Microbiological records for abscesses and blood cultures, medical records, and neuro-imaging findings using pre-existing standardized evaluation forms were retrospectively reviewed for patients with bacterial brain abscesses admitted to Kaohsiung Chang Gung Memorial Hospital, a 2482-bed acute-care teaching hospital providing primary and tertiary referral care, in the period 1986–2007.
Diagnostic criteria of bacterial brain abscess
The diagnostic criteria of bacterial brain abscess were: 1) characteristic computerized tomography (CT) and/or magnetic resonance imaging (MRI) findings; 2) evidence of brain abscess seen during surgery or histopathologic examination; and 3) classical clinical manifestations such as headache, fever, localized neurological signs and/or consciousness disturbance
]. Brain abscesses were defined as nosocomial according to the 1988 guidelines of the Centers for Disease Control
]. Brain abscesses related to head trauma with skull fracture or neurosurgical procedures were classified as a post-neurosurgical form. Otherwise, patients who presented with no distinctive characteristics or those who had not undergone invasive procedures were classified as the spontaneous form.
Diagnostic criteria and therapeutic regimens of nasopharyngeal carcinoma (NPC)
In all patients, experienced pathologists diagnosed NPC histologically, while multi-disciplinary teams administered treatment. The hospital’s therapeutic protocol was according to the National Comprehensive Cancer Networks (NCCN) Clinical Practice Guidelines in Oncology-Head and Neck Cancers (USA) and the Kaohsiung CGMH Head and Neck Oncology Group of Chang Gung Memorial Hospital Cancer Center. The therapeutic strategies were as follows: patients with stage I-IIA (AJCC system) were given radiotherapy alone whereas those with stage IIB-IV were given concurrent chemo-radiotherapy (CCRT).
Exclusion and inclusion criteria
Patients with evidence of brain abscesses not due to bacterial pathogens were excluded from this study. Patients who were initially treated in other hospitals but subsequently transferred to the study hospital for further therapy were included in this study, with the initial clinical data collected in those hospitals used for analysis.
The Glasgow coma scale (GCS) score was determined by neurosurgeons or neurologists as the patient arrived at the emergency room. All of the patients received brain CT scans shortly after arriving at the emergency room. Follow-up brain CT and/or MRI studies were performed for any clinical deterioration, including acute onset of focal neurologic deficits, seizures or status epilepticus, and progressively disturbed consciousness, as well as post-neurosurgical procedures. Hydrocephalus was judged retrospectively by a dilated temporal horn of the ventricle without obvious brain atrophy and/or an Evan’s ratio (the ratio of the ventricular width of the bilateral frontal horn to the maximum bi-parietal diameter) >0.3 on initial CT scans.
The volumes of brain abscesses on admission CT scans were measured. An experienced radiologist who was blinded to the patients’ clinical and biochemical data analyzed the CT scans for volumetric measurements of brain abscess volumes. All images were processed using the imaging processing software (Vitrea version 18.104.22.168, Vital images, Minnesota, U.S.A.) running on an off-line workstation. The volumes were calculated using a semi-automated process. The examiner manually drew regions-of-interest (ROI) in each slice throughout the brain abscess. Contiguous voxels were automatically summed to yield a brain abscess volume. The observer drew the brain abscess twice, at an interval of one month.
A trained research assistant performed the measurements again. Maps of the ROI used for measurement were stored and confirmed by a neurosurgeon. Intra- and inter-observer reproducibility of these measurements was evaluated using intra-class correlation coefficients. For brain abscess volume measurements, the intra-observer agreement was r
0.99, while the inter-observer agreement was r
0.99. The “volumes of brain abscesses” indicated the sum of all volumes of brain abscesses if at least two were found.
Combined surgical intervention and antibiotic therapy were the mainstays of treatment of bacterial brain abscesses. In the study hospital, surgical management consisted of imaging-guided stereotactic aspiration or craniotomy with complete excision. The contents of the abscess were aspirated using a ventricular catheter via burr hole or through a small craniotomy, which left the capsule alone. Craniotomy and resection of the abscess were defined as excision. The choice of one procedure over another was based on the patient’s age and neurologic condition, the location, stage and type of abscess, and the presence of multiple lesions.
Stereotactic aspiration was the simplest and safest method of obtaining pus for culture. It allowed for the precise localization and decompression of the abscess cavity using a minimally invasive technique, and was valuable in treating deep-seated lesions, lesions in eloquent areas, and multiple abscesses
]. The combination of third-generation cephalosporins and metronidazole for 8–12
weeks was the mainstay of initial empiric antimicrobial treatment for bacterial brain abscesses. The choice of final antibiotics was guided by the final culture results.
All of the materials from cerebrospinal fluid and/or blood, and/or drainage from the ear or sinuses were cultured for aerobic and anaerobic bacteria, Mycobacterium, and fungi. Antibiotic susceptibility was determined using the Kirby-Bauer disc diffusion method (Mueller-Hinton II agars; Becton Dickinson Microbiology Systems, Cockeysville, MD).
Evaluation of therapeutic outcome after discharge and a minimum 18-month follow-up was based on the Glasgow outcome score (GOS) as follows: good recovery, moderate disability, severe disability, persistent vegetative state, and death
Two separate statistical analyses were performed. First, the demographic data between the NPC and non-NPC groups were compared. Categorical variables were compared using the chi-square test or Fisher’s exact test, as appropriate. Continuous variables within the two groups were compared using the independent t-test for parametric data and the Mann–Whitney U test for non-parametric data. Second, significant variables (p
0.05) associated with an NPC patient group were entered into a forward stepwise logistic-regression analysis model that allowed for simultaneous control of multiple factors. Variables with a zero cell count in a 2-by-2 table were eliminated from logistic analysis, while only variables with a strong association with fatality rate (p
0.05) were included in the final model. All statistical tests were two-tailed and were conducted using the SAS software package, version 13.0 (2002, SAS Statistical Institute, Cary, North Carolina).