Specific patterns of volatile organic compounds (VOCs) in urine samples can identify patients with pancreatic ductal adenocarcinoma (PDAC), researchers report in the February issue of Gastroenterology. The noninvasive technology, based on ion mobility spectrometry, can be used to discriminate between healthy individuals vs individuals with early- or advanced-stage cancer.
VOCs are metabolic products of microbes that can be detected in urine, breath, and feces. Ramesh P. Arasaradnam et al investigated whether alterations in cell functions and microbes in patients with PDAC lead to altered patterns of VOCs that can be detected in urine specimens.
In a video, the authors describe their procedure. The authors collected urine samples from 81 patients with pancreatic cancer (different stages) and 81 healthy individuals. A 5 mL urine sample from each individual was added to special glass vial, leaving some air above the sample. The samples were placed in the ion mobility spectrometry unit and ion mobility was measured in the air above the samples (known as the headspace).
Once all the samples were analyzed, 4 different classifiers were applied to the data (sparse logistic regression, random forest, gaussian process classifier, and a support vector machine algorithm).
Arasaradnam et al report that this technology identified patients with PDAC with 91% sensitivity and 83% specificity, with an area under the curve value of 0.92, using the support vector machine algorithm.
The results were validated by randomly splitting the data into the training and test set, and produced similar results, with an AUC of 0.92, 90% sensitivity, and 81% specificity.
When the authors compared samples from patients with early-stage disease (I/II) vs healthy individuals, the analysis identified the samples with 91% sensitivity, 78% specificity, and an AUC of 0.89.
When they compared samples from patients with early-stage I/II vs late-stage disease (III/IV), the analysis identified the samples with 82% sensitivity and 89% specificity, with an AUC of 0.92, again using a support vector machine algorithm.
Arasaradnam et al say that the advantages of this test are that it requires only a 5-mL urine sample, and it is more rapid and more cost-effective than other gas analysis technologies. The technology provide an exciting new approach to identify biomarkers of early-stage PDAC and provide earlier treatment.