Abstract 624: Clinical evaluation of the MasSpec Pen technology for rapid diagnosis and margin assessment in pancreatic cancer surgery

Precise removal of pancreatic ductal adenocarcinoma (PDAC) with microscopically negative margins, commonly assessed by frozen section analysis, is associated with longer disease-free survival. However, histologic complexities and tissue-processing artifacts can render frozen section analysis of PDAC margins a challenging and time-consuming task, with reported accuracies dependent on the skillset and subspecialty of the pathologist on call. We developed the MasSpec Pen, a handheld device coupled to a mass spectrometer, for rapid (~15 s) and nondestructive molecular analysis and diagnosis of tissues. The MasSpec Pen supplies a discrete water droplet onto a tissue's surface, allowing diagnostic metabolites and lipids to be extracted into the droplet and then transmitted into a mass spectrometer for analysis. Here, we evaluate the performance of the MasSpec Pen for intraoperative diagnosis of PDAC in human pancreatic and bile duct margins. Pancreatic and bile duct tissue samples (N=157) were obtained from the Cooperative Human Tissue Network and Baylor College of Medicine and stored at -80°C prior to analysis. A Q Exactive mass spectrometer (Thermo Scientific) coupled to the MasSpec Pen was used for analysis of thawed samples in the negative ion mode. Tissues were then cryo-sectioned, H&E stained, and blindly evaluated by a pathologist. Based on the distinct molecular profiles acquired, we generated two statistical classifiers using lasso penalized logistic regression for distinguishing PDAC from healthy pancreas and bile duct tissue based on a sparse set of molecular features indicative of disease state. For distinguishing normal pancreas from PDAC, an overall accuracy of 91.5%, sensitivity of 95.5%, and specificity of 89.7% was achieved for training, validation, and test sets. Classification results for discriminating normal bile duct from PDAC had an overall accuracy of 95%, sensitivity of 92%, and specificity of 100% in training and validation. We have begun clinical testing of the MasSpec Pen in human surgeries following its successful translation to an operating room at Texas Medical Center. To date, the MasSpec Pen has been used to analyze in vivo and fresh ex vivo tissue in 19 pancreatic surgeries. When predicting on 64 intraoperative analyses using classification models built on banked data, 93.8% agreement with final postoperative pathology reports was achieved. While further validation studies are needed, our results show that the MasSpec Pen can di