Crohn’s disease is a chronic inflammatory condition of the human gastrointestinal tract. It affects approximately one in five hundred individuals in Australia, predominantly young adults. It can cause significant morbidity from bowel obstruction, bowel perforation, perianal abscess or perianal fistula formation, or chronic bowel inflammation. A significant proportion of patients with Crohn’s disease develop irreversible bowel damage, damage which requires surgical resection of the affected bowel to relieve symptoms.
Medical therapies used to treat Crohn’s disease modify the host immune system and reduce bowel inflammation, reducing symptoms of pain and diarrhoea, and possibly reducing progression to irreversible bowel damage. Medical therapies increase the risk of opportunistic infection, the risk of developing skin cancer or lymphoma, and carry a risk of drug induced effects such as hepatitis or bone marrow suppression. These risks are likely to increase with increasing strength of immunosuppression.
Therapy in Crohn’s disease needs to appropriately weigh risks and benefits for individual patients. Accurate and objective prediction of likely outcome for patients with Crohn’s disease would aid in selection of appropriate therapy. This body of work aimed to define objective, longitudinal tools to improve prediction of outcome in Crohn’s disease.
There were two critical features which defined our approach in this work. The first was the recording of objective clinical data in a longitudinal fashion. This was achieved by designing a database with dated datafields, and endeavoring to minimize subjectivity in datapoint recording. Further objective longitudinal data were obtained through data linkage to laboratory databases. The second feature was definition of an outcome which was reversible. This feature of outcome definition meant that analysis of longitudinal information was able to occur at many timepoints in a patient’s disease course.
A poor outcome was defined as the formation of a bowel stenosis, perforation or fistula. Resolution of an outcome was defined as the passage of 2 years without further observation of the outcome. Resolution could occur following surgery, or passively with the passage of time. Perianal fistula formation was considered as an independent outcome in a separate analysis.
A consistently low albumin level < 37 g/L -1, a platelet count >370 x109/L, an MCV < 86 fL and a neutrophil count >8.6 x109 /L were identified to be associated with subsequent bowel stenosis, fistula formation or perforation. Additionally, an albumin level consistently < 38 g/L -1 or a CRP consistently > 11 mg/L-1 were associated with subsequent perianal fistula formation. This information may lead to enhanced outcome prediction in Crohn’s disease, and improved tailoring of therapy for individual patients. These findings require validation in an external cohort.