David’s focus is actually wider than breath testing. He’s also looking at biomarkers in blood using one of the world’s biggest bio-banks for oesophageal tissue specimens, which runs over four Australian states and which Flinders University has been collaborating in for the past 14 years. For David, it’s a race to see which solution—blood or breath?—is the best. Or perhaps clinicians could use a combination of the two.
For their breath test, the challenge is not only identifying the markers of early stage oesophageal cancer. It’s also in reducing the size of the equipment used to test breath samples (currently the size of a washing machine), while maintaining its high sensitivity and ability to detect small trace quantities. Roger uses a selected ion flow tube mass spectrometer (SIFT MS) to identify individual gases within a complex mixture. It’s the only machine of its type being used in Australia for clinical research—so it’s quite unique in that way. It will need to be much smaller if it will ever be used in clinical or remote environments.
This issue of practicality is why Australia doesn’t currently screen for oesophageal cancer using the current technique, an endoscopy; it’s just not cost effective even in people who present with the condition known as Barrett’s oesophagus, which can develop into cancer. One of David’s passions is health economics. Are we being efficient? Are we wasting money? When it comes to this type of cancer, David says that our resources are being poorly managed. Perhaps, if we knew which people were likely to develop oesophageal cancer, we could start screening and catch the disease earlier. With a less resource intensive test, the standard practices could change for the better. A test like the ones they’re trying to develop (blood or breath) could shift the current cost structure and reduce the endoscopy workload, as well as save lives. At each stage, he and Roger ask themselves: does it work, will it be used by clinicians and patients, can it be commercially produced, and is it affordable?