Abstract Purpose: New-onset diabetes (NOD) is an early indicator of pancreatic ductal adenocarcinoma (PDAC). There remains a need to establish biomarkers that can reliably identify NOD patients at risk of having occult PDAC. Here, we assessed the predictive performance of single timepoint CA19-9 as well as an established algorithm that considers repeat CA19-9 testing for risk prediction of PDAC among a prospective cohort of patients with NOD. Methods: CA19-9 was assayed in 6,516 serially collected pre-diagnostic plasma samples collected from 2,121 NOD patients from the Consortium of Chronic Pancreatitis Diabetes and Pancreatic Cancer (CPDPC)-initiated NOD study who completed the 3-year study follow-up period. The specimen set included 25 pre-diagnostic samples from the 12 PDAC cases diagnosed during study follow-up. We applied a single threshold (ST) method, which considers CA19-9 levels at a single time point, as well as a previously established parametrical empirical Bayes (PEB) algorithm referred to as PEBCA19-9, which considers prior CA19-9 results, with ‘case’ calls made based on pre-specified cutoffs corresponding to 1% 1-year risk. Resultant CA19-9 continuous data as well as case calls were provided to the EDRN Data Management and Coordinating Center as part of a Prospective-sample-collection-Retrospective-Blinded-Evaluation (ProBE)-compliant Phase 3 biomarker validation study. Sensitivity, specificity, population-level positive predictive value (PPV), and negative predictive value (NPV) are reported. Results: The 3-year incidence of PDAC in the NOD cohort was 0.57%. At the patient level, CA19-9 yielded sensitivity of 83.3% at 97.2% specificity, with respective PPV and NPV of 14.7% and 99.9%. Sensitivity for early-stage (I-II) disease was 100%. In a subset of patients, CA19-9 first tested ‘positive’ at a median (interquartile range IQR) of 7 months (4 to 14 months) prior to clinical PDAC diagnosis. Of the two PDAC cases missed by CA19-9 using the ST method, one (diagnosed with stage III PDAC) was detected using the PEBCA19-9 algorithm. Conclusion and Relevance: In the setting of adult new onset diabetes, CA19-9 is a readily available and promising biomarker that can be leveraged for earlier detection of an underlying pancreatic cancer. Citation Format: Johannes Fahrmann, Camden Lopez, Ehsan Irajizad, Suresh Chari, Jody Vykoukal, Rachelle Spencer, Jennifer B. Dennison, Eugene Jon Koay, Florencia McAllister, Michael Paul Kim, Jo Ann Rinaudo, Phil A. Hart, William E. Fisher, Stephen Van Den Eeden, Bechien Wu, Ziding Feng, Samir M. Hanash, Anirban Maitra. Utility of CA19-9 for prediction of asymptomatic pancreatic cancer among patients with new onset diabetes abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 4020.
Fahrmann et al. (Fri,) studied this question.