Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Jul 1.
Published in final edited form as: J Cyst Fibros. 2018 Nov 23;18(4):507–515. doi: 10.1016/j.jcf.2018.10.016

Urinary Metabolomics Reveals Unique Metabolic Signatures in Infants with Cystic Fibrosis

BT Kopp a,b,*, E Joseloff c, D Goetz d, B Ingram e, SL Heltshe g,h, DH Leung f, BW Ramsey g,h, K McCoy a, D Borowitz c,d
PMCID: PMC6533170  NIHMSID: NIHMS1514345  PMID: 30477895

Abstract

Background:

Biologic pathways and metabolic mechanisms underpinning early systemic disease in cystic fibrosis (CF) are poorly understood. The Baby Observational and Nutrition Study (BONUS) was a prospective multi-center study of infants with CF with a primary aim to examine the current state of nutrition in the first year of life. Its secondary aim was to prospectively explore concurrent nutritional, metabolic, respiratory, infectious, and inflammatory characteristics associated with early CF anthropometric measurements. We report here metabolomics differences within the urine of these infants as compared to infants without CF.

Methods:

Urine metabolomics was performed for 85 infants with predefined clinical phenotypes at approximately one year of age enrolled in BONUS via Ultrahigh Performance Liquid Chromatography-Tandem Mass Spectroscopy (UPLC-MS/MS). Samples were stratified by disease status (non-CF controls (n=22); CF (n=63, All-CF)) and CF clinical phenotype: respiratory hospitalization (CF Resp, n=22), low length (CF LL, n=23), and low weight (CF LW, n=15).

Results:

Global urine metabolomics profiles in CF were heterogeneous, however there were distinct metabolic differences between the CF and non-CF groups. Top pathways altered in CF included tRNA charging and methionine degradation. ADCYAP1 and huntingtin were identified as predicted unique regulators of altered metabolic pathways in CF compared to non-CF. Infants with CF displayed alterations in metabolites associated with bile acid homeostasis, pentose sugars, and vitamins.

Conclusions:

Predicted metabolic pathways and regulators were identified in CF infants compared to non-CF, but metabolic profiles were unable to discriminate between CF phenotypes. Targeted metabolomics provides an opportunity for further understanding of early CF disease.

Keywords: metabolites, biomarkers, CF

INTRODUCTION

Care for patients with cystic fibrosis (CF) remains challenging due to altered expression of the cystic fibrosis transmembrane conductance regulator (CFTR) in multiple organ systems throughout the body.[1] Although the advent of newborn screening for early detection of CF and the institution of preventative therapies has dramatically increased survival in CF,[2] infants and young children with CF have limited therapeutic options as treatments that increase CFTR expression are not yet widely available in these populations.[3] While evidence of early systemic inflammation in CF has been demonstrated in all CF genotypes,[4-6] it remains unclear why infants with similar genotypes have varied phenotypic expression of respiratory and nutritional disease. Additionally, biologic pathways and specific metabolic mechanisms underpinning early systemic disease in CF are poorly understood and have not been examined in a large-scale manner.

The Baby Observational and Nutrition Study (BONUS) was a prospective multi-center study of infants with CF with the primary aim to examine the current state of weight gain and linear growth in the first year of life. Its secondary aim was to prospectively explore concurrent nutritional, metabolic, respiratory, infectious, and inflammatory characteristics associated with early CF anthropometric measurements. We report here metabolomics differences within the urine of these well-characterized infants compared to infants without CF. We hypothesized that exploratory analysis of urinary metabolite profiles would generate new testable hypotheses, differentiate between pre-specified CF phenotypes of poor growth and respiratory complications, and identify key biochemical pathways that differentiate CF infants from infants without CF.

METHODS

Patients and Study Design

BONUS is a prospective cohort study conducted at 28 US Cystic Fibrosis Foundation (CFF) accredited care centers within the CFF Therapeutic Development Network (TDN), with methods previously described.[7] Samples from full-term infants who had a negative newborn screen (healthy non-CF infants) were collected contemporaneously to BONUS at a single general pediatric practice site during the same time period. [8] Non-CF infant samples were collected using the exact collection protocol as the CF infants. Length and weight measurements were performed at each visit for the BONUS cohort by trained and certified staff.[9] Length and weight measurements were performed by clinical staff for the healthy non-CF infant cohort. Clinical and demographic information was recorded at each visit. Urine for this study was collected at 12 months of age and immediately flash frozen in liquid nitrogen and transported for storage at −80°C for metabolite preservation. Samples were collected over the three-year recruitment period of BONUS and stored for approximately two years before analysis. No freeze-thaw cycles occurred prior to metabolite determination. Patients were not required to fast at urine collection. Written informed consent was obtained from all participating parents/guardians, and all participating sites received Institutional Review Board approval. The BONUS study was registered at ClinicalTrials.gov (NCT01424696).

Growth definitions:

Attained weight and length for age z-scores were calculated using World Health Organization standard growth curves. [10, 11] In the absence of a standard definition of ‘poor growth’ in infants with CF, persistently low weight or low length were defined as a z-score < −1.28 (10th percentile) for more than half the yearly measurements AND at least one of these low measurements occurring between 6 and 12 months of age.[7]

Metabolomics analyses:

We sought to identify biochemical differences in the urine of 85 infants with and without CF with varying, but well-defined clinical phenotypes according to the BONUS study design definitions. Groups included all subjects with 12-month urine specimens available who met these criteria: 1) CF infants with low length but normal weight (CF LL, n = 23), 2) CF infants with low weight but normal length (CF LW, n = 15) and 3) CF infants hospitalized for respiratory conditions with normal weight and length (CF Resp, n=25). CF groupings were pre-specified to designate groups with the largest expected phenotypic differences at one year of age. A large enough group of asymptomatic CF infants who did not fit any of these 3 sub-groupings was not available for comparison. Healthy non-CF infants were also included (Non-CF, n = 22). Urine was analyzed for 872 compounds of known identity (named biochemicals) identified on a global metabolomics platform (Metabolon, Inc., Morrisville, NC) and 819 compounds of known identity identified on the complex lipids platform. All CF and non-CF samples were analyzed simultaneously to avoid batch effects.

Ultrahigh Performance Liquid Chromatography-Tandem Mass Spectroscopy (UPLC-MS/MS):

Details regarding sample preparation, data collection, and compound identification can be found in the online data supplement and have been previously published.[12]

Statistical analysis:

Metabolite levels were normalized according to urine osmolality. The normalized data was analyzed following log transformation and imputation of missing values, if any, with the minimum observed value for each compound. Welch’s two-sample t-test was used to identify biochemicals that differed significantly between experimental groups. An estimate of the false discovery rate (FDR) was calculated to account for multiple comparisons (q-value < 0.10) to determine final significance. P values were determined statistically significant at less than 0.05. Random forest analysis was carried out as previously reported.[13] To determine which variables (biochemicals) made the largest contribution to the group classification, a “variable importance” was computed using Mean Decrease Accuracy. Sparse Partial Least Squares discriminant analysis (sPLS-DA) was performed with standard methodology. Standard statistical analyses are performed in ArrayStudio on log transformed data. For those analyses not standard in ArrayStudio, the programs R (http://cran.r-project.org/) or JMP were used. Pathway analysis was performed using Ingenuity Pathway Analysis software (IPA) and MetaboAnalyst (4.0).

RESULTS

Demographics

BONUS enrolled 231 infants with CF (48% female, 57% Phe508del homozygous, 91% pancreatic insufficient; a completed description of the overall cohort is available [7]). A total of 63 CF patients and 22 non-CF controls had urine available for untargeted metabolomics and targeted lipidomics analyses. Demographics of this cohort are listed in Table 1. Briefly, these included CF and non-CF infants enrolled at approximately 2 months of age, with urine collected and analyzed at 12 months for metabolomics and correlation with clinical phenotypes. CF infants were mostly pancreatic insufficient with severe genotypes. Both groups were predominantly non-Hispanic Caucasians.

Table 1:

Patient Demographics

CF
(n=63)		 Non-CF
(n=22)
Gender, female 46.0% 50.0%
Age enroll (months) 2.5 ± 0.7 2.1 ± 0.2
Severe genotype 85.7% N/A
Pancreatic insufficiency 95.2% N/A
Pseudomonas 28.6% N/A
MSSA 69.8% N/A
Open in a new tab

Severe genotype: 2 mutations in class I-III

Global CF urine metabolomics profiles are heterogeneous, but unique compared to non-CF

Metabolites were normalized to urine osmolality and compared between CF groups and to non-CF controls. The number of significant biochemical differences between the CF and non CF cohorts was found to be significantly greater than that observed when comparisons were made between the individual CF groups, with greater differences noted in the global metabolomics platform (E-table 1, up to 254 significant biochemicals) compared to the complex lipid platform (E-table 2, up to 95 significant biochemicals). A complete listing of the significant and nonsignificant biochemicals is available in E-table 3. We then used sPLS-DA to predict discriminative differences in metabolomes classified by grouping. There was significant overlap and heterogeneity amongst the CF sample groups by 2D sPLS-DA (Fig. 1A) and 3D sPLS-DA (Fig. 1B), with separation of CF and non-CF groups in both models. Loadings plots for the variables selected by the sPLS-DA model for the top four components of the 3D sPLS-DA are displayed in Figure 2, with variables ranked by the absolute values of their loadings. There was no change in separation among CF groups when up to 15 components were chosen for analysis.

Figure 1: Global CF urine metabolomics profiles are heterogeneous, but unique compared to non-CF.

Figure 1:

Open in a new tab

2-D Sparse Partial Least Squares discriminant analysis (sPLS-DA) of serum metabolomics profiles for patients with CF and non-CF controls. CF patients were grouped by low length but normal weight (CF LL, n = 23), low weight but normal length (CF LW, n = 15) and hospitalization for respiratory conditions with normal weight and length (CF Resp, n=25). Grouping are indicated by colors and shading. Metabolomics performed by UPLC-MS/MS. B) 3-D sPLS-DA with 4-component analysis for groupings in 1A.

Figure 3: Targeted metabolite analysis differentiates CF phenotypes from non-CF.

Figure 3:

Open in a new tab

Random forest biochemical importance plot of metabolites found to be important in discriminating between A) CF Resp versus non-CF, B) CF LL versus non-CF, and C) CF LW versus non-CF. The top 30 biochemicals are presented in order of increasing importance to group separation. Random Forest Confusion Matrixes of predicted classification for each comparison are also presented. Figure 3A predicted accuracy 87%, 3B predicted accuracy 91%, and 3C predicted accuracy 92%.

Figure 2: Variables selected by the sPLS-DA model for a given component.

Figure 2:

Open in a new tab

Loadings plots for the ten metabolites selected for each of the 4 components chosen in Figure 1B sPLS-DA. Color plots demonstrate relative expression of individual metabolites within groups from high (dark red) to low (dark green). LL, LW, and Resp represent CF groups compared to non-CF controls.

Random forest analysis of metabolic profiles differentiates CF phenotypes from non-CF

Next, we used random forest classification to determine if the CF groups could be distinguished from each other and the non-CF based on their global metabolic profiles. Random forest resulted in high predictive accuracies when discriminating between the profiles of each CF group and the non-CF groups (87-92% as compared to 50% by random chance alone, Figs. 3A-C). The biochemical importance plots in the figures show the top 30 metabolites based on their importance in separating the experimental groups in each comparison. The metabolites are represented in descending order from the top with respect to their importance to group separation. The top 2 metabolites for the CF LL and CF LW groups were the same, 4-ethylphenylsulfate and 4-hydroxyphenylacetylglutamine. 4-hydroxyphenylacetylglutamine is a derivative of glutamic acid (glutamate) while 4-ethylphenylsulfate is an organic acid derivative. Additionally, 12/30 top metabolites overlapped between the CF LL and CF LW groups in comparison to non-CF. The CF Resp group had different top metabolites from CF LL and CF LW compared to non-CF including 1,3-propanediol and asmol. 1,3-propanediol is a building block for polymers that can be bioconverted from glycerol by bacteria and plays a role in pulmonary vascular permeability. Asmol is a metabolite of the bronchodilator albuterol.

In contrast to the comparisons involving CF and non-CF groups, there were limited differences obtained via random forest analysis when comparing within CF groups (E-Fig. 1). Random forest confusion matrixes generated for each comparison demonstrated low association accuracy (35-60%, E-Fig. 1A-C) for differentiating between CF groups. The top 30 metabolites identified for each within CF group comparison are displayed in E-Fig. 1.

Metabolomics identifies unique canonical pathways altered in infants with CF

To further determine the biologic significance of altered urinary metabolites in CF compared to non-CF, pathway analysis was performed. The top canonical pathways altered in CF in comparison to non-CF are shown in Table 2. These included changes in transfer RNA (tRNA) charging and methionine degradation. tRNA charging occurs when an amino acid is loaded on a tRNA so that a ribosome can transfer the amino acid from the tRNA onto a growing peptide. Methionine is an essential amino acid in mammals that has been demonstrated to be decreased in CF plasma during exacerbations. [14] However, total methionine levels were not significantly different between groups.

Table 2:

Top altered canonical pathways in CF

Canonical Pathway Significant Groups vs Non-CF
tRNA charging *All CF, CF Resp, CF LL
(S)-reticuline biosynthesis *All CF, CF Resp, CF LL
4-hydroxybenzoate biosynthesis *All CF, CF Resp
Methionine degradation *All CF, CF Resp, CF LL
Phenylalanine degradation *All CF, CF LW
L-DOPA degradation CF Resp, CF LL
Catecholamine biosynthesis CF LL
Bile acid biosynthesis CF LW
Glycine degradation CF LW
TCA cycle CF LW
Glycine betaine CF LW
Open in a new tab
*

All CF denotes a significant pathway when comparing the entire BONUS cohort versus non-CF

Pathway analysis was further interrogated to determine the top predicted regulator effect networks. Adenylate cyclase activating polypeptide 1 (ADCYAP1) was the top regulator network in CF Resp and CF LL groups in comparison to non-CF (Fig. 4A). Activation of ADCYAP1 is predicted to lead to increases in signaling molecules such as dopamine and cyclic AMP that may activate downstream pathways of neuroendocrine stress responses such as cell stimulation, nucleotide metabolism, and calcium concentrations. Huntingtin was the top regulator biochemical in the CF LW group compared to non-CF (Fig. 4B). Huntingtin is found is many tissues throughout the body, but is highly concentrated in the brain and is required for normal brain development. Activation of Huntingtin is predicted to activate neuronal responses, cell stimulation, and inhibit T lymphocytes activation.

Figure 4: Top regulator effect networks in CF.

Figure 4:

Open in a new tab

Pathway analysis was performed on significantly altered metabolites between CF groups and non-CF. The top regulators and their effect networks are displayed for A) CF Resp and CF LL versus non-CF and B) CF LW versus non-CF. A color-coded prediction legend displays predicted activation/inhibition as well as intensity of measurement.

CF is associated with altered urinary bile acid metabolites

We have previously demonstrated altered bile acid metabolism in CF children[15] and adults. [12] Bile acid signaling in CF can alter bacterial composition and tolerance to antibiotics, [16] therefore we examined bile acid metabolism in CF infants. Significant increases in primary bile acids were seen in all three CF groups compared to non-CF (pathway E-Fig. 2A, results E- Fig. 2B). The observed increases in glycocholate are similar to our prior studies.[12, 15] Both significantly increased and significantly decreased secondary bile acid metabolites were seen in CF groups compared to non-CF, while the CF Resp group displayed significant increases in the secondary bile acids deoxycholate, glycodeoxycholate, and glycodeoxycholate sulfate compared to the CF LW group (E-Fig. 2B). The observed secondary bile acid metabolite changes are unique to the CF infants in comparison to prior studies, but the CF Resp group may more closely resemble adult CF patterns.11

Increased fat and non-fat soluble vitamin metabolites in CF urine

CF is associated with fat malabsorption, and patients with CF and pancreatic insufficiency are required to take fat-soluble vitamin replacement therapy.[17] We analyzed fat and non-fat soluble vitamin metabolites in the CF groups compared to non-CF. Over 95% of the CF infants were on vitamin supplementation at the time of collection. Significant increases in the non-fat soluble vitamin C (ascorbate) were observed in all three CF groups compared to non-CF along with significant increases in some of its downstream degradation products (oxalate and threonate, Fig. 5A). Additionally, significant increases in fat-soluble vitamin E (tocopherol) metabolites and its degradation products (α-CEHC, α-CEHC glucuronide) were observed in all 3 CF groups compared to non-CF (Fig. 5B). Other fat-soluble vitamin metabolites (A, D, K) were not detected in CF urine.

Figure 5: Increased fat and non-fat soluble vitamin metabolites in CF urine.

Figure 5:

Open in a new tab

A) Changes in non-fat soluble vitamin C pathway metabolites with corresponding box-plots for each metabolite within the displayed pathway. Individual metabolites are compared between CF groups and non- CF. B) Changes in fat-soluble Vitamin E pathway metabolites with corresponding box-plots for each metabolite within the displayed pathway. Individual metabolites are compared between CF groups and non-CF.

Infants with CF have altered urinary pentose sugar metabolites

Lastly, we examined metabolite profiles for changes related to nutrient metabolism in our CF LL and CF LW groups. No significant differences were observed specific to these two groups. However, all three CF groups demonstrated increased urinary pentose sugars such as xylose (Fig. 6A) and fucose (Fig. 6B). Overall, nine out of the 10 urinary pentose sugars detected were significantly altered within CF groups (Fig. 6C). Pentose sugars are simple sugars that do not require pancreatic enzymatic digestion prior to absorption.

Figure 6: CF has altered urinary pentose sugar metabolites.

Figure 6:

Open in a new tab

Differences in the urinary pentose sugar metabolites A) xylose and B) fucose between the CF and non-CF groups. Metabolites differences displayed as box plots with log scaled intensity. B) Heatmap showing differences in urinary pentose sugar metabolites. Numerical values represent fold change in metabolites between group comparisons. A color legend indicates significance of p values for fold changes greater than 1.00 (red).

DISCUSSION

The pathophysiology of CF begins early in life with inflammation, recurrent and persistent infections, and structural organ changes that often precede outward symptoms. Thus, prompt, preventative, therapeutic interventions are preferable to remedial therapies. Nevertheless, CF is a heterogeneous disease, and reliable tools to predict both the onset and progression of early disease do not exist, despite well-characterized genotyping. These deficits are crucial since early treatment is essential to prevent permanent damage. Metabolomics is an example of a modality that has helped determine markers of disease progression in older CF children and adults. Information provided by metabolomics analysis has been used to monitor health or treatment responses in multiple facets of CF disease.[12, 18-32] To this end, we report for the first time urinary metabolite changes within a large cohort of infants with CF that differentiate them from infants without CF. We also highlight metabolic pathways altered early in the course of CF. Understanding how metabolic pathways change during the first year of life in CF and identification of potential novel therapeutic pathways can shed light on the early pathophysiology of CF.

Pathway analysis of altered metabolites revealed several shared pathways among the CF phenotypes studied compared to infants without CF, as well as pathways unique to CF infants with poor growth. In particular, the CF LW group had alterations in the canonical pathways of bile acid biosynthesis, glycine degradation, and the tricarboxylic acid (TCA) cycle. Alterations in bile acid metabolism have been reported in several other CF metabolomics studies of older children and adults, including skin metabolomics.[12, 15, 33] Bile acid signaling after lung aspiration has been also linked with the establishment of chronic lung infections in CF.[16] The presence of urinary bile acid alterations in our study would suggest an early GI source of altered bile acid metabolism in CF such as altered enterohepatic circulation, but it does not rule out concomitant airway changes since airway metabolomics were not performed. Potential sources of bile acid dysregulation in CF have been previously described,[12], but the association of bile acid changes with the CF LW group would suggest the importance of further study of bile acid metabolism in young children with CF as a therapeutic target.

The CF LW group also was predicted to have increased activation of Huntingtin and its downstream regulatory network. A role for Huntingtin has not been previously described in CF, but Huntingtin aggregation may impair the ubiquitin-proteasome system in a similar fashion to CFTR.[34] Protein catabolism via the ubiquitin-proteasome system is central to numerous cellular processes, so it is difficult to determine specifically why it may be increased in the CF LW group. However, proteasomal impairment by Huntingtin could further hinder altered CFTR biosynthesis. [35] In addition to activation of Huntingtin in the CF LW group, ADCYAP1 was found to be the top regulator effect network for both the CF Resp and CF LL groups. ADCYAP1 has been poorly described in CF and not previously mentioned in CF metabolomics studies, but it has been proposed to have a role in CFTR-dependent chloride secretion.[36] ADCYAP1’s downstream effects include the ability to modulate hormonal signaling involved in growth and metabolism. Increased activation of ADCYAP1 in infants with CF will need verification in independent cohorts and further study to determine if it is a pathophysiologic modifier of early CF disease, or a consequence of ongoing disease. Predicted activation in both the CF Resp and CF LL groups suggests that ADCYAP1 is more globally involved in CF, and not just involved in settings of poor growth.

Unique to this study, we found increased urinary fat and non-fat soluble vitamin metabolites as well as increased pentose sugar metabolites in CF infants compared to non-CF. While we have previously reported alterations in the vitamin E isoform gamma-tocopherol in serum from older CF children[15] in relation to altered energy metabolism, we found multiple vitamin C and E metabolites altered in urine from CF infants. CF’s association with fat malabsorption necessitates daily intake of fat-soluble multivitamins (which also contain non-fat soluble vitamins) for most patients with CF, but alterations in non-fat soluble vitamins are not commonly reported. However, because prior CF metabolomics studies have not utilized urine, it is unclear if the overall changes in CF vitamin metabolites are due to differences in bioavailability, increased supplementation, or wasting. Of note, vitamin C supplementation above the recommended daily allowance is not recommended and oxidative stress is high in people with CF, making this an intriguing observation. In addition, increases in urinary pentose sugars were noted. Altered urine fucose concentrations in children with CF were described in the 1960s,[37] but have not been described in infants. Increased urinary pentose metabolites could correlate with inefficient absorption, increased intake, or changes in kidney function/retention of metabolites. Further studies on both pancreatic and non-pancreatic-mediated absorptive pathways are needed in young children with CF.

Despite the discovery of dysregulated metabolites and pathways unique to infants with CF compared to those without, both global and targeted metabolomics profiles were unable to clearly differentiate within our chosen CF phenotypes. This finding raises several important questions. First, is metabolomics able to differentiate phenotypes of early CF disease? The established CF metabolomics literature would suggest metabolomics is quite useful in older populations as listed earlier. It is possible that we did not see differences within CF as all 3 groups had evidence of systemic disease either as poor growth or pulmonary complications, and larger sub-groups may be needed to detect meaningful differences. Metabolomics may be more useful for predicting asymptomatic vs symptomatic infants rather than within symptomatic groups, but is rare to find infants with more severe genotypes who do not have pulmonary or nutritional manifestations. The second question raised is what is the optimal source for metabolomics responses in early CF disease? Urine was chosen due to an interest in nutritional markers and ease of repeated sampling in a young population where more invasive sampling is difficult. It is possible that urine is not the right source for differentiating infants with pulmonary vs nutritional complications as certain lipids and other metabolites are not excreted in urine. Blood, bronchoalveolar lavage, sweat, or exhaled breath may have improved diagnostic capabilities compared to urine. However, single urine metabolites may be better biomarkers than grouped profiles in early CF disease, and our results would not preclude the use of urine metabolomics in early CF studies. Lastly, what is the optimal timing for metabolomics sampling in early CF disease? If used to predict changes to trigger intervention, metabolomics may be more useful when measured over time starting from diagnosis or near diagnosis, or in response to therapeutic initiation. Further studies are needed to help clarify all of these questions.

This study was limited by the observational design, which predisposes to indication bias. As stated earlier, we were limited to a single collection time point and lacked a CF validation cohort due to the design of the BONUS trial. Longitudinal metabolomics would help to verify and monitor identified metabolite changes over time. Verification studies are also needed to support our identified pathway analysis findings, as these can only be considered hypothesis-generating findings until they are independently verified or refuted. Although pathway analysis uses machine learning to determine plausible biologic associations, confirmation of predicted pathways is required. However, our findings did corroborate bile acid changes observed in prior CF metabolomics studies.[12, 15] Further, certain fatty acids (e.g. arachidonic acid metabolites) that are associated with CF disease were not identified by this study and are better assessed with targeted approaches using both blood and urine. Additionally, we were strengthened by a multi-CF center approach, which limits geographic variations in diet and care that may influence metabolite findings in smaller or single-center studies. However, non-CF controls were limited to a single-site, which can influence comparisons to CF patients from multi-sites and is subject to the aforementioned diet variations. Dietary choices can alter the presence or absence of certain metabolites, but participants were not able to be fasted during the study due to the young age of enrollees. Fasting and diet may also be important factors when exploring the interaction of the human and bacterial metabolomes.

In summary, we identified alterations in predicted novel metabolic pathways in infants with CF compared to infants without CF, but cross-sectional metabolic profiles were unable to discriminate between CF phenotypes at one year of age. Targeted metabolomics provides an opportunity for further understanding of early CF disease.

Supplementary Material

Highlights:

  • Urine metabolomics discriminates CF infants from non-CF

  • ADCYAP1 and huntingtin are predicted regulators of altered metabolic pathways in CF

  • CF infants have alterations in bile acid and vitamin metabolism

ACKNOWLEDGEMENTS

Funding for this study was provided through CFFT BONUS11KO, NIH R01DK095738, NIH P30DK089507, and individual TDN Sites. We would like to thank all of the patients and their families for participation. Special thanks to Russell Vaughn from Amherst Pediatrics in Amherst, NY for collection of urine from infants without CF.

ABBREVIATIONS:

(ADCYAP1)

Adenylate cyclase activating polypeptide 1

(BONUS)

Baby Observational and Nutrition Study

(CF)

cystic fibrosis

(CFF)

Cystic Fibrosis Foundation

(CF LL)

CF infants with low length but normal weight

(CF LW)

CF infants with low weight but normal length

(CF Resp)

CF infants hospitalized for respiratory conditions with normal weight and length

(CFTR)

cystic fibrosis transmembrane conductance regulator

(FDR)

False discovery rate

(sPLS-DA)

Sparse Partial Least Squares discriminant analysis

(TDN)

Therapeutic Development Network

(UPLC-MS/MS)

Ultrahigh Performance Liquid Chromatography-Tandem Mass Spectroscopy

Footnotes

Conflict of Interest statement: BI works for Metabolon, a precision metabolomics company

Trial Registration: United States ClinicalTrials.Gov registry NCT01424696 (clinicaltrials.gov).

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

REFERENCES

  • [1].Chin M, Aaron SD, Bell SC. The treatment of the pulmonary and extrapulmonary manifestations of cystic fibrosis. Presse Med 2017;46:e139–e64. [DOI] [PubMed] [Google Scholar]
  • [2].Rosenfeld M, Sontag MK, Ren CL. Cystic Fibrosis Diagnosis and Newborn Screening. Pediatr Clin North Am 2016;63:599–615. [DOI] [PubMed] [Google Scholar]
  • [3].Guimbellot J, Sharma J, Rowe SM. Toward inclusive therapy with CFTR modulators: Progress and challenges. Pediatr Pulmonol 2017;52:S4–S14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [4].Laguna TA, Williams CB, Nunez MG, Welchlin-Bradford C, Moen CE, Reilly CS, et al. Biomarkers of inflammation in infants with cystic fibrosis. Respir Res 2018;19:6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [5].Dickerhof N, Turner R, Khalilova I, Fantino E, Sly PD, Kettle AJ, et al. Oxidized glutathione and uric acid as biomarkers of early cystic fibrosis lung disease. J Cyst Fibros 2017;16:214–21. [DOI] [PubMed] [Google Scholar]
  • [6].Pillarisetti N, Williamson E, Linnane B, Skoric B, Robertson CF, Robinson P, et al. Infection, inflammation, and lung function decline in infants with cystic fibrosis. Am J Respir Crit Care Med 2011;184:75–81. [DOI] [PubMed] [Google Scholar]
  • [7].Leung DH, Heltshe SL, Borowitz D, Gelfond D, Kloster M, Heubi JE, et al. Effects of Diagnosis by Newborn Screening for Cystic Fibrosis on Weight and Length in the First Year of Life. JAMA Pediatr 2017;171:546–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [8].Mann M, Waz W, Poupoulos J, Kadle R, Korn A, Vaughan R, et al. Urinary sodium to creatinine ratio in healthy infants. Clin Pediatr (Phila). 2012;51:852–5. [DOI] [PubMed] [Google Scholar]
  • [9].Coburn-Miller C, Casey S, Luong Q, Cameron N, Hocevar-Trnka J, Leung DH, et al. Standardization of Research-Quality Anthropometric Measurement of Infants and Implementation in a Multicenter Study. Clinical and translational science. 2015;8:330–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [10].WHO Child Growth Standards: length/height-for-age, weight-for-age, weight-for-height and body mass index-for-age: methods and development. WHO Mulicenter Growth Reference Study Group. Geneva: World Health Organization; 2006. [Google Scholar]
  • [11].Grummer-Strawn LM, Reinold C, Krebs NF. Use of World Health Organization and CDC growth charts for children aged 0–59 months in the United States. MMWR Recomm Rep 2010;59:1–15. [PubMed] [Google Scholar]
  • [12].Kopp BT, McCulloch S, Shrestha CL, Zhang S, Sarzynski L, Woodley FW, et al. Metabolomic responses to lumacaftor/ivacaftor in cystic fibrosis. Pediatr Pulmonol 2018. [DOI] [PubMed] [Google Scholar]
  • [13].Breiman L Random Forests. Machine Learning 2001;45:5–32. [Google Scholar]
  • [14].Laguna TA, Reilly CS, Williams CB, Welchlin C, Wendt CH. Metabolomics analysis identifies novel plasma biomarkers of cystic fibrosis pulmonary exacerbation. Pediatr Pulmonol 2015;50:869–77. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [15].Joseloff E, Sha W, Bell SC, Wetmore DR, Lawton KA, Milburn MV, et al. Serum metabolomics indicate altered cellular energy metabolism in children with cystic fibrosis. Pediatr Pulmonol 2014;49:463–72. [DOI] [PubMed] [Google Scholar]
  • [16].Reen FJ, Flynn S, Woods DF, Dunphy N, Chroinin MN, Mullane D, et al. Bile signalling promotes chronic respiratory infections and antibiotic tolerance. Sci Rep 2016;6:29768. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [17].Debray D, Mas E, Munck A, Gerardin M, Clouzeau H. [Liver disease, gastrointestinal complications, nutritional management and feeding disorders in pediatric cystic fibrosis]. Arch Pediatr 2016;23:12S5–S20. [DOI] [PubMed] [Google Scholar]
  • [18].Polineni D, Esther CR, Pace RG, Stonebraker JR, Jones LC, Guo X, et al. In Vivo Human Nasal Lavage Metabolomics Suggest Biomarkers of Lung Disease Severity in Cystic Fibrosis. Pediatric Pulmonology. 2015;50:277-. [Google Scholar]
  • [19].Esther CR. Metabolomics of Airway Inflammation in Cf. Pediatric Pulmonology. 2014;49:126–7.23765632 [Google Scholar]
  • [20].Joseloff E, Sha W, Bell SC, Wetmore DR, Lawton KA, Milburn MV, et al. Serum Metabolomics Indicate Altered Cellular Energy Metabolism in Children With Cystic Fibrosis. Pediatric Pulmonology. 2014;49:463–72. [DOI] [PubMed] [Google Scholar]
  • [21].Muhlebach MS, Sha W. Lessons learned from metabolomics in cystic fibrosis. Molecular and cellular pediatrics. 2015;2:9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [22].Esther CR Jr., Turkovic L, Rosenow T, Muhlebach MS, Boucher RC, Ranganathan S, et al. Metabolomic biomarkers predictive of early structural lung disease in cystic fibrosis. Eur Respir J. 2016;48:1612–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [23].Esther CR Jr., Coakley RD, Henderson AG, Zhou YH, Wright FA, Boucher RC. Metabolomic Evaluation of Neutrophilic Airway Inflammation in Cystic Fibrosis. Chest. 2015;148:507–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [24].Wolak JE, Esther CR Jr., O'Connell TM. Metabolomic analysis of bronchoalveolar lavage fluid from cystic fibrosis patients. Biomarkers. 2009;14:55–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [25].Montuschi P, Paris D, Melck D, Lucidi V, Ciabattoni G, Raia V, et al. NMR spectroscopy metabolomic profiling of exhaled breath condensate in patients with stable and unstable cystic fibrosis. Thorax. 2012;67:222–8. [DOI] [PubMed] [Google Scholar]
  • [26].Montuschi P, Paris D, Montella S, Melck D, Mirra V, Santini G, et al. Nuclear magnetic resonance-based metabolomics discriminates primary ciliary dyskinesia from cystic fibrosis. Am J Respir Crit Care Med. 2014;190:229–33. [DOI] [PubMed] [Google Scholar]
  • [27].Wetmore DR, Joseloff E, Pilewski J, Lee DP, Lawton KA, Mitchell MW, et al. Metabolomic profiling reveals biochemical pathways and biomarkers associated with pathogenesis in cystic fibrosis cells. J Biol Chem. 2010;285:30516–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [28].Robroeks CM, van Berkel JJ, Dallinga JW, Jobsis Q, Zimmermann LJ, Hendriks HJ, et al. Metabolomics of volatile organic compounds in cystic fibrosis patients and controls. Pediatr Res. 2010;68:75–80. [DOI] [PubMed] [Google Scholar]
  • [29].Garg N, Wang M, Hyde E, da Silva RR, Melnik AV, Protsyuk I, et al. Three-Dimensional Microbiome and Metabolome Cartography of a Diseased Human Lung. Cell Host Microbe. 2017;22:705–16 e4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [30].Quinn RA, Lim YW, Mak TD, Whiteson K, Furlan M, Conrad D, et al. Metabolomics of pulmonary exacerbations reveals the personalized nature of cystic fibrosis disease. PeerJ 2016;4:e2174. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [31].Quinn RA, Phelan VV, Whiteson KL, Garg N, Bailey BA, Lim YW, et al. Microbial, host and xenobiotic diversity in the cystic fibrosis sputum metabolome. The ISME journal 2016;10:1483–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [32].Alvarez JA, Chong EY, Walker DI, Chandler JD, Michalski ES, Grossmann RE, et al. Plasma metabolomics in adults with cystic fibrosis during a pulmonary exacerbation: A pilot randomized study of high-dose vitamin D3 administration. Metabolism. 2017;70:31–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [33].Esteves CZ, de Aguiar Dias L, de Oliveira Lima E, de Oliveira DN, Rodrigues Melo CFO, Delafiori J, et al. Skin Biomarkers for Cystic Fibrosis: A Potential Non-Invasive Approach for Patient Screening. Front Pediatr 2017;5:290. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [34].Bence NF, Sampat RM, Kopito RR. Impairment of the ubiquitin-proteasome system by protein aggregation. Science. 2001;292:1552–5. [DOI] [PubMed] [Google Scholar]
  • [35].Ramachandran S, Osterhaus SR, Parekh KR, Jacobi AM, Behlke MA, McCray PB Jr., SYVN1, NEDD8, and FBXO2 Proteins Regulate DeltaF508 Cystic Fibrosis Transmembrane Conductance Regulator (CFTR) Ubiquitin-mediated Proteasomal Degradation. J Biol Chem 2016;291:25489–504. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [36].Derand R, Montoni A, Bulteau-Pignoux L, Janet T, Moreau B, Muller JM, et al. Activation of VPAC1 receptors by VIP and PACAP-27 in human bronchial epithelial cells induces CFTR-dependent chloride secretion. Br J Pharmacol 2004;141:698–708. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [37].Talamo RC, Raunio V, Gabriel O, Pallavicini JC, Halbert S, Disantagnese PA. Immunologic and Biochemical Comparison of Urinary Glycoproteins in Patients with Cystic Fibrosis of the Pancreas and Normal Controls. J Pediatr 1964;65:480–94. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

NIHMS1514345-supplement.docx (1.1MB, docx)

RESOURCES