Early non-invasive detection of breast cancer using exhaled breath and urine analysis
Graphical abstract
Introduction
Breast cancer (BC) is the most commonly diagnosed malignancy among females and the leading cause of death around the world. In 2016, breast cancer constituted 29% of all the identified new cases of cancer in the US, and 14% of the deaths caused by cancer [1]. The mortality of cancer in general, and BC in particular, is strongly connected with the sensitivity of tumor detection methods used [2]. Consequently, the development of new early tumor detection methods has been a highly active area of research for several decades. Development of new tumor detection schemes requires improved accuracy that can lead to detection of smaller tumors. However, the new scheme has also to be simple and inexpensive for implementation. Screening mammography is currently the main approach for early detection and has been proven to reduce breast cancer mortality. However, mammography has limitations that are associated with its ability to detect small tumors in dense breast tissue. The overall sensitivity of mammography is 75%–85%, which can decrease to 30%–50% in dense breast tissue [3]. Thus, new methods that can overcome these limitations are needed to identify tumor development at earlier stages of the cancer. One such method is the Dual-energy digital mammography [4,5]. This approach consists of high- and low-energy digital mammograms following administration of an iodine based contrast agent. In this method, the breast is exposed to the low- and high-energy X-ray beams during a single breast compression in mediolateral-oblique (MLO) projection. The breast is then released from compression, and the contrast agent is injected. Following a 3 min delay, the breast is compressed again, and another low- and high-energy exposures is performed to create pre- and post-contrast dual-energy images. These images allow to evaluate the contrast agent kinetics of uptake and washout. Subtraction of the images allows canceling the soft-tissue contrast common to both images and to isolate the iodine signal in the region of angiogenesis. Dual-energy acquisitions are chosen to maximize and minimize, respectively, the ratio of the attenuation of the breast tissue to that of the iodine. Dual-energy enhanced mammography is an inexpensive technique useful in identification of lesions in dense breasts and capable of demonstrating cancers that are not visible at standard mammography. However, the improved resolution of this method is achieved by the exposure of the breast to an increased dose of X-ray irradiation.
An additional method for breast cancer detection is based on magnetic resonance imaging (MRI) imaging [6]. MRI imaging became increasingly important in the detection and delineation of breast cancer in daily practice. The main diagnostic value of MRI relies on specific situations such as detecting cancer in dense breast tissue and recognition of an occult primary breast cancer in patients presenting with cancer metastasis in axillary lymph nodes, among others. Nevertheless, the development of new MRI technologies such as diffusion-weighted imaging, proton spectroscopy and higher field strength 7.0 T imaging offer a new perspective in providing additional information in breast abnormalities. However, a major drawback of the MRI imaging technique is its high cost.
After tumor detection, in most cases a detailed analysis of the tumor tissue is performed following biopsy [7,8]. This procedure is invasive, and requires a high level of expertise and expensive equipment. Moreover, this approach can be used only to confirm BC after the tumor was identified. Another possibility is to use serum for the identification of BC biomarkers [[9], [10], [11], [12], [13], [14], [15]]. These methods are invasive, require very high degree of expertise, and can be implemented only in specialized laboratories.
Recently there have been attempts to detect various cancers including BC using analysis of exhaled breath and urine samples [[16], [17], [18], [19], [20], [21], [22], [23], [24]]. This type of diagnostic methods has important advantages. They are non-invasive, usually easy to implement and in many cases inexpensive. The analysis of body fluids can be performed using different techniques. One possibility is to examine the chemical composition and to identify biomarkers of the illness studied. This can be achieved using either gas or liquid chromatography coupled to mass spectrometry [16,[23], [24], [25]]. Other possibilities are to use electrochemical sensors [26] or different gas sensors (electronic noses, ENs) [16,[20], [21], [22], [27], [28],27,28]. In this approach, the measurement of the exhaled breath sample yields a set of signals, the output of the sensors on the EN used, without details related to the chemical composition of the sample. The association between the outcome of all measurement types and the medical state of the individual examined is achieved by performing statistical analysis of the data collected. A wide variety of statistical methods can be utilized in the data analysis, including multivariate regression [29], principle component analysis [20,30,31], artificial neural networks (ANN) [[32], [33], [34], [35]], fuzzy logic [19,35] and other methods.
The present article describes a scheme for data analysis based on artificial neural networks (ANN) that can be used to develop a reliable predictive model. The method is applied to results obtained in a pilot study in which samples of urine and exhaled breath were analyzed from women with initial stages of BC and from a control group of healthy individuals. In this pilot study, the breath samples were analyzed using two different commercial electronic noses. The urine samples were analyzed using gas chromatography with mass spectroscopy (GC-MS) and detected the volatile compounds in the urine. The main goal of the study is to demonstrate that analysis of the raw data leads to very poor models while application of feature extraction and feature selection to the measured data leads to highly accurate models that allow to detect early stages of BC.
Section snippets
Electronic noses
The exhaled breath analysis was performed using two different commercial ENs. Both ENs contain sensors whose electrical conductivity change when they are exposed to different gas mixtures. The first EN used was the MK4 model (E-Nose Pty Ltd) that contains 12 solid state oxide sensors that have different sensitivities to various gases. The second EN used was the Cyranose 320 (by Sensigent Intelligent Sensing Solutions) that has 32 polymer-based sensors each with a different sensitivity to
Exhaled breath data
The signals obtained from the sensors in the two ENs used have quite different shapes. Typical signals are shown in Figs. SM-1 and SM-2 in the Supplementary materials section. The signal preprocessing started with subtraction of the sensor's initial conductivity from the entire signal. This leads, in most cases, to a zero baseline for all sensors in the ENs. Elimination of the baseline also allows to connect the signals obtained by the different sensors in an EN to obtain a string of signals
Summary and conclusions
This study examined the possible detection of breast cancer at its initial stage by analysis of both exhaled breath and urine. Exhaled breath was analyzed using two different commercial ENs while urine samples were analyzed by GC-MS. For all types of measurements, very poor ANN models were obtained when pre-processed raw data were used. To improve data modeling, feature extraction and feature selection processes were applied to the exhaled breath measured data while only feature selection was
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Ethical approval
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Conflicts of interest
The authors declare that they have no conflict of interest.
Declarations of interest
None.
References (38)
- et al.
A feature extraction method for chemical sensors in electronic noses
Sensor. Actuator. B Chem.
(2003) - et al.
Cancer statistics
CA A Cancer J. Clin.
(2016) - et al.
The benefits and harms of breast cancer screening: an independent review
Lancet
(2012) - et al.
Breast cancer detected with screening US: reasons for non-detection at mammography
Radiology
(2014) - et al.
Dual-energy contrast-enhanced digital mammography: initial clinical results of a multireader, multicase study
Breast Canc. Res.
(2012) - et al.
Application of BI-RADS descriptors in contrast-enhanced dual-energy mammography: comparison with MRI
Breast Care
(2017) - et al.
Magnetic resonance imaging in breast cancer: a literature review and future perspectives
World J. Clin. Oncol.
(2014) - et al.
CCR7 and CXCR4 as novel biomarkers predicting axillary lymph node metastasis in T1 breast cancer
Clin. Canc. Res.
(2005) - et al.
Basal-like breast cancer defined by five biomarkers has superior prognostic value than triple-negative phenotype
Clin. Canc. Res.
(2008) - et al.
A pilot study of circulating mirnas as potential biomarkers of early stage breast cancer
PLoS One
(2010)
Circulating microRNA in body fluid: a new potential biomarker for cancer diagnosis and prognosis
Canc. Sci.
Proteomics and bioinformatics approaches for identification of serum biomarkers to detect breast cancer
Clin. Chem.
Serum microRNA-155 as a potential biomarker to track disease in breast cancer
PLoS One
Serum proteomic biomarker discovery reflective of stage and obesity in breast cancer patients
J. Am. Coll. Surg.
The level of circulating miRNA-10b and miRNA-373 in detecting lymph node metastasis of breast cancer: potential biomarkers
Tumor Biol.
Independent validation of candidate breast cancer serum biomarkers identified by mass spectrometry
Clin. Chem.
Detection of cancer through exhaled breath: a systematic review
Oncotarget
Volatile markers of breast cancer in the breath
Breast J.
Prediction of breast cancer using volatile biomarkers in the breath
Breast Canc. Res. Treat.
Cited by (48)
Cancer biomarkers and their biosensors: A comprehensive review
2023, TrAC - Trends in Analytical ChemistryCirculatory RNA based non-invasive detection and diagnosis of breast cancer
2023, Materials Today: ProceedingsCitation Excerpt :It was observed that at the time of diagnosis, more than 90 % of breast cancers are not metastatic [1]. In general, the sensitivity of tumor detection techniques is closely related to the mortality of carcinoma, and particularly breast cancer [3]. For early detection of breast carcinoma, mammography has been widely used.
Detection of breath acetone by semiconductor metal oxide nanostructures-based gas sensors: A review
2022, Materials Science in Semiconductor ProcessingCitation Excerpt :It has now been determined that there are more than 3500 chemical species in exhaled air, and most of them are VOCs in low concentrations (from parts per million (ppm) to parts per billion (ppb)) [4]. The biomarkers that are present in exhaled air are used as indicators of various illnesses, including lung cancer [4,5], bronchial asthma [6,7], persistent obstructive pulmonary sickness [8,9], breast cancer [10,11], and diabetes [12,13]. According to the latest research on breath evaluations, breath biomarkers indicate abnormal concentrations or the presence of specific species of gases in the breath that are particularly related to certain diseases or metabolic disorders.
A review on chemi-resistive human exhaled breath biosensors for early diagnosis of disease
2022, Materials Today: ProceedingsCitation Excerpt :Nearly 3500 different VOCs are found in exhaled human breath [47]. Exhaled breath biomarkers can indicate a variety of medical problems, including lung cancer [48], asthma [49], chronic obstructive pulmonary disease (COPD) [50], and breast cancer [50] as well as diabetes [51,52]. Multiple stages are involved in the exhaled-breath analysis procedure.
Application of chemoresistive gas sensors and chemometric analysis to differentiate the fingerprints of global volatile organic compounds from diseases. Preliminary results of COPD, lung cancer and breast cancer
2021, Clinica Chimica ActaCitation Excerpt :In this regard, Díaz de León et al. were able to find a separation between the chemical fingerprint of women with breast cancer (n = 262) and clinically healthy women (n = 181) with a correct classification of 98% and a sensitivity and specificity of 100% using the Cyranose 320 [22]. Also, Herman-Saffar found an accuracy of 86.1%, a sensitivity of 87.5% and a specificity of 85% when evaluating 48 women with breast cancer and 45 healthy women using the same equipment [23]. These data are also consistent with our results in obtaining a correct classification of 100% in patients with breast cancer with respect to healthy subjects.