Press "Enter" to skip to content

E-sniffer predicts non-responders to lung cancer immunotherapy

An electronic “nose” that detects chemicals in the breath of lung cancer patients may help distinguish between those who will and will not respond to immunotherapy, a new study suggests.

In the small study, the eNose was better at identifying the patients likely to benefit from immunotherapy than the current gold standard test, researchers report in the Annals of Oncology, online September 17.

“We can differentiate between responders and non-responders in patients with non-small cell lung cancer with an accuracy of 85%,” said study coauthor Dr. Mirte Muller, a researcher at the Netherlands Cancer Institute in Amsterdam.

That means that for 24% of patients in the study, immunotherapy could be skipped because it would not help, Muller said in an email.

The eNose didn’t sniff out all the non-responders, Muller said, noting that “44% of patients who did not respond (to immunotherapy) had a positive test.”

Lung cancer is the leading cause of cancer death in men and women, according to the American Cancer Society. Most lung cancers, 85%, are non-small cell lung cancers, which are slower growing than small-cell lung cancers.

The researchers suspected that chemicals patients exhale while breathing could provide clues to whether immunotherapy might be beneficial. The eNose contains sensors to detect volatile organic compounds, which make up about 1 percent of the air people exhale.

To determine whether the sniffer could use those chemicals to identify patients would respond to immunotherapy, the researchers found patients who were about to undergo immunotherapy with anti PD-1 drugs, pembrolizumab or nivolumab, and who were willing to join the study.

Overall, 143 patients volunteered. Before immunotherapy, volunteers were asked to take a deep breath, hold it for five seconds and then slowly exhale into the eNose.

The researchers used the breath data and actual three-month outcomes from the first 92 patients to calibrate, or “train,” the algorithm that analyzed eNose data to predict responders and non-responders. Then the data from the other 51 patients was used to determine whether the electronic sniffer could correctly identify who would and would not respond to immunotherapy.

While immunotherapy “is well-tolerated, 10% to 20% of patients develop side effects, which can be severe, such as hepatitis or colitis,” Muller said. “Patients who do not respond to immunotherapy can still develop a side effect. Therefore it would be better not to give therapy to patients who will not respond.”

The new results are “intriguing” but “preliminary,” said Dr. Edward Garon, director of thoracic oncology at the University of California, Los Angeles, Jonsson Comprehensive Cancer Center.

And while there “is a great enthusiasm about the idea of predictive biomarkers, you would want to see larger datasets studied in a prospective fashion before looking at this as a diagnostic tool,” Garon said.

“It would also be helpful to have a mechanistic explanation of why this would be predictive of immunotherapy outcomes,” added Garon, who was not involved with the new research.


Ann Oncol 2019.

Our Standards:The Thomson Reuters Trust Principles.