JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

A nomogram for predicting the diagnosis of central malignant tumors based on preoperative clinical characteristics and laboratory indicators: a diagnostic study

  • 0Guangdong Cardiovascular Institute, Guangdong Provincial People's Hospital, Guangdong Academy of Medical Sciences, Guangzhou, China.
International Journal of Surgery (london, England) +

|

July 28, 2025

|

July 28, 2025

View abstract on PubMed

Summary

This summary is machine-generated.

This study developed predictive models for central nervous system lymphoma and glioma, and glioblastoma. These tools aid clinicians in diagnosing brain tumors and personalizing treatment strategies.

Area Of Science

  • Neuro-oncology
  • Medical imaging analysis
  • Machine learning in medicine

Background

  • Accurate preoperative diagnosis of central nervous system (CNS) tumors is crucial for effective treatment planning.
  • Differentiating between primary CNS lymphoma (PCNSL) and glioma, and between glioblastoma and non-glioblastoma, presents a significant clinical challenge.

Purpose Of The Study

  • To develop and validate a nomogram for predicting the preoperative diagnostic probabilities of PCNSL versus glioma.
  • To develop and validate a nomogram for predicting the preoperative diagnostic probabilities of glioblastoma versus non-glioblastoma.

Main Methods

  • Retrospective analysis of 712 patients with PCNSL or glioma.
  • Development of diagnostic models using LASSO and multivariate logistic regression.
  • Validation using internal training/validation sets and external prospective cohorts from multiple centers.
  • Performance evaluation with seven machine learning models, AUC, and decision curve analysis.

Main Results

  • The PCNSL versus glioma model included age, KPS, NEUT, NEUT1, and MONO, achieving optimal AUCs of 0.877 (internal) and ranging from 0.716 to 0.743 (external).
  • The glioblastoma versus non-glioblastoma model included age, NEUT1, and MONO, achieving optimal AUCs of 0.861 (internal) and ranging from 0.710 to 0.842 (external).
  • Seven machine learning models demonstrated robust performance, with AUCs ranging from 0.778 to 0.889.

Conclusions

  • Validated diagnostic probability models for differentiating CNS lymphoma from glioma and glioblastoma from non-glioblastoma.
  • These models can assist clinicians in improving preoperative diagnostic accuracy for brain tumors.
  • Facilitates personalized treatment strategies for patients with central malignant tumors.

Keywords: