Published: August 2010
Brain abscess is defined as purulence and inflammation in one or more localized regions within the brain parenchyma. It is one of several forms of severe intracranial infection, with other examples including subdural empyema and intracranial epidural abscess. Brain abscesses are uncommon but are life-threatening. Optimal assessment of the likely pathogenesis of the lesion and causative pathogens is essential for a favorable outcome. With the advent of routine neuroimaging and less invasive neurosurgical techniques, the mortality rate of brain abscesses has fallen, but an understanding of the complexities of this infection is crucial for management.
There are approximately 1500 to 2000 cases of brain abscess diagnosed in the United States annually, with an estimated 1 in 10,000 being hospitalized for a brain abscess. The infection tends to occur in young men, although infection can occur in all age groups; the male-to-female ratio varies between 2 : 1 and 3 : 1. In some series, children account for up to 25% of cases.
There have been several trends in the epidemiology of brain abscess over recent decades. One trend is that there appears to be a trend toward a decreasing incidence of brain abscess. In a population-based study of residents of Olmstead County, Minnesota, the incidence rate was 1.3 per 100,000 patient-years from 1935 to 1944 compared with 0.9 per 100,000 patient-years from 1965 to 1981.1 In this series, brain abscess rates were highest in children and in adults older than 60 years. Another epidemiologic trend has been the decrease in the incidence of brain abscesses as a complication of otitis media, which has been attributed to the early use of antimicrobial agents for respiratory infections. A third epidemiologic trend is the increasing number of brain abscesses occurring in immunocompromised hosts. These patients are at a higher risk of opportunistic fungal and parasitic infections of the central nervous system (CNS). Most case series still report that most brain abscesses are caused by common bacteria, but the trend for more unusual pathogens as the cause in the immunocompromised patient is likely to increase in the future. Finally, there has been a marked decrease in mortality rate associated with brain abscess. For a number of decades, the mortality associated with brain abscess remained constant, in the 40% to 50% range in the pre-antibiotic era, and did not improve substantially for a number of years during the post-antibiotic era. Antibiotics alone did not appear to be the key factor to improve survival, and neurosurgical drainage had been described since the late 1800s.2
However, after the mid-1970s, the case-fatality rate improved, with several case series reporting mortality rates below 20%. This improvement has been attributed to the routine use of brain computed tomography (CT) scanning for initial diagnosis and follow-up of brain abscesses. In one series, the mortality rate from brain abscess from 1970 to 1974, before routine CT scanning, was 36%, but fell to 0% for brain abscesses managed from 1974 to 1977, after routine CT scanning.3 This improved mortality rate was attributed to multiple factors. including earlier diagnosis, fewer cases with multiple abscesses, more accurate localization of abscesses, greater incidence of total abscess removal, and rapid detection of postoperative complications.
Brain abscesses tend to occur in immunocompetent patients in the general population, with most being polymicrobial with aerobic and anaerobic bacteria. The major predisposing factors in this group include contiguous focus of infection, such as from paranasal sinusitis; hematogenous spread from an extracranial site of infection, such as infective endocarditis or lung abscess; recent neurosurgical procedure; and penetrating head trauma. Brain abscess from a contiguous focus of infection accounts for about 50% of total cases. The epidemiology varies with the location of the primary focus; for example, with sinusitis, older children and young adults are most commonly affected but, with otogenic brain abscesses, young children and older adults are usually involved. There may be a trend away from this particular risk factor, with one study of 100 brain abscesses reporting only 21% associated with ear infections or paranasal sinusitis.4 In this European series, postneurosurgical cases accounted for 28% of cases. The widespread and early use of antibiotics for otitis and sinusitis has been cited as one reason for this possible shift in epidemiology.
An important risk factor for the development of a brain abscess is an acute or chronic immunosuppressed state, particularly patients who are solid organ transplant recipients, bone marrow transplant recipients, or persons with the acquired immunodeficiency syndrome (AIDS). From epidemiologic and clinical perspectives, brain abscesses that develop in these groups should be considered separately from healthy persons with brain abscesses from traditional risk factors. In persons with AIDS, Toxoplasma encephalitis has been the most common focal brain disorder. The annual incidence in the 1980s through the early 1990s was about 0.7 per 100 person-years. Risk factors for CNS toxoplasmosis included low CD4+ cell counts, prior opportunistic infections, injection drug use, and lack of prophylaxis. The annual incidence of toxoplasmosis fell between 1992 and 1996, likely because of the routine use of prophylactic regimens. A further decrease in the annual incidence was seen after the introduction of highly active antiretroviral therapy (HAART) in the mid-1990s. However, in a report from Italy, Toxoplasma encephalitis remained a highly prevalent disorder, even during the late HAART era.5
In transplant recipients, opportunistic mold or yeast infections predominate and can originate from primary infection or from reactivation of latent infection. In a large retrospective study from Pittsburgh, brain abscesses occurred in 0.61% of solid organ transplant recipients.6 There appeared to be two distinct populations of patients. The first population consisted of acutely immunosuppressed transplant recipients at risk for a number of fungal infections, including fungal brain abscesses (mainly Candida and Aspergillus spp.). The lung was the primary site of fungal dissemination for most of these patients. Important factors in this high-risk group included antirejection therapy, need for retransplantation, recent bacteremia or viremia, and multiorgan failure. The second population was made up of transplant recipients who were generally healthy and developed a brain abscess long after transplantation. These late-presentation abscesses were usually of nonfungal origin, including nocardiosis and toxoplasmosis.
Fungal brain abscesses have also been reported in hematopoietic stem cell transplant recipients, despite the routine use of antifungal prophylaxis. At the Fred Hutchinson Cancer Research Center in Seattle, brain abscesses were diagnosed in 58 patients undergoing marrow transplantation between 1984 and 1992.7 Fungi were isolated in 92% of cases, with Aspergillus spp. being the most common isolate, followed by Candida spp. Concomitant pulmonary aspergillosis was common. Risk factors for Candida brain abscess were prior candidemia and prolonged neutropenia. The incidence of brain abscess in bone marrow transplant recipients overall has been estimated at 1 in 49 persons.
In a substantial number of cases, the cause of brain abscess is unknown. The percentage of cases of brain abscess in which no primary focus of infection can be identified ranges from 10% to over 60%, although in recent case series the rates of idiopathic brain abscess are in the lower range (≤15%). Direct spread of infection from a site contiguous to the CNS remains the most common route of infection in most case series, comprising about 50% of brain abscess cases. Historically, the common primary sites have been sinusitis, otitis, and dental abscesses. Brain abscess has been described as a complication of infection involving any of the paranasal sinuses, with brain abscesses located mainly in the frontal lobe (frontal and ethmoidal sinusitis), temporal lobe (sphenoid sinusitis), or both, depending on the sinuses involved. Sphenoid sinusitis is particularly problematic, because it carries a higher rate of CNS complications compared with the other paranasal sinuses; it can spread to several contiguous areas, including the temporal lobe, pituitary, cavernous sinus, and sometimes the frontal lobe.
When there is spread of infection from the mastoids or middle ear to the CNS, the cerebellum temporal lobes, or both, are most often involved.8 The most common predisposing condition is chronic otitis media. Several case series have reported involvement of the cerebellum four times more frequently than temporal lobes,9 but others have not.10 Cholesteatoma is seen in 40% to 100% of cases.8,11 Direct spread from a dental source can account for up to 10% of brain abscesses, although this varies considerably in case series and is highest in developing countries. Dental infection or dental procedures, or both, can also cause brain abscess via a hematogenous route, rather than by direct extension. Brain abscess from direct spread from a dental abscess, often a molar, usually involves the frontal lobe.
Several mechanisms have been proposed to explain the spread from a contiguous focus of infection to the brain: extension from the primary focus to the adjacent bone and then to the brain (e.g., frontal sinusitis complicated by osteomyelitis of the posterior wall of the sinus); spread of local infection to the brain via emissary veins of the skull—bacteria or fungi in the sinuses can spread retrograde from the vasculature of the sinus mucosa, through the extensive valveless network of veins and venous sinuses, and into the cerebral venous system; spread via lymphatics; spread via the internal auditory canal or cochlear structures (for otitis media); and deep inoculation into the brain, such as with neurosurgery.
It should be noted that brain abscess rarely results from meningitis. The finding of a possible brain abscess on neuroimaging should not automatically prompt a lumbar puncture to rule out meningitis, because meningitis is a rare cause of brain abscess. This is particularly true in adults, for whom large series of community-acquired bacterial meningitis have reported no cases of secondary brain abscess. One exception is meningitis caused by Listeria monocytogenes, in which small abscesses in the brainstem (rhomboencephalitis) have occasionally been described. In the newborn, some forms of gram-negative meningitis are often complicated by brain abscess; about 70% of neonates with Citrobacter diversus meningitis will develop secondary brain abscesses. Conversely, a brain abscess can occasionally rupture into a cerebral ventricle, leading to sudden deterioration and a meningitis-like picture. This is a rare event, with about 140 cases reported in the literature, but with a lethality rate of more than 80%.12
Hematogenous spread of infection from a distant source to the CNS represents the second most commonly identified mechanism. This route of infection accounts for about 25% of brain abscesses. However, it should be noted that the vast majority of bacteremic events do not result in brain abscess or any other CNS infections. When bacteremia does result in brain abscess, there is usually an additional predisposing factor. Historically, pulmonary infections were the most common conditions associated with hematogenous brain abscess. The most common pulmonary conditions are lung abscess, empyema, and bronchiectasis. Pulmonary venous malformation is another underlying condition associated with a relatively high rate of brain abscess. Patients with Osler-Weber-Rendu disease, hereditary hemorrhagic telangiectasia, are known to be at higher risk of brain abscess,13 presumably because bacteremia may result in small infected emboli that travel through the pulmonary arteriovenous malformations and enter the arterial circulation of the brain directly, without first being filtered by the pulmonary vasculature. Cyanotic congenital heart disease is also considered a traditional risk factor for brain abscess because of the increased risk of spread of a blood-borne pathogen to the brain.
Less commonly, brain abscess has also been reported with concomitant liver abscess. Most of these reported cases represent underlying pyogenic liver abscess, with hematogenous spread to the brain. One emerging trend is simultaneous liver and brain abscesses caused by Klebsiella pneumoniae in diabetics. This particular association has been described in several series from Taiwan, in which K. pneumoniae liver abscess was found to be a leading cause of community-acquired brain abscess.14,15 Brain abscess has also been described with other gastrointestinal conditions and procedures, such as following sclerosis of esophageal varices. It has also been described as a complication of tongue piercing.16
Experimental studies of brain abscesses have led to a four-stage model of disease development.17-20 Direct inoculation of bacteria into brain parenchyma results in focal inflammation and edema. This has been termed the first stage, or early cerebritis stage, and develops in the first 1 to 3 days after inoculation. Typically, there is neutrophil accumulation, edema, and some tissue necrosis. Astrocytes and microglia are activated early on, and this activation persists afterward. The area of cerebritis expands and a necrotic center develops, termed late cerebritis, on days 4 to 9. Macrophages and lymphocytes predominate in the infiltrate. The third stage is characterized by the development of a capsule that is vascularized and ring enhancing on CT scan, days 10 to 14, the early capsule stage. In the fourth stage, the host immune response causes the capsule to wall off, and there is destruction of some surrounding healthy brain tissue in an attempt to sequester the infection. This model can vary considerably based on factors such as the size of the bacterial inoculum, microorganism, immune state of the host, and antimicrobial agent. This model was derived primarily from experimental rodent studies but seems to correlate well with human disease. In both animal modeling and human disease, the key requirements for the formation of a pyogenic brain abscess are a focus of necrotic or ischemic brain tissue, combined with a virulent microorganism.
Animal studies examining the neuroimmunology of brain abscess have shown that the local immune response may contribute to parenchymal brain destruction.19 In animal and human brain abscess studies, the area of inflammatory response and tissue destruction often appears significantly larger than the more localized area of the initial bacterial infection. In a Staphylococcus aureus brain abscess model, bacteria are recognized by Toll-like receptor 2, which leads to activation of astrocytes and microglia, which in turn results in production of proinflammatory cytokines.20 Cytokine release causes disruption of the blood-brain barrier, with subsequent entry of large molecules into brain parenchyma, such as immunoglobulin G (IgG) and albumin. Adhesion molecules are expressed, such as the intracellular adhesion molecule (ICAM), which facilitate entry of neutrophils, T cells, and macrophages from the circulation into the growing abscess. These immune cells are activated by bacteria and cytokines, leading to a repetitive cycle of inflammation.
The pathogens involved in pyogenic brain abscess often can be predicted by the predisposing condition, host immune status, and mode of acquisition. When the predisposing condition is a contiguous focus of infection, such as paranasal sinusitis, otitis media, or dental abscess infection, it is usually polymicrobial (>50%-60% of cases), particularly if culture specimens from the brain abscess are submitted optimally on anaerobic media. The usual pathogens reflect the oral and upper respiratory flora, in which anaerobes are an important component of the normal flora: aerobic and anaerobic streptococci, Fusobacterium spp., and anaerobic gram-negative bacilli (e.g., Bacteroides, Prevotella). The paranasal sinuses can also be colonized and/or infected with other bacteria, such as S. aureus and Enterobacter spp., but this is less common than the normal flora.
Following neurosurgery or head trauma, such as a gunshot wound, the likely pathogens for brain abscess are S. aureus, coagulase-negative staphylococci, Pseudomonas aeruginosa, Enterobacter spp., and some streptococci. Clostridium spp. can be seen following contaminated penetrating head trauma.
The spectrum of organisms varies somewhat when the associated condition is a chronic pulmonary infection, such as lung abscess, bronchiectasis, or chronic empyema. Common organisms are Fusobacterium spp., anaerobic gram-negative bacilli such as Bacteroides spp. (not B. fragilis), various aerobic and anaerobic streptococci, and some Actinomyces and Nocardia spp. For brain abscesses occurring in the setting of bacterial endocarditis or congenital heart disease, the most common organisms are viridans streptococci, including the Streptococcus milleri group (S. constellatus, S. anginosus, S. intermedius, which have a marked tendency for abscess formation), S. aureus, Haemophilus spp., and occasionally enterococci.
The frequency of these causes of brain abscess varies considerably depending on geography, but usually accounts for only a small percentage of cases in most series, including developing countries. Streptococcus pneumoniae is the most common cause of community-acquired bacterial meningitis, but is a rare cause of brain abscess.21
In persons with AIDS, brain abscesses are more likely to be from opportunistic infections, although bacterial brain abscesses have been reported in a small percentage of cases. Common pathogens in AIDS-related brain abscess include Toxoplasma gondii, Listeria monocytogenes, and Cryptococcus neoformans. Toxoplasma encephalitis is the most common cause of a focal intracranial lesion in AIDS, even in the current era of HAART. In one study from Italy, toxoplasmosis remained the most prevalent cause of HIV-related neurologic disorders.5 However, the differential diagnosis of one or more focal cerebral lesions in a person with AIDS is wide. The most common causes are T. gondii (about 50%), but this is followed by CNS lymphoma (about 20%-30%).
Other pathogens account for the remainder. Rhodococcus equi, an intracellular gram-positive coccobacillus, is a known but rare cause of brain abscess in those with AIDS.22 This zoonotic infection typically causes a chronic cavitary or nodular pulmonary infection and brain abscess has been reported, usually in the setting of a current or prior Rhodococcus pulmonary infection. Mycobacterium tuberculosis is recognized as a cause of brain abscesses in AIDS patients,23,24 and has been reported as an isolated ring-enhancing lesion or multiple lesions in the setting of widely disseminated mycobacterial infection. Tuberculosis can infect the CNS in a number of ways, including tubercular meningitis, tuberculous encephalopathy, tuberculous arteritis, tuberculoma, and tuberculous abscess. A true tuberculous abscess is a rare manifestation of CNS tuberculosis, even in countries in which tuberculosis is common. Other causes of focal mass lesions in AIDS include C. neoformans, H. capsulatum, A. fumigatus, and others.
In solid organ transplant recipients, the major organisms of concern are fungal (Aspergillus, Mucorales, Candida, Cryptococcus) and Nocardia spp. Occasionally, other opportunistic infections may be seen, such as Toxoplasma gondii,25-29 as well as polymicrobial bacterial brain abscess. In the neutropenic patient, opportunistic mold infection is the most common cause of brain abscess, particularly Aspergillus and Mucorales spp. Additional pathogens are Pseudomonas aeruginosa and Enterobacter spp.
In general, the clinical and radiographic appearances of brain abscess in the immunocompromised host are nonspecific. Microbiologic confirmation from brain tissue is necessary to establish a causative diagnosis in the vast majority of cases.
The clinical presentation depends on a number of factors, including the host’s immune status, specific pathogen involved, contiguous sites of infection or distant foci of infection, size of the abscess, and location of the lesions. The most common symptom is a headache (about 70% of cases), although the headache tends to be nonspecific and not always localizing. In general, the headaches are dull and subacute or chronic in nature, and not sudden and severe. One exception to this is rupture of a preexisting brain abscess into an adjacent ventricle, which manifests with an acute illness more suggestive of meningitis. Patients may also present with symptoms of increased intracranial pressure such as nausea, vomiting, and advancing lethargy. The classic triad of fever, headache, and altered mental status is unfortunately not reliable and is present in 50% or fewer cases. It is important to note that fever is absent in over 50% of patients with brain abscess at the time of the initial presentation. This is also true for persons with AIDS with CNS toxoplasmosis; fever is not reliably present. Similarly, focal neurologic deficits such as hemiparesis and aphasia are absent in 50% of patients with bacterial brain abscess. Papilledema is also an insensitive sign, being absent in about 75% of cases on presentation. Other clinical manifestations include generalized seizures and neck stiffness if the abscess is close to the meninges.
Certain focal neurologic abnormalities may provide a clue to the location of a space-occupying lesion. A sizable abscess in the frontal lobe may lead to hemiparesis and motor speech deficits. A lesion in the temporal lobe may lead to an ipsilateral headache, aphasia, and possibly a visual field defect. Lesions in the cerebellum may cause ataxia, dysmetria, and nystagmus. However, these relationships are not dependable and, as previously noted, patients with brain abscesses often have no localizing signs on presentation. Thus, the bedside diagnosis of a brain abscess can be challenging, with a number of patients presenting without fever or focal deficits.
Routine laboratory studies are not helpful for the diagnosis of brain abscess. Leukocytosis may be absent; in some series about 40% of patients have a normal peripheral white blood cell count. Acute-phase reactants are moderately helpful but nonspecific. The C-reactive protein level is elevated in almost all patients, but the sedimentation rate can be only moderately elevated and sometimes is normal. Samples for blood cultures should be obtained in all suspected cases; although the yield is low, a positive result can be extremely valuable.
In immunocompromised patients, several tests may be useful. The tuberculin skin test is an often overlooked screening test that should be administered to an immunosuppressed patient with a brain or lung lesion. The limitations of the tuberculin skin test are false-negative reactions caused by cutaneous anergy from chronic steroids, other medications, and often the primary infection itself. The Toxoplasma IgG level is of some potential use in assessing the AIDS patient with afocal CNS lesions. The seroprevalence rate of a positive Toxoplasma IgG level is high in the general population, with considerable geographic and international variations, and a positive serum level in isolation is not diagnostic of active toxoplasmosis. The Toxoplasma IgG serum level is used to support a presumptive diagnosis of toxoplasmosis when there are also compatible clinical and radiographic features in an individual case. A negative serum Toxoplasma IgG level is suggestive of diagnoses other than Toxoplasma encephalitis, but does not rule out this diagnosis completely. AIDS patients with a negative serum toxoplasmosis IgG level in the setting of Toxoplasma encephalitis have been reported. A negative serum IgG test result may place the diagnosis of toxoplasmosis lower in the differential diagnostic considerations but does not exclude the diagnosis entirely.
Lumbar puncture is often contraindicated in persons with suspected brain abscess. Cerebrospinal fluid (CSF) results, when reported in the literature, have shown nonspecific abnormalities in most cases, usually with mild elevations of the CSF protein level, and varying numbers of CSF leukocytes. The yield of a pathogen from CSF examination with suspected brain abscess is low, less than 10%. There may be an occasional patient in whom the abscesses are small on CT or magnetic resonance imaging (MRI), the lesions exert minimal mass effect, the third and fourth ventricles are open, and there are no other signs of cerebral edema In these cases, lumbar puncture is sometimes considered mainly to obtain CSF for cytologic and flow cytometry studies (to rule out metastases), cryptococcal antigen detection, and polymerase chain reaction (PCR) assay for T. gondii. However, this should be undertaken with great caution. Given the risks of brain herniation in the setting of elevated intracranial pressure, and the low yield of diagnostic tests, lumbar puncture should be avoided in the vast majority of cases.
Most patients with a brain abscess will undergo some type of CT-guided aspiration or open evacuation of the abscess. Operative specimens should be routinely submitted for Gram staining, routine culture, and anaerobe culture. Purulent fluid should usually be submitted in a sterile container, in addition to the use of standard Culturette swabs, because the microbiologic yield of larger tissue samples and fluids is significantly higher than the yield of the small specimen submitted for swab cultures. In addition, some microbiology laboratories do not accept specimens submitted on a swab culture for certain organisms, especially acid-fast bacillus (AFB). Because of the important role of anaerobes in brain abscesses, special effort should also be made to optimize recovery of anaerobes. The local microbiology laboratory should be consulted beforehand regarding the preferred specimen collection technique for anaerobes, which generally requires inoculation of specimens directly into anaerobic bottles or rapid transport of a specimen from the operating room to the laboratory using appropriate transport systems. Other stains and cultures include acid-fast staining with mycobacterial culture, fungal stains and cultures, Nocardia stain and culture, and culture for Rhodococcus equi. Samples for these special stains and cultures should be obtained routinely when the patient is immunocompromised, but should be strongly considered in some immunocompetent patients as well.
PCR molecular testing is available through many reference laboratories for various pathogens from CSF and tissue specimens, particularly Mycobacterium tuberculosis and T. gondii. It should be emphasized that a number of molecular tests that are potentially useful for the diagnosis of meningitis in the AIDS patient or transplant recipient are not useful for the diagnosis of brain abscess in these patients. For example, cytomegalovirus and varicella zoster virus are known causes of meningoencephalitis in AIDS patients, but do not cause brain abscesses. In patients from whom CSF or tissue may have been obtained to diagnose Toxoplasma brain abscesses, several PCR-based tests have been developed, with some commercially available assays able to detect as few as 100 copies of DNA per mL (or down to three oocysts). Commercially available PCR tests for M. tuberculosis have a lower limit of detection of 400 cells/mL in CSF and tissue. Many tertiary centers have developed in-house PCR detection systems for these and other unusual organisms. Standardization of the numerous PCR assays reported in the literature, and also the many variations of PCR assays in local laboratories, may have had only in-house validation. Many lack rigorous standardization and validation for multiple specimens, and thus the false-negative and false-positive results for a number of these tests remain unclear. Generally, the decision to order multiple PCR assays on blood, CSF, and brain tissue should be done judiciously, avoiding a shotgun approach when ordering molecular diagnostics, even for an immunocompromised host.
Specimens should also be submitted routinely for cytopathologic and histopathologic testing for special organism stains and also to rule out malignancy. Touch preparations from tissue specimens may be useful for detecting T. gondii rapidly. Special stains should be used for pathology specimens, including silver stains for fungi, AFB stains for mycobacteria, and mucicarmine stain for Cryptococcus neoformans in select cases. Tissue biopsy will also be useful in diagnosing the rare case of parasitic brain abscess other than T. gondii, such as those caused by Entamoeba histolytica (extraintestinal amebiasis).
Because of the difficulties in clinical diagnosis of brain abscess, neuroimaging is essential.30-37 In general, radiographic abnormalities depend on the particular stage of the brain abscess. One of the uses of neuroimaging is to estimate the age of the brain abscess. A lesion in the early or late cerebritis stage may be managed somewhat differently compared with a mature, walled-off abscess (see later, “Treatment”). CT scanning with contrast during stage 1 (early cerebritis) may show only edema—an area of hypodensity—which may or may not enhance with contrast. If done very early in the course of infection, a contrast-enhanced CT may be normal. During later stages, there is the development of a space-occupying lesion with a hypodense center and later a ring-enhancing rim, which is often surrounded by a large area of edema. Occasionally, ring enhancement can be seen with late cerebritis; delayed administration of contrast that fills in the central hypodensity is suggestive that the lesion is still in the cerebritis stage. Although contrast-enhanced CT scanning is considered sensitive for the detection of brain abscesses, it is not specific. Brain abscesses tend to have smooth thin-walled capsules, whereas tumors tend to have more irregular capsules. There are additional characteristics of brain tumors, but some overlap exists with brain abscesses. It is important to note that brain abscesses and brain tumors may have an identical appearance on the CT scan.
MR imaging is more sensitive than CT, and MRI can usually detect infection in the early cerebritis stage. The MRI scan in focal cerebritis usually shows an area of hypointensity on T1- and T2-weighted imaging. The characteristic appearance of a mature brain abscess on MRI is a focal lesion with low intensity on a T1-weighted image, with a smoothly marginated capsule that enhances with IV gadolinium (Fig. 1). On T2-weighted images, the central abscess is hyperintense and the surrounding capsule is hypointense. There is extensive surrounding edema in most cases. The finding of a capsule that is hypointense on a T2-weighted image and mildly hyperintense on a T1-weighted image is suggestive of an abscess capsule.
The finding of multiple cerebral abscesses conforming to a vascular distribution is suggestive of a central embolic source. Brain abscesses from a hematogenous source, such as infective endocarditis, tend to be multiple and distribute widely in the middle cerebral artery. Multiple cerebral lesions are present in about 50% of cases of hematogenously acquired brain abscess.38 However, a single brain abscess can also be seen with endocarditis.
In immunocompromised persons, it is not possible to distinguish a bacterial brain abscess from an opportunistic CNS infection on the basis of conventional MR imaging alone. With Toxoplasma encephalitis, MRI typically shows multiple, small, ring-enhancing lesions. Any region of the brain may be involved, but the basal ganglia is the most common site. There is frequently some degree of local edema and mass effect as well, which are helpful in distinguishing Toxoplasma encephalitis from other brain lesions in AIDS, which lack mass effect (e.g., progressive multifocal leukoencephalopathy, cytomegalovirus encephalitis). CNS lymphoma in AIDS patients also manifests with a ring-enhancing lesion, but tends to be solitary rather than multiple, although multifocal CNS lymphoma does occur. Given the difficulty in distinguishing toxoplasmosis from CNS lymphoma, a number of additional radiographic techniques have been studied, including thallium single-photon emission computed tomography (SPECT) and positron emission tomography. These tests can be useful adjunctive studies in difficult cases.
For transplant recipients, diffusion-weighted imaging in fungal brain abscesses has been reported in one small study of six patients with proven fungal (Aspergillus) cerebral abscesses.39 Conventional MR imaging showed typical ring-enhancing lesions indistinguishable from pyogenic brain abscesses. Most of the patients had restricted diffusion in the center of the abscess, similar to prior reports of diffusion-weighted imaging (DWI) in bacterial brain abscesses. The DWI hyperintensity (restriction) was attributed to cellular infiltration and proteinaceous fluid in the abscess. There were several additional patients who presented with Aspergillus and Rhizopus cerebritis; these individuals had a fulminant course and expired rapidly. Conventional MRI showed large, nonenhancing cerebral lesions, which did not show ring enhancement. On DWI, the lesions showed a more heterogeneous pattern with some areas of decreased and increased diffusion. Pathology demonstrated cerebritis rather than a purulent abscess. In transplant patients presenting with neurologic changes and ring-enhancing brain lesions, a fungal cause should be high in the differential diagnosis; conventional MRI and DWI abnormalities appear to be similar to those of bacterial brain abscesses.
The optimal treatment of brain abscess is complex and requires a coordinated approach among multiple teams. There are no large, randomized, clinical trials comparing different antimicrobial regimens or different neurosurgical approaches, but large retrospective series have demonstrated dramatic improvement in survival using a combination of surgical drainage and a prolonged course of IV antibiotics. In most patients, particularly those with abscesses larger than 2.5 cm, stereotactic aspiration or open brain biopsy with evacuation should be strongly considered for microbiologic diagnosis, given the wide range of possible pathogens and the prolonged course of IV antibiotics required. One approach has been to treat neurologically stable patients with early infections in the cerebritis stage with IV antibiotics alone, with aspiration attempted when the infection later appears more encapsulated and liquefied, to maximize the yield of an aspiration procedure. Alternatively, an area of focal cerebritis without a frank abscess may still be aspirated using a stereotactically guided needle procedure through a burr hole, but the volume of fluid and tissue may be small. Large abscesses (>2.5 cm) should generally be aspirated, drained, or completely excised.
For patients with multiple abscesses of varying sizes, one approach would be to aspirate and/or evacuate large collections, and treat smaller abscesses medically. Serial CT scans need to be obtained afterward to assess the stability of the remaining abscesses, because repeat aspirations may be necessary. Patients with multiple small (<2 cm) abscesses may respond to medical therapy alone, particularly if the lesions are smaller than 1 cm, but these patients also need serial neuroimaging. Medical treatment is usually attempted when abscesses are also in deep locations, such as the brainstem. However, a medical approach carries a significant risk of selection of incorrect antibiotics. This is particularly problematic for the immunocompromised patient, in whom the spectrum of pathogens is even more unpredictable. The diagnostic algorithm is somewhat different for immunocompromised patients with suspected cerebral toxoplasmosis (see later).
In general, the antibiotics used for the treatment of brain abscess should have the following characteristics: favorable brain or CSF drug levels, or both, even in the absence of inflamed meninges; bactericidal activity; minimal side effects over many weeks of treatment; good penetration and activity in the abscess fluid; and good in vitro activity against the most common pathogens, based on the most likely pathogenesis of the infection.
Penicillin G has excellent activity against most aerobic and anaerobic streptococci found in brain abscesses associated with paranasal sinusitis, otitis media, mastoiditis, and dental abscesses. Penicillin G also penetrates relatively well into the CSF and brain tissue, and maintains activity in abscess fluid. Most isolates of Prevotella, Fusobacterium, and Porphyromonas species are susceptible to penicillin G, but there are some resistant clinical strains. Penicillin G is also not reliably active against most anaerobic gram-negative bacilli. Thus, penicillin G is not used by itself for the initial empirical coverage of brain abscesses, and is usually used in combination with a second agent with broader anaerobic coverage, such as metronidazole.
Third-generation cephalosporins, such as ceftriaxone, also achieve effective levels in various compartments of the CNS and are highly active against aerobic streptococci and most gram-negative bacilli, but lack anaerobic activity and generally need to be used in combination with an agent such as metronidazole, because many brain abscesses include anaerobes. Metronidazole has many favorable characteristics for the treatment of brain abscess. It is reliably active against gram-positive and gram-negative anaerobes, retains activity in abscess cavities, and crosses the blood-brain barrier, even when the meninges are not inflamed. Metronidazole has no activity against aerobes, including aerobic streptococci, and thus needs to be used in combination with other agents. The beta-lactam–beta-lactamase inhibitor drugs (e.g., ampicillin-sulbactam, ticarcillin-clavulanate, piperacillin-tazobactam) are effective against most organisms that cause contiguous and hematogenous brain abscess. These agents cross the blood-brain barrier and can be used as monotherapy in many cases.
For the treatment of S. aureus, antistaphylococcal penicillins such as oxacillin are the drugs of choice for susceptible isolates. The semisynthetic penicillins are rapidly bactericidal for S. aureus and penetrate into the CNS, although not as well as penicillin G. However, many hospital-associated isolates of S. aureus are methicillin-resistant and antimicrobial susceptibility tests are not available for several days after brain tissue is submitted for culture. The most experience in treating methicillin-resistant S. aureus (MRSA) brain abscesses is with vancomycin, which is also active against aerobic and anaerobic streptococci and Clostridia spp. The penetration of vancomycin into the CNS is limited with noninflamed meninges. Ceftriaxone has the best activity against S. aureus compared with other third-generation cephalosporins. However, its role in the treatment of life-threatening S. aureus infections is not clear, and agents with better in vitro activity and clinical use are available.
The carbapenems (e.g., imipenem, meropenem, ertapenem) have the advantage of excellent anaerobic coverage, along with activity against aerobic streptococci and Enterobacter spp. Imipenem has good penetration into CSF and brain pus. Anecdotal reports have suggested that imipenem may be effective for bacterial brain abscess, but concerns have been raised regarding seizures induced by the carbapenems, particularly imipenem. In one series of 15 patients, imipenem was used successfully to treat brain abscesses from various gram-positive and gram-negative organisms, with one seizure.40 In addition, imipenem has been used for the treatment of Nocardia brain abscesses without undue reports of seizures. Carbapenems do not appear contraindicated in the setting of brain abscess, but their use should be weighed against the small seizure risk and the risks and benefits of alternative regimens. Quinolones (e.g., ciprofloxacin, levofloxacin) penetrate well into the CNS but experience with this class of antibiotics in CNS infections is still limited. Ciprofloxacin has unreliable activity against a number of gram-positive infections, including S. pneumoniae and other streptococcal spp. In general, the quinolones should be reserved for special cases, such as for those with multiple drug allergies.
The decision to treat empirically with antibiotics before a stereotactic or open brain biopsy is not always straightforward and can depend on a number of factors, such as the severity of the illness, presence of mass effect or impending herniation, timing of surgery, and comorbid medical conditions. Administration of empirical antibiotics for several days before a brain biopsy probably does decrease the yield of pathogens in culture, but the exact impact is not known. In practice, a number of patients will receive empirical antibiotics before microbiologic culture, and every effort should be made to submit cultures on appropriate media, including anaerobic cultures at the time of surgery.
For the empirical treatment of brain abscess associated with paranasal sinusitis, otitis, or mastoiditis, one preferred regimen is a third-generation cephalosporin (e.g., ceftriaxone, ceftizoxime) combined with IV metronidazole.41-45 This regimen covers the common pathogens in this setting but has no activity against MRSA and suboptimal activity against methicillin-susceptible S. aureus. However, S. aureus is not a pathogen associated with otitis media and dental abscesses, and remains relatively uncommon with brain abscess secondary to paranasal sinusitis. The addition of antistaphylococcal therapy in this setting is optional but may be considered in patients at higher risk for S. aureus sinusitis, such as those with recent endoscopic sinus surgery, chronic sinusitis, and recent nasal packing.
Brain abscesses from dental infections or dental procedures can be treated with IV penicillin in combination with metronidazole. Aerobic gram-negative bacilli are not important causes of odontogenic infection and do not require empirical coverage with a third-generation cephalosporin.
S. aureus is a major pathogen when brain abscess is associated with trauma or neurosurgical procedures, and must be covered empirically in this situation. Gram-negative bacilli, including Pseudomonas aeruginosa, are also common nosocomial pathogens in neurosurgical infections and should be treated pending culture results. A common empirical regimen in this setting would be vancomycin, or an antistaphylococcal penicillin, in combination with an antipseudomonal third- or fourth-generation cephalosporin, such as ceftazidime or cefepime. Brain abscess associated with lung abscess or bronchiectasis can be empirically treated with a third-generation cephalosporin in combination with metronidazole or with IV penicillin G in combination with metronidazole.
When brain abscess is associated with congenital heart disease, a third-generation cephalosporin covers the usual associated pathogens, Streptococcus and Haemophilus spp. This regimen does not cover enterococci, which is an occasional pathogen in this setting; IV ampicillin may be added to the third-generation cephalosporin in some cases pending microbiologic information. The empirical treatment of brain abscess associated with endocarditis usually is determined by the treatment of the underlying endocarditis, and depends on whether the patient has native valve or prosthetic valve endocarditis. Empirical regimens include the following: for native valve endocarditis, IV ampicillin and gentamicin; and for prosthetic valve endocarditis, IV vancomycin and gentamicin, with or without rifampin.
The length of therapy for the treatment of bacterial brain abscesses in all these scenarios has not been formally studied. Large series have generally reported good outcomes with prolonged courses of IV antibiotics for 4 to 6 weeks, or longer. One group described cures using an early switch to oral antibiotic therapy in a small number of patients who had refused IV antibiotics.46
Aspergillus fumigatus brain abscess usually occurs in the setting of disseminated Aspergillus, in which there is often evidence of pulmonary and other organ involvement. CNS aspergillosis is often rapidly fatal, even with surgical intervention. Several clinicians have reported a good clinical response to voriconazole for fungal brain abscesses,47-50 but there have been no comparative trials to date between voriconazole and amphotericin. Animal modeling has shown that voriconazole has high penetration into brain tissue and CSF, with steady-state drug levels in the CNS twice those of plasma levels, and limited human data have suggested favorable pharmacokinetics.51 Various regimens have been described at a case report level for successful salvage therapy for CNS, including amphotericin B and voriconazole, voriconazole and caspofungin, intrathecal amphotericin administered via an Ommaya reservoir, and others. Overall, the data are sparse and do not permit a specific recommendation for salvage therapy, although a number of options exist.
Toxoplasma encephalitis in AIDS patients is generally treated empirically when there is a compatible presentation (advanced AIDS, positive Toxoplasma IgG serum, lack of prophylaxis, multiple ring-enhancing basal ganglia lesions). The regimens of choice are sulfadiazine 6 to 8 g orally, in four divided doses, plus pyrimethamine, 200-mg loading dose orally, followed by 75 mg/day. An alternative regimen is clindamycin, 600 to 1200 mg IV, four times daily (or oral equivalent) with pyrimethamine. Because of the rapid radiographic improvement when Toxoplasma encephalitis is treated correctly, serial CT scans should demonstrate improvement within several weeks.