Perinatal inflammation relates to early respiratory morbidity and lung function at 12 years of age in children born very preterm

2.1 Study population and neonatal data collection

The study population used here was a previously described prospective cohort of very preterm infants,¹⁸ born 2001–2003 and admitted at birth to the neonatal intensive care unit in Lund, Sweden. Inclusion criteria were gestational age <32 weeks, antenatal consent and no congenital anomalies, and acceptance rate was 86%. Clinical data were collected as described in Appendix S1.

Blood samples were obtained from the umbilical cord at birth and from an arterial line at 6, 24, and 72 h of age. Levels of pro-inflammatory (tumour necrosis factor-α, interferon (IFN)-γ, interleukin (IL)-1β, IL-2, IL-6, IL-8, IL-12) and modulatory (IL-4, IL-10) cytokines were quantified in plasma (see Appendix S1). Plasma levels of the respective cytokines were used to calculate area under the curve (AUC) as an assessment of cytokine burden over time (0–72 h) Levels of cytokines and the calculated AUCs were logarithmically transformed to achieve a more normal distribution for the statistical calculations.¹⁸

Clinical data (presented in Tables 1 and 2) were collected as described in Appendix S1. Bronchopulmonary dysplasia (BPD) was defined as treatment with supplemental oxygen at 36 weeks postmenstrual age.

2.2 Pulmonary function testing (PFT)

Out of 74 infants enrolled at birth, 71 survived to 36 weeks postmenstrual age. In two infants, the parents declined further participation after discharge from the neonatal unit, and one child was lost to follow-up. At 12 years of age, the remaining 68 children were invited for PFT and 54 agreed to participate (a detailed flow chart is shown in Appendix Figure S1). Reasons for not coming were parental decline, family having moved outside the reach of the investigators, or that testing would be impossible because of physical or mental disability.

The 54 children were examined at median 12.6 (range 12.3–13.5) years of age. They were asked not to take bronchodilators for at least 24 h before PFT. Height and weight were measured, and a history of respiratory symptoms, medication and allergy was obtained from a questionnaire completed by the accompanying parent.

The examination consisted of four parts: measurement of exhaled nitric oxide, spirometry, measurement of static lung volumes by whole body plethysmography and measurement of the diffusion capacity of the lung for carbon monoxide. The individual tests performed and the number of children who completed the different tests are shown in Table 3. The test procedures are detailed in Appendix S1. Thirty-nine children completed all four parts of the examination, 10 children completed three, and three children only completed two parts. One child did not cooperate and generated no results.

Fifty children underwent a second spirometry 15 min after inhalation of 200 µg salbutamol. A significant airway response to salbutamol (positive reversibility) was defined as an increase in the absolute value of forced expiratory volume in 1 s (FEV₁) with ≥12%. Except for this reversibility analysis, pre-bronchodilator spirometry results were used for all the statistical evaluations.

Results of spirometry and body plethysmography are reported as per cent of predicted by height and sex using reference equations from Zapletal.¹⁹ Results of diffusing capacity are reported as per cent of predicted by age, height and sex according to standards from the Global Lung Initiative (GLI).²⁰

2.3 Ethics

Both parts of the study, in the neonatal period and at 12 years of age, were approved by the Regional Ethical Review Board in Lund. All parents and children received both oral and written information about the study, and signed parental informed consents were obtained.

2.4 Statistics

Statistical analysis was performed using IBM® SPSS® Statistics version 26. Nonparametric tests were used due to presence of asymmetric distribution in some variables even after logarithmical transformation. Spearman's rank test was used for correlation analyses, Mann–Whitney U-test was used for comparisons between two groups, and Wilcoxon's signed-rank test was used for paired comparisons. For all correlations, associations caused by outliers were considered as non-significant. Linear or logistic regression was used to adjust for gestational age (in analyses of neonatal comorbidities). Chi-squared test or Fisher's exact test was used to explore relationships between categorical variables. p-values < 0.05 were considered significant.

3.1 Population characteristics and respiratory morbidity

The perinatal characteristics of the study population are shown in Table 1 and associations with BPD in Table 2. Infants with BPD had lower gestational age and lower birth weight than infants without BPD. They had more often been exposed to suspected maternal infection, but maternal infection was also associated with a more preterm birth (median 26.3 vs. 27.9 weeks, p = 0.004), and the association with BPD did not persist after adjustment for this. A diagnosis of BPD was preceded by a more severe lung disease, as judged by a higher need for surfactant, mechanical respiratory support and postnatal corticosteroids.

3.2 Inflammatory biomarkers and neonatal respiratory morbidity

Overall, perinatal systemic inflammation quantified as a higher plasma cytokine burden (AUC from 0 to 72 h) was associated both with a more severe neonatal lung disease in terms of need for respiratory support and with a diagnosis of BPD.

Infants who were given surfactant (n = 40; within the first hour in 25 cases) had higher AUC for IL-2 (p = 0.04), IL-6 (p < 0.001) and IL-8 (p < 0.001) compared to infants not given surfactant. These associations remained significant for IL-6 and IL-8 also after adjustment for gestational age.

The need for mechanical ventilation was mainly dependent on a low gestational age, with 18/28 (64%) infants born before 27 weeks needing ventilation for more than 10 days in contrast to none of the 43 infants born after 27 weeks or more (p < 0.001). However, cytokine burden was a significant contributor, as AUCs for IL-2 (r_s=0.33, p = 0.005), IL-6 (r_s=0.55, p < 0.001), IL-8 (r_s=0.49, p < 0.001) and IL-10 (r_s=0.24, p = 0.042) all correlated with number of days on mechanical ventilation, and for IL-6 (p = 0.001), IL-8 (p = 0.015) and IL-10 (p = 0.006), the correlations remained significant after adjustment for gestational age (Figure 1).

Infants who were later diagnosed with BPD had higher AUCs for IL-6 (p = 0.002), IL-8 (p < 0.001) and IL-10 (p < 0.001) than infants without BPD (Figure 2A-C). Individual plasma levels of IL-6, IL-8 and IL-10 were also significantly higher in infants with BPD in cord blood and at most of the following time points when the cytokines were sampled. All these differences remained significant after adjustment for gestational age.

The individual cytokine patterns were as follows: In infants with BPD, the pro-inflammatory cytokine IL-6 remained equally high at 6 h as at birth, but then fell significantly to 24 h of age (p = 0.002) and remained unchanged at 72 h. Levels of IL-6 were higher in infants with BPD at all time points except at 24 h. The other pro-inflammatory cytokine IL-8 did not change significantly from birth to 72 h but was higher in infants with BPD at all time points. In contrast, the anti-inflammatory cytokine IL-10 increased significantly from birth to 6 h of age (p < 0.001) and then fell to 24 h (p < 0.001) and further to 72 h (p = 0.001), with plasma levels at 72 h still higher than at birth (p < 0.001). The size of the initial increase and then decrease in IL-10 were not significantly different between infants with or without BPD, but levels were higher in infants with BPD at all time points except at 6 h. With the exception of a higher IFN-γ in cord blood (p = 0.022) in infants with BPD, there were no differences in plasma levels or AUCs for any of the other cytokines between infants with or without BPD.

AUC for IL-8 decreased with increasing gestational age (r_s = −0.37, p = 0.002) and with increasing birth weight (r_s = −0.49, p < 0.001), but there was no similar correlation for the other cytokine AUCs. There was no association between preeclampsia, premature rupture of membranes (PROM), suspected maternal infection or clinical chorioamnionitis and cytokine AUCs except that AUC for IFN-γ was higher after PROM (p = 0.009).

3.3 History of airway disease and lung function at school age

Fifty per cent of the 54 children who came for PFT reported at least one of four symptoms (history of wheezing, exercise-induced wheezing, sleep disturbed by wheezing, nocturnal cough), and 63% had taken medication for airway disease at some time (Appendix Table S1). Taken medication was more common in children born after PROM (81% vs. 52%, p = 0.029) or after suspected maternal infection (83% vs. 53%, p = 0.028) and less common in children born after preeclampsia (29% vs. 75%, p = 0.002) or born small for gestational age (33% vs. 74%, p = 0.005).

More children with a previous diagnosis of BPD had a diagnosis of asthma (44% vs. 17%; p = 0.03) or a history of allergy (33% vs. 10%, p = 0.04). Seventeen per cent of children had ongoing inhalation treatment, with no difference between children with or without BPD. The proportion of children reporting at least one of the four airway symptoms mentioned above was not significantly different between those having and those not having a diagnosis of BPD (60% vs. 41%, p = 0.17; see Appendix Table S1).

The background characteristics of the children who did PFT (n = 53) were similar to those of the whole cohort of survivors (n = 71, Table 1). The 18 children who did not do PFT were born at a lower gestational age and had more often been treated with systemic corticosteroids than the 53 children who did PFT (Appendix Table S2). Physical characteristics and results of PFT are shown in Table 3. Values below the lower limit of normal were more common for tests evaluating airway obstruction than for lung volumes; FEV₁/FVC was <0.75 in 21% and FEF_25-75 below 70% of predicted in 48% of the children. In contrast, measurements of lung volumes showed few results lower than normal: FVC below 80% of predicted was only seen in one child, TLC below 80% of predicted was seen in three children (7%) and alveolar volume below 80% of predicted in none. DLCO below the lower limit of normal (77% of predicted) was seen in two children, and KCO below the lower limit of normal (77% of predicted) was also seen in two children.

Median values for the different lung function measures were not significantly different between children with or without a previous diagnosis of BPD. Median FEV₁, FEV₁/FVC and FEF_25-75 were lower in boys than in girls, though not significant (p = 0.11, p = 0.10 and p = 0.07, respectively).

After inhalation of salbutamol, FEV₁ increased by a median of +6% (range −10% – +23% and p < 0.001 when comparing before and after salbutamol), but a clinically significant reversibility (≥12% increase) was seen only in 10/49 children (20%). Children who reported at least one of the four symptoms of airway obstruction mentioned above had significantly lower FEV₁ as a ratio of forced vital capacity (FEV₁/FVC; p = 0.029) and lower FEF_25-75 (p = 0.022) than children with none of these symptoms. They also had a lower diffusion coefficient for carbon monoxide (KCO, p = 0.039) and a tendency towards lower diffusing capacity of the lung for carbon monoxide (DLCO, p = 0.055).

FeNO₅₀ was below 20 ppb in 45/52 children (87%), and above 35 ppb²¹ in four children only (8%), two with and two without a diagnosis of BPD; these four infants all reported a history of wheezing and had taken medication for airway disease.

3.4 Perinatal morbidity, inflammatory biomarkers and lung function at school age

There was no association between PROM, suspected maternal infection or clinical chorioamnionitis and lung function at school age.

Within the span studied, 23.9 – 31.7 weeks, there was no association between gestational age or birth weight and PFT results, except that forced vital capacity (FVC) as per cent of predicted decreased with increasing gestational age (r_s = −0.28, p = 0.047).

Out of the three cytokines significantly associated with development of BPD (IL-6, IL-8 and IL-10; Figure 1), IL-6 correlated with a lower FEV₁ (r_s = −0.30, p = 0.030) and a lower FEF_25-75 (r_s = −0.29, p = 0.034) as per cent of predicted (Figure 3). For the other cytokines, AUCs did not correlate with lung function at school age.

Cytokine AUCs from birth to 72 h of age were not significantly different in children with a later history of airway disease (having a diagnosis of asthma, having had at least one of the symptoms of airway obstruction mentioned above, or having ever taken medication for airway obstruction) than in children without this history. A higher AUC for IL-4 correlated with a higher FeNO₅₀ (r_s = 0.30, p = 0.033), but not with any lung function measurement. None of the other cytokines correlated with FeNO₅₀, alveolar or bronchial flux of NO.

There were no associations between indicators of morbidity in the neonatal period (treatment with surfactant, levels of PCO₂ and FiO₂ at 6, 24, 48 and 72 h of age, days with mechanical ventilation, sepsis, ligation of ductus arteriosus, cumulative dose of corticosteroids) and lung function measurements at school age.