Dr Morad Nakhleh, our Head Commercial Biomarker Strategist and Dr Jason Kinchen, Lead Biomarker Discovery Scientist, presented a webinar discussing the developments and applications of Breath Biopsy® OMNI, our most advanced solution for reliable global breath biomarker analysis.
Following the presentation, attendees asked our team a number of great questions that we thought might have wider interest for our community – so we're sharing them here too. During the live Q&A session, questions were also answered by Julia Greenwood, Technical Program Manager and Jeff Buckthal, Senior Business Development Executive.
Your questions answered
How did you go about selecting the standard compounds and how representative are they of the VOCs you can detect on breath?
OMNI uses two sets of standards, a set of 52 that we use to generate calibration curves during each GC-MS analytical sequence and eight internal standards that are added to each sample prior to analysis.
The set of 52 compounds were chosen because they’re commonly found in breath samples, including some that are on breath so occur at much higher levels in breath samples compared to ambient blanks. The mix of on-breath and other compounds is useful for understanding the behaviour of possible biomarkers and contaminant compounds.
The internal standards are deuterated so they’re uniquely identifiable in samples and don’t overlap with their undeuterated equivalents. These were chosen as a set of compounds that are distributed across the GC-MS chromatograms allowing us to ensure we’re getting suitable analytical performance throughout every sample.
Why were patients with fibrosis not included in the limonene breath test study in collaboration with the Cleveland Clinic?
This study was conceived as a bookend design – that is, a pilot study where subjects with more severe disease are used because effect size in fully-developed (decompensated) disease is generally larger compared to early disease. To give some perspective, many of the patients here are transplant candidates, with average MELD score in this cohort being 13 (with transplant usually initiated at 15). As we plan the next phase of studies, or aim is to to expand observations here to include patients with earlier / less severe disease.
More details on this study are included in our research poster.
Could you describe the sampling system?
We use the Breath Biopsy Collection Station to collect breath samples. This includes the ReCIVA® Breath Sampler that actually collects the samples, and the CASPER® Portable Air Supply which helps to exclude contaminants coming from the surrounding environment.
ReCIVA collects each sample onto a Breath Biopsy Cartridge consisting of four sorbent tubes, which are easy to store and transport for analysis. Removing samples from sorbent tubes requires high temperatures, which means samples can be transported under ambient conditions, even for overseas transport back to the Breath Biopsy Lab.
In terms of size ReCIVA is a handheld device and CASPER is about the size of a small printer. The Collection Station also includes a trolley, that allows the whole system to be easily moved around, and also a laptop with the Breath Biopsy Collect software that connects to and operates ReCIVA during sampling. We provide a 90 minute training session, which can be done online, to help new teams to get started with the Collection Station.
Are there any standard compounds affected by circadian rhythm and is daytime standardization necessary when considering breath sampling in clinical studies?
Metabolomic datasets often include indicators of circadian effects – as a result, time-dependent shifts in substrate availability can impact data collection and interpretation. For example, glucose availability after fasting impacts the need to use lipid reserves, which can affect metabolites related to lipid oxidation. The ketones described in this talk may be another example.
Circadian effects are interesting from an academic standpoint, but often the goal in breath studies is to minimize these effects to reduce variation which may confound biomarker identification. We can recommend a few best-practices:
- Dietary intake is a driver of circadian metabolic shifts. Where possible, collecting samples from fasted subjects can minimize diet- and circadian-related noise in the data.
- Ensuring that case and control samples are collected at standard times or randomized across the day will also reduce circadian effects on subsequent analysis.
How long does it take to process a sample?
One of the great things about using ReCIVA to collect VOCs from breath is that we can improve sensitivity to low abundance compounds by collecting samples for longer. We’ve typically found that we get the best results if each sample collection is between 12 and 15 minutes.
Our priority for OMNI is early-stage biomarker research, so our priority is sensitivity and consistency of results allowing us to more reliably identify and validate prospective biomarkers. This means that for each study we store all of the samples until sample collection is complete so we can analyse them all at the same time. By doing this we minimize the effects of variation in the analytical system over time. As such, the length of time it takes to report the results of a study is largely dependent on how long it takes to collect all the samples.
Of course, once biomarkers have been identified and we’re closer to clinical applications, we will be able to deliver results much faster. Our digestive health breath testing pipeline, for example, returns results within a few days. For certain biomarkers it may even be possible to produce point of care devices that may return results immediately or within a matter of minutes.
Can BoB studies be used with other breath analysis methodologies?
Certainly this is something that can be applied more broadly, we wanted to develop something that’s independent of method and we’ve used it extensively to improve our own approaches. Comparing samples to background isn’t an uncommon concept but hasn’t been widely applied in breath. Our focus on on-breath compounds is what makes our approach unusual, and we’d encourage others to use the definition of three standard deviations above the background, to determine on-breath compounds and to use this in reporting their results.
We talk more about the use of BoB studies in our BoB blogpost.
Can you identify where noise in a specific sample set is coming from e.g. differences in drugs or diet?
The answer there is a qualified yes, in some cases – it really depends on what type of metadata you have available to you. When we run projects we try and gather as much of the clinical metadata as we can so we can look for those effects in the data and then compensate for them in our statistical analyses. Generally, it's very dependent on having access to descriptive variables where we can identify what those factors are that are driving differences, so it’s often not so easy.
From our work, we have identified several variables that contribute noise to studies, including recent dietary intake, smoking habit, and presence of halitosis. Co-morbidities (e.g. diabetes) also have potential to add noise to studies as well – power in a study can become limiting if a large number of these factors are present. Inclusion and exclusion criteria in clinical protocols need to be crafted to ensure these effects are limited in a given dataset.
Is there any evidence so far of a VOC signature in a discovery cohort that's been validated in a validated cohort in any disease using OMNI?
Limonene as a marker of liver disease is perhaps one of the most promising examples of a prospective VOC biomarker on breath. It was initially reported by Fernandez del Rio et al. 2015 and O’Hara et al. 2016, showing elevated limonene in patients with cirrhosis which recovered following liver transplant.
Our published work (Ferrandino et al.) provided an independent validation of this association, which was further backed up by our work with Cleveland Clinic presented here by Jason. The exciting part about limonene is that it has exogenous origins which makes it suitable as an exogenous VOC probe (EVOC® Probe) that can be administered in a standard dose to stimulate a measurable metabolic response.
OMNI is an evolving solution, and the above work was performed using a similar method to the current OMNI approach. We have performed a number of internal studies that demonstrate the reproducibility of the OMNI method as we’re initiating a number of exciting projects in different disease areas. Since OMNI uniquely differentiates on breath VOCs from background signals, we also expect it to reveal some proposed biomarkers that are actually false positives and cannot be validated.
How comprehensive is the ability to use breath to monitor the whole body?
Blood moves around the whole body and we know that many volatile compounds from blood exchange readily into breath within the lungs. Since all our blood circulates through the body and back to the lungs once every minute, and we’re collecting for more than ten minutes, we can be pretty confident that Breath Biopsy samples can contain VOCs from all over the body. This is then validated by the literature, where breath biomarkers have been identified in a number of disease contexts, including the liver (as described here) and the gastrointestinal space.
Hydrogen methane breath tests (HMBT) are one example of clinical breath tests that detect biomarkers generated in the gut that are carried to the lungs by the blood.
How can biomarkers associated with common processes such as oxidative stress and inflammation provide information that is useful in differentiating between diseases?
The products of biochemical reactions are dependent on substrate availability – for example, one of our most interesting readouts in breath are products of oxidized lipids. These are generated from species including complex lipids (phospholipids) and/or free fatty acids. While every organ will have many of the same lipid species present, the composition (ie, ratio) is something that is specific to a given organ. Disease impact may then produce a high level of one derivative in one organ, but a different derivative in another using the same underlying biochemical process.
We discussed the opportunities and challenges of targeting common processes as sources of biomarkers in our inflammation blogpost.
Have you ever thought about using breath analysis for investigating environmental exposure?
Exposure research is actually an area we have a lot of interest in, we’re even involved in EPHOR an EU-funded consortium looking at the impact of occupational exposures on long-term health. We’ve previously presented some of our work in this area including a study detecting compounds associated with smoking.
At the moment we haven’t applied the OMNI method to an exposure study but we do have some work underway in this space. It’s certainly an area where OMNI can be applied and where breath tests have potential to both monitor exposures themselves and to help assess their impacts on metabolism and health.
There’s more information about our work on our exposure research page.
Do you rely on HRAM for compound identification or are you using other processes?
We have our own in-house HRAM library that we can use for the identification of certain compounds, which allows us to have high confidence in the identity of those compounds. For compounds that aren’t in the library, we can also use the NIST library and we’ve done a lot of work to show that we still get a really good match rate when we use NIST with Orbitrap data.
Is there a list of VOCs in human breath available anywhere?
This would be a really great resource to have, and it’s something we’re very interested in, but currently there isn’t anything that’s suitably complete, accurate and up to date. As we work on developing OMNI we learn more and more about breath and the VOCs it contains. One of our goals is to collect enough data to generate a Human VOC Atlas as mentioned by Morad, which would reflect VOCs on breath in different populations and can be used to compare to different disease groups.