Abstract LB-135: Adaptive dynamic artificial poly-ligand targeting (ADAPT) enables plasma-based exosome profiling with potential diagnostic utility

Abstract

Abstract Introduction: Single stranded DNA (ssDNA) libraries consisting of several trillion oligodeoxynucleotides (ODNs) can adopt a nearly infinite number of three-dimensional structures. These structures can potentially bind any biomolecule and can be screened for specificity toward important biomarkers by employing suitable enrichment schemes. Since no prior knowledge on the binding partner is required, massively parallel biomarker identification is possible even on complex matrices like biological fluids and across a wide range of biological conditions. Here we present Adaptive Dynamic Artificial Poly-ligand Targeting (ADAPT) as a platform for biomarker and target discovery. We employed ADAPT for the molecular profiling of exosome-associated proteins in small volume plasma samples from women with breast cancer and healthy donors. Results: Random ssDNA-libraries of 1011 unique ODNs were subjected to a number of selection and counter-selection steps on pooled blood plasma of breast cancer and healthy women. Several positive and negative enrichment schemes were employed, and exosome isolation and ODNs library partitioning were performed by ultracentrifugation and/or PEG precipitation. After library enrichment reduction of complexity to 106-107), ODN libraries were used to probe an independent set of individual plasma samples from women with or without breast cancer. Two thousand differentially-binding aptamers with significant p-values were re-synthesized and combined in equimolar amounts to create a profiling library (L2000). The L2000 library was used to probe plasma samples from 323 individuals (206 from breast cancer patients and 117 from healthy donors) in triplicate. Using Next Generation Sequencing, we quantitated bound ODN from each plasma sample. ANOVA revealed 350 aptamers with significant p-values in distinguishing plasma samples from cancer patients and healthy donors, far in excess of the number of ODNs that would have achieved statistical significance by random sampling of the 2000 ODNs. Generalized linear model showed an AUC in a ROC curve of 0.94 for the training set. Random forest modelling was used to assess classification performance and revealed an AUC of 0.73 (p<0.002 in a permutation test) in Out-of-bag validation. A more strict 10-fold cross-validation yielded an AUC of 0.63 (p<0.01 in a permutation test). Conclusions: We have demonstrated the feasibility of aptamer library enrichment directly on blood plasma and have identified a set of 2000 DNA aptamers that distinguish plasma from women with breast cancer from women without breast cancer. This liquid biopsy approach requires only 200 microliters of plasma and is amenable to high-throughput processing. By employing a number of statistical approaches including rigorous cross-validation, we consistently achieve ROC AUC values >0.6. Further optimization of the aptamer library and testing on additional samples is ongoing. Upon complete validation, an ADAPTTM - derived breast cancer test may serve as a vital diagnostic adjunct that can be easily incorporated into standard clinical practice. Citation Format: Valeriy Domenyuk, Zhenyu Zhong, Jie Wang, Adam Stark, Nianqing Xiao, Mark Miglarese, George Poste, Michael Famulok, Günter Mayer, David Spetzler. Adaptive dynamic artificial poly-ligand targeting (ADAPT) enables plasma-based exosome profiling with potential diagnostic utility. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr LB-135.