Pathophysiology

Pyogenic infections of the meninges originate either by hematogenous spread of bacteria or infected thrombi or by direct extension from bacterially colonized cranial structures adjacent to the meninges. Typical sources of direct extension include surgical or traumatic breaches of the paranasal sinuses or mastoid air cells, osteomyelitic foci within the skull, or congenital sinus tracts.¹ Iatrogenic infection via an infected shunt, intraoperative contamination, or even a lumbar puncture needle is also possible. In animal models of bacteremia, placement of a needle into the subarachnoid space causes clustering of bacteria at the site of injury. It is tempting to extend this by analogy to humans, although there is no proof that this occurs in systemic human bacterial infection.

Cerebral tissue itself is relatively resistant to infection, and direct injection of virulent bacteria into the brains of animals seldom yields abscess formation. Some coincident infarction of brain tissue either by venous or arterial occlusion or by direct injury is the usual (and possibly requisite) antecedent event. Bacteria in the CSF, however, attract a robust inflammatory response and require no such additional injury. Parameningeal foci of infection, whether lodged in meningeal or superficial cerebral vessels or in adjacent sinus cavities or bone, simply require a physical breach in the arachnoid membrane to initiate bacterial colonization of CSF and result in frank meningitis. When septic material embolizes from an infected lung or congenital heart lesion or when it extends directly from the ears or sinuses, more than one type of bacterium can be found in the CSF. By contrast, hematogenous infections usually permit only one bacterial type to gain entry into the subarachnoid space. Bacterial pathogens differ by age of the patient (Table 44-1), and therapy initiated before identification of the pathogen is generally chosen to cover the typical pathogens found within the age group of the patient. Once bacteria enter the CSF space, they initiate a cascade of events within that space that extends throughout its reach. Thus, meningitis starting in the spine can easily cause cranial nerve dysfunction, and the converse is equally true. CSF within the ventricles is not spared because bacteria may enter it (and cause ventriculitis) either directly from infective emboli to the choroid plexus or by reflux of bacteria through the foramina of Magendie and Luschka.

TABLE 44-1 Bacterial Pathogens Causing Meningitis

The first effect of bacteria within the CSF is hyperemia of the meningeal vessels, followed rapidly by migration of neutrophils into the subarachnoid space. The exudates increase quickly and extend to the sheaths of cranial and spinal nerves and into the perivascular spaces of the cortex. At first, neutrophils predominate, but over the next few days lymphocytes and histiocytes show a gradual increase. Exudation of fibrinogen and other proteins from blood continues while the meningitis is active and leads to the typical increase in CSF protein seen in spinal taps performed during the acute phase of bacterial meningitis. Toward the end of the second week plasma cells appear, and thereafter they increase as well. On a microscopic level, the exudate disperses into two layers, with the outer one immediately beneath the arachnoid consisting of neutrophils and fibrin and the inner one adjacent to the pia consisting of lymphocytes, macrophages, and plasma cells. Eventually, the exudate organizes and arachnoid fibrosis ensues with loculation of small pockets of cellular exudate that may allow recrudescence of the meningitis if antibiotic treatment is not prolonged.

During resolution of meningitis, the inflammatory cells disappear in the same order in which they came. The last to go are lymphocytes, macrophages, and plasma cells, which disappear more slowly than neutrophils and may remain for several months in decreasing numbers. Infections controlled early may leave no trace on arachnoid structure, whereas those treated after the infection has become solidly established may leave behind a thickened, cloudy, and adherent arachnoid membrane. The presence of this cellular immune reaction within the CSF also leads to changes in the small blood vessels on the cortical surface. Within 2 to 3 days of the onset of infection, endothelial swelling occurs and compromises the diameter of the vascular lumen, and the adventitia is infiltrated by neutrophils. Occasionally, necrosis of the arterial wall develops and may result in subarachnoid bleeding. The vessels ultimately demonstrate subintimal fibrosis, which is seen with any long-standing infection of the meninges but most typically with tuberculous meningitis. A similar process occurs in the veins, although the subintimal cellular infiltration is not seen; rather, the immune cells tend to infiltrate the entire wall diffusely. This can lead to necrosis of the wall, mural thrombus formation, and infectious thrombophlebitis by the end of the second week of infection. Ultimately, such vascular changes can culminate in ischemia and stroke.

Dysfunction of cranial and spinal nerves occurs as the purulent exudate surrounds the nerves and then, over a period of several days, infiltrates the perineurial sheaths. The endoneurium is typically not infiltrated, and the nerve fibers themselves are not usually damaged. Disruption of nerve function is also promoted by the vascular changes alluded to earlier should they affect the vasa nervorum and cause ischemia within the nerves.

Infection does not generally spread outward from the subarachnoid space to the subdural space. Occasionally, however, a subdural effusion is produced, more often in infants than adults. Because they tend to resolve spontaneously as the infection fades with medical therapy, these effusions are tapped only if they produce a significant mass effect on adjacent brain tissue.²

In the spine, fibrinopurulent exudate can accumulate around the spinal cord and roots of the cauda equina in sufficient quantity to block the spinal subarachnoid space (and cause chronic root pain). Blockage of the foramina of Magendie and Luschka or the basal cisterns can lead to hydrocephalus through interference with the normal circulatory pattern of CSF. Here, too, treatment of the hydrocephalus is generally deferred unless it is significant enough to produce symptomatic intracranial hypertension and endanger neurological function; it is usually mild in degree and will settle as the infection and inflammation subside and CSF flow is reestablished.

The effects of meningitis on the brain itself are minimal in the early stages, but an increase in cortical microglia and astrocyte numbers occurs after several days because of either diffusion of bacterial toxins from the meninges or disturbances in blood flow (but not the presence of bacteria within the brain parenchyma). The encephalopathy that ensues, as well as any associated seizures, should be considered indirect effects of the infection and probably associated with cerebral edema or changes in local vascular permeability and flow. In like fashion, bacteria within the ventricles initially show little effect on the ependyma and subependymal zone, but analogous changes are seen at later stages. The concomitant hydrocephalus, with resulting disruption of ependymal integrity, probably promotes subependymal proliferation of microglia and astrocytes.

Types of Bacterial Meningitis

The three most common bacteria causing meningitis are Haemophilus influenzae, Neisseria meningitidis, and Streptococcus pneumoniae, which together account for 75% of cases (Fig. 44-1). Less frequently seen are Staphylococcus aureus, Staphylococcus epidermidis, and Streptococcus group A, which usually occur after head trauma or neurosurgical procedures or with a brain or epidural abscess; Streptococcus group B, which is seen in newborns; and the Enterobacteriaceae (Klebsiella, Proteus, and Pseudomonas spp.), which occur after lumbar puncture or shunt placement.

FIGURE 44-1 Causes of acute meningitis worldwide and for all ages combined.

Pneumococcal, Haemophilus, and meningococcal meningitides have a seasonal pattern, with the majority of cases occurring in fall, winter, and spring and with a male preponderance. Because of the overuse of antibiotics prevalent in medicine today, drug-resistant strains occur increasingly, and it is important to know the patterns of drug resistance in general and the resistance of the bacterium in question in particular to form an effective therapeutic plan. Meningococcal meningitis occurs most often in children and adolescents but can be seen in adulthood until the age of 50, after which it sharply declines in incidence.^3,4 Pneumococcal meningitis and Haemophilus meningitis both predominate in the very young and in elderly adults.

These three most common meningeal pathogens colonize the nasopharynx in a significant percentage of the general population and need not be eradicated if found on routine nasal culture. Antecedent viral infection of the upper respiratory tract may predispose the patient to hematogenous spread to the meninges, but this is by no means certain. Certainly, these three microorganisms have a tropism for the meninges not shared by other bacteria.

Clinical Features

Clinical Testing

Lumbar puncture is the key test in diagnosing meningitis.⁶ This test should be done before instituting antibiotic therapy, and a positive diagnostic yield is increased by multiple taps. Those with bleeding disorders should, if possible, have hematologic correction of their coagulopathy before lumbar puncture. Those with increased intracranial pressure should undergo computed tomography (CT) or magnetic resonance imaging (MRI) of the head before lumbar puncture. Some will have hydrocephalus sufficient to require a ventriculostomy, and CSF can be acquired by that route rather than by a redundant lumbar puncture. In general, the fear of herniation when a spinal tap is performed in the presence of an intracranial mass lesion is much overstated. It is actually very unusual for transtonsillar herniation to occur when small amounts of CSF are withdrawn from the lumbar cistern. However, it is wise to use a small (25-gauge) needle in preference to a large-bore needle when intracranial pressure is elevated because this will permit more gradual removal of CSF and may further lessen the chance of such herniation.

The CSF findings in the various types of meningitis are shown in Table 44-2. Spinal fluid pressure is typically elevated, and a normal pressure in a patient strongly suggests that meningitis is not present. Pleocytosis is the most diagnostic finding, with the typical white blood cell count in CSF ranging from 1000 to 10,000 cells/mm³. A very high count (>50,000) should raise the possibility of a bacterial abscess that has ruptured into the ventricles. Such patients will be very ill and may require intraventricular as well as intravenous antibiotics. In bacterial (as opposed to viral or fungal) meningitis, a neutrophilic predominance is seen (85% to 95% of the total cell population), but later in the course of the illness the percentage of mononuclear cells rises.

Protein levels are high in 90% of patients, in the range of 100 to 500 mg/dL. Glucose levels fall to a level of less than 40 mg/dL. In hyperglycemic patients, the diagnostic finding is a drop in glucose to less than 40% of the blood glucose level measured simultaneously. Gram stain of spinal fluid sediment may show the bacterium in patients who have not previously been treated with antibiotics. It is invariably negative in those who have been so treated. CSF samples for culture are always best obtained before treatment because they too will be affected by the presence of previously administered circulating antibiotic. In untreated cases, cultures should be expected to be positive in 20% to 90% of patients. In culture-negative patients, counterimmunoelectrophoresis may help detect bacterial antigens that linger after the bacteria themselves have disappeared. Other adjunctive methods include radioimmunoassay and enzyme-linked immunosorbent assay (ELISA), also done to detect bacterial antigens, and polymerase chain reaction (PCR) amplification of nucleic acid sequences specific for bacterial species.⁷

Additional measurements may include lactate dehydrogenase (LDH) in CSF, which consistently rises in patients with bacterial meningitis. LDH isozymes 4 and 5 contribute the most to this rise. In patients who sustain brain injury from meningitis, LDH isozymes 1 and 2 rise sharply; they are only slightly elevated in those without such injury and thus can be used to predict neurological outcome. Lactic acid may also be measured because it is consistently elevated in meningitis caused by bacteria or fungi but remains normal in viral meningitis. Levels of lactic acid in CSF higher than 35 mg/dL are considered elevated.

Blood cultures should always be done because they are positive in 40% to 70% of patients with Haemophilus, meningococcal, and pneumococcal meningitis. If CSF cultures are negative, blood cultures may provide the only clue to the bacterial etiology. Cultures of the nasopharynx are occasionally helpful in that the absence of Haemophilus or meningococcus in a patient’s nose makes it unlikely that the meningitis is caused by either of these bacteria.

Radiologic Studies

Radiographs of chest, skull, and sinuses are useful in any patient suspected of having bacterial meningitis without a known source. CT of the skull base may also be useful in showing sinus infection or mastoiditis. Both CT and MRI will allow the diagnosis of hydrocephalus, infarction, brain abscess, or subdural empyema (or effusion), and such a scan should be performed at some point during the hospital stay of any patient being treated for bacterial meningitis.

Postcraniotomy Meningitis

The risk of CSF contamination leading to meningitis begins during surgery, but seeding of bacteria can occur after surgery in patients with either controlled CSF drainage or uncontrolled CSF leakage. The most common organisms cultured in patients after craniotomy have been S. aureus and S. epidermidis, with gram-negative organisms also being common.^8,9 Patients have fever, neck stiffness, and altered levels of consciousness, as do those with standard meningitis. However, those in whom meningitis develops after surgery have an increased risk for stroke secondary to venous infarction, an event precipitated by septic thrombophlebitis and encouraged by the dehydration that is often present. A lumbar puncture can be done safely in most postoperative patients, particularly those from whom a mass lesion has been resected and in whom the brain has been decompressed. CSF should be evaluated in the usual fashion for cell count, protein, and glucose. One study of meningitis in the postoperative period defined it (somewhat arbitrarily) as 100 white cells/mm³ with a minimum of 50% polymorphonuclear cells or 400 white cells/mm³ regardless of the polymorphonuclear percentage.¹⁰ However, normal lumbar puncture parameters have never been defined for specific times after craniotomy and would additionally depend on the nature of the disease for which the craniotomy was performed. It is likely that the measurements are changed by the operation itself, by anesthesia, by the use of steroids, and by the disruption of cerebral tissue that occurs. In the absence of precise information for each type of craniotomy, interpretation of the CSF results should not rely on borderline values or on one parameter alone but must be a synthesis of all relevant measurements. Gram stain of CSF can help guide selection of antibiotics because culture may take 2 or more days to yield bacterial identity and sensitivity to treatment. In patients in whom the clinical symptoms and signs point to meningitis, treatment is instituted after a spinal tap is performed. If subsequent cultures show the meningitis to be aseptic, antibiotic administration may be stopped. However, it is common to continue them because perioperative administration of antibiotics may interfere with CSF cultures done in the first few days after craniotomy.

The role of prophylactic antibiotics in preventing meningitis after craniotomy remains controversial. A study by Barker, a meta-analysis of six trials involving 1729 patients, showed that meningitis accounted for 32% of the 102 infections reported after craniotomy.¹¹ The incidence of meningitis in antibiotic-treated patients was 1.1% versus 2.7% in untreated controls. Because statistical analysis suggested no heterogeneity among the different trials, the author concluded that antibiotics conferred significant benefit in preventing meningitis after surgery. A second study by Korinek and associates drew a different conclusion.¹² In a series of 6243 consecutive craniotomies they showed an overall 1.5% incidence of meningitis and identified the presence of CSF leakage, concomitant infection of the surgical incision, and surgical duration to be independent risk factors. Antibiotic prophylaxis reduced incisional infection by half but did not change the incidence of meningitis. Such conflicting results leave the question of prophylaxis for meningitis unsettled, but it is widely accepted that prophylaxis should be used to prevent postcraniotomy infection in a more general sense. The main issue would seem to be the choice of antibiotics to allow suppression of the broadest possible range of types of postoperative infection.

Recurrent Meningitis

Meningitis will recur if the source of the bacteria remains active after treatment and suppression of the meningitis itself. Such recidivism should raise suspicion of an ongoing CSF leak through a previous basilar skull fracture or surgical procedure affecting the frontal, sphenoid, or ethmoid sinuses or the cribriform plate (Fig. 44-2). In the absence of previous trauma, a congenital fistula between the nasal sinuses and subarachnoid space may be suspected. In either case, CT with thin cuts through the skull base on bone windows is very helpful. Difficult cases are those in which the CSF leak is intermittent or very slow. We have found the best method of detecting small leaks to be injection of radioactive tracer (typically ⁹⁹Tc or ¹¹¹In) into the lumbar subarachnoid space with placement of nasal pledgets.¹³ The patient is then imaged over the next 24 to 48 hours to observe gradual movement of the tracer to the head (and if leak is present, into the nose and stomach). The pledgets are removed after the first day and also scanned; if a very slow leak is present, they may be the only source of positive detection. Finally, the kinetics of disappearance in CSF have been determined in normal patients, and an overly quick decay of tracer activity beyond the normal range suggests that a leak is present even if anatomic localization is lacking.¹⁴ Ultimately, if a fistula is found that is causing repeated bouts of meningitis, it should be closed surgically.¹⁵

FIGURE 44-2 Recurrent meningitis caused by persistent leakage of cerebrospinal fluid (CSF) through the skull base. This 53-year-old woman with Cushing’s disease underwent subtotal resection through a transbasal approach for tumor resection, and then a CSF leak developed 1 month later. This leak persisted over the next 9 months, during which time she suffered three separate bouts of meningitis, the first of which was culture positive for Streptococcus intermedius. This magnetic resonance image (axial T1 weighted, contrast enhanced) was obtained during the third episode. It shows diffuse enhancement over the brain convexities and falx cerebri, a classic finding with active bacterial meningitis. The expanded extracerebral space is consistent with ongoing intracranial hypotension from the loss of CSF. Of note, the frontal sinus shows mucosal thickening and enhancement consistent with paranasal sinusitis. However, there is no breach in the posterior wall allowing direct contact between the sinus cavity and intracranial compartment, and the meningitis must be ascribed to the ongoing CSF leak.

Treatment

Treatment of specific pathogens is presented in Table 44-3. The antibiotics suggested are only starting points because ongoing evolution of bacterial resistance to antibiotics has changed the patterns of antibiotic use significantly over the past 20 years.¹⁶ In all cases the best antibiotics are those with good penetration into CSF, a penetration that in some instances is enhanced by the presence of meningeal irritation. Selection of an antibacterial agent should take into consideration the need for bactericidal activity and for adequate penetration into CSF. The antibacterial concentration should be at least 10 to 20 times the minimal bactericidal concentration of the agent in question. Penetration of CSF by antibiotics is enhanced by a low molecular weight and simple chemical structure of the drug, low degree of ionization at physiologic pH, high lipid solubility, and low degree of protein binding. Initial therapy is empirical according to the most likely organism suspected for the age of the patient.¹⁷ The duration of therapy should be 10 to 14 days for those with bacterial meningitis in general and 21 days for patients with gram-negative bacteria and Listeria. If a shunt is present, it will typically require externalization or removal, or both, with reinsertion done only when the meningitis has been completely treated. Thus, prolonged ventriculostomy drainage will be needed in patients in whom meningitis develops but who are heavily shunt dependent. The ventricular drains should be changed at least weekly because their rate of infection (thereby perpetuating the meningitis) goes up after that time point.¹⁸ When a persistent parameningeal focus of infection is present, this too must be cleared (medically or surgically) before the patient’s meningitis can be declared cured. Treatment is invariably by intravenous administration, and in refractory cases or in those with profound ventriculitis, intraventricular therapy may additionally be needed.¹⁹ It is not necessary to repeat lumbar puncture at intervals during therapy as long as progressive clinical improvement suggests that the disease is clearing. CSF glucose in particular remains low after other signs of infection have disappeared and must be interpreted in the context of the patient’s clinical status.

TABLE 44-3 Treatment of Specific Pathogens Causing Meningitis