August 23, 2018
As a group, breath tests to detect cancer have a relatively high level of sensitivity and specificity, despite the fact that they are still only in experimental stages of development. This is the conclusion from a new meta-analysis of early trials of the concept.
The study was published online August 16 in JAMA Oncology.
A breath test detects and quantifies preidentified, named volatile organic compounds (VOCs) within exhaled breath in order to diagnose cancer, explain the study authors, led by George Hanna, MD, PhD, a surgeon at the Imperial College London in the United Kingdom.
The approach is still relatively immature and under development by multiple teams in oncology. "There are currently no breath tests used in oncology of cancer screening," Hanna told Medscape Medical News.
However, the concept is already in use in other areas of clinical practice, including breathalyzer devices for ethanol detection, carbon 13 urea breath testing for Helicobacter pylori, and exhaled nitric oxide testing in asthma.
In the new analysis, Hanna and colleagues identified 63 studies, all of which were phase 1 biomarker trials. They report that the studies showed that for cancer diagnosis, the mean (SE) area under the ROC curve for breath VOC analysis was 0.94 (0.01). The team also reports a pooled sensitivity of 79% (95% confidence interval [CI], 77% - 81%), and a pooled specificity of 89% (95% CI, 88% - 90%).
Medscape Medical News asked Hanna to compare the accuracy of the breath test with that of three common cancer screening tests.
He answered that the pooled results compare favorably to existing screening tests for cervical cancer (Pap test: sensitivity, 73%; specificity, 90%), prostate cancer (prostate-specific antigen test: sensitivity, 21%; specificity, 91%) and breast cancer (mammography: sensitivity, 68%; specificity, 75%).
Notably, the meta-analysis included only studies that identified named VOCs altered within the exhaled breath of patients with cancer. Studies were excluded if they involved sensors and pattern recognition techniques that did not report on specific VOCs.
The new study is "a nice compilation of all the [breath test] trials that have been done over the last decade," said Victor van Berkel, MD, PhD, a thoracic surgeon at the University of Louisville in Kentucky and a breath test researcher who was asked for comment.
However, van Berkel was uncertain about the value of the new findings.
"My concern with this review is that it takes a bunch of very different tests, looking at different volatile compounds in different cancers via different mechanisms, and lumps all the results together for an overall sensitivity and specificity. I am not sure what to conclude from those results," he told Medscape Medical News.
"I think it is true that a lot of work still needs to be done with regards to breath tests for cancer," he said. He emphasized the need for standardization and quality control in larger validation trials.
Big Steps Underway, Especially in UK
van Berkel also reported that he and colleagues are in the midst of securing funding for a large, multicenter trial of their own breath test technology, which uses carbonyl cancer markers, to diagnose lung cancer. Their team has reported preliminary results with the test (J Thorac Cardiovasc Surg. 2015;150:1517-1524).
London's Hanna and his colleagues are also on a roll, in terms of taking next steps in the breath test development process.
They externally validated the breath test for esophagogastric cancer in a multicenter clinical trial that was published in May (JAMA Oncol. 2018;4:970-976). In the trial of 335 patients, which was conducted in 2015-2016, their breath test distinguished esophageal and gastric cancers from benign diseases with 85% accuracy (area under the curve, 0.85).
Hanna described some of the hallmarks of his group's work so far. "Crucially, biomarkers were different between different tumor sites, allowing for a single breath test for use as a risk-assessment tool to determine independent referral pathways for suspected esophagogastric, colorectal, or pancreatic cancers," he said.
A confirmatory prospective study has been planned to determine the diagnostic accuracy of a breath test to detect cancer in an enriched population of about 18,000 patients referred from primary care for rapid assessment for esophageal and gastric cancers, in accordance to guidelines from the UK's National Institute for Health and Care Excellence.
"We have also established the framework for standardization and quality control measures for breath analysis, which have traditionally been barriers to the wider adoption of breath testing in clinical practice," he said.
It's unclear how breath tests may be ultimately used in oncology. In their article, Hanna and coauthors review the possibilities.
First, an exhaled breath test could act as a triage investigation to direct patients who present with nonspecific common symptoms that could indicate cancer to have invasive or more specialized investigations, they write.
Second, breath tests may have a role in screening programs (provided they are accurate enough). The noninvasive nature of the test and its acceptability by patients and practitioners would increase the uptake of screening, they add.
Third, breath tests may have a role in monitoring the response to cancer therapy and in detecting disease recurrence, the meta-analysis coauthors believe.
Louisville's van Berkel agrees that much is unknown in the field, including which cancer is the best candidate for breath detection.
"We don't really know what cancers will best be evaluated by breath tests. There has been a lot of focus on lung cancer, with the supposition that since the tumors are directly contacting the breath, they may be more markers identified. But that is purely an assumption," he said.
The study authors and Dr van Berkel have disclosed no relevant financial relationships.