Statistical and bioinformatical methods to differentiate chronic obstructive pulmonary disease (COPD) including lung cancer from healthy control by breath analysis using ion mobility spectrometry

A systematic review of the diagnostic accuracy of volatile organic compounds in airway diseases and their relation to markers of type-2 inflammation

BACKGROUND

Asthma and COPD continue to cause considerable diagnostic and treatment stratification challenges. Volatile organic compounds (VOCs) have been proposed as feasible diagnostic and monitoring biomarkers in airway diseases.

AIMS

To 1) conduct a systematic review evaluating the diagnostic accuracy of VOCs in diagnosing airway diseases; 2) understand the relationship between reported VOCs and biomarkers of type-2 inflammation; 3) assess the standardisation of reporting according to STARD and TRIPOD criteria; 4) review current methods of breath sampling and analysis.

METHODS

A PRISMA-oriented systematic search was conducted (January 1997 to December 2020). Search terms included: "asthma", "volatile organic compound(s)", "VOC" and "COPD". Two independent reviewers examined the extracted titles against review objectives.

RESULTS

44 full-text papers were included; 40/44 studies were cross-sectional and four studies were interventional in design; 17/44 studies used sensor-array technologies (e.g. eNose). Cross-study comparison was not possible across identified studies due to the heterogeneity in design. The commonest airway diseases differentiating VOCs belonged to carbonyl-containing classes (i.e. aldehydes, esters and ketones) and hydrocarbons (i.e. alkanes and alkenes). Although individual markers that are associated with clinical biomarkers of type-2 inflammation were recognised (i.e. ethane and 3,7-dimethylnonane for asthma and α-methylstyrene and decane for COPD), these were not consistently identified across studies. Only 3/44 reported following STARD or TRIPOD criteria for diagnostic accuracy and multivariate reporting, respectively.

CONCLUSIONS

Breath VOCs show promise as diagnostic biomarkers of airway diseases and for type-2 inflammation profiling. However, future studies should focus on transparent reporting of diagnostic accuracy and multivariate models and continue to focus on chemical identification of volatile metabolites.

BALSAM-An Interactive Online Platform for Breath Analysis, Visualization and Classification

The field of breath analysis lacks a fully automated analysis platform that enforces machine learning good practice and enables clinicians and clinical researchers to rapidly and reproducibly discover metabolite patterns in diseases. We present BALSAM-a comprehensive web-platform to simplify and automate this process, offering features for preprocessing, peak detection, feature extraction, visualization and pattern discovery. Our main focus is on data from multi-capillary-column ion-mobility-spectrometry. While not limited to breath data, BALSAM was developed to increase consistency and robustness in the data analysis process of breath samples, aiming to expand the array of low cost molecular diagnostics in clinics. Our platform is freely available as a web-service and in form of a publicly available docker container.

On-Line Analysis of Exhaled Breath Focus Review

On-line analysis of exhaled breath offers insight into a person's metabolism without the need for sample preparation or sample collection. Due to its noninvasive nature and the possibility to sample continuously, the analysis of breath has great clinical potential. The unique features of this technology make it an attractive candidate for applications in medicine, beyond the task of diagnosis. We review the current methodologies for on-line breath analysis, discuss current and future applications, and critically evaluate challenges and pitfalls such as the need for standardization. Special emphasis is given to the use of the technology in diagnosing respiratory diseases, potential niche applications, and the promise of breath analysis for personalized medicine. The analytical methodologies used range from very small and low-cost chemical sensors, which are ideal for continuous monitoring of disease status, to optical spectroscopy and state-of-the-art, high-resolution mass spectrometry. The latter can be utilized for untargeted analysis of exhaled breath, with the capability to identify hitherto unknown molecules. The interpretation of the resulting big data sets is complex and often constrained due to a limited number of participants. Even larger data sets will be needed for assessing reproducibility and for validation of biomarker candidates. In addition, molecular structures and quantification of compounds are generally not easily available from on-line measurements and require complementary measurements, for example, a separation method coupled to mass spectrometry. Furthermore, a lack of standardization still hampers the application of the technique to screen larger cohorts of patients. This review summarizes the present status and continuous improvements of the principal on-line breath analysis methods and evaluates obstacles for their wider application.

Bedside Measurement of Volatile Organic Compounds in the Atmosphere of Neonatal Incubators Using Ion Mobility Spectrometry

Background: Early and non-invasive diagnosis of common diseases is of great importance in the care of preterm infants. We hypothesized that volatile organic compounds (VOC) can be successfully measured in the neonatal incubator atmosphere. Methods: This is a feasibility study to investigate whether the discrimination of occupied and unoccupied neonatal incubators is possible by bedside measurement of volatile organic compounds (VOCs) on the neonatal intensive care unit. VOC profiles were measured in the incubator air using ion mobility spectrometry coupled to multi-capillary columns (BreathDiscovery B&S Analytik GmbH, Dortmund, Germany). Results: Seventeen incubators occupied by preterm infants (50 measurements) and nine unoccupied neonatal incubators were sampled, using 37 room air measurements as controls. Three VOC signals that allow the discrimination between occupied and unoccupied incubators were identified. The best discrimination was reached by peak P20 exhibiting a sensitivity, specificity, positive predictive value and negative predictive value of 94.0, 88.9, 97.3, and 72.3%, respectively. Use of a decision tree improved these values to 100.0, 88.9, 98.0, and 100.0%, respectively. Discussion: A bedside method that allows the characterization of VOC profiles in the neonatal incubator atmosphere using ion mobility spectrometry was established. Occupied and unoccupied incubators could be discriminated by characterizing VOC profiles. This technique has the potential to yield results within minutes. Thus, future studies are recommended to test the hypothesis that VOCs within neonatal incubators are useful biomarkers for non-invasive diagnostics in preterm neonates.

Carotta: Revealing Hidden Confounder Markers in Metabolic Breath Profiles

Computational breath analysis is a growing research area aiming at identifying volatile organic compounds (VOCs) in human breath to assist medical diagnostics of the next generation. While inexpensive and non-invasive bioanalytical technologies for metabolite detection in exhaled air and bacterial/fungal vapor exist and the first studies on the power of supervised machine learning methods for profiling of the resulting data were conducted, we lack methods to extract hidden data features emerging from confounding factors. Here, we present Carotta, a new cluster analysis framework dedicated to uncovering such hidden substructures by sophisticated unsupervised statistical learning methods. We study the power of transitivity clustering and hierarchical clustering to identify groups of VOCs with similar expression behavior over most patient breath samples and/or groups of patients with a similar VOC intensity pattern. This enables the discovery of dependencies between metabolites. On the one hand, this allows us to eliminate the effect of potential confounding factors hindering disease classification, such as smoking. On the other hand, we may also identify VOCs associated with disease subtypes or concomitant diseases. Carotta is an open source software with an intuitive graphical user interface promoting data handling, analysis and visualization. The back-end is designed to be modular, allowing for easy extensions with plugins in the future, such as new clustering methods and statistics. It does not require much prior knowledge or technical skills to operate. We demonstrate its power and applicability by means of one artificial dataset. We also apply Carotta exemplarily to a real-world example dataset on chronic obstructive pulmonary disease (COPD). While the artificial data are utilized as a proof of concept, we will demonstrate how Carotta finds candidate markers in our real dataset associated with confounders rather than the primary disease (COPD) and bronchial carcinoma (BC). Carotta is publicly available at http://carotta.compbio.sdu.dk [1].

Volatile Organic Compounds during Inflammation and Sepsis in Rats

Background:

Multicapillary column ion-mobility spectrometry (MCC-IMS) may identify volatile components in exhaled gas. The authors therefore used MCC-IMS to evaluate exhaled gas in a rat model of sepsis, inflammation, and hemorrhagic shock.

Methods:

Male Sprague–Dawley rats were anesthetized and ventilated via tracheostomy for 10 h or until death. Sepsis was induced by cecal ligation and incision in 10 rats; a sham operation was performed in 10 others. In 10 other rats, endotoxemia was induced by intravenous administration of 10 mg/kg lipopolysaccharide. In a final 10 rats, hemorrhagic shock was induced to a mean arterial pressure of 35 ± 5 mmHg. Exhaled gas was analyzed with MCC-IMS, and volatile compounds were identified using the BS-MCC/IMS-analytes database (Version 1209; B&S Analytik, Dortmund, Germany).

Results:

All sham animals survived the observation period, whereas mean survival time was 7.9 h in the septic animals, 9.1 h in endotoxemic animals, and 2.5 h in hemorrhagic shock. Volatile compounds showed statistically significant differences in septic and endotoxemic rats compared with sham rats for 3-pentanone and acetone. Endotoxic rats differed significantly from sham for 1-propanol, butanal, acetophenone, 1,2-butandiol, and 2-hexanone. Statistically significant differences were observed between septic and endotoxemic rats for butanal, 3-pentanone, and 2-hexanone. 2-Hexanone differed from all other groups in the rats with shock.

Conclusions:

Breath analysis of expired organic compounds differed significantly in septic, inflammation, and sham rats. MCC-IMS of exhaled breath deserves additional study as a noninvasive approach for distinguishing sepsis from inflammation.

Exhaled breath analysis for lung cancer detection using ion mobility spectrometry

BACKGROUND

Conventional methods for lung cancer detection including computed tomography (CT) and bronchoscopy are expensive and invasive. Thus, there is still a need for an optimal lung cancer detection technique.

METHODS

The exhaled breath of 50 patients with lung cancer histologically proven by bronchoscopic biopsy samples (32 adenocarcinomas, 10 squamous cell carcinomas, 8 small cell carcinomas), were analyzed using ion mobility spectrometry (IMS) and compared with 39 healthy volunteers. As a secondary assessment, we compared adenocarcinoma patients with and without epidermal growth factor receptor (EGFR) mutation.

RESULTS

A decision tree algorithm could separate patients with lung cancer including adenocarcinoma, squamous cell carcinoma and small cell carcinoma. One hundred-fifteen separated volatile organic compound (VOC) peaks were analyzed. Peak-2 noted as n-Dodecane using the IMS database was able to separate values with a sensitivity of 70.0% and a specificity of 89.7%. Incorporating a decision tree algorithm starting with n-Dodecane, a sensitivity of 76% and specificity of 100% was achieved. Comparing VOC peaks between adenocarcinoma and healthy subjects, n-Dodecane was able to separate values with a sensitivity of 81.3% and a specificity of 89.7%. Fourteen patients positive for EGFR mutation displayed a significantly higher n-Dodecane than for the 14 patients negative for EGFR (p<0.01), with a sensitivity of 85.7% and a specificity of 78.6%.

CONCLUSION

In this prospective study, VOC peak patterns using a decision tree algorithm were useful in the detection of lung cancer. Moreover, n-Dodecane analysis from adenocarcinoma patients might be useful to discriminate the EGFR mutation.

Current breathomics--a review on data pre-processing techniques and machine learning in metabolomics breath analysis

We define breathomics as the metabolomics study of exhaled air. It is a strongly emerging metabolomics research field that mainly focuses on health-related volatile organic compounds (VOCs). Since the amount of these compounds varies with health status, breathomics holds great promise to deliver non-invasive diagnostic tools. Thus, the main aim of breathomics is to find patterns of VOCs related to abnormal (for instance inflammatory) metabolic processes occurring in the human body. Recently, analytical methods for measuring VOCs in exhaled air with high resolution and high throughput have been extensively developed. Yet, the application of machine learning methods for fingerprinting VOC profiles in the breathomics is still in its infancy. Therefore, in this paper, we describe the current state of the art in data pre-processing and multivariate analysis of breathomics data. We start with the detailed pre-processing pipelines for breathomics data obtained from gas-chromatography mass spectrometry and an ion-mobility spectrometer coupled to multi-capillary columns. The outcome of data pre-processing is a matrix containing the relative abundances of a set of VOCs for a group of patients under different conditions (e.g. disease stage, treatment). Independently of the utilized analytical method, the most important question, 'which VOCs are discriminatory?', remains the same. Answers can be given by several modern machine learning techniques (multivariate statistics) and, therefore, are the focus of this paper. We demonstrate the advantages as well the drawbacks of such techniques. We aim to help the community to understand how to profit from a particular method. In parallel, we hope to make the community aware of the existing data fusion methods, as yet unresearched in breathomics.

A mobile instrumentation platform to distinguish airway disorders

Asthma and chronic obstructive pulmonary disease (COPD) are distinct but clinically overlapping airway disorders which often create diagnostic and therapeutic dilemmas. Current strategies to discriminate these diseases are limited by insensitivity and poor performance due to biologic variability. We tested the hypothesis that a gas chromatograph/differential mobility spectrometer (GC/DMS) sensor could distinguish between clinically well-defined groups with airway disorders based on the volatile organic compounds (VOCs) obtained from exhaled breath. After comparing VOC profiles obtained from 13 asthma, 5 COPD and 13 healthy control subjects, we found that VOC profiles distinguished asthma from healthy controls and also a subgroup of asthmatics taking the drug omalizumab from healthy controls. The VOC profiles could not distinguish between COPD and any of the other groups. Our results show a potential application of the GC/DMS for non-invasive and bedside diagnostics of asthma and asthma therapy monitoring. Future studies will focus on larger sample sizes and patient cohorts.

Computational methods for metabolomic data analysis of ion mobility spectrometry data-reviewing the state of the art

Ion mobility spectrometry combined with multi-capillary columns (MCC/IMS) is a well known technology for detecting volatile organic compounds (VOCs). We may utilize MCC/IMS for scanning human exhaled air, bacterial colonies or cell lines, for example. Thereby we gain information about the human health status or infection threats. We may further study the metabolic response of living cells to external perturbations. The instrument is comparably cheap, robust and easy to use in every day practice. However, the potential of the MCC/IMS methodology depends on the successful application of computational approaches for analyzing the huge amount of emerging data sets. Here, we will review the state of the art and highlight existing challenges. First, we address methods for raw data handling, data storage and visualization. Afterwards we will introduce de-noising, peak picking and other pre-processing approaches. We will discuss statistical methods for analyzing correlations between peaks and diseases or medical treatment. Finally, we study up-to-date machine learning techniques for identifying robust biomarker molecules that allow classifying patients into healthy and diseased groups. We conclude that MCC/IMS coupled with sophisticated computational methods has the potential to successfully address a broad range of biomedical questions. While we can solve most of the data pre-processing steps satisfactorily, some computational challenges with statistical learning and model validation remain.

Integrated statistical learning of metabolic ion mobility spectrometry profiles for pulmonary disease identification

Exhaled air carries information on human health status. Ion mobility spectrometers combined with a multi-capillary column (MCC/IMS) is a well-known technology for detecting volatile organic compounds (VOCs) within human breath. This technique is relatively inexpensive, robust and easy to use in every day practice. However, the potential of this methodology depends on successful application of computational approaches for finding relevant VOCs and classification of patients into disease-specific profile groups based on the detected VOCs. We developed an integrated state-of-the-art system using sophisticated statistical learning techniques for VOC-based feature selection and supervised classification into patient groups. We analyzed breath data from 84 volunteers, each of them either suffering from chronic obstructive pulmonary disease (COPD), or both COPD and bronchial carcinoma (COPD + BC), as well as from 35 healthy volunteers, comprising a control group (CG). We standardized and integrated several statistical learning methods to provide a broad overview of their potential for distinguishing the patient groups. We found that there is strong potential for separating MCC/IMS chromatograms of healthy controls and COPD patients (best accuracy COPD vs CG: 94%). However, further examination of the impact of bronchial carcinoma on COPD/no-COPD classification performance is necessary (best accuracy CG vs COPD vs COPD + BC: 79%). We also extracted 20 high-scoring VOCs that allowed differentiating COPD patients from healthy controls. We conclude that these statistical learning methods have a generally high accuracy when applied to well-structured, medical MCC/IMS data.