Identification of symptoms associated with the diagnosis of pancreatic exocrine and neuroendocrine neoplasms: a nested case-control study of the UK population


Session type:

Weiqi Liao1, Ashley Clift1, Martina Patone1, Carol Coupland2, Stephen P Pereira3, Julia Hippisley-Cox1
1University of Oxford, 2University of Nottingham, 3University College London (UCL)



Pancreatic cancer has the worst survival among all cancers, and most patients are diagnosed at a late stage in the UK.  This study aimed to investigate the symptoms associated with the diagnoses of pancreatic ductal adenocarcinoma (PDAC) and neuroendocrine neoplasms (PNEN), and comparatively characterise the symptomatology between the two tumours to inform earlier diagnosis in primary care.


We designed a nested case-control study and used data from the QResearch database. Patients aged ≥25 years and diagnosed with PDAC or PNEN during 2000-2017 were the cases. Up to 10 patients without PDAC and PNEN (controls) from the same general practice were matched with each case by age, sex, and calendar year using incidence density sampling. Multiple imputation with chained equations was used to impute missing values (ten imputations were conducted). Multivariable conditional logistic regression was used to investigate the association between the 42 shortlisted symptoms and the diagnoses of PDAC and PNEN in seven different timeframes (from <3 months to 5 years) relative to the index date (date of cancer diagnosis for the case), adjusting for patients’ sociodemographic characteristics, lifestyle, and relevant comorbidities.


Through data linkage, 23,640 patients diagnosed with PDAC and 596 with PNEN were identified, which made it the largest population-based study of this kind. Twenty-three symptoms were significantly associated with PDAC, and nine symptoms with PNEN (OR>1.2 and P<0.01). PDAC and PNEN had overlapping symptom profiles, although the odds ratios for the significant symptoms were different. Jaundice and gastrointestinal bleeding were the two alarm symptoms for both tumours. Thirst and dark urine were the two new symptoms identified for PDAC. The risk of unintentional weight loss may be longer than two years before the diagnosis of PNEN.


Compared with the current QCancer (Pancreas) model, thirteen additional symptoms were identified. The new findings enable us to conduct further work on updating the QCancer prediction models for PDAC and PNEN, which will help GPs to make clinical decisions, especially when patients present with multiple non-specific symptoms.

Impact statement

A comprehensive symptom profile for PDAC and PNEN is gained from a large representative UK primary care population.