Infections of the Spine

Published on 26/03/2015 by admin

Last modified 26/03/2015

Print this page

This article have been viewed 1673 times

CHAPTER 276 Infections of the Spine

Infections of the spinal axis have been recognized throughout history. Evidence of tubercular disease of the spine has been found in Egyptian ¹ and South American mummies.² Social changes and advances in medical technology have changed the spectrum of causative organisms and the characteristics of patients who acquire these infections. However, the uneven nature of these socioeconomic changes across the globe, combined with variations in the endemic microbial flora, has led to regional variability in the epidemiology of spinal infections. The growth of medical technologies has enhanced the ease of detection and the options for definitive management of spinal infections. Ironically, technologic advances have also promoted the development of spinal infections by increasing the number of patients with iatrogenic immunosuppression, thereby enhancing the life expectancy of patients with chronic medical illness, and by increasing the complexity of spinal procedures and thus the potential for infectious complications.³

From a neurosurgical perspective, infections can be subdivided into three broad categories—spontaneous pyogenic infections, iatrogenic infections, and infections caused by mycobacteria, fungi, unusual bacteria, and parasites. We use the term “pyogenic” restrictively to refer to bacterial infections that produce a predominantly neutrophilic infiltrate and purulence. Although mycobacteria, Actinomyces, Nocardia, and certain fungal infections can produce purulence, this is a lymphocyte-predominant immune response, the so-called cold pus. The clinical behavior and management of such infections differ enough from that of pyogenic infections to merit discussion in a separate section.

Spontaneous Pyogenic Spine Infections

Traditionally, categorization of spinal infections has been based on the anatomic locus of the infection. The terminology used—osteomyelitis, diskitis, and epidural abscess—suggests that the infection is more or less limited to a particular anatomic structure. Advanced evaluation with magnetic resonance imaging (MRI) has made it clear that such terminology often does not accurately depict the nature of the infection. Most spinal infections spread to involve more than one of these anatomic structures and, not uncommonly, affect all of them.⁴^–¹⁰ In addition, all three subcategories of infection share similar epidemiology, risk factors, and clinical findings and are diagnosed with the same laboratory tests and imaging studies. Furthermore, a classification based on these anatomic boundaries does not contribute much to determining the optimal management for an individual patient. Although a longer duration of antibiotic therapy is typically needed to cure osteomyelitis than an infection of the disk or the epidural space, the role of surgical intervention, the incidence of neurological deficits, and the chance of failure with maximal medical management are poorly predicted by these traditional anatomy-based descriptors. Instead, we find it more useful to categorize these infections by the neurological condition of the patient (presence or absence of deficits) and by the vertebral level affected by the infection. A classification based on these descriptors is concordant with the biology of these infections and facilitates decisions determined on the basis of the patient’s clinical condition and the radiographic findings rather than on an arbitrary anatomic label.

Epidemiology

The incidence and demographics of pyogenic spinal infections (PSIs) have been significantly influenced by social changes, advances in medical technology, and the acquired immunodeficiency syndrome (AIDS) pandemic. The incidence of spinal infections appears to be on the rise.^11,¹² In years past, the incidence of epidural abscess was estimated to be around 0.2 to 2 per 10,000 hospital admissions,^4,⁷ but most recent authors have cited higher rates.¹³^–¹⁶ The annual incidence of all PSIs is now probably between 5 and 10 cases per million individuals, with a male preponderance.^11,¹⁷^–¹⁹ Older series describe two age-related spikes, one during the first decade of life and the second around the fifth or sixth decade,^20,²¹ which reflects the proclivity of the very young and the older population to acquire such infections. Infections still occur in these demographic groups, perhaps with no change in frequency,²² but are overshadowed in number by those that occur in adults with frequent bacteremia, significantly impaired host defenses, or both.^13,²³^–²⁶ Intravenous drug abuse and AIDS, the most common risk factors for PSI,^16,²⁷^–³² have reached epidemic proportions in modern times.^33,³⁴ Medical advances have led to an increased prevalence of individuals who are susceptible to spinal infection. Patients with diabetes, end-stage renal disease, and cirrhosis live longer than in years past.³⁵ Immunosuppression of transplant recipients,³⁶ long-term steroid therapy for autoimmune diseases, chemotherapy, chronic indwelling catheters for venous access,³⁷ splenectomy, genitourinary instrumentation, and other medical interventions also unfortunately enhance a patient’s risk for contracting an infection of the spine.^23,^28,³⁸^–⁴⁰ In susceptible patients, remote pyogenic infections (e.g., skin, genitourinary tract, lungs, gastrointestinal tract) can result in bacterial seeding of the spine. Spontaneous bacteremia with the potential to inoculate the spine is probably a relatively common event that is ordinarily rendered inconsequential by competent immune mechanisms. A Danish database of all patients with staphylococcal bacteremia in that country suggests that hematogenous vertebral osteomyelitis developed in 145 patients—roughly 1% of all those who had clinically apparent bacteremia.¹³ In some patients with predisposing conditions for PSI, no overt source of primary infection is found. The same can also be said of patients in whom infections occur in the absence of any particular risk factors. Increased awareness of spinal infections as a cause of fevers of unknown origin and the widespread availability of MRI scanners may both have contributed to the increase in the rate of diagnosis of this condition and hence a greater reported incidence.

Pathogenic Mechanisms

Infective organisms can be carried to the spine by four routes: via the arterial blood supply, retrograde by the vertebral venous plexus,^41,⁴² by direct inoculation (a contaminated surgical instrument/needle or a penetrating injury), or by direct extension from an adjacent nidus of infection (e.g., pulmonary abscess or sacral decubitus ulcer). In adults, the nutrient artery of the metaphyseal end plate is an end artery derived from the periosteal arteries. Infected thrombi that lodge in this nutrient artery, the metaphyseal artery, produce avascular necrosis of a portion of the metaphysis, which in turn creates a sizable nidus for infection.⁴³ Small anastomotic arteries that do branch off from the metaphyseal arteries are unable to supplant the blood flow to an ischemic metaphysis. These connect the metaphyseal plates at opposite ends of a vertebra and are the probable pathway for spread of the infection to transequatorial metaphyses while sparing the intervening equatorial region of the vertebra.⁴⁴ The equatorial region of the vertebra is supplied by multiple branches from the main segmental artery, thus making it very vascular and relatively resistant to the processes that result in infarction of the metaphysis. In addition to devitalization of the metaphyseal bone, thrombosis of the metaphyseal artery gives rise to ischemia of the intervertebral disk, which results in an infection of the disk, as well as chronic aseptic necrosis. This leads to gradual loss of disk height and, occasionally, the production of frank pus in the disk space. Purulence in the disk or the bone can result in septic thrombosis of draining veins, which in turn relay the infection to the epidural venous plexus, thereby leading to the formation of an epidural abscess.

The constellation of findings of an infection of the vertebral body, disk, and epidural space is seen commonly but not constantly, and the relative predominance of infection in a single region has been the basis for the traditional categorization of these lesions. The microvascular anatomy of the vertebrae and disk is slightly different in children, as a consequence of which the predominant type of infection in children tends to be limited to the disk space, as discussed in the later section on pediatric spinal infections.

Not all infections arrive in the spine through an arterial route—transvenous dissemination may also occur, somewhat akin to metastasis of genitourinary and gastrointestinal malignancies to the spine. Infections involving the left kidney may be more likely to spread to the spine because the left renal vein often communicates with Batson’s plexus.⁴⁰ Occasionally, infection may be restricted to the epidural veins, which act as a portal to the spine and can result in the production of contiguous or heterotopic epidural abscesses over multiple spinal segments without significant involvement of the bony or cartilaginous components of the spine.⁶ Similar mechanisms may be involved in the infrequent occurrence of infections arising within the posterior elements of the spine.⁴⁵^–⁴⁷ Less than 5% of spinal infections are isolated to the posterior spinal elements.^6,^48,⁴⁹

Bacteria can also reach the spine through contaminated needles or other instruments used during diskography, epidural anesthesia, or surgery; by contiguous spread from surrounding infected tissues; and by penetrating injuries or after cutaneous breakdown (especially over the sacrum and lower lumbar region). The clinical findings and management of such infections are addressed in a subsequent section.

Neurological deficits develop as a result of compression or ischemia (or both) of the spinal cord or cauda equina in 5% to 50% of all patients with PSI.^27,^28,^48,⁵⁰^–⁵² Such compression and ischemia can be caused either by an expanding epidural abscess or, more commonly, by kyphosis of the spine as a result of the loss of bone integrity and subsequent bony compression and distortion of neural elements. The lumbar spinal canal is somewhat capacious relative to the space requirement of the cauda equina. This allows the accommodation of sizable epidural collections and retropulsed bony fragments in the lumbar spinal canal in comparison to the cervical or thoracic spinal canal, where very little “spare” room is available around the cord. For this reason, even though PSI affecting the lumbar region is relatively common,^27,^53,⁵⁴ spinal infections that produce neurological deficits are seen predominantly with infections of the cervical and thoracic regions.^6,^51,⁵⁵ In addition, neurological deficits are more likely to develop in immunocompromised patients as a consequence of a spinal infection.²⁷

Microbiology

The most common organism causing PSI is Staphylococcus aureus.^4,^6,^8,^9,^25,^30,^38,^{52–54,56–60} S. aureus together with other gram-positive organisms such as Staphylococcus epidermidis, Streptococcus viridans, Streptococcus pneumoniae, Streptococcus faecalis (enterococcus), Propionibacterium, and diphtheroids account for the vast majority of PSIs. Gram-negative organisms such as Escherichia coli, Pseudomonas, Salmonella, Enterobacter, Klebsiella, Haemophilus, and Proteus are less frequent and may be associated with gastrointestinal or genitourinary sources of infection.^15,^50,^60,⁶¹ Infections in intravenous drug abusers are most likely to be caused by staphylococcal species as well,^14,^59,^62,⁶³ but Pseudomonas infection may be relatively more common in this patient group.^27,^29,^55,⁶² ^,64 Rarely, anaerobes such as Peptostreptococcus and Bacteroides may cause spinal infections,⁴ ^,65 also with relatively greater frequency in intravenous drug abusers. Anaerobic infections are more likely than aerobic infections to be polymicrobial. Care should be taken to process all material submitted for culture, both aerobically and anaerobically.⁶⁶

Clinical Findings

The diagnosis of PSI can easily be missed unless a high index of suspicion is maintained.* Given the large number of visits to emergency departments for complaints of back pain, recognizing this comparatively rare, sometimes subtle, but treatable condition is challenging. Additionally, patients with a history of substance abuse, who are especially prone to the development of such infections, can be poor historians and manifest drug-seeking behavior, thereby further confounding matters.⁶²

Early in the course of their disease, patients typically have isolated back pain ^27,⁵³ that they may relate to strenuous activity or a minor injury. Systemic signs of infection such as fever or an elevated leukocyte count may be absent. Such patients may be discharged from emergency departments or physician’s offices without a definitive diagnosis, only to return later with progressive symptoms, deformity, or neurological deficits.

Heusner described the clinical evolution of epidural abscess in four stages: pain, radiculopathy, weakness, and paralysis.⁷ In practice, however, such distinct progression of the infection rarely occurs. Patients can be seen anywhere along a spectrum that ranges from predominantly local manifestations of focal spinal or radicular pain at one end to systemic manifestations of infection (e.g., fevers, chills, malaise, and night sweats) at the other end. The presence of systemic manifestations may herald bacteremia and is an opportune time to obtain blood for culture. Unusual characteristics of the back pain, such as a midline location over the thoracic or upper lumbar spine that worsens with recumbency, especially at night, and the association of a thoracic radiculopathy should prompt consideration of a nondegenerative cause of the pain. Careful screening for risk factors such as intravenous drug abuse, AIDS, diabetes, recent steroid therapy, or the presence of other immunocompromised states helps detect patients who merit further evaluation. Neurological symptoms are more commonly seen with infections of the cervical and thoracic spine than with infections in the lumbar region. The rate of progression of infection and neurological deficits is variable—some infections are relatively indolent, whereas others can progress rapidly and result in profound neurological deficits in just a matter of hours.

On examination, there is usually exquisite tenderness with percussion of the affected spinal segment. When the infection has a prominent bone component, a gibbus deformity may be clinically obvious. A dermatomal level of sensory, motor, or combined deficits is common in patients with deficits. The examination is usually consistent with an acute spinal cord injury with bladder and bowel involvement. Measurement of postvoid bladder residual followed by bladder catheterization provides an objective measure of the urologic dysfunction and prevents secondary injury of the detrusor muscle. Rarely, with relatively indolent infections, chronic compression of the spinal cord may occur and long-tract signs may be present.⁶⁷

Occasionally, patients may initially be seen with florid sepsis ⁶⁸ and an altered level of consciousness and be unable to provide any history. This is somewhat more common in immunoincompetent patients, those in whom medical attention is delayed, and those with a previous, more rostral spinal cord injury who are insensate to the pain and have no neurological function below the infected level.⁶⁹ In such cases, the diagnosis of PSI may be delayed for hours or days until the patient is stable enough to undergo MRI.

Diagnosis

Laboratory Markers

The diagnosis of PSI is suggested by the clinical features described earlier. Laboratory markers of acute inflammation—leukocyte count, erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP)—are helpful in screening patients for further evaluation and in establishing the diagnosis in cases in which the imaging changes are nondiagnostic. Of these markers, the leukocyte count is the least commonly affected,^11,^27,^53,⁷⁰ whereas the ESR is usually affected, often in dramatic fashion.^6,^16,^20,^50,^53,^56,^70,⁷¹ According to Carragee and colleagues, the ESR is elevated in more than 90% of patients with spinal infections.⁷⁰ Measurement of the ESR is inexpensive and reasonably sensitive, although nonspecific. It is an excellent parameter to monitor in determining the response to therapy and should be measured at the initial evaluation even if the diagnosis is already clear. Given its lack of specificity, the ESR should always be interpreted in the context of the patient’s overall condition. This lack of specificity is especially an issue in the management of patients with PSI and a preexisting elevated baseline ESR—such as those with cirrhosis—in which the elevated ESR is refractory to treatment of the infection and it is hard to determine whether the infection is responding. The diagnosis of spinal infection is also strongly suggested by a persistently elevated CRP. Measurement of CRP may be useful in detecting early infections because serum levels may increase within hours of a bacterial infection.^72,⁷³ Although both the ESR and CRP will become elevated after elective spine surgery, CRP normalizes postoperatively much more quickly than the ESR does, thus making it potentially more useful in the evaluation of postoperative patients.⁷⁴

Imaging

The diagnosis of PSI requires that the physician first establish the presence of the infection and, second, arrive at a bacteriologic diagnosis. The first of these goals is dependent on adequate imaging data. The imaging diagnosis is usually established by MRI. Plain radiographs and computed tomography (CT) can also play a role in the diagnosis. Occasionally, MRI is either not feasible or is nondiagnostic. In such cases, the diagnosis needs to be made by an overall interpretation of the patient’s clinical picture, laboratory parameters, and radiographic findings. Additional imaging data such as radionuclide scans may be useful in such cases.

Plain radiography and CT typically show few if any changes during the very early stages of infection.²⁶ These imaging modalities, however, are valuable in the evaluation of more advanced infections in which bone changes are apparent and in the evaluation of vertebral bony integrity and spinal stability in patients in whom surgical management is being considered. Changes that are seen on plain radiographs several weeks after onset of the infection increasingly consist of prevertebral and paravertebral soft tissue volume, loss of disk height, trabecular erosion, and, eventually, destruction of the entire vertebral end plate on either side of a disk. Vertebral collapse, loss of normal lordosis around the affected level, and the development of a kyphotic deformity occur with advanced infection.^75,⁷⁶

CT is generally more sensitive and specific than plain radiographs. Contrast-enhanced CT reveals inflammation of the prevertebral and paravertebral soft tissues (Fig. 276-1), visible as stranding and loss of the normal tissue planes, in infections that have been present for several days. Enhancing epidural collections may also be visible on a high-quality, contrast-enhanced CT of the spine. CT is an appropriate modality for the detection and percutaneous management of psoas abscesses and paravertebral abscesses that result from unchecked progression of the prevertebral and paravertebral components of a PSI. CT guidance is useful for percutaneous aspiration of disk spaces, paravertebral fluid collections, and necrotic bone and to provide specimens for a bacteriologic diagnosis. Myelography followed by CT provides another means of visualizing spinal cord or cauda equina compression in situations in which MRI cannot be performed.

FIGURE 276-1 Paraspinous muscle involvement in a patient with a pyogenic spine infection as seen on an abdominal computed tomographic scan obtained after the administration of contrast material.

MRI is the diagnostic test of choice for the detection of PSI and should be performed in all patients unless contraindicated. Imaging should be carried out without and with the administration of paramagnetic contrast agents. Unenhanced T1-weighted images reveal a hypointense signal in the vertebral body, especially at the end plates; the normal hyperintense fat signal in the vertebral bone marrow is lost. Disk height is reduced and may be markedly diminished. T2-weighted imaging reveals high signal (edema) in the disk space and occasionally in the bone and paravertebral soft tissues.^8,^57,⁷⁷ Gadolinium-enhanced T1-weighted imaging is perhaps the most diagnostic MRI sequence (Fig. 276-2)—enhancement of the vertebral end plates, the vertebral body, the prevertebral and paravertebral soft tissues, and the epidural space can be seen.⁷⁶ The entire spine should be imaged if an infection is detected because much like metastatic tumors, spinal infections can occasionally be multifocal (Fig. 276-3 to 276-5).^78,⁷⁹

FIGURE 276-2 Gadolinium-enhanced T1-weighted imaging with fat suppression in a patient with L4-5 diskitis. There was no osteomyelitis or epidural involvement. The patient was treated with intravenous antibiotics, which led to resolution of all the symptoms.

FIGURE 276-3 A and B, Axial and sagittal magnetic resonance images of a patient with a large, dorsal epidural abscess that extended from T11 to the sacrum. The patient complained of urinary retention and back pain. The abscess, which grew coagulase-positive staphylococci, was drained via laminectomy.

FIGURE 276-4 Typical findings of a spinal infection with osseous involvement as seen on magnetic resonance imaging. There is hypointense signal in the vertebral body on T1-weighted images (A and D) and enhancement of the vertebral end plates, prevertebral and paravertebral soft tissues, and epidural space after the administration of contrast material (B and E). T2-weighted imaging reveals increased signal (edema) within the disk space, bone, and paravertebral soft tissues (C).

FIGURE 276-5 A and B, Magnetic resonance image of T8-9 diskitis and destruction of the adjacent end plates. Computed tomography–directed biopsy was positive for Enterococcus species.

Occasionally, it is not feasible to evaluate patients with MRI because of the presence of incompatible devices such as cardiac pacemakers, ferromagnetic aneurysm clips, or shrapnel, which constitute absolute contraindications. Other problems such as patient size, claustrophobia, ongoing mechanical ventilation, or other monitoring may render the performance of high-quality MRI difficult or time-consuming. Furthermore, MRI may be less specific in patients with extensive or pronounced spondylosis or in those who have suffered recent trauma. Finally, the scans of some patients with MRI-compatible, metallic implants are affected by sufficient artifact to render them nondiagnostic. In all cases in which MRI is either not possible or not diagnostic, CT myelography and nuclear medicine scans should be considered.

Radionuclide studies have a high degree of sensitivity in early infection. Gallium 67 and technetium 99 both have reasonable sensitivity for the detection of PSI. Gallium binds to iron-binding proteins at the site of inflammation, whereas technetium reflects blood flow to the bone. Gallium scans are therefore more specific than technetium scans for PSI, but both of these scanning methods are not very specific.⁸⁰ Focal uptake can be seen with spondylosis, after trauma, and in tumors. In comparison, scans using white blood cells tagged with radionuclide are more specific for the detection of infection, but their sensitivity is much lower.⁸¹ Leukocytes from the buffy layer of the patient’s blood are tagged with indium 111 and reinjected into the patient (Fig. 276-6). These labeled cells localize to sites of ongoing inflammation; focal uptake in the spine strongly suggests the diagnosis of a PSI, although false-positive results can occasionally be seen with a few conditions such as tumors, especially hematogenous malignancies involving the spine. Chronic infections can occasionally lead to false-negative results with indium scanning. Fluoro-2-deoxy-D-glucose (FDG)-labeled positron emission tomography (PET) has proved to be highly sensitive for spinal infections, although it is also relatively nonspecific.⁸²^–⁸⁴ Recently, scans using indium-labeled biotin have been used for the diagnosis of osseous infections.⁸⁵ Biotin, a growth factor widely used by bacterial species, shows early promise as a more specific indicator of infection. Various other potential peptide markers are being investigated to aid in early diagnostic imaging.

FIGURE 276-6 A, Magnetic resonance imaging was not diagnostic of a pyogenic spine infection. B, The diagnosis was made with an indium 111–tagged white blood cell scan that revealed focal uptake.

Bacteriologic Diagnosis

The second component of the diagnostic evaluation is bacteriologic characterization of the infection. As noted earlier, the spine may be hematogenously seeded from other sites of infection such as the respiratory tract, urinary tract, or an endovascular source. Cultures of urine and sputum should be performed in patients with these potential sources. The causative organism is typically isolated from either blood or spinal tissue. Blood should be obtained for culture in all cases, and if possible, multiple sets should be collected to coincide with spikes in the patient’s temperature. If the blood cultures are positive, the causative organism is identified in 25% to 59% of cases.⁸⁶ Appropriate therapy may be started without the need for further, more invasive testing. Unfortunately, blood cultures are negative in 40% to 75% of cases, possibly because some infections are indolent, although most commonly in patients who have received antibiotics before blood is drawn for culture. In such cases, biopsy of the vertebra or the disk space under CT or fluoroscopic guidance has a higher rate of success in culturing the organism. This may need to be repeated if no growth results from culturing the aspirate. A larger bore needle that obtains a core of tissue may yield microbiologic results superior to that of aspirates of fluid from bone. The use of a nucleotome for percutaneous suction-aspiration of the infected disk space has also been described.^86,⁸⁷ Biopsy specimens consisting of a core of tissue can also be submitted for histopathologic analysis to confirm the diagnosis. Closed biopsy techniques have reported accuracy rates ranging from 60% to 100%. If these measures fail and the imaging diagnosis is reasonably definitive for PSI, open biopsy may be necessary. This is best done in an operating suite under fluoroscopic guidance. A bacteriologic diagnosis is made in about 80% of open biopsies.⁸⁶

It is difficult to establish a bacteriologic diagnosis if empirical antibiotic therapy is begun before obtaining blood for culture. It is important to exert restraint and withhold the administration of antibiotics until it is clear that an organism has been cultured, unless the patient is septic or has major systemic manifestations, in which case delaying therapy may be inappropriate.⁶¹ Even a single dose of a broad-spectrum intravenous antibiotic may significantly decrease the probability of culturing an organism. In patients with neurological deficits, antibiotics should be withheld until specimens are collected intraoperatively. If antibiotic treatment is initiated before a bacteriologic diagnosis is established in patients who are neurologically intact, it may be reasonable to terminate this empirical therapy and obtain fresh samples for culture without the influence of antimicrobials. In a small number of cases, no organism can be cultured despite multiple attempts.^6,⁸⁷ Mycobacterial or fungal infections should be considered in such cases. Once this possibility is reasonably excluded, empirical antibiotic therapy is the only option. Patients treated empirically with antibiotics need to be monitored closely to confirm a response to the treatment being administered.

Diagnosis of Associated Conditions

Patients with PSI are at risk for concomitant pyogenic infections elsewhere in the body. Because the risk for bacterial endocarditis is especially high, patients should be examined for cardiac murmurs and evidence of embolization to the retina and the skin. If endocarditis is suspected, echocardiography, preferably via a transesophageal route, should be performed to look for valvular vegetations.

Frequently, a cutaneous pyogenic lesion that represents the index location of the infection is still present when the patient is evaluated by a neurosurgeon.⁸⁸ Such lesions can often be used to obtain material for culture to make a bacteriologic diagnosis. Patients should also undergo assessment for risk factors responsible for the infection. Certain risk factors may be obvious, whereas others may be discovered only after careful evaluation. For instance, patients may be reluctant to admit to intravenous drug abuse. In some cases of indolent infection, the drug abuse may even have occurred several months previously. It is reasonable to offer intravenous drug abusers testing for human immunodeficiency virus (HIV) and type B and C viral hepatitis, as well as appropriate counseling together with such testing.

It should be emphasized that evaluation of a PSI is incomplete unless the patient is assessed for extraspinal manifestations of the infection. Infections can track along fascial planes adjacent to the infected vertebrae and result in psoas abscess, paraspinous muscle abscess, empyema, sympathetic pleural effusion, and retropharyngeal abscess.⁸⁹ Infections can breach the dura spontaneously or with the unintended assistance of a biopsy or lumbar puncture needle. Changes in mental status, nuchal rigidity, or emesis in a patient with a known or suspected PSI should lead to the consideration of meningitis or parameningeal inflammation as a diagnostic possibility. The diagnosis of concomitant meningitis is best confirmed by obtaining cerebrospinal fluid (CSF) from a cisternal rather than a lumbar puncture—unless the locus of spine infection is clearly remote from the site of the puncture.⁶²

Differential Diagnosis

Occasionally, patients with severe back pain and laboratory studies consistent with acute inflammation have spinal MRI scans that fail to reveal changes diagnostic of a PSI. In such cases it is prudent to wait for definitive imaging changes to develop or to obtain radionuclide scans if the diagnosis of a PSI is strongly suspected. It is also essential to consider alternative diagnoses that can mimic PSI.

Conditions that can have clinical findings similar to PSI but can be differentiated with imaging include pyogenic arthritis of the hip, septic or autoimmune sacroiliitis, pyelonephritis, primary psoas abscess, autoimmune spondylitis, spinal trauma, osteoporotic compression fractures, spinal epidural hematoma, spontaneous spinal subarachnoid hemorrhage, and leptomeningeal metastatic disease. Certain conditions may occasionally be a little harder to distinguish with imaging alone. Nonpyogenic (tubercular or fungal) spinal infections can occasionally mimic PSI rather closely, as can tumors metastatic to adjacent vertebral levels. Involvement of the vertebral body more than the disk space and the development of paravertebral abscesses rather early in the course of the infection suggest a tubercular rather than a pyogenic etiology.⁷⁰ A definitive diagnosis can usually be established by needle biopsy in such cases. Tubercular infection and lymphomas should always be considered in the differential diagnosis in patients with AIDS.⁹⁰ The degenerative changes seen in patients with advanced spondylosis⁹¹ can at times be confused with infections because both are preferentially localized to the vertebral end plates. They can generally be differentiated from infection; a degenerated disk is usually dehydrated and therefore hypointense, whereas an infected disk is hyperintense on T2-weighted imaging. Enhancement of the disk itself is also indicative of PSI, but enhancement of vertebral bone can be seen with either entity. The presence of gas within the disk, the vacuum disk phenomenon, is much more suggestive of degeneration than infection.⁷⁶ A rare entity that may mimic PSI is avascular necrosis of the vertebral body, which is usually associated with significant collapse of the vertebral body and intravertebral vacuum clefts. Changes in the intravertebral vacuum clefts are seen as a consequence of spinal loading and unloading. T2-weighted imaging performed immediately after the patient lies supine on the scanner reveals a hypointense signal because of the presence of air, but as fluid enters the cleft, this signal becomes hyperintense.^92,⁹³

Management

Once a diagnosis of PSI has been established, an important decision to be made is whether operative intervention is indicated. Although all patients need antibiotic therapy, the term “medical management” in the present context refers to the use of antibiotics without planned surgical débridement at the site of infection. “Surgical management,” in contrast, implies operative intervention consisting of débridement of necrotic tissue, decompression of neural elements, correction of deformity, fusion, and instrumentation as indicated by the clinical situation, along with appropriate antimicrobials.

Surgical Treatment

The decision to proceed with surgery should be made after consideration of the patient’s neurological status, vertebral level of involvement, extent of vertebral destruction, and findings on MRI. In the past, the decision to undertake emergency surgery was often made solely on the basis of an enhancing epidural component. The guiding principle has been that an epidural abscess constitutes a neurosurgical emergency. It has become increasingly clear that there is heterogeneity in the composition of epidural collections. Entirely liquid “abscesses” are rare, and in most cases a phlegmon with minimal, if any, liquid abscess is seen (Fig. 276-7). Such heterogeneity also applies to the clinical manifestations of epidural collections—some produce rapidly progressive neurological deficits, whereas others produce no deficits. Furthermore, as discussed in the section on the pathogenesis of PSI, neurological deficits occur more often as a result of spinal instability or deformity than as a result of compression of the cord or cauda equina by an abscess component. Thus, there is often poor correlation between an imaging diagnosis of “epidural abscess” and the development of neurological deficits. This lends further credence to the notion that management decisions in PSI are best made by taking into account multiple clinical and imaging criteria rather than a simple anatomic classification of the infection.

FIGURE 276-7 A-C, Osteomyelitis of C5 and C6 in a 51-year-old woman with neck pain of 2 weeks’ duration and quadriparesis. Magnetic resonance imaging shows a prevertebral abscess and cord compression from a ventral phlegmon. Anterior débridement and reconstruction with an expandable cage, allograft, and an anterior plate resulted in solid fusion with complete resolution of the neurological symptoms.

The presence of neurological deficits is the most important of these criteria. Emergency surgical intervention should be considered in all patients with neurological deficits, regardless of the duration of the weakness, unless the deficits are minimal (e.g., radiculopathy) or the patient’s medical comorbid conditions (coagulopathy, sepsis) preclude rapid surgical intervention. If a patient has minimal neurological deficits and it is decided to proceed with nonsurgical management, the patient must be carefully monitored to detect any early progression. Deficits may sometimes progress rapidly in a few hours.

Surgical intervention for neurological deficits needs to address the location of the compressive lesion, such as ventral or dorsal to the spinal cord or cauda equina. Simplistic though this sounds, ignoring this principle may result in destabilization of an already compromised spine, with worsening deficits.^39,^49,^51,⁹⁴ The nature of the compressive lesion—liquid pus versus a mass of granulation tissue or retropulsed bone—is also an important consideration in determining the optimal surgical approach. Although pus may be accessed and drained by various routes, simple laminectomy does not adequately afford decompression of a solid ventrally situated extradural lesion and can exacerbate the deficits produced by a kyphotic deformity. Finally, the various anatomic regions of the spine dictate the potential approaches available and the likelihood of postoperative instability.

In the cervical spine, the surgical approach usually coincides with the location of the compressive lesion (i.e., an anterior approach for ventral compression and a posterior approach for dorsal compression).^9,⁵⁵ An exception may be cited for ventral abscesses without major bone involvement extending over more than two or three levels. In these cases, pus can usually be drained from a posterior approach without the morbidity of multilevel corpectomy and fusion. Infections of the odontoid, though rare, have been reported. If deemed to have produced instability, they are best managed by occipitocervical fusion and a transoral biopsy or decompression of the thecal sac. Stable lesions at this level can be managed medically.^95,⁹⁶ At the cervicothoracic junction and in the upper thoracic spine, anterior lesions may be difficult to access from a ventral approach because of the presence of the great vessels. Although partial sternotomy or manubrial resection may provide adequate access in such cases, technical challenges with débridement and reconstruction of the anterior column remain. Furthermore, a kyphotic deformity produced by the infection can make access to the apex of the deformity via an anterior approach more difficult. Transpedicular, lateral extracavitary,⁹⁷^–⁹⁹ or periscapular^100,¹⁰¹ approaches may be used in these cases. These approaches can be used to decompress the ventral aspect of the spinal cord, and potential or apparent segmental instability can be addressed by concurrent posterior thoracic fusion with instrumentation.^39,^49,⁹⁴ Recent technologic advancements have increased the options available for fixation over the cervicothoracic junction.

Surgical approaches for treating PSI of the midthoracic spine are best tailored to the site of compression. Thoracotomy approaches offer excellent visualization of the ventral and ventrolateral aspects of the spinal canal. Anterior reconstruction after vertebrectomy is readily performed via this exposure. Alternatively, the lateral extracavitary approach ⁹⁷^–⁹⁹ or the retropleural approach¹⁰¹ can be used. The temptation to perform a laminectomy for ventral disease in the thoracic spine, other than liquid pus, should be resisted because it can result in the cord being draped over the compressive lesion along with concomitant loss of the stability offered by the posterior tension band.^49,^51,⁹⁴ The extent of spinal instrumentation required to restore stability is a function of the number of segments involved, the degree of kyphotic deformity, the patient’s bone stock, and the integrity of the posterior tension band.

In the lower thoracic and upper lumbar spine, anterior débridement via a thoracoabdominal approach affords excellent exposure for resection of the involved vertebral bodies and reconstruction of the anterior and middle columns.^22,^27,⁵⁶ When the posterior elements are intact, an anterior approach alone may suffice. Anterior débridement and fusion followed by posterior instrumentation and posterolateral fusion^102,¹⁰³ may be an option in selected patients in whom concern for appropriate placement of instrumentation from the anterior approach used for the decompression is especially high. Infections of the middle and lower lumbar spine may be approached through either a retroperitoneal or a transperitoneal approach for débridement and anterior reconstruction. Below the conus, a posterior approach can be used to decompress the neural elements; however, reconstruction of the anterior and middle columns is difficult with this approach. Transpedicular instrumentation can provide a measure of stability in such cases, but it may occasionally fail if anterior column reconstruction is not performed. After fusion and instrumentation for spinal infections, an external orthotic device appropriate for the level in question should be prescribed for approximately 3 months.

In addition to the treatment of patients with neurological deficits, surgical intervention is also indicated for the management of those who have failed medical therapy, the treatment of chronic pain after medical management, and the treatment of patients with prominent deformity or overt instability.^6,^32,¹⁰⁴ Relapses of infection can be treated either with a second course of antibiotics or by surgery, depending on the clinical scenario and the patient’s preference. It has been postulated that relapses occur because of the presence of necrotic bone (sequestra) within a vertebral body that lacks blood supply and thus provides a nidus for persistent infection. Surgical treatment in these cases therefore includes débridement of the vertebral body that is infected (i.e., corpectomy), followed by reconstruction (Figs. 276-8 and 276-9). Finally, some surgeons tend have a bias toward immediate surgical management of patients with a prominent bony component of their infection because they believe that bacteriologic cure rates are then higher and thus a cure is effected earlier. This bias still remains to be validated by clinical trials. However, the fact that aggressive débridement eliminates a sizable portion of infected and necrotic bone lends credence to this viewpoint.

FIGURE 276-8 A-C, Destruction of the third and fourth cervical vertebrae by an Enterobacter cloacae infection in an intravenous drug abuser. The spine was grossly unstable, and C3-4 corpectomy was performed, followed by fusion with an iliac crest autograft and an anterior cervical plate. The postoperative radiograph (D), which was obtained at 2 years, demonstrates some subsidence but solid incorporation of the graft and a stable spine.

FIGURE 276-9 A and B, Pyogenic infection caused by Staphylococcus aureus at T9-10 in a 38-year-old man with paraparesis. Treatment consisted of anterior débridement, T9 and T10 corpectomy, and reconstruction with a Harms cage, autograft, and a Kaneda-type construct.

Timing of Surgery

Surgery needs to be performed on an emergency basis in patients with rapidly progressive neurological deficits.⁵⁸ In cases in which the deficits have been slowly progressive or the patient has significant medical comorbidity (septicemia, coagulopathy, endocarditis with cardiac failure), the timing of surgical intervention should be more carefully considered. Surgery can be carried out on an elective basis in patients with no deficits, in whom the goal is to stabilize the spine or débride a necrotic focus of infection.

The role of steroids, at dose levels used for the acute management of spinal cord injury,^105,¹⁰⁶ is questionable in that there are no prospectively collected data to address the issue, nor are there likely to be any given the relatively small numbers of such cases seen at any one institution. Our experience, though anecdotal, suggests that the use of steroids at high doses in patients with a significant neurological deficit is safe in the perioperative period.