Breath Analysis: A Promising Tool for Disease Diagnosis-The Role of Sensors

Abstract

Early-stage disease diagnosis is of particular importance for effective patient identification as well as their treatment. Lack of patient compliance for the existing diagnostic methods, however, limits prompt diagnosis, rendering the development of non-invasive diagnostic tools mandatory. One of the most promising non-invasive diagnostic methods that has also attracted great research interest during the last years is breath analysis; the method detects gas-analytes such as exhaled volatile organic compounds (VOCs) and inorganic gases that are considered to be important biomarkers for various disease-types. The diagnostic ability of gas-pattern detection using analytical techniques and especially sensors has been widely discussed in the literature; however, the incorporation of novel nanomaterials in sensor-development has also proved to enhance sensor performance, for both selective and cross-reactive applications. The aim of the first part of this review is to provide an up-to-date overview of the main categories of sensors studied for disease diagnosis applications via the detection of exhaled gas-analytes and to highlight the role of nanomaterials. The second and most novel part of this review concentrates on the remarkable applicability of breath analysis in differential diagnosis, phenotyping, and the staging of several disease-types, which are currently amongst the most pressing challenges in the field.

Keywords: breath analysis; differential diagnosis; nanomaterials; sensors; volatile organic compounds.

Figure 1

Diagram summarizing the correlation of VOCs present in the exhaled breath, with oxidative stress and inflammatory conditions; metabolic breakdown of larger molecules leads to the formation of exhaled VOCs. Reprinted with permission from ref. [17]. Copyright © 2012 John Wiley & Sons Ltd.

Figure 2

Schematic representation of the working principle of selective sensors and artificially intelligent cross-reactive sensor arrays. Selective sensors contain highly selective elements in order to detect a specific gas-analyte in the presence of a composite gas-mixture. Cross-reactive arrays feature sensors that are sensitive to the majority of the gases present in the gas-mixture. In any case, detecting analyte concentration above a critical value leads to the differentiation between sick and healthy subjects. The response of gas-sensing arrays can be then processed by employing artificial intelligence, machine-learning, and pattern recognition techniques. Reprinted with permission from Ref. [6] Copyright © 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.

Figure 3

Statistical analysis of the response of a nanomaterial-based, cross-reactive chemiresistor for real-world samples of sick and healthy subjects. The use of PCA permits the differentiation of the groups. Notably, relative humidity compensation reduces the dispersion of different clusters thereby improving the discrimination between healthy and sick subjects. Representative 2D breath-analysis PCA plots for prostate cancer diagnosis: (a) without relative humidity compensation; (b) with relative humidity compensation. PCA plots for breast cancer diagnosis: (c) without relative humidity compensation; (d) with relative humidity compensation. Adapted with permission from Ref. [42] Copyright © 2012, American Chemical Society.

Figure 4

Schematic representation of (A) a chemiresistor under a constant bias and with an overlying metallic or semiconducting sensing layer which acts as the sensitive gas-sensing layer; (B) a field-effect transistor where the conductivity of the channel is sensitive to gas-analytes exposure; (C) an electrochemical sensor (potentiometric, amperometric, or conductometric) composed of a working (sensing) electrode on which the analyte reacts (redox reaction), a counter electrode (with respect to which electrical signal is measured) and a reference electrode of “reference” potential. Reprinted with permission from Ref. [13] Copyright © 2021 Wiley-VCH GmbH.

Figure 5

Polar plot of a 2 × 4 sensor array response for (a) 10 ppm of acetone; (b) 1 ppm of NO₂ and; (c) 1 ppm of H₂S and operation temperatures 300, 150, and 250 °C; (d) schematic representation of the parts of the sensor (2 × 4 sensor array, back heater, chip carrier, and Au wires). The array shows good recovery and repeatability as well as high performance at sub ppb level, facilitating discrimination between the three biomarkers. Adapted with permission from Ref. [99]. Copyright © 2021 by the authors. Licensee MDPI, Basel, Switzerland.

Figure 6

Schematic representation of a MCNPs-based chemiresistor, programmed according to the VOCs identified as inflammatory bowel and irritable bowel syndrome disease biomarkers. The sensor was exposed to simulated, HC and patient breath samples. Time-dependent and reversible shift in the sensor’s resistance is associated with the MCNPs–VOCs interactions. The variability of molecular ligands leads to varying sensing responses; pattern recognition methods such as ANN, permit the development of effective diagnostic classifiers. Reprinted with permission from ref. [111]. Copyright © 2016 WILEY-VCH Verlag GmbH & Co. KgaA, Weinheim.

Figure 7

(a) SEM image of an array of polymer-coated PtNPs-based chemiresistors, developed for pesticide detection, TEM image of PtNPs layer, with surface coverage 46% and mean nanoparticle diameter of 5 nm; (b) schematic representation of the sensor array. Different polymer susceptibility towards water/pesticide vapors leads to a gas-sensing array that is capable of identifying each of the gas-analytes. Reprinted with permission from Ref. [41] Copyright © 2018 Elsevier Ltd.

Figure 8

Schematic representation and SEM images of (a) carboxylated SWCNTs aligned between Au electrodes and (b) Au-decorated SWCNTs, as sensing films for gas sensors (FETs), for enhanced H₂S sensing. Au NPs play a crucial role in device-performance by modulating the mobility after the gaseous-molecule interactions. Reprinted with permission from Ref. [156]. Copyright © 2011 Wiley-VCH Verlag GmbH & Co. KgaA, Weinheim.

Figure 9

Schematic of (a) surface modification of SiNW FET sensors (SSM, single-step modification; TSM, two-step modification); (b) sensor exposure to synthetic samples of potential organic volatile biomarkers for each disease (asthma/COPD, LC, GCa); (c) sensor exposure to real breath samples of patients with selected diseases, in comparison to HC; (d) ANN analysis representation. Increased accuracies were obtained for the discrimination of GCa vs. LC and LC vs. asthma and COPD patients. Reprinted with permission from Ref. [43]. Copyright © 2016, American Chemical Society.

Figure 10

Schematic representation of Au-MoS₂ nano-flakes sensing mechanisms for (a) hydrocarbons using MoS₂, leading to small signal alternation, due to dipole scattering through the electron clouds of –CH₂– groups; (b) oxygen-based VOCs using MoS₂, which donate electrons to the MoS₂ (left) and are oxidized by the adsorbed oxygen species, releasing electrons to MoS₂ (right); (c) oxygen-based VOCs using MoS₂ decorated with Au NPs, which increase electron concentration and, thus, oxygen species adsorption on MoS₂; (d) returning more electrons upon interaction with the oxygen-based VOCs. Reprinted with permission from ref. [69]. Copyright © 2019, American Chemical Society.

Figure 11

Representation of the sensing mechanism of gas detection using (a) a QCM sensor where the oscillation of the quartz crystal substrate and propagation of the transverse acoustic wave through the quartz substrate are caused by the alternating electric field applied over the electrodes; sensing layer-gas interactions change the mass on the substrate and hence wave amplitude and velocity, leading to a resonance frequency shift (Δm → Δf); (b) SAW sensors where a surface wave confined within one acoustic wavelength of the surface of the piezoelectric material is induced by an input RF-voltage applied across the interdigitated transmitter (IDT); mechanical energy is transformed back into radio frequency as an output when the SAW reaches the receiving IDTs. Analyte adsorption on the piezoelectric coating induces mass variations and ultimately a shift in frequency. Reprinted with permission from Ref. [172]. Copyright © 2021 Elsevier B.V. All rights reserved.

Figure 12

Sensitivity values (1/ppm) of (a) sensors S1–S3, (b) sensor S1–S4–S5 in radar plots, and (c) sensors S1–S5 in bar chart representation, for six different analytes under 0% RH provided with responses to 100% humidity. Inserted graph: magnified view of data below 2 × 10⁹ 1/ppm. (S1: SH-Calix[4]arene, S2: AuNRs, S3: Calix[4]arene modified AuNRs, S4: AgNCs, and S5: Calix[4]arene modified AgNCs, ISP: isoprene, ACE: acetone, n-HEX: n-hexane, EtOH: ethanol, CHL: chloroform, TOL: toluene). Calix[4]arene modification (S3, S5) increased the sensitivity, under 0% of humidity, especially for TOL and CHL, while thiol terminated calix[4]arene (S1) exhibited increased response towards CHL. S5 exhibited the highest responsiveness towards TOL among other VOCs and the highest among all sensors (π–π interactions leading to host–guest complexes). Reprinted with permission from Ref. [49]. Copyright © 2021 Elsevier B.V. All rights reserved.

Figure 13

Detailed color difference maps of six VOCs at three volumes (0.75, 1.5, and 3 mL saturated vapor, respectively) saturated analyte vapor at 20 °C, acquired by metalloporphyrin-AuNRs and dyes-based optical chemical sensor. Metalloporphyrin-AuNRs exhibited increased responsiveness and high sensitivity and selectivity due to their high affinity. Reprinted with permission from Ref. [141] Copyright © 2014 Elsevier B.V. All rights reserved.

Figure 14

(a) Presentation of the cross-validation percentages of the differentiation of asthma, COPD, LC patients, and HC, using SpiroNose; (b) PCA plot of breathprints collected from asthmatic patients at the Academic Medical Center (AMC), Amsterdam and Medical Spectrum Twente (MST), Enschede, for which no significant differentiation is observed (p = 0.892). Adapted with permission from Ref. [91]. Copyright © 2015 IOP Publishing Ltd.

Figure 15

DFA plots representing the discrimination of (a) LC patients from HC; (b) SCLC from NSCLC patients; and (c) SCC from ADC patients, using a 6-sensor array of UV-irradiated (394 nm) pristine or metal-doped WO₃NWs. The arrays achieved the detection of lung cancer but also the prediction of LC histological subtypes. Reprinted with permission from Ref. [246]. Copyright © 2020 Published by Elsevier B.V.