New Study Points to Feasibility of Breath-Based Diagnostic Test for C. diff

Clostridioides difficile infection (CDI) is the leading cause of hospital-acquired infective diarrhea, but there are limitations associated with the currently available methods for diagnosing the condition. A recently published study, “Detection of Clostridioides difficile infection by assessment of exhaled breath volatile organic compounds (VOCs),” explores the potential of a diagnostic test that could evaluate the presence of specific VOCs in a patient’s breath. The study, which appeared in Journal of Breath Research, illustrates how the research group built a quadratic discriminant model that could discriminate between predictors of CDI with moderate sensitivity and specificity.

“Clinicians have noticed that the stools of patients with CDI have a distinct odor and that this odor can help them distinguish between patients with and without CDI,” explains Nabin Shrestha, MD, a staff physician in Cleveland Clinic’s Department of Infectious Diseases and a coauthor on the paper. “This raised the question of whether there are VOCs that can help make this distinction. We tried to examine that in one of our earlier studies by testing the air around the stools and people’s breath.”

He continues, “The idea for this study was to look at the entire collection of compounds that are present in the breath to determine if patients with CDI can be identified. Our research group identified nine specific VOCs that show promise in terms of helping diagnose CDI, but this was a very small study. We still need to determine if these findings can be replicated in a larger study.”

Study design

The prospective, cohort study enrolled patients with diarrhea who were hospitalized at Cleveland Clinic. Participants were at least 18 years or older and were admitted for or who developed diarrhea during hospitalization. All participants were tested for CDI by polymerase chain reaction (PCR) – one of the two most common methods for diagnosing CDI. Patients with a positive PCR test were matched to a patient of similar age and sex who tested negative that same day. Demographic and clinical information including comorbid conditions (diabetes mellitus, chronic kidney disease, chronic liver disease, inflammatory bowel disease, malignancy and transplantation) were collected from each patient’s medical records.

The researchers used thermal desorption-gas chromatography-mass spectrometry to study breath samples from 17 patients with CDI and 17 patients without CDI (34 total samples).

“Because these samples were collected at a hospital, this population had a very high rate of comorbidities,” explains Raed Dweik, MD, the Inaugural Chief of the Integrated Hospital Care Institute (IHI) at Cleveland Clinic and senior author on the paper. “In fact, every single one of the participants had at least one comorbidity, but most comorbidities appeared more frequently among patients with CDI.”

Findings

In total, 65 VOCs were detected and qualified for inclusion in the final dataset. Among these VOCs, four displayed evidence of association with CDI at a level of significance of 0.1, and 12 VOCs had a difference of abundance that neared significance (P < .2) between CDI patients and controls. Using quadratic discriminant analysis (QDA), nine VOCs were identified and yielded a model with a final cross-validated accuracy of 0.74, a sensitivity of 71%, a specificity of 76% and a receiver operating characteristic area under the curve of 0.72.

These nine VOCs include: MF2 n-Hexane, MF18 Tetrachloroethylene, MF53 2-Phenyl-2-propanol, MF26 o-Xylene, MF25 4-Heptanone, MF41 2-Ethyl-1-hexanol, MF46 2,6-Dimethyl-7-octen-2-ol, MF31 1,2,4-Trimethyl-benzene and MF50 3-7-Dimethyl-3-octanol.

Looking ahead

“Essentially, there are two ways one might imagine doing breath testing for diagnosis,” explains Dr. Shrestha. “One would be to collect breath samples and send them to a lab for analysis to determine if the patient has CDI or not. But that’s a cumbersome and time-consuming process that isn’t that much different or easier than how we analyze stool samples now. What would be more beneficial and convenient would be a type of handheld device that could assess patients at their bedsides and provide immediate test results; something similar to the breathalyzer device police officers use.”

Dr. Shrestha notes that he believes something like this is still a long way off. The device would need to select a few molecules to test but which molecules remains a question.

“Every study that I’ve seen that’s tried to identify specific molecules have all returned different results,” he says. “I don’t think we’re at a point yet where we can accurately identify the specific molecules — although it’s clear that the breath composition from patients with CDI and those without are different. To really determine the specific molecules involved, we will need to perform a much larger study with much larger numbers of patients. But I believe the results from our recent study demonstrate the feasibility of testing VOCs in breath samples, and that moving forward on a larger, more expensive study is worthwhile.”