Abstract B20: Development and validation of diagnostic biomarker model for detection of early stage pancreatic cancer

Abstract

Abstract Background: Although resectable pancreatic cancers are associated with better survival, only about 10% of pancreatic patients present with localized disease. Imaging modalities, notably endoscopic ultrasound and Magnetic resonance cholangiopancreatography, allow detection of early stage pancreatic cancer or small pancreatic cysts. However, these imaging modalities are not suitable for screening in terms of throughput, cost effectiveness, or invasiveness. Blood-based biomarkers could be ideal for screening of early stage pancreatic cancer. The performance of CA19-9 as a pancreatic cancer biomarker is limited and moreover CA 19-9 is not detectable in 5-10% of subjects with fucosyltransferase deficiency. As a result there is a need for additional markers that complement CA 19-9 for reliable detection of early stage pancreatic cancer. We have previously identified and obtained initial validation data for a set of blood based biomarkers which distinguished cases from control in the pre-diagnostic setting (Faca et al PLoS Medicine, 2008). We have undertaken further validation of this marker panel augmented with novel candidates including autoantibodies to tumor antigens identified by recombinant protein array. Methods: The selected biomarker candidates as well as CA19-9 were further assayed in a training set, consisting of plasmas from 138 pancreatic cancer patients and 81 controls (52 healthy subjects and 29 subjects with chronic pancreatitis) resulting in a combination rule which was tested in an independent cohort consisting of plasmas from 42 early stage pancreatic cancer patients, 50 healthy controls, 29 subjects with chronic pancreatitis, and 14 subjects with benign pancreatic cysts. Results: A logistic regression model was built in the training set using 5 markers. The AUCs of the model and CA19-9 alone were 0.80 (95% CI = 0.71-0.89) and 0.76 (95% CI = 0.67-0.86) (P value for difference in AUC = 0.014) in the training set. In the test set, the AUCs of the model in a comparison of early stage pancreatic cancer vs. (1) healthy controls, (2) subjects with chronic pancreatitis, and (3) subjects with benign pancreatic cysts were 0.91, 0.85, and 0.94 respectively. The marker panel substantially improved the negative predictive values (NPV) at 98% sensitivity in comparison of early stage pancreatic cancer vs. (1) healthy controls and (2) subjects with chronic pancreatitis (NPVs of the panel = (1) 0.96, (2) 0.95, and (3) 0.88; NPVs of CA19-9 alone = (1) 0.83, (2) 0.83, and (3) 0.88). Stratification based on CA19-9 positivity (cut off = 37 units/ml) yielded AUCs of (1) 0.79, (2) 0.72, and (3) 0.90, and NPVs were (1) 0.99, (2) 0.98, and (3) 0.95 in CA19-9 negative subjects. Conclusions: We have validated a marker combination with a potential to differentiate early stage pancreatic cancer from healthy controls, subjects with chronic pancreatitis and subjects with pancreatic cysts. Citation Format: Ayumu Taguchi, Michela Capello, Yang Zhao, Ingrid Babel, Gary Goodman, Margaret A. Tempero, Matthew A. Firpo, Matthew H. Katz, Ziding Feng, Samir Hanash. Development and validation of diagnostic biomarker model for detection of early stage pancreatic cancer. [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer: Innovations in Research and Treatment; May 18-21, 2014; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2015;75(13 Suppl):Abstract nr B20.