Nanoanalysis of plasma volatile organic compounds using novel DNA-decorated carbon nanotube vapor sensors to noninvasively distinguish ovarian and pancreatic cancer from benign and control samples.

Abstract

5544 Background: All cells release volatile organic compounds (VOCs) which emanate from body fluids. Our previous preliminary proof of concept study demonstrated that VOCs released from tissue and plasma from ovarian cancer patients are distinct from those released from samples of patients with benign tumors and controls. We seek to create a sensitive and specific, high-throughput screening test for cancer based on analysis of VOCs using novel nanosensors, first targeting cancers with limited clinical screening modalities. In this study we use these sensors to distinguish vapor characteristics in plasma samples from patients with ovarian and pancreatic cancer from benign specimens and controls. Methods: VOCs emanating from.5 mL of thawed, previously banked plasma samples from 93 total individuals were analyzed using a 10-channel nanoelectronic olfaction (“e-nose”) system based on single-stranded DNA-decorated single-walled carbon nanotube (DNA-NT) vapor sensors. Analysis was performed on samples from 20 patients with ovarian cancer, 20 with benign ovarian tumors and 20 age-matched women as well as 13 patients with pancreatic cancer, 10 patients with benign pancreatic disease, and 10 age- and sex-matched controls. All ovarian cancer patients and comparators were non-smokers, while 1 pancreatic patient and 1 corresponding control were current smokers. The sample set included cancer patients with both early- and late-stage disease. All cancer specimens were obtained proximal to initial diagnosis and prior to initiation of therapy. With a test time of approximately 20 minutes per sample, the array output for each individual sample creates a vector in a 10-dimensional sensor space. The ability of the nanosensor array to discriminate between malignant, benign, and healthy groups was investigated using linear discriminant analysis (LDA), support vector machine (SVM), k-nearest neighbors (KNN), and random forest classification algorithms. Each algorithm was trained and tested according to leave-one-out and repeated stratified k-fold cross-validation methods. Results: Compared to their corresponding benign and control specimens, the DNA-NT sensor array was able to discriminate the VOCs from ovarian cancer with 95% accuracy and pancreatic cancer with 90% accuracy. Plasma samples from patients with early-stage ovarian and pancreatic cancers were correctly identified by the algorithms. Conclusions: Nano-enabled DNA coated vapor sensors were able to distinguish the VOC pattern between cancer, benign and control samples in both ovarian and pancreatic cancer. We provide strong evidence that ovarian and pancreatic cancer alters the VOC pattern emanating from plasma. Our results provide optimism that a diagnostic approach based on vapor detection of ovarian and pancreatic cancer is achievable.