A Nomogram for Predicting the Diagnosis of Central Malignant Tumors Based on Preoperative Clinical Characteristics and Laboratory Indicators:

Overview

The purpose of this study is to develop a nomogram to predict the diagnostic probability of preoperative central lymphoma and glioma, as well as the diagnostic probability of glioblastoma and non glioblastoma in central malignant tumors. The author retrospectively analyzed patients with central lymphoma and glioma who received treatment in the neurosurgery department of Guangdong Provincial People's Hospital from 2016 to 2024. Eligible patients were randomly divided into training and validation sets in a 7:3 ratio. By integrating the least absolute shrinkage and selection operator \[LASSO\] and multivariate logistic regression analysis, the key variables for establishing a nomogram were identified. Moreover, seven models including logistic regression, decision tree, random forest, support vector machine (SVM), neural network, XGBoost, and lightGBM were used to calculate the area under the receiver operating characteristic curve (AUC)

this study aimed to develop and validate a novel diagnostic model, based on a large cohort of clinical data, for accurate differentiation between PCNSL and glioma, and GBM and non-GBM. the Ethics Review Committee of Guangdong Provincial People's Hospital granted ethical approval for this study (KY-Z-2022-231-01). This retrospective cohort study included 127 patients with newly diagnosed primary central nervous system diffuse large B-cell lymphoma (PCNSL) and 586 patients with newly diagnosed glioma, treated at the Department of Neurosurgery, Guangdong Provincial People's Hospital, between 2016 and 2024. Patients included in this study met the following criteria: (1) HIV-seronegative status; (2) no history of immunosuppression or organ transplantation; (3) availability of all clinical indicators prior to the first surgery; (4) sufficient clinical and data availability. All participants were informed of the study's purpose and provided signed informed consent prior to participation, in accordance with the principles outlined in the Declaration of Helsinki. Basic demographic and clinical data were collected prior to initial diagnosis and first surgical treatment. Demographic data included age and sex, while clinical data encompassed peripheral white blood cell count (WBC), neutrophil-to-lymphocyte ratio (NEUT1), lymphocyte ratio (LYMPH1), neutrophil count (NEUT), lymphocyte count (LYMPH), monocyte count (MONO), international normalized ratio (INR), prothrombin time (PT) and albumin. Patients were randomly assigned to training and validation cohorts in a 7:3 ratio for both diagnostic models, comprising 713 and 586 patients, respectively. No significant differences in demographic or clinical characteristics were observed between the training and validation sets. To identify potential predictors, LASSO regression analysis was performed on the training set, effectively eliminating several irrelevant or multicollinear independent variables to reduce dimensionality. Subsequently, multivariable logistic regression analysis was employed to determine the characteristic variables. Finally, a nomogram was constructed based on the training set and validated in the validation set. The predictive performance of the model was assessed and validated using machine learning and deep learning techniques. Model discrimination was evaluated using receiver operating characteristic (ROC) curve analysis. Decision curve analysis (DCA) was employed to assess the model's net benefit and clinical utility. Categorical variables were described using frequencies and percentages. Chi-squared tests or t-tests were used to compare data between the two cohorts. A LASSO regression model was applied to mitigate potential multicollinearity. All variables with complete data in the training cohort were included in the LASSO regression analysis using the "glmnet" package in R software. The model achieved optimal performance at Lambda.min, thus identifying candidate variables. Subsequently, multivariable Cox proportional hazards regression analysis was performed to identify independent predictors associated with the binary outcome. These factors were then integrated to construct a nomogram prediction model using the "rms" package in R software. Seven additional models-logistic regression, decision tree, random forest, support vector machine (SVM), neural network, XGBoost, and LightGBM-were constructed using the "caret" package, and decision curves were generated using the "rmda" package. All statistical analyses were performed using R software version 4.4.2, and a p-value \< 0.05 was considered statistically significant.