4229 Background: Early diagnosis remains the major unmet challenge in pancreatic ductal adenocarcinoma (PDAC), which accounts for approximately 3% of all cancers but nearly 8% of cancer-related deaths in the United States, reflecting its exceptionally poor prognosis. Overall five-year survival remains below 15%, largely because the majority of patients present with unresectable or metastatic disease at diagnosis. Curative surgical resection is feasible in fewer than 20% of cases. There is a critical need to improve risk stratification and diagnostic triage among high-risk populations, including individuals with pancreatic cystic lesions, chronic pancreatitis, and new-onset diabetes over the age of 50. However, the low short-term incidence of PDAC in these groups limits the feasibility of widespread surveillance using current imaging modalities. Methods: In this study, a spectroscopic blood test has been evaluated as an alternative strategy for PDAC detection. This technology employs infrared (IR) spectroscopy to interrogate blood samples, generating disease-specific spectral signatures sensitive to cancer-associated biochemical alterations. When integrated with machine learning, the approach enables detection by capturing global changes across all biomolecular components of the sample. Initially, spectra from 166 patients with PDAC were classified against those from 459 symptomatic patients with non-cancer diagnoses. The trained algorithms are then independently tested on an additional clinical dataset focused on the higher risk population (n = 975). Results: The initial receiver operating characteristic (ROC) curve reported an area under the curve (AUC) value of 0.84. The diagnostic algorithm reported 92% sensitivity with 52% specificity. Importantly, the model did not seem to be affected by cancer stage. The detection rates with the sensitivity-tuned model were 88% stage I, 94% stage II, 99% stage III and 95% stage IV. Validation testing provided additional confirmation of diagnostic ability in the intended use population. The ROC curve for PDAC versus all control patients reported an AUC of 0.81, and PDAC classified against high-risk non-cancers produced an AUC of 0.83. Conclusions: Earlier detection of PDAC has the potential to substantially improve patient outcomes and survival. While advances in genetic sequencing have enabled the identification of tumor-derived biomarkers—including circulating tumor DNA, exosomes, and microRNA—these approaches are limited in early-stage PDAC due to low tumor burden and weak signal from circulating markers. A spectroscopy-based, multi-omic blood test represents a novel strategy that may overcome these limitations, particularly in high-risk populations.
Butler et al. (Wed,) studied this question.