Search

Pancreatic ductal adenocarcinoma (PDAC) is currently the 3rd leading cause of cancer-related deaths in the US. Although the all-stage 5-year survival rate is ~10%, early-stage 5-year survival is markedly superior and in excess of 40%. Hence, early detection of PDAC via blood-based liquid biopsies holds promise to reduce morbidity and mortality. PrognomiQ’s multi-omics platform performs deep and unbiased molecular profiling of blood samples to detect proteins, metabolites, lipids, mRNA, miRNA, cfDNA fragmentation and copy-number, and CpG methylation. Here we report results from training and validation of a classifier on a subset of that multi-omic data with the potential to enable the development of high sensitivity and specificity tests for early detection of PDAC.We conducted a case-control study comprising 146 subjects across 16 clinical sites, including 63 pathology-confirmed, untreated PDAC cases (12 stage I, 8 stage II, 4 stage III, 36 stage IV, and 3 stage unknown) and 83 age- and gender- matched controls without any known cancer. For each subject, venous blood samples including plasma were collected. Unbiased LCMS was used to detect and quantify proteins, and targeted, multiplexed MRM-LCMS assays were used for both metabolites and lipids. After data processing, we detected 54,114 proteomic features, 898 lipids, and 373 metabolites. 445 proteomic features, 170 lipids, and 37 metabolites were found to be significantly different as determined by Bonferroni-corrected Wilcoxon tests with FWER < 0.05. For classification, the dataset was split into training (37 cases and 37 controls) and validation (26 cases and 46 controls) sets, with control for collection site and date, age, and gender. XGBoost models were constructed for each analyte class using ten repeats of 10-fold cross-validation. To improve specificity to PDAC, all proteomic features which mapped to GOBP terms associated with acute-phase response, inflammation, and immune response were excluded prior to training. The best-performing hyperparameters were used for a final model built on the full training set and then used for inference on the validation set. At 99% specificity, the proteomic classifier had sensitivities of 77%, 57%, and 88% for Stages 1-4, Stages 1-2, and Stages 3-4, respectively, estimated by bootstrap re-sampling of the validation results. Metabolomics had sensitivities of 81%, 71%, and 88%. Lipidomics had sensitivities of 65%, 71%, and 65%. A joint, multi-omic model was constructed by averaging the scaled probabilities of all models. This joint model improved performance at 99% specificity with sensitivities of 92%, 86%, and 94%, highlighting the synergy of multi-omics data, particularly phenotypically related omics such as those described here. Multi-omic classifiers such as these can serve as the foundation for blood-based liquid biopsies for the early detection of PDAC. Citation Format: John Blume, Ghristine Bundalian, Jessica Chan, Connie Chao-Shern, Jinlyung Choi, Rea Cuaresma, Kevin Dai, Sara N. Golmaei, Jun Heok Jang, Manoj Khadka, Ehdieh Khaledian, Thidar Khin, Yuya Kodama, Ajinkya Kokate, Joon-Yong Lee, Manway Liu, Hoda Malekpour, Megan Mora, Nithya Mudaliar, Preethi Prasad, Madhuvanthi Ramaiah, Saividya Ramaswamy, Peter Spiro, Kavya Swaminathan, Dijana Vitko, James Yee, Brian Young, Susan Zhang, Chinmay Belthangady, Bruce Wilcox, Brian Koh, Philip Ma. A multi-omics classifier achieves high sensitivity and specificity for pancreatic ductal adenocarcinoma in a case-control study of 146 subjects [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 6597.

Pancreatic cancer is the third leading cause of cancer-related deaths in the United States. Disease biomarkers quantified from blood-based assays may help reduce mortality by enabling early detection, treatment selection, or response and resistance assessment. PrognomiQ has developed a multi-omics assay and analysis platform that comprehensively profiles blood samples to detect proteins, metabolites, lipids, mRNA, miRNA, cfDNA fragments, and methylation at CpG sites. This platform can provide deep insights into the biology of pancreatic cancer and could enable the development of high sensitivity and specificity tests for the early detection of pancreatic cancer. We conducted a case-control study comprising 196 subjects: 92 with untreated pancreatic cancer and 104 matched controls without pancreatic cancer. For each subject, blood was collected in assay-specific tubes and processed to provide 7 different `omics readouts. cfDNA and mRNA were isolated from samples and assayed following standard NGS protocols. cfDNA fragments were processed to estimate fragment-length disorder and copy-number variation along with CpG site methylation. In addition, targeted and untargeted LCMS were used to detect and quantify proteins, metabolites, and lipids. After normalization, non-parametric univariate analyses of cases versus controls were performed to identify differentially abundant features on all available samples for each assay. Unsupervised learning was used to investigate the separation of subjects into groups based on disease status for the subset of 157 subjects for which complete data on all 7 readouts were available. We detected 2,812 proteins, 811 lipids, 373 metabolites, and 110,864 mRNA transcripts in all samples where data for each assay was available. Of these, 275 proteins, 232 lipids, 49 metabolites, and 3385 mRNA transcripts were significantly different (FWER < 0.05) in cases versus controls. From cfDNA data, we identified 35 non-contiguous genomic regions associated with fragment-length disorder, 557 with copy-number variation, and 5 with multiple, differentially methylated CpGs (FWER < 0.05) that aggregately span 307 protein-coding genes; of these, the overlap with the differentially expressed proteins included E-cadherin (tumor suppressor) and N-cadherin (involved in epithelial-to-mesenchymal transition). Statistically significant genes and proteins were found to be associated with processes including Wnt signaling, regulation of focal adhesion assembly, and actin cytoskeleton organization. Multi-omics, unsupervised learning showed separation of early- and late-stage cases and controls. High-dimensional bioinformatics analyses systematically decomposed each `omics data type into joint and orthogonal components associated with pancreatic cancer. Ongoing multivariate analyses, including supervised machine learning, will further elucidate the biology of pancreatic cancer development, and serve as the foundation for high-sensitivity blood tests for the early detection and monitoring of pancreatic cancer. Citation Format: Ehdieh Khaledian, Preethi Prasad, John Blume, Ghristine Bundalian, Connie Chao-Shern, Jinlyung Choi, Rea Cuaresma, Jared Deyarmin, Jun Heok Jang, Manoj Khadka, Thidar Khin, Yuya Kodama, Ajinkya Kokate, Joon-Yong Lee, Manway Liu, Nithya Mudaliar, Madhuvanthi Ramaiah, Saividya Ramaswamy, Peter Spiro, Kavya Swaminathan, Preston Williams, Mi Yang, James Yee, Brian Young, Robert Zawada, Susan Zhang, Chinmay Belthangady, Bruce Wilcox, Philip Ma. High-dimensional, multi-omics analyses of proteins, metabolites, transcripts, and genes enable biomarker discovery in early- and late-stage pancreatic cancer [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer; 2022 Sep 13-16; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2022;82(22 Suppl):Abstract nr A038.

To date, 70 different TGFBI mutations that cause epithelial-stromal corneal dystrophies have been described. At present one commercially available test examines for the five most common of these mutations: R124H, R124C, R124L, R555W, and R555Q. To expand the capability of identifying the causative mutation in the remaining cases, 57 mutations would need to be added. The aim of this study was to obtain a better understanding of the worldwide distribution and population differences of TGFBI mutations and to assess which mutations could be included or excluded from any potential assay. A total of 184 published papers in Human Gene Mutation Database (HGMD) and PubMed from 34 countries worldwide reporting over 1600 corneal dystrophy cases were reviewed. Global data from 600,000 samples using the commercially available test were analyzed. Case studies by University College of London (UCL), Moorfield's Corneal Dystrophy Study data and 19 samples from patients with clinical abnormality or uncertainty for which the current test detected no mutation were used to predict an achievable detection rate. Data from the literature search showed no difference in the spectrum and frequency of each mutation in different populations or geographical locations. According to our analysis, an increase to the worldwide detection rate in all populations from 75 to 90% could be achieved by the addition of six mutations-H626R, A546D, H572R, G623D, R124S, and M502V-to the currently available test and that may be beneficial for LASIK pre-screening worldwide.