Bronchopulmonary dysplasia (BPD) is one of the most common chronic lung diseases in infants, and the incidence has continued to rise with the increasing survival of extremely premature neonates [1]. With limited options for effective and sustained postnatal therapeutic interventions, it continues to be a frustrating clinical syndrome for many neonatologists. Initially described by Northway et al. in 1967 [2] with the advent of mechanical ventilation in preterm neonates, the definition and characteristics of BPD have evolved over the past fifty years. Once thought to predominantly stem from insults to the lung, BPD is now believed to result from an arrest in lung development. The “classical” form of BPD, characterized by airway injury, pulmonary edema, inflammation, fibrosis, and smooth muscle hypertrophy, was largely attributed to prolonged mechanical ventilation and oxygen exposure in the preterm infant [3]. Over time, this has transitioned into the concept of “new” BPD, characterized by a progressive deterioration in pulmonary function, a significant decrease in alveolarization and vasculogenesis, and heterogeneous, coarse densities in the lungs [1, 4, 5]. Since June 2000, the diagnosis of BPD has changed from an assessment of an oxygen requirement at 28 days of life, to the current severity-based stratification occurring at 36 weeks post-menstrual age, which defines BPD on the infant’s ongoing need for respiratory support [6, 7]. However, despite this more nuanced definition, it has become apparent that BPD is a diverse and variable disease, not only between infants, but even within one infant throughout his or her clinical course.
There is significant morbidity and mortality associated with BPD, with affected infants showing persistent pulmonary impairment even as children and adults. They have a higher risk of developing asthma and of requiring hospitalizations for pulmonary exacerbations, as well as developmental disorders, all leading to higher healthcare costs [1]. Consequently, there have been many studies focused on addressing the pathophysiology of BPD and its possible management. Despite advances in neonatal-perinatal medicine, including antenatal steroid use, postnatal surfactant therapy, and various ventilation strategies, BPD remains a serious and persistent clinical problem. As with other common complications of prematurity, the development of BPD is multifactorial, with an infant’s predisposition, outcome severity, and treatment response widely believed to be affected by complex gene-environment interactions. As such, there have been a variety of approaches aimed at identifying specific genes or pathways that could be implicated in the development of BPD.
Multiple studies have utilized twin cohorts to address genetic and environmental influences on preterm birth outcomes, including BPD. While some have looked at the effects of birth order, gender, and intrauterine growth retardation on mortality and morbidity rates of VLBW twins, in 1996, Parker et al. was one of the first studies to significantly associate the occurrence of BPD between twins. Even after adjusting for birth weight, gestational age, gender, severity of respiratory distress syndrome (RDS), patent ductus arteriosus (PDA), pneumothorax, infection, and antenatal steroids using multiple logistic regression modeling, this study of 108 VLBW twin pairs showed that there was significant concordance of BPD status between twins. When BPD was diagnosed in the first born twin, it also occurred in 65% of the second born twins. In contrast, when BPD was not diagnosed in the first born twin, only 8% of the second born twins were diagnosed with BPD [8]. Although zygosity was unable to be established in order to delineate the effects of genetics from shared environmental factors, this novel study strongly suggested that BPD has a genetic component.
In 2006, Bhandari et al. furthered this hypothesis by broadening their study to 252 preterm twin pairs across four institutions. Not only was the concordant development of BPD between twins supported, but the authors also analyzed a subgroup for which zygosity was available. While twins typically share the same in utero environment and often, similar postnatal treatments, monozygotic twins are unique in having a higher number of shared genes when compared to dizygotic twins. Therefore, a higher incidence of similarities between monozygotic twins implies a genetic component is present. After adjusting for multiple demographic factors, Bhandari et al. demonstrated that monozygotic twins were more likely than dizygotic twins to develop BPD and require a longer length of stay in the NICU. They showed that genetic factors are a significant contributor to BPD, proposing that they account for 53% of the observed variance [9].
With the new definition of BPD proposed by the National Institute of Child Health and Human Development (NICHD)/National Heart, Lung, and Blood Institute Workshop in June 2000 based on gestational age and severity, which was clinically validated by Ehrenkranz et al. in 2005 [7], Lavoie et al. estimated heritability in 159 preterm twin pairs by calculating the contributions of genetic, shared environmental, and nonshared environmental factors. Using model-fitting analyses, the authors found an even stronger genetic component for 36-week oxygen need-based BPD and NICHD-defined BPD compared to previous studies [10]. However, a recent study published by Parad et. al. in 2018 showed no significant heritability pattern in 250 preterm twin pairs [11]. Overall, the heritability component from the majority of these twin studies is estimated to be between 50-80%, strongly suggesting that genetic influences play a role in the development of BPD. Despite this, no specific genes or heritable factors have successfully been identified.
The human genome is a complex organization of approximately 20,000 genes that encode proteins, as well as a variety of non-coding regulatory components. Only about 0.1% of the genome differs between individuals, contributing to variability. This manifests phenotypically with variations in characteristics such as height or eye color, but is also thought to contribute to differences in disease susceptibility. While some disease-causing mutations are severe, these typically occur in specific genes and are overall rare. Much more prevalent, but less understood, are minor sequence variations that may have subtle influences on the predisposition to disease development. The most common of these are single nucleotide polymorphisms (SNP), which are individual base pair substitutions that occur in approximately 1 in 1000 base pairs. Starting in 2002, international researchers collaborated to create a haplotype map, known as the International HapMap Project, which was aimed at cataloging common SNP variations in the human genome.
Genome Wide Association (GWA) studies are used to widely examine millions of SNPs between a “diseased” population and a “control” population, and have been successfully used in diseases such as diabetes, Crohn’s disease and coronary artery disease. This is inherently more difficult in the analysis of BPD since GWA studies are dependent on a large sample size and the presumption that the at-risk, or diseased, alleles, will be passed on to the patients’ offspring often enough to affect population frequency. However, BPD is often a major cause of neonatal mortality, limiting the inheritability rate of the at-risk alleles. To date, three GWA studies on BPD have been published. In 2011, Hadchouel et al. analyzed 418 extremely premature infants of Caucasian-French or African-French descent and found that allele C of the rs1245560 SNP of the SPOCK 2 (SPARC/osteonectin, CWCV, and Kazal-like domains proteoglycan 2) gene was significantly associated with moderate-severe BPD even after adjustment for perinatal factors. These findings were subsequently replicated in a Finnish cohort and further studies showed that SPOCK2 mRNA was significantly increased during the alveolar stage of development in rat pups, particularly when they were exposed to hyperoxia [12]. Wang et al. performed a GWA study on 1726 VLBW extremely premature infants, predominantly of Mexican-Hispanic origin. Their data did not show any SNP to have a statistically significant association with BPD, including those from prior studies [13]. The most recent study, by Ambalavanan et al., used 834 infants to combine their GWA with pathway-based approaches, including early death as a competing outcome for BPD. No SNP was found to be significant for any of their comparisons, although a few were close to statistical significance, including adenosine deaminase (ADARB2) and CD44. However, their data were notable for the implication that the pathways associated with severe BPD or death are different than those for mild or moderate BPD [14].
In contrast, candidate gene studies focus on polymorphisms for specific genes that are hypothesized to be involved in the development of particular diseases. In the case of BPD, success has been limited, partly due to small sample sizes but also to the prevailing idea that BPD is likely to be polygenic. There are published results for genes encoding for surfactant proteins, mannose-binding lectin, tumor necrosis factor, interferon gamma, angiotensin converting enzyme, interleukins, growth factors, superoxide dismutases, toll-like receptors, macrophage migration inhibitory factor, human leukocyte antigens, and vascular endothelial growth factor, among others. Overall, these have had inconsistent results. Although some studies may express significant results for a particular gene, the results have generally not been reproducible, with statistical significance eliminated when using larger sample sizes in follow-up studies [15].
A more recent genetic approach has focused on small genomic components known as microRNA (miR), which are a relatively new family. First discovered in the 1990’s in C. elegans, miRs are endogenous, single-stranded, non-coding RNAs that are only about 20 nucleotides in length. It was initially believed that there were approximately 200 to 1000 miR genes in the mammalian genome, contributing to 1 to 3% of known genes, but as of 2009, there were at least 11,000, with the number continuing to increase with time. Recent discoveries have shown that after transcription and processing in the nucleus, and subsequent exportation to the cytosol, mature miR is incorporated into the RNA-induced silencing complex (RISC) that post-transcriptionally regulates gene expression via partial or complete complementarity to the 3’ untranslated region of target messenger RNA molecules. Depending on the level of complementarity, this process results in an inhibition of the initiation of translation, an inhibition of peptide elongation, or an induction of mRNA degradation [16, 17]. In the past two decades, miRs have been found to have important roles in a variety of biological processes such as developmental timing, proliferation, differentiation, signaling, inflammation, tissue morphogenesis, and cell death. In addition, they have been implicated in a wide variety of diseases, such as cancer, cardiovascular diseases, and a number of childhood diseases, including pediatric respiratory diseases. Early studies have shown a change in expression of a few miR during lung development [18].
Of note, a number of miRs have been implicated in the regulation of autophagy, and they have begun to be considered as novel therapeutic targets for lung diseases [19]. Autophagy is induced by hyperoxia in lung epithelial cells and neonatal mouse lungs, and thus, appears to play an important, protective, anti-apoptotic role in the development of BPD [20]. Autophagy was initially described in rat liver cells in the 1960’s and autophagy-related genes were first discovered in the 1990’s. Stemming from the Greek words for “self” and “eating”, autophagy is an evolutionarily conserved, intracellular catabolic process where a eukaryotic cell self-cannibalizes long-lived proteins, protein aggregates, dysfunctional organelles, and even intracellular pathogens in a lysosomal-dependent manner to allow for reuse of the degraded products. Although this occurs at a basal level under normal homeostatic conditions, it is activated in response to physical and chemical stresses, such as starvation, hypoxia, and infection, and has been found to be actively induced in almost all tissues during the early neonatal period [21]. Previous studies have reported that autophagy can inhibit or promote cell death both in developmental and diseased conditions, including cancer, neurodegenerative disorders, and metabolic and infectious diseases [22]. It has been shown that the deletion of essential autophagy genes (beclin [BECN] 1, and autophagy-related [ATG] 5) in mice causes early embryonic or neonatal lethality, underscoring the importance of autophagy to the developmental process. It is also believed to be involved in the pathogenesis of various preclinical models of pulmonary diseases, as well as in humans [20].
Briefly, the process of autophagy involves a complex reorganization of subcellular membranes to form a lipid isolation membrane that elongates and sequesters cytoplasm and organelles into a double-membrane cytosolic vesicle known as an autophagosome. The autophagosome fuses with a lysosome to form an autophagolysosome, inside which lysosomal hydrolases degrade the sequestered material to be recycled by the cell [22]. Unsurprisingly, the complex process of autophagy is regulated by a diverse network that consists of different signaling pathways and autophagy-related genes, which were first discovered in the 1990’s in yeast. Much of the known regulation occurs at the post-translational level, predominantly via phosphorylation or acetylation, but transcriptional control of autophagy genes is an area of more recent discovery. Interestingly, recent literature has focused on the regulation of autophagy via non-coding RNAs, especially microRNAs (miR) [19].
One study demonstrated that miR-34a levels are significantly increased both in the lungs of neonatal mice and Type 2 alveolar epithelial cells of humans exposed to hyperoxia. In addition, miR-34a overexpression worsens the pulmonary phenotype and BPD-associated pulmonary arterial hypertension in BPD mouse models [23].