of air trapping.²²

Single-breath determination of carbon monoxide (CO) uptake in the lung during a 10 s breath-hold at maximal inspiration (diffusion capacity of the lung, DL_CO) was done using the MasterScreen PFT equipment (Erich Jaeger). The accessible alveolar volume was simultaneously measured using methane as a biologically inert tracer gas. The rate constant for uptake of CO (K_CO) was calculated as DL_CO divided by alveolar volume.²³

Multiple breath washout (MBW) of nitrogen was performed with an Exhalyzer D (Eco Medics, Duernten, Switzerland). The number of lung volume turnovers required to reach 5% and 2.5% of initial end-tidal N₂-concentration are termed lung clearance index (LCI_5.0 and LCI_2.5, respectively) and are measures of overall ventilation inhomogeneity.²⁴ S_cond×VT and S_acin×VT are indices of ventilation inhomogeneity in conductive and intraacinar airways, respectively. For all these indices, higher values indicate a more inhomogeneous ventilation.

Predicted normal values and limits of normal for spirometry, static lung volumes and diffusing capacity were calculated according to the Global Lung Initiative.^25–27 Data are presented as per cent of predicted (median and IQR) and as proportion of children outside of normal limits: upper (ULN) and lower (LLN) limit of normal. For impulse oscillometry, airway resistance by body plethysmography, and MBW, normal values were defined from our previous study²⁸ (see online supplemental file 1). In the tables, measurements are reported as absolute values, per cent of predicted normal values (FEV₁), z-scores or number of individuals <LLN, >ULN, <5th centile and >95th centile.

Statistics

Data were analysed using IBM SPSS Statistics V.28. Spearman’s r was used for correlations. Mann-Whitney U test was used to compare two independent samples. P values <0.05 were considered significant.

Results

Tracheal aspirate biomarker levels

Tracheal aspirate had been collected from a subgroup of the children (see lung function of these children, n=20, in online supplemental table S1). The median (IQR) level of CC16 in tracheal aspirate was 130 (43–384) ng/ml, IL-1β was 12.6 (1.3–51.4) pg/mL, IL-6 was 299 (73–1610) pg/mL, IL-10 was 2.1 (0.97–5.0) pg/mL, MCP-1 was 264 (147–478) pg/mL, TNF-α was 3.1 (0.7–8.3) pg/mL, VEGF was 20 (10–46) pg/mLl and MMP-9 was 20.2 (1.6–68.9) ng/mL.

Deficient lung function

The most common abnormalities of lung function in the children were expiratory flow limitation and dysfunction of peripheral airways. Lung function results of all children in the cohort who performed lung function tests and the children where tracheal aspirate had been collected are shown in online supplemental table S1, both as values and as a number of children that had results outside the normal range.

CC16 was associated with airway obstruction

Lower levels of CC16 in tracheal aspirate at birth were associated with increased airway obstruction during lung function testing at school age, shown as a positive correlation between CC16 and FEV₁/FVC, and negative correlations between CC16 and the impulse oscillometry parameters R₅–R₂₀, F_res and AX (table 2, figure 1). Even though CC16 is known to be increased during development, there were no correlations between gestation age and levels of CC16 in tracheal aspirate in the present study.

Table 2.

Correlations between lung function measurements at 12 years of age and CC16, IL-1β and TNF-α (and ratios) in tracheal aspirate at birth

	CC16 in tracheal aspirate	IL-1β in tracheal aspirate	TNFα in tracheal aspirate	IL-1β/CC16 ratio in tracheal aspirate	TNF-α/CC16 ratio in tracheal aspirate
Spirometry	(n=20)	(n=18)	(n=18)	(n=18)	(n=18)
FEV1, L	r=0.03, p=0.90	r=−0.47, p=0.048	r=−0.55, p=0.019	r=−0.52, p=0.028	r=−0.51, p=0.029
FEV1, z-score	r=0.27, p=0.26	r=−0.13, p=0.60	r=−0.21, p=0.41	r=−0.21, p=0.41	r=−0.37, p=0.13
FEV1/FVC	r=0.51, p=0.021	r=−0.17, p=0.49	r=−0.09, p=0.74	r=−0.63, p=0.005	r=−0.68, p=0.002
FEV1/FVC, z-score	r=0.50, p=0.025	r=−0.22, p=0.39	r=−0.18, p=0.47	r=−0.62, p=0.006	r=−0.71, p=0.001
FEF25–75, L/s	r=0.33, p=0.16	r=−0.39, p=0.11	r=−0.36, p=0.15	r=−0.73, p<0.001	r=−0.73, p<0.001
FEF25–75, z-score	r=0.39, p=0.094	r=−0.25, p=0.32	r=−0.21, p=0.40	r=−0.51, p=0.031	r=−0.58, p=0.011
Impulse oscillometry	(n=20)	(n=18)	(n=18)	(n=18)	(n=18)
R5, kPa×s/L	r=−0.19, p=0.43	r=0.09, p=0.71	r=−0.23, p=0.36	r=0.18, p=0.47	r=0.001, p=1.00
R20, kPa×s/L	r=−0.02, p=0.93	r=0.18, p=0.49	r=−0.12, p=0.64	r=−0.01, p=0.97	r=−0.16, p=0.53
R5–R20, kPa×s/L	r=−0.49, p=0.028	r=0.14, p=0.58	r=0.09, p=0.72	r=0.57, p=0.013	r=0.62, p=0.006
X5, kPa×s/L	r=0.37, p=0.11	r=−0.29, p=0.24	r=−0.18, p=0.48	r=−0.61, p=0.008	r=−0.53, p=0.023
Fres, 1 /s	r=−0.65, p=0.002	r=0.02, p=0.95	r=0.02, p=0.95	r=0.54, p=0.022	r=0.69, p=0.002
AX, kPa/L	r=−0.57, p=0.008	r=0.26, p=0.31	r=0.16, p=0.53	r=0.72, p<0.001	r=0.73, p<0.001
Body plethysmography	(n=18)	(n=17)	(n=17)	(n=17)	(n=17)
Rinsp, kPa×s/L	r=−0.27, p=0.27	r=0.17, p=0.52	r=0.19, p=0.47	r=0.36, p=0.15	r=0.53, p=0.029
Rexp, kPa×s/L	r=−0.26, p=0.30	r=0.30, p=0.25	r=0.25, p=0.34	r=0.60, p=0.012	r=0.60, p=0.011
Rtot, kPa×s/L	r=−0.31, p=0.21	r=0.27. p=0.30	r=0.22, p=0.39	r=0.55, p=0.023	r=0.61, p=0.010
Diffusion capacity	(n=18)	(n=17)	(n=17)	(n=17)	(n=17)
DLCO, mmol/(min×kPa)	r=0.24, p=0.33	r=−0.07, p=0.80	r=−0.03, p=0.91	r=−0.12, p=0.65	r=−0.13, p=0.61
DLCO, z-score	r=0.46, p=0.056	r=0.36, p=0.16	r=0.35, p=0.18	r=0.15, p=0.57	r=0.081, p=0.76
Alveolar volume, L	r=−0.24, p=0.34	r=−0.28, p=0.28	r=-041, p=0.10	r=−0.29, p=0.26	r=−0.15, p=0.58
Alveolar volume, z-score	r=0.09, p=0.72	r=0.06, p=0.82	r=0.17, p=0.52	r=0.13, p=0.62	r=0.26, p=0.31
KCO, mmol/(min×kPa×L)	r=0.60, p=0.008	r=0.27, p=0.30	r=0.39, p=0.12	r=0.028, p=0.91	r=−0.083, p=0.75
KCO, z-score	r=0.42, p=0.086	r=0.28, p=0.27	r=0.22, p=0.39	r=0.020, p=0.94	r=−0.15, p=0.58
Multiple breath washout	(n=19)	(n=17)	(n=17)	(n=17)	(n=17)
LCI2.5	r=−0.30, p=0.22	r=−0.09, p=0.74	r=−0.40, p=0.12	r=0.21, p=0.43	r=−0.037, p=0.89
LCI5.0	r=−0.33, p=0.17	r=−0.31, p=0.23	r=−0.53, p=0.030	r=0.022, p=0.93	r=−0.15, p=0.58
Scond×VT	r=−0.21, p=0.39	r=−0.08, p=0.76	r=0.07, p=0.79	r=0.18, p=0.48	r=0.26, p=0.31
Sacin×VT	r=−0.09, p=0.72	r=−0.53, p=0.030	r=−0.54, p=0.027	r=−0.25, p=0.33	r=−0.25, p=0.33

The table shows correlations (regression coefficients and p values) by Spearman's r between CC16, IL-1β, TNF-α, IL1-β/CC16 and TNFα/CC16 in tracheal aspirate and the various lung function measurements. The actual number (n) is given for the respective test, except for FEF_25–75 where data from one child are missing. P values <0.05 were considered significant and are marked as bold.

AX, area under the reactance curve between 5 Hz and F_res; CC16, club cell secretory protein; FEF_25-75, mean forced expiratory flow between 25% and 75% ; FEV₁, forced expiratory volume in 1 s; F_res, resonant frequency; FVC, forced vital capacity; IL-1β, interleukin-1β; LCI_5.0 and LCI_2.5, lung clearance index, defined as the number of lung volume turnovers required to reach 5% and 2.5%; R_exp, expiratory flow resistance of the airways; R_insp, inspiratory flow resistance of the airways; R_5, R₂₀, respiratory resistance at 5 and 20 Hz, respectively; R₅-R₂₀, frequency dependence of resistance; R_tot, total flow resistance of the airways; S_cond x VT, an index of ventilation inhomogeneity in conductive airways; TNFα, tumour necrosis factor-α; X₅, respiratory reactance at 5 Hz.

Figure 1.

FEV₁/FVC (A) measured by spirometry and F_res (B), R₅-R₂₀ (C) and AX (D) measured by impulse oscillometry at 12 years of age in relation to CC16 in tracheal aspirate after intubation in the first 24 hours of life. Open circles are participants who had lung function test results within the normal range while filled circles are participants with results outside the normal range. A lower FEV1/FVC and a higher F_res, R₅-R₂₀ and AX are signs of airway obstruction. The fit lines are by least squares regression including all data values. The r values and p values come from Spearman’s non-parametric correlation. FEV₁/FVC, forced expiratory volume in 1 s/forced vital capacity.

Proinflammatory cytokines were associated with airway obstruction

Higher levels of the proinflammatory cytokines IL-1β and TNF-α in tracheal aspirate were both associated with some measures of airway obstruction. Thus, IL-1β and TNF-α both showed a significant negative correlation with FEV₁ at school age (table 2). There were also significant negative correlations between IL-1β and TNF-α at birth and S_acin at school age.

There were no significant correlations between the proinflammatory cytokine IL-6, the anti-inflammatory cytokine IL-10, the chemokine MCP-1 or the protease MMP-9 in tracheal aspirate and lung function test results.

Inflammatory imbalance in tracheal aspirate

We calculated the ratios IL-1β/CC16 and TNFα/CC16 as indices of inflammatory imbalance in tracheal aspirate. Higher values of these ratios (=overweight of proinflammatory mediators) correlated significantly with more airway obstruction, that is, lower FEV₁, FEV₁/FVC, FEV₁/FVC as z-score, FEF_25–75, and FEF_25–75 as z-score. Higher ratios also correlated with higher R₅–R₂₀, lower X₅, higher F_res and higher AX as measured by impulse oscillometry, all signs of small airway dysfunction (table 2, figures 2, 3). Finally, higher ratios correlated with higher expiratory and total resistance measured by body plethysmography, but there was no correlation to static lung volumes (online supplemental table S2).

Figure 2.

FEV₁/FVC (A) and FEF_25-75 (B) measured by spirometry and R₅-R₂₀ (C) and AX (D) measured by impulse oscillometry at 12 years of age in relation to the IL-1β/CC16 ratio in tracheal aspirate after intubation in the first 24 hours of life. Lower FEV₁/FVC and FEF_25–75, and higher R₅–R₂₀ and AX, indicate increased obstruction. Open circles are participants with a result of the respective lung function test within the normal range while filled circles are participants with a result outside of the normal range. The fit lines are by least squares regression including all data values. The r values and p values come from Spearman’s non-parametric correlation. FEF_25–75, forced expiratory flow between 25% and 75%; FEV₁/FVC, forced expiratory volume in 1 s/forced vital capacity.

Figure 3.

FEV₁/FVC (A) and FEF_25–75 (B) measured by spirometry and R₅–R₂₀ (C) and AX (D) measured by impulse oscillometry at 12 years of age in relation to the TNF-α/CC16 ratio in tracheal aspirate after intubation in the first 24 hours of life. Lower FEV₁/FVC and FEF_25–75, and higher R₅–R₂₀ and AX, indicate increased obstruction. Open circles are participants with a result of the respective lung function test within the normal range while filled circles are participants with a result outside of the normal range. The fit lines are by least squares regression including all data values. The r values and p values come from Spearman’s non-parametric correlation. FEF_25–75, forced expiratory flow between 25% and 75%; FEV₁/FVC, forced expiratory volume in 1 s/forced vital capacity.

The ratio IL-1β/CC16 in tracheal aspirate was higher in newborns who later in childhood had impulse oscillometry measurements of R₅ (p=0.035), R₅–R₂₀ (p=0.038), F_res (p=0.035) and AX (p=0.019) above the normal range compared with those within normal range. The same was true for TNFα/CC16 in relation to R₅ (p=0.049), F_res (p=0.049) and AX (p=0.026), as well as FEF_25-75 (p=0.044) which was below the normal range.

The ratio of the proinflammatory IL-1β and TNFα over the anti-inflammatory cytokine IL-10 did not show any significant correlations to any of the lung function variables.

VEGF and CC16 were associated with diffusion capacity

Lower levels of VEGF in tracheal aspirate were associated with lower DL_CO as z-score (p=0.001, r=0.71, figure 4A), and a tendency to association to lower K_CO, that is, ratio of diffusion capacity per alveolar volume (p=0.095, r=0.42).

Figure 4.

DL_CO as z-score (A) and K_CO (B), measured at 12 years of age in relation to VEGF and CC16, respectively, in tracheal aspirate after intubation in the first 24 hours of life. A lower DL_CO and K_CO are signs of deficient diffusion capacity and subsequently oxygen uptake. It should be noted, however, that none of the measurements of DL_CO are below the lower limit of normal. The fit lines are by least squares regression including all data values. Dotted line: y=0. The r values and p values come from Spearman’s non-parametric correlation. DL_CO, diffusion capacity of the lung; VEGF, vascular endothelial growth factor.

In addition, lower levels of CC16 were associated with lower K_CO (p=0.008, r=0.60, figure 4B, table 2), and a tendency to association with both lower DL_CO as z-score (p=0.056, r=0.46) and K_CO as z-score (p=0.086, r=0.42).

Airway symptoms

Among the children who had tracheal aspirate collected (n=20), 10 children (50%) had a history of wheezing, 7 (35%) reported disturbed sleep because of wheezing, 4 (20%) had experienced nocturnal cough and 4 (20%) reported exercise-induced wheezing. Four children (20%) reported having allergy, eight (40%) had received a diagnosis of asthma while only four (20%) were on continuous inhalation treatment.

There were no correlations between the analysed biomarkers in tracheal aspirate on the first day of life and reported symptoms at school age.

Discussion

This study supports our hypothesis that an inflammatory imbalance in the lung in the first day of life of very preterm infants may have consequences for airway function at school age. Low tracheal fluid levels of the secretory protein CC16, believed to have anti-inflammatory properties, and high levels of the proinflammatory cytokines IL-1β and TNF-α were associated with airway obstruction at school age. This seemed to be a specific effect on airway patency as measures of lung volumes were not affected. This is in line with our previous findings of an association between high levels of the inflammatory cytokine IL-6 in blood during the first days of life in very preterm born infants and airway obstruction at school age¹⁹. In addition, low levels of VEGF and CC16 were associated with impaired DL_CO, suggesting a dysfunctional pulmonary vasculature at the age of 12 in these preterm born children.

Indices of ventilation inhomogeneity (LCI and S_cond) are sensitive markers of small airway dysfunction and have previously been reported to be increased in children born preterm when examined at school age.²⁹ A substantial proportion of children (15%–23%) in our study had values of these indices above the 95th percentile, but this was not related to any of the biomarkers in tracheal aspirate.

Impaired alveolar development goes hand in hand with dysmorphic pulmonary microvessel growth, and there is increasing evidence suggesting that lung blood vessels actively promote alveolar growth during development.³⁰ Angiogenic growth factors, such as VEGF, contribute to normal alveolar development and are expressed in lower levels during BPD.⁹ The present study shows that the low VEGF expression during infancy is associated with a lower diffusion capacity and thereby a deficient oxygen uptake during school age in these children. In addition, CC16 was associated with diffusion capacity and may also be involved in the process of angiogenesis.

There are few studies to compare with ours and, generally, it has been difficult to show that exposures or treatments that could potentially alter pulmonary outcome in very preterm infants are important for lung function at school age or in adolescence.^{31 32} The strength of our study is the long-term follow-up, but this is also a weakness as the originally relatively large group of participants diminished over time due to drop-outs. Since tracheal aspirate could only be obtained from intubated infants there could be a bias of investigating the younger and the sicker children in the total population. However, the lung function at the age of 12 was similar in the children with tracheal aspirate collected and those without (as shown in online supplemental table S1).

Tracheal aspirates are heterogeneous, and even if a strict protocol was followed, the degree of dilution may have varied somewhat between samples. Unfortunately, the only reference substance available was total protein, which also increases during inflammation. We repeated correlations after adjusting biomarker concentrations to total protein content but got very similar results (data not shown). The most consistent results were achieved by using ratios between the cytokines IL-1β and TNF-α, respectively, and CC16 in the correlations, thus eliminating the effect of varying dilution.

The comprehensive lung function testing introduced the problem of multiple comparisons in a relatively small cohort. The validity of our findings is, however, supported by the distinct and biologically plausible pattern connecting the main exposure, an imbalance between certain proinflammatory cytokines and the anti-inflammatory protein CC16 in the lung, with a distinct pattern of outcomes, that is, obstruction of large and small airways, significant with three different techniques of lung function measurement, that is, spirometry, impulse oscillometry and body plethysmography (table 2). As shown in the figures, the association was evident also among the majority of infants with lung function within a normal range. However, there are many postnatal exposures that may affect lung function over the years, and it is probably a variety of factors that influence individual outcome.

Data availability statement

Data are available on reasonable request. The data that support the findings of this study are not publicly available because that possibility was not included in the original consents. Anonymised data are available from the corresponding author on reasonable request.

Ethics statements

Patient consent for publication

Consent obtained from parent(s)/guardian(s).

Ethics approval

This study involves human participants and the Regional Ethical Review Board in Lund, Sweden, approved separately the neonatal study part and the 12-year follow-up (approval number LU 423-99 and 2013/779, respectively). All parents received oral and written information and signed written informed consents before their baby was born and, for the later part of the study, at the time of lung function testing. At the 12-years follow-up, all children received oral and written information about the study.