Monitoring for 5-aminosalicylate toxicity: prognostic model development and validation

Abstract

Background and aimTo develop and validate a prognostic model for risk-stratified monitoring of 5-aminosalicylate (5-ASA) toxicity.MethodsThis nationwide retrospective cohort study used data from the Clinical Practice Research Datalink (CPRD) Aurum and Gold for model development and validation respectively. It included adults newly diagnosed with inflammatory bowel disease (IBD) and established on 5-ASAs between 01/01/2007 and 31/12/2019. 5-ASA discontinuation with abnormal monitoring blood test result was the outcome of interest. Patients prescribed 5-ASAs for ≥6 months i.e., established on treatment, were followed-up for up to five years. Penalised Cox-regression was used to develop the risk equation. Model performance was assessed in terms of calibration and discrimination. Statistical analysis was performed using STATA (StataCorp LLC).Results14,109 and 7,523 participants formed the development and validation cohorts with 401 and 243 events respectively. 185, 172, and 64 discontinuations were due to cytopenia, elevated creatinine and elevated liver enzymes respectively in the derivation cohort. Hazardous alcohol intake, chronic kidney disease, thiopurine use, and blood test abnormalities before follow-up were strong prognostic factors. The optimism adjusted R2Din development data was 0.08. The calibration slope and Royston D statistic (95% Confidence Interval) in validation cohort were 0.90 (0.61-1.19) and 0.57 (0.37-0.77) respectively.ConclusionThis prognostic model utilises information available during routine clinical care and can be used to inform decisions on the interval between monitoring blood-tests. The results of this study ought to be considered by guideline writing groups to risk-stratify blood test monitoring during established 5-ASA treatment.What is already known?Renal, hepatic and blood toxicity are uncommon during long-term 5-aminosalicylate (5-ASA) treatment.There are no mechanisms to predict the risk of these toxicities during established treatment that may be used to risk stratify blood-test monitoring.What this study adds?Using a large national dataset originated during routine care, this study developed a prognostic model that discriminated patients at varying risk of 5-ASA toxicity during established treatment with good performance characteristics validated.Most patients were at low-risk of toxicity due to 5-ASAs and could continue with annual monitoring blood-tests while others at high risk may require more frequent monitoring.This prognostic model can be used to make an informed decision on the interval between monitoring blood tests and the findings ought to be considered by guideline writing groups to bring about equitable and sustainable change in clinical practice.