Spinal Canal Infections
Although uncommon, infections of the spinal canal have profound clinical implications and pose a risk for paralysis. Their early diagnosis and prompt, effective treatment are therefore very important in maintaining neurologic function and quality of life. Improvements in radiographic imaging, most notably the widespread availability of magnetic resonance (MR) imaging, allow such conditions to be localized because they typically cause the gross structural abnormality of an abscess rather than the more subtle microscopic findings of meningitis. These infections have actually become more common in recent decades as a consequence of the increasing number of patients with chronic or immunosuppressive illness whose life expectancy modern medicine has extended.
The spectrum of such disorders falls into three categories: (1) spinal epidural abscess (SEA), typically bacterial in origin and the most common of the three; (2) spinal subdural abscess, which is much less common; and (3) intramedullary abscess of the spinal cord, which is also rare, although more often reported than subdural abscess in the spinal compartment. Each of these infections is generally pyogenic (i.e., producing a purulent and mainly neutrophilic infiltrate), but infections caused by mycobacteria, fungi, and Nocardia species are also occasionally seen. This chapter specifically excludes discussion of focal infection of the vertebrae, known as vertebral osteomyelitis, which is addressed in a separate chapter.
Spinal Epidural Abscess
Infections in the epidural space can be isolated but are often associated with infection in the vertebral body, disk space, or both. Thus, an anatomic distinction is somewhat artificial and not the principal factor in determining therapy, although an important distinction is that the medical cure of osteomyelitis requires a longer duration of antibiotic administration than does infection of the disk or epidural space. The first pathologic description of a bacterial vertebral osteomyelitis was published in 1820 by Bergamaschi. By the 1890s, surgery had been reported for pyogenic vertebral infections by Ollier and Chipault; in 1896, Makins and Abbott associated vertebral infection with epidural suppuration and advocated both removal of the infected bone and evacuation of the epidural pus.1 This concept was also promoted by Ramsay Hunt in 1904 in the first paper devoted to establishing epidural abscess as a separate entity.2 The first attempt at surgical drainage of an isolated SEA without vertebral involvement was performed by Delorme in 1892; the patient succumbed to bacterial endocarditis quickly thereafter, and functional recovery after successful drainage was first reported in 1901 by Barth.3 These early papers promoted SEA as a distinct condition and began the trend toward surgical drainage of such abscesses, which continues to this day. Because such abscesses are now frequently diagnosed in immunocompromised and elderly patients, many of whom are medically fragile, there has been a reassessment of the need for surgery in all cases and a reconsideration of medical treatment in selected patients, which is now performed with some success ( Table 13.1 ).
Anatomy
The spinal epidural space under normal conditions is filled with fat, a loose areolar connective tissue network, and an exuberant venous plexus. This space sits between the spinal dura and the bony ring that forms the vertebral canal. The dura is relatively adherent to the posterior longitudinal ligament in the anterior aspect of the canal, and thus the dorsal and lateral components of the epidural space are more prominent and are the typical sites for abscess. Because of the larger diameter of the spinal cord in the cervical spine, the space is minimal in the neck; however, it enlarges in the midthoracic region (T4–T8), where it has a depth of 5 to 7 mm, then narrows gradually to accommodate the lumbar cord enlargement, with progressive widening caudal to L2 until the dura ends at S2. It has been suggested that this anatomic variation promotes the formation of SEAs in the thoracolumbar area. However, the distribution of SEAs is relatively even along the spine. Epidural abscesses found in the anterior epidural space are almost always associated with diskitis or vertebral osteomyelitis extending past the posterior longitudinal ligament into the epidural compartment. Because the anterior dural attachments are stronger, epidural abscesses in the anterior space tend to be more limited in extent, whereas those in the posterior aspect of the canal extend over more segments. A typical SEA covers three to five segments, but many more can be involved.
Epidemiology
The incidence of SEA has risen in recent years, likely through a combination of increased detection, increased life expectancy for immunocompromised patients, and aging of the population. The incidence is variously reported as between 0.2 and 2 cases per 10,000 hospital admissions.4,5 The elderly are prone to such infections, with the peak incidence in the sixth and seventh decades; SEAs are also relatively uncommon in the pediatric age range. The two most common correlates are frequent bacteremia and significant impairment of immune function. Both acquired immunodeficiency syndrome (AIDS) and intravenous drug use are thus very significant risk factors for SEA, as are diabetes mellitus, end-stage renal disease, and hepatic cirrhosis. In addition, any medical treatment that uses immunosuppression to a therapeutic end or provokes it as a side effect (including long-term steroid therapy, chemotherapy, and suppression of the antigraft response in transplant recipients) can promote SEA.6,7 In such patients, the spontaneous bacteremia that all people harbor from time to time (and that is cleared by competent immune mechanisms) is more likely to progress unchecked, resulting in SEA.
Microbiology
The most common causative organism is Staphylococcus aureus,8 which in conjunction with other gram-positive microbes (e.g., Staphylococcus epidermidis, Streptococcus pneumoniae, Streptococcus viridans, Enterococcus, and Propionibacterium acnes) causes most epidural abscesses. Gram-negative species are less often cultured from SEAs, but abscesses containing Escherichia coli, Pseudomonas aeruginosa, Salmonella, Klebsiella, and other species are sometimes found after infection of the urinary or gastrointestinal tract. Anaerobes rarely cause spinal epidural infections in the general population but sometimes cause infection in intravenous drug abusers. Unfortunately, 10 to 20% of patients show no growth on culture of the abscess material, even in the absence of prior antibiotic therapy.6,7
Streptococcus species are the second most common bacteria (after Staphylococcus species) to cause SEA and are often seen in association with recent pneumonia or after dental procedures. The gram-negative species dominate in intravenous drug abusers, while Mycobacterium tuberculosis is quite common in Asia and Africa, and such cases are starting to seep into North America as well. Fungal infections are also becoming slightly more common.
Mechanisms of Pathogenesis
The venous route of infection from pelvis to vertebral column, proposed by Batson in 1957, has had much traction over the years but is now largely discredited. Hematogenous dissemination of bacteria likely occurs most often through arterial transfer, particularly to the metaphyseal area near the anterior longitudinal ligament. Abscesses in the epidural space can arise through bacterial embolization to this part of the vertebral body with secondary propagation into the adjacent epidural space, or through direct spread to the epidural space posteriorly. Adjacent foci of infection, such as pulmonary, retropharyngeal, or intraabdominal abscess, are also occasional sources of SEA.9 Infection and thrombosis of the nutrient metaphyseal artery produce avascular necrosis of bone and thus create a nidus for infection. Ischemia of the disk occurs through the same mechanism, rendering it more susceptible to infection. Both bone and disk can act as sources of infection in the neighboring epidural space. Some infections may emanate from the epidural veins alone, thus allowing involvement of several spinal segments without concomitant infection of bone or disk.
Bacterial contamination can occur directly in patients undergoing open surgery with exposure of the epidural space. The dead space left after instrumentation is a particularly suitable environment for the growth of bacteria implanted at the time of surgery. Foreign bodies placed in the disk space (e.g., epidural stimulators or catheters) can also carry bacteria into this compartment, as can penetrating injuries or direct breakdown, as occurs in a decubitus ulcer.
The development of neurologic deficits follows the onset of SEA in 5 to 50% of patients. Such deficits occur through direct compression caused by an expanding abscess; this is further exacerbated when spinal deformity occurs as a consequence of biomechanical impairment of the spine as infected bone leads to vertebral wedging. It has long been noted, however, that cord dysfunction on a clinical level is often significantly more evident than the degree of compression alone should cause. Some patients acquire a central cord syndrome that is hard to explain as a result of mass effect.4 In addition, the rapidity and occasional irreversibility of the neurologic deterioration associated with the onset of SEA suggests that a vascular mechanism underlies these clinical phenomena.
Autopsy studies by Russell et al showed thrombosis of the epidural arteries and veins as well as venous infarction and edema within the cord in patients with SEA.10 However, experimental model systems (as exemplified by the rabbit model of Feldenzer et al) show an explicit lack of microangiopathy in the cord except in animals with severely compressive lesions, and they found no thrombosis or venous occlusion.11–13 Such findings argue in favor of the compressive theory, yet the clinical course of the disease in humans does not fully support it. This contradiction may simply reflect the limits of the experimental rabbit model in mimicking the human clinical situation, as well as the heterogeneity of such infections in their microbial pathogenesis and the underlying clinical conditions that predispose patients to SEA. Unfortunately, this ambiguity of etiology and of substrate has prevented a consensus on the appropriate method of therapy across all groups.
Clinical Features
A broad range of symptoms can signify an SEA, and classic features are not always present. The diagnosis can be missed unless the index of suspicion is high.14 Although a very small fraction of patients with back pain harbor a spinal infection, a pattern can be gleaned that should point to the diagnosis of epidural abscess and drive the performance of studies to determine whether one is present. The first symptom of an SEA is isolated back pain, and more generalized infection is often absent. Because the symptoms are clarified over time, it is typical for patients to seek medical attention repeatedly before the diagnosis is made. Symptoms usually progress through four stages: (1) an aching pain in the spine, (2) root pain or radiculopathy, (3) weakness, and (4) paralysis.11 Usually, a patient′s symptoms do not follow this pattern perfectly and may consist either of local pain manifestations or systemic signs of infection (fever, malaise, night sweats). The interval between the onset of pain and the onset of neurologic deficit also varies.4,15,16 The combination of fever and back pain in a patient with immunocompromise or a history of drug abuse should lead to the suspicion of SEA. The back pain is not always associated with local tenderness of the spine, and the pain is usually severe. Fever is often slight and may be masked by the patient′s use of analgesics for the back pain or through the prior use of antibiotics. In some, septicemia occurs, and this can be the primary presentation. One-third of patients are neurologically intact at presentation. The rate of progression of the neurologic deficits varies greatly, with some advancing slowly and others causing paralysis within hours.
About 75% of patients have signs of spinal cord or nerve root dysfunction when the diagnosis is made. Monoparesis is common, and paraparesis or quadriparesis is occasionally seen. Patients with severe back pain or encephalopathy may be more difficult to examine, and confusion is noted in 15 to 35% of patients. Loss of bowel or bladder control and sensory deficit can also occur, but a complete loss of sensation below a sensory level is usually seen only with profound and often irreversible loss of motor function. In patients with urinary retention or incontinence, measuring postvoid residual volumes with subsequent Foley catheter placement provides clarity as to bladder function.17
The presence of fever and back pain, particularly in immunocompromised patients or in those with a history of intravenous drug abuse, should prompt testing for SEA, especially when recent bacteremia is documented.
Secondary extraspinal infection is occasionally seen in people with SEA. Infection in a vertebra can seep into the adjacent fascial layers to cause abscess in the psoas or paraspinal muscles, can cause retropharyngeal abscess, and sometimes provokes a sympathetic pleural effusion.
Diagnosis
Laboratory Studies
Serum markers of acute inflammation are helpful in screening patients for SEA. It is important to remember, however, that at least 10% of patients later proven to have an epidural abscess show no inflammatory markers. The three markers generally used are the white blood cell count, the erythrocyte sedimentation rate (ESR), and the C-reactive protein (CRP) level. Most sensitive is the ESR, elevated in 90% of patients with spinal infection, often dramatically so.18 This readily available test can be used to monitor response to therapy, and it should always be measured during initial evaluation. However, it is not particularly specific, and some patients have an elevated basal ESR from other causes even before onset of the infection. One such example would be a patient with hepatic cirrhosis. Such patients are prone to acquiring SEA; however, their ESR elevations do not correct even when treatment of the infection is successful. The CRP level is less sensitive but somewhat better in the detection of early infections because serum levels rise within hours after the initiation of a bacterial infection. It is also useful in the postoperative setting in patients suspected of having an SEA after spinal surgery. Both the ESR and the CRP level rise after spine surgery, but the CRP level returns to normal more quickly and thus may have more utility in the work-up of patients in the postoperative period. The ESR can take 2 to 3 months to normalize after successful treatment ( Fig. 13.1 ). Lumbar puncture is not usually performed when an epidural infection is suspected because it may inoculate infected material into the subarachnoid space and because the cerebrospinal fluid (CSF) from most patients with SEA demonstrates evidence of a nonspecific parameningeal process, with a high protein level and a mild leukocytosis.
Imaging
Confirmation of the presence of infection and determination of its site in the spine are usually achieved by MR imaging. The multiplanar capacity of MR imaging and its ability to distinguish infection from other pathology make it the imaging method of choice. Many patients will have had preceding plain X-rays or computed tomography (CT) scans. CT is most useful when it includes three-dimensional reconstructions of the spinal anatomy. The main findings of SEA on plain films are indirectly associated anomalies of the bone or disk, including narrowing of the disk space, erosion of the vertebral end plate, and lytic erosion of the vertebral bodies. Plain films can also show spinal alignment and thus may be useful to establish a baseline status. Bone scans (radionuclide scintigraphy) are not very specific and have a sensitivity of only 70%,15 although they sometimes show early foci of osteomyelitis before plain films do. Gallium 67 and technetium 99 tracers both have reasonable sensitivity, with gallium slightly more specific than technetium as a tracer, but neither has high specificity over all. Sensitivity can be raised by performing a tagged white blood cell scan, but even this shows a lack of sensitivity that impairs its utility. Other markers, including biotin labeled with indium-111, are now being investigated as more specific indicators of infection.19 Plain films and CT show no change during the earliest stages of infection. It takes several weeks for a disk to lose height and for trabecular erosion to occur on either side of a disk from which infection will ultimately spread to the epidural space. Vertebral collapse and loss of lordosis occur only when infection is advanced.
CT is most often used in guiding the percutaneous aspiration of disk spaces, vertebral bodies, or paraspinal fluid collections to provide microbiological identification of infection. CT myelography is still occasionally needed for patients in whom metallic implants degrade the quality of MR images. With the advent of open MR imaging, the previous contraindication of obesity is no longer as important in driving patients toward CT. The absence of ionizing radiation in MR imaging is another advantage of this modality compared with CT.
Contrast-enhanced MR imaging should thus be performed in all patients suspected of having an SEA unless contraindications exist. Imaging before and after the infusion of gadolinium will provide images that are sufficient to distinguish between different pathologies and to determine their location. T1-weighted images before contrast show bone infection as a hypointense signal, with particular focus on the end plates, as well as loss of the normal hyperintensity caused by fat in the vertebral marrow. On T2-weighted images, hyperintensity represents edema in the disk space, which is sometimes found in the vertebra and paravertebral region as well. In post-contrast studies, T1-weighted images show enhancement of the vertebral body, its end plates, the paravertebral soft tissues, and the epidural space20 ( Fig. 13.2 ). Because spinal infections are sometimes multifocal, the entire spine should be imaged.
Identifying the Pathogen
Once imaging has suggested an epidural abscess, the causative agent(s) must be identified to direct the antibiotic choices. Because hematogenous spread from other infected sites is the usual cause of SEA, cultures of urine and septum should be obtained, as should blood cultures in multiple sets. Ideally, blood cultures should be done during a septicemic phase or when temperature spikes occur. Blood cultures identify the organism in only half of cases, however, so needle biopsy of the involved vertebral body or disk space should be carried out under radiographic control. Once this is done, broad-spectrum antibiotic coverage of the usual pathogens (Staphylococcus species and Streptococcus species) should be initiated. Percutaneous biopsy yields material for successful culture in 60 to 80% of cases. When percutaneous attempts fail to yield a positive specimen, open biopsy is the next step and has a similar rate of success. Withholding antibiotics until all necessary material has been taken for culture will maximize the success of such cultures. If cultures are negative in the face of obvious clinical infection, broad-range polymerase chain reaction (PCR) techniques can be used to enhance the sensitivity of detection.20 When no bacterial organism can be identified after several attempts, mycobacterial and fungal infections should also be considered, particularly in at-risk patients.
Treatment
Surgery
The mainstay of treatment for SEA has traditionally been surgery, and it remains so despite inroads that have been made by medical therapy in recent years.6,7 Factors to be considered in the choice of treatment (surgical versus medical) include degree of sepsis, neurologic status, extent of cord compression, degree of vertebral destruction, and presence of spinal instability. Most epidural abscesses require urgent although not emergent care; however, any patient with significant myelopathy should receive prompt surgical treatment. The heterogeneity of the clinical course in SEA makes it difficult to standardize treatment decisions. Delay in treatment can be entertained only when neurologic deficits are minimal or when comorbidities such as sepsis and coagulopathy require correction before surgery. Careful and frequent monitoring should be used for any patient undergoing medical management alone because deficits can progress rapidly and so may require a surgical strategy instead.
Planning surgery requires consideration of the level or levels of spinal involvement, the degree of cord compression, the location of the abscess within the spinal canal (anterior versus posterior), and the physical nature of the material compressing the cord. The strategies for dealing with pus and retropulsed bone are quite different. In addition, simple laminectomy may be effective for washing out a posteriorly or posterolaterally situated phlegmon and for providing cord decompression, but it may also worsen spinal instability when vertebral involvement is profound at the same level. The goals of surgery are decompression of neural structures, including cord and roots; removal of purulent material from the epidural space; removal of any granulation tissue present; and, when necessary, correction of deformity and instability caused by disruption of the vertebral column.
The great majority of patients are still treated primarily with surgery, with antibiotic therapy to follow for a period of several weeks.21 In general, the probability of success is maximized by the removal of as much infectious material as possible. Doing this may induce or worsen instability; therefore, the surgeon must be prepared to restabilize the spine at the same sitting. Tissue adherent to the dura should not be removed too vigorously because breaching the dura may allow the infected material to be introduced into the subarachnoid space, resulting in meningitis. Once the solid material has been removed from the epidural compartment, copious irrigation with antibiotic solutions is performed. With anterior decompression, a fusion will typically be required, which is commonly done with an interposition autograft or allograft. When necessary, reconstruction and stabilization can be delayed to a second-stage procedure. It is also important to remember that in patients with only a slight loss of lordosis or slight kyphosis, fusion may occur over time during antibiotic therapy without instrumentation. Therefore, when it is not certain that instrumentation is necessary, delaying a decision on its placement (and using an external brace in the meantime) is reasonable.
Further treatment difficulties arise when the abscess is largely ventral. In the cervical spine, a ventrally placed abscess without adjacent bony involvement can usually be treated safely by laminectomy with maintenance of the facet joints. When vertebral body involvement is more extensive, the best option is generally a formal diskectomy and vertebrectomy via an anterior approach, followed by bone grafting or cage placement with anterior plating. Trough corpectomies can be used to limit bone removal and preserve the biomechanical stability in cases of ventral SEA.22 In the thoracic spine, a transpedicular or lateral extracavitary approach to an anterior phlegmon may be preferable, often with posterior instrumentation being necessary. Anterior infections within the midthoracic or low thoracic spine respond well to transthoracic approaches performed through a thoracotomy, but the lateral extracavitary approach can be utilized in these cases as well. A laminectomy alone for ventral thoracic disease is not advisable.
Ventral abscesses of the lower thoracic and upper lumbar spine generally require débridement by a thoracoabdominal approach, which allows bone or cage reconstruction with plating as well as removal of the phlegmon. Those patients with infection in the midlumbar or low lumbar spine can undergo decompression via a retroperitoneal approach when the infection is anteriorly placed, with the reconstruction similar to that used at higher levels. Below L2, a posterior approach is generally used.
When infection persists despite antibiotics or when it recurs, reoperation may be necessary. In this scenario, a search for sequestered and devascularized bone is needed, and thus a full corpectomy is commonly performed with reconstruction thereafter.
When patients have significant compression of the cord and thus a clinical myelopathy, steroid administration is considered both safe and useful at high doses.
Current studies suggest that the spine can be reconstructed and stabilized with safety and effectiveness even within a surgical wound contaminated by an epidural infection. As long as necrotic bone and infectious material are removed to the maximum degree, bone grafts can be used as either autografts or allografts and can fuse to native bone. By putting the fusion construct under compression, the placement of plates, rods, and screws as needed promotes bone fusion in this circumstance. As long as the infectious material is well removed and a sufficiently long course of antibiotics is given, it is rarely necessary to reoperate to remove infected or colonized instrumentation or bone graft.17,23–26
Related posts:
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
