Development and validation of a preoperative glycolipid metabolism-based nomogram for predicting postoperative recurrence in primary glioma: a retrospective cohort study.
Glioma recurrence after surgery remains prevalent, significantly impacting patient survival. Tumor progression is closely linked to metabolic reprogramming, especially abnormalities involving glycolipid metabolism. The triglyceride-glucose (TyG) index accurately indicates insulin resistance (IR) and metabolic disturbances. Although these metabolic indicators are prognostically valuable in various cancers, their role in forecasting glioma recurrence is still insufficiently investigated.
The medical records of 302 primary glioma patients who received surgical treatment at Linyi People's Hospital from 2016 to 2024 were retrospectively reviewed. Participants admitted to one ward (n = 236) were randomly assigned to either a training set (n = 141) or an internal validation set (n = 95). Another distinct ward provided patients (n = 66) for an independent internal validation group. In the training cohort, essential glycolipid metabolic parameters were identified via Bootstrap resampling combined with Least Absolute Shrinkage and Selection Operator (LASSO) regression, yielding a stabilized Bootstrap-LASSO Score (BSL-Score). Clinical variables alongside this score were subjected to univariate Cox regression analysis, and variables with statistical significance (P < 0.05) progressed into multivariate Cox regression to pinpoint independent prognostic indicators. Subsequently, these independent indicators were integrated into a nomogram to forecast 1-, 2-, and 3-year postoperative recurrence-free survival (RFS). Model performance was confirmed through concordance index (C-index) evaluation, time-dependent receiver operating characteristic (ROC) analyses, calibration curves, and decision curve analysis (DCA), with Bootstrap correction utilized for the C-index.
In the training cohort (n = 141), the nomogram achieved a C-index of 0.747 (95% CI: 0.676-0.818) and area under the curve (AUC) values of 0.832, 0.732, and 0.732 for 1‑, 2‑, and 3‑year RFS, respectively. In internal validation (n = 95), the C-index was 0.703 (95% CI: 0.584-0.807); in independent internal validation (n = 66), it was 0.785 (95% CI: 0.694-0.874). Calibration curves showed good agreement, and decision curve analysis confirmed clinical net benefit. The BSL‑Score, derived from routine metabolic parameters (TyG, triglyceride‑to‑high‑density lipoprotein cholesterol ratio (TG/HDL‑C), and TyG‑body mass index (TyG‑BMI)), was an independent predictor of recurrence (multivariate Cox, P < 0.05). Risk stratification by the median nomogram score significantly distinguished high‑risk from low‑risk patients (log‑rank P < 0.001).
The established nomogram effectively integrates preoperative glycolipid metabolic indicators with key clinical factors, accurately stratifying recurrence risk in postoperative glioma patients. It serves as a valuable reference for personalized postoperative monitoring, where risk-adapted surveillance and intervention strategies could optimize patient outcomes.
The medical records of 302 primary glioma patients who received surgical treatment at Linyi People's Hospital from 2016 to 2024 were retrospectively reviewed. Participants admitted to one ward (n = 236) were randomly assigned to either a training set (n = 141) or an internal validation set (n = 95). Another distinct ward provided patients (n = 66) for an independent internal validation group. In the training cohort, essential glycolipid metabolic parameters were identified via Bootstrap resampling combined with Least Absolute Shrinkage and Selection Operator (LASSO) regression, yielding a stabilized Bootstrap-LASSO Score (BSL-Score). Clinical variables alongside this score were subjected to univariate Cox regression analysis, and variables with statistical significance (P < 0.05) progressed into multivariate Cox regression to pinpoint independent prognostic indicators. Subsequently, these independent indicators were integrated into a nomogram to forecast 1-, 2-, and 3-year postoperative recurrence-free survival (RFS). Model performance was confirmed through concordance index (C-index) evaluation, time-dependent receiver operating characteristic (ROC) analyses, calibration curves, and decision curve analysis (DCA), with Bootstrap correction utilized for the C-index.
In the training cohort (n = 141), the nomogram achieved a C-index of 0.747 (95% CI: 0.676-0.818) and area under the curve (AUC) values of 0.832, 0.732, and 0.732 for 1‑, 2‑, and 3‑year RFS, respectively. In internal validation (n = 95), the C-index was 0.703 (95% CI: 0.584-0.807); in independent internal validation (n = 66), it was 0.785 (95% CI: 0.694-0.874). Calibration curves showed good agreement, and decision curve analysis confirmed clinical net benefit. The BSL‑Score, derived from routine metabolic parameters (TyG, triglyceride‑to‑high‑density lipoprotein cholesterol ratio (TG/HDL‑C), and TyG‑body mass index (TyG‑BMI)), was an independent predictor of recurrence (multivariate Cox, P < 0.05). Risk stratification by the median nomogram score significantly distinguished high‑risk from low‑risk patients (log‑rank P < 0.001).
The established nomogram effectively integrates preoperative glycolipid metabolic indicators with key clinical factors, accurately stratifying recurrence risk in postoperative glioma patients. It serves as a valuable reference for personalized postoperative monitoring, where risk-adapted surveillance and intervention strategies could optimize patient outcomes.