Abstract Background Surgical resection is controversial in treating primary central nervous system lymphoma (PCNSL). This study aimed to elucidate the prognostic role of resection based on treatment- and survival-related tumor localizations. Methods Magnetic resonance imaging (MRI) data of 122 untreated PCNSL patients from January 2007 to October 2021 were analyzed. Images were registered to a standard anatomical template, followed by lesion segmentation. Analysis of differential involvement (ADIFFI) identified statistically significant voxel-wise clusters using Pvalue heatmaps. Overall survival (OS) and progression-free survival (PFS) were evaluated via Kaplan-Meier analyses. Multivariate Cox regression was performed to identify independent prognostic factors. Results Favorable but non-significant trends for OS and PFS were detected in patients receiving resection compared to biopsy. ADIFFI revealed that clusters in superficial frontal and parietal areas were more likely to receive resection, whereas deep-seated clusters were more inclined to be biopsied. Superficial localization also predominated in patients with OS 2 years, suggesting the spatial confounding effect. Crucially, by controlling for the dual confounders of tumor location and age, significantly prognostic advantages for resection over biopsy were identified in patients with superficial PCNSLs aged ≤ 70 years. Further Cox regression in this subgroup demonstrated that surgical resection, particularly gross total resection, emerged as an independent protective factor for OS. Conclusions The historically perceived limited benefit of resection in PCNSL may be confounded by tumor localization and patient age. For selected patients with superficial lesions aged ≤ 70 years, resection appears to provide survival advantages, offering a reference to refine surgical decision-making.
Dou et al. (Wed,) studied this question.