Upper gastrointestinal cancers (including cancers of the oesophagus, stomach, pancreas and liver) have poor survival, and therefore preventing their occurrence is vitally important. It is plausible that some medications could have unintended consequences and increase upper gastrointestinal cancer risk.
We previously systematically investigated 250 prescription medications in relation to upper gastrointestinal cancer risk using General Practice data from Scotland. We identified four medications which increase upper gastrointestinal cancer risk and for which a plausible biological reason could be found. However systematically investigating large numbers of medications in this way can lead to the identification of medications by chance and which are not truly associated with cancer risk. Therefore, replicating the observed associations in a different study population is necessary.
The aim of this study is to determine the associations between these four medications and upper gastrointestinal cancer using the Clinical Practice Research Datalink dataset to determine whether these associations still hold. If so, current use of such medications may need to be reconsidered.
Background: The poor survival of upper gastrointestinal cancers (including cancers of the oesophagus, stomach, pancreas and liver) highlight the need for their prevention. In a screening study we systematically examined the association between 250 prescription medications and upper gastrointestinal cancer risk in the Scottish Primary Care Clinical Informatics Unit Research (PCCIUR) database. We identified four medications which were associated with increased upper gastrointestinal cancer risk and for which there were plausible clinical mechanisms.
Aim: To determine whether associations identified between four medications and upper gastrointestinal cancer in a screening study are apparent in the Clinical Practice Research Datalink (CPRD).
Methods: A nested case-control study will be conducted, estimating the association between each medication and upper gastrointestinal cancer. In the primary analysis cases with cancer will be determined from National Cancer Registration and Analysis Service [NCRAS] data on cancer registrations. Controls will be matched to cases using age, gender, year of registration and GP practice. The index date within each matched set will be the date of the first diagnosis of a primary cancer in the case. Prescription usage will be determined prior to the index date for both cases and controls. Conditional logistic regression will be used to calculate odds ratios (ORs) and 95% confidence intervals (95%CIs) comparing the risk of cancer among users with non-users adjusting for comorbidity and lifestyle confounders.
Potential: The study could determine whether associations between medications and upper gastrointestinal cancer identified in PCCIUR are also found in CPRD. If so, these medications may require more detailed study in relation to any potential causal relationship. Medications which increase cancer risk may require revision of licensing and usage.
Diagnosis for upper gastrointestinal cancer (oesophagus, stomach, liver, pancreas) will be determined using NCRAS cancer registration data.
Chris Cardwell - Chief Investigator - Queen's University Belfast
Chris Cardwell - Corresponding Applicant - Queen's University Belfast
Carmel Hughes - Collaborator - Queen's University Belfast
Peter Murchie - Collaborator - University of Aberdeen
Ronald McDowell - Collaborator - Queen's University Belfast
NCRAS Cancer Registration Data;Patient Level Index of Multiple Deprivation