Prognostic biomarkers to identify patients destined to develop severe Crohn's disease who may benefit from early biological therapy: Protocol for a systematic review, meta-analysis and external validation

Abstract

Background: It is believed increasingly that patients with severe Crohn's disease are best treated early with biological therapy, which may ameliorate subsequent disease course and diminish long-term complications. However, we cannot predict currently which new presentations of Crohn's disease are destined to develop severe disease so treatment cannot be targeted to the most appropriate patients. Accordingly, via systematic review and meta-analysis we aim to identify if biomarkers of disease activity are able to predict development of severe disease. Methods/design: We will search the primary literature and conference proceedings for studies of biomarkers of all types including clinical, endoscopic, radiological, faecal, urinary, serological, genetic, and histological. Precise definition of "severe" disease is elusive so we will include sensitivity analysis to account for different definitions. We will use the CHARMS checklist to frame our question and to extract data. We will extract the study design, setting, participant characteristics, biomarker(s) investigated, and study outcomes. Bias will be assessed via the PROBAST tool. We will present the results using narrative and graphical methods. We will present the summary by meta-analysis where there are sufficient studies with reasonable homogeneity, using methods appropriate to the type of data extracted. Heterogeneity will be presented via Forest and ROC plots. Discussion: If this systematic review and meta-analysis identifies biomarkers that appear sufficiently predictive for subsequent severe disease course, we aim to combine them in a predictive model, followed by external validation using individual patient data. A predictive model able to identify new presentations of Crohn's disease destined to develop severe disease subsequently would have considerable clinical utility for patient management. Systematic review registration: PROSPERO CRD42016029363.