P292 Development of a predictive model for the diagnosis of Crohn’s disease when ileocolonoscopy is insufficient: a multicentre study
P. Boal Carvalho*1, M. Marques2, B. Rosa1, H. Cardoso2, F. Machado3, G. Macedo2, J. Cotter1, 4, 5
1Hospital Senhora da Oliveira, Gastroenterology, Guimarães, Portugal, 2Hospital de São João, Gastroenterology, Porto, Portugal, 3ACES Porto Ocidental, Public Health Unit, Porto, Portugal, 4University of Minho, Life and Health Research Institute, Braga/Guimarães, Portugal, 5PT Government Associate Laboratory, ICVS/3B’s, Braga/Guimarães, Portugal
The diagnosis of Crohn’s disease (CD) usually requires not only a strong clinical suspicion but also a combination of laboratorial markers and endoscopic procedures. Small bowel capsule endoscopy (SBCE) is a useful tool for establishing a diagnosis of CD, particularly in the setting of an inconclusive ileocolonoscopy. We aimed to develop a predictive model for the diagnosis of CD using SBCE, sociodemographic and analytical variables.
Multicentric retrospective study including patients with suspected CD and a negative or inconclusive ileocolonoscopy, who subsequently underwent SBCE. Patients’ clinical and demographic characteristics were assessed, and a blood panel was obtained. Inflammatory activity in SBCE was evaluated through the Lewis score (LS), calculated from the presence and severity of villous oedema, small bowel ulcers, and stenosis.
Patients were followed during 12 months, and the primary outcome was the confirmation of the diagnosis of CD, based on the cumulative evidence of all available follow-up data, including clinical assessment and further diagnostic procedures when applicable, as determined by the assisting physician. Statistical analysis was performed using SPSS v21.0, and a p-value < 0.05 was considered statistically significant. Multivariate analysis was performed, and a logistic regression model was developed for the prediction of CD diagnosis during follow-up.
Included 137 patients, 60.6% (n = 83) women, with mean age 43 (range 18–66) years. Small bowel inflammatory activity was observed in 40.6% of the patients (n = 56), mean LS 371 (range 0–6 400), inflammatory activity was highest in the third tertile (mean LS 243 and range 0–6400), and inflammatory activity throughout the entire small bowel was present in 11% (n = 15) of the patients. Diagnosis of CD during follow-up was established in 53 patients (39.7%).
In the multivariate analysis, thrombocytosis (80.0% vs 36.9%, p = 0.045), lower mean albumin levels (4.1 vs 5.2 mg/dL, p = 0,004), elevated c-reactive protein (54.4% vs 27.9%, p = 0.012), and a higher mean LS (819 vs 61, p = 0.029) were independently associated with a subsequent diagnosis of CD. A predictive logistic regression model using these 4 variables demonstrated an excellent discriminative power (AUC = 0.82, p < 0,001) for the diagnosis of DC.
In this multicentric study, the presence of thrombocytosis (p = 0.045), elevated C-reactive protein (p = 0.012), lower levels of serum albumin (p = 0.004), and higher inflammatory activity were assessed during SBCE using the LS (p = 0,029) were associated with the diagnosis of DC during follow-up. A regression model including these variables accurately predicted the diagnostic outcome in 82% of the patients.