Analysis of Volatile Metabolites Using Vibrational Spectroscopy ^†

Abstract

Metabolites are the signature of biochemical reactions that occur in living cells and carry the chemical information of the cell. Analysis of metabolites in biofluids not only provides the information about the internal chemistry of the body but also the body’s state. This can be an efficient route for the development of minimal-invasive or non-invasive diagnosis even in an early disease state if a reliable metabolite detection technique is developed. In particular, volatile metabolites-based breath diagnosis is most attractive since exhaled breath samples can be collected in a fully non-invasive way. Infrared spectroscopy provides a non-destructive, label-free, molecular identification technique, which makes it unbeatable over other existing techniques. It is already applied to the diagnosis of different life-threatening diseases like cancer, diabetes, bacterial infection, and many more. Technical detail of the infrared diagnosis of gaseous biofluids will be presented.

1. Introduction

Early diagnosis of a disease is crucial for survival from many life-threatening diseases. In this regard, cancer, diabetes, heart disease, stroke, chronic obstructive pulmonary disease (COPD), Alzheimer’s, sepsis, etc., are the most vulnerable [1,2,3]. In general, these diseases develop silently at their early disease state, and symptoms only appear when the disease reaches its advanced stage [4]. Even at the advanced stage of the disease, most of the available diagnostic techniques require invasive sample collection or radiative detection of affected tissues, which in practice are avoided to prevent further consequences of physical and psychological stress [5]. Non-invasive and non-radiative diagnostics may be able to overcome this problem as there is no physical impact on sample collection. In this regard, metabolite-based diagnoses from human biofluids are the most promising alternatives [6].

2. Biofluids

In terms of the non-invasive collection of biofluids, exhaled breath and urine are the best choices, as both need to be released at regular intervals of time from the body. Breath is a gaseous biofluid and urine is a liquid biofluid. The article deals with the metabolites in the gas phase only. In practice, urine headspace also provides a significant number of volatile metabolites (VOC), therefore, urine is also considered as a potential biofluid for volatile metabolites analysis. Both the biofluids have many common metabolites; however, the concentration of metabolites is different from breath to urine headspace, which makes one advantageous to the other. In addition, many metabolites are unique to the particular biofluids. Therefore, breath and urine are considered to complement each other and together they strengthen the diagnosis [7]. Both biofluids have several unbeatable advantages, e.g., both can be collected fully non-invasively, are patient-friendly and may be processed rapidly, allowing the collection and measurement of many samples per day.

3. Volatile Metabolite Detection Techniques

Several experimental methods are developing towards technical development as well as diagnosis method development. Among them, different mass-spectrometry (MS) techniques made a significant contribution to the analysis of the volatile metabolites from biofluids [8]. However, MS techniques suffer from poor accuracy and do not show reproducibility in VOC analysis due to the complex and not completely controlled process of the sample preparation. Another very promising technology is the electronic nose, which consists of several chemical sensors to mimic the human nose. Compare to MS devices, it is extremely small in size and cost [9]. However, e-nose neither produces consistent results in different experimental groups, nor it is able to reveal metabolites, as a result, a question mark appears in its applicability. Comparing these techniques, vibrational spectroscopy already demonstrated convincing results on a metabolites-based diagnosis [2]. The diagnostic strategy is explained in the following sections.

4. Discussion on Experimental Method and Data Analysis

4.1. Challenge in Biofluid Analysis

The main challenge of vibrational spectroscopy for gaseous biofluid analysis is the water in the sample. Water strongly absorbed infrared radiation, especially in the spectral regions where most of the biological molecules provide their characteristic spectral signature [10,11,12]. Due to the strong absorption of infrared intensity by water, metabolites receive significantly less infrared radiation, and as a result, the strength of the characteristic molecular signatures becomes significantly low [13]. In addition, water absorption spectra make a shielding effect on the top of biological molecular signatures and it is almost impossible to reveal them. The recent development of a water suppression technique from gaseous biofluids opens an enormous opportunity to analyze volatile metabolites from biofluids [14]. The sample preparation, spectroscopic measurement, and data analysis scheme are presented as a flowchart in Figure 1.

4.2. Water Suppressed Gaseous Biofluids

Biofluids are collected in two forms, gas phase, e.g., exhaled breath; and liquid phase, e.g., urine, blood, saliva, etc. Gaseous biofluids are directly sent through a water suppression unit (see Ref. [14]) to prepare an almost water-free sample suitable for infrared spectroscopic analysis. For liquid biofluids, a headspace is prepared and collected in a special technical process and finally sent through the water suppression unit. A strong water reduction (factor of more than 2500) is achieved when the temperature of the water suppression unit is set at −60 °C.

4.3. Infrared Spectra of Gaseous Biofluids

Infrared spectra of the biofluids are collected by an FTIR spectrometer in conjunction with a multipass 4 m-long “White cell” of volume 2 L in the spectra range from 500–4000 cm${}^{-1}$. Since the metabolites are present in trace amounts, their characteristic absorption peaks are very low intensity and comparable to the noise. To generate a noise-free absorption spectra, a large number of spectra need to be averaged. In practice, an average of 100 spectra yields reasonably good spectra for medical diagnosis.

4.4. Data Analysis

Spectroscopic data are analyzed by two methods, e.g., component analysis and statistical analysis. Component analysis reveals the metabolites present in the biofluids and statistical analysis allows the development of a diagnostic method. In component analysis, significant spectral features are identified by visual inspection [15]. A rigorous spectral search has been performed to find a reasonable spectral match with the identified spectral feature. This is a highly probable guess and several molecules may be guessed. Spectral features are then fitted with experimental or calculated molecular spectra of guessed molecules [16,17]. The best-fitted molecule is considered as the probable metabolite.

Due to the electronic noise and uneven detection sensitivity of the infrared detector, the baseline of the spectra is distorted significantly. This behavior of the spectra prevents the application of statistical analysis to the full spectra. A selective spectral region where most of the spectra show a reasonably steady baseline are considered for a set of statistical analysis. In practice, the statistical analysis around the identified spectral features gives the best classification for different groups in medical diagnosis. Applying such statistical methods significantly high sensitivity, and specificity have been achieved for cerebral palsy [18,19] and prostate cancer cases [20,21,22].

5. Conclusions

The article draws a brief sketch of vibrational spectroscopic analysis of volatile metabolites from human biofluids. Gaseous biofluids like exhaled breath and liquid phase biofluids like urine, blood, saliva, etc., are considered for infrared spectroscopic analysis. In the case of liquid-phase biofluids, headspace is collected as gaseous form. Finally, gaseous biofluids are sent through a water suppression unit to prepare water-free samples for FTIR measurements. Infrared spectra are analyzed by component analysis as well as statistical analysis. Disease-specific biomarkers are revealed by component analysis. Combining component analysis and statistical analysis, a highly accurate non-invasive diagnostics method is under development. Still, a significant development is required before it can be used as a clinical diagnostic method. The research group is working with different disease groups to reveal the disease-specific biomakers as well as the development of statistical methods.