Work

Multimodal preoperative risk stratification of pancreatic mass lesions using clinical variables, imaging radiomics and plasma cfDNA methylation: protocol for an ambispective multicentre diagnostic prediction model study

Year: 2026

Type: article

Abstract: INTRODUCTION: Preoperative risk stratification of pancreatic mass lesions is challenging because malignancies, premalignant, inflammatory and indolent lesions can overlap on conventional assessment. Clinical variables, imaging radiomics and plasma cell-free DNA (cfDNA) methylation may offer complementary information. We aim to develop and externally evaluate a multimodal model for clinically high-risk lesions, define interpretable risk strata and assess whether cfDNA methylation adds value beyond clinical and CT-radiomics information. METHODS AND ANALYSIS: This ambispective diagnostic prediction model study will include approximately 3000 adults at eight Chinese tertiary hospitals: 1500 at a lead development centre and 1500 at seven external evaluation centres. The primary target condition is a prespecified diagnosis-based composite, assigned independently of management and predictor values. A hierarchical reference standard prioritises surgical pathology, followed by definitive biopsy or cytology and blinded adjudication with longitudinal follow-up for non-operated participants. Predictors will comprise clinical variables, contrast-enhanced CT radiomics and plasma cfDNA methylation markers measured using a prespecified targeted methylation panel. The primary multimodal model will integrate these three domains using elastic-net logistic regression; MRI radiomics will be analysed in a separate secondary pipeline. Threshold-selection rules will be fixed before fitting; the final preprocessing pipeline, model and two cut-points will be locked before external evaluation. Performance will be assessed by discrimination, calibration, decision-curve analysis and model-based descriptive analyses. Predicted probabilities will define low-, intermediate- and high-risk strata for multidisciplinary assessment. ETHICS AND DISSEMINATION: Ethics approval has been obtained from three participating centres. Ethics approval, institutional authorisation or documented reliance will be obtained from the remaining centres before local prospective enrolment or data contribution, where required. Written informed consent will be obtained from prospectively enrolled participants; consent may be waived for eligible retrospective cases where permitted. Findings will be disseminated through peer-reviewed publication and conference presentation. TRIAL REGISTRATION NUMBER: NCT07045571.

Source: BMJ Open

DOI: 10.1136/bmjopen-2026-124165