You are here
Abstract: Breath volatile organic compound analysis for the diagnosis of oesophago-gastric cancer; multi-centre blinded validation clinical trial
S. Markar1, T. Wiggins1, S. Antonowicz1, J. Lagergren2, M. Mughal3, G. Hanna1
1Imperial College London, Department of Surgery & Cancer, London, United Kingdom;
2Karolinska Institutet-, Department of Molecular Medicine & Surgery, Stockholm, Sweden;
3University College London, Department of Surgery, London, United Kingdom
Background: Only one-third of patients with oesophago-gastric cancer (OGC) are treated curatively, due to late presentation, giving five-year survival of 15%. Previous research conducted within the group developed a 13 Volatile Organic Compound (VOC) model from exhaled breath, analysed with selected ion flow-tube mass spectrometry (SIFT-MS) for the diagnosis of OGC [PMID:25575255]. Re-analysis of previously collected data reduced the model to 5 VOCs, maintaining a sensitivity of 84% and specificity of 88%. The aim of this study was to determine the accuracy of this breath model for the diagnosis of OGC in a multi-centre blinded validation study.
Methods: Based upon a 1:1 cancer:control ratio, and maintaining a sensitivity and specificity of 80% the sample size required was 325 patients. Breath samples were collected within steel breath bags from 3 sites, and
returned to a central laboratory for SIFT-MS analysis. The risk of cancer was established based upon the previously generated 5-VOC model by a statistician blinded to the diagnosis of the patient. Quality assurances
measures included sampling room air, training all researchers to take the breath samples reliably, and calibration to water.
Results: 335 patients were included; 172 patients as control and 163 patients with OGC. Of the OGC group 69% were T3/4, and 65% had nodal positive disease on clinical staging, all patients were on a curative treatment pathway. 4 of the 5 VOCs previously identified, were significantly dysregulated in the OGC compared to the control group. This association with OGC persisted following adjustment with regression for confounders
including patient age, medical comorbidities and medications. The predictive probabilities generated by this 5 VOC diagnostic model were used to generate a ROC curve, with good diagnostic accuracy, area under the curve of 0.85. This translates to a sensitivity of 80% and specificity of 81% for the diagnosis of oesophago-gastric cancer.
Conclusion: This study shows the potential of breath analysis in noninvasive diagnosis of OGC. The potential benefits of this technology to patients may be early diagnosis and improved of survival. If placed as a endoscopy triage test, the benefits to the healthcare system may include cost-saving through reducing the number of negative endoscopies. However these findings must be further validated in an un-enriched larger
population of patients undergoing diagnostic endoscopy and in false negative patients the value of repeat testing should be established.
Funding: This study was supported by the NIHR (NIHR-DRF-201407–088). Trial registration: This study has been registered on the National Institute for Health Research clinical trials portfolio (UKCRN 18063).
No conflict of interest
European Journal of Cancer, Volume 72, Supplement 1, Pages S3–S4
© 2017 Elsevier Ltd. All rights reserved