Application of metabolomics in acute respiratory distress syndrome: A narrative review

Abstract

Acute respiratory distress syndrome (ARDS) is a clinical condition characterized by damage and inflammation of the alveolar-capillary membrane, which is associated with high incidence and mortality rates. Metabolomics, a significant field of systematic biology, provides a novel perspective for understanding the pathophysiological mechanisms underlying ARDS by analyzing small molecule metabolites in biological samples. This narrative review explores the application of metabolomics in ARDS.

Introduction

Acute respiratory distress syndrome (ARDS) is a serious lung condition often triggered by risk factors such as pneumonia, nonpulmonary infections, trauma, blood transfusions, burns, aspiration or shock, resulting in acute and widespread inflammatory lung damage. ¹ Studies show that patients with ARDS make up about 10% of the total population in intensive care units, with mortality rates between 35% and 45%. ² Additionally, the expensive treatment for ARDS places a significant financial strain on both individuals and society. ³ Even after many years of research, ARDS remains a major challenge for healthcare providers. Metabolomics, a developing research area, has shown its unique significance in studying tumors, cardiovascular diseases, endocrine disorders, and other ailments.^4–6 This field provides fresh insights for disease diagnosis, deepens our understanding of disease mechanisms, and helps identify possible therapeutic targets. ⁷ Recently, the use of metabolomics in ARDS research has seen considerable progress, and this narrative review will explore the latest developments in this field.

Methods

According to the Scale for the Assessment of Narrative Review Articles (SANRA) ⁸ and to ensure methodological rigor, a structured literature search was conducted using PubMed. We used the following search terms (including both free-text and, where applicable, MeSH terms): “acute respiratory distress syndrome,” “acute lung injury,” “ARDS,” “ALI,” and “metabolomics.” To ensure comprehensive coverage, we also screened reference lists from key research articles and reviews for additional sources.

Current status of the diagnosis and treatment of ARDS

More than 50 years have elapsed since Ashbaugh et al. ⁹ first proposed the term “adult respiratory distress syndrome” in 1967. Over these decades, the definition of ARDS has significantly evolved. In 1988, Murray et al. ¹⁰ introduced a scoring system for lung injury. Then, in 1994, the American-European Consensus Conference (AECC) established modern definitions and diagnostic criteria for ARDS to create a unified understanding. ¹¹ The Berlin criteria, released in 2011, provided a more extensive definition of ARDS. ¹² However, clinicians still face challenges in diagnosing ARDS, especially during the COVID-19 pandemic when the limitations of the Berlin criteria became evident. On 24 July 2023, the American Thoracic Society published a new global definition of ARDS, which includes diagnostic criteria for nonintubated patients and those in resource-limited settings to promote early detection and treatment. ¹ Nevertheless, broadening the diagnostic criteria may lead to a higher likelihood of misdiagnosis. To date, the criteria for diagnosing ARDS have mainly relied on clinical presentation and imaging, with no reliable biomarker available for accurate diagnosis.

The treatment options for ARDS are still limited, with respiratory support, mainly mechanical ventilation, forming the foundation of management. ¹³ While many studies have explored pharmacological treatments for ARDS, significant progress remains elusive.^14–16 There is ongoing debate about the effectiveness of glucocorticosteroids, which are frequently used in clinical settings. The 2024 American Thoracic Society guidelines support the use of glucocorticosteroids in patients with ARDS, ¹⁷ yet they do not specify the particular drug, dosage, or duration of treatment. Furthermore, extensive research has shown that ARDS presents as a heterogeneous disease, indicating that a one-size-fits-all treatment may not enhance outcomes for every patient. Therefore, classifying patients with ARDS is crucial.

Although research on ARDS has progressed significantly in the past 50 years, the diagnostic criteria's low specificity and lack of precision in treatment still present major challenges for clinicians. As a result, there is a growing interest in innovative strategies to improve our understanding of ARDS, with metabolomics being recognized as a promising field of study.

What is metabolomics?

Metabolomics involves the detailed examination of metabolites in the fluids, cells, and tissues of living organisms and is commonly utilized for discovering biomarkers. ¹⁸ Its fundamental idea is that disease conditions induce significant alterations in metabolites, reflecting corresponding changes in the genome, transcriptome, or proteome during such conditions. ¹⁹ This approach to forecasting illnesses through the observation of metabolite variations in biological tissues and fluids has roots in ancient Greece. The swift progression of metabolomics can be attributed mainly to remarkable advancements in mass spectrometry and nuclear magnetic resonance spectroscopy, coupled with a deeper understanding of multivariate statistical methods. ²⁰ By leveraging these sophisticated techniques, researchers can quickly identify and measure thousands of small-molecule metabolites, including organic acids, amino acids, carbohydrates, peptides, vitamins, and steroids in patient samples, associating shifts in metabolites with various disease states, significantly enriching the comprehension of disease mechanisms.

Research process in metabolomics

Study design

A carefully structured experiment is the key element of a metabolomics study. Elements like sample size, the alignment of experimental and control groups, the choice of specimens, their collection and storage techniques, and the assay technology utilized will all greatly affect the results of a metabolomics experiment. ²¹

Selection and handling of specimens

Metabolomics research usually selects samples from biological fluids, cells, or tissue extracts. Commonly used biological fluids, like urine, serum, and plasma, are prevalent in both animal and human studies. Moreover, specific research areas have also incorporated other fluids, such as semen, amniotic fluid, cerebrospinal fluid, digestive juices, and alveolar lavage fluid. ²² In addition to biological fluids, many studies have made use of tissue cells as samples.^23,24 Regardless of the specimen type, metabolomics studies require strict adherence to handling protocols, as the handling process significantly affects the quality of data analysis and the reliability of findings. Some studies have shown that different sample processing methods significantly affect the quantitative analysis of metabolites in samples. ²⁵ In 2021, the International Organization for Standardization (ISO) published specifications for pre-examination processes related to metabolomics in urine, venous blood serum, and plasma. ²⁶ Biobanks involved in metabolomics should utilize ISO standards to guarantee high-quality data. For samples collected through nonstandard operating procedures, metabolites that can be influenced by pre-analytical treatments should be excluded from analysis. ²⁷

Selection of detection techniques

A range of techniques for metabolite detection is utilized in metabolomics research, including gas chromatography–mass spectrometry (GC-MS), liquid chromatography–mass spectrometry (LC-MS), and nuclear magnetic resonance spectroscopy (NMR). GC-MS is known for its high efficiency and reproducibility in detecting gaseous and volatile compounds. In contrast, LC-MS stands out for its superior sensitivity, resolution, and quantification abilities, along with a wider detection range than GC-MS. ²⁸ Notably, NMR can quickly identify and quantify various metabolites in biological fluids. It is particularly adept at detecting minute quantities of small molecules within complex mixtures and does not require any physical or chemical sample treatment prior to analysis. This feature makes NMR valuable for examining metabolite levels within intact tissues.^19,20 The detection methods are divided into targeted and untargeted strategies; the former concentrates on specific metabolites while the latter includes all metabolites present in the sample.

Data processing and analysis

Metabolite assay data can be quite complex. Therefore, preprocessing is often essential before performing statistical analyses. Popular software for mass spectrometry data preprocessing includes Mass Profiler Professional and MassLynx, while for NMR data, options include Chenomx NMR Suite, MestReNova, TopSpin, and MATLAB. Once the data is preprocessed, a statistical analysis of the metabolite dataset is required. Commonly used univariate methods in metabolomics involve the t-test, analysis of variance (ANOVA), Kruskal-Wallis test, and Mann-Whitney U test, among others. For predictive modeling, multivariate methods are employed, notably principal component analysis (PCA) and partial least squares-discriminant analysis (PLS-DA) (Figure 1). ²⁹

Figure 1.

Research process in metabolomics. ANOVA: analysis of variance; GC–MS: gas chromatography–mass spectrometry; LC–MS: liquid chromatography–mass spectrometry; NMR: nuclear magnetic resonance spectroscopy; PCA: principal component analysis; PLS–DA: partial least squares–discriminant analysis.

Application of metabolomics in ARDS

Basic research

Metabolomics has shown considerable promise in various facets of basic ARDS research, including the creation of animal models and investigation into pathological mechanisms. Serkova et al. ³⁰ suggested that NMR-based metabolomics can quantify inflammatory lung injury, assisting in the development of more consistent and predictive lung injury models in mice. Additionally, Bos et al. ³¹ studied exhaled breath from mice with LPS-induced acute lung injury using GC-MS, pinpointing several potential biomarkers like hexanal and pentadecane, thus showcasing metabolomics’ potential for diagnosing acute lung injury. Finally, Gouellec et al. ³² employed high-resolution magic angle spinning nuclear magnetic resonance spectroscopy to examine lung tissues from Pseudomonas aeruginosa-infected mice of varying virulence, enabling not only identification of infection severity but also evaluation of early therapeutic effects.

Diagnosis and classification

Metabolomics is essential for diagnosing and classifying ARDS. Many studies have used metabolomics to pinpoint potential biomarkers for ARDS, such as octane, ³³ purine, ³⁴ aspartates, carbamate, ³⁵ citrates, creatine, ³⁶ and phosphatidylcholine. ³⁷ However, given the complexity and diversity of ARDS, it is unlikely that a single metabolite would serve as a universal biomarker for all cases. Thus, other research has focused on screening multiple metabolites in assays to create diagnostic models. For example, Bos et al. ³⁸ used GC-MS to analyze exhaled breath from 42 patients with ARDS and 52 patients with non-ARDS in the ICU, identifying octane, acetaldehyde, and 3-methyl heptane for their diagnostic model, which yielded an area under the curve (AUC) of 0.80 for the training cohort and 0.78 for the validation cohort. Similarly, in another study involving a larger sample size (n = 499), Zhang et al. ³⁹ screened five metabolites: 1-methylpyrrole, 13,5-trifluorobenzene, methyloxyacetic acid, 2-methylfuran, and 2-methyl-propanol, which resulted in an AUC of 0.71 for the training cohort and only 0.63 for the validation cohort. This variation may be attributed to the larger sample size, which highlights the heterogeneity of ARDS. Therefore, metabolomics might be more valuable in uncovering ARDS subphenotypes.

There are several methods to classify ARDS. Based on the underlying causes, ARDS can be separated into direct and indirect forms. Those with direct ARDS usually present higher lung injury scores. ⁴⁰ Additionally, ARDS can be categorized into high-inflammatory and low-inflammatory subphenotypes based on biomarker levels. The high-inflammatory subphenotype shows severe inflammation, shock, and metabolic acidosis, which are linked to significantly worse clinical outcomes. ⁴¹ Furthermore, ARDS can be classified into focal and nonfocal ARDS based on computed tomography findings, with nonfocal ARDS exhibiting poorer lung compliance and higher mortality rates. ⁴² Different ARDS subphenotypes respond distinctly to therapeutic approaches, including fluid management, drug choices, and mechanical ventilation strategies.^43–45 Currently, there are not many metabolomics studies focusing on these subphenotypes. Chang et al. ³⁷ highlighted sphingolipid metabolism as a critical pathway for distinguishing between lung-derived and extra-pulmonary ARDS through metabolomic techniques. Moreover, Metwaly et al. ⁴⁶ identified two ARDS subphenotypes using GC-MS: a high-inflammatory subphenotype and a low-inflammatory counterpart. Similarly, Alipanah-Lechner et al. conducted a metabolomics analysis on plasma from 93 patients with sepsis-induced ARDS, confirming both high-inflammatory and low-inflammatory subphenotypes. Patients with the high-inflammatory subphenotype had significantly lower plasma lipid concentrations and increased levels of glycolytic metabolites such as lactate and pyruvate compared to those with hyperinflammatory ARDS. ⁴⁷

Some studies have also examined the metabolic characteristics of ARDS caused by various pathogens. For example, Lorente et al. applied high-resolution magic angle rotational NMR spectroscopy to study serum samples from 18 patients with ARDS due to COVID-19 and 20 with ARDS due to H1N1. They found notable differences between the two groups, with COVID-19 patients displaying significant alterations in energy metabolism, while patients with H1N1 showed a more marked inflammatory and oxidative stress response. ⁴⁸ Izquierdo-García et al. ⁴⁹ contrasted the metabolic profiles of H1N1 and Streptococcus pneumoniae-induced ARDS using NMR and identified glutamine, methylguanidine, and phenylalanine as potential biomarkers for H1N1-induced ARDS, whereas lactate and creatine were suggested as markers for S. pneumoniae-induced ARDS. Additionally, Batra et al. ⁵⁰ performed metabolomics analysis on urine samples from 42 patients with COVID-19-induced ARDS and 17 with bacterial sepsis-induced ARDS via ultra-high-performance liquid chromatography-tandem mass spectrometry, uncovering 150 differential metabolites. Similarly, Ferrarini et al. ⁵¹ noted significant differences in metabolic patterns among three groups of patients with ARDS caused by COVID-19, H1N1, and bacterial pneumonia, respectively.

Grading and prognostic prediction

Metabolomics can also be utilized to evaluate the severity of ARDS and forecast patient outcomes. Viswan et al. classified 176 patients with ARDS into mild, moderate, and severe categories and performed a metabolomics analysis of the patients’ serum using NMR. They developed a model that accurately distinguished ARDS severity among the patients. ⁵² In another study, 36 patients with ARDS were segmented into mild and moderate/severe groups. Researchers analyzed the patients’ alveolar lavage fluid via NMR to create a prognosis prediction model based on six distinguishing metabolites: proline, lysine, arginine, taurine, threonine, and glutamate, achieving an AUC value of up to 0.95. ⁵³ In a larger study, Wu et al. combined metabolomics and proteomics to analyze serum samples from 130 patients with ARDS, establishing an early prognosis prediction model with an AUC of 0.893, outperforming clinical risk prediction models based on the Sequential Organ Failure Assessment (SOFA) score and oxygenation index. ⁵⁴ Furthermore, other research has indicated changes in lipid and phenylalanine metabolism associated with mortality rates in patients with ARDS.^55,56

Precision therapy

The diversity of ARDS has created an urgent demand for precise treatment options. Metabolomics could serve as a crucial instrument, as it identifies changes in key metabolites associated with ARDS, clarifies its pathogenesis, deepens the understanding of the metabolic disturbances linked to ARDS, uncovers new drug targets for prevention and treatment, and aids in assessing therapeutic outcomes in ARDS.^57,58 In fact, metabolomics can be applied throughout the drug development process, from target discovery to ADMET testing, clinical trials, efficacy monitoring, and dose optimization. ⁷ Nevertheless, despite these advancements, achieving precision treatment for ARDS poses ongoing challenges. Integrating clinical data with multi-omics strategies to investigate ARDS subphenotypes more comprehensively could be essential for realizing precise treatments.

Conclusion

ARDS is a severe illness marked by significant variability, and its mortality rate remains troublingly high despite substantial research efforts. Recent developments in metabolomics, a novel research approach, have shown promise in enhancing the diagnosis, classification, treatment, and prognosis prediction of ARDS. Nonetheless, most current studies have been limited to single centers and small sample sizes, lacking uniform criteria for sample selection and confounding variable identification, which leads to low reproducibility. Future investigations should focus on optimizing the metabolomics research approach and integrating other histological techniques to create a comprehensive database that aids in classifying ARDS. It's crucial to design specific treatment strategies tailored to the histological characteristics of different ARDS subphenotypes. Our goal is to shift ARDS treatment from traditional methods to a more precise and individualized medical model, which we believe will improve both survival rates and the quality of life for patients with ARDS.