Improving early diagnosis of bronchopulmonary dysplasia

ABSTRACT

Introduction

Bronchopulmonary disease (BPD) is associated with long-term neurodevelopmental and cardiorespiratory complications, often requiring significant use of resources. To reduce this healthcare burden, it is essential that those at high risk of BPD are identified early so that strategies are introduced to prevent disease progression. Our aim was to discuss potential methods for improving early diagnosis in the first week after birth.

Areas covered

A narrative review was undertaken. The search strategy involved searching PubMed, Embase and Cochrane databases from 1967 to 2024. The results of potential biomarkers and imaging modes are discussed. Furthermore, the value of scoring systems is explored.

Expert opinion

BPD occurs as a result of disruption to pulmonary vascular and alveolar development, thus abnormal levels of factors regulating those processes are promising avenues to explore with regard to early detection of high-risk infants. Data from twin studies suggests genetic factors can be attributed to 82% of the observed difference in moderate to severe BPD, but large genome-wide studies have yielded conflicting results. Comparative studies are required to determine which biomarker or imaging mode may most accurately diagnose early BPD development. Models which include the most predictive factors should be evaluated going forward.

KEYWORDS:

• Biomarkers
• bronchopulmonary dysplasia
• chronic lung disease
• genetics
• imaging
• prediction scoring
• prematurity

1. Introduction

Bronchopulmonary disease (BPD) is a form of chronic lung disease first described by Northway et al. [1] over 50 years ago. It is the most common complication of very preterm birth [2]. It is usually diagnosed in infants who have an oxygen requirement at 28 days after birth and then the severity determined at 36 weeks post-menstrual age (PMA) according to the level of respiratory support required at that age [3]. That definition, however, can result in difficulty reaching a timely diagnosis when the pulmonary phenotype is often variable [4]. A study of 1340 infants born between 23 and 27 weeks of gestation identified three patterns of lung disease which tended to emerge within the first two weeks in extremely low gestational age newborns (ELGANS): minor lung disease and rapid recovery, early and persistent pulmonary dysfunction and pulmonary deterioration i.e. those who recover in week one but rapidly deteriorate in week two [5]. They found that nearly 40% of ELGANS experienced pulmonary deterioration in the first two weeks after birth and half of them developed BPD. A subsequent study, using the same definitions described four distinct patterns of respiratory disease with different prevalences of BPD. The risk factors for early deterioration were male sex, lower gestational age, higher incidence of severe RDS and lack of postnatal caffeine [6] In a study of 2677 infants, 26% of whom developed serious respiratory morbidity or died, 18 prespecified, revised definitions of BPD were evaluated as to which best predicted death or serious respiratory morbidity through 18 to 26 months’ corrected age. It was found the best predictor was the definition which categorized disease severity according to the mode of respiratory support administered at 36 weeks PMA regardless of supplemental oxygen use [7].

Several antenatal and postnatal risk factors have been identified. Maternal risk factors include maternal pregnancy-induced hypertensive disorders [8], chorioamnionitis [9] and smoking [10]. Weight and gestational age at birth and intrauterine growth restriction all increase the risk of developing BPD [11]. The incidence increases as gestational age and weight at birth decreases. Improvements in neonatal care have resulted in increased survival of very preterm infants leading to a higher incidence of BPD [12]. Postnatally, volume induced lung injury secondary to respiratory support [13], the duration of exposure to mechanical ventilation [14], late onset sepsis [15], patent ductus arteriosus [16] and genetics [17] all play a role. Twin studies, however, have shown that not all at-risk premature infants develop BPD [18], suggesting BPD is a unique lung injury syndrome characterized by the amalgamation of antenatal exposures, postnatal oxygen and ventilation-mediated injury and other insults to an immature and developing lung [19].

BPD occurs as a result of disruption to pulmonary vascular and alveolar development, leading to ineffective gas exchange [20]. Histopathology studies show that a BPD lung is characterized by large, simplified alveolar structures, a dysmorphic capillary configuration and variable interstitial cellularity and/or fibroproliferation [21].

BPD is associated with long-term neurodevelopmental and cardiorespiratory complications, often requiring significant use of resources and funding. To reduce this healthcare burden it is, therefore, essential that those at high risk of BPD are identified early, as there may be a critical window in which to introduce strategies to prevent disease progression [22]. The aim of this review is to discuss potential methods for improving early diagnosis within the first week after birth.

2. Methods

The search strategy of this narrative review involved searching PubMed, Embase and Cochrane databases from 1967 to January 2024 for articles published in peer reviewed journals. An initial search was performed using PubMed to identify the majority of relevant papers published on the subject. The following search strategy was used: (‘Bronchopulmonary disease’ OR ‘Neonatal respiratory disease’ OR ‘Neonatal lung disease’ OR ‘Respiratory distress syndrome’) AND (‘Early diagnosis’ OR ‘Newborn screening’ OR ‘Diagnostic techniques’ OR ‘Biomarkers’ OR ‘Imaging techniques’ OR ‘Lung function tests’ OR ‘Pulmonary function tests’ OR ‘Neonatal intensive care’ OR ‘investigations’) AND (‘First week of life’ OR ‘Early infancy’ OR ‘less than 7 days’)

The search process was iterative, with refinements made based on initial findings and insights gained from key articles. We opted to concentrate on the outcomes of randomized controlled trials (RCTs), well-designed retrospective studies and systematic reviews, with a particular focus on results published after 2010 and studies with large sample sizes. Animal studies were incorporated to establish biological plausibility. Through this method, a diverse range of articles, reviews, and clinical trials were summarized to inform the understanding of early diagnostic approaches in bronchopulmonary disease. We excluded non-peer reviewed articles and those not published in English.

3. Risk factors

3.1. Prenatal risk factors

Disrupted placentation leading to compromised placental perfusion is a common factor contributing to early fetal growth restriction (FGR), gestational hypertension and preeclampsia [11]. In the French EPIPAGE-2 study involving preterm infants born below 32 weeks of gestation, there was a more than sixfold increased risk of moderate/severe BPD in infants with FGR, odds ratio [OR] 6.6; confidence interval [CI] (4.1–10.7) [23]. The study also linked FGR to a specific vascular phenotype of BPD associated with more pulmonary hypertension leading to increased morbidity and mortality. Evidence from retrospective cohort studies attributed an increased risk of BPD in mothers with placental vascular abnormalities on histology, (OR 2.6, CI 1.4–4.8) [24]. This was thought to be secondary to an imbalance of angiogenic growth factors, inflammation and oxidative stress which potentially could be measured as early as day one after birth [25].

Lung capillary angiogenesis and maturation is a critical process regulated by the angiopoietin (ANG)/Tie-2 ligand/receptor system interacting with the vascular endothelial growth factor (VEGF) pathway. VEGF acts on cells through two distinct tyrosine kinase receptors, VEGF receptor (VEGFR)-1 and VEGFR-2 [26]. In a prospective study involving 102 preterm infants (gestational age (GA) <32 weeks), it was observed that ANG-1 concentrations in umbilical cord blood (UCB) were lower in infants with BPD compared to those without BPD (p < 0.001) [27]. The ANG-1 levels in UCB were found to have an inverse relationship with endostatin levels in newborns diagnosed with BPD (r = −0.48; p = 0.008). Endostatin, a 20-kDa angiogenesis inhibitor derived from the C-terminal, non-triple helical domain of collagen XVIII, plays a role in angiogenesis. Amongst very low birth weight (VLBW) infants with a gestational age <32 weeks, those who developed BPD exhibited higher levels of endostatin in UCB than those who did not develop BPD (100.7 ± 29.7 ng·mL⁻¹ versus 85.6 ± 28.7 ng·mL⁻¹; p = 0.029) [28]. Those findings suggest that prenatally reduced ANG-1 and elevated endostatin levels in UCB are risk factors for the development of BPD. Angiopoietin-2 (ANG-2) is an angiogenic growth factor shown to destabilize blood vessels in preterm infants [29]. Elevated ANG-2 levels in tracheal aspirates (TA) within the first week of life have been associated with an increased incidence of BPD or mortality in ventilated preterm newborns [30]. Moreover, administration of dexamethasone significantly decreased ANG-2 levels. Endothelin-1 (ET-1), a potent endothelial-derived vasoconstrictor and bronchoconstrictor is found at highest concentrations in the lungs [31]. The concentrations of ET-1 on day 3 after birth in newborns who developed BPD were significantly higher than those who did not develop BPD (p < 0.0001). Furthermore, higher levels of Fibroblast Growth Factor-2 in the tracheal aspirate on day one were associated with a poorer prognosis for the development of BPD or death [32].

Placental growth factor (PlGF), a member of the VEGF family, regulates angiogenesis by modifying VEGF activity [33]. Murine studies have demonstrated that the overexpression of PlGF in transgenic mice led to increased alveolar type II cell apoptosis, resulting in expanded airspace and pulmonary emphysema, pathologically comparable to BPD [34]. Higher UCB levels of PlGF were significantly and independently associated with an increased risk of BPD [33]. More recent studies have shown that higher plasma VEGF and PlGF levels were associated with BPD and death within the first 28 days of life [35]. Extremely preterm infants with BPD associated pulmonary hypertension at a corrected age of 36 weeks had lower UCB levels of PlGF and granulocyte colony-stimulating factor (G-CSF) than subgroups of infants with BPD only and those without BPD or pulmonary hypertension. Low VEGF levels in tracheal aspirates combined with high soluble VEGFR (sVEGFR)-1 levels on the first day after birth were associated with BPD development in VLBW infants requiring oxygen and mechanical ventilation [36]. Additionally, preterm newborns who later developed BPD exhibited significantly increased sphingosine 1-phosphate concentrations in tracheal aspirates on day one after birth [37]. Whilst these factors help us to understand the pathophysiology of the development of BPD, the results are from research studies. Thus, further work is necessary to determine which of them might be cost-effective in routine practice in the early diagnosis of infants at highest risk of BPD development.

3.2. Genetics

Data from twin studies suggests genetic factors could be attributed to 82% of the observed difference in moderate to severe BPD, although no association was found with mild BPD. In an exome-sequencing study of 50 pairs of twins, rare genetic variants associated with BPD were identified [38]. These rare variants showed an enrichment in pathways related to lung development in murine models. Retrospective studies of individual genes also suggested that polymorphisms within the endothelial nitric oxide synthase pathway were independent predictors of an increased risk of developing BPD [39]. A large genome-wide study (GWAS) in extremely low birthweight infants investigated over 1.2 million genotyped SNPs and an additional seven million imputed single nucleotide polymorphisms (SNPs) using a DNA repository of ELBW infants [40]. They identified that multiple SNPs in the CD44 and phosphorus oxygen lyase pathways were significantly increased in infants who developed BPD. Assessing the results of genetic studies is challenging due to the effects of epigenetics, copy number variations, the effects of multiple SNPs or interactions between them. A GWAS study failed to identify any SNPs at the genome-wide significance level associated with BPD in 899 infants with moderate to severe BPD compared to 827 controls [41]. Similar results were observed in a replication cohort. An ancestry and genome-wide association study aimed to identify variants linked to survival without BPD in 387 high-risk infants who underwent inhaled nitric oxide treatment in the Trial of Late Surfactant [42]. A broader African genetic ancestry was correlated with an elevated chance of survival without BPD, with admixture mapping pointing to chromosome bands 18q21 and 10q22 in maternal African American infants. The most significant individual variant identified was located within the intron of NBL1, a gene expressed in mid-trimester lungs. In a separate GWAS study conducted by the GEN-BPD Study Group, associations were identified between SNPs located near the gene responsible for C-reactive protein (CRP) and BPD development [43]. Those associations were observed in two cohorts, one Finnish and the other French African which strengthened the findings.

4. Biomarkers

4.1. Umbilical cord blood (UCB) markers

Epithelial cells, known as club cells, line the respiratory bronchioles and are thought to be markers of alveolar remodeling and hypoplasia in BPD. CC16 is a club cell secretory protein which belongs to the secretoglobulin family [44]. CC16 significantly reduces pulmonary inflammation and enhances airway regeneration in animal studies [45]. Moreover, it stimulates epithelial proliferation and provides protection against oxidative stress [46]. Preterm infants with low concentrations of CC16 in UCB were at increased risk of developing BPD (p < 0.01) [47]. Reduced CC16 levels may indicate the onset of lung damage, a crucial factor in the progression of BPD.

Interleukin (IL)-6 is associated with inflammation and immunological control. It binds to receptors and acts on the signaling membrane glycoprotein 130 (gp130), initiating a cascade of downstream signals and regulating gene expression [48]. A study assessing the levels of IL-6 and soluble gp130 in the UCB of 134 preterm infants indicated that those biomarkers might serve as predictors of BPD (area under the receiver operator curve (AUROC) = 0.849) [49].

CD4+ regulatory T-cells (Tregs) have an immunosuppressive function and are crucial in maintaining immune homeostasis. Studies show they are involved in various chronic, inflammation-mediated respiratory disorders [50]. In newborns with a gestational age less than 29 weeks, increased Treg frequencies in preterm infants who developed BPD were found in days four to ten after birth (p = 0.034) [51]. In another study assessing UCB mononuclear cells in preterm infants with a gestational age of less than 32 weeks, the total number of Tregs and non-Tregs in infants who went on to develop moderate BPD was significantly lower than in those with mild or no BPD [52].

4.2. Gastric aspirate

Spectral data of gastric aspirate samples taken from birth in a multicenter study of 61 preterm infants born between 24 and 31 weeks of gestation were analyzed [53]. Artificial intelligence was used to develop a predictive algorithm for development of BPD using the gastric aspirate results and clinical data. The point-of-care test developed using this new software algorithm had a sensitivity of 88% and specificity of 91% for early diagnosis of BPD.

4.3. Urine

Urinary metabolomics and spectral analysis identified five discriminant urinary metabolites in the urine of 18 preterm infants born below 29 weeks of gestation who developed BPD compared to matched controls [54]. Importantly, urine samples were collected within 24 and 36 hours of birth. One of the metabolites identified, myoinositol, has been associated with malformation of lung tissue during maturation [55]. In a study of 96 preterm infants, urinary Beta-2-microglobulin (B2M) levels were collected at birth and 34% of the infants went on to develop BPD. Multivariate logistic regression demonstrated urinary B2M levels correlated with a diagnosis of BPD [56]. Of note, mothers of infants who developed BPD had a higher occurrence of chorioamnionitis which is consistent with inflammation which results in B2M release. The urine proteomics of ELGANs who developed BPD at a single center was investigated. Urine was collected within 72 hours of birth and they were able to identify several BPD associated changes in the urine proteome which mirrored blood proteome changes [57].

A prospective pilot study investigated whether 8-hydroxy-2’-deoxyguanosine (8-OHdG), a marker of oxidative stress, and N-terminal pro-brain natriuretic peptide (NT-proBNP) concentrations in the urine were predictors of BPD in 165 preterm infants [58]. Levels of 8-OHdG and NT-proBNP were significantly higher in the BPD group compared to the control group from days 7 to 28. Moreover, there was a positive correlation between 8-OHdG and NT-proBNP levels (r: 0.655–0.789, p < 0.001), and both 8-OHdG and NT-proBNP levels were positively correlated with the duration of mechanical ventilation and supplementary oxygen exposure (r: 0.175–0.505, p < 0.05) from days 7 to 28.

4.4. Tracheal aspirates and oral secretions

Assessment of tracheal aspirate transcriptomic and microRNA (miRNA) signatures in 55 mechanically ventilated extreme preterm infants with BPD and in 27 term babies receiving invasive mechanical ventilation for elective procedures, identified 22 miRNAs and 33 genes differentially expressed in those with BPD [59]. In a similar study in which mRNA was isolated from mesenchymal stromal cells derived from tracheal aspirates, SPARC, a specific matricellular protein, was identified as an independent predictor of BPD or death (p < 0.02) [60]. Flow cytometry on myeloid cell populations derived from tracheal aspirates prospectively collected from intubated patients born before 29 weeks of gestation and less than 30 days old was performed to identify and characterize CD14+CD16+ (double-positive) and CD14+CD16− (single-positive) monocytes. Their gene expression profiles were examined through RNA sequencing and quantitative polymerase chain reaction (PCR). Expression of interleukins including IL-1A, IL-1B, and IL-1 receptor antagonist mRNA was increased in monocytes collected on days 7, 14 and 28 compared with those collected on day three. In a study investigating permissive hypercapnia in preterm infants, interleukin levels in tracheal aspirates were not modified by high pCO₂ levels [61]. Data from a multicenter trial, however, demonstrated that inhalation of nitric oxide in preterm infants with evolving BPD was associated with reduced interleukin levels in tracheal aspirates, suggesting inflammation was reduced, but there was no improvement in survival without BPD [62].

MiRNAs are a class of small, non-coding RNAs which regulate target gene expression. They play important roles in modulating oxidative stress, cell differentiation, apoptosis, inflammatory responses and angiogenesis [63]. Dysregulation of MiRNAs has been found in infants with BPD [64]. Their use as early predictors of the development of BPD is enhanced as they can be found in blood and tracheal aspirates and indeed clinical studies have demonstrated blood levels of MiRNA were increased on day 28 in those with severe BPD [65]. Furthermore, certain MiRNAs were decreased in tracheal aspirates of premature infants with severe BPD [66,67]. Clearly, further studies are required to determine the robustness of early prediction of BPD from tracheal aspirates, but an important caveat is that tracheal aspirates are only robustly sampled when infants are intubated and nowadays even very prematurely born infants are usually supported by noninvasive ventilation.

In a single center study, using previously banked samples, oral secretions (OS) and tracheal aspirates (TA) proteomes were compared in infants admitted to the NICU born at less than 32 weeks of gestation [68]. In samples collected in the first month, 607 proteins unique to OS and 327 proteomes unique to TA were identified. Thirty-seven OS proteins that showed significantly differential abundance between BPD cases and controls were demonstrated. The utility of OS needs to be explored in further studies.

4.5. Exhaled breath condensates

Nitrite levels in exhaled breath condensate (EBC) in ventilated premature infants during the first week after birth correlated with tracheal aspirate nitrite levels (r = 0.45, p = 0.025) and both levels were higher in infants who developed BPD compared to those who did not [61]. Assessment of EBC is advantageous in that it is noninvasive and it will be interesting to see if there are proteomic and metabolic markers in those with BPD, as have been found in those with pneumonia or congenital diaphragmatic hernia [69].

4.6. End-tidal carbon monoxide

Several studies have investigated whether end-tidal carbon monoxide (ETCO) is a biomarker for BPD [70]. In a cohort of 50 preterm infants born at a gestational age of less than 32 weeks, sequential ETCO measurements revealed elevated levels in infants who later developed BPD [71]. The study identified ETCO levels exceeding 2.15 ppm on day 14 after birth as a potential predictor of BPD. Another study by our group, involving 78 preterm infants with a GA less than 33 weeks categorized into mild BPD (n = 12), moderate BPD (n = 15), severe BPD (n = 12) and no BPD (n = 39), demonstrated that ETCO on day 14 was a sensitive and specific predictor of BPD [72]. Additionally, a study exploring the association of ETCO levels with the severity of respiratory distress syndrome and BPD found that infants who developed BPD had elevated ETCO levels (>2.5 ppm) during the first 12 hours after birth compared to those who did not [73]. Those results suggest this noninvasive assessment might be give an early prediction of the development of BPD.

5. Imaging

5.1. Chest radiograph

Northway et al. reported four stages of chest X-ray (CXR) findings. As the definition of BPD has evolved with increased survival of preterm infants, so have the corresponding radiographic changes. Toce et al. devised a scoring system encompassing lung expansion, interstitial densities, focal emphysema, and cardiovascular abnormalities [74] and proposed that the most specific time for diagnosing BPD on CXR was two weeks after birth. After the introduction of surfactant, CXR changes in infants with BPD were again re-defined [75]. Arai et al. reported that a bubbly/cystic appearance on the chest radiograph at 28 days was an independent risk factor for prolonged oxygen dependency and mechanical ventilation in a study of over 8000 infants [76]. A retrospective cohort study found that an interstitial pneumonia pattern on the day seven chest radiograph was independently associated with BPD or death before 36 weeks PMA (odds ratio [OR] 4.0, 95% confidence interval [CI] 1.1–14.4), with a specificity of 98% [77].

5.2. Computed tomography scans (CT)

A number of groups have investigated whether changes on CT scans might be diagnostic. A scoring system for CT findings correlated with the clinical severity of BPD [78]. Significant associations were found between CT scores and clinical BPD severity (r = 0.855; p < 0.001) [79]. A different high-resolution CT scoring system correlated better with the clinical severity of BPD than a CXR score [80]. Furthermore, an association between CT scores and the requirement for home oxygen therapy following discharge from the NICU was noted. An alternative PRAGMA-BPD scoring system was developed and its associations with early postnatal growth and ventilatory function investigated [81]; 95.5% of the chest CT scans at six months of age showed architectural distortion of the lung in infants born at less than 32 weeks of gestation who developed severe BPD. There are currently no studies which have assessed CT findings in the first week after birth. Dynamic-volume CT, also known as cine CT, involves utilizing CT scans to observe the lungs over a brief timeframe, employing a minimal dose for each rotation to gather imaging data [82]. The benefit lies in its ability to depict alterations in the lungs and large airways throughout a respiratory cycle. May et al. described this new approach as a technique for determining positive end-expiratory pressure in infants with BPD; it is possible this could be used both diagnostically in the future for early prediction of BPD [80].

5.3. Lung ultrasound

An expanding body of literature has highlighted the potential role of lung ultrasound (LUS) carried out between the first three and 28 days after birth in predicting BPD at 36 weeks PMA [83]. Recently, the use of lung ultrasound has evolved into what is referred to as the ‘functional’ method, employing semi-quantitative lung ultrasound scores (LUS) [84]. A multicenter, longitudinal cohort study in neonates with a gestational age of 30 + 6 weeks or younger demonstrated that the mean LUS on days 1, 7, 14 and 28 of life differed significantly between 72 infants who developed BPD and the 75 who did not. There were significant correlations between LUS scores, oxygenation and the work of breathing at all time points (p < 0.0001) [85]. Adjusted for gestational age, LUS scores at days 7 (AUROC 0.826–0.833; p < 0.0001) and 14 (AUROC 0.834–0.858; p < 0.0001) significantly predicted BPD development. Furthermore, the severity of BPD and gestational age adjusted LUS scores were significantly correlated at 7 and 14 days (p < 0.0001). Another study found that the LUS trajectory was dependent on gestational age, highlighting that clinicians must bear this in mind if using lung ultrasound as a tool for monitoring [86]. The predictability of LUS in the development of BPD was investigated in a meta-analysis of the results of seven studies which involved over 1,000 neonates less than 32 weeks gestational age and demonstrated good prediction of BPD at 7 and 14 days (AUROC, 0.85–0.87; pooled sensitivity, 70–80%; pooled specificity, 80–87%) [87].

5.4. Lung MRI

Ultra-short echo time MRI carries no radiation risk and provides high-definition 3D visualization of lung structures as well as results of respiratory dynamics [88]. A significant correlation was found between MRI scores and BPD severity [89]. Moreover, in multivariable models, MRI scores were the best predictor of the duration of respiratory support, surpassing clinical data such as birth weight and gestational age. A limitation is that MRI scans have usually been performed at 36 weeks PMA or older. Further studies are required to evaluate early MRI scanning in the prediction of BPD.

5.5. Echocardiography

Echocardiography can be used to detect early signs of pulmonary vascular disease associated with BPD, namely pulmonary hypertension (PH). In a prospective study, PH on the day seven echo was identified in 42% of 277 preterm infants with birthweights between 500 and 1,250 g. Early PH was a risk factor for increased BPD severity (relative risk, 1.12; 95% confidence interval, 1.03–1.23) [90]. Infants with late PH that is at 36 weeks PMA had a greater duration of supplementary oxygen and increased mortality in the first year after birth [88]. In another study, right ventricular (RV) areas and right ventricular fractional area change (RV-FAC) from the first 24 hours after birth to 36 weeks PMA in preterm infants with no or mild BPD were compared to those with moderate to severe BPD [91]. Although no significant differences were noted between the two groups during the initial three days after birth in RV end-diastolic and end-systolic areas, both areas were significantly increased at 32 weeks PMA in infants with moderate or severe BPD. Similarly, no significant differences in RV-FAC between these groups was observed on days one and three; however, by 32 weeks PMA, RV-FAC notably decreased in infants with moderate and severe BPD compared to the control cohort. Those differences persisted at 36 weeks PMA. Similar observations at 36 weeks were documented by Sehgal et al. [92], but other studies did not confirm those findings [93]. Another measure of cardiac function is pulsed wave myocardial performance index (PW-MPI). Elevated RV PW-MPI has been documented at 7, 21, and 28 days of life in infants who developed BPD compared to those who did not [94]. Higher PW-MPI values were associated with increased severity of BPD after 28 days [95], but not during the initial two weeks after birth. In multivariate analysis, RV PW-MPI on days 7 [94] and 28 [96] emerged as independent predictors of BPD or death. Other measures include tricuspid annular plane systolic excursion (TAPSE) which has been shown to be significantly lower on day seven in infants who died or developed BPD compared to controls who did not develop BPD [97]. Dasgupta et al. evaluated currently used echo measures and concluded that all premature infants with evolving BPD should be screened prior to 36 weeks PMA with functional echocardiography which should encompass left ventricular systolic eccentricity index (LV-sEI), pulmonary artery acceleration time (PAAT), TAPSE, RV FAC, strain imaging, tissue Doppler, and the Tei index [98]. This is in addition to the standard echocardiographic parameters such as tricuspid regurgitation (TR) jet velocity and evaluation of the interventricular septum curvature during systole. The most appropriate timing of echocardiography and sequence of longitudinal assessments to detect premature infants with evolving BPD and PH remains unclear and further studies are needed to guide current practices.

6. Prediction scores

APGAR scores of less than 8 at 1 and 5 minutes have been shown to correlate with increasing severity of BPD [99]. Several models have identified Apgar scores as significant predictors for BPD development [100,101]. Dassios et al. used cluster analysis to conduct a retrospective whole population study of all 10,197 extremely preterm infants born in UK neonatal units between 2014 and 2019 [101]. Gestational age, Apgar score at 5 minutes and duration of mechanical ventilation were used as input variables with BPD as a primary outcome measure. Infants were classified into four discrete phenotypic clusters with Cluster 4 having the highest incidence of BPD, lowest gestational age and longest duration of ventilation.

The Eunice Kennedy Shriver National Institute of Child Health developed a predictive tool for BPD using data from over 3000 infants born at 17 centers between 23 and 30 weeks of gestation [102]. They validated models for six postnatal ages using risk factors which included gestational age, birth weight, race and ethnicity, sex, respiratory support, and fraction of inspired oxygen (FiO₂). The accuracy of prediction increased with progressing postnatal age, rising from a C statistic of 0.793 on day one to a peak of 0.854 on day 28. Gestational age was found to significantly enhance outcome prediction on postnatal days one and three, while the type of respiratory support demonstrated an improvement on postnatal days 7, 14, 21, and 28.

Clinicians are often keen to close a PDA in ventilator dependent infants to prevent development of BPD, but the evidence to close all PDAs is weak. A meta-analysis of 32 studies used Bayesian modeling to create Bayes factors, where BF₁₀ is the ratio of the probability that PDA was associated with BPD [103]. Evidence for an association with any PDA was weakly positive (BF₁₀ = 2.90; 10 studies), association with a hemodynamically significant PDA was moderate (BF₁₀ = 3.77; 3 studies), but the association with surgically ligated or catheter-occluded PDA was very strong (BF₁₀ = 294.9; 16 studies). PDA exposure time correlated with BPD development and should be tested in risk prediction models. In a recently reported multicenter RCT of 653 infants less than 28 weeks and 6 days of gestation with large PDAs (diameter > 1.5 mmm early treatment (≤72 hours after birth) with ibuprofen was not associated with a significant reduction in the primary outcome of death or moderate or severe BPD. Unfortunately, serial echocardiograms were not collected so this limited the interpretation of the effect of short duration on outcomes [104].

In a national study, data from a cohort of infants born between 2009 and 2010 at less than 32 weeks of gestation or with a birth weight of less than 1501 g were analyzed [105]. A scoring system was developed using predicting factors. The score was validated on the national cohort born between 2014 and 2015. Gestational age, birth weight, antenatal corticosteroids, surfactant administration, proven infection, patent ductus arteriosus, and duration of mechanical ventilation were identified as autonomous predictors of BPD at 28 days. In the derivation cohort, the AUROCs for BPD risk scores were 0.90 and 0.89 for BPD development at 28 days and BPD at 36 weeks PMA, respectively. These figures were mirrored in the validation cohort, where the AUCs were 0.92 and 0.88 for BPD at 28 days and BPD at 36 weeks, respectively.

In a study of 435 preterm infants from a single center fourteen variables that could potentially predict BPD were assessed. The three most promising predictors, gestational age, duration of mechanical ventilation and the serum concentration of N-terminal-pro-brain natriuretic peptide (NT-proBNP) in the first week after birth were selected through screening for the training set. Utilizing these data, the researchers developed a nomogram to evaluate the risk of BPD on day seven after birth resulting in an AUC of 0.85 [106].

In a study using data in the first three days after birth from 652 infants, birth weight, gestational age, gender, presence of respiratory distress syndrome, patent ductus arteriosus, intraventricular hemorrhage and hypotension were the most important risk factors for BPD. Infants with scores below 4 constituting 4.1% of the cohort (18 out of 436), and of those with scores above 9, 100% (29 out of 29) developed BPD. The successful validation of the score was conducted in a subset of 172 infants [107].

Most recently, exome sequencing on a cohort of 245 infants born before 32 weeks, of which 131 were diagnosed with BPD and 114 without BPD were controls, was performed [108]. A gene burden test was employed to identify risk gene sets, characterized by loss-of-function mutations or overrepresented missense mutations in BPD and severe BPD patients. Using machine learning, two predictive models were subsequently devised and validated assessing the risk of BPD and severe BPD, integrating both clinical and genetic features. Thirty genes were identified in the BPD cohort, while 21 genes were identified in the severe BPD cohort. The predictive model for BPD, which integrated both the risk gene sets and fundamental clinical risk factors, exhibited superior discrimination compared to a model solely reliant on clinical features (AUROC, 0.915 versus AUROC, 0.814, p = 0.013) in an independent testing dataset. A similar trend was observed in the predictive model for severe BPD (AUROC, 0.907 versus AUROC, 0.826; p = 0.016).

7. Artificial intelligence

Over the past decade, there has been investigation into the use of machine learning models and artificial intelligence in early prediction of BPD. In one study, a predictive model was devised using clinical factors such as gestational age, blood transfusions, frequency of surfactant administration and data pertaining to patent ductus arteriosus [109]. Machine learning tools, including regression coefficients, were employed to establish a categorical risk scoring system, which in turn facilitated the creation of risk categories. The strengths of the study lie in its high AUC (0.932) and its application in clinical settings. In a more recent study, multivariate logistic regression analysis was conducted using a stepwise approach for selecting risk factors, while the least absolute shrinkage and selection operator (LASSO) method was utilized for factor selection [110]. A nomogram model was then developed using the R Package ‘rms’ to forecast BPD outcomes. The model exhibited strong performance, with an AUC of 0.910 in the training set and 0.9051 in the validation cohort. This study compared two predictor selection methods and external validation of the risk factors. Both of the studies [109,110], however, were limited by being single center studies. Another recent study similarly employed the LASSO method via five-fold cross-validation to identify predictive blood serology proteins for BPD prognosis [111]. It had an AUC of 0.96 in the test cohort suggesting the integration of computational and molecular techniques may become useful in early diagnosis of BPD.

A recent systematic review and meta-analysis with external validation of models identified 64 studies using 53 prediction models developed using predictors available before day 14 after birth in very preterm infants [112]. Data from 274,407 infants were analyzed. Most studies were single center, conducted before 2010 and the review found that 97% of the studies had a high risk of bias, particularly in the analysis domain. Thirty-three models were externally validated and meta-analysis identified Laughon’s day one model as the most promising in predicting BPD and death, with a C-statistic of 0.76 and good calibration. The results were corroborated in a further a systematic review and meta-analysis published a few months later which reported a median C-statistic of 0.77 in the 65 studies reviewed [113]. Overall, the studies highlighted the scarcity of contemporary, validated and dynamically predictive models for BPD in very preterm infants.

8. Conclusion

The most robust markers of infants at high risk of BPD development in the first week after birth remain to be identified. The results of a number of prenatal research studies have further explained the mechanisms resulting in BPD development, but are unlikely to influence current clinical care. Lung ultrasound is a promising emerging technique due to its availability at the cotside. Given the varying phenotypes of BPD, a personalized approach to prevention would seem the best option, but further genetic studies are required to guide such an approach.

9. Expert opinion

Bronchopulmonary disease (BPD) is associated with long term neurodevelopmental and cardiorespiratory complications. To reduce this healthcare burden, it is essential that those at high risk of BPD are identified early so that strategies, preferably in the first week after birth are introduced to prevent disease progression. BPD occurs as a result of disruption to pulmonary vascular and alveolar development and abnormal levels of factors regulating those processes have been found which help explain the mechanisms of BPD development. Those results are, however, unlikely to influence current clinical practice unless further studies identify the most accurate predictor and the assays are available outside a research setting. Although data from twin studies suggests genetic factors can be attributed to 82% of the observed difference in moderate to severe BPD, the results of large genome-wide studies are conflicting, some studies suggesting the no SNPS are associated with BPD development. At this stage, the results of genetic studies are unlikely to influence personalized medicine. A number of possible biomarkers have been investigated from samples obtained in the first week, but comparative studies are required to identify which may give the most accurate diagnosis. The results of imaging techniques can discriminate between infants with different degrees of BPD severity, but this is usually after the first week after birth and even as late as 36 weeks PMA. Scoring systems are attractive as they give a numerical assessment of an individual infant, but to date there has been no consistent inclusion of the various risk factors. Indeed, the net should may be widened to include the results of biomarkers and exome sequencing.

More immediately clinically relevant predictors are lung ultrasound and echocardiography. Lung ultrasound is a promising technique as it can be performed at the cotside and is noninvasive, but studies determining whether the results are predictive in the first week need to be performed. Echocardiography is invaluable in determining the onset and progression of pulmonary hypertension in infants with BPD, but its usefulness in the first week to identify the most high-risk population requires further investigation.

If it were possible to predict those infants who would go on to develop BPD, targeted treatments that have been shown to prevent BPD should be offered to improve their respiratory outcomes. Such infants may particularly benefit from volume targeted ventilation. A Cochrane review of use of dexamethasone after seven days to facilitate extubation demonstrated that dexamethasone reduced the incidence of BPD [114]. Some individual trials exploring the prophylactic administration of azithromycin to premature infants have shown a modest improvement in rates of BPD [115]; however, when analyzed in a meta-analysis, the benefit was attenuated [116]. The prophylactic treatment of PDAs within the whole population of infants born prematurely has not been shown to reduce BPD [117], however, having a PDA for more than 10 days was shown to be a risk factor for BPD development [118]. A further area of interest is use of stem cells, in one trial including ventilated infants born between 22 and 28 weeks of gestation, severe BPD was significantly reduced from 53% to 19% [119]. Vitamin A supplementation may reduce BPD, particularly in those with low vitamin A uptake, although this would have to be balanced by the need for repeated intramuscular injections [120]. Caffeine has become a mainstay of treatment within neonatal units due to its effect on apnea of prematurity, as well as the reduction in rates of BPD [121].

Despite many trials designed to assesses whether treatments reduced the rates of BPD, few have shown significant benefits with limited evidence for postnatal sildenafil, diuretics and inhaled corticosteroids and nitric oxide [122]. Studying such therapies in a high-population identified by prediction scoring may identify a treatment effect that is lost in studies that incorporate all prematurely born infants.

Importantly, it should be considered are we asking the correct question, that is how can BPD be robustly diagnosed in the first week after birth? BPD is associated with long-term abnormalities at follow-up and increased healthcare costs, but infants who did not develop BPD can also have problems at follow-up including long term respiratory morbidity. Indeed, strategies which have reduced the development of BPD have not resulted in lower rates of chronic respiratory morbidity and other interventions which have not reduced the development of BPD have improved respiratory outcomes at follow-up [123]. Over the next five to ten years, we need to determine what are the robust markers in the first week after birth of the development of chronic respiratory morbidity. Such results will inform the development of relevant strategies and their appropriate evaluation. Using BPD as an outcome has been attractive as the results are available relatively quickly and using chronic respiratory morbidity as the outcome will have funding implications but provide more important results for the infants, their carers and healthcare.

Article highlights

• BPD occurs as a result of disruption to pulmonary vascular and alveolar development resulting in large, simplified alveolar structures and a dysmorphic capillary configuration.

• Abnormal levels of factors regulating those processes have been found. Lung capillary angiogenesis and maturation is regulated by the angiopoietin (ANG)/Tie-2 ligand/receptor system interacting with the vascular endothelial growth factor (VEGF) pathway. ANG-1 concentrations in umbilical cord blood (UCB) were lower in infants that went on to develop BPD, whereas Angiopoietin-2 (ANG-2) levels, which destabilizes blood vessels in preterm infants, were elevated in tracheal aspirates (TA) in the first week after birth. Placental growth factor (PlGF) regulates angiogenesis by modifying VEGF activity. Higher levels of PlGF in UCB were associated with an increased risk of BPD.

• Data from twin studies suggests genetic factors can be attributed to 82% of the observed difference in moderate to severe BPD, although no association was found with mild BPD. Large genome-wide studies (GWAS), however, have yielded conflicting results some failing to identify any single nucleotide polymorphisms (SNPs) associated with BPD development.

• Possible biomarkers can be harvested from UCB, gastric or tracheal aspirates and exhaled breath condensates. The attractiveness of such sources is they can be accessed in the first week after birth, but as yet there are no comparative studies to identify which may give the most accurate diagnosis.

• The results of imaging techniques may also be useful and include chest radiographs, computed tomography (CT) scans, magnetic resonance imaging (MRI), lung ultrasound and echocardiography, but there is insufficient evidence to determine which will be most useful in the first week.

• There have been a number of predictive models reported incorporating different risk factors, including biomarkers and the results of exome sequencing may improve their accuracy.

Declarations Of interest

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Additional information

Funding

This paper was not funded.