Airway Proteins in First Week of Life Predict Severity of Preterm Lung Disease

For the tiniest babies, the difference between a lifetime of healthy breathing and chronic lung disease may be written into the airways within days of birth. A new clinical study from the University of Michigan suggests that the molecular signature of bronchopulmonary dysplasia, the most common chronic lung disease of prematurity, is not a single inflammatory storm but a set of distinct protein patterns that differ depending on how severe the disease will ultimately become. The findings, published in Pediatric Research, point to two proteins in particular, thrombospondin 1 and periostin, as early biomarkers that could one day help neonatologists predict which ventilated preterm infants face the gravest risk.

Bronchopulmonary dysplasia, or BPD, arises when the immature lungs of infants born at or before 32 weeks of gestation are injured by mechanical ventilation, oxygen therapy, and inflammation during a critical window of development. Instead of forming the millions of delicate alveolar sacs that allow efficient gas exchange, the lungs of affected infants develop simplified architecture with impaired blood vessel growth. The consequences can last a lifetime, ranging from repeated respiratory infections and asthma-like symptoms in childhood to reduced lung function in adulthood. Yet the disease is strikingly heterogeneous: some infants recover with minimal support while others remain dependent on intensive respiratory care well past their due date, and the biological reasons for this spectrum have remained poorly understood.

The research team, led by Saly N. Essoh, Fernando A. Munoz, and senior author Antonia P. Popova, hypothesized that the answer might lie in the very first days of life. They collected tracheal aspirates, fluid samples obtained during routine suctioning of breathing tubes, from 81 preterm infants who were mechanically ventilated for respiratory distress syndrome within their first week at the C.S. Mott Children’s Hospital Newborn Intensive Care Unit. Infants with congenital anomalies or neonatal sepsis were excluded. The researchers then measured a panel of pro-inflammatory cytokines and matricellular proteins, non-structural extracellular molecules that regulate interactions between cells and their surrounding matrix and modulate immune signaling, and compared the results with each infant’s eventual BPD outcome at 36 weeks postmenstrual age.

The severity classification followed the 2019 NICHD criteria, which grade BPD by the level of respiratory support an infant requires: no BPD if none is needed, grade 1 for low-flow nasal cannula oxygen, grade 2 for higher-flow oxygen or non-invasive positive pressure, and grade 3 for invasive mechanical ventilation. In the cohort, 36 infants had no BPD, 24 had grade 1, 8 had grade 2, and 13 had grade 3. Because of the smaller numbers in the two most severe categories and no significant protein differences between them, grades 2 and 3 were analyzed together. As expected, gestational age and birth weight fell with increasing severity, from a median of 28.7 weeks and 1,190 grams in infants without BPD to 25.9 weeks and 755 grams in the grade 2/3 group.

What emerged from the protein measurements was not one uniform inflammatory signature but three clearly separable patterns. The first pattern included cytokines elevated in BPD of any severity, most prominently interleukin-17A and its upstream regulator interleukin-23, along with TNF-alpha, IL-6, and the matricellular protein galectin 1. The second pattern was specific to grade 1 disease: IL-12p70, IL-12p40, the dendritic cell growth factor FLT3L, SPARC, and periostin were elevated only in infants who went on to develop mild BPD. The third pattern was the most surprising. In infants with grade 2/3 BPD, levels of thrombospondin 1, TGFBI, and CYR61, all matricellular proteins involved in angiogenesis and tissue repair, were lower than in infants with no BPD or mild disease, not higher.

Correlation analysis reinforced the idea that these proteins operate as coordinated networks. IL-17A and IL-23 moved together almost inseparably, with a correlation coefficient of 0.88, while IL-6 and TNF-alpha formed another tightly linked pair. FLT3L, IL-12p40, IL-12p70, and IL-2 formed a third cluster consistent with a dendritic cell-driven feedforward loop, and SPARC paired with periostin. Most intriguingly, the relationships of thrombospondin 1 to these networks shifted with disease outcome. In infants who never developed BPD, THBS1 correlated positively with IL-12p70. In the grade 1 group, it correlated negatively with the IL-12 family cluster. In the grade 2/3 group, it swung back to strong positive correlations with galectin 1, SPARC, TGFBI, and IL-5, suggesting that its biological role changes depending on the inflammatory context of the injured lung.

The predictive power of these measurements was striking. Using receiver operating characteristic analysis, the team found that a thrombospondin 1 cutoff of 3 nanograms per milliliter discriminated grade 2/3 BPD with 75 percent sensitivity and 86 percent specificity, yielding an area under the curve of 0.84. Infants with THBS1 below that threshold had a 7.3-fold higher risk of severe disease, a risk that remained 4.9-fold elevated after statistical adjustment for birth weight and sex. Periostin told the opposite story for mild disease: levels above 5.5 nanograms per milliliter identified grade 1 BPD with an area under the curve of 0.78, and the elevated risk of roughly threefold persisted after adjustment. Secretory component of immunoglobulin A, measured as a reference protein to confirm that differences were not due to sample dilution, was similar across all groups.

The biology behind these patterns may explain why low THBS1 is a red flag rather than a reassuring sign. Thrombospondin 1 is a key mediator of mechanotransduction, helping lung tissue adapt to mechanical stress, and experimental models show that its deficiency leads to impaired alveolar and vascular development, heightened inflammation, and worse injury. Notably, THBS1 deficiency has been shown to enhance IL-17A-driven inflammatory responses, providing a plausible mechanism by which low levels of the protein in the first week of life could permit unopposed IL-17 inflammation and fibrosis in the most severely affected infants. Periostin and SPARC, by contrast, regulate collagen assembly and tissue remodeling, and their selective elevation in grade 1 BPD may reflect a more contained, milder repair response, though the authors caution that in more severe disease these proteins may accumulate later or remain bound to the matrix, escaping detection in soluble airway samples.

The study carries important caveats. The cohort was modest in size, sample collection spanned 13 years, and not every infant was tested for every protein due to limited sample volumes. The cross-sectional design cannot establish causality, and the statistical thresholds have not yet been validated in an independent population. Tracheal aspirates, while a direct window into airway inflammation, may not capture immune activity in the lung interstitium or the systemic circulation. Still, the authors argue that the work lays a foundation for a new generation of early biomarkers. If confirmed in larger cohorts, a simple protein measurement in the first days of life could identify which ventilated preterm infants are destined for severe BPD, opening a window for targeted anti-inflammatory therapies or matricellular protein-based interventions before irreversible damage is done, and offering hope that the earliest molecular whispers of chronic lung disease can be heard, and answered, in time.

Subject of Research: Protein biomarkers in tracheal aspirates of preterm infants associated with bronchopulmonary dysplasia severity

Article Title: Pro-inflammatory and matricellular protein expression patterns associated with bronchopulmonary dysplasia severity

Article References: Essoh, S. N., Munoz, F. A., Christner, L. P., Cui, T. X., Zhang, Y.-J., Hershenson, M. B., & Popova, A. P. (2026). Pro-inflammatory and matricellular protein expression patterns associated with bronchopulmonary dysplasia severity. Pediatric Research. https://doi.org/10.1038/s41390-026-05584-2

Image Credits: AI Generated

DOI: 10.1038/s41390-026-05584-2

Keywords: bronchopulmonary dysplasia, preterm infants, tracheal aspirate, thrombospondin 1, periostin, interleukin-17A, matricellular proteins, neonatal lung disease, biomarkers, pulmonary inflammation, tissue repair, Pediatric Research