A nomogram for predicting 3-year total weight loss percentage following LSG: insights from visceral adipose tissue inflammatory methylation sites

Abstract

Background: Obesity is a chronic low-grade inflammatory condition. Laparoscopic sleeve gastrectomy (LSG) is a widely recognized intervention for weight management; however, the percentage of total weight loss (%TWL) achieved varies significantly among patients.

Objective: This study aims to develop a nomogram based on methylation sites associated with the inflammatory (INF) in intraoperative visceral adipose tissue (VAT) to predict %TWL at three years post-LSG.

Methods: Patients undergoing LSG were categorized into two groups based on their%TWL three years post-LSG: satisfactory (%TWL ≥25) and unsatisfactory (%TWL<25). Comparative analyses of 850K methylation microarrays from VAT samples were performed to identify methylation sites associated with INF-related genes. Differentially methylated sites were analyzed using least absolute shrinkage and selection operator, random forest, and support vector machine with recursive feature elimination analyses to identify key predictive methylation sites. A nomogram was subsequently developed using these hub methylation sites. The model's performance was assessed through receiver operating characteristic (ROC) curve analysis with bootstrap resampling, calibration curves, decision curve analysis (DCA), and clinical impact curves (CIC).

Results: Among 25 patients (11 satisfactory and 14 unsatisfactory), 151 differential INF-related methylation sites were identified. Two hub methylation sites, cg14027957 and cg20666492, were selected as predictors for the nomogram. Internal validation demonstrated excellent predictive performance, with an area under the curve (AUC) of 96.8%. The model also showed strong calibration and clinical utility.

Conclusion: The nomogram, based on two hub methylation sites, effectively predicts%TWL outcomes three years post-LSG. Its high predictive accuracy and clinical relevance suggest significant potential for guiding personalized treatment strategies in patients undergoing LSG.

Keywords: Inflammatory response methylation sites; Laparoscopic sleeve gastrectomy; Nomogram; Percentage of total weight loss.

Fig. 1

Design and workflow of the study

Fig. 2

Identification of hub methylation sites A volcano plot depicting 151 differentially methylated β-values of INF-related methylation sites between the satisfactory group and unsatisfactory groups. B LASSO regression analysis results. C SVM-RFE technique algorithm output. D RF algorithm output. E Venn plot showing the overlap of reliable biomarkers identified by LASSO, SVM-RFE, and RF INF, inflammatory; LASSO, least absolute shrinkage and selection operator; SVM-RFE, support vector machine recursive feature elimination; RF, random forest

Fig. 3

Validation of the Hub Methylation Sites A ROC curves assessing the diagnostic ability of cg14027957 and cg20666492. B Box plots showing differential expression analysis of cg14027957. C Box plots showing differential expression analysis of cg20666492 ROC, receiver operating characteristic

Fig. 4

Construction and Validation of the Nomogram A Nomogram for predicting long-term outcomes of LSG. B ROC curves evaluating the discrimination ability, with internal validation using bootstrap resampling (1,000 iterations). C Calibration curve. D DCA assessing the clinical utility of the diagnostic model. E CIC evaluating the clinical impact of the diagnostic model ROC, receiver operating characteristic; DCA, decision curve analysis; CIC, clinical impact curve