• Susanne Roth, Promita Bose, Mohamed S S Alhamdani, Shakhawan A Mustafa, Christine Tjaden, Katharina Zamzow, Ulf Hinz, Christoph W Michalski, John P Neoptolemos, Jörg D Hoheisel, Markus W Büchler, and Thilo Hackert.
• Department of Surgery, Heidelberg University Hospital, Heidelberg, Germany.
• Ann. Surg. 2021 Jun 1; 273 (6): e273-e275.

ObjectiveTo propose a noninvasive diagnostic approach, which allows reliable distinction between low- and high-risk pancreatic intraductal papillary mucinous neoplasms (IPMNs).BackgroundIPMNs are identifiable precursor lesions of pancreatic cancer, of which surgical resection is warranted prior to the development of invasive carcinoma, but low-grade IPMNs should not be unnecessarily resected. However, diagnostic tools that preoperatively enable accurate risk stratification of IPMNs are missing.MethodsThis single-center, retrospective cohort study included 56 patients who underwent surgical resection for IPMN including 18 low-risk (low-grade) and 38 high-risk (high-grade/invasive carcinoma) IPMNs, from whom clinical features and serum samples were prospectively obtained. An antibody microarray platform was used to analyze the serum proteome. Based on serum markers and selected clinical characteristics support vector machine models were constructed to predict the risk of IPMN malignancy.ResultsA serum protein signature discriminating low- and high-risk IPMN patients was identified. Combinations of established clinical features and the newly identified serum biomarkers correctly distinguished low- and high-risk IPMNs in 93% on 1000-fold cross-validation.ConclusionsThis study highlights the synergistic predictive value of combining a novel serum protein signature with conventional clinical characteristics to risk-stratify IPMN patients. If these findings are supported by larger validation studies, they might enable more rational decision-making in clinical management of IPMN patients in conjunction with clinical guidelines.Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.

16 keywords...

hide…