A Multi-marker Model Based on Routine Laboratory Parameters for Predicting Intestinal Fibrosis in Crohn's Disease

Abstract

Background: This study aimed to evaluate the diagnostic value of complete blood count, biochemical, and coagulation parameters in identifying intestinal fibrosis in patients with Crohn’s disease. Methods: A retrospective study was conducted, including 280 patients with intestinal fibrosis and strictures due to Crohn's disease from the Sixth Affiliated Hospital of Sun Yat-sen University as the case group, and 104 patients with benign intestinal diseases as the control group. ROC curve analysis was performed to evaluate the performance of each individual laboratory parameter. Finally, a multi-parameter combination with a high area under the receiver operating characteristic curve (AUC), high sensitivity, and high specificity was selected to assess its predictive diagnostic performance for intestinal fibrosis and strictures in Crohn's disease. Results: Among the 35 laboratory parameters evaluated, eight highly predictive biomarkers were identified-each exhibiting an AUC > 0.70 and a statistically significant P-value < 0.0001. Fasting glucose (GLU) demonstrated the strongest discriminative performance (AUC = 0.7683). Predictive models integrating multiple biomarkers significantly outperformed single-biomarker models in diagnostic accuracy. A total of 226 combinatorial models were constructed; the optimal 8-biomarker model achieved an AUC of 0.9197. Following systematic elimination of redundant features, a parsimonious 5-biomarker model-comprising fasting glucose (GLU), mean corpuscular volume (MCV), platelet distribution width-standard deviation (PDW-SD), activated partial thromboplastin time (APTT), and lymphocyte count (LYM)-yielded an AUC of 0.9169. This value was not significantly different from that of the 8-biomarker model (ΔAUC = 0.0028; P > 0.05), indicating comparable diagnostic accuracy and strong translational potential for clinical implementation. Conclusion: This study assessed the predictive value of 35 routine complete blood count, biochemical, and coagulation parameters for intestinal fibrosis in Crohn’s disease, identified eight parameters with robust diagnostic performance, and developed a multi-parameter predictive model. A five-marker combination-GLU, MCV, PDW-SD, APTT, and LYM-was ultimately determined to be the optimal non-invasive predictive tool. Future studies may integrate novel molecular biomarkers associated with fibrogenesis to further enhance prediction accuracy.