Chapter 70. Cerebrospinal Fluid Shunt Infections

Cerebrospinal fluid (CSF), or ventricular, shunts are the predominant mode of therapy for children with hydrocephalus. Common causes of hydrocephalus in children include myelomeningocele, meningocele, obstructive or communicating hydrocephalus, intraventricular hemorrhage, congenital cyst, and central nervous system tumors.1 The majority of shunts are inserted in the perinatal period. The shunts divert CSF away from the ventricles, preventing increases in intracranial pressure that lead to neurologic sequelae. The typical CSF shunt has a proximal portion that enters the CSF space, an intermediate reservoir that lies outside the skull but underneath the skin, and a distal portion that terminates in either the peritoneal (ventriculoperioteal [VP] shunt), vascular (ventriculoatrial shunt; VA shunt), or pleural space (Figure 70–1).

Infection develops in 5–15% of all CSF shunts;2,3 most infections occur within 6 months of shunt placement.3,4 Factors associated with CSF shunt infections include premature birth, young age, neuroendoscope use during shunt insertion, prior shunt infection, and hospital stay more than 3 days at the time of shunt insertion.1,5–7 Insertion of a VP shunt in a premature neonate (age <3 months) has been associated with a nearly five-fold increase in the risk of shunt infection. Patients younger than 1 year at the time of shunt placement also have a substantially higher risk of shunt infection than those older than 1 year at the time of shunt placement.8,9 Insertion of a shunt after a previous shunt infection is associated with a four-fold increase in the risk of shunt infection.

The etiologic agents associated with CSF shunt infections are shown in Table 70–1.10 Staphylococcal species, especially coagulase-negative Staphylococcus and Staphylococcus aureus, account for almost two-thirds of all shunt infections.7,11 The remaining infections are produced by a wide variety of organisms, including gram-negative bacilli. Among 92 patients with VP shunts, prior S. aureus shunt infection (OR, 5.9; 95% CI: 1.4–25.9) independently increased the odds that S. aureus was the causal pathogen.5 Gram-negative organisms (e.g., Escherichia coli, Klebsiella pneumoniae, Pseudomonas aeruginosa) tend to have a delayed onset, suggesting inoculation after surgery.1,10Propionibacterium acnes has been isolated more often in recent series of VP shunt infections; this bacterium generally causes low-grade, indolent infections.12 The apparent increase in P. acnes infection is probably a result of the more frequent use of anaerobic culture media and prolonged (up to 7 days) incubation times. Candida species should be considered in premature infants and other immunocompromised patients as well as in those patients receiving parenteral nutrition or prolonged corticosteroid therapy.13

There are four common mechanisms of shunt infection: (1) local inoculation of bacteria at the time of surgery, (2) skin breakdown overlying the shunt with subsequent bacterial entry, (3) hematogenous shunt inoculation, and (4) retrograde infection from the distal end of the shunt. The most common mechanism of infection, local inoculation of bacteria at the time of surgery, usually manifests within several weeks of the operation. Bacterial entry following breakdown of skin overlying the shunt may occur if the incision fails to properly heal or if the patient disrupts the healing process by scratching the open wound; gram-positive bacteria are more likely in this scenario. Children, who are relatively immobile, such as those with severe neurologic disability, may develop an overlying decubitus ulcer, which permits bacteria direct access to the shunt. Rarely, accessing the shunt by needle puncture introduces colonizing skin bacteria into the shunt system. Children with shunts in their vascular system (e.g., ventriculoatrial shunts) are continually at risk of infection from bacteremia with retrograde spread to the ventricles. Finally, retrograde infection from the distal end of the shunt as a consequence of viscus (e.g., bowel, gallbladder) perforation may lead to distal catheter contamination; gram-negative bacteria are most commonly isolated in the context of bowel perforation.

History and Physical

The patient with a ventricular shunt requires a review of the symptoms and signs associated with increased intracranial pressure and central nervous system infection. Historical information obtained from the patients or their caregivers should include indications for shunt placement, postoperative course, and history of shunt infection or malfunction (Table 70–2).

In infants, the physical examination should include a measurement of head circumference and assessment of size and softness of the anterior fontanelle. The skull and scalp should be palpated and inspected for signs of fluid accumulation or soft tissue infection such as erythema and tenderness to palpation. Characteristics of the burr hole(s) should be noted and all incision sites on the skull, neck, chest, and abdomen should be inspected. Attention should be directed toward swelling and fluctuance, which may represent CSF or a purulent fluid collection.

The extracranial portions of the shunt system should be examined to assess for shunt patency (see Box 70–1). Digital compression of the valve is an integral part of shunt examination. Shunt pumping is easy to perform, does not require special equipment, and can support the diagnosis of overt shunt malfunction. See Box 70–1 for a step-by-step approach to shunt assessment.

A complete neurological examination should also be performed, including cranial nerve assessment and fundoscopy to detect papilledema and optic atrophy (which suggest elevated intracranial pressure). The neck should be palpated to detect cervical and posterior auricular adenopathy which may occur with infections of the shunt insertion site.

Signs and Symptoms of Proximal Shunt Infection

The clinical features of CSF shunt infection depend on the mechanism of infection, the causative pathogen, and the type of shunt. The most common clinical symptoms are fever, headache, nausea, and lethargy (Table 70–3).2,7 Shunt infection is the cause for shunt malfunction in 3–8% of cases.14Table 70–4 presents common signs and symptoms associated with CSF shunt malfunction.14

Signs of meningitis such as meningismus and photophobia are less common because infected CSF from the ventricles does not easily communicate with CSF in the subarachnoid space. Children with infections caused by indolent organisms such as P. acnes may have an insidious course with few overt symptoms. Additionally, children with gram-negative organisms have been described as “well appearing” although half have mild alterations of mental status.15

Infections of the external surface are less frequent and usually present with signs of local soft tissue inflammation such as focal swelling, pain, erythema, and purulent drainage from the incision site. Surface shunt infection is usually a complication of surgery because of the direct inoculation of the insertion site at the time of placement.

Signs and Symptoms of Distal Shunt Infection

Signs and symptoms of distal shunt infection depend on the location of the shunt tip and whether the internal lumen or the external surface is infected. Intraluminal infection of a ventriculoatrial shunt can result in bacteremia and systemic signs of toxicity, including fever, chills, and tachycardia. Rarely, compression of the reservoir or catheter track of an infected VA shunt can lead to intermittent bacteremia accompanied by fever and chills; this phenomenon has been referred to as the “shampoo clue” by some authors since some cases of VA shunt infection were suspected after inadvertent manipulation of the catheter track during hair washing caused fever and rigors.16 Severe sepsis or septic shock is uncommon. Intraluminal infection of a ventriculoperitoneal shunt usually produces signs of focal infection.

Other complications of CSF shunts are summarized in Table 70–5.17–19 Abdominal peritoneal pseudocysts develop as a consequence of clinical or subclinical infections that cause an inflammatory reaction around the catheter tip (Figure 70–2). The pseudocysts may grow quite large since the CSF encased within the pseudocyst cannot be resorbed by the peritoneal cavity. Pseudocysts complicate VP shunt placement in 0.7–4.5% of cases; usually as a late complication, occurring >12 months after initial shunt placement.17,20,21 Small pseudocysts are more likely than large cysts to be associated with infection, presumably because pseudocysts associated with infection cause symptoms earlier than noninfected pseudocysts. Among patients with abdominal pseudocysts, the abdominal symptoms (Table 70–5) precede central nervous system complaints such as lethargy, headache, and visual disturbances by several days or weeks.

If shunt infection goes untreated for extended periods, immunologic sequelae, such as shunt nephritis can occur. Nephritis in the context of VA shunt infection is caused by deposition of antibody–antigen complexes in the renal glomeruli. “Shunt nephritis,” which can be difficult to distinguish from bacterial endocarditis, occurs in 5–15% of VA shunt infections.

Shunt infection may occur with or without shunt malfunction. The differential diagnosis should focus on distinguishing shunt-related complications from other causes of headache or altered mental status. In addition to shunt-related infection, other acute infectious causes of headache include meningoencephalitis, brain abscess, sinusitis, orbital disease, and cranial neuralgias (e.g., herpes zoster). Other causes include stroke, subarachnoid hemorrhage, hypoglycemia, hypertension, collagen vascular disease, and migraine headaches.

Diagnosis of a CSF shunt infection requires either isolation of a pathogen from ventricular fluid, lumbar CSF, or blood (for VA shunts) or the presence of CSF pleocytosis (usually defined as >50 white blood cells/mm3 in the context of a CSF shunt) in combination with either shunt malfunction or one or more of the signs or symptoms listed in Table 70–3.

CSF Studies

The CSF should be sent for cell count, glucose, protein, Gram stain, and aerobic and anaerobic bacterial culture (Table 70–6).22,23 A CSF fungal culture should also be performed in premature infants and in children with other immunocompromising conditions.13 A mild CSF pleocytosis, low CSF glucose level (hypoglycoracchia), and elevated CSF protein are usually present in cases of ventricular infection. CSF white blood cell counts typically range from 100 to 2500/mm3 in VP shunt infections although normal CSF parameters (including CSF white blood cell count) have been reported in 17–35% of children with VP shunt infections.4,10,22 CSF pleocytosis alone is not diagnostic of infection. Mild to moderate pleocytosis (20–500 white blood cells/mm3) also occurs as a consequence of postsurgical or foreign body (i.e., shunt)-associated inflammation. Furthermore, infections caused by indolent organisms such as P. acnes may fail to induce a vigorous inflammatory response.

The types of white blood cells present in the CSF also facilitate the diagnosis of infection. McClinton et al.24 found that the presence of >10% CSF neutrophils had a specificity of 99% for shunt infection (i.e., almost all patients without shunt infection had very few CSF neutrophils). Furthermore, the positive predictive value of >10% CSF neutrophils was 93% (i.e., almost all patients with >10% CSF neutrophils had a shunt infection).24 CSF eosinophilia (>5% of total CSF white blood cell count) has also been associated with both shunt infection and malfunction but may also occur in response to intrathecal antibiotics or as a reaction to the shunt catheter.24

Ideally, fluid from the reservoir should be obtained by percutaneous aspiration under sterile conditions. Shunt drainage should be performed by a neurosurgeon or a clinician with experience in performing this procedure. The potential complications of draining CSF directly from the shunt include bleeding at the puncture site, CSF leakage, mechanical damage to the valve, and introduction of infection. In addition, draining CSF too rapidly may cause intraventricular or subdural bleeding. Bacteria are identified by Gram stain of CSF obtained from the reservoir in up to 80% of cases although the likelihood of a positive Gram stain depends on the causative organism. S. aureus and aerobic gram-negative rods such as E. coli typically have positive Gram stain results while P. acnes, coagulase-negative staphylcoccci, and viridans group streptococci are positive in <40% of cases.10 Therefore, a negative Gram stain does not exclude the diagnosis of shunt infection. Although most bacteria causing shunt infections grow within 48–72 hours, cultures should be held for 5–7 days since fastidious organisms such as P. acnes may take longer to grow. Contamination and true infection cannot be readily differentiated when bacteria are identified by culture in the context of normal CSF parameters. In such cases, infection should be strongly considered and shunt aspiration should be repeated; a positive culture with the same bacteria usually indicates true infection.

Isolation of bacteria from CSF obtained by lumbar puncture suggests CSF shunt infection in the appropriate context. However, children requiring a CSF shunt often have impaired CSF flow. As a consequence, the ventricular fluid may have little or no communication with the lumbar spinal fluid and CSF obtained by lumbar puncture may not suggest infection despite the presence of ventriculitis.

Other Laboratory Studies

Blood should be routinely obtained for culture from patients evaluated for suspected shunt infection. While a negative peripheral blood culture does not rule out a shunt infection, a positive blood culture often influences the choice of antimicrobial therapy. Among patients with confirmed VP shunt infection, blood cultures are positive in 20–30% of cases.4,10 Peripheral cultures are more likely to be positive in patients with VA shunt infection where blood cultures are positive in 90% of cases.1,7 Laboratory manifestations of shunt nephritis include anemia, azotemia, hypocomplementemia, as well as hematuria and proteinuria.

Neuroimaging

Neuroimaging studies including X-rays of the skull, neck, chest, and abdomen (the “shunt series”) and computed tomography (CT) should be performed as part of the routine evaluation of a child with a suspected CSF shunt infection. Neuroimaging studies may provide evidence of shunt malfunction that accompanies some cases of infection. Specific abnormalities that can be visualized on the shunt series include disconnection of the distal catheter, retraction of the distal catheter tip, and discontinuity near the proximal shunt bulb. Routine performance of shunt series has a low overall yield but on rare occasions detects abnormalities that are missed by CT.25 Both CT and MRI of the head will detect increased ventricular size; this finding may reflect either increased intracranial pressure or hydrocephalous ex vacuo, a condition where the increased ventricle size reflects shrinkage of brain parenchyma rather than an increase in the intracranial pressure. Ventriculitis and meningitis can be visualized on CT and MRI as enhancement of the ventricular ependymal lining or cerebral cortical sulci.26 In rare cases, subdural empyema or brain abscess may be the first indication of shunt infection. Radiologic imaging of other areas should be considered depending on the location of the distal catheter tip. CT or ultrasound of the abdomen may identify abdominal peritoneal pseudocysts at the distal portion of a VP shunt (Figure 70–2). Some free fluid in the peritoneal cavities is normal but larger amounts should raise concern for infection. Chest radiography detects pleural effusions associated with ventriculopleural shunt infection.

The child with a ventricular shunt infection should be managed in consultation with neurosurgical and Infectious Diseases specialists. Success rates with various treatment strategies are as follows:

• Intravenous antibiotics without shunt removal, <25%;
• Intravenous and intraventricular antibiotics without shunt removal, 40%;
• Intravenous antibiotics with shunt removal and immediate replacement, 75%; and
• Intravenous antibiotics with shunt removal and delayed replacement, >90%.11,27

Therefore, optimal management of a CSF shunt infection includes intravenous antibiotics and removal of all components of the infected shunt with placement of a temporary external ventricular drain until the CSF is sterile.4,8,11,27,28 The external ventricular drain facilitates resolution of the ventriculitis and permits continued monitoring of CSF parameters. Infection complicates fewer than 5% of closed external drainage systems; routine changing of the drainage catheter does not appear to reduce the infection rate.

In cases of distal shunt infection, some neurosurgeons prefer to externalize only the distal portion of the shunt. This strategy still maintains CSF flow and still offers the ability to perform frequent ventricular fluid sampling without subjecting the patient to a more extensive surgical procedure. However, early infection of the proximal portion of the shunt may be obscured by antibiotic treatment and become active after discontinuation of therapy and reinsertion of the distal portion of the shunt.

Until an organism is isolated, patients should be treated with empiric antibiotic therapy that covers the range of potentially causative pathogens.28 Reasonable options include vancomycin in combination with ceftazidime, cefepime, or meropenem. Linezolid was successfully used to treat a woman with Staphylococcus epidermidis VP shunt infection and a history of Stevens-Johnson syndrome attributed to prior vancomycin therapy.29 Quinupristin–dalfopristin has also been reported to successfully cure patients with CSF shunt infections caused by coagulase-negative staphylococci.

Situations that may warrant additional measures include cases of delayed ventricular fluid sterilization (>3 days) and cases where the patient cannot safely undergo surgical catheter removal. First, intraventricular antibiotic administration should be considered. No antibiotic has been approved by the U.S. Food and Drug Administration for intraventricular use. However, commonly used intraventicular antibiotics include vancomycin, gentamicin, tobramycin, and amikacin.30–32 Polymixin B and colistin have also been administered directly into the ventricles to treat ventricular infections caused by gram-negative bacteria resistant to many commonly used antibiotics.33–36 Penicillin and cephalosporins should not be instilled directly into the ventricles since intraventricular administration of these antibiotics has been associated with increased neurotoxicity, including seizures. Second, rifampin has excellent CSF penetration and should be administered orally (in addition to an intravenous antistaphylococcal agent such as vancomycin) when the infection is caused by susceptible staphylococci.37,38 Third, neuroimaging should be performed to diagnose an intracranial abscess or empyema. Magnetic resonance imaging is preferred because of its higher sensitivity but contrast-enhanced CT is sufficient in many cases. Finally, either the trough ventricular antibiotic concentration or the ventricular fluid bactericidal titer should be measured to assess the adequacy of antibiotic therapy. No standardized values exist but many experts agree that the trough antibiotic concentration should exceed the minimum inhibitory concentration of the organism by 10-fold or more; lower values indicate suboptimal ventricular fluid antibiotic concentrations. Bactericidal titer measurements may not be readily available since they are technically difficult and time-consuming to perform; if no turbidity is observed after 24 hours of growth, reflecting failure of bacteria to grow at a dilution of 1:8 or higher (i.e., more dilute), then the ventricular antibiotic concentrations are probably sufficient.11

Management strategies are summarized in Figure 70–3. Duration of antibiotic therapy varies depending on the causative organism, the time to CSF sterilization, the extent of CSF inflammation, and the patient's clinical response. However, few studies have rigorously analyzed the relationship of duration of therapy to clinical outcome. In general, coagulase-negative staphylococcal infections can be treated for 5–7 days after the first negative culture. Other bacteria should be treated for longer periods after CSF sterization: 7–10 days for streptococci and P. acnes; 10 days for S. aureus; and 10–21 days for gram-negative rods. The isolation of bacteria in the absence of CSF pleocytosis suggests colonization of the shunt catheter; in this situation, some experts recommend a shorter course of therapy than discussed above. While some authors39 recommend a 3-day period off antimicrobial therapy to verify clearing of infection prior to shunt reimplantation, such a strategy does not appear to benefit the patient.40 Patients with a complicated course or suppurative complications such as brain abscess or intracranial empyema may require longer therapy. Additional therapy is not typically required once the shunt has been replaced. If the shunt was not initially removed, therapy should be continued for an even longer period of time than mentioned above, although the optimal duration in such cases is not known.

The mortality associated with ventricular shunt infections is low. Potential morbidity includes new or more frequent seizures and worsening neurologic impairment. Infections caused by S. aureus and Candida species have a substantially higher rate of recurrence despite adequate therapy than infections caused by other organisms.

• Laboratory findings of anemia, azotemia, hematuria, and proteinuria suggest shunt nephritis in a child with a ventriculoatrial shunt.
• Impaired CSF flow can lead to normal CSF findings despite the presence of ventriculitis in specimens obtained by lumbar puncture.