Combination of serological biomarkers and clinical features to predict mucosal healing in Crohn's disease: a multicenter cohort study

Abstract

Purpose: Mucosal healing (MH) has become the treatment goal of patients with Crohn's disease (CD). This study aims to develop a noninvasive and reliable clinical tool for individual evaluation of mucosal healing in patients with Crohn's disease.

Methods: A multicenter retrospective cohort was established. Clinical and serological variables were collected. Separate risk factors were incorporated into a binary logistic regression model. A primary model and a simple model were established, respectively. The model performance was evaluated with C-index, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and accuracy. Internal validation was performed in patients with small intestinal lesions.

Results: A total of 348 consecutive patients diagnosed with CD who underwent endoscopic examination and review after treatment from January 2010 to June 2021 were composed in the derivation cohort, and 112 patients with small intestinal lesions were included in the validation cohort. The following variables were independently associated with the MH and were subsequently included into the primary prediction model: PLR (platelet to lymphocyte ratio), CAR (C-reactive protein to albumin ratio), ESR (erythrocyte sedimentation rate), HBI (Harvey-Bradshaw Index) score and infliximab treatment. The simple model only included factors of PLR, CAR and ESR. The primary model performed better than the simple one in C-index (87.5% vs. 83.0%, p = 0.004). There was no statistical significance between these two models in sensitivity (70.43% vs. 62.61%, p = 0.467), specificity (87.12% vs. 80.69%, p = 0.448), PPV (72.97% vs. 61.54%, p = 0.292), NPV (85.65% vs. 81.39%, p = 0.614), and accuracy (81.61% vs. 74.71%, p = 0.303). The primary model had good calibration and high levels of explained variation and discrimination in validation cohort.

Conclusions: This model can be used to predict MH in post-treatment patients with CD. It can also be used as an indication of endoscopic surveillance to evaluate mucosal healing in patients with CD after treatment.

Keywords: Crohn’s disease; Endoscopic; Mucosal healing; Nomogram; PLR.

Fig. 1

ROC curve analysis of simple model (model-1) and primary model (model-2) in training group (A); ROC curve analysis of primary model in validation group (B)

Fig. 2

Nomogram for evaluation of MH rate in a given patient, constructed using as weights the coefficients derived from multivariate analysis. To calculate the probability of MH, we first obtained the value of each evaluator by drawing a vertical line straight upward from that factor to the points’ axis, then summed the points achieved for each factor and located this sum on the total points’ axis of the nomogram, where the probability of MH can be located by drawing a vertical line downward. MH, mucosal healing

Fig. 3

Calibration curves for primary model in (A) training cohort and (B) validation cohort. The x-axis represents the predicted MH while y-axis represents actual MH rate. The 45-degree dotted lines represent a perfect prediction. The solid line represents the performance of the evaluation models. The closer solid line fits to the dotted line, the better accuracy of the model shows. MH, mucosal healing