Accuracy of volatile urine biomarkers for the detection and characterization of lung cancer. Issue 1 (December 2015)

• Record Type: Journal Article
• Authors: Mazzone, Peter; Wang, Xiao-Feng; Lim, Sung; Choi, Humberto; Jett, James; Vachani, Anil; Zhang, Qi; Beukemann, Mary; Seeley, Meredith; Martino, Ray; Rhodes, Paul
• Is Part Of: BMC cancer. Volume 15:Issue 1(2015)
• Page Start: 1
• Page End: 6
• Publication Date: 2015-12
• Subjects: Volatile organic compounds -- Urine -- Biomarker -- Lung cancer; Cancer -- Periodicals; 616.994005
• DOI: 10.1186/s12885-015-1996-0
• Languages: English
• ISSNs: 1471-2407

Abstract

Abstract Background The mixture of volatile organic compounds in the headspace gas of urine may be able to distinguish lung cancer patients from relevant control populations. Methods Subjects with biopsy confirmed untreated lung cancer, and others at risk for developing lung cancer, provided a urine sample. A colorimetric sensor array was exposed to the headspace gas of neat and pre-treated urine samples. Random forest models were trained from the sensor output of 70 % of the study subjects and were tested against the remaining 30 %. Models were developed to separate cancer and cancer subgroups from control, and to characterize the cancer. An additional model was developed on the largest clinical subgroup. Results 90 subjects with lung cancer and 55 control subjects participated. The accuracies, reported as C-statistics, for models of cancer or cancer subgroups vs. control ranged from 0.795 – 0.917. A model of lung cancer vs. control built using only subjects from the largest available clinical subgroup (30 subjects) had a C-statistic of 0.970. Models developed and tested to characterize cancer histology, and to compare early to late stage cancer, had C-statistics of 0.849 and 0.922 respectively. Conclusions The colorimetric sensor array signature of volatile organic compounds in the urine headspace may be capable of distinguishing lung cancer patients from clinically relevant controls. The incorporation of clinical phenotypes into the development of this biomarker may optimize its accuracy.