Perioperative risk stratification after resection of brain metastases: internal development and validation of the dominant lesion surgery score in a 20-year single-center cohort

To examine whether intracranial lesion count remains associated with survival after resection of brain metastases and to develop an internally validated perioperative risk-stratification model for early mortality in a surgically treated cohort. We conducted a retrospective single-center cohort study of adults who underwent surgical resection for histologically confirmed brain metastases between 2002 and 2024. Overall survival was analyzed using Kaplan–Meier methods and multivariable Cox regression. The Dominant Lesion Surgery Score (DLSS) was derived from variables available during the perioperative period that were associated with early mortality and was evaluated for 6- and 12-month mortality using receiver operating characteristic analysis. Internal validation was performed with bootstrap resampling, and model performance was further assessed using calibration and decision-curve analysis. Among 189 surgically treated patients, lesion count was not the strongest variable associated with survival after multivariable adjustment, whereas extent of resection and histology-defined tumor subtype showed stronger associations within this selected cohort. DLSS demonstrated moderate discrimination for 6- and 12-month mortality, with stable optimism-corrected performance after bootstrap validation. Calibration analysis showed acceptable agreement between predicted and observed mortality. Decision-curve analysis suggested potential net benefit across clinically relevant threshold probabilities. A DLSS cutoff of ≥ 2 identified a subgroup with higher 12-month mortality. Among surgically treated patients with brain metastases, lesion count alone may be insufficient to characterize postoperative risk. DLSS should be regarded as an exploratory internally derived model for perioperative risk stratification rather than a standalone tool for treatment selection, and external validation is required before broader clinical use.