Abstract 9O
Background
Delayed diagnosis contributes to poor survival in pancreatic cancer (PC). Early detection may improve outcomes, but symptom-based diagnostic approaches are challenging, because most symptoms are non-specific and more likely to be associated with a non-cancer cause. Approximately forty percent of people diagnosed with PC have diabetes at the time of PC diagnosis. Few PC prediction models exist for this population, with limited performance and high risk of bias. We aim to refine performance in predicting PC in individuals diagnosed with type 2 diabetes by identifying and weighting commonly used routine blood markers, which could potentially enhance identification of individuals at high risk of PC in primary care with a simple blood test.
Methods
We identified individuals from the AMORIS cohort study with newly diagnosed diabetes based on the Swedish National Diabetes Registry and patient registry. Individuals aged 50 or older with type 2 diabetes were included. All participants presented a routine biomarker panel measured after diagnosis of diabetes, including liver enzymes, glucose and lipids profile, markers of inflammation and sugar metabolism. Possible confounding factors such as age, sex, and educational level were considered. Initial multivariate Cox proportional hazards analysis was conducted.
Results
A total population of 101,051 individuals with newly diagnosed diabetes were identified. 1390 PC events occurred during the follow up time (meanyears=25.38). Preliminary results identified increased risk of PC with higher total cholesterol (mmol/L) (HR (95%CI): 1.11 (1.03-1.19)) and Alkaline Phosphatase (ALP) (μkat/L) (HR (95%CI): 1.10 (1.05-1.16)) and lower albumin (g/L) (HR (95%CI): 0.95 (0.92-0.96)). Table: 9O
Multivariate biomarker panel continuous
Variable | HR | CI (95%) | p-value | |
Albumin (g/L) | 0.946 | 0.916 | 0.976 | 0.0006 |
Alkaline phosphatase (ALP) (μkat/L) | 1.102 | 1.045 | 1.161 | 0.0003 |
fS-glucose (mmol/L) | 1.048 | 0.995 | 1.104 | 0.0758 |
Cholesterol (mmol/L) | 1.106 | 1.025 | 1.194 | 0.0098 |
Conclusions
Routine blood markers could be a non-invasive cost-effective avenue to identify diabetic patients at higher risk of PC and improve early detection. Work is ongoing to evaluate lag time, discrimination analysis and prediction scores.
Editorial acknowledgement
Clinical trial identification
Legal entity responsible for the study
The authors.
Funding
Cancer Research UK and Pancreatic Cancer Research.
Disclosure
All authors have declared no conflicts of interest.
Resources from the same session
6O - Insights into tumour dissemination and progression using longitudinal imaging and ctDNA in the TRACERx lung cancer study
Presenter: Wing Kin Liu
Session: Proffered Paper session 2
Resources:
Abstract
7O - Molecular drivers and therapeutic targets for neuroendocrine transformation in lung cancer
Presenter: Triparna Sen
Session: Proffered Paper session 2
Resources:
Abstract
8O - Clinical implications of clonal hematopoiesis in the tumor microenvironment of non-small cell lung cancer
Presenter: Oriol Pich
Session: Proffered Paper session 2
Resources:
Abstract
Invited discussant
Presenter: Gerard Zalcman
Session: Proffered Paper session 2
Resources:
Slides
Webcast
Invited discussant
Presenter: Alice Boileve
Session: Proffered Paper session 2
Resources:
Slides
Webcast
Q&A
Presenter: Mariam Jamal-Hanjani
Session: Proffered Paper session 2
Resources:
Webcast