Background: Only 25% of pancreatic ductal adenocarcinoma (PDA) patients with clinically-confirmed localized disease survive five years post-resection. It is hypothesized PDA undergoes dissemination at the earliest stages of tumor formation, driving formation of occult metastases which go undetected using conventional screening methods. Development of a high specificity, high sensitivity biomarker is critical to improving patient mortality. Circulating tumor DNA (ctDNA) has gained increasing acceptance as a non-invasive prognostic in metastatic disease. However, the sensitivity of most targeted ctDNA assays precludes reliable detection of localized and resected disease. Here, we present a digital droplet PCR assay for multiplexed enrichment and detection of KRAS mutations, the most commonly mutated oncogene in PDA. This assay preserves ctDNA allelic frequency in the original sample, while increasing the molecular signal over 50-fold. This study demonstrates the feasibility of using our strategy for non-invasive ctDNA detection in low DNA input samples from localized cancer plasma samples. Methods: We retrospectively analyzed ninety-eight plasma samples from 51 patients with various stages of localized disease. We also measured ctDNA levels in 20 patients with metastatic PDA. The MED-Amp (Multiplex Enrichment using Droplet Pre-Amplification) assay was used as previously described to quantify abundance of KRAS codon 12 mutant ctDNA. We correlated the presence and quantity of ctDNA with overall survival (OS) as well as progression-free survival (PFS). We also used serial plasma draws to correlate changes in ctDNA allelic frequency and PDA progression. Results: KRAS-positive ctDNA was detected in 75% of metastatic and 52.9% of localized PDA samples tested. We were able to detect single copies of mutant template using DNA inputs as low as 2 ng. Presence of KRAS mutant ctDNA was correlated with worse OS for all disease stages (p = 0.02). High ctDNA levels also correlated with significantly worse OS (533 days vs 1090 days) and PFS (192 days vs 787 days) in patients with localized PDA. We also used serial plasma draws to measure the correlation between ctDNA dynamics and PFS. We found 83% of patients with increased fold change in mutant KRAS experienced disease progression. In contrast, 75% of patients with a fold change decrease did not experience disease progression (p=0.03). These data suggest that MED-Amp is a flexible and cost-effective approach for measurement of ctDNA in patients with localized cancer.