Oral Presentation The Pancreas Summit 2025

Predicting Pancreatic Cancer Risk in People With New-Onset Diabetes Mellitus (125886)

Sitwat Ali 1 2 , Mary Waterhouse 1 , Arrisonia Doubatty 1 , Andrew Metz 3 , Benedict Devereaux 4 5 , Daniel Croagh 6 , Joel Rhee 7 , John Windsor 8 , John Zalcberg 9 10 , Karen Canfell 11 , Michael Caruana 12 , Paul Grogan 13 , Louisa Collins 1 2 14 , Susan Jordan 2 , Rachel Neale 1 2
  1. Population Health Department, QIMR Berghofer, Brisbane, Queensland, Australia
  2. School of Public Health, The University of Queensland, Brisbane, Queensland, Australia
  3. Jreissati Pancreatic Centre, Epworth Medical Foundation, Melbourne, Victoria, Australia
  4. Department of Gastroenterology and Hepatology, Royal Brisbane and Women's Hospital, Brisbane, Queensland, Australia
  5. Faculty of Health, Medicine and Behavioural Sciences, University of Queensland, Brisbane, Queensland, Australia
  6. Department of Upper GI and Hepatobiliary Surgery, Monash Medical Centre, Melbourne, Victoria, Australia
  7. Discipline of General Practice, School of Clinical Medicine, Faculty of Medicine and Health, University of New South Wales, Sydney, New South Wales, Australia
  8. Surgical and Translational Research Centre, University of Auckland, Auckland, New Zealand
  9. Department of Medical Oncology, Alfred Health, Melbourne, Victoria, Australia
  10. Cancer Research Program, Monash School of Public Health,, Monash University, Melbourne, Victoria, Australia
  11. School of Public Health, Faculty of Medicine and Health, The University of Sydney, Sydney, NSW, Australia
  12. The Daffodil Centre, The University of Sydney, a Joint Venture with Cancer Council New South Wales, Sydney, Australia
  13. School of Public Health, Faculty of Medicine and Health, The University of Sydney, Sydney, NSW, Australia
  14. Cancer Prevention and Survivorship, Cancer Council Queensland, Brisbane, Queensland, Australia

Background and aim

Pancreatic cancer is associated with poor prognosis. Surveillance in individuals at a higher risk of pancreatic cancer could result in earlier diagnosis and people with new-onset diabetes may represent a potential target group. However, differentiating pancreatic cancer–related diabetes from typical type 2 diabetes remains a challenge. This study aimed to develop and validate a prediction model for identifying pancreatic ductal adenocarcinoma (PDAC) in individuals with new-onset diabetes.

Methods

We used national administrative health databases linked by the Australian Institute of Health and Welfare (the Panlink database). We conducted a retrospective cohort study among people newly diagnosed with diabetes, identified through use of diabetes medications, or use of diabetes services primarily administered in primary care settings, or hospital records. We used a Cox proportional hazards model to predict the risk of PDAC within one year of diabetes diagnosis, using a stepwise approach that combined forward selection and backward elimination. We used age at diabetes diagnosis, sex, prescription medications, some healthcare services item codes, and an indicator of ‘severity’ of diabetes as potential predictors.

Results

Among 905,402 people aged 50 years and older with newly diagnosed diabetes, 1,803 (0.2%) were diagnosed with PDAC within one year of follow-up. People with PDAC were older at the time of diabetes diagnosis (72.6 vs 65.9 years) and more likely to be men (56% vs 44%). The one-year risk prediction model showed good discrimination and calibration (c-index 0.78). Key predictors included older age, poor response to initial diabetes treatment, use of acid disorder medications, pancreatic enzyme supplements, and medications used in people with metabolic disorders (which were protective).

Conclusion

Our model identified older age and indicators of more severe diabetes as significant predictors of pancreatic cancer. Pancreatic cancer occurred more commonly in individuals who were not taking medications used to treat metabolic disorder but who presented with symptoms suggestive of pancreatic cancer. This model may serve as a useful initial screening tool to help identify people with newly diagnosed diabetes who could benefit from further investigations such as abdominal imaging.