Relationship between bronchopulmonary dysplasia phenotypes with high-resolution computed tomography score in early preterm infants

Qiong Yao, Quanli Shen, Guoying Huang, Xihong HuFrontiers in Pediatrics 2022 · DOI

TL;DR

This study demonstrates that high-resolution computed tomography (HRCT) scoring is a powerful tool for phenotyping bronchopulmonary dysplasia (BPD) in preterm infants. By quantifying lung damage into hyperaeration and parenchymal scores, researchers found that specific CT features can not only assess BPD severity but also help predict secondary complications like pulmonary hypertension (PH).

What Problem Does This Paper Address?

Bronchopulmonary dysplasia (BPD) is a chronic lung disease common in premature infants, but it isn't a "one size fits all" diagnosis. It is highly heterogeneous, involving varying degrees of lung tissue (parenchymal) damage, vascular impairment (pulmonary hypertension), and airway collapse (tracheobronchomalacia).

Currently, BPD severity is mostly graded based on how much supplemental oxygen an infant needs at 36 weeks postmenstrual age. However, this clinical metric doesn't reveal the underlying structural damage. The authors sought to determine if HRCT—a high-precision imaging tool—could provide a more objective "structural fingerprint" to classify BPD phenotypes and improve clinical management.

Methods

The researchers conducted a retrospective, single-center study of 81 preterm infants (born at ≤32 weeks) diagnosed with BPD. They utilized a modified HRCT scoring system that evaluated six pulmonary lobes across seven types of lesions, categorized into two main groups:

  1. Hyperaeration Score (HS): Focused on "air trapping" markers like decreased attenuation, mosaic attenuation, and bullae/blebs.
  2. Parenchymal Score (PS): Focused on structural tissue changes like linear lesions, consolidation, bronchial wall thickening, and bronchiectasis.

The Total Score (TS) was the sum of all seven parameters. These scores were then correlated with clinical data, including BPD severity (mild, moderate, severe) and the presence of complications like pulmonary hypertension (PH) and tracheobronchomalacia (TBM).

Datasets

The study used a retrospective clinical dataset from the Children’s Hospital of Fudan University covering the years 2013 to 2020.

  • Initial pool: 376 BPD cases.
  • Final cohort: 81 infants who met strict inclusion criteria (gestational age ≤32 weeks, HRCT performed between 40–50 weeks postmenstrual age, and high-quality imaging).
  • Demographics: 56 males, 25 females; average gestational age of ~29 weeks.

Key Insights and Ideas

  • Hyperaeration as a Severity Marker: The Hyperaeration Score (HS) was a strong indicator of overall BPD severity. This suggests that air trapping and alveolar simplification are the most sensitive structural changes in "new" BPD.
  • Imaging the Vasculature Indirectly: One of the most significant findings was that parenchymal features—specifically linear densities and consolidation—correlated with pulmonary hypertension. This suggests that structural scarring in the lung tissue is closely linked to the impairment of the pulmonary vascular bed.
  • The "Triple Threat" Phenotype: Infants who suffered from a combination of parenchymal disease, PH, and TBM exhibited the highest total HRCT scores, highlighting the cumulative impact of these multi-system complications.
  • The Limitation of Static Imaging: Interestingly, HRCT scores did not significantly differ for infants with TBM. The authors suggest this is because TBM is a dynamic airway collapse, while the HRCT scans were static (taken at end-inspiration), likely missing the collapse that occurs during expiration.

Key Results

  • Severity Correlation: Infants with severe BPD had significantly higher Total Scores ($p = 0.01$), Hyperaeration Scores ($p = 0.02$), and Parenchymal Scores ($p = 0.02$) compared to mild/moderate cases.
  • Pulmonary Hypertension (PH): PH was present in nearly half the cohort (49.38%). Higher Total Scores ($r = 0.28$) and Parenchymal Scores ($r = 0.35$) were positively correlated with a PH diagnosis.
  • Specific Lesions: Linear densities ($p < 0.01$) and mosaic attenuation ($p = 0.03$) were specifically higher in more severe BPD groups.
  • Reliability: The scoring system showed high inter-observer and intra-observer agreement (Cronbach’s $\alpha$ between 0.85 and 0.90), suggesting the scoring method is robust and reproducible.

Limitations and Future Work

  • Selection Bias: Because this was a retrospective study, infants who did not receive an HRCT (often those with milder BPD) were excluded, potentially overestimating the prevalence of certain abnormalities.
  • Sample Size: The number of infants with TBM was relatively small (n=20), which may have limited the statistical power to find correlations with CT scores.
  • Radiation Concerns: HRCT involves radiation exposure, which is a significant concern in neonates. The authors suggest that future work should focus on "ultra-low dose" CT protocols or dynamic expiratory imaging to better capture airway issues like TBM.
  • Lack of Longitudinal Data: The study did not analyze long-term clinical outcomes (e.g., lung function at school age), which would be the "gold standard" for validating these imaging biomarkers.

Summary updated

Abstract

HRCT scores correlated with BPD severity and PH in our study. HS might be a useful tool in the assessment of BPD severity while linear densities and consolidation might be helpful in predicting PH.
BibTeX
@inproceedings{yao2022relationship,
  author = {Qiong Yao and Quanli Shen and Guoying Huang and Xihong Hu},
  title = {{Relationship between bronchopulmonary dysplasia phenotypes with high-resolution computed tomography score in early preterm infants}},
  booktitle = {Frontiers in Pediatrics},
  year = {2022},
  doi = {10.3389/fped.2022.935733},
}
Figure