Clinical Outcomes and Prognostic Factors in Cerebral Abscess: Hyperglycemia as a Predictor of the Need for Prolonged Antibiotic Therapy

Introduction

Cerebral abscesses are life-threatening intraparenchymal infections that frequently result in significant neurological deficits and substantial morbidity and mortality [1]. Despite advances in neurosurgical techniques and the availability of broad-spectrum antibiotics, timely diagnosis and effective management remain challenging due to variable clinical presentations, atypical imaging findings, and the increasing prevalence of comorbidities such as diabetes and immunosuppression [2]. Modern neuroimaging modalities, including computerized tomography (CT) and magnetic resonance imaging (MRI), have improved early detection; however, controversies persist regarding optimal surgical intervention, duration of antimicrobial therapy, and management strategies for patients with deep-seated or multiple abscesses [3].

Previous studies have reported divergent outcomes and inconsistent recommendations, particularly concerning prognostic factors such as neurological status on admission, abscess volume, microbiological profile, and metabolic conditions like hyperglycemia [4–6]. Furthermore, while most literature focuses on small or single-center cohorts, long-term data encompassing contemporary management practices are limited [7]. These gaps highlight the need for a comprehensive evaluation of patient characteristics, treatment strategies, and predictors of both clinical outcomes and antibiotic therapy duration.

The present study aims to address these knowledge gaps by retrospectively analyzing 2 decades of data from 112 patients treated for cerebral abscess at a tertiary medical center. Specifically, we sought to: 1) characterize the clinical, radiological, and microbiological features of cerebral abscess cases seen in our center; 2) evaluate treatment outcomes and identify factors influencing prognosis; and 3) assess determinants of the need for prolonged antimicrobial therapy, with particular attention to the role of hyperglycemia on admission. By integrating these findings, we aimed to provide practical guidance for optimizing management strategies and improving patient outcomes in contemporary clinical practice.

Material and Methods

PATIENT SELECTION:

The study included 112 patients treated with image-guided abscess drainage from December 2003 to June 2023. The Institutional Review Board of our institute approved this study. The informed consent was waived due to the retrospective design. Cerebral abscesses were diagnosed as single or multiple round lesions, with central hypodense areas and peripheral contrast enhancement on CT or MRI. Additionally, lesions were considered eligible if they had at least 1 of the following features: microorganisms grown from intracerebral materials and/or blood, and histopathology of the intracerebral lesion compatible with an abscess. Subjects who underwent craniotomy for cerebral abscess excision and those with epidural and/or subdural abscesses were excluded.

DATA COLLECTION:

A thorough review of patient medical records was conducted to determine a variety of parameters, including patient age, sex, past medical illness, symptom duration before admission, neurological deficits, consciousness status [evaluated using the Glasgow Coma Scale (GCS)] at admission, predisposing factors, and laboratory data [including blood count, C-reactive protein (CRP) level, erythrocyte sedimentation rate (ESR), procalcitonin level, and blood sugar level]. Blood glucose levels were assessed as random plasma glucose concentrations measured at the time of admission. A prior diagnosis of diabetes mellitus was identified from patients’ clinical histories and medical records and was included as a covariate in the statistical analyses. Microorganisms isolated from the abscess material, blood cultures, antibiotic regimens, and duration of antimicrobial therapy were also examined. Radiological characteristics, including the number of abscess pockets, locations, and pre- and postoperative abscess volumes were assessed. The abscess volumes were further estimated according to the widely used formula: [ABC/2], where A, B, and C represent the maximum diameters of the 3 dimensions of the abscess (Figure 1) [8].

SURGICAL PROCEDURE:

Abscess aspiration was performed using a navigation system (Stealth Station, Medtronic, MN, USA). We conducted the operation at the earliest period, if indicated, before antibiotic administration, to avoid sterilizing the cultures. Empirical antibiotics were initiated prior to surgical intervention in patients who underwent delayed surgery. During surgery, the abscess content was aspirated maximally, while a proximal catheter was connected to distal plastic bags, which were kept for at least 12 hours. Postoperative residual volumes were measured from immediate postoperative CT scans obtained for all patients following abscess aspiration. Follow-up imaging studies were subsequently performed at 1–2 week intervals to monitor further resolution. Pus collected during the operation was cultured for bacteria and fungi. Antimicrobial susceptibility testing was performed on the identified microorganisms.

ANTIMICROBIAL THERAPY:

We administered empirical antibiotics intravenously immediately after abscess aspiration until isolating and identifying the causative microorganisms. The antibiotic regimen was adjusted based on the target microorganisms identified in the culture test. Patients with cultures that failed to grow were treated with antibiotics directed at the most likely organisms, in consideration of the patient’s medical history and presumed primary infection focus. The duration of therapy was guided by a combination of factors, including the patient’s clinical improvement, normalization of laboratory markers (such as white blood cell count, ESR, and CRP), and radiological evidence of abscess resolution or marked reduction on follow-up imaging.

OUTCOME ASSESSMENT:

Patient outcomes were assessed using the Glasgow Outcome Scale (GOS) at discharge, categorized as either favorable (GOS ≥4) or unfavorable (GOS <4) to explore factors associated with treatment outcome.

STATISTICAL ANALYSIS:

All statistical analyses were performed using SPSS software (version 17.0; SPSS, Inc., Chicago, IL, USA). Results are presented as medians with ranges or means with standard deviations. Univariate analysis employed the chi-square test, t-test, and Fisher’s exact test to compare parameters between patients with good outcomes or early responses to antimicrobials, and those with poor outcomes or delayed antibiotic responses. A multivariate logistic regression method was applied to variables with P<0.1 in the univariate analyses to assess their prognostic significance for clinical outcomes, as well as the duration of antibiotic administration. Pearson’s correlation coefficients were calculated to assess correlations between continuous variables. Statistical significance was set at P<0.05.

Results

PATIENT POPULATION:

The patient characteristics are outlined in Table 1. This study included 90 male and 22 female patients. The median patient age was 54 years (range: 11–83 years). The most common symptom was headache (57 patients). Hemiparesis was observed in 42 patients, while fever was noted in 21 patients during their initial hospital visit. Other symptoms included dysarthria or aphasia (17 patients), confusion (16 patients), visual compromise (13 patients), convulsions (10 patients), general weakness (9 patients), sensory abnormalities (8 patients), and cognitive decline (7 patients). Further, 90 (80%) patients had clear or mildly decreased consciousness (GCS ≥13), whereas 22 (20%) presented with moderately or severely decreased consciousness (GCS <13). The interval between symptom onset and hospital visit varied, with a median duration of 7 days (range: 0–49 days). Possible predisposing factors were identified in 52 patients, as follows: pulmonary infection (15 patients), paranasal infection (9 patients), profound dental caries/periodontitis/gingivitis (8 patients), previous neurosurgery or trauma (6 patients), otologic infection (6 patients), infective endocarditis (3 patients), liver abscess (2 patients), cutaneous infection (2 patients), and deep muscle infection (1 patient). Twenty patients had been previously diagnosed with diabetes mellitus. Ten patients were immunocompromised due to: chemotherapy for malignant neoplasms (5 patients), hemodialysis for chronic renal disease (3 patients), or steroid therapy for other systemic diseases (2 patients; 1 with Addison’s disease, and 1 with lupus). Two patients had both cancer and diabetes mellitus.

LABORATORY AND RADIOLOGICAL CHARACTERISTICS:

The mean white blood cell counts among the 112 patients were 11 467±4419 cells/μL, and the mean blood CRP and ESR levels were 42.1±62.6 mg/L and 34.6±29.8 mm/h, respectively. Mean plasma glucose values at admission were 144.7±53.0 mg/dL. Fifty patients (45%) were hyperglycemic (≥140 mg/dL) upon admission.

On imaging studies, most of the patients (96 patients, 86%) had a single lesion. The abscess cavities were predominantly located in the frontal lobe (65 patients, 58%), followed by the temporal (29 patients, 26%), parietal (24 patients, 21%), and occipital (15 patients, 13%) lobes. Other locations included the thalamus (n=5, 4%), basal ganglia (n=4, 4%), cerebellum (n=3, 3%), and brain stem (n=3, 3%). The mean volume of the initial abscess cavity was 19.1±14.0 cc, while the mean postoperative residual volume of the abscess was 7.0±6.9 cc (Table 2).

MICROBIOLOGICAL CHARACTERISTICS:

Pathogens were isolated from the abscess material of 56 patients (50%). A single microorganism was identified in 45 patients, while multiple pathogens were found in 11 patients. Gram-positive bacteria were identified in 48 (43%) patients and gram-negative bacteria were identified in 18 (16%) patients. Fungus (Aspergillus) was found in 1 patient. The most frequently identified gram-positive bacteria were Streptococcus spp. (32 patients, 29%), Staphylococcus spp. (5 patients, 4%), Parvimonas micra (4 patients, 4%), Gemella morbillorum (4 patients, 4%), and Peptostreptococcus spp. (2 patients, 2%). The most common gram-negative organisms were Fusobacterium spp. (6 patients, 5%), Klebsiella pneumoniae (6 patients, 5%) and Prevotella spp. (3 patients, 3%) (Table 3). Blood cultures were conducted for all 21 febrile patients upon admission, among whom positive results were obtained in 6 (5%). Three bacterial species, Klebsiella pneumoniae (2 patients), Streptococcus anginosus (1 patient), and Staphylococcus aureus (1 patient), grew in both blood and abscess material. In another 2 patients, Streptococcus viridans and Acinetobacter haemolyticus, respectively, were isolated from blood culture, although the abscess material culture was negative. These organisms were presumed to be the causative pathogens, and appropriate antibiotic treatment was initiated on this basis.

CLINICAL OUTCOMES OF TREATMENTS:

The median time between admission and surgical drainage was 4 days (range: 0–21 days). Twelve patients (11%) underwent a second surgery because of a considerable residual abscess after the first surgery. Empirical antibiotics were administered intravenously immediately after abscess drainage until the presumed microorganisms were identified. Antibiotic therapy was initiated before abscess drainage in 21 patients with cerebritis. The most frequently used antimicrobial regimen comprised a combination of metronidazole and ceftriaxone (44 patients, 40%), which was often combined with vancomycin or gentamycin. The median duration of intravenous antimicrobial use was 43 days (range: 8–66 days). Following completion of intravenous antimicrobial therapy, none of the patients were prescribed oral antimicrobials. Steroid agents were applied with caution in 18 (16%) patients with severe perilesional edema. At discharge, 87 patients (78%) had a favorable outcome (GOS ≥4), while 25 patients (22%) had unfavorable outcomes (GOS 1–3) (Table 4). Despite our treatment efforts, 15 patients (13%) died. Of these, 5 (4%) deaths were from the abscess itself, while 10 (9%) deaths were believed to be due to the underlying illness. Among the latter group of patients whose death may have been caused by the worsening of preexisting disease, 5 with pneumonia died from septic shock, 2 with a congenital atrial septum defect died from pulmonary embolism, 2 with chronic renal failure died from sudden cardiac arrest, and 1 with hepatocellular carcinoma died from cancer bleeding. Of the 97 survivors, 10 showed unfavorable outcomes (GOS 2 or 3). Four patients had abscesses involving the eloquent area. Of these, 1 abscess in the motor cortex, 1 abscess in the thalamus, and 1 abscess in the brain stem, resulted in serious neurological deficits incompatible with independent living, while 1 patient’s outcome was a tenacious vegetative condition caused by ventriculitis following abscess rupture.

PROGNOSTIC FACTORS:

The results of the comparative analysis of patient outcomes are presented in Table 5. Patients with unfavorable outcomes had a shorter mean symptom duration (4.4 days vs 8.8 days, P=0.033). The prevalence of patients with GCS <13 at admission (P<0.001) and those under immunocompromised status (P=0.028) was higher in the unfavorable group. The mean CRP (P=0.010) and blood sugar level (P=0.031) of the patients with unfavorable outcome were higher than those of the favorable group. In the multivariate logistic model, a GCS score of <13 (P=0.002, 95% confidence interval [CI] 2.045–24.348) emerged as a unique independent predisposing factor for unfavorable outcomes. To investigate the clinical and imaging parameters related to the treatment response to antibiotics, we categorized the subjects into 2 groups: (1) early responders (duration of antibiotic administration less than 6 weeks), which included 54 patients, and (2) delayed responders (duration of antibiotic administration >6 weeks or until death), which included 58 patients. The mean white blood cell count and plasma glucose level of early responders were significantly lower than those of delayed responders (10 288/μL versus 12 565/μL, P=0.005; 125.7 mg/dL vs 162.4 mg/dL, P<0.001). Elevated plasma glucose level upon admission was further identified as an independent variable for long-term antibiotic use in the multivariate statistical model (P=0.004, 95% CI 1.005–1.029, Table 6). A positive relationship was also observed between the initial plasma glucose level and duration of antimicrobial administration (P<0.001, r=0.455, Figure 2).

Discussion

Four decades ago, cerebral abscesses were associated with mortality rates as high as 40% [4], largely due to delayed diagnosis, limited imaging capabilities, and the absence of effective antimicrobial therapy. The introduction of advanced neuroimaging techniques has significantly improved early detection and precise localization of abscesses, enabling timely intervention. Alongside these diagnostic advances, refinements in neurosurgical procedures and the development of broad-spectrum antibiotics have collectively contributed to a substantial reduction in both mortality and morbidity. In a recent study involving 139 patients, Kameda et al [9] reported an overall mortality rate of 11.5%, with most deaths attributable to underlying comorbidities, such as uncontrolled systemic infections or exacerbation of preexisting malignancies, rather than the abscess itself. In our cohort, the overall mortality was 13%, with the majority of deaths linked to systemic complications, cardiac disease, or worsening of chronic conditions. Notably, the cerebral abscess itself was identified as the primary cause of death in only 5 patients, emphasizing that contemporary management strategies must address not only local abscess control but also the patient’s capacity to withstand systemic challenges.

Several patient-related factors have consistently been associated with unfavorable outcomes. Previous studies [4,6] have reported that male sex, advanced age, altered mental status at presentation, and concurrent pulmonary infection are significant predictors of poor prognosis. Additionally, prolonged duration from symptom onset to hospital admission and reduced consciousness levels correlate with increased morbidity and mortality. Our findings are in agreement with these observations, showing that patients with preserved mental status upon hospital arrival are more likely to achieve favorable outcomes, particularly after controlling for confounding variables in multivariate analyses. These results highlight the importance of early recognition and prompt referral to specialized centers for improving patient outcomes.

Cerebral abscesses develop via multiple mechanisms, with contiguous spread from infections of the ear, paranasal sinuses, or teeth representing a major source [10,11]. Approximately half of all cerebral abscesses reported in prior studies arose from adjacent infectious foci. In our cohort, 23 patients (21%) had a contiguous infection focus, including 9 cases of sinusitis, 8 dental infections, and 6 cases of otitis media, consistent with recent literature reporting that roughly 23% of cerebral abscesses originate from head and neck infections [8]. Aggressive management of these primary infections through surgical debridement, drainage, and targeted antimicrobial therapy is critical for preventing secondary abscess formation and ensuring comprehensive patient care.

Comorbidities represent an additional key determinant of susceptibility to cerebral abscesses and influence clinical outcomes. Conditions such as diabetes mellitus, malignancy, immunodeficiency, autoimmune disorders, chronic renal or hepatic disease, and the use of immunosuppressive therapies increase infection risk and complicate management [8,12,13]. In our series, 28 patients (25%) had underlying medical conditions contributing to immunocompromise: 20 with diabetes mellitus, 10 with immunosuppression due to cancer, chronic renal failure, Addison’s disease, or systemic lupus erythematosus, and 2 with both diabetes and cancer. These findings align with previous studies identifying diabetes as a major predisposing factor [8,14,15]. While some reports suggest that comorbidities such as diabetes or liver disease are associated with higher mortality [14], our analyses did not demonstrate a statistically significant correlation between comorbidity type or presence and outcomes, indicating that other factors – including abscess characteristics, timeliness of intervention, and institution-specific management protocols – may have a stronger prognostic influence.

Accurate identification of the causative microorganism is essential for effective management of cerebral abscesses [3,16]. Blood cultures are particularly valuable when neurosurgical aspiration is not performed or when abscess fluid cultures yield negative results, and are indispensable in cases suspected of hematogenous spread. The development of stereotactic and neuro-navigation systems has enabled precise, minimally invasive image-guided abscess aspiration, allowing rapid drainage and identification of pathogens with minimal procedural risk [17,18]. According to the guidelines of the European Society of Clinical Microbiology and Infectious Diseases [3], neurosurgical intervention, preferably stereotactic aspiration or excision, is strongly recommended whenever feasible. Conservative management with intravenous antibiotics alone may be considered for small, asymptomatic lesions, although this approach requires close clinical and radiological monitoring. While the guidelines emphasize early surgical intervention, they do not provide absolute size or volume thresholds for intervention. At our institution, treatment decisions are made according to a structured protocol that integrates abscess size, anatomical location, neurological status, and patient risk profile, as supported by the existing literature [14,19–23]. Small abscesses (<1–2 cm in diameter or <4 mL in volume) without mass effect or neurological deterioration are generally managed conservatively with intravenous antibiotics. In contrast, larger abscesses (>2–2.5 cm in diameter or >4 mL in volume), particularly those associated with mass effect, elevated intracranial pressure, or progressive neurological deficits, are managed with stereotactic image-guided aspiration or drainage. Lesions located in non-eloquent or surgically accessible regions are prioritized for aspiration to achieve both decompression and microbiological diagnosis, whereas deep-seated or brainstem abscesses in high-risk patients may initially be managed conservatively under close clinical and radiological observation. Abscess volume estimation, typically calculated using the ellipsoid formula (length×width×height×0.5), provides a practical and reproducible method for guiding therapeutic decision-making. Evidence from previous studies indicates that abscesses exceeding 10 mL rarely respond adequately to antibiotic therapy alone and almost invariably necessitate surgical drainage [24,25]. Incorporating volume measurement into clinical assessment enhances objectivity in treatment planning, facilitates stratification between conservative and surgical management, and aids in predicting the likelihood of repeat interventions. Although some reports suggest that complete excision, including abscess wall removal, may shorten the clinical course and reduce repeat intervention rates [26,27], contemporary evidence indicates no additional benefit over minimally invasive stereotactic aspiration [9,17,28]. Minimally invasive procedures offer advantages including lower operative morbidity, shorter recovery, and sufficient abscess material for microbiological evaluation, informing targeted antibiotic therapy.

Laboratory markers predictive of cerebral abscess outcomes remain limited. Hyperglycemia has been associated with adverse outcomes in various systemic and neurological conditions, including sepsis, ischemic stroke, and traumatic brain injury [29–32]. Tight glucose control has been shown to reduce morbidity and mortality in critically ill patients [33]. Mechanistically, hyperglycemia increases circulating cytokines and impairs immune responses, reflecting systemic stress [34]. Central nervous system injury itself may contribute to dysregulated glucose metabolism, as the hypothalamus plays a key role in glucose homeostasis [35]. Studies indicate that hypothalamic injury or altered autonomic regulation via structures such as the insular cortex can precipitate hyperglycemia, potentially impacting outcomes [35–37]. In our cohort, 45% of patients presented with hyperglycemia (≥140 mg/dL) at admission. Elevated glucose levels correlated with prolonged antibiotic therapy (>6 weeks) and showed a statistically significant association in multivariate analysis. These observations suggest that hyperglycemia may occur as a consequence of central nervous system injury, causing impaired glucose regulation. It could thus serve as a potential indicator for predicting treatment outcomes and antibiotic responsiveness in patients with a cerebral abscess, providing practical guidance for clinicians by emphasizing the value of early metabolic assessment and close monitoring, which can help anticipate treatment duration, optimize management strategies, and implement timely interventions for high-risk patients.

Our study had several limitations. Firstly, because this was a retrospective study, it was inherently prone to potential confounding factors that may have influenced our observations. Additionally, we did not include patients who underwent craniotomies for abscess excision, which may have affected our analyses. We further measured random blood glucose levels upon admission to determine hyperglycemia. Identification of diabetes mellitus in our cohort relied on clinical histories and medical records. The possibility of undiagnosed diabetes cannot be excluded, as glycosylated hemoglobin levels or oral glucose tolerance test results were not systematically available in this retrospective dataset. Therefore, the association observed between admission hyperglycemia and treatment outcomes may partly reflect contributions from unrecognized chronic hyperglycemia. Future prospective studies incorporating standardized metabolic profiling are warranted to clarify the interplay between acute hyperglycemia, chronic diabetes, and clinical outcomes, and to determine whether targeted glycemic control could improve therapeutic responses.

Conclusions

In our study, the mental status of patients upon admission was a strong predictor of treatment outcomes, with lower GCS scores associated with unfavorable prognosis. The overall mortality rate was 13%, though most deaths were related to preexisting medical conditions rather than the abscess itself. Among survivors, combination therapy with image-guided abscess aspiration followed by targeted antimicrobial treatment demonstrated high efficacy. Additionally, hyperglycemia at admission was independently associated with prolonged antibiotic use, suggesting that early assessment of metabolic status may help guide treatment planning and anticipate therapy duration.