Sarcopenic obesity and cachexia as nutritional risk phenotypes and histology-specific outcomes after intracranial tumor resection: a histology-stratified NSQIP analysis of 27,057 cases

Abstract Preoperative risk stratification in neurosurgery typically uses body mass index (BMI), albumin, or frailty as single-marker predictors. We assessed whether a composite phenotype predicts outcomes after cranial neuro-oncologic surgery and whether effects differ by tumor histology. A retrospective analysis of 27,057 patients undergoing craniotomy for meningioma, malignant glioma, or brain metastasis was performed using the American College of Surgeons National Surgical Quality Improvement Program database (2016–2023). The composite phenotype incorporated BMI and serum albumin: adequately nourished, obese-replete, sarcopenic-obese, and cachectic-malnourished. Outcomes included major 30-day morbidity, 30-day mortality, failure to rescue (FTR), and non-home discharge. Associations were estimated using inverse probability of treatment weighting (IPTW), histology-stratified analyses, and interaction testing. Compared with adequately nourished patients, sarcopenic-obese and cachectic-malnourished phenotypes conferred elevated odds of major 30-day morbidity (IPTW odds ratio [OR] 1.47 and 1.60, respectively) and 30-day mortality (OR 2.28 and 1.96). Obese-replete showed no mortality increase (OR 1.06) and reduced FTR (OR 0.76), most pronounced in the metastasis cohort (OR 0.58). Phenotype effects differed by histology for mortality (interaction p = 0.044) and non-home discharge ( p = 0.009), with sarcopenic obesity conferring a 3.61-fold mortality increase among meningioma patients. The composite phenotype achieved discrimination comparable to but not exceeding serum albumin alone (area under the curve [AUC] 0.75 vs. 0.76 for mortality). A biomarker-defined nutritional phenotype separated distinct high-risk subgroups but did not improve discrimination beyond serum albumin alone, with histology-specific effects. These findings support histology-aware nutritional phenotyping for preoperative risk communication and hypothesis generation rather than superior prediction.