Imaging-based predictors of progression in brain abscess: the role of loculation and ADC gradient

Brain abscess remains a life-threatening condition despite advances in neurosurgical and antimicrobial therapy, and reliable imaging-based markers to guide postoperative risk stratification and surveillance remain limited. This study aimed to identify clinical and radiographic factors associated with abscess progression in neurosurgically treated patients, with particular emphasis on the apparent diffusion coefficient (ADC) gradient.

We retrospectively reviewed 53 patients with brain abscess who underwent neurosurgical treatment at Gachon University Gil Medical Center between January 2005 and April 2025. Clinical characteristics, microbiology, surgical approach, and imaging findings were analyzed. Univariable and multivariable logistic regression analyses were performed to identify factors associated with progression. ADC gradients were evaluated as continuous variables and dichotomized using a receiver operating characteristic (ROC)-derived cutoff.

The cohort included 38 men and 15 women, with a median age of 56.4 years. During a mean follow-up of 28.0 months, progression occurred in 17 patients (32%). Abscess loculation was significantly associated with progression in univariable analysis. Absolute ADC values at baseline or during follow-up were not predictive. ROC analysis demonstrated good discriminatory performance of the ADC gradient (AUC 0.857), with higher gradients associated with a significantly reduced risk of progression. The ADC gradient remained independently associated with progression in multivariable analysis.

Abscess loculation and the ADC gradient are radiographic features associated with progression in surgically managed brain abscess and may serve as useful imaging markers for clinical monitoring. Incorporation of ADC gradient assessment into postoperative imaging surveillance may help identify cases with a higher likelihood of treatment resistance and support clinical decision-making. However, these findings should be interpreted with caution given the retrospective and single-center design of the study, and further validation in prospective multicenter cohorts is needed.