Advantages of breath testing for the early diagnosis of lung cancer

What could be more attractive than a simple noninvasive breath test for the presence of tumors in the lungs at their early stages of growth? Reaching this challenging objective has now become the focus of a growing body of scientists involved in breath analysis.

Most of the quantitative work in breath analysis research has been carried out to identify the trace gas compounds present in the exhaled breath of healthy volunteers. This has purpose in itself, because the ‘normal’ levels of these compounds and their concentrations need to be established if abnormal levels, which may be indicators of adverse clinical conditions, are to be recognized. For example, it has been shown that abnormally high level of breath ammonia can be an indicator of uremia, which is characteristic of patients suffering from kidney and/or liver disease, and an abnormally high breath acetone may be indicative of the onset of diabetes, although this can also be due to fasting. Breath analysis is a relatively new area of research and much more needs to be done. Nevertheless, this research is establishing breath analysis as a worthy technique for physiological studies and as a valuable addition to the armamentarium of the clinician for disease diagnosis and for monitoring therapy [1–7].

Lung cancer, and indeed other forms of cancer, needs to be recognized at a very early stage of growth of the tumors if therapeutic intervention is to cure the condition or improve the prognosis. The first approach of the clinician is to use imaging techniques, usually x-rays, to establish the presence of tumors, but when they can be ‘seen’ on an x-ray plate, they often have developed to a dangerous size. Therefore, if it can be established that early-stage tumors release volatile biomarkers into the breath at measurable levels then, in principle, the analysis of the breath for these biomarkers would be of enormous clinical value in the battle to diagnose and then cure this dreadful disease. What is so attractive about this approach is that exhaled-breath analysis is totally noninvasive and painless to the individual with no undesirable side effects and, if the breath analysis can be accomplished in real time by simply exhaling into an instrument rather than by sample collection and subsequent analysis, the data will be immediately available to the clinician.

Therefore, a growing effort is being given by physical scientists and clinicians, in harness, to the search for volatile biomarkers of tumors in exhaled breath. Current progress in the development of a diagnostic test for lung cancer through the analysis of breath volatiles has been reviewed recently by Mazzone [8,9]. Central to this research is the availability and development of suitably sensitive analytical instrumentation. Over the last two decades, most use has been made of gas chromatography mass spectrometry (GC-MS) notably by Phillips and colleagues in the USA [10,11]. They have advanced this topic to the point that they have observed patterns of trace compounds in exhaled breath, specifically long-chain and branched-chain alkanes, and found that these patterns are different in the breath of patients suffering from lung cancer than for healthy controls. However, these seminal observations need to be substantiated by other research groups. A major challenge relates to the very low levels of individual alkanes, which are in the region of parts-per-trillion by volume of the exhaled breath. At these very low levels, a large number of compounds are present in ambient air and so an important aspect of this work is to identify which of the compounds in the exhaled breath are truly endogenous and which have been inhaled (exogenous). The biochemical routes to the production of alkanes by tumor cells are not entirely clear; why (nonpolar) alkanes and why not more biologically active (polar) organic molecules? These comments are not intended to cast doubt on these important observations but rather, they are a plea for more research work. Most interestingly, the recent study involving a combination of GC-MS and proton-transfer reaction mass spectrometry (PTR-MS) [12] has revealed eight potential biomarkers for lung cancer confirming four alkanes and also four polar compounds (two alcohols, a ketone and an aldehyde).

Our involvement in trace gas analysis of exhaled breath and the headspace of biological fluids (urine, serum and cell cultures) began with the development of selected ion flow tube mass spectrometry (SIFT-MS) [3,5,7]. A major objective was to develop SIFT-MS to a stage that exhaled breath could be analyzed in real time, obviating sample collection into bags or onto some form of adsorption trap, as used in GC-MS analyses. Sample collection can be time consuming and the composition of humid breath can change during storage. SIFT-MS has now been developed to a state by which the analysis of single-breath exhalations can be accomplished to high sensitivity. Data flow is great – the breath of dozens of patients/volunteers can be analyzed for several volatile metabolites in a single day, providing immediate results. Our early involvement in oncology concerned the analysis of the volatile compounds in the headspace of the urine from patients diagnosed with bladder and prostate cancer. These analyses revealed that formaldehyde was present in the urine of the cancer patients, but was not present in the urine of the healthy controls [13]. The role of formaldehyde generated during metabolism in cancer cell proliferation has been described [14]. There have been reports that formaldehyde is elevated in the exhaled breath of patients with breast cancer [15], although this has not been substantiated. A variant of SIFT-MS has been used to determine levels of formaldehyde in the headspace of breast cancer, leukemia and cervical cancer cells by Bierbaum and colleagues [16]. Latterly, there have been reports that formaldehyde has been detected in the breath of patients with lung cancer using PTR-MS [17]; the search for this and other compound in the exhaled breath is continuing using SIFT-MS, PTR-MS and GC-MS. It is our belief that it is inherently better if specific compounds present at readily measured levels can be identified as biomarkers of various tumor types rather than using more-complex pattern-recognition methods to characterize groups of trace compounds, but this may turn out to be a forlorn hope. Nevertheless, this is the excitement and the focus of our SIFT-MS work.

In parallel with, and in support of, our breath analysis work, we have initiated a study of the release of volatile compounds from lung cancer cell lines incubated in vitro with the expectation that, should characteristic biomarkers be released by these cells, they can act as a focus for the study of the exhaled breath of lung cancer patients. Our earliest studies revealed that incubated CALU-1 and SKMES lung cancer cell lines release molecules of another aldehyde, in this case acetaldehyde, in amounts proportional to the number of cells in the growth medium [18]. This is an exciting and potentially very important observation that deserves much more research. Very recently, we repeated the CALU-1 studies and extended the study to include NL20 normal lung epithelial cells and 35FL121 telomerase-positive (Tel+) lung fibroblast cells. Thus, SIFT-MS has been used to quantify acetaldehyde and, additionally, carbon dioxide in the headspace of the cell cultures [19]. The results showed that acetaldehyde was generated by the CALU-1 and NL20 cell cultures in proportions to the number of cells in the medium. However, following incubation, the acetaldehyde levels in the headspace of the 35FL121 Tel+ cell cultures were much lower than those present in the headspace of the medium alone. The amount of carbon dioxide generated by the CALU-1 and 35FL121 Tel+ cells indicated that they were respiring normally, but much less was produced by the NL20 cells, presumably indicating that normal cell metabolism was inhibited. These rather perplexing results indicate that acetaldehyde release is not specifically the action of malignant cells and so this compound alone cannot be used with confidence as a tumor diagnostic. Even so, we collected bag samples of breath from patients with non-small-cell lung cancer, which, on analysis using SIFT-MS, was shown to contain acetaldehyde. But, it was quickly realized that the major fraction of the acetaldehyde was being emitted from the bag surfaces – another warning of the dangers of bag sampling. However, a very interesting discovery was made that acetic acid was elevated in the breath samples of the cancer patients compared with breath samples from healthy controls. This prompted further investigation of the CALU-1 in vitro cell cultures, when it was found that they also emitted acetic acid when the medium was made acidic. Therefore, it seems likely that the lung tumors are emitting acetic acid, and this may be an important discreet biomarker [Smith D et al. Manuscript in preparation]. However, much more work is needed to substantiate these preliminary results.

We are optimistic that, with further research work, definite biomarkers of lung (and other) tumors will be detected and quantified in exhaled breath, conditional on having sufficient analytical sensitivity. So, what is the way forward? The research carried out to date indicates that cancer biomarkers in breath will generally be at relatively low levels, perhaps at sub-parts-per-trillion by volume levels, and so greater detection sensitivity of SIFT-MS will be required; such improvements are ongoing. Additionally, parallel GC-MS analyses of volatile compounds extracted from patients’ breath should be carried out in order to confirm the exact chemical identity of potential cancer biomarkers detected and quantified by SIFT-MS. It is then important to recruit a larger cohort of patients with diagnosed lung tumors at various stages to provide breath samples on line to the analytical instrument whenever possible. This will be realized by locating the SIFT-MS instruments in the clinic. Further to this, it is recommended that, instead of the conventional sampling of exhaled breath via the mouth, which is known to introduce serious contamination [20], it should be sampled when exhaled via the nose. This will distinguish those trace gases that originate in the lungs/circulation and lower airways from those generated in the oral cavity. With this noninvasive, patient-friendly approach to breath analysis, which is readily accessible to children and frail people, the likelihood of detecting biomarkers of early-stage lung tumors and, indeed, biomarkers of other pathophysiological conditions, will be greatly increased. Then, it will be important to make a greater effort to understand the processes involved in the generation of volatile compounds by both normal cells and cancer cells cultured in vitro, and by different tumor types and grades in vivo, and how the levels of the biomarkers might depend on their growth rate and the tumor microenvironment [Smith D et al. Manuscript in preparation].

Financial & competing interests disclosure

David Smith is Professor in the Medical School at Keele University, UK and a Director of Trans Spectra Limited, UK. Patrik Španel is Research Scientist at the J. Heyrovsky Institute of Physical Chemistry, Czech Republic, Professor in the Medical School at Keele University, UK and a Director of Trans Spectra Limited, UK. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.