Sulfonamides

Sulfonamide neurotoxicity can occur in the premature and newborn period because sulfonamides have the potential to displace bilirubin from albumin, with the resultant free bilirubin being deposited in the basal ganglia and subthalamic nuclei and resulting in kernicterus.² Moreover, there are reports in the literature of other CNS toxicities, including ataxia, depression, and psychosis with associated visual and auditory hallucinations.^3,4

Quinolones

The incidence of neurotoxicity from quinolones ranges from 1% to 2%, and symptoms may include headache, dizziness, and insomnia.^5–8 Additionally, instances of delirium, acute psychosis, and seizures have been reported.^5,6,9 Moreover, there have been reports of the development of demyelinating polyneuropathy,¹⁰ exacerbation of myasthenia gravis,^11–15 and peripheral sensory disturbances¹⁶ with fluoroquinolone use. Signs of pseudotumor cerebri may develop in infants and young children with high doses of nalidixic acid, the first quinolone to be introduced.¹⁷ These CNS effects should generally resolve with the discontinuation of therapy.

Penicillins

Neurotoxicity after parenteral administration of penicillin G is most likely to occur in patients with renal insufficiency, intracranial lesions, or alteration of the blood-brain barrier (BBB).¹⁸ This toxicity can occur when the concentration of penicillin G in cerebrospinal fluid (CSF) exceeds 10 µg/mL and may be manifested as lethargy, confusion, twitching, multifocal myoclonus, or seizures.¹⁸ Arachnoiditis and encephalopathy may follow the intrathecal injection of penicillin G, and therefore intraventricular and intrathecal administration should be avoided.¹⁸ The development of symptoms that may resemble panic attacks or acute psychosis with seizures or hallucinations is known as Hoigne’s syndrome and can follow the intravascular injection of penicillin.^19,20 Furthermore, ampicillin may exacerbate weakness in patients with myasthenia gravis.²¹

Cephalosporins

Many cephalosporins have been associated with neurotoxicity, and ceftazidime, in particular, has been reported to cause hallucinations,²² confusion, encephalopathy,²³ and status epilepticus.^22,24 There have also been published reports of cephalosporin-associated recurrent aseptic meningitis.²⁵

A hypersensitivity reaction is the most common side effect of cephalosporin administration and may be manifested as anaphylaxis, bronchospasm, urticaria, fever, or maculopapular rash.

Carbapenems

The most common adverse events encountered with carbapenem administration are nausea and vomiting.²⁶ Other frequent adverse events include diarrhea, rash, fever, and laboratory abnormalities, such as elevated liver enzyme levels, eosinophilia, thrombocytopenia, and increased prothrombin time.²⁶ Patients at risk for seizures are those with preexisting renal insufficiency or an intracranial mass lesion who are given high doses of imipenem.²⁷ Furthermore, there are reports that concomitant administration of imipenem with theophylline, quinolones, metronidazole, ganciclovir, or cyclosporine may reduce the seizure threshold.^28–30 In contrast to imipenem, meropenem is less likely to induce seizures.^31–34

Aminoglycosides

The most important and serious side effects of aminoglycosides are nephrotoxicity and ototoxicity. The other dose-related adverse effect of clinical importance is neuromuscular blockade, which occurs rarely and is usually related to an underlying condition.³⁵ The first signs of ototoxicity may be seen histologically, with the outer hair cells of the cochlea affected first, followed by the inner hair cells.³⁶ Damage to the hair cells results in high-frequency hearing loss, followed by progressive loss of hearing at lower frequencies.³⁵ After degeneration of the hair cells, there can also be damage to nerve fibers.³⁷ The first clinical symptom of cochlear damage is often high-pitched and continuous tinnitus.³⁸ In regard to the vestibular apparatus, hair cell damage occurs along with deterioration of the otoconial membrane and otolith structures.^39,40 Clinical signs and symptoms of vestibular damage include disequilibrium, ataxia, transient positional vertigo, and oscillopsia.^35,38 In rare instances, aminoglycosides have the potential to cause neuromuscular blockade and paralysis.³⁵ Patients with preexisting myasthenia gravis are at a higher risk for the development of neuromuscular blockade.⁴¹ Risk factors for ototoxicity include renal insufficiency, preexisting impaired hearing, old age, sepsis, dehydration, fever, previous aminoglycoside exposure, and concomitant use of vancomycin, cisplatin, or carboplatin.³⁵ Anosmia after aminoglycoside therapy has also been described in the literature, with the sense of smell returning after time.^42,43 Intrathecal or intraventricular administration is used rarely because it may cause local inflammation and can result in aseptic meningits.⁴⁴

Polymyxins

The major adverse drug effects of polymyxins are nephrotoxicity, neurotoxicity, and neuromuscular blockade. Renal insufficiency and giddiness may result when polymyxins are used parenterally, and pain, numbness, paresthesias, confusion, coma, and convulsions may result when they are used intrathecally.^45–47

Neurotoxicity from parenteral polymyxin B and colistin is seen most frequently in patients with compromised renal function, with an overall incidence of approximately 7.3%.⁴⁸ With increasing doses of polymyxin, patients may experience circumoral paresthesias, convulsions, apnea, distal paresthesias, ptosis, diplopia, dysphagia, dysphonia, areflexia, ataxia, and dizziness.^49–51 Intrathecal dosages greater than 50,000 U/day have been reported to cause chemical meningitis.^18,52

Vancomycin

Neurotoxicity with the administration of vancomycin can include both vestibular damage and cochlear damage, which can result in tinnitus and sensory hearing loss.^53,54 Tinnitus may be an early symptom and can indicate the development of deafness.⁵⁵ Ototoxicity in the early stages, characterized by tinnitus and dizziness, appears to be reversible.⁵⁶ However, by the time that the patient experiences a noticeable hearing deficit, the toxicity is often irreversible.⁵⁷ The mechanism by which vancomycin causes ototoxicity is thought to be direct damage to the auditory nerve, which leads to irreversible loss of the sensory hairs in the cochlea that initially affects high-frequency sensory hearing.⁵⁷ Subsequently, lower frequencies are affected, and eventually, total hearing loss may result.⁵⁸

Tetracyclines

Neurotoxicity is a well-recognized side effect of tetracycline therapy, and there are several reports in the literature documenting increased intracranial pressure with medium- to long-term use.^59–62 In infants, a bulging fontanelle may be apparent but resolves on discontinuation of the therapy.⁶³ Vestibular symptoms are also reported with tetracycline administration and may include dizziness and ataxia.^64–66 Additionally, tetracyclines are thought to block the neuromuscular junction by both prejunctional and postjunctional effects that depress the sensitivity of muscle to acetylcholine.^3,67 This neuromuscular blockade has been reported to exacerbate myasthenia gravis.¹⁸

Chloramphenicol

Neurotoxic effects of chloramphenicol administration are rare. Optic neuropathy and peripheral neuropathy have been cited in the literature in children with cystic fibrosis.^68–71 Symptoms of visual toxicity may include blurred vision followed by loss of visual acuity and impaired red-green color discrimination, whereas peripheral nerve toxicities result in burning, tingling, or numbness of extremities.³ The aforementioned symptoms generally resolve after discontinuation of the therapy.³ Additionally, there have been reports of chloramphenicol-induced encephalopathy that can progress to delirium.⁷²

The most feared toxicities of chloramphenicol administration include bone marrow suppression, aplastic anemia, and gray baby syndrome.⁷³

Macrolides

Neurotoxicity is mostly associated with erythromycin and may result in neuropsychiatric symptoms or ototoxicity.⁷⁴ The neuropsychiatric symptoms that have been reported in the literature include confusion, hysteria, anxiety, and nightmares, all of which disappeared after therapy was terminated.^75,76 The ototoxicity that is induced by erythromycin is a result of high doses, or it can occur in patients with preexisting hepatic or renal insufficiency.^74,77 In contrast to aminoglycosides, the hearing loss with macrolides is reversible with cessation of the drug.⁷⁸ There are also reports cited in the literature of the potential for erythromycin to exacerbate weakness in patients with myasthenia gravis.^79,80

Linezolid

Neurotoxicity can result in peripheral neuropathy with prolonged use of linezolid.^81–83 Most cases of peripheral neuropathy have occurred when linezolid is used beyond the maximal recommended duration of therapy of 28 days.⁸³ Additionally, there have been several reports documenting linezolid-associated optic neuropathy.^84–86

Rifampin

The neurotoxic effects of rifampin may include ataxia, confusion, dizziness, numbness, muscular weakness, inability to concentrate, and headache.^18,87

The adverse reactions most commonly reported include rash, fever, headache, general malaise, gastrointestinal symptoms, nausea, and vomiting.^88,89

The Blood-Brain and Blood–Cerebrospinal Fluid Barriers

The BBB is formed by endothelial cells of the cerebral vasculature, supported by astroglia and pericytes, with tight junctions between the endothelial cells and minimal fenestrations or bulk transport across the cells. For the BCSFB, this barrier is located at the epithelial layer of the choroid plexus, not at the endothelium, but is similar in character, with tight junctions between the epithelial cells. For both barriers, active influx and efflux transporters located on the endothelial/epithelial cell surface may drastically alter the distribution of an antibiotic into the desired compartment.^90,91 Many factors can contribute to the permeability of any substance across the BBB:

Each of these factors affects a substance’s ability to cross the brain barriers and reach its intended site of action. Increasing molecular weight, ionization, plasma protein binding, and metabolism at the barrier and the presence of efflux transporters decrease the permeability of substances across the brain barriers.^91,92 Increased lipophilicity, influx transporters, and inflammation can increase the permeability of the barriers to antibiotics.⁹³ The inflammation associated with infections may decrease over the course of and in response to treatment. Antibiotics may cross into the CSF or brain parenchyma more readily at the initiation of treatment, but as the inflammatory response to the infection abates, either from the antibiotic treatment or from the administration of other medications such as dexamethasone, the ability of an antibiotic to cross the brain barriers diminishes.^94,95 However, this effect may not alter outcome.^96–100

The overall goal of antibiotic treatment is to deliver an adequate concentration of the drug to the proper compartment. Antibiotics that are bactericidal are generally preferred for the treatment of CNS infections as well because of the low concentration of immunologic proteins such as complement and immunoglobulins and the relatively low numbers of phagocytic cells, although this contention may be changing.^101–103 This goal may be accomplished in several ways. First, the dose of the drug may be increased. This method is helpful for drugs with low systemic toxicity and relatively low permeability across the brain barriers, such as β-lactam antibiotics. Second, the choice of antibiotic may be changed to a drug that has greater penetration into the CNS, such as chloramphenicol or quinolones. Third, antibiotics may be delivered directly across the brain barriers (usually by indwelling ventricular or lumbar catheters). This method is especially helpful when using antibiotics that have higher systemic toxicity and poor permeability across the brain barriers, which may limit the systemic dose that can be administered; such drugs include vancomycin and aminoglycosides. For example, administration of chloramphenicol via an intraventricular route does not provide the same advantage that intraventricular administration of vancomycin does because chloramphenicol’s penetration of the brain barriers already exceeds that needed to provide a therapeutic concentration. In addition, bypassing the brain barriers does not provide a significant advantage because chloramphenicol also needs to be hydrolyzed to be effective, which is usually done in the liver. Although hydrolysis has been shown to occur in the CSF, intraventricular administration is probably not necessary.¹⁰⁴ Alternatively, using higher doses of β-lactam antibiotics to reach bactericidal concentrations is easily accomplished and generally associated with minimal systemic toxicity. Intraventricular administration of β-lactams is probably more toxic than systemic administration, although whether the source of the toxicity is the antibiotic or the underlying infection is not clear.¹⁰⁴

Pharmacokinetics of Antibiotic Delivery

To achieve effective antibiotic dosing, the pharmacokinetics of antibiotic administration must be understood. For infections occurring outside the barriers of the CNS, the principles of systemically administering antibiotics are less complicated. Peak concentrations depend on bioavailability, the amount delivered, the volume of distribution, and elimination via metabolism and excretion.¹⁰³ However, the pharmacokinetics of antibiotic administration to the CNS depends on both systemic pharmacokinetics and the behavior of the antibiotic in its access to and elimination from the CNS. The sum of these factors can be ascertained by experimentally determining the proportion of antibiotic that reaches the CNS. Most data relating to antibiotic pharmacokinetics in the CNS come from studies on CSF and meningitis. Much less data exist on these parameters for the brain parenchyma itself.^105,106

In determining this proportion, careful interpretation of experimental data is required. Many studies looking at the proportion of antibiotic reaching the CNS use simple plasma-CSF ratios at a single time point. These ratios can vary widely during a dosing cycle and can be quite misleading.^106,107 The most useful data are derived from using plasma-CSF AUC (AUC is the area under the drug concentration–versus-time curve) ratios in intermittent dosing or steady-state concentrations during continuous infusion. For β-lactam antibiotics, the AUC ratio generally ranges from 0.01 to 0.1. Less hydrophilic antibiotics, such as rifampicin, trimethoprim-sulfamethoxazole, and the fluoroquinolones, have ratios that range from 0.1 to 0.9. Vancomycin and the aminoglycosides also exhibit low penetration into CSF, with ratios of less than 0.1.^106,108 However, data suggest that these ratios may be different in the treatment of infections of the brain parenchyma (i.e., brain abscess). One study showed equivalent levels of antibiotic in abscess fluid and plasma 6 hours after administration; however, these are also point ratios, not AUC ratios.¹⁰⁹ Whether the antibiotics are as effective in abscess fluid is a separate consideration. Concentrations of antibiotics throughout the CSF are not constant. Ventricular CSF will have a lower concentration of protein and antibiotic than will lumbar CSF because the CSF produced in the ventricles has not yet mixed with exuded extracellular fluid from the brain parenchyma. Therefore, CSF concentrations of antibiotics rely on the permeability of both the BBB and the BCSFB. Penetration of antibiotics through the blood-lesion barrier (specifically, the blood-abscess barrier) will vary with the stage of formation of the abscess, the relative vascularity of the lesion, and even the cause of the lesion. However, it is impossible to differentiate the individual contributions of the blood-lesion barrier and the surrounding BBB to antibiotic concentrations by measuring antibiotic levels within abscesses.^{105,109–111}

The half-life of the antibiotic in CSF is also an important consideration. Most antibiotics are not metabolized in CSF. Elimination is achieved either by diffusion back through the BBB and BCSFB or from turnover of the CSF. Generally, the CSF half-life of antibiotics is significantly longer than the plasma half-life. The CSF half-life of antibiotics may also be increased in CNS infections because of decreased turnover of CSF. Conversely, in patients with CSF shunts or external CSF drains, the CSF half-life may be quite variable because the circulation of CSF is altered by the presence of the shunt or drain.^112,113

Central Nervous System Toxicity of Antibiotic Therapy

Antibiotics may be toxic to the CNS when administered systemically, as discussed earlier, or when administered directly into the CNS via an intrathecal route or by means of antibiotic irrigation.

Intrathecal antibiotics may also have significant neurological toxicities, although the most commonly used intrathecal antibiotics, vancomycin and gentamicin, appear to have relatively low toxicity when administered intrathecally. Additionally, discerning these effects may be difficult in patients with a coexisting serious CNS infection. Intraventricular vancomycin appears to be relatively free of toxicity, even at high CSF levels.¹¹⁴ Intraventricular gentamicin may lead to CNS toxicity and cause ototoxicity or epilepsy; however, these effects are not clearly related to CSF levels of gentamicin.^115,116 Other antibiotics less commonly used intraventricularly, such as the β-lactam antibiotics, may cause similar effects, especially seizures, as when they are administered systemically.^117,118

Systemic Antibiotic Prophylaxis

Antibiotic prophylaxis should be considered in terms of the inherent risk for infection associated with the procedure being contemplated. The standard approach to estimating risk for infection at the surgical site is the classification endorsed by the Centers for Disease Control and Prevention (Table 42-3).¹¹⁹

TABLE 42-3 Classification of Surgical Site infection

WOUND CLASS	DESCRIPTION	EXAMPLES
Clean	Uninflamed, uncontaminated, no trauma or infection, primarily closed with no break in sterile technique	Craniotomy for tumor Microlumbar diskectomy

Clean-contaminated Entry into the alimentary, respiratory, or genitourinary tract under controlled circumstances; no contamination; minor break in sterile technique Transnasal hypophysectomy Contaminated Nonpurulent inflammation, recent trauma, gastrointestinal tract contamination, major break in sterile technique

Clean Neurosurgical Procedures

The value of systemic antibiotic prophylaxis in clean neurosurgical operations is supported by level I evidence from multiple randomized clinical trials and high-quality meta-analysis.¹²² The same is true of systemic antibiotic prophylaxis for shunt operations.^123,124 Additional meta-analyses have supported its value in preventing meningitis after craniotomy¹²⁵ and in spine neurosurgery.¹²⁶

The value of systemic antibiotic prophylaxis in clean-contaminated operations has not been adequately studied to allow a confident conclusion to be reached. The current accepted practice in transnasal surgery varies from limited perioperative use¹²⁷ to use as though the procedure were contaminated (i.e., therapeutic doses for a therapeutic duration).¹²⁸

It is not feasible to study the differential effectiveness of different antibiotics. If one antibiotic reduces the expected infection rate to 1% and another is twice as good (infection rate of 0.5%), a study consisting of more than 5000 patients would be required to have a reasonable chance (power of 0.8) of finding that result to be statistically significant (P ≤ .05). Although the duration of prophylactic antibiotic administration in neurosurgery has not been specifically studied in a randomized trial, the general principles of systemic prophylactic antibiotic administration are well established in many disciplines¹¹⁹:

The value of systemic antibiotic prophylaxis in reducing the rate of infection after neurosurgical operations is well established with evidence of the highest quality. Failure to use antibiotics in this way requires justification with evidence of similar quality.

External Ventricular Drains

There is insufficient evidence to support a firm conclusion about the value of systemic antibiotic prophylaxis in reducing infections associated with external ventricular drains.¹³⁰ One underpowered trial compared systemic antibiotic prophylaxis with placebo for ventriculostomy and found no difference in the infection rate.¹³¹ A single trial compared short-term and long-term antibiotic administration and suggested that long-term use reduced infection rates but selected for resistant organisms.¹³² The infection rates in this study were high (extracranial infection rates of 40% with short-term use and 20% with long-term use). This is an area in which further study could be useful.

Cerebrospinal Fluid Fistula

A recent Cochrane Systematic Review examined the value of systemic antibiotic prophylaxis in preventing meningitis in patients with basilar skull fracture and confirmed the long-held view that the practice is ineffective.¹³³