Abstract
Metachronous brain metastasis is a clinically significant pattern of recurrence after curative-intent surgery for non-small cell lung cancer (NSCLC). The extent to which contemporary clinicopathological and molecular characteristics predict postoperative intracranial relapse, and whether routine postoperative neuroimaging might aid early detection, remains uncertain. We performed a systematic review and meta-analysis of studies reporting postoperative brain metastasis following curative resection for NSCLC. Fourteen retrospective cohort studies published between 2010 and 2025 were included, comprising 13,414 patients and 883 documented intracranial recurrence events. Data extraction focused on clinicopathological factors, molecular features, follow-up duration, and multivariable-adjusted hazard ratios (HRs). Only HRs derived from multivariable Cox regression models were pooled; studies reporting odds ratios without time-to-event adjustment were not combined quantitatively, ensuring comparability of effect estimates and minimizing confounding. The overall incidence of metachronous brain metastasis was approximately 6.6%. Routine postoperative brain imaging was implemented in only 28.6% of studies, whereas the majority relied on symptom-triggered neuroimaging. Adenocarcinoma was the predominant histology, accounting for 45-100% of cases. Although EGFR mutation was variably associated with increased intracranial recurrence, the pooled estimate demonstrated substantial heterogeneity (I² = 86.1%), limiting interpretability. In contrast, traditional clinicopathological factors-including nodal involvement, visceral pleural invasion, tumor differentiation, smoking history, and chronic pulmonary comorbidities-showed more consistent associations across multivariable models with low to moderate heterogeneity. Considerable heterogeneity in imaging strategies and molecular reporting was observed. Metachronous brain metastasis remains an important postoperative recurrence pattern in surgically treated NSCLC. Clinicopathological and host-related factors appear to be more consistently predictive than molecular features alone. Whether selected high-risk patients may benefit from scheduled rather than purely symptom-driven postoperative brain imaging warrants further prospective investigation. Standardized surveillance protocols and integration of clinical, pathological, and molecular determinants are essential to improve risk stratification and to facilitate earlier detection of intracranial relapse.