We hope you enjoy the ‘Non-invasive Functional Assessment of the Microbiome from Exhaled Breath’ webinar 

To turn on subtitles, click play, then click the “CC” button at the bottom right of the YouTube window.

Please make sure you accept cookies to ensure this webinar appears in your browser, please see our cookie policy for more options

  • 00:35 Dr Elizabeth Crone with an introduction to the Breath Biopsy OMNI Platform and why breath?
  • 14:23 Dr Robert Mohney – Applications for Microbiome Research & Understanding Microbiota Function in Human Health
  • 36:36 Dr Anthony Hobson, Clinical Director at Functional Gut Clinic
  • 44:48 Q&A Disccussion


For more details on Breath Biopsy OMNI read our whitepaper, or visit our gastrointestinal diseases information hub. If you would like to incorporate Breath Biopsy technology into your research, get in touch to speak to one of our experts. 

Question & Answer session

Answered live

You mentioned using exhaled breath to measure metabolites from the gut microbiome and differentiating between small and large intestine. How do you differentiate your read outs from those coming from the oral cavity or the lung microbiome? 

Owlstone’s collection system collects the lower breath fraction, excluding the dead space which includes the oral cavity. Our samples are thus enriched for VOCs coming from the blood and lung. 

What are the pros and cons compared to faecal sample analysis? 

  • Fecal analysis reflects bacterial metabolism from a few hours-days ago, while breath gives an almost instant read out, similar to blood, as gut metabolic products will diffuse quickly into the blood then out into the breath 
  • Fecal samples are often challenging to get from patients – they are off-putting and can harm recruitment rates 
  • Fecal samples can only be obtained as and when the patient produces them – longitudinal time courses to monitor what’s happening over the course of minutes to hours is not possible. Breath samples can be collected at frequent intervals 

Is your in vitro assay setup for non-human, non-primate samples (i.e., murine)? 

Our headspace analysis platform could use any type of species samples (e.g. cells or tissues). For sampling VOCs from an entire organism, this is not something we can currently offer as a standard service, however, it is an interesting area and for projects where this is critically needed we can explore how to do this with you. 

How amenable is your breath collection method to high frequency serial sampling in the same individuals? 

Extremely amenable. There are no limits on thefrequency of breath sampling. This is well demonstrated by hydrogen and methane testing, where a substrate that microbes metabolize is ingested, then samples are taken over a three hoursperiod to determine where in the gut approximately the metabolising microbes are.  

Could volatile compounds also indicate patient response to the microbiome and microbial metabolites? 

Yes – breath VOCs also reflect host response and other metabolic processes in humans – for example, they have been associated with lipid peroxidation and inflammation in the past. This paper demonstrates host VOC production in response to microbial triggers nicely. 

Are the VOCs from bacteria metabolism or from interaction between bacteria and host cells? 

VOCs can come from both processes – from direct bacterial metabolism as well as host response, the question directly above helps to answer this too. 

How long is each breath sample collection session? 

One of the benefits of breath VOCs is that they can be concentrated – thus, taking a longer breath sample allows us to analyse even molecules that are very low concentration in the breath. For biomarker discovery, where we want to be able to analyse as broad a range of molecules as possible (both those at high and low concentrations in the breath), we typically collect a breath sample over 13-15 minutes – this gives us 5 L of breath in total, spread across two samples; we analyse 2.5L of breath VOCs (from the lower respiratory tract) on the GC-MS.  

How to separate the VOC produced by gut infection and lung diseases? 

A breath sample will reflect VOCs from the blood (and thus from any tissue in proximity to blood) as well as the lungs. Use of parallel headspace fecal analysis could help to clarify if VOCs that are seen in the breath are also seen in the gut.  
How many typical breaths are required to collect detectable volumes of SCFAs? How are data compared between participants in a study – is the analysis based on equivalent volumes of breath? What is that typical volume and number of breaths?

In our experiments so far we have used 2.5L of breath (specifically the lower fraction of breath). However, there were indications in the data that lower volumes of breath would be sufficient. Owlstone has put a lot of effort into standardizing breath collection to make sure samples are as comparable as possible, with minimum technical variability. We collect a fixed volume of the lower breath fraction to enable this. It takes around 13-15 minutes on average to collect 5 L of breath over 4 sorbent tubes. Breath collection time will vary patient by person as everyone breaths differently. More details are here.

Do you have any studies/examples comparing SCFA measured by breath vs. SCFA measured by a stool sample? Our typical method of assessing SCFA is by stool sample. Would like to see how this compares with a breath analysis. 

Is what is being exhaled through breath (reflecting faecal microbiome) similar to the volatile compounds found in headspace analysis of faeces? 

(These were answered together) 

Owlstone has recently been doing work looking at fecal headspace sampling to look at SCFAs, and we see many of the same SCFAs in both fecal headspace and breath. Looking at the samples of both matrices from the same patient has not yet been done, however, due to the temporal differences the samples would not be expected to be identical – breath is more similar to blood in that it reflects metabolism at the time of collection, while fecal samples reflect metabolism a few hours-days after.  

Does breath reflect the gut or oral microbiome better? 

Owlstone’s collection parameters focus on lower breath, which is enriched for compounds from the bloodstream. Thus, our breath samples should better reflect VOCs from the rest of the body, including the gut, versus the oral cavity.  

Are the VOCs completely coming from metabolism of metabolites that pass through the blood, liver, and into the lungs or are they directly regurgitated from the GI tract? If the VOCs are generated in the first mechanism, then changes in microbiome in other areas of the body, e.g. joints, skin can generate these VOCs either by themselves or in combination with microbiome in the gut. 

What we see in the breath will be coming more from the blood and lungs, versus direct regurgitation from the digestive tract. You are correct that changes not just to the microbiome will be observed in breath – we will see VOCs from anywhere around the body in contact with the blood.  

Are there any contaminations brought by the plastic parts of the [ReCIVA] Breath sampler you developed? 

Owlstone has taken care during development to avoid materials that leach or adsorb a lot of VOCs. Any contamination that is still present (it’s impossible to absolutely avoid this) can be examined by comparison to system blank samples taken. These are samples taken with just the equipment.  

Does breath VOCs collected on sorbent tubes deteriorate over time? 

The sorbent tubes used stabilize VOCs, which helps with this. Owlstone also process sorbent tubes within two weeks of collection by dry-purging them – this removes residual water vapour that could speed up degradation of the molecules.  We believe that some molecules will never be seen in TD-GC-MS – they are so reactive that they will degrade immediately after exhalation, however we see many molecules persisting in breath samples stored over long time periods. As long as samples are taken regularly in clinical studies, data processing can also help to correct for degradation of specific molecules.  

Is Hydrogen Sulphide detected in all people or are there individuals whose microbiome do not generate Hydrogen Sulphide?

Hydrogen Sulphide was detected in all patients [sampled by the Functional Gut Clinic], however, the levels of it varied significantly.  

How do you assess the correlation analysis between the VOCs in exhaled breath and microbiota (e.g., faecal) statistically? How to infer the results and estimate this correlation effectively? 

The relationship between VOCs in exhaled breath and those observed in fecal samples (e.g., through headspace analysis) can be examined through a variety of methods that assess correlation or covariance. (For example, network analysis is one method that is commonly used, which not only considers the relationship of the individual components but also their interconnections.) 

In a similar manner, one could also examine the association of breath VOCs (or fecal headspace VOCs) and microbial species (using data generated from 16S rRNA, shotgun metagenomics, or whole-genome microbial sequencing). These types of analytical approaches are referred to as metagenome-wide association studies. 
There are numerous software and web-based tools available that scientists can utilize to assess the relationships between volatile metabolites or between volatile metabolites and (meta)genomic data. 

Additional questions we did not have time to cover

Has the VOC technology been correlated to colonoscopy biopsy samples? 

Yes, there is a paper  here that demonstrates this. In this case, the researchers correlated VOCs from headspace analysis of a small number of gut biopsy samples with breath samples from the patients. They found a number of VOCs that were produced from both samples, with some differences also observed. Owlstone’s headspace analysis service can be used with colon biopsy samples, alongside use of our breath collection and analysis service, so if this is a subject of interest we could help.  

What evidence is there that these kinds of compounds could be specific for specific bacterial species, or even strains? 

Some of the references in slide 14 give evidence on this.  In many cases, genomic sequencing indicates the presence of genes responsible for the metabolism of specific molecules in some species but not others – see here for a good review. Multi-omic approaches that combined VOC and microbial sequencing analysis (e.g. 16S rRNA, shotgun metagenomics, or whole-genome microbial sequencing) will be informative in this regard, though these results are most enlightening when this approach is applied to very large (n=1000s) sample numbers. From our work looking at a healthy population, and the finding that there are many VOCs correlated with known microbial metabolites, we believe there are additional VOCs undiscovered as of yet that may help with this.  

How well do plasma levels of SCFAs correlate with breath SCFAs? 

We expect the levels of SCFAs in blood will correlate with levels on breath.  However, we have not identified peer-reviewed published data to confirm this hypothesis. Owlstone is discussing a pilot study to directly examine this question (and other related questions) with a well-respected academic and clinical research group. We believe this will make for an informative and interesting study! 

With regard to ethanol fermentation by the gut, couldn’t high fermenters modify their diet to correct the issue? 

Diet is one approach to reducing the symptoms, but it doesn’t solve the problem. Similarly, treatment with probiotics containing ethanol-consuming bacteria could be another route to remedy the issue.  (Indeed, there is a clinical trial underway that is assessing the utility of ethanol-consuming probiotic bacteria for reducing fatty liver.) However, it is likely that antibiotic treatment would be warranted to reduce/eliminate the pathogenic ethanol-producing bacteria in the gut and restore proper balance. 

Can you discuss whether there are breath collection devices suitable for infants and young children? 

Owlstone’s current breath sampling technology (the ReCIVA Breath Sampler) is optimised for use on adults; however, it has been evaluated for use in school-age children – see here . There are sampling components (e.g. a mask) in a small size that could be used for younger children. Infants are currently not possible. If this is something of high interest to you, we can discuss to explore how we could help you test if our systems could be used on your target population. 

As much as this platform goes far beyond traditional hydrogen breath testing… does this platform measure hydrogen, too? 

Owlstone has a home-based test that can be used for hydrogen and methane testing [see slides 27-28 of the microbiome webinar]. Our OMNI platform is used for larger VOCs. 

Can breath sampling be done at home? 

Owlstone’s OMNI platform for broad breath VOC analysis is optimised to reduce sources of technical variability and is best suited for use in clinical sites. However, once we are confident in the use of specific VOCs for an application, we can work with clinicians to develop home-based breath sampling solutions. We are already developing options for some of our internal work, and of course offer hydrogen and methane breath tests, which are sent to a patient’s home after a healthcare practitioner referral. Upon completion of the test in the comfort of the patient’s home, the test is mailed back to Owlstone for analysis, where a trained clinical scientist will compile the test results into a report that is sent to the patient and their doctor. 

What GC-MS instrument / column do you use to detect hydrogen and methane? 

We use Selected Ion Flow Tube Mass Spectrometry (SIFT-MS) – you can find more in this presentation from our conference last year. 

Is your collection device suitable for connection to the circuit of a ventilator? The reference you provided by Ahmed and colleagues suggested samples can be collected in mechically ventilated patients. 

Our collection device is not suitable for this at present. Waqar Ahmed et al, working with Stephen Fowler in Manchester, developed their own approach to do this. While ventilator sampling isn’t something we’re currently developing, we’ll be following developments by others with interest, and if this is of high importance for studies we could collaborate with, for example Waqar and Steve, to work out how to make this happen.  

Could this technique be used in conjunction with nutritional-provision research, particularly in seniors? 

Absolutely! This technology is ideally suited to such applications, where you are very limited with samples that can be taken (e.g. limited blood volumes). Breath sampling is easy and has high acceptability with patients, even those with breathlessness, read this review to find out more.

Being that you use CO2 sensors to “eliminate” lung specimens from your dataset, would it be possible to target lung samples specifically using your system? 

It is very challenging to target lung specimens specifically with breath sampling – this would require removing the alveolar fraction of breath, and we can’t control any mixing or diffusion within the airway, so getting precise, narrow fractionation outside of the airway will always be challenging. Everyone has different lungs, and removing such a small fraction (the alveolar part) to make this work would be extremely challenging for an algorithm, and with sensor and pump response times. It would likely result in very variable samples between patients which would make drawing any conclusions from the data challenging. Currently we remove the dead space with our algorithm which we believe is more reliable, resulting in breath samples with low variability but enriched for signal versus noise.  

Have you applied your home-sampling breath tests in pharma industry driven clinical trials in phase II or III with unattended (non-supervised) sampling with hundreds to thousands of patients? If so what was the experience from this concept? 

We don’t currently have a home-sampling system for VOC discovery, so unfortunately this specific situation has not been tested. However, our clinic-based breath sampling system has, and is, being deployed in multiple clinical trials – including those with both hundreds and thousands of patients and with multiple clinical sites – one of our studies has collected around 4000 samples. These studies are typically observational and are aiming to find biomarkers, however, we have been involved in phase II studies with more planned. Our hydrogen and methane breath tests are home based – patients are sent out kits in the post, which are sent back to us for analysis.  

With possible plans to conduct breath sampling at home, what are some ways to preserve breath samples without altering its metabolite composition? 

Sampling onto sorbent tubes is our strategy for breath sampling in the clinic, and this would translate to the collection of breath samples at home. Breath volatile metabolites are stable on sorbent tubes (you can send them in the post at room temperature), so this is not a major concern for us. Owlstone has already developed a sampling system focussed on collected a specific VOC with sorbents, which is suitable for home-based used. 

We might have an interest in studying VOCs in in-vitro systems in the future (e.g. cell culture + /- drugs). May you please comment on the background introduced by culture media? Does it introduce a major challenge for the in vitro analysis? What do you recommend?  

Great question – culture medium can certainly introduce background into samples as it will emit its own set of VOCs. Suitable experimental design here is critical – control samples of media must be analysed alongside samples of the cells being cultured in it. VOCs that are seen as ‘above media background’ can then be focussed on for data analysis. This is similar to the principle of taking system blank samples for comparison with breath, more here.

Can it be used on e-cigarette smokers to find possible VOCs and biomarkers to find possibility of lung cancer? 

This could certainly have applications here. Critical would be designing a suitable study to help identify any relevant VOCs. If this is of interest, please do reach out to us.  

When detecting breath VOCs, how do we know that the VOCs picked up are not from non-pathogenic bacteria or from bacteria consumed by subjects e.g. probiotics? 

It sounds like you’re wanting to look specifically at pathogens here. In a breath sample you will get VOCs produced from around the body and from many processes – work would be required to understand where they are coming from. Understanding which VOCs are coming specifically from pathogenic bacteria versus from the healthy microbiome could be in a few different ways. First, headspace work comparing cultures of the target pathogenic bacteria could be compared to common non-pathogenic bacteria/probiotics. Second, clinical breath studies could be designed to identify VOCs specific to the pathogenic bacteria and its effects on the body. For example, taking a diverse population of patients – one group with the pathogenic bacteria and one without the pathogenic bacteria, could help here as long as there is equivalent diversity of healthy microbiome and probiotic use between each group.  

Do you have info about the SCFAs concentration levels determined in breath of healthy population and perhaps ones with increased GIT bacterial activity or after the exercise? 

A recent scientific paper by Neyrick et al. examined SCFA levels in breath (and feces!) from 15 healthy subjects before ingestion of a fermentable substrate or a fully digestible (control) carbohydrate. For exercise-induced changes, you may wish to review the work of Henderson et al., though total SCFAs on breath are presented in the main text of the paper and acetic acid alone is presented in the supplemental data.

Do you also work with 13C labelled substrates to be able to see the metabolic pathways without interference from the endogenous background in breath? Such as 13C-Urea Breath Test? 13C Octanoic acid breath test and more? 

We think this is a fantastic idea, especially to help understand how specific nutrients are metabolised by the gut microbiome to demonstrate gut specificity for VOCs, and it is perfectly feasible with our platform. For development of clinical tests, we prefer an approach that doesn’t require the use of 13C-labelled compounds as this makes the clinical use of these tests significantly more challenging (see the inspiration behind our EVOC Probe strategy). Additionally, many 13C-labelled tests rely on detection of 13C-labelled CO2 – thus, only one probe for one pathway can be used at once. Our approach allows for multiple metabolic pathways to be explored from one breath sample. 

You mentioned about unique VOCs in specific pathogenic bacteria. Are such unique VOCs known for Helicobacter pylori?

A somewhat surprisingly small number of studies have examined exhaled breath VOCs for H. pylori infection, and this may be the result of an existing breath test for the condition. In 1996, the FDA approved a 13C-urea breath test for the diagnosis of H. pylori that measures an increase in urease activity when infection is present, as measured by 13C-labelled CO2 on exhaled breath. (Bacterial urease activity is low in the absence of infection.) With an existing test in place, it’s plausible that fewer studies have been carried out in search of H. pylori-specific biomarkers on breath.

In the few studies that have characterized breath VOCs associated with H. pylori infection (e.g., Ulanowska et al. and Leja et al.) or even headspace analysis of reference bacterial strains (Lechner et al.), VOCs exclusive to H pylori were not observed. Rather, alterations in more ubiquitous VOCs were identified. Thus, while the gut microbiota may not always generate unique volatile markers, changes in the levels of hundreds of volatile compounds may produce bacterial strain-specific VOC profiles.

Get expert support from our team at every stage of your research with Breath Biopsy Services