Historically, brain abscesses were usually caused by contiguous infection following sinusitis or middle-ear infection and carried a poor prognosis 1 2 , but as a result of improved neuro-imaging, surgery and antibiotic treatment, brain abscesses are now relatively rare, and mortality has improved.

Several aspects of management remain controversial, including the need for surgery and optimal surgical approaches, type and length of antibiotic treatment, and need for monitoring during treatment. No randomized controlled prospective trials are available.

We describe brain abscesses diagnosed during a 15-year period. We report microbial and treatment characteristics in relation to outcome for patients with pyogenic brain abscesses treated with a combination of surgery and antibiotics or antibiotics alone. A particular focus is placed on a detailed analysis of diagnostic delays as well as type and length of antimicrobial treatment in relation to outcome.

We retrospectively reviewed adult patients ≥ 16 years of age, diagnosed with confirmed or suspected pyogenic brain abscess at the Departments of Neurosurgery, Infectious Medicine and Neurology at Rigshospitalet University Hospital, Copenhagen, between January 1994 and April 2009. Cases were ascertained through ICD 10 discharge codes. Rigshospitalet is a tertiary, regional and national specialist referral hospital. During the study period, we estimate that approximately 1.7 million people at risk of brain abscess were primarily served by Rigshospitalet. Medical records were reviewed by paper and electronic data query. Data were entered by custom made data entry forms in Epidata and cross-validated 3 . Contiguous infection was defined as infection secondary to otitis media, mastoiditis, sinusitis or dental infections. One patient was lost to long-term follow-up by emigration, 6 months after hospital discharge. Complete follow-up for the remaining patients was available to January 2011. Inclusion criteria were characteristic computed tomography (CT)/magnetic resonance imaging (MRI) findings and evidence of bacterial brain abscess from surgery, autopsy and/or appropriate microbiological specimens (Cerebrospinal (CSF), blood, sputum) or characteristic CT/MRI findings, a clinical history and treatment response compatible with pyogenic brain abscess in patients with negative microbial findings. Patients with mycobacterial, parasitic or fungal abscesses and cases with subdural empyema were excluded.

A total of 102 patients were diagnosed with brain abscess between January 1994 and April 2009. Median age was 47 years and 65% were male. This yields an estimated incidence of 0,4/100.000/year. Table 1 provides baseline characteristics.

*: 12 patients had more than one comorbidity. GCS: Glasgow Coma Score, IVDU: Intravenous Drug User.

Despite diagnostic advancements with the introduction of CT and later MRI, as well as better antimicrobial agents and neurosurgical procedures, brain abscess continues to be a serious, potentially life threatening condition. Mortality continues to be relatively high. In our study almost a quarter of patients had a poor outcome.

The key prognostic factors were level of consciousness at admission and presence of comorbidity. These findings confirm previous observations. Mortality has generally been reported to be higher with severe mental state changes 4 7 , and rapidly progressing neurological impairment 8 9 and with predisposing illness 9 10 . Intraventricular rupture was associated with a poor prognosis; of 10 patients, 3 died and 3 had severe neurological deficits at discharge. This observation confirms previous findings by Takeshita and co-workers 11 12 . In particular, increased meningeal irritation and/or localized enhancement of the ventricular wall adjacent to the abscess are predictors of subsequent rupture and need to be recognized and managed particularly aggressively 11 12 .

Our study has several strengths. In particular, the complete availability of detailed diagnoses, neuroimaging, microbiological and follow-up data provided the opportunity for detailed analysis of the characteristics and management of brain abscess in relation to outcome. We believe all patients diagnosed with pyogenic brain abscess at our institution were identified, including both surgically and non-surgically treated patients as well as patients who died very early after admission.

A similar Danish study by Nielsen et al. of 200 patients, diagnosed with cerebral abscess at Rigshospitalet between 1935 and 1976 including patients managed before the availability of penicillin and CT scans, showed a decrease in mortality from 78% before 1945 to 37% between 1945 and 1958 and to 17% between 1958 and 1976 13 . Of note, this mortality rate is similar to the 17% 3-months mortality rate observed in our study, suggesting that the overall outcome has changed little in recent years. However, as patients with brain abscess are heterogeneous, the outcome is critically dependent on baseline characteristics. In agreement with other studies from the last three decades we observed an increased rate of patients with unknown source of infection (25%) and/or concurrent morbidity (46%), compared to the previous study at by Nielsen in which only 15% had unknown source of infection and 25% concurrent morbidity 13 .

Recent retrospective studies from East Asia and Europe have reported marked variations in the outcome of brain abscess with mortality rates ranging from 6% to 35% and adverse outcome ranging from 10 to 58% 4 5 6 7 8 9 10 14 15 16 17 18 19 20 21 22 23 24 25 . The highest mortality rates have been reported in case-series with a high proportion of non-surgically managed patients 9 15 , primarily reflecting the overall poor prognosis of patients with severe septic shock, meningitis, multiple abscesses and/or staphylococcal infections. In contrast, case series with a high proportion of surgically managed streptococcal abscesses tend to have a much better overall outcome 6 14 19 23 .

A substantial number of patients (n=36) were investigated by lumbar puncture. The CSF findings varied considerably with CSF cellularity ranging from normal to high, emphasizing the need for early adequate neuroimaging in patients suspected of meningitis and low level (< 1000 cells/μl) pleocytosis, as 39% had CSF leucocytes between 15 and 1000 and 19% normal values. In agreement with previous studies, the microbial yield of lumbal puncture was low, with only 16% being culture positive. Furthermore, lumbar puncture is generally contraindicated in patients with suspected intracranial mass lesions, such as an abscess due to risk of herniation. Blood cultures were also rarely positive, but should yet always be considered, as they are minimally invasive and may yield the pathogens in some instances 4 26 .

Diagnostic delay has generally been believed to be a key contributing factor to the severity and outcome of brain abscess 6 27 . In our study, the majority of patients presented with uncharacteristic symptoms and diagnosis was frequently delayed. The usual signs of infection such as fever, increased WBC and increased phase reactants were only present in 60% at presentation. Although fever was the single most common presenting symptom, 42% displayed no febrile symptoms. It is a common misconception that fever is necessarily a part of the presenting scheme of cerebral abscess. In several series fever has been reported to occur in about 50% or less of cases, and its absence should never be used to exclude the diagnosis of brain abscesses 5 10 16 18 27 28 .

The significant delays observed in timely diagnosis and treatment were caused by both “doctors delay” (delay or misinterpretation of neuro-imaging, referral and definitive surgery) and “technical delays” (caused by inadequate initial choice of imaging, e.g. CT without contrast). For example, in 10 patients, the initial CT imaging study was interpreted as being compatible with brain tumour or stroke. Importantly, access to urgent MRI was not always readily available. The differentiation between abscess and brain tumours is not always straightforward. A typical clue on MRI is the presence of a hypointense capsule on T2-weighted images, however a capsule may be absent and while advanced techniques such as Magnetic Resonance Spectroscopy (MRS) or brain- Positron Emission Tomography(PET)–CT, may improve the differential diagnosis between brain abscess and brain tumours, no single technique achieves perfect specificity and sensitivity, and a definitive diagnosis of brain abscess may not be reached until surgery 29 30 31 .

In agreement with other studies, contiguous, haematogenous and unknown primary source of infection constituted approximately a third of cases, respectively 18 19 . While several textbooks have suggested that the location of the abscess may predict the source of the infection, we found that abscess location had an overall limited predictive value for the primary source of infection. For instance, we observed that cerebellar abscesses were often not associated with otic infection. Similar observations have been reported by Kowleesar et al. 14 .

Despite a relatively heterogeneous spectrum of bacterial species in this study, the predominant species remained streptococci. In agreement with Xiao 9 , the duration of antibiotic treatment before operation was not directly related to the culture positive rate. Of note, 15 patients with positive surgical cultures had received antibiotics for more than 5 days prior to surgery, implying that surgical intervention is often essential to attain full therapeutic response. Compared to other studies, the presence of anaerobes was relatively infrequent (17% of cases).

For several years in Denmark, high dose penicillin in combination with metronidazole was recommended as first-line treatment for most cases of brain abscess, since penicillin resistant streptococci and MRSA are uncommon in Denmark. In recent years, broader-spectrum agents have been increasingly used based on the good outcome of third line cephalosporins or carbapenems in small non-controlled studies and the increased rate of patients with culture negative disease or unknown source of infections 19 20 .

In our study, the efficacy of individual antibiotic choice is difficult to evaluate as changes of antibiotics were common during treatment. Recent data based on 16sRNA PCR suggest that brain abscesses contain a high number of fastidious and/or unculturable species 32 . However, we found little indication that broad spectrum treatment with cephalosporin or meropenem was superior to standard high-dose penicillin combined with metronidazole therapy in patients managed by surgery. Hence, the extent to which more detailed microbial identification could contribute to improved treatment is unknown.

Although the use of steroid treatment in brain abscess is controversial, corticosteroids may be beneficial in patients with raised intracranial pressure and potentially life-threatening complications such as impending cerebral herniation 26 33 . Our policy is to reserve corticosteroid treatment to patients with significant cerebral edema with mass effect, compromising mental or neurological status, despite maximal surgical treatment. Almost a third of patients received steroids before surgery, including a few patients initially suspected of brain tumours and more than half of surgically treated patients received at least temporary treatment with steroids after surgery. We were unable to demonstrate any clear benefit or risk associated with steroid use in agreement with previous studies 7 8 34 .

The appropriate duration of antibiotic treatment has been debated 35 36 . In the absence of solid evidence, most recommendations are vague and based on minimum duration of therapy. In a recent British study, inadequate duration (< 3 weeks) or choice of oral antibiotic therapy, after switch from intravenous therapy was reported to be associated with recurrence of abscess among eight patients 21 . Nevertheless, treatment duration as short as two to three weeks after definitive surgical intervention has been used in selected patients with success 35 37 . American textbook recommendations 38 are primarily based on a review paper by Mathisen 33 , which suggests high-dose intravenous agents for 6 to 8 weeks followed by oral antimicrobial therapy for 2 to 3 months if an appropriate therapy is available. UK guidelines recommend a minimum of 4–6 weeks of therapy if the abscess has been excised or aspirated and 6–8 weeks if treated conservatively 36 .

Most patients had a prolonged duration of antibiotic treatment. While we observed no cases of recurrence, the rationale for extended duration of therapy may be questioned. Among 18 patients, for which the duration of postsurgical antibiotic treatment was limited to less than 6 weeks, we observed no cases of recurrence. In our study, treatment duration was usually guided by regression of abscess as verified by CT or MRI and all patients had at least one or more follow-up neuroimaging studies. Paradoxically, while the use of MRI was often critical for a timely diagnosis, subsequent MRI monitoring was frequently associated with prolongation of treatment. Even with effective treatment, MRI radiologic changes such as the disappearance of contrast enhancement lag behind clinical improvement and enhancement may persist for months. In several cases, the continued presence of such signs prompted clinicians to extend the duration of therapy despite clinical improvement.

The majority (89%) of patients were treated surgically. The patients who were treated conservatively with antibiotics in our series often had multiple abscesses, poorly accessible abscesses, such as in the cerebellum, or a poor premorbid conditions. Seventy-six% of patients requiring surgical treatment had needle aspiration performed, while 24% of those requiring surgical treatment underwent a craniotomy with excision of the abscess. These proportions are similar to those reported in other series 17 27 , but are still a matter of debate and highly dependent on surgeon preference. It is likely that in our series, an even higher proportion of needle aspirations would have been sufficient, had the preoperative diagnosis of brain abscess been less equivocal, as a differential diagnosis of intracerebral tumour, warranting a full craniotomy, was initially suspected in several cases preoperatively. Additionally, in the rare cases that abscess developed as a complication to a prior craniotomy, was a craniotomy usually performed for abscess excision. Several studies suggest that needle aspiration is as effective as abscess excision in the management of the majority of intracerebral abscesses and excision can be reserved for abscesses that fail to regress despite aspiration, or that are caused by resistant pathogens 33 . In a recent retrospective literature review, the mean mortality appeared considerably lower in patients managed by aspiration compared to surgical excision, but did not provide information on measures such as need for reoperation or recurrence 23 . In our series, the need for reoperation was not surprisingly higher in patients managed by aspiration compared to excision, however, the duration of antibiotic treatment was similar.

In conclusion, despite a decline in mortality rates in the era of advanced neuroradiological studies and better antibiotics, brain abscess continue to be a serious condition with high mortality risk. The clinical signs of brain abscess are often unspecific and diagnosis is often delayed, particularly in patients initially investigated by CT compared to MRI. Key determinants of poor outcome are decreased GCS at admission, the presence of comorbidities and intraventricular rupture of brain abscess.

The authors declare that they have no financial or non-financial competing interests.

JH-L, AA, AL and JB planned the study. JH-L, AA, HR, JE and AL undertook the data extraction. JH-L was responsible for data analysis, the initial manuscript draft, correction of the article and submission. All authors contributed to the discussions and interpretation of data, and rviewed the final report.