Vertebral osteomyelitis is an uncommon but increasingly recognized infection that can be difficult to diagnose and can be complicated by potentially devastating neurologic or vascular catastrophies. It represents 2% to 4% of all cases of osteomyelitis. In adults, hematogenous dissemination is the most common method of spread. Recent data suggest that the risk for vertebral osteomyelitis as a complication of Staphylococcus aureus bacteremia is increasing. The reasons for increased recognition of this disease include a higher prevalence of parenteral drug use and nosocomial infection (primarily related to the use of IV catheters), the prevalence of diabetes, an aging population, better imaging techniques, and a greater understanding of this illness.
The disease is more common in adults than in children, and a recent study demonstrated that men were affected in 95% of cases. Most cases occur in persons over the age of 50 years; this may be explained in part by the increased frequency of urinary tract infections in the elderly and the presence of Batson’s plexus, a low-pressure, valveless venous plexus that drains the pelvis toward the vertebral column and potentially allows passage of infection into this area. Thus, it differs from other forms of osteomyelitis in that it has a higher male-to-female ratio, occurs in an older population, and is more closely associated with diabetes mellitus. With regard to pyogenic vertebral osteomyelitis, at least 50% of cases involve the lumbar spine (possibly related to the noted anatomic considerations). The thoracic spine (approximately 30% of cases) and cervical spine are involved in decreasing order of frequency. Tuberculous vertebral osteomyelitis most commonly involves the lower thoracic vertebrae (Pott’s disease).
The clinical presentation is often subtle, and the physician must have a high index of suspicion to make an early diagnosis. Although about 10% of cases present acutely, with positive blood cultures substantiating the diagnosis, subacute or chronic back pain is the most common presentation and is noted in more than 90% of cases. The duration of complaints is frequently longer than 3 months. Unusual clinical presentations may occur as a result of extension into surrounding tissues. Pleural empyema and retropharyngeal abscess have been recently described as initial presentations of pyogenic vertebral osteomyelitis. Neurologic complaints often imply the presence of spinal epidural abscess. IV substance abusers with vertebral osteomyelitis tend to have a more truncated course. Additionally, two cases of gas-forming osteomyelitis caused by enteric gram-negative bacilli have been reported. In both cases, diabetic patients were involved and the course was fulminant. Fever is present in up to 90% of cases but is usually low-grade. Frank rigors are unusual.
Initial assessment must include a careful history to evaluate for a primary focus of infection. Clinical conditions predisposing to vertebral osteomyelitis include a history of IV substance abuse, diabetes mellitus, trauma to the back (including surgery), distant skin or soft-tissue infection, and history of urinary tract infection or instrumentation. In patients currently or recently hospitalized, the IV catheter must be strongly considered. The history should also include an epidemiologic assessment for unusual pathogens, such as Myobacterium tuberculosis, fungi, and Brucella species. A recent investigation demonstrated that M. tuberculosis was the pathogen in approximately 33% of patients with vertebral osteomyelitis, and all came from endemic areas. However, in at least 33% of cases, the predisposing factors for infection will remain undetermined.
The physical examination must include an assessment for localized tenderness over a vertebral body and a search for feeding foci, such as an infected IV line, stigmata of IV substance abuse, and a urinary tract infection. The examination should also assess for neurologic complications.
Initial laboratory data are generally nonspecific. Leukocytosis is seen in fewer than 50% of cases. The erythrocyte sedimentation rate is elevated in more than 80% of cases, but this is a nonspecific finding. However, a decline of this parameter by 33% to 50% during antimicrobial therapy may help to predict cure.
The suspected area should always be assessed radiographically. Plain films may not reveal positive findings for 2 to 8 weeks after clinical presentation, and initial findings are often nonspecific. They may include paravertebral swelling or sclerosis of the vertebral end-plates. However, more advanced cases of vertebral osteomyelitis may involve a disk and adjacent vertebrae. The finding of disk space narrowing with adjacent changes should prompt consideration of this disease. This is in contradistinction to metastatic tumor, which is characteristically confined to the vertebral body and spares the disk. The clinician must have a high index of suspicion when the course of metastatic cancer is other than expected and consider additional diagnoses. Magnetic resonance imaging (MRI) has proved especially useful in this regard. Observations have also been made recently of patients with confirmed vertebral osteomyelitis in the setting of underlying osteoporosis; they presented with localized vertebral disease and resultant collapse mimicking compression fracture, which caused diagnostic delay. Alternatively, findings on routine plain films are usually abnormal when osteomyelitis is of granulomatous origin. This is presumably because of the more indolent nature of the disease and longer time course before clinical presentation.
Gallium, indium 111In-leukocyte, or technetium bone scanning is more sensitive than standard radiography, and positive results may appear within 7 days. Gallium or 111In-leukocyte scanning has the capacity to demonstrate complications within adjacent tissues; however, it is not as specific. Computed tomography (CT) and MRI are the best tests for diagnosing vertebral osteomyelitis in a timely fashion, and one or the other should be used early in management. MRI changes suggestive of vertebral osteomyelitis are noted in 90% of patients with symptoms of less than 2 weeks’ duration, and in 96% of patients with symptoms lasting beyond 2 weeks.
MRI may prove to be particularly useful in diagnosing vertebral osteomyelitis in the known setting of metastatic tumor to a vertebral body. Changes considered diagnostic of vertebral osteomyelitis include erosion of end-plates and occult paravertebral swelling. Similarly, both MRI and CT can provide timely information concerning complications of the spinal canal or aorta. Thus, in the case of patients with obscure back, chest, or abdominal pain or in the evaluation of fever of unknown origin, CT or MRI can provide invaluable information concerning the diagnosis and potential complications and should be performed early in the evaluation. MRI may also suggest differences between pyogenic and tuberculous osteomyelitis. In the former, fewer vertebral bodies are involved, paravertebral abscesses are smaller, and the magnetic intensity of vertebral bodies is more homogenous. The author recommends initial plain films and then MRI as in the diagnosis of vertebral osteomyelitis. Bone, gallium, and labeled-WBC scans are of limited value.
Once osteomyelitis is suspected, bacteriologic confirmation should be sought. Although blood and urine cultures should be performed, they are unlikely to provide confirmatory data. The best approach is percutaneous bone aspiration of the involved site under CT guidance before the initiation of antimicrobial therapy. This approach may also allow simultaneous drainage of any paravertebral abscesses. Material obtained should be sent for (a) Gram’s stain and acid-fast stain; (b) routine (aerobic and anaerobic), mycobacterial, and fungal cultures; and (c) histopathologic analysis. If results of the initial assessment prove negative, routine cultures should be held for several weeks so that pathogens such as Brucella species are not overlooked. The yield from percutaneous bone aspiration will decrease if the anterior portion of the bone is involved. In these circumstances, a decision will be needed regarding more aggressive surgery (typically an anterior approach) or empiric therapy.
Bacteriology is variable and relates to the underlying cause of disease. A recent investigation demonstrated that the presence of HIV/AIDS did not affect the bacteriology of vertebral osteomyelitis. Most cases involve single isolates. Occasionally, cultures remain sterile, perhaps because of previous antimicrobial therapy. S. aureus is the most commonly identified pathogen. Gram-negative enteric bacilli, especially Escherichia coli and Salmonella species, account for about 30% of cases. The latter are especially noted in patients with sickle cell disease and in those whose illness is complicated by aortic mycotic aneurysm. Pseudomonas aeruginosa has been commonly identified in patients whose vertebral osteomyelitis is associated with IV substance abuse. Infections caused by species of Brucella should be suspected in persons from selected areas of the world. Tuberculous vertebral osteomyelitis usually involves thoracic or upper lumbar vertebrae and often is associated with evidence of pulmonary tuberculosis. The duration of symptoms is longer in patients with M. tuberculosis infection than in those with pyogenic vertebral osteomyelitis. A history of active tuberculosis at other sites should be sought. In some series, it has accounted for almost 40% of cases of vertebral osteo-myelitis. M. tuberculosis infection should be suspected in patients whose epidemiologic history would suggest exposure to endemic disease. If attempts to identify a bacterial pathogen fail, the search for it should be intensified.
Parenteral antimicrobial therapy active against the offending pathogen is the cornerstone of therapy for uncomplicated pyogenic vertebral osteomyelitis. In most instances, therapy can be withheld until a pathogen is identified. Treatment is usually continued for at least 4 weeks, and the sedimentation rate may be employed to gauge response to therapy. For gram-positive infections, a single b-lactam agent usually suffices. It is important to monitor for adverse reactions, such as interstitial nephritis, toxic hepatitis, or neutropenia or thrombocytopenia, which can occasionally be observed with prolonged b-lactam therapy. Treatment of infections caused by gram-negative bacilli should be guided by susceptibility data. It is the opinion of the author that many cases can be treated with full doses of oral quinolones as monotherapy. Such treatment has been successfully employed with other forms of osteomyelitis. In occasional cases, empiric therapy has been required, as a specific etiologic diagnosis cannot be made. In these instances, the author has had success with a quinolone plus rifampin, given orally for at least 4 weeks. Should tuberculosis also be a consideration, then isoniazid may be added.
Uncomplicated cases of vertebral osteomyelitis are usually managed medically. Surgical or nonsurgical drainage is generally indicated for spinal epidural abscess and other neurologic or vascular complications, or if frank paravertebral abscess is demonstrated by CT. The focal paravertebral collection may be successfully managed with percutaneous drainage.
Immobilization has also generally been recommended; however, this is now generally interpreted to mean modest bed rest early in treatment. Most authorities do not recommend braces or casting for the routine case, although this may occasionally be indicated if spinal instability is noted.
With appropriate diagnosis and treatment, mortality and neurologic sequelae should be modest. Diagnosing this infection remains the major challenge for the clinician, as the presentation may mimic that of numerous other infectious and noninfectious conditions. (R.B.B.)
Abbey DM, Hosea SW. Diagnosis of vertebral osteomyelitis in a community hospital by using computed tomography. Arch Intern Med 1989;149:2029–2035.
This report evaluates the radiographic diagnosis of vertebral osteomyelitis in 20 persons with the disease. It points out the failure of routine plain films to document this process and the value of CT in delineating changes consistent with the diagnosis. The most commonly identified abnormality was destruction of the vertebral end-plates. The authors conclude that CT should be performed early in the evaluation of patients who possibly have this disease.
An HS, et al. Differentiation between spinal tumors and infections with magnetic resonance imaging. Spine 1991;16(Suppl 8):S334–S338.
The authors studied 30 patients with proven spinal tumors or infections. MRI correctly diagnosed 29 of 30 processes. Involvement of disk spaces was seen only with infectious processes, as was involvement of contiguous vertebrae. The authors believe MRI to be the best imaging modality to differentiate infection from tumor.
Arnold PM, et al. Surgical management of nontuberculous thoracic and lumbar vertebral osteomyelitis: report of 33 cases. Surg Neurol 1997;47:551–561.
Indications for surgery were either neurologic deficit or failure of antibiotic therapy. Long-term outcomes following surgery were good.
Carragee EJ. The clinical use of magnetic resonance imaging in pyogenic vertebral osteomyelitis. Spine 1997;22:780–785.
One hundred three cases of vertebral osteomyelitis were reviewed by the author. Within the first 2 weeks of symptoms, MRI findings were either positive or suggestive of the diagnosis in about 90% of cases. After 2 weeks, this rate rose to about 96%. The author feels that MRI is valuable for the initial diagnosis of vertebral osteomyelitis. However, its role in follow-up is uncertain, as the MRI often demonstrates worsening disease while clinical improvement has been noted.
McHenry MC, et al. Vertebral osteomyelitis presenting as spinal compression fracture. Six patients with underlying osteoporosis. Arch Intern Med 1988;148:417.
The authors identified six patients with underlying osteoporosis complicated by pyogenic vertebral osteomyelitis (one patient was infected with Nocardia asteroides). All demonstrated infection of a single vertebra, which initially was felt to represent compression fracture rather than infection. In the experience of the authors, this clinical presentation represented 10% to 15% of cases of vertebral osteomyelitis and 2% to 3% of cases of presumed compression fracture. The presentation was associated with significant sequelae because of long delays in diagnosis. The presence of elevated sedimentation rate, fever, severe pain, or discomfort that does not improve with standard treatment should make the clinician suspect the presence of infection.
McHenry MC, et al. Vertebral osteomyelitis and aortic lesions: case report and review. Rev Infect Dis 1991;13:1184–1194.
This report summarizes data on 70 patients whose vertebral osteomyelitis was complicated by involvement of the aorta. Mortality was more than 70%. Infected aneurysms were most commonly noted, and pathogens included an extended list of enteric gram-negative bacilli (including Salmonella species), gram-positive cocci, and mycobacteria. The best outcomes were noted with a combination of targeted antimicrobial therapy plus aortic resection.
Patzakis MJ, et al. Analysis of 61 cases of vertebral osteomyelitis. Clin Orthop 1991; 264:178–183.
The authors compared diagnostic imaging techniques in 61 patients with confirmed vertebral osteomyelitis and conclude that MRI is the most sensitive technique for making the diagnosis. Further prospective studies are necessary before their recommendation of MRI as a front-line study for this disease can be totally accepted.
Perronne C, et al. Pyogenic and tuberculous spondylodiskitis (vertebral osteomyelitis) in 80 adult patients. Clin Infect Dis 1994;19:746–750.
This retrospective study demonstrated that almost 40% of patients had tuberculous disease, based primarily on the population that was studied. This points out the need for a careful epidemiologic history. Cases of both tuberculosis and Staphylococcus infection were associated with histories of prior active infection, which could help the clinician in making the diagnosis. In this investigation, more than 50% of patients with pyogenic disease had positive blood cultures, and about 75% had positive bone aspirate cultures.
Sapico FL, Montgomerie JZ. Pyogenic vertebral osteomyelitis: report of nine cases and review of the literature. Rev Infect Dis 1979;5:754–776.
An excellent review of the subject that evaluates 318 cases. Elderly men and diabetic patients were particularly predisposed, and the most commonly identified predisposing focus was the urinary tract. Parenteral antimicrobials for more than 4 weeks plus bed rest were the major therapeutic modalities. Treatment of lesser duration increased failure rates. Failure of the sedimentation rate to decline by at least one third was an adverse prognostic indicator.
Sapico RL, Montgomerie JZ. Vertebral osteomyelitis. Infect Dis Clin North Am 1990; 4:539–550.
The authors review data on the diagnosis, bacteriology, and management of vertebral osteomyelitis. They rightfully conclude that diagnosis may be subtle, and all attempts should be made to isolate a pathogen rather than treat empirically.
Torda AJ, Gottleib T, Bradbury R. Pyogenic vertebral osteomyelitis: analysis of 20 cases and review. Clin Infect Dis 1995;20:320–328.
In this investigation, most patients were elderly and only 30% presented with fever. Nosocomial infection, often associated with IV cannula use and methiciliin-resistant S. aureus, was common. MRI or CT was the most useful radiographic modality.