Clinical manifestations

Presentation and course

Spinal epidural abscesses may be acute or chronic. Most patients with acute spinal epidural abscess present within 1 to 3 weeks of onset (29; 28; 97; 112; 117). The initial symptom is characteristically localized, severe pain; this is followed by radicular pain and then by signs of spinal cord compression including impairment of position and vibration sense, the appearance of corticospinal tract findings, the development of a sensory level to vibration, position, light touch, or pinprick, and finally impairment of bowel and bladder control (48; 29; 82; 89; 28). Nuchal rigidity may be present, in particular, in cases of cervical epidural abscess (49). Epidural abscess may occasionally present with abdominal pain, suggesting appendicitis (17; 38). Cases in which infection involves the high cervical epidural space or extends to involve the clivus may present with cranial nerve palsies (12; 110), and a case has been reported in which a cervical epidural abscess presented as cluster headache (73). Mental status may be impaired (13; 29; 82; 89). Older patients (> 65 years of age) may present later in their clinical course and are less apt to have fever but may present with delirium or hypotension (07).

In chronic epidural abscess, the onset of symptoms may be extremely insidious, and the history may suggest an extrinsic spinal cord neoplasm (26; 48; 82). Duration of symptoms in such cases may be 5 months or longer (82). Fever, leukocytosis, and other signs of systemic infection are common but not invariable in acute spinal epidural abscess (54). Fever and leukocytosis are often absent if the infection is chronic and may also be absent in immunosuppressed patients (19; 106; 54; 114).

Prognosis and complications

Mortality in spinal epidural abscess in most current series has been 6% to 15% (89; 28; 97; 117). A study by Du and colleagues recorded a 30-day mean mortality of 3.7%, but a range from 0.3% to 37.5% based on numbers of adverse risk factors (34). Adverse outcome tends to be significantly higher in cervical epidural abscess than in epidural abscesses involving thoracic or lumbar spine (42). Prognosis for neurologic recovery worsens in direct proportion to the degree of neurologic deficit and correlates both with the length of the abscess and the degree of cord compression (116; 89). Prognosis is excellent in epidural abscess if treatment is given before signs of cord compression occur (29; 82; 61; 31; 28; 97; 42). Conversely, however, prognosis is guarded if signs of cord compression are already present, and full recovery is much less likely if the duration of paralysis is greater than 24 hours. Likelihood of poor outcome is worse in diabetics and the elderly and may be worse in cases associated with methicillin-resistant Staphylococcus aureus (MRSA) than with methicillin sensitive strains (51; 07; 117). Adverse factors in patients treated with antibiotics alone include age over 65 years and diabetes, MRSA infection, and the presence of neurologic deficits (63; 07). Patients acquiring epidural abscess following intravenous drug abuse, which is a risk factor that accounts for 19% of spinal epidural abscesses, can be particularly challenging because of patient compliance (125; 02). In their analysis of 30-day mortality, Du and colleagues listed adverse risk factors of older age, diabetes, hypertension, respiratory comorbidities, renal comorbidities, metastatic cancer, and thrombocytopenia (34).

Clinical vignette

A 57-year-old man was in his usual state of health until 20 days prior to his admission when he stooped to move a heater. He felt a twinge of soreness in his upper back and because of this took a hot shower and went to bed. He awoke 2 hours later with profuse sweating and back pain so severe that he had difficulty breathing. He treated the pain initially with ibuprofen. The back pain continued, however, and the patient developed circumferential abdominal pain, which he described as "pulling from the spine and shoulders downward." Treatment with ice, heat, and analgesics was unhelpful. The patient noted that his legs were becoming progressively weaker and number, forcing him to use a cane to ambulate. He was unable to move his bowels and was able to urinate only by manually compressing his abdomen. Six days prior to admission the patient was seen at another facility, where lumbar CT was negative and where he received cephalexin, a muscle relaxant, and codeine. These medications relieved the patient’s pain, but he became unable to ambulate 5 days prior to admission. At the time of his admission he had been unable to walk for 5 days and had not had a bowel movement for 20 days.

His past health was noncontributory. The patient’s mother had diabetes, but the patient had never been diagnosed as diabetic. The patient was divorced and had been homeless before coming to his present situation, which involved working as an assistant dog groomer in exchange for room and board. The patient consumed one fifth of vodka per week but denied intravenous drug abuse, homosexual activity, or contact with prostitutes.

On examination the patient was acutely ill. Pulse was 64 per minute, blood pressure 123/73, respirations 24, and temperature 95.9°F. General physical examination was essentially normal without focal back pain but with nuchal rigidity. The patient was alert and oriented. Cranial nerves were normal. Motor examination revealed mild left arm weakness and unequivocal bilateral lower extremity weakness. Light touch was intact, but the patient had diminished pinprick sensation to the lower abdomen and absent vibration sensation to the costal margin. Station and gait could not be tested. Reflexes were not increased, but Babinski sign was present bilaterally. Hematocrit was 36 and white blood cell count was 18,000. Erythrocyte sedimentation rate was 57 mm per hour. MRI showed an enhancing mass lesion posterior to the cord, extending from C7 to T12, consistent with an epidural abscess.

The patient was placed on cefotaxime, vancomycin, and metronidazole and underwent emergency C7, T1, T11, and T12 laminectomy and drainage, with catheter-directed lavage of the epidural space. Cultures of the abscess grew Streptococcus pneumoniae, sensitive to penicillin. Blood cultures were negative. The patient was initially treated with penicillin and then, to allow less frequent dosing, with cefuroxime, totaling 8 weeks of intravenous antibiotic therapy. The patient regained bladder control with a gradual and partial return of his lower extremity function.

Biological basis

Etiology and pathogenesis

Etiology. Spinal epidural abscess due to Streptococcus pneumonia is unusual. The major organism associated with spinal epidural abscess is Staphylococcus aureus; this organism is isolated in 50% to 90% of cases of acute epidural abscess and may also, somewhat surprisingly, be found in chronic abscesses (26; 13; 57; 29; 82; 71; 91; 89; 60; 28; 42; 08; 84; 101). Up to 40% of Staphylococcus aureus isolates from epidural abscesses may be methicillin-resistant strains (28; 23; 08). Aerobic and facultative streptococci may be found in 5% to 10% of cases (13; 66; 82; 91; 08; 84; 101). Gram-negative aerobic and anaerobic organisms, including Escherichia coli, Proteus, Pseudomonas aeruginosa, and Bacteroides fragilis occur in 8% to 17% of cases (13; 29; 08). Staphylococcus epidermidis is most commonly found in cases associated with neurosurgical procedures (29). Spinal epidural abscess in the postpartum period may be associated with group B streptococci (53). Mycobacterium tuberculosis may be associated with chronic epidural abscess and is of particular concern in patients with AIDS or with a history of intravenous drug abuse (57; 65; 67; 44). Tuberculous epidural abscess may also occur as the sole sign of reactivated tuberculosis (57). In rare instances, epidural abscess has been associated with Streptococcus pneumoniae during bacterial endocarditis (22). Occasional cases have been caused by Neisseria gonorrhoeae (74), Group G streptococci (94), Listeria monocytogenes (03), Nocardia asteroides (09), Echinococcus (57), Brucella (108; 43), Candida albicans and Aspergillus (123). Melioidosis may be an occasional cause of epidural abscess in developing countries (80). Multiple organisms are found in approximately 10% of patients (13; 29; 82). A mixed flora of Staphylococcus aureus and Cryptococcus neoformans has been reported in a patient with AIDS (86), and epidural abscess may rarely be caused by Blastomyces dermatitidis (14). Spinal epidural abscess in children is also most commonly associated with Staphylococcus aureus but has been associated with group B streptococcus in a pediatric patient with chronic vesicoureteral reflux (10). Bartonella henselae, the agent of cat scratch fever, has been isolated from the epidural abscess of a 5-year-old child (33).

In 2012 multiple cases of epidural abscess, meningitis, and central nervous system vasculitis due to unusual agents were reported in patients receiving epidural injections of contaminated methylprednisolone, resulting in 753 cases and 61 deaths (55; 59; 105; 58). Aspergillus fumigatus was isolated in one patient. However, most isolates have been other fungi, either Exserohilum rostratum or Cladosporium species (55; 59; 58). Recurrence of infection following treatment has been described after up to 24 months (90). More recent cases of infection with these agents following contaminated methylprednisolone have not been reported.

Pathogenesis. The spinal canal, unlike the skull, contains a fat-filled epidural space that provides little resistance to the spread of infection. Organisms may enter this space as a complication of bacteremia, by direct spread from vertebral osteomyelitis, by extension of infection from superficial cutaneous infections such as furuncles, or by lymphatic spread from infections within the retropharyngeal space, posterior mediastinum, or retroperitoneal space (13; 29; 28). Secondary cases have long been recognized as a complication of surgery; cases of secondary spinal epidural abscess, however, have become more frequent with the increased use of spinal interventions for pain management (60). The resultant epidural abscess usually occupies one to five vertebral segments and may extend the entire length of the cord. The abscess may be anterior to the cord, posterior to the cord, or circumferential. The narrow diameter of the spinal canal allows epidural abscess to produce spinal cord compression and necrosis while still relatively small, and longitudinal spread of infection in either condition can result in cord necrosis extending over multiple vertebral segments, both from direct cord compression and from vascular compromise; importantly, vascular compromise may cause spinal cord infarction in the absence of radiologically detectable cord compression (20; 13; 39; 119). Infection of the intervertebral disc at one or more levels is present in up to 80% of acute cases (81). Accompanying osteomyelitis can be detected in up to 65% of cases (13; 29; 82; 93).

Epidemiology

The incidence of spinal epidural abscess has traditionally been estimated between 0.2 and 1.2 cases per 10,000 hospital admissions (13; 29). Estimates of incidence in individual institutions have been higher, possibly reflecting more frequent recognition of the condition (111), and a case series reported an incidence of 5.1 cases for each 10,000 admissions (117). The condition is more common in males than in females (89; 08; 101).

Prevention

The major risk factor for development of epidural abscess is diabetes mellitus, followed closely by intravenous drug abuse (117; 115). Other risk factors include back trauma, hepatic cirrhosis, alcoholism, and pregnancy (47; 13; 57; 65; 29; 82; 89; 24; 28; 97; 08; 101; 117; 01), and prior spine surgery (91). Spinal epidural abscess in young individuals without known risk factors may follow extensive skin lacerations (49). Occasional cases have been reported in patients with AIDS (86). Risk of epidural abscess may also be increased in patients undergoing hemodialysis (122). Cases have been reported in patients receiving tumor necrosis factor-alpha (TNFα) antagonist therapy for psoriasis or rheumatoid arthritis (104; 27) and in two patients receiving rituximab (11). A history of back trauma is present in 10% to 30% of cases (13; 29). Epidural abscess may also be a consequence of psoas or other deep abscesses, lumbar puncture, spinal anesthesia, epidural catheter placement, or infections following abdominal procedures (13; 29; 46; 100; 64; 95; 103; 89; 28). Rare cases have been associated with esophageal stenting (16), prostatic biopsy (04; 56; 76), acupuncture (124; 69), and tattooing (21). Likelihood of perioperative epidural abscess may be increased by the use of epidural methylprednisolone at the time of microdiscectomy (75). Both cervical and thoracic epidural abscesses have been reported in patients with SARS-CoV-2 (COVID-19) infections (107; 113; 40). Severity of COVID-19 infection has ranged from mild to severe and requiring intubation. The majority of these cases have been due to S aureus, with one case being due to S capitis.

Differential diagnosis

Spinal epidural abscess is sufficiently rare that the diagnosis is often not considered, yet it is common enough to constitute a real diagnostic consideration and is not an uncommon cause of lawsuits for malpractice (05). This is of particular importance for physicians at the point of initial contact, so emergency room and other nonoperative physicians are more likely to be sued than are surgeons (05; 32; 99). The great challenge in epidural abscess is not treatment, which is highly successful if instituted early, but rather prompt diagnosis allowing for appropriate therapy before development of severe neurologic injury (89; 31; 37; 30; 111; 08; 84; 101). Epidural abscess should be suspected in any patient presenting with acute, severe spinal or radicular pain, particularly if there is evidence of active infection or if there is a history of diabetes mellitus, intravenous drug abuse, back trauma (which may be minor), or pregnancy (28; 72; 97; 117; 115). Davis and colleagues have suggested additional risk factors, which include immunosuppressed states, chronic liver or kidney disease, indwelling vascular catheter, recent spinal surgery (especially with indwelling hardware), or known systemic infection (30). Diagnostic accuracy may be significantly increased by use of an algorithm combining presence of these risk factors with clinical presentation and elevation of erythrocyte sedimentation rate (ESR) and/or C-reactive protein (30). In the study by Davis and colleagues, ESR was elevated to higher than 20 m/hour in all patients with spinal epidural abscess as opposed to 33% of patients with spine pain but without abscess. Of interest, the mean ESR value for 86 patients with spinal epidural abscess was 76.5 mm/hour as opposed to 20.1 mm/hour in patients in whom epidural abscess was not found, indicating that significant elevation of the ESR should increase suspicion that a spinal epidural abscess may be present (30).

As mentioned above, spinal epidural abscess may occasionally present as an acute abdomen (17; 38). Spinal osteomyelitis may duplicate the findings of focal pain but will not by itself produce radicular or long-tract findings. Spinal subdural empyema, which may mimic spinal epidural abscess, is extremely rare. Spinal epidural hemorrhage, rapidly growing spinal epidural tumor or, more rarely, spinal epidural cysts are also diagnostic considerations (50; 70). The absence of fever in a patient presenting with back pain does not exclude the possibility of epidural abscess (54; 72). The possibility of staphylococcal cervical or thoracic epidural abscess should be kept in mind when dealing with patients with SARS-CoV-2 infection who develop radicular pain or signs of spinal cord injury; it is important to keep in mind that the associated COVID-19 infection may not be severe and that epidural abscess can develop in patients who do not require intubation (107; 113; 40).

Diagnostic workup

MRI is the procedure of choice because of its ability to visualize the cord over its entire length (29; 81; 92; 41; 91; 116; 89; 83; 28; 97). It can also indicate the site of origin of an epidural abscess. Both acute epidural abscess and adjacent osteomyelitis may exhibit increased signal on diffusion-weighted imaging and be dark on ADC maps (36; 79). (Myelography supplemented by CT scan should be used if MRI is not available.) Myelography carries some risk of introducing the offending organism into the subarachnoid space and causing meningitis. For this reason, the spinal puncture should be conducted at a level distant from that believed to be occupied by the abscess or empyema. Thus, a lateral cervical puncture should be used if lumbar infection is suspected, and a lumbar approach should be used if the cervical or high thoracic spine is involved. Plain radiography may be useful in cases of suspected discitis and vertebral osteomyelitis. Blood and material from infected sites should be cultured for aerobic and anaerobic organisms. Blood cultures can be expected to be positive in approximately 50% of acute cases (29). Infection of spinal fluid is unusual (29).

Management

In almost all cases, spinal epidural abscess is an absolute indication for urgent open surgery or, in selected cases, microsurgery or CT-guided aspiration (89; 77; 15; 42; 23; 35; 87; 117; 96). Occasional spinal epidural abscesses may be treated with antibiotics alone if the abscess is small and the patient is neurologically completely normal. Ideally, in such patients, CT-guided biopsy should be used to identify the causative organism (68; 45; 121; 67; 97; 117). Rarely, more extensive spinal epidural abscesses have been treated successfully with antibiotics alone, but this approach remains the exception in larger abscesses (118). Outcome of patients treated with antibiotics alone, as compared to those treated surgically as well, has varied from series to series, and the two approaches have not been subjected to controlled trial (102; 25; 85; 87; 109). In one series, however, 41% of patients treated initially with antibiotics alone required subsequent surgical drainage (84). Thus, it must be kept in mind that not all patients respond to antibiotic therapy alone and that epidural abscess may fail to resolve and may actually progress in the face of antibiotic treatment (121; 06; 08). For this reason, patients treated with antibiotics alone are at potentially higher risk and, thus, require meticulous clinical observation and frequent neuroradiological follow-up (112; 98). Progression of neurologic signs in the face of antibiotic therapy alone is an absolute indication for surgical intervention. Predictors of a poor response to antibiotic therapy alone include patient age greater than 65 years, patients with diabetes mellitus, infection due to methicillin-resistant Staphylococcus aureus (MRSA), and patients manifesting neurologic deficits (63; 07). Shah and colleagues evaluated the occurrence of treatment failure in patients with spinal epidural abscess whose initial treatment was nonoperative (98). Of 367 patients, 99 treatment failures occurred. Independent predictors of failure included motor deficit at presentation; sensory changes; diabetes mellitus; pathologic or compression fracture in affected levels; active malignancy, and dorsal location of abscess. Laboratory values suggesting that medical management may be unsuccessful include C-reactive protein greater than 115 and white blood count greater than 12,500 (84). Hunter and colleagues have reviewed post-treatment fall in C-reactive protein as a prognostic indicator and have found that failure to reduce C-reactive protein by 50% by day 4 to 5 after initiation of treatment was associated with longer hospital stay, worse functional outcome, and greater mortality risk within 2 years, suggesting that failure of C-reactive protein to fall within 4 to 5 days should prompt a search for other foci of infection and/or consider modification of treatment regimen (52). It must also be kept in mind that spinal cord injury in epidural abscess may be due to vasculitis and may occur before severe spinal cord compression is detected.

Initial antibiotic therapy where the causative agent is unknown should be directed against S. aureus and the streptococci and should include vancomycin until sensitivity to oxacillin or nafcillin is determined (91). Vancomycin should be continued if the organism is found to be methicillin-resistant or there is a history of penicillin allergy. Khatib and colleagues have reported a patient who developed a spinal epidural abscess while receiving vancomycin for a methicillin-resistant Staphylococcus aureus infection. In this case, resolution of infection was achieved with a combination of gentamicin and vancomycin (62). Ceftriaxone or cefotaxime should be added if there is any question of a gram-negative organism. Ceftazidime or meropenem should be used if there is strong evidence for infection with Pseudomonas aeruginosa. These third-generation cephalosporins should be supplemented with systemic gentamicin or amikacin (120). Metronidazole should be used if infection by Bacteroides fragilis is suspected. Aminoglycoside agents such as gentamicin and amikacin cross the blood-brain barrier poorly and for this reason systemic gentamicin is not used in treatment of gram-negative meningitis in adults. The spinal epidural space lies outside the blood-brain barrier, however, and although no controlled studies exist to support a role for gentamicin in spinal epidural abscess, consideration should be given to its use in epidural abscess caused by gram-negative organisms. Final selection of antibiotics is based on isolation of organisms from abscess material and determination of antibiotic sensitivity.

Holospinal epidural abscesses represent an area of particular concern, with overall high morbidity. Treatment always involves surgery and often extensive laminectomy. In a study of eight patients who underwent holospinal epidural abscesses, patients received an average of 8.6 laminectomies--one requiring laminectomy of the entire spinal column, three skip laminectomies, and three additional surgeries (18). In this series, 50% of the treated patients improved, 37.5% remained neurologically unchanged, and 12.5% worsened.

Treatment of spinal epidural abscess and of other complications arising from epidural injections of contaminated methylprednisolone has not been subjected to controlled trials. Initially, treatment involved voriconazole, and in some patients, voriconazole was used in combination with liposomal amphotericin B (55; Kerkering al 2013). Subsequently, this combined therapy was used only in the most ill patients, most other patients undergoing voriconazole monotherapy for 6 months or longer (58). Delayed recurrence of epidural infection has been reported in one patient treated over time with both voriconazole and liposomal amphotericin B (90).

Special considerations

Pregnancy

The incidence of spinal epidural abscess is increased during pregnancy (13; 29; 28; 97).

Anesthesia

Spinal epidural abscess is not a contraindication for the use of general anesthesia. Use of spinal anesthesia of any sort should be approached with caution to avoid introducing infectious organisms into the subarachnoid space. As mentioned above, epidural abscess may represent a consequence of epidural administration of anesthetic agents (64).