A new model for the discrimination between ulcerative colitis and Crohn's disease

Abstract

Distinguishing ulcerative colitis (UC) from Crohn's disease (CD) is sometimes difficult in a clinical setting. The purpose of this study was to identify a series of independent serum markers capable of distinguishing between UC and CD. 140 UC and 174 CD patients hospitalized at The First Affiliated Hospital, College of Medicine, Zhejiang University were recruited into this study. A panel of serum markers was quantified for each patient and the Bayesian information criterion (BIC) was used to determine a discrimination model. The receiver operating characteristic (ROC) was used to evaluate the performance of the model, and the area under the ROC curve (AUC) was used to evaluate the accuracy of the model. Serum albumin (Alb), total cholesterol (TC), total calcium (TCa), platelet (Plt), glycyl proline dipeptidyl aminopeptidase (GPDA) and their ratios (Alb: Plt, Alb: GPDA, TCa: TC, and Plt: GPDA) were selected into the diagnosis model using BIC. The resulting CD/UC Index (CUI) is CUI = 1.901 + 0.425 Alb - 3.324 TC - 7.444 TCa + 0.018 Plt + 0.087 GPDA - 0.0007 Alb: Plt - 0.004 Alb: GPDA + 1.839 TC: TCa + 0.003 Plt: GPDA, with CUI > 0 incrementally favored a diagnosis of UC, while CUI < 0 corresponded to a higher likelihood of a diagnosis of CD. An average value of the AUC for the CUI model is 0.73 (95% confidence interval: 0.67-0.80). The CUI, derived from commonly available serum biomarkers, could try to differentiate UC from CD in patients with unclear clinical features as a new approach to diagnosis.

Keywords: Crohn’s disease; Ulcerative colitis; diagnostic model; serum biomarkers.

Figure 1

Estimated density functions of CUI for CD and UC patients.

Figure 2

Average ROC curves for null model (without GPDA) and alternative model (with GPDA) based on 100 sample splitting.

Figure 3

Component + Residual plots for working predictors and interactions in the alternative model.