Prognostic modeling for diffuse midline glioma: development and validation of a risk stratification nomogram using...

Diffuse midline glioma (DMG) is a rare and highly aggressive central nervous system tumor with limited treatment options and poor survival outcomes. Reliable prognostic models are urgently needed to guide risk-adapted therapy.

We retrospectively analyzed 409 DMG patients from the SEER database (2018-2021). Independent prognostic factors were identified using multivariate Cox regression analysis. A nomogram was developed to estimate overall survival, and its performance was evaluated using the concordance index (C-index), time-dependent ROC curves, calibration plots. Risk stratification was based on nomogram total scores. Subgroup survival comparisons were conducted using Kaplan-Meier and log-rank tests. External validation was performed using an independent institutional cohort of 22 patients.

An age-dependent anatomical distribution was observed: brainstem tumors predominated in children, while non-brainstem tumors were more common in adults. Multivariate Cox regression identified older age, higher household income, and cerebellar location as independent prognostic factors. These variables were incorporated into a nomogram that demonstrated good discriminative ability and calibration. Based on total risk scores, patients were stratified into high- and low-risk groups with significantly different survival outcomes. Combined chemoradiotherapy significantly improved survival compared to radiotherapy or chemotherapy alone, while chemotherapy alone showed no added benefit. Surgical resection extent was not associated with prognosis. In an external validation cohort of 22 patients, survival was better in the low-risk group than in the high-risk group, although the difference was not statistically significant (P = 0.188).

This study presents the first large-scale, SEER-based nomogram for DMG, offering reliable prognostic stratification and reinforcing the survival benefit of combined chemoradiotherapy. The model's clinical utility is further supported by real-world institutional validation, underscoring its potential to inform individualized treatment strategies in DMG.