Clinicoradiological Features, Surgical Management, and Prognostic Factors in Expansive Brainstem Lesions: A 16-Year Comparative Analysis of Intrinsic versus Infiltrating Pathologies at a Tertiary Referral Center

BACKGROUND: Expansive brainstem lesions (BSLs) present significant diagnostic and management challenges. This study evaluates the clinical landscape and prognostic factors of BSLs. PURPOSE: To report our institutional experience with expansive brainstem lesions in Tunisia, describing their epidemiological, clinical, radiological, therapeutic, and prognostic features, with a comparative analysis of intrinsic versus infiltrating lesions. PATIENTS AND METHODS: Retrospective study of 48 patients treated over 16 years (2000-2015). Univariate associations were tested using Chi-square, Fisher's exact, and t-tests. Survival was estimated via Kaplan-Meier. Due to the small sample size, multivariable modeling was not feasible. RESULTS: The cohort (mean age 25 years) showed a 1:1 sex ratio. Common pathologies: gliomas (38%), medulloblastomas (19%), and cavernomas (13%). Complete resection (n=15) was associated with lower recurrence rates (56%) than incomplete resection (71%, p=0.37) or no surgery (75%, p=0.42), though not statistically significant. Postoperative complications occurred in 75%, including meningitis (30%) and cranial neuropathies. Overall mortality was 54%. On univariate analysis, a consultation delay of less than six months (75% mortality vs. 20% for >6 months, p=0.003) and presence of pyramidal syndrome (62% mortality vs. 25%, p=0.021) were associated with higher mortality. Crucially, shorter consultation delay is interpreted as a surrogate marker of aggressive tumor biology rather than a modifiable patient factor. CONCLUSION: In our local experience, managing BSLs remains high-risk. While total resection may improve local control, the risk-benefit ratio for high-grade or infiltrating tumors requires careful individualization, including consideration of a primarily diagnostic or palliative role. Our findings may not be generalizable.