Analysis of Location-Specific Volumetric Cutoffs in Intraparenchymal Hemorrhage: A Case Series of 94 Patients

Accurate prognostic assessment is essential for guiding management in intracerebral hemorrhage (ICH). While hematoma volume and location are key determinants, recent evidence highlights the need for a more precise topographic evaluation. We investigated whether newly proposed location-specific volumetric cutoffs improve the prediction of 6-month functional outcome, assessed with the modified Rankin Scale, and 30-day mortality.

We retrospectively analyzed 94 patients with spontaneous ICH, performing semi-automated hematoma volume segmentation and categorizing its topographic location as lobar, external capsule/putamen, internal capsule/globus pallidus, thalamus, cerebellum, or brainstem. Ordinal logistic regression was used to evaluate the predictive performance of location-specific volumetric cutoffs for 6-month disability (modified Ranking Scale), both as standalone predictors and when incorporated into the ICH and functional outcome scores. Binary logistic regression evaluated their ability to predict 30-day mortality, both as standalone and when integrated into the ICH score.

Location-specific cutoffs improved prediction of 6-month disability compared to the traditional 30-mL threshold (Nagelkerke's R² [R²N]: 0.170 vs. 0.092). Their integration enhanced the performance of both the ICH score (R²N: 0.267 vs. 0.253) and the FUNC score (R²N: 0.262 vs. 0.244). For 30-day mortality, location-specific cutoffs performed better as standalone predictors (area under the curve: 0.757 vs. 0.715) but reduced accuracy when integrated into the ICH score (area under the curve: 0.864 vs. 0.929).

Location-specific volumetric cutoffs improve the prediction of long-term disability and enhance the performance of established prognostic scores. However, this benefit does not extend to short-term mortality, for which the traditional ICH score remains superior. Larger multicenter studies are needed to validate their clinical applicability.