B-340 Validation of a five-biomarker test in serum for detection of early-stage pancreatic ductal adenocarcinoma

Abstract Background Diagnosing pancreatic ductal adenocarcinoma (PDAC) before it can metastasize is key to improving its current 13% 5-year survival rate. Diagnostic imaging technologies have limitations in detecting early-stage tumors, and no biomarkers are approved for PDAC diagnosis. We recently reported a 5-plex serum biomarker signature that detected Stage 1 and 2 PDAC with 85% sensitivity at 98% specificity in a cohort of diagnosed patients and genetic/familial high-risk controls. This signature is comprised of carbohydrate antigen 19-9, measured on a Roche Cobas instrument, and of tissue inhibitor of metalloproteinases 1 (TIMP1), intracellular adhesion molecule 1 (ICAM1), cathepsin D (CTSD), and thrombospondin 1 (THBS1), which are measured using ELISA assays. Here, we report the analytical and clinical performance of the test. Methods Precision in serum was measured across the concentration range for each analyte by running 4 replicates of each of three QC levels daily for at least 20 days (n 80 per level). Linearity, matrix effects, hook effect, analytical specificity, interference, analyte stability at room temperature, refrigerated conditions, freeze/thaw cycles and preliminary long-term storage were also evaluated. Acceptance criteria for precision and linearity were %CV 15%. Analytical specificity, interference, and freeze/thaw stability were acceptable if they showed 20% bias from control samples. Matrix effects were assessed using a parallelism experiment. Long-term stability was preliminarily assessed by performance of the biomarker signature during the clinical validation study. Confidence in the algorithmic score was modeled using clinical validation data in a Monte-Carlo simulation that incorporated the precision of the individual biomarkers. Results All analytes met acceptance criteria for inter- and intra-day precision and linearity. Matrix effects did not impact quantitation nor was hook effect present. The biomarkers were stable in serum at room temperature for at least 24 hours, at refrigerated conditions for 72 hours, and through multiple freeze/thaw cycles. For interference, most experiments met acceptance criteria except: low concentrations of CTSD and ICAM spiked with hemoglobin showed 20% bias, high and low concentrations of CTSD spiked with bilirubin demonstrated 22.9% and -21.0% biases, respectively, and high concentration ICAM1 samples spiked with bilirubin showed a -21.4% bias. Clinical validation performance showed that early-stage PDAC was detected with 82% sensitivity and 95% specificity in samples stored 5 years. Performance declined to 75% sensitivity and 84% specificity in samples stored 5 years. The Monte-Carlo simulation modeling showed that only 1.7% of the clinical validation scores were more than 20% uncertain in the confidence of the negative or positive call. Conclusion All protein biomarkers for PDAC demonstrated acceptable analytical precision, linearity, specificity, and stability. The quantitative impact of any interferences measured was minimal, and unlikely to affect the performance of the test. Although a decline in clinical performance of samples stored 5 years was found, in practice the test will analyze freshly-collected samples thus final test performance is not impacted. The algorithm confidence modeling showed that the measured biomarker precision gives high confidence in the positive or negative result of the test.