Use of Circulating MicroRNAs for Early Detection and Risk Assessment for Pancreatic Cancer

Detecting pancreatic cancer at an early, asymptomatic stage is a top priority for reducing incidence and mortality of this most fatal disease. MicroRNAs (miRNAs) regulate gene expression and play a critical role in carcinogenesis and cancer progression. In 2014, two blood miRNA-based diagnostic indices capable of distinguishing pancreatic cancer patients from healthy controls and chronic pancreatitis patients (AUC: 0.83 to 0.75) were reported. The utility of these indices for pancreatic cancer early detection, however, has not been evaluated. Recently, we completed a pilot study in which we measured 800 miRNAs in pre-diagnostic plasma samples from 185 pancreatic cancer cases and individually-matched controls selected from the Prostate, Lung, Colorectal and Ovary Cancer Screening Trial (PLCO). We applied the algorithms of the two reported diagnostic indices and found that one index is reasonably effective in discriminating pancreatic cancer patients from controls within 2 years (AUC= 0.73) but not from 2- <5 years (AUC=0.68) before cancer diagnosis. The AUC increased to 0.83 for both <2 years and 2- <5 years prior to cancer diagnosis when 4 miRNAs selected from our own pilot study were applied as the classifier; and for <2 years, further increased to 0.92 when CA19-9 and known risk factors were added to the classifier. To validate these highly promising findings and to evaluate the time window during which the discriminative/predictive miRNAs are most informative for cancer early detection and/or risk assessment, we propose a comprehensive and confirmatory study using additional PLCO resources as well as resources from four prospective cohort studies: the Southern Community Cohort Study (SCCS), the Multiethnic Cohort Study (MEC), the Shanghai Women's Health Study (SWHS), and the Shanghai Men's Health Study (SMHS). Specific aims for the study are: (1) To evaluate all miRNAs showing a significant association with pancreatic cancer risk in our pilot study and from publications in an independent set of 341 case-control pairs using plasma collected within 5 years prior to cancer diagnosis. (2) To evaluate the association of miRNAs replicated in Aim 1 in an independent set of 924 case-control pairs with plasma samples collected >5 years prior to cancer diagnosis by time windows (e.g., 5-9 years and 10+ years). (3) To develop indices for early detection (based on miRNAs showing a stronger association in the time window closer to diagnosis) and risk assessment (based on miRNAs showing a long-term association, e.g., 5+ years prior to cancer diagnosis) using the PLCO data and validate them using samples from the SCCS, SWHS, SMHS and MEC. (4) To assess the added value of monitoring secular changes in levels of replicated miRNAs for cancer early detection using repeated pre-diagnosis plasma samples from the PLCO cases and controls. Performance of miRNA-based indices will be compared to those based on CA19-9 and TIMP1 and known risk factors. Built upon strong pilot data and unique pre-existing resources from cohorts that include multiple ethnic populations, this innovative study has significant potential to generate highly impactful and translatable results.