Surgical Neurology 65 (2006) 557 562 www.surgicalneurology-online.

com

Infection

Brain abscess in 142 patients: factors influencing outcome and mortality

Jen-Ho Tseng, MDa, Ming-Yuan Tseng, MD, MPhil a,b,4
b a Division of Neurosurgery, Department of Surgery, National Taiwan University Hospital, Taipei 100, Taiwan Academic Department of Neurosurgery, Addenbrookes Hospital, University of Cambridge, CB2 2QQ Cambridge, United Kingdom Received 11 May 2005; accepted 10 September 2005

Abstract

Background: With the introduction of CT, stereotactic techniques, and broad-spectrum antibiotics, the outcome for brain abscess has dramatically improved. The purpose of this study was to identify prognostic factors by reviewing data on 142 patients with brain abscess. Methods: Clinical data, including age, sex, medical history, duration of symptoms, initial neurological status, associated predisposing factors, laboratory data, treatment, and abscess characteristics, were considered as potential prognostic factors. A comparison was made between patients with favorable (GOS: moderate disability or good recovery) and those with unfavorable (GOS: death, persistent vegetative status, or severe disability) outcomes at discharge. Univariate (v 2 analysis or Fishers exact test) and multivariate logistic regression analyses were used to identify prognostic factors. Data were considered significant when the 2-tailed P value was lower than .05. Results: There were 98 male and 44 female patients (male/female ratio, 2.2). Their average age at diagnosis was 41.5 years (range, 2-84 years). There were 105 patients with a favorable outcome and 37 with an unfavorable outcome. Both univariate and multivariate analyses indicated that patients who were male, had an initial GCS score N 12, had no other septic complication, or had Grampositive cocci grown in abscess cultures had better outcomes. No association was found between outcome and other factors, including age, focal neurological deficits, seizures, laboratory findings, characteristics of the abscesses, associated factors, and treatment modalities. Conclusions: With the advancement of imaging studies and broad-spectrum antibiotic therapies, the outcome of brain abscess depends on prompt awareness of the diagnosis and effective infection control. D 2006 Elsevier Inc. All rights reserved.
Brain abscess; Outcome; Prognostic factor

Keywords:

1. Introduction Brain abscesses are relatively uncommon because of abundant blood supply and the protection of the blood-brain barrier in the brain. Before antibiotics were discovered, brain abscesses were almost uniformly lethal [20]. With the introduction of CT scanning and stereotactic techniques, the

outcome of brain abscess has dramatically improved [22]. The purpose of this study was to review data on 142 patients with brain abscess to identify clinical, biologic, and therapeutic characteristics that are important for outcomes.

2. Methods This is a retrospective study achieved by reviewing medical records coded with International Classification of Diseases, Ninth Revision 324.0 (intracranial abscess). From January 1, 1986, to December 31, 2004, 165 patients records were available for review at the National Taiwan University Hospital (Taipei, Taiwan). It is a universityaffiliated tertiary care teaching hospital in Taipei covering a population of 2.64 million (at the midpoint year 1995).

Abbreviations: CSF, cerebrospinal fluid; CT, computerized tomography; GCS, Glasgow Coma Scale; GOS, Glasgow Outcome Scale; IICP, increased intracranial pressure; MRI, magnetic resonance imaging; OR, odds ratio; SD, standard deviation. 4 Corresponding author. Academic Department of Neurosurgery, Addenbrookes Hospital, University of Cambridge, CB2 2QQ Cambridge, United Kingdom. Tel.: +44 1223 410634; fax: +44 1223 216926. E-mail address: myt22@cam.ac.uk (M.-Y. Tseng). 0090-3019/$ see front matter D 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.surneu.2005.09.029

558

J.-H. Tseng, M.-Y. Tseng / Surgical Neurology 65 (2006) 557 562 Table 1 Summary of the initial presentation of the patients Initial presentation Total (N = 142) Favorable outcome (n = 105) 81 (77.1) 24 (22.9) 40.0 F 20.9 56 (53.3) 62 (59.0) 15 (14.3) 38.6 F 0.6 Unfavorable P (v 2) outcome (n = 37) 20 (54.1) b .001 17 (45.9) 45.8 F 20.8 .147a 17 (45.9) .439 20 (54.1) 4 (10.8) 38.5 F 0.4 .597 1.0b .521a

Brain abscess was defined as a localized brain parenchymal lesion with ring enhancement by contrast medium and with perilesional edema visible on cranial CT scan or MRI and was associated with at least 1 of the following 3 characteristics: positive blood cultures, positive cultures of intracerebral materials, and histology of the intracerebral lesions suggesting brain abscess. Subdural and epidural empyemas were excluded. One hundred forty-two patients satisfied these criteria; their clinical data included age at diagnosis, sex, medical history, duration of symptoms before diagnosis, initial neurological status (clinical signs associated with IICP, GCS, seizures, focal neurological deficits), predisposing factors, laboratory data, surgical procedures, abscess location, and treatment modalities. Level of consciousness was arbitrarily divided into GCS N 12 and GCS V 12 for analysis. The outcome was assessed according to the GOS at the time of discharge [8]. Favorable outcome was defined as moderate disability (having a disability but being independent) and good recovery. Unfavorable outcome was defined as death, persistent vegetative status, and severe disability (conscious but with disability). A comparison was made between patients with a favorable outcome and those with an unfavorable outcome using STATA Intercooled 8.0 for Windows (College Station, Tex). Univariate analysis was performed using the v 2 test; Fishers exact test was used when any number of the cell in the n 2 table was less than 5. Multivariate logistic regression was used to identify significant factors affecting outcome. Data are presented as mean F SD and considered significant when the 2-tailed P value was lower than .05.

Male [n (%)] 98 (69.0) Female [n (%)] 44 (31.0) Age (y, mean F SD) 41.5 F 20.9 Signs suggesting 73 (51.4) IICP [n (%)] Focal neurological 82 (57.7) deficits [n (%)] Seizure [n (%)] 19 (13.0) Temperature 38.5 F 0.5 (8C, mean F SD) GCS [n (%)] N 12 106 (74.6) V 12 36 (25.4) Duration of symptoms 13.9 F 20.0 (d, mean F SD) Hematogenous origin 39 (27.5) [n (%)] Contiguous origin 25 (17.6) [n (%)] Diabetes mellitus 10 (7.0) [n (%)] Immune compromised 20 (14.1) [n (%)] Blood culture positive 13 (9.2) [n (%)] White blood cell 92 (64.8) count N 109/L [n (%)]
a b

88 (83.8) 19 (51.4) b .001 17 (16.2) 18 (48.6) 14.4 F 20.3 12.5 F 19.1 .623a 31 (29.5) 21 (20.0) 5 (4.8) 10 (9.5) 8 (7.6) 69 (65.7) 8 (21.6) 4 (10.8) 5 (13.5) 10 (27.0) 5 (13.5) 23 (62.1) .354 .315b .074 .008 .285 .697

t Test. Fishers exact test.

3. Results 3.1. Clinical presentation There were 98 male and 44 female patients (male/female ratio, 2.2) (Table 1). Their average age at diagnosis was 41.5 F 20.9 years (range, 2-84 years). There were 17 children ( b 15 years) and 19 older adults ( N 65 years). On admission, most patients had neurological manifestations such as focal neurological deficits (57.7%) and signs associated with IICP (headache, nausea, vomiting, papilledema; 51.4%). Seizures (generalized or partial) occurred in only 19 patients (13.4%). Fever was common and occurred in 87 patients (61.3%); 29 (20.4%) had a temperature of 38.58C or higher. However, the classic triad of brain abscess (fever, headache, and focal neurological deficits) was present in only 24 patients (16.9%). Consciousness was clear (GCS = 15) in 83 patients (58.5%), mildly disturbed (GCS = 13 or 14) in 23 (16.2%), moderately disturbed (GCS = 9-12) in 23 (16.2%), and severely disturbed or comatose (GCS = 3-8) in 13 (9.2%). The mean duration of symptoms before being diagnosed by image studies was 13.9 F 20.0 days, with the median duration being 10 days. The longest duration, 133 days, occurred in a patient with Aspergillus brain abscess.

A total of 87 patients (61.3%) presented with known predisposing factors, including hematogenous metastasis (chest infection, cyanotic congenital heart disease, or other right-to-left shunting), concomitant contiguous infections (otogenous, rhinogenous, penetrating head injury, or postneurosurgical infection), diabetes mellitus, and compromised immunity. A total of 39 patients (27.5%) had brain abscesses related to hematogenous spread, including 21 with cyanotic congenital heart disease, 3 with pulmonary arteriovenous fistula, 6 with subacute bacterial endocarditis, 8 with lung infections (pneumonia, lung abscess, pleural empyema), and 1 with liver abscess. Cyanotic congenital heart diseases were particularly prevalent among the children (adults vs children = 10.6% vs 80.0%; Fishers exact test P b .001). A total of 25 patients (17.6%) had brain abscesses associated with contiguous infection, including 10 with chronic otitis media, 8 with frontoethmoidal sinusitis, 5 with penetrating head injury, and 2 who underwent craniotomy. Of all the patients, 20 (14.1%) had immunity compromised, with 15 receiving steroid treatment for systemic lupus erythematosus, nephrotic syndrome, or graft-vs-host disease after bone marrow transplantation; 1 infected with human immune deficiency virus and acquired immunodeficiency syndrome; and

J.-H. Tseng, M.-Y. Tseng / Surgical Neurology 65 (2006) 557 562

559

4 having hematological disorders such as chronic myelogenous leukemia, Waldenstrfms macroglobulinemia, and aplastic anemia. Ten patients (7.0%) had poor-controlled diabetes mellitus. The average white blood cell count was 13.7 F 1.8/L; 92 patients (64.8%) had a white blood cell count higher than 10.0/L. 3.2. Characteristics of the abscess All patients had brain CT scans, which showed single abscesses in 113 (79.6%) and multiple ones in 29 (20.4%) (Table 2). The most common location for single abscesses was the frontal lobe (35.2%), followed by the temporal (18.30%), parietal (14.1%), and occipital (5.6%) lobes. No association was found between multiple abscesses and hematogenous origin. 3.3. Microbiological findings Lumbar punctures (LPs) were performed on 21 patients under the clinical suspicion of bacterial meningitis but without signs of IICP. None of them had immediate clinical deterioration caused by the LP. The microbiological studies consisted of cultures of brain abscess materials in 122 patients, blood cultures in 49, and CSF cultures in 21. Cultures of the abscess materials remained sterile for 39 patients (negative rate, 32.0%), including 6 who had had antibiotic treatment for other infections. Among the 84 patients with positive cultures from brain abscess

materials, 64 had single pathogens and 20 had multiple pathogens. Sixty-two bacteria were isolated. The most common pathogens were of Gram-positive cocci (streptococci, staphylococci), followed by anaerobes and Gram-negative enteric bacilli. Other pathogens, such as Hemophilus , Pseudomonas , diphtheroid species, Candida albicans , and Aspergillus fumigatus , were also found, albeit there were relatively few of them. Blood cultures were performed for 49 patients who had a clinical presentation of systemic infection (fever and leukocytosis). Only 13 of those patients had blood cultures that grew bacteria (positive rate, 26.5%); 7 of them had the same pathogen in both blood and brain abscess cultures. Of the 21 CSF cultures, 5 grew bacteria (positive rate, 23.8%) of the same species as that in the brain abscess, suggesting that ruptures of the brain abscess into the ventricle had occurred. 3.4. Treatment All patients were treated with parenteral antibiotics for at least 4 weeks. A total of 122 patients (85.9%) underwent surgical treatment: 74 had drainages by craniotomy, 34 had stereotactic aspirations, and 14 had excisions. The drainage or excision by craniotomy was performed for larger ( N 2.5 cm) or superficial brain abscesses. The stereotactic aspirations were performed in patients with a brain stem abscess, a small and deep-seated abscess, or multiple small abscesses. Of the 2 patients who had a brain stem abscess, 1 had undergone excision of a vestibular schwannoma but with postoperative CSF leakage. The brain stem abscess was treated with stereotactic aspiration and antibiotics. The other was a 2-year-old child with the tetralogy of Fallot. The brain stem abscess was managed with a microsurgical aspiration and antibiotic treatment. Both recovered with minimal neurological deficits. A total of 20 patients received medical treatment only: 12 had multiple small abscesses, 7 had single abscesses smaller than 2 cm in diameter, 2 were at a terminal stage of cancer, and 1 had ventriculitis after the abscess had ruptured into the ventricle. 3.5. Outcome

Table 2 Summary of the brain abscess characteristics of the patients Abscess characteristics Total (N = 142) Favorable outcome (n = 105) 40 (38.1) 25 (23.8) 13 (12.4) 21 (20.0) 1 (1.0) 6 (5.7) 35 21 18 5 1 5 20 14 (33.3) (20.0) (17.1) (4.8) (1.0) (4.8) (19.0) (13.3) Unfavorable outcome (n = 37) 18 (48.6) 2 (5.4) 8 (21.6) 3 (8.1) 3 (8.1) 3 (8.1) 15 5 2 3 1 2 9 6 (40.5) (13.5) (5.4) (8.1) (2.7) (5.4) (24.3) (16.2) .402a P (v 2)

Pathogens [n (%)] Sterile or no specimen 58 (40.8) Gram-positive cocci 27 (19.0) Gram-negative 21 (14.8) enteric bacilli Anaerobes 24 (16.9) Fungus 4 (2.8) Others 9 (6.3) Single abscess location [n (%)] Frontal 50 (35.2) Temporal 26 (18.3) Parietal 20 (14.1) Occipital 8 (5.6) Basal ganglia 2 (1.4) Posterior fossa 7 (4.9) Multiple abscesses 29 (20.4) 20 (14.1) Medical treatment only [n (%)] Excision of the 14 (9.9) abscess [n (%)] Drainage by 74 (52.1) craniotomy [n (%)] Stereotactic 34 (23.9) aspiration [n (%)] Sepsis [n (%)] 14 (9.9)
a

.007a

.454

11 (10.5) 58 (55.2) 22 (21.0) 1 (1.0)

3 (8.1) 16 (43.2) 12 (32.4) 13 (35.1) b .001a

Fishers exact test.

A total of 105 patients (73.9%) had a favorable outcome: 63 (44.4%) had full recovery and 42 (29.6%) had mild deficits. A total of 37 patients (26.1%) had an unfavorable outcome: 24 (16.9%) died during hospitalization and 13 (9.2%) had moderate or severe disability. The deaths were caused by underlying systemic infection or the terminal stage of cancer and not by the abscess itself. Univariate analysis (Tables 1 and 2) revealed that male patients had better outcomes than female patients (v 2 test P b .001). Patients with favorable outcomes also had an initial GCS score N 12 (v 2 test P b .001), without compromised immunity (v 2 test P = .008) or sepsis (Fishers exact test P b .001), or had Gram-positive cocci grown in the abscess materials. No association was found between outcome and other factors, including age, focal neurological

560

J.-H. Tseng, M.-Y. Tseng / Surgical Neurology 65 (2006) 557 562

deficits, seizures, laboratory findings, characteristics of the abscesses, and treatment modalities. For mortality, female sex (v 2 test P = .027), initial GCS score V 12 (v 2 test P = .011), associated predisposing factors (v 2 test P = .048), compromised immunity (v 2 test P = .020), uncontrolled diabetes (v 2 test P = .043), systemic infection (v 2 test P b .001), negative brain abscess cultures (Fishers exact test P = .015), and higher white blood cell count (22.7 F 40.7/L vs 12.0 F 10.9/L; t test P = .017) were significant contributing factors. No significant difference in the outcome was found among various treatment modalities. Multivariate logistic regression confirmed that being male (OR, 9.81; P = .002), having a GCS score N 12 (OR, 6.20; P = .019), and being sepsis free (OR, 761.49; P b .001) were the main factors of a favorable outcome (Table 3). Others did not have significant effects, except that the Gram-positive cocci (OR, 42.3; P = .013) grew on brain abscess cultures, which was a positive prognostic factor for the favorable outcome. High fever (temperature, z 38.58C) seemed to have negative effects on the outcome (OR, 0.25; P = .078).
Table 3 Logistic regression analysis of factors contributing to favorable outcome in the patients Variables Male GCS N 12 Without sepsis Age (per 10 y) Temperature z 38.58C Signs of IICP Focal neurological deficits Seizure Duration of symptoms b 7 d White blood cell count N 10 9/L Hematogenous origin Contiguous origin Diabetes mellitus Compromised immunity Abscess pathogen No growth Gram-positive cocci Gram-negative enteric bacilli Anaerobes Fungus Others Abscess location Frontal Temporal Parietal Occipital Basal ganglia Posterior fossa Multiple Treatment Medical Excision Drainage by craniotomy Stereotactic aspiration OR 9.81 6.20 761.49 1.18 0.25 0.82 0.69 0.33 2.03 1.74 2.78 1.52 0.26 0.44 1.00 42.3 1.49 3.81 17.2 0.06 1.00 1.30 10.20 0.58 0.05 0.47 2.67 1.00 0.45 0.54 0.40 95% Confidence interval 2.29- 42.07 1.35 - 28.39 19.90 - 29 144.37 0.80 - 1.74 0.05 - 1.17 0.17 - 3.98 0.14 - 3.37 0.06 - 1.80 0.39 - 10.67 0.39 - 7.77 0.40 - 19.44 0.18 - 12.82 0.02 - 4.24 0.06 - 3.33 P .002 .019 b .001 .411 .078 .807 .651 .201 .401 .469 .302 .701 .348 .429

4. Discussion Although brain abscesses are uncommon, even with the advancement of imaging technologies and antibiotic treatment, the mortality is still relatively high (range, 17% -37%) [2,9,15,22]. Common presentations of brain abscesses include IICP, focal neurological deficits, and clinical signs of infection [10,22]. Initial manifestations often correlate with the size, location, and progression of the abscess, virulence of infecting pathogens, and underlying host conditions [14]. Fever is a common finding, but its absence should not be used to exclude the diagnosis [5]. Focal neurological deficits depend on the location and the mass effect of the abscess; absence of such presentation also frequently causes the brain abscess to be overlooked. The classic triad (fever, headache, and focal neurological deficits) is relatively uncommon, and ruptures of the brain abscess into the ventricle often present as sudden worsening of preexisting headache accompanied by a new onset of meningeal irritation, which can cause a high mortality [19,21]. The trend of epidemiology of brain abscess is a decreasing incidence of otogenous abscess and an increasing incidence among patients with compromised immunity [14]. Brain abscess caused by hematogenous spread is common in patients with a right-to-left shunt or cyanotic congenital heart disease [9]. Cultures of brain abscess materials in these patients usually grow streptococci, the most common microorganisms in silent bacteremia [5]. Only patients with signs of systemic infection had blood cultures in our unit to increase the positive rate. Despite a 26.5% positive rate in our series, merely 14.2% of blood cultures grew the same pathogen as that in the abscess materials, which is similar to the findings of other series [5,18]. In patients with a right-to-left shunt, decreased arterial oxygen saturation combined with increased blood viscosity owing to increased hemoglobin levels may cause focal brain ischemia, which could serve as an excellent nidi for infection [7]. Furthermore, the right-to-left shunt can bypass the filter effect of pulmonary circulation; therefore, it has been advised that earlier correction of the shunt may reduce the likelihood of developing brain abscess in these patients, instead of carrying out palliative procedures alone [7]. Lumbar puncture has been considered hazardous in patients with brain abscess [4,5,16,18]. It is usually performed under a strong suspicion of concomitant meningitis and/or ventriculitis [5,16] and yields only 10% to 30% positive CSF cultures compatible with abscess cultures [4,5,18]. Although a significant proportion of the mortality has been thought to be caused by LP during early work [4,16], a recent study using multivariate regression failed to reveal such hazard [17]. Therefore, we believe that LP in patients with brain abscess could be justified in the absence of IICP and with clear manifestations of meningitis and/or ventriculitis. It should be judged upon individual merits as

2.23 - 803.72 0.18 - 12.07 0.40 - 35.89 0.02 - 17 804.92 0.005 - 0.83

.013 .707 .243 .422 .036

0.18 - 9.47 0.42 - 248.12 0.05 - 6.87 0.00 - 3.53 0.03 - 7.52 0.37 - 19.04

.794 .154 .666 .171 .593 .327

0.03 - 6.06 0.07 - 4.16 0.04 - 4.11

.551 .550 .437

J.-H. Tseng, M.-Y. Tseng / Surgical Neurology 65 (2006) 557 562

561

our analysis reveals that the outcome depends on the effectiveness of infection control (systemic and/or central nervous system). Careful culturing of abscess materials obtained at the time of surgery provides the best opportunity to make a microbiological diagnosis, although the incidence of sterile culture is high for patients already receiving antimicrobial therapy [9,13]. However, an initial broad-spectrum antibiotic coverage for the most common organisms should be the first choice for all critically ill patients, even if the culture results are being awaited. Because the infecting pathogens of brain abscess may be the same as those of other septic foci, the initial antimicrobial coverage could therefore be chosen according to the most likely ones at the suspected site of entry. In cases caused by these more resistant pathogens, such as fungi or Nocardia species, complete excision is more favorable than simple drainage [14]. In this study, the outcome is less favorable in female patients, although all potential confounders have been adjusted for in the multivariate analysis. This sex difference was not seen in other studies [1,13]. Another finding is that death is mainly caused by systemic infection or terminal cancer rather than directly by the brain abscess itself, suggesting that, although the current therapeutic strategy for controlling these abscesses is highly effective, the decisive factor for patients survival still depends on the host capability to combat offending pathogens [19]. Since 1975, some studies have advocated nonsurgical treatment for patients who are poor candidates for surgery or for those with surgically inaccessible lesions [3]. The major drawback to this approach is the potential toxicity involved in prolonged administration of empirical antimicrobial therapy [14]. Therefore, except during the stage of cerebritis, surgical treatment should be attempted, which achieves not only reducing the mass effect but also obtaining abscess materials for identifying infecting pathogens and, thus, for facilitating selection of antibiotics [11]. The following management policy has been recommended by Mamelak et al [12]: aggressive surgical drainage of abscesses larger than 2.5 cm in diameter, followed by 6 to 8 weeks of intravenous antibiotics and biweekly imaging studies for monitoring evidence of abscess re-expansion or failure to resolve despite antibiotics. With the availability of better imaging studies, stereotactic techniques, and more broad-spectrum antibiotics, the role of aspiration has increased because of its simplicity. Even in hardly accessible regions such as brain stems or eloquent areas among patients with high risk for surgery or with multiple brain abscesses, stereotactic aspiration combined with antibiotic treatment still can be applied [6]. Although the excision of abscess has been considered to shorten the clinical course [13], no further improvement in the outcome has been observed [11,22]. With the advancement in imaging technologies and broad-spectrum antibiotics, our data reveal that outcome depends on prompt awareness of the diagnosis and effective

infection control. No difference in the outcome is found among various treatment strategies and choices of surgical procedures. Identifying the offending pathogens is crucial to manage abscess more efficiently. Thus, we recommend the treatment strategy of Mamelak et al [12]; however, in addition, surgical procedures for obtaining culture materials should always be attempted when the clinical condition is permitted, even if the abscess is small ( V 2.5 cm), deeply seated, and/or multiple. Acknowledgment We thank Mr. E. Merchant for his assistance with the manuscript preparation. References
[1] Aydin IH, Aladag MA, Kadioglu HH, et al. Clinical analysis of cerebral abscesses. Zentralbl Neurochir 1988;49:210 - 9. [2] Bagdatoglu H, Ildan F, Cetinalp E, et al. The clinical presentation of intracranial abscesses. A study of seventy-eight cases. J Neurosurg Sci 1992;36:139 - 43. [3] Barsoum AH, Lewis HC, Cannillo KL. Nonoperative treatment of multiple brain abscesses. Surg Neurol 1981;16:283 - 7. [4] Carey ME, Chou SN, French LA. Experience with brain abscesses. J Neurosurg 1972;36:1 - 9. [5] Chun CH, Johnson JD, Hofstetter M, et al. Brain abscess. A study of 45 consecutive cases. Medicine (Baltimore) 1986;65:415 - 31. [6] Dyste GN, Hitchon PW, Menezes AH, et al. Stereotaxic surgery in the treatment of multiple brain abscesses. J Neurosurg 1988;69:188 - 94. [7] Fischbein CA, Rosenthal A, Fischer EG, et al. Risk factors for brain abscess in patients with congenital heart disease. Am J Cardiol 1974;34:97 - 102. [8] Jennett B, Bond M. Assessment of outcome after severe brain damage. Lancet 1975;1(1):480 - 4. [9] Kagawa M, Takeshita M, Yato S, et al. Brain abscess in congenital cyanotic heart disease. J Neurosurg 1983;58:913 - 7. [10] Kao PT, Tseng HK, Liu CP, et al. Brain abscess: clinical analysis of 53 cases. J Microbiol Immunol Infect 2003;36:129 - 36. [11] Kratimenos G, Crockard HA. Multiple brain abscess: a review of fourteen cases. Br J Neurosurg 1991;5:153 - 61. [12] Mamelak AN, Mampalam TJ, Obana WG, et al. Improved management of multiple brain abscesses: a combined surgical and medical approach. Neurosurgery 1995;36:76 - 85. [13] Mampalam TJ, Rosenblum ML. Trends in the management of bacterial brain abscesses: a review of 102 cases over 17 years. Neurosurgery 1988;23:451 - 8. [14] Mathisen GE, Johnson JP. Brain abscess. Clin Infect Dis 1997;25: 763 - 81. [15] Miller ES, Dias PS, Uttley D. CT scanning in the management of intracranial abscess: a review of 100 cases. Br J Neurosurg 1988;2: 439 - 46. [16] Morgan H, Wood MW, Murphey F. Experience with 88 consecutive cases of brain abscess. J Neurosurg 1973;38:698 - 704. [17] Qureshi HU, Habib AA, Siddiqui AA, et al. Predictors of mortality in brain abscess. J Pak Med Assoc 2002;52:111 - 6. [18] Schliamser SE, Backman K, Norrby SR. Intracranial abscesses in adults: an analysis of 54 consecutive cases. Scand J Infect Dis 1988;20:1 - 9. [19] Seydoux C, Francioli P. Bacterial brain abscesses: factors influencing mortality and sequelae. Clin Infect Dis 1992;15:394 - 401. [20] Stephanov S. Surgical treatment of brain abscess. Neurosurgery 1988;22:724 - 30.

562

J.-H. Tseng, M.-Y. Tseng / Surgical Neurology 65 (2006) 557 562

[21] Takeshita M, Kagawa M, Yato S, et al. Current treatment of brain abscess in patients with congenital cyanotic heart disease. Neurosurgery 1997;41:1270 - 8. [22] Yang SY. Brain abscess: a review of 400 cases. J Neurosurg 1981; 55:794 - 9.

Commentary This article is a review of data on 142 patients treated for brain abscess. The outcome was favorable in 75% of the patients. There were twice as many males as there were females. Fever was common, but the classic signs of abscess fever, focal deficit, and headache were present in only 17% of the patients. The average interval between onset of symptoms and imaging studies was approximately 2 weeks. Half the patients were seen within 10 days, but there still seemed to be a period wherein patients with a more aggressive infection were selected out before a proper referral could be made. Our experience suggests that we are seeing patients within 2 to 3 days of the onset of symptoms. Perhaps we are to assume that mortality would approach zero if the systemic disease is controllable and malignancy is excluded. I suspect that the assumption would be true if we could more effectively manage the early cerebritis period. Our experience has not been as good as that of the authors. It is interesting to note that the therapeutic approach selected, be it antibiotic therapy alone or intervention (either stereotactically or by craniotomy), had minimal effect on the final outcome. Harry O. Cole, MD Neurosurgical Associates Chesterfield, MO 63017, USA

I congratulate the authors for reporting on a large series of intracranial abscesses treated at one institution. Although the series is subject to the intrinsic problems associated with any retrospective review, by considering the International Classification of Diseases coding of all hospital admissions, it would seem that the analysis includes patients treated medically as well as surgically. The relative infrequency (16.9%) of the classic triad of headache, fever, and focal deficit in the presentation of cerebral abscess is illustrated by this series, and it highlights the need for clinical suspicion and subsequent scan in patients with any combination of these symptoms and/or seizures. It is particularly noteworthy that although 61% of the patients had fever, only 20.4% had a temperature in excess of 38.58C, which is the level at which most clinicians suspect serious infection. The white blood cell count was a

little more helpful in suggesting infection, although even this was not elevated in 35.2% of the cases. Patients presenting with a deficit or seizures and a ringenhancing cerebral lesion, posing the diagnostic dilemma of tumor vs abscess, are well known to neurosurgical residents and neurosurgeons alike. This article illustrates that a relatively long history of symptoms, the absence of fever, and a normal white blood cell count cannot be relied on to effectively rule out the diagnosis of abscess; in addition, it underscores the necessity of conducting biopsies promptly in these cases. I am relieved that the 21 cases of LP in the presence of abscess resulted in no significant morbidity. The authors conclude that this may therefore be an acceptable investigation in selected cases of cerebral abscess. Although the LPs in this series were all performed on patients with a meningitic presentation (and are, therefore, a selected group of the total abscess population with the highest chance of positive culture), they provided no positive growth in 74% of the cases. Therefore, considering the low yield and the devastating consequences of cerebral herniation subsequent to LP (not to mention the unenviable medicolegal predicament), I would caution against LP in all but the most exceptional circumstances of cerebral abscess. Indeed, this series highlights the value of routine CT prior to LP. Conversely, I find myself disagreeing with the authors regarding blood cultures, and I would strongly recommend that they be performed routinely in cerebral abscess patients prior to antibiotic commencement. The relatively low yield of blood cultures (26.5% positive) needs to be considered from the perspective that they are almost risk free, are cheap, and take only a few minutes. The authors finding that only 14.2% of blood cultures cultured the same organism as the abscess aspirate is consistent with the 26.5% and 68% culture positivity from blood cultures and abscess aspirates, respectively (which would suggest an 18% positivity for both if all aspirated abscesses had blood cultures taken). As expected, the abscess aspirate is the most likely source of positive cultures (68%); considering this as well as the accuracy and safety of stereotactic aspiration, the authors have provided a compelling argument for the initial treatment of all abscesses surgically. The results of 73.9% independent outcome and 9.2% moderate or severe disability are excellent, particularly when it is considered that all deaths (16.9%) resulted from systemic sepsis or cancer and not from cerebral abscess per se. John D. Laidlaw, MD Private Medical Center Royal Melbourne Hospital Parkville VIC 3050, Australia