Related Experiment Video
Updated: Jul 4, 2026
07:08
Reprogramming Pancreatic Ductal Adenocarcinoma to Pluripotency
Published on: February 2, 2024
Enriching New-onset Diabetes for Pancreatic Cancer (ENDPAC): External validation using English sentinel network.
Claire A Price1,2, Hugh Claridge3,2, Simon de Lusignan4,3
1University of Surrey Faculty of Health and Medical Sciences, School of Health Sciences, Guildford, United Kingdom of Great Britain and Northern Ireland claire.a.price@surrey.ac.uk.
Summary
The Enriching New-Onset Diabetes for Pancreatic Cancer (ENDPAC) algorithm shows moderate ability to identify high-risk patients in UK primary care. This tool can help stratify risk for new-onset diabetes patients potentially indicating pancreatic cancer.
Area of Science:
- Oncology
- Endocrinology
- Primary Care Medicine
Background:
- Pancreatic cancer diagnosis is often delayed due to non-specific symptoms, leading to poor survival rates.
- Early detection in primary care is challenging, necessitating improved risk stratification tools.
- The Enriching New-Onset Diabetes for Pancreatic Cancer (ENDPAC) algorithm, developed in the USA, requires validation in other populations.
Purpose of the Study:
- To validate the ENDPAC algorithm in a UK primary care setting.
- To assess the predictive utility of ENDPAC for identifying pancreatic cancer risk in new-onset diabetes patients.
- To evaluate ENDPAC's performance using key metrics like discrimination, calibration, sensitivity, and specificity.
Main Methods:
- A retrospective cohort study was conducted using the UK's national primary care sentinel network (ORCHID).
- Adults aged 50 and above with new-onset diabetes (NOD), glycated haemoglobin (HbA1c), and weight data were included.
- ENDPAC scores were calculated, and model performance was evaluated using discrimination, calibration, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV).
Main Results:
- The study included 70,050 individuals, with 185 (0.26%) diagnosed with pancreatic cancer.
- ENDPAC demonstrated an area under the curve (AUC) of 0.733, indicating moderate discrimination.
- An optimal cutoff (≥3) identified 27.6% as high-risk, achieving 62.6% sensitivity, 72.3% specificity, a 0.6% PPV, and a 99.9% NPV.
Conclusions:
- The ENDPAC algorithm shows moderate predictive utility for pancreatic cancer risk in UK primary care new-onset diabetes patients.
- Despite a low positive predictive value, ENDPAC offers a scalable, low-cost method for automated risk stratification within sequential diagnostic pathways.
- Integration into routine primary care systems could enhance early identification of individuals with new-onset diabetes at increased risk for pancreatic cancer.