Development of a Predictive Model for Colorectal Cancer Using Exhaled Volatile Organic Compounds

BACKGROUND: Colorectal cancer is lacking non-invasive and efficient diagnostic markers. Recently, volatile organic compounds present in exhaled breath have emerged as potential diagnostic indicators for various diseases.

OBJECTIVE: This study aimed to establish a predictive model including exhaled volatile organic compounds for the diagnosis of colorectal cancer.

DESIGN: Prospective comparative study.

SETTINGS: Single academic medical center.

PATIENTS: Exhaled breath was collected prospectively from 112 colorectal cancer patients and 51 healthy controls. The exhaled breath of the research participants was collected under the control of various conditions, such as eating or smoking prior to collection.

INTERVENTIONS: A total of 53 volatile organic compounds was analyzed in exhaled breath using thermal desorption-gas chromatography/mass spectrometry.

MAIN OUTCOME MEASURES: After correcting various clinical variables that may affect volatile organic compound, the concentrations detected in the colorectal cancer group and the healthy control group were compared.

RESULTS: Four predictive models were established using both matched clinical variables and volatile organic compound values to differentiate between colorectal cancer and healthy controls. In the four models, clinical variables of age, gender, room temperature, heating and cooling, fasting duration, intake of grilled meat or fish, and bowel preparation were included, along with levels of eight volatile organic compounds (ethanol, decanal, 1-iodononane, dodecane, hexanal, 1-nonene, (1S)-(-)-alpha-pinene, and acetophenone). The sensitivity of the four prediction models ranged from 84.0% to 98.2%, the specificity ranged from 89.2% to 93.8%, and the area under the receiver operating characteristic curve ranged from 92.1% to 98.3%.

LIMITATIONS: The demographics of the participants in the groups were different.

CONCLUSIONS: Eight types of volatile organic compounds in the exhaled breath of the colorectal cancer group were significantly different from those of the healthy control group. Our results support the diagnosis of colorectal cancer using a predictive model combining volatile organic compounds and specific clinical variables. See Video Abstract.