Relationship Between Airway Metabolites and Structural Damage in Young Children with Cystic Fibrosis Open Access

Portelli, Alexandria (Spring 2018)

Permanent URL:


Background: Cystic Fibrosis (CF) is a genetic disease affecting over 70,000 people worldwide. Airway disease, the main cause of morbidity and mortality in CF, has been found to begin soon after birth. The Perth-Rotterdam Annotated Grid Morphometric Analysis (PRAGMA-CF) method, used to score chest computed tomography (CT) scans, is a sensitive and reproducible measure of the extent of lung disease in CF children. There is limited knowledge of relationships between PRAGMA-CF scoring of chest CT scans and molecular biomarkers of disease measured by metabolomics of bronchoalveolar lavage fluid (BALF) in CF children.


Objective: Identify significant statistical correlations and relationships between the PRAGMA-CF score of overall structural airway damage (PRAGMA-%Dis, or %Dis) in CF children and specific BALF metabolites.


Methods: Univariate and multivariate biostatistical methods were used to assess a longitudinal dataset from a prospective study of CF children (I-BALL study) to identify BALF metabolites associated with airway damage. Pearson and Spearman correlation coefficients of %Dis and metabolite concentration were calculated for cohorts at ages 1, 3, and 5 years old. Linear mixed models assessed the response of metabolite concentration to covariates including %Dis, age, total BALF protein concentration and % BAL neutrophils.


Results: Significant correlations and linear relationships between the concentration of specific BALF metabolites and %Dis were identified. Trends in the Pearson and Spearman correlations coefficients change in the first 5 years of children born with CF. The linear mixed model including %Dis and total BALF protein concentration was chosen because it had the lowest Akaike information criterion (AIC) overall across most metabolites. %Dis showed significant positive associations with diacyl (aa) and acyl ether (ae) phosphatidylcholines (PCs) aa C30:0, aa C34:1, aa C36:2, ae C34:0 and ae C34:1, and sphingomyelin SM C16:0 when adjusting for total BALF protein concentration.


Conclusions: These results add to our growing understanding of early CF pathogenesis, and how metabolomics can be used to generate clinically-relevant molecular outcomes for disease monitoring at a stage when conventional biomarkers remain at low to undetectable levels. More extensive investigation of age as a covariate is needed on progression of early CF per the %Dis outcome.

Table of Contents




Study Cohort: I-BALL Study

Data Acquisition

Targeted Metabolomics

PRAGMA-CF Scoring of Chest CT Scans

Data Cleaning

Metabolomics Data



Descriptive Characteristics of Study Cohort

Univariate Correlation Analysis

Pearson Correlation

Spearman Correlation

Linear Mixed Modeling

Assessment of Reproducibility of Metabolite Measurements

Computing Environment


Summary Statistics

Pearson Correlations

Spearman Correlations

Linear Mixed Modeling

Akaike Information Criterion

Linear Mixed Model with PRAGMA and Total Protein

Lin’s Concordance Correlation Coefficient


Interpretation of Results


Further Research



Appendix 1: List of Metabolites Detected by AbsoluteIDQ® p180 Kit

Appendix 2: Histograms Comparing Distributions

Appendix 3: Scatterplots of Significant Metabolites


About this Master's Thesis

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
  • English
Research Field
Committee Chair / Thesis Advisor
Committee Members
Partnering Agencies
Last modified

Primary PDF

Supplemental Files