Pancreatic cancer has the lowest survival rate of all major cancer types. Less than 10% of the patients facing this disease are alive five years after the diagnosis. Limited therapeutic options and particularly tumour detection at an advanced stage due to the lack of early and specific clinical symptoms lead to this dismal prognosis. Therefore, to substantially improve life expectancy, early detection based on lab values in screenings would be helpful. Diabetes mellitus, an early manifestation of pancreatic cancer, represents a decisive clue for selecting patients at significant risk. However, given the increasing incidence of diabetes mellitus type II these days, the success of defining a population eligible for screening distinctly depends on the identification of criteria suggesting pancreatic cancer associated diabetes. Hence, our aim is to characterise pancreatic cancer patients with regard to previously recorded hyperglycaemia indicative of diabetes. We will assess in particular temporal changes in blood glucose and in glycosylated haemoglobin (HbA1c) levels as well as in body weight prior to the cancer diagnosis as potentially important factors for differentiation.
While new-onset diabetes mellitus can be a symptom of pancreatic cancer, only 1% of diabetic patients will develop the disease within 3 years after diabetes detection. Using a case-control design, we intend to characterise 30-89 years old patients with a first-time pancreatic cancer diagnosis and a matched comparison group without pancreatic cancer focusing particularly on their hyperglycaemia history. We will assess the timing of hyperglycaemia onset (</= or > 2 years prior to the cancer diagnosis) and the temporal pattern of blood glucose and HbA1c values up to 5 years prior to the cancer diagnosis or the corresponding date in the controls. We will classify blood glucose and HbA1c values into quartiles or into a priori defined categories of low, normal or high levels, and we will evaluate whether different categories are differentially associated with pancreatic cancer performing conditional logistic regression analysis. We will further compare body weight and BMI in hyperglycaemic pancreatic cancer and in control patients. Using conditional logistic regression analysis, we will assess the association between the risk of pancreatic cancer and changes in body weight by the time of cancer diagnosis.
Health Outcomes to be Measured: