CHAPTER 40 Postoperative Infections of the Head and Brain
Before Lister’s 1867 introduction of surgical antisepsis, nearly 80% of operations were followed by infections at the surgical site and almost half of the patients died after operation.1 Despite considerable advances in our understanding of the pathogenesis of surgical infection, the introduction of rigorous aseptic practices within the operating room, and the use of prophylactic antibiotics for clean operations, infection after neurosurgical intervention remains an all too frequent occurrence. Although mortality rates have decreased markedly, postcraniotomy infections commonly require prolonged antibiotic treatment and additional surgical interventions for successful eradication and frequently result in significant morbidity, prolonged hospitalization, and increased health care expenses. The economic burden of postoperative infections is significant: the estimated average cost of a surgical site infection (SSI) attributable to methicillin-resistant Staphylococcus aureus (MRSA) is almost $100,000,2 and the overall cost of SSIs is believed to account for up to $10 billion annually in health care expenditures.3
Epidemiology and Etiology
Postoperative infections are typically categorized according to anatomic site. The Centers for Disease Control and Prevention (CDC) defines superficial incisional infections as those limited to the skin and subcutaneous tissue, whereas deep incisional infections may involve the subgaleal space and bone flap. Deep organ space infections include subdural empyema, brain abscess, and meningitis/ventriculitis. According to data from the CDC’s National Nosocomial Infection Surveillance (NNIS) program, superficial infections are responsible for 60% of SSIs after craniotomy. Meningitis is the most common deep organ space infection and represents 22% of postcraniotomy infections, whereas other intracranial infections, including subdural empyema and brain abscess, account for 14% of infections.4 Estimating the infection rate after craniotomy from the neurosurgical literature is difficult because of differences in definitions and methodology. Several large prospective studies have reported infection rates ranging from 1% to 8%.4–10 McClelland and Hall reviewed the postoperative courses of 1587 patients who underwent elective cranial operations over a 15-year period performed by a single surgeon and found an impressively low rate (0.8%) of postoperative infection.7
In accord with other studies,11,12 McClelland and Hall identified S. aureus as the causative agent for approximately half of the infections that develop after craniotomy.7 Data from the NNIS also demonstrated S. aureus to be the most common pathogen after craniotomy, followed by coagulase-negative staphylococci. Other bacteria frequently causing postcraniotomy infection included enterococci, Streptococcus spp., Pseudomonas aeruginosa, Acinetobacter spp., Citrobacter spp., Enterobacter spp., Klebsiella pneumoniae, Escherichia coli, miscellaneous other gram-negative bacilli, and yeast; each of these organisms accounted for less than 10% of episodes.13,14 Although direct spread from contiguous areas of infection is common, the causative agents tend to vary depending on the site of infection. Yang and colleagues retrospectively identified 31 patients with brain abscesses after neurosurgical procedures and found gram-negative bacilli and polymicrobial infections to be the most frequent pathogens isolated.15 Gram-negative bacilli are also the most common cause of postoperative meningitis and account for 29% to 38% of nosocomial episodes.16,17 Isolation of Propionibacterium acnes, an anaerobic gram-positive bacillus, from neurosurgical specimens has been dismissed as a contaminant because it is commensal scalp flora. However, the role of P. acnes as a causative agent of postcraniotomy infections is increasingly being recognized. Earlier reports probably underestimated its pathologic role because of its often indolent clinical manifestation, as well as difficulties associated with microbiologic isolation of the organism, specifically the need for anaerobic culture held for 10 days.18
Risk Factors for Infection and Preventive Strategies
Multiple factors combine to affect the risk for development of an SSI after craniotomy. Although it is unlikely that all postoperative infections can be completely prevented, many of the factors influencing the development of infection may be modifiable, including those attributable to the patient and those related to the surgical intervention itself. The majority of postsurgical infections are due to contamination of the wound with bacteria from the patient’s skin. Although the magnitude of contamination and the virulence of the contaminating organism certainly contribute to the rate of infection, all surgical wounds become inoculated with bacteria to some extent at the time of surgery, but in only a small percentage of patients does this contamination lead to clinical infection.19 Host defense mechanisms represent the primary barrier to establishment of infection, and these defenses may be impaired in patients undergoing craniotomy. Low levels of antibody and complement contribute to make the brain less efficient than other organs of the body at eradicating infection, and many of the underlying pathologies leading to neurosurgical intervention may significantly impair immune function. For example, patients with malignant gliomas express a variety of immune defects, including increased secretion of immunosuppressive cytokines and an increased fraction of regulatory T cells.20 Additionally, many of the adjunctive therapies used for treating brain tumors, such as corticosteroids, chemotherapy, and radiation, may result in immune compromise. Other frequent indications for craniotomy, such as trauma, have also been shown to be profoundly immunosuppressive.21 General surgical and infection control studies have identified other host factors that influence the risk for SSI, including advanced age, obesity, hypoalbuminemia, diabetes mellitus, and poor functional status.3,22–25 Gianotti and associates demonstrated the importance of nutritional status in oncologic surgery by showing that malnourished patients had improved resistance to infection after as little as 5 days of enteral nutrition.26 The increased rate of SSIs associated with these factors has been attributed to nonspecific deficits in host defense. Even though earlier reports suggested an increased rate of postoperative infection after general surgical procedures in patients infected with human immunodeficiency virus (HIV),27,28 several more recent retrospective studies performed since the advent of highly active antiretroviral therapy (HAART) have failed to demonstrate an association between SSI rates and HIV infection.29–31 Although the influence of these intrinsic factors on the rate of SSI after neurosurgical intervention has not been established in prospective studies, optimization of immune function through minimization of corticosteroid use, adequate nutritional support, and optimized perioperative glucose control may all be potentially helpful in the prevention of postcraniotomy infections.
Several factors specific to craniotomy have been identified as increasing the risk for postoperative infection. In a prospective multicenter trial, Korinek identified postoperative CSF leakage and early subsequent reoperation as independent risk factors for SSI, thus suggesting that careful attention to closure techniques and meticulous hemostasis may potentially result in lower rates of postoperative infection.5 Several other studies confirmed CSF leakage as a major risk factor for infection.8,32–35 Additionally, Korinek identified four independent predictors of postoperative infection after craniotomy: surgery lasting longer than 4 hours, emergency surgery, clean-contaminated and contaminated surgery, and neurosurgical intervention in the preceding month.5 Valentini and coworkers also observed an increased relative risk of 24.3 for postoperative infection in elective clean craniotomies lasting longer than 3 hours.36 The association between longer duration of surgery and infection has not been defined precisely, but plausible explanations include greater complexity of the surgery and prolonged exposure of the wound to bacterial contamination.37
A variety of other risk factors associated with infection after craniotomy have been less reliably demonstrated, including placement of drains or intracranial pressure monitors, poor neurological status, paranasal sinus entry, diabetes mellitus, and foreign body implantation (other than shunts).5,8,38 Synthetic dural substitutes are foreign bodies and might represent a potential risk factor for infection in comparison to autologous graft materials such as pericranium, temporalis fascia, or fascia lata. Actual evidence demonstrating increased rates of infection with their use, however, is limited. Malliti and coauthors reported a nonsignificantly increased incidence of deep wound infections after craniotomy with the use of a nonresorbable polyester urethane synthetic dural graft (Neuro-Patch, B. Braun, Boulogne, France).39 Postoperative CSF leaks were also significantly more frequent when using the synthetic dural substitute, thus limiting the ability of this study to determine whether use of the Neuro-Patch independently increased the risk for infection. The presence of nonresorbable dural substitutes may also impair the potential for an infected wound to be successfully treated because these grafts may become chronically colonized and could require removal to eradicate the infection.40 A variety of nonautologous, resorbable collagen dural substitutes are currently available, and their relationship to surgical infection has not been well explored. McCall and coworkers reported the uncomplicated use of several of these materials in a small number of patients in the setting of contaminated wounds, a finding suggesting that they may not impede clearance of infection.40 The use of Gliadel wafers (MGI Pharma, Inc., Bloomington, MN), which contain 1,3-bis-(2-chloroethyl)-1-nitrosurea, for the treatment of malignant gliomas has also been associated with an increased incidence of postoperative infection. McGovern and coauthors reported a 29% rate of infection in cases associated with insertion of Gliadel wafers between 1996 and 1999.41 Subsequent reports with larger patient populations, however, have not revealed statistically significant differences in the rate of infection with Gliadel use.41,42
Multiple prospective randomized clinical studies and a meta-analysis have validated the effectiveness of preoperative antibiotics in reducing the incidence of SSIs after craniotomy.11,33,34,43,44 Hugh Cairns described the first trial of a modern prophylactic antibiotic in neurosurgery in 1947 when he reported sprinkling a “light frosting” of penicillin powder directly onto the brain in 670 patients and thought that the results were superior to those of historical controls.45,46 In 1979, Malis demonstrated the ability of a prophylactic antibiotic regimen (vancomycin and an aminoglycoside) to reduce the incidence of SSI after craniotomy.47 Since these initial reports, a variety of antibiotic regimens have been used for effective surgical prophylaxis, and the choice of agent should be guided by individual institutional data on frequently recovered pathogens and their resistance profiles. In general, the antibiotic chosen for prophylaxis should be safe, provide an appropriate narrow spectrum of coverage against relevant bacteria, and be administered for a defined, brief course. The Surgical Infection Prevention (SIP) project has recommended three performance measures for monitoring appropriate antimicrobial prophylaxis use: selection of an appropriate antibiotic, administration within 1 hour before incision (2 hours is allowed for the administration of vancomycin and fluoroquinolones), and discontinuation of the antibiotic within 24 hours after surgery is completed.48 Antibiotics with short half-lives such as cefazolin should be readministered every 3 to 4 hours during surgery to ensure adequate drug levels throughout the operation, including the time of wound closure.49 Prolonged use of antibiotics beyond 24 hours postoperatively has not shown a greater benefit, may increase the risk for other nosocomial infections, and might encourage the emergence of multidrug-resistant pathogens.50,51 Despite wide acceptance of these basic measures, compliance with them continues to remain poor in the United States.52 Prophylactic antibiotics for craniotomy are covered in more detail elsewhere in this book.
Additional perioperative factors that may potentially reduce the risk for postoperative infection include maintenance of normothermia and supplemental oxygenation. Several prospective randomized trials evaluating active warming of patients during colorectal surgery to maintain normothermia have shown decreased rates of infectious complications; the proposed mechanism of action is support of adequate blood flow and tissue oxygenation at the surgical site.53–55 Supplemental administration of oxygen may also assist in preventing infection by increasing tissue oxygen levels and facilitating oxidative killing of bacteria by neutrophils. Belda and colleagues conducted a prospective trial to evaluate the postoperative infection rate in patients randomized to receive either 30% oxygen or 80% oxygen during elective colorectal surgery.56 The group receiving 80% oxygen had a 54% reduction in wound infections. Studies evaluating the role of supplemental oxygen or temperature control for neurosurgical interventions have yet to be performed.
Other perioperative risk reduction considerations include surgical site preparation and environmental control within the operating room. Although no evidence has been found that preoperative hair removal reduces the incidence of postoperative infection, any hair removal that is performed should be done as close to the time of surgery as possible and clippers used rather than a razor to minimize the number of bacteria that colonize the inevitable small cuts and abrasions that develop from shaving.57–60 Several antiseptic skin preparations have been used (chlorhexidine, iodophor compounds, alcohol), but no agent has been shown to be more effective than another.61 To provide effective antisepsis, these agents must remain on the skin until they dry naturally, with avoidance of any pooling. Theoretically, adhesive barrier drapes with antiseptic embedded within the adhesive may prevent regrowth of bacteria at the surgical site throughout the operative procedure; however, their ability to reduce the incidence of SSI has not been proved.62 Similarly, preoperative bathing or showering with an antiseptic skin product has no evidence in support of it.63 The operating room environment represents another important consideration in the reduction of SSIs. The number of health care workers within the operating room and traffic throughout the procedure should be kept to a minimum because bacterial shedding increases with activity and can potentially result in increased airborne contamination.58 Ensuring adequate ventilation minimizes the particulates and bacteria in the perioperative environment, and the use of high-efficiency particulate air (HEPA) filters has been shown to reduce the rate of SSI development after orthopedic implant surgery.52
Principles of Treatment
Immune defenses within the brain are rarely adequate to control infection once it has been established. Postoperative infections tend to be particularly difficult to resolve because of the complex anatomic changes resulting from craniotomy and the frequent involvement of virulent organisms. Early and decisive intervention is critical to limit morbidity, and the keystone of successful treatment is effective source control (i.e., drainage of abscesses and infected fluid collections and débridement of necrotic tissue).64 Once source control has been achieved, initiation of appropriate antibiotic therapy is necessary to eliminate any residual local infection.
The ability of antimicrobials to treat postcraniotomy infections successfully is a function of multiple factors. Selection of an antibiotic regimen should be based on the capacity of the antibiotic to penetrate the infected tissue effectively and to exhibit activity against the suspected pathogen. Bactericidal rather than bacteriostatic agents are generally preferred because of the inefficient opsonization and phagocytic capabilities within the brain.65 Maximal bactericidal activity is achieved only when the peak antibiotic concentration at the site of infection exceeds the minimal inhibitory concentration of the causative organism by at least 5- to 10-fold.66,67 Most antibiotic agents enter the CNS predominantly by passive diffusion down a concentration gradient, with physical barriers such as the blood-brain and blood-CSF barriers functioning as the primary determinants of drug distribution. Inflammation at the site of infection may facilitate entry of drugs across these barriers and into the brain, but not all postoperative infections are accompanied by marked inflammation, and concomitant treatment with corticosteroids may further impair drug entry.68 Other inherent physiochemical properties of the antimicrobial agent may affect its penetration into the CNS, including molecular weight, lipophilicity, protein binding, and ionization state. Optimal antibiotic administration and dosing rely on an understanding of the pharmacodynamic properties of the agent and the susceptibility profile of the microorganism. In the absence of data from prospective randomized clinical trials evaluating the success rates of specific antibacterial agents, recommendations for the treatment of postcraniotomy infections are based largely on the results of previous experience, along with consideration of the complex physiologic, bacteriologic, and pharmacologic factors involved.
Empirical treatment of postoperative infections should include broad coverage for the full spectrum of potential pathogens, including resistant gram-positive organisms (e.g., MRSA) and gram-negative bacilli (e.g., Pseudomonas and Acinetobacter spp.). Failure to include an antibacterial agent with activity against the responsible bacterium may result in severe neurological sequelae or death.65 Suitable empirical regimens for postcraniotomy infections typically include a combination of vancomycin and a second drug such as a third- or fourth-generation cephalosporin having antipseudomonal activity (e.g., ceftazidime, cefepime) or a carbapenem (e.g., meropenem). Antibiotic selection can be tailored once speciation and susceptibility testing from a microbiologic specimen are available.
Vancomycin has weaker activity against staphylococcal infections relative to β-lactams69 and decreased penetration into the CNS because of its large molecular weight (1449 daltons).65 Even in the presence of significant inflammation, concentrations of vancomycin may be critically low at the site of infection,70 and substitution of a β-lactamase–resistant penicillin (e.g., nafcillin, oxacillin) for vancomycin is appropriate, except in the setting of resistance or hypersensitivity. First-generation cephalosporins (e.g., cefazolin) have relatively poor CNS penetration and are not recommended for the treatment of deep wound infections.
Third-generation cephalosporins (specifically cefotaxime, ceftriaxone, and ceftazidime) are often used for the treatment of CNS and postcraniotomy infections because of their low toxicity, good CNS penetration, and excellent in vitro activity against many of the responsible bacterial pathogens. Administration of these agents in high doses achieves therapeutic concentrations within brain abscess cavities.71,72 Carbapenems such as imipenem (with cilastatin) and meropenem also cover a broad antimicrobial spectrum and have been used successfully for the treatment of bacterial brain abscesses.73–75 These agents, principally imipenem-cilastatin, are associated with an increased seizure risk, and their use in patients with postcraniotomy infections should be considered primarily for resistant pathogens. From a pharmacokinetic viewpoint, fluoroquinolones (levofloxacin, ciprofloxacin, moxifloxacin) are attractive agents for the treatment of CNS infection because of their lipophilicity and low molecular mass. The usefulness of these agents, however, is limited by a high rate of bacterial resistance, increased seizure potential (albeit modest), and a relative paucity of data regarding their clinical effectiveness for postoperative CNS infections.76
Newer agents that may prove useful for the treatment of resistant staphylococcal infections include linezolid and daptomycin. Linezolid has bacteriostatic activity against both MRSA and vancomycin-resistant enterococci and bactericidal activity against most streptococci. Linezolid may be administered intravenously or orally and has excellent bioavailability. Experience with this agent for the treatment of postcraniotomy infections is limited, and potential side effects include reversible myelosuppression and irreversible peripheral neuropathy.69 Daptomycin is a novel cyclic lipopeptide antibiotic that shows better in vitro microbicidal activity against MRSA than either vancomycin or linezolid and has been used primarily for the treatment of skin and soft tissue infections. Animal models of meningitis suggest that it may be an effective therapeutic agent in a setting of meningeal inflammation77–79; human studies to date are lacking.
Rifampin is a broad-spectrum antimicrobial that may have a role in the adjunctive treatment of infections associated with foreign body implantation or bone flap osteomyelitis. These types of infections are notoriously difficult to eradicate because of their resistance to host defense mechanisms and poor penetration of antimicrobials. Most foreign body infections are caused by staphylococci growing in biofilms consisting of bacteria clustered together in an extracellular matrix attached to the foreign body.80 Depletion of metabolic substances within the biofilm causes the microbes to enter a slowly growing (sessile) state, which renders them up to 1000 times more resistant to most antimicrobial agents than their free-living (planktonic) counterparts.69,81–84 Rifampin is one of just a few agents that can effectively penetrate biofilms and kill organisms in the sessile phase of growth. Because of the rapid emergence of resistance, rifampin must always be used in combination with a second active agent. In vitro data, experimental animal models, and several randomized clinical trials suggest that dual therapy that includes rifampin may be better than monotherapy for orthopedic hardware–related staphylococcal infections in terms of bone sterilization and cure rates.69,85,86 This experience makes adjunctive therapy with rifampin an attractive consideration for difficult postcraniotomy staphylococcal infections associated with retained hardware or osteitis. Caution must be used with rifampin therapy because of its very large number of drug interactions. Through cytochrome P-450 enzyme induction, rifampin increases the metabolism of many substrates, including antiseizure drugs, anticoagulants, and immunosuppressive and chemotherapeutic agents.87
Aminoglycosides have excellent activity against aerobic gram-negative bacilli, including P. aeruginosa, as well as synergistic activity with β-lactams against aerobic gram-positive cocci. Systemic use of aminoglycosides is limited by their toxicity profile and a narrow therapeutic window. Penetration into CSF and across the blood-brain barrier is poor.88 Polymyxins (e.g., colistin) also have activity against a broad array of gram-negative bacilli but fell out of favor because of nephrotoxicity.89 As a result of the retained activity of polymyxins against multidrug-resistant gram-negative bacilli, including P. aeruginosa and Acinetobacter baumannii, this class again plays a role in difficult to treat infections. Similar to the aminoglycosides, the distribution of systemically administered polymyxins to CSF is poor. Intraventricular antibiotic administration bypasses the blood-brain barrier, can achieve much higher CSF concentrations than with systemic administration, and has been used successfully in multiple case reports.90–92 Intraventricular antibiotic dosing has been associated with neurotoxicity, however, in experimental animal models and a small number of case reports.93,94 Currently, there are no well-established data to support adjunctive intraventricular administration when a systemically delivered antimicrobial can achieve adequate microbicidal concentrations in CSF.
Superficial Infections and Bone Flap Osteomyelitis
Clinical Manifestations
Superficial infections are the most frequent infectious complication after craniotomy. Although every surgical patient is at risk for postoperative infection, a variety of factors may contribute to create an environment that is suboptimal for wound healing and more favorable for infection, including repeat operative intervention, poor tissue quality, impaired vascular supply, radiation injury, nutritional deficiencies, and the presence of foreign bodies. The role of foreign material in facilitating infection was first reported by Elek and Conen, who demonstrated a 10,000-fold increased risk for skin abscesses in the presence of suture material.95,96 Continuous activation of granulocytes by foreign bodies may lead to local impairment of phagocytic ability, thereby reducing the amount of bacterial contamination needed to establish infection.18,66
Superficial infections are typically manifested as local erythema, swelling, and tenderness at the craniotomy site or as wound breakdown and suppurative drainage. With progressive infection, systemic signs such as malaise, fever, or chills may develop. The presence of neurological symptoms such as meningismus, altered mental status, or new focal deficits strongly suggests the coexistence of deep wound infection. The most common pathogenic agents of superficial wound infections are gram-positive cocci, including S. aureus, coagulase-negative staphylococci, and P. acnes.35,53,58,97 Infection of the bone flap most often results from either direct bacterial inoculation at the time of surgery or extension of infection from the adjacent subgaleal or epidural compartments.
Diagnostic Imaging and Laboratory Data
The presence of superficial wound infection is often clinically apparent; however, imaging studies can frequently assist in defining the anatomic extent of infection (especially extension through the dura), as well as possible precipitating factors, such as entry into the mastoid air cells or paranasal sinuses during craniotomy. Computed tomography (CT) or magnetic resonance imaging (MRI) may reveal fluid collections in the subgaleal or epidural spaces that require surgical evacuation or extension of infection beyond the dura and into the subdural space or brain parenchyma. Imaging studies may also show evidence of bone flap destruction suggestive of osteomyelitis. Unfortunately, diffusion-weighted imaging, which is very sensitive to spontaneous intracerebral abscesses, is frequently unreliable in diagnosing the presence of superficial infection after craniotomy.98
Measurement of the erythrocyte sedimentation rate (ESR) or C-reactive protein (CRP) level may provide some assistance in detecting infection and monitoring the response to therapy. These acute-phase reactants are normally elevated after craniotomy and return toward baseline levels by the fifth postoperative day.99 Although these markers are highly nonspecific, prolonged elevation or a secondary increase in their levels may indicate the development of bone flap infection. Failure of the ESR or CRP level to decline after institution of therapy may signify persistent infection and the need for prolonged antibiotic therapy or further diagnostic imaging to detect recurrence.
Treatment
Treatment of superficial wound infections depends on the extent of infection. Superficial cellulitis, a spreading infection of subcutaneous tissue without deeper infection of the subgaleal space or bone flap, is generally treated with oral or intravenous antibiotic therapy. Oral agents typically used to treat gram-positive bacteria include first-generation cephalosporins (e.g., cefazolin) or β-lactamase–resistant penicillins (e.g., dicloxacillin). In patients with rapidly spreading infection, prominent systemic symptoms, or significant comorbidity, initial antibiotic administration should be by the intravenous route until the symptoms improve and fever abates.100
Devitalization and devascularization of the bone flap at the time of craniotomy present a unique challenge in the treatment of infection because of impaired delivery of host defense mechanisms and antibacterial agents. Treatment options include antibiotic therapy alone, débridement with replacement of the bone flap, or surgical débridement with removal of the bone flap. Prolonged antibiotic therapy may control the clinical manifestations of infection but rarely leads to complete eradication, with frequent recrudescence after discontinuation of the antibio. Removal of the infected bone flap followed by delayed cranioplasty allows the best chance of clearing the initial infection; however, this treatment approach entails multiple surgical interventions and at least temporary cosmetic deformity while predisposing to the possibility of subsequent brain injury with a long-term risk for cranioplasty infection.13,101–103