Thirty-five patients aged 14–74 years (average, 54 years) who had brucellar spondylitis were treated between January 1991 and December 1997. The time from onset of symptoms to diagnosis of spondylitis ranged from 1 week to 8 months (median, 9 weeks). Back or neck pain (100% of patients), fever (66%), and constitutional symptoms (57%) were the most common symptoms. Cultures of blood specimens from 26 patients (74%) were positive for Brucella melitensis. The duration of antimicrobial therapy (median, 120 days; range, 45–535 days) varied according to clinical response and the presence of epidural and paravertebral masses. One of the 35 patients underwent surgical treatment of a spinal epidural abscess. Therapy failed for 9 patients (26%; 95% confidence interval [CI], 12%–43%), and 5 (14%; 95% CI, 5%–30%) had a relapse. There were no deaths or severe sequelae in this study. Brucellar spondylitis causes considerable suffering and absenteeism from work, but long-term clinical responses are favorable.
Brucellosis, a zoonosis with a worldwide distribution, is a systemic infection caused by facultative intracellular bacteria of the genus Brucella that can involve many organs and tissues . Osteoarticular disease is the most common complication of brucellosis and has been described in 10%–85% of patients [2, 3]. The spectrum of bone and joint lesions includes arthritis, bursitis, tenosynovitis, sacroiliitis, spondylitis, and osteomy-elitis . The type of skeletal involvement depends in part on the patient's age and the Brucella species involved [2, 4]. Spondylitis is the most prevalent and important clinical form of osteoarticular involvement in adults with infection due to Brucella species [2–7]. Spondylitis may be difficult to diagnose [8, 9] and can be complicated by potentially devastating neurological [10, 11] or vascular conditions [12, 13].
This report details clinical manifestations, diagnostic evaluations, treatment modalities, prognostic indicators, and final outcomes for 35 patients with brucellar spondylitis. It also describes the utility of MRI in the practical management of these patients. To provide a comprehensive description of this entity, we discuss our results in light of the results of other series of cases that have been reported in medical literature [14–17].
Patients and Methods
Study population. From 1 January 1991 to 31 December 1997, a total of 285 cases of brucellosis were diagnosed and followed up prospectively at the General Hospital of Albacete, Albacete, Spain. This secondary care institution is located in the south of Spain (population, 346,195 inhabitants), where there is a tradition of sheep farming. The incidence of brucellosis was 27.73–18.24 cases per 100,000 inhabitants during the period 1991–1997 .
In the present study, we retrospectively evaluated clinical manifestations, treatment modalities, final outcomes, and parameters associated with complete response to therapy for 35 patients with spondylitis. The diagnosis of brucellosis was made by isolating Brucella species from blood or other body fluid or tissue specimens or by standard tube agglutination testing, revealing a titer of antibodies to Brucella of ≥1:160 in addition to compatible clinical findings. The diagnosis of spondylitis was based on clinical symptoms confirmed by MRI [19, 20].
Clinical assessment and definitions. All patients were assessed prospectively according to the protocol described in our 3 previous reports [21–23]. This protocol includes demographic, clinical, and laboratory data. Radiological and other diagnostic imaging studies were performed according to the symptoms of the patients. The patients were assessed initially, on days 7 and 45, and at the end of therapy. After the end of therapy, patients were reassessed (as outpatients) at months 1, 2, 3, 6, 9, and 12 and annually thereafter, as well as whenever clinical symptoms reappeared.
Patients were classified into 3 groups according to the clinical outcome of brucellar spondylitis: patients who recovered and, at the end of the follow-up, did not have any symptoms or signs of infection; patients for whom therapy failed who continuously had symptoms or signs that suggested persistent infection after 45 days of treatment; and patients with relapse who had initially recovered after treatment but had clinical relapse (there were associated positive blood cultures for 3 of 5 patients).
The severity of clinical sequelae was classified according to the patient's functional status at the end of follow-up: normal, no pain or neurological deficits remained; mild sequelae, no neurological deficits remained but pain with exercise that did not interfere with work was present; moderate sequelae, pain interfered with work or milder motor or sensorial deficits remained; and severe sequelae, permanent and excruciating pain (requiring rest and analgesics) or motor or sensorial deficits remained.
Microbiological studies. Standard tube agglutination testing, the rose bengal test, and the Coombs' test for antibodies to Brucella species were done according to standard methods  with commercial reagents (Knickerbocker, Barcelona, Spain). Blood cultures were performed, as reported elsewhere , and incubated for 30 days; BACTEC NR 730 or BACTEC 950 (Becton Dickinson, Madrid, Spain) was used. All isolates were identified as recommended by Hausler et al. . The isolated strains were sent to a reference center (Laboratorio Regional de Brucelosis, Valladolid, Spain) for confirmation and biotyping. All Brucella isolates were identified as Brucella melitensis.
Interpretation of radiographs and MRIs. Radiographs, CTs, and MRIs were assessed independently by 2 radiologists (L.A. and E.L.). Plain radiographs consisted of at least an antero-posterior and lateral view of the spine in the area of interest. A Vectra 0.5-T scanner (General Electric Medical Systems, Milwaukee, WI) and an MRT-50A 0.5-T scanner (Toshiba Medical Systems, Otawara, Japan) were used to perform MRI. All patients were studied in the sagittal plane with spin-echo T1-weighted and T2-weighted sequences. Axial T1-weighted sequences were obtained through the levels of interest. Typical MRI findings for osteomyelitis or diskitis that were sought by imaging studies included decreased signal intensity in the vertebral bodies on T1-weighted images, increased signal in the vertebral bodies on T2-weighted images, increase in signal intensity in disks on T2-weighted images, and loss of end-plate definition on T1-weighted images .
Two forms of spinal brucellosis have been considered: focal and diffuse . Focal vertebral osteomyelitis was defined as focal areas of abnormal signal intensity usually localized in the anterior aspect of an end plate of a vertebra at the diskovertebral junction. Diffuse vertebral osteomyelitis was defined as diffuse abnormal signal intensity of adjacent vertebrae and the intervening disk. Gadopentetate dimeglumine (0.1 mmol of Omniscan/kg; Nycomed, Madrid, Spain) was administered, and imaging studies of the sagittal and axial planes in 17 patients were performed.
Statistical analysis. The χ2 test, Fisher's exact test, t test, and Wilcoxon-Mann-Whitney rank-sum test were used as appropriate . Two-tailed P values were calculated; P <.05 was considered statistically significant. Confidence intervals (CIs) for response and relapse were calculated with use of the normal approximation to the binomial by means of Epi-Info Version 6 .
Of the 285 patients with brucellosis, 35 (12%; 95% CI, 8.6%–16.6%) were diagnosed with spondylitis. The demographic and clinical characteristics of these patients are compared with those of all patients with brucellosis who did not have spondylitis in table 1. Patients with spondylitis were more likely to be older (P < .0001), have longer durations of symptoms before therapy (P = .0007), and have erythrocyte sedimentation rates (ESRs) higher (P<.0001) than those of patients with brucellosis who did not have spondylitis. Host factors predisposing to spine infection included diabetes mellitus (3 cases), dialysis for chronic kidney insufficiency (1), alcoholism (2), and previous surgery on the lumbar spine (1).
Presenting symptoms and signs. The time from onset of symptoms to diagnosis of spondylitis was 1–8 months (median, 9 weeks). For 17 patients (49%), the diagnoses of brucellosis and spinal osteomyelitis were made almost simultaneously (within 2 weeks of each other). In these 17 patients, the disease process was more acute, and blood cultures were positive for 16 (94%). Thirteen patients (37%) were diagnosed with brucellosis and, at least 2 weeks later, were also diagnosed with spondylitis. Five patients (14%) were diagnosed with a vertebral lesion 1–8 months before they were diagnosed with brucellar spondylitis. For these 5 patients, initial diagnoses were malignancy (2), degenerative conditions (2), and cholecystitis (1).
The symptoms reported at presentation are shown in table 2. None of the patients was asymptomatic. Back pain, fever, and constitutional symptoms were the most common symptoms. In 18 (51%) of the patients, the pain irradiated (to the shoulders in 2 cases and to 1 of the lower limbs, mainly as sciatica, in the rest). Five patients also had pain in 1 of the sacroiliac joints.
Laboratory data. ESRs were measured for all 35 patients; ESRs ranged from 6 to 94 mm/h (median, 35 mm/h). Twenty-nine patients (83%) had ESRs >20 mm/h, and 15 (43%) had ESRs >40 mm/h. C-reactive protein levels were measured in 20 patients (median, 32 mg/L; range, 6–168 mg/L). Anemia (hemoglobin concentration <11.5 g/dL in females and <13.5 g/dL in males) was found in 7 patients (20%). Leukocytosis (>10,500 WBCs/mm3) was found in 4 patients, and leukopenia (<4500 WBCs/mm3) was found in 1 patient. Thrombocytopenia (<150,000 platelets/mm3) was found in 4 patients (11%). A slight to moderate increase in serum hepatic transaminase levels was found in 18 patients (51%). Elevated serum γ-glutamyl transpeptidase (GGT) and alkaline phosphatase concentrations with normal bilirubin level (dissociated cholestasis) were found in 22 patients (63%).
Standard tube agglutination testing of initial samples from 32 (91%) of the 35 patients was positive for antibodies to Brucella (titer, ≥1:160). Blood cultures were positive for the 3 patients for whom these titers were <1:160. The Coombs' test revealed titers of antibody to Brucella of ≥1:320 for all patients. Thus the Coombs' test was positive for 100% of the patients. All 35 patients underwent the rose bengal test; the results were positive for all of them. Cultures of blood specimens from 26 (74%) of 35 patients with spondylitis were positive for Brucella species. Four patients with negative blood cultures had received antibiotic therapy previously. A Brucella species was isolated from synovial fluid from a patient with elbow bursitis.
Findings of MRI and other diagnostic imaging studies. All patients underwent MRI at least once (81 abnormal MRIs altogether: lumbar, 71; thoracic, 4; and cervical, 6). The interval between the onset of symptoms and the initial MRI was 6–298 days (median, 64 days). Thirty-eight levels of vertebral osteomyelitis were diagnosed in 35 patients. Thirty-two patients (91%) had involvement of only a single spinal region. Two of these 32 patients had contiguous involvement at multiple levels (>2 vertebral bodies). Three patients (9%) had noncontiguous multifocal spinal involvement. Two patients (6%) had focal spondylitis (figure 1), 31 had diffuse disease (89%), and 2 had focal spondylitis and diffuse disease at another level.
The frequency of typical MRI findings for vertebral osteomyelitis and diskitis is summarized in table 3. Abnormality of the signal intensity of the vertebral body marrow was evident in all patients at the lesion site. The subchondral parts of vertebrae were most commonly involved (figure 2).
Twenty-six patients had 2 or more MRIs (17 had 2; 3, 3; 4, 4; 1, 6; and 1, 7). In the first weeks after the diagnosis, there was an increase in signal abnormalities in vertebral bodies and worsening of the disk space despite clinical improvement (figure 3). A decrease in soft-tissue masses was associated with a favorable outcome, whereas persistence of epidural masses tended to have an unfavorable course. In cases of focal spondylitis, 2 patients evolved to diffuse disease, 1 had persistent focal disease, and 1 did not have a subsequent control MRI.
All patients underwent radiography of the spine at least once (total, 91 studies). Only 23 patients underwent radiography before the diagnosis of spondylitis was made by MRI. Of these patients, 14 underwent radiography within 30 days after the onset of symptoms; 8, between 30 and 90 days after the onset of symptoms; and 1, 90 days after the onset of symptoms. Of the first 14 patients, changes suggestive of vertebral osteomyelitis were observed in 6 (43%). Seven (50%) of these 14 patients had degenerative changes at other levels. In 2 cases, radiograms were normal. In the group of 8 patients who underwent radiography between 30 and 90 days after the onset of symptoms, 6 had erosion and sclerosis; 5, narrowing of the disk space; and 1, no abnormalities. The patient who had an x-ray 90 days after the onset of symptoms showed sclerosis, erosion, and narrowing of the disk space.
Technetium bone scanning was positive in all 6 cases in which it was performed. In 1 case, both technetium and gallium scans were obtained, and both demonstrated increased uptake at the affected area. CT scans of the spine were obtained in 6 cases. There were changes suggestive of spondylitis in 4 cases.
Treatment. All patients received antibiotic therapy with a combination of agents. The median duration of antimicrobial therapy was 120 days (range, 45–535 days), and 83% of patients received therapy for ≥90 days. Duration of therapy varied according to clinical response and the presence of epidural and paravertebral masses (table 4). Twenty-eight patients (80%) received a combination of oral doxycycline (100 mg every 12 h) and an im aminoglycoside (1 g of streptomycin/day for the initial 14–21 days, 12 patients; 240 mg of gentamicin/day for the initial 7–15 days, 11; and 300 mg of netilmicin/day for the initial 7–15 days, 5). Seven patients (20%) received a combination of doxycycline (100 mg every 12 h) and rifampin (900 mg/day orally).
Response to treatment varied (table 4). Of the 28 patients initially treated with the doxycycline/aminoglycoside combination, treatment was switched to doxycycline and rifampin for 2, doxycycline and trimethoprim-sulfamethoxazole for 1, trimethoprim-sulfamethoxazole for 1 because of no response; treatment was switched to rifampin and trimethoprim-sulfamethoxazole for 1 because of a doxycycline-related rash. For the 7 patients initially treated with the doxycycline/rifampin combination, treatment was switched to doxycycline and aminoglycoside for 3 (43%) because of no response. One of the 35 patients underwent surgical treatment of a spinal epidural abscess with left L5 and S1 radicular compression (figure 4). This treatment consisted of decompressive laminectomy and drainage of the abscess together with debridement of infected bone.
Outcome. All patients were available for follow-up for at least 12 months (median, 4.0 years). The period of defervescence for 22 inpatients with fever at the beginning of therapy was between 1 and 15 days (median, 6 days). Twenty-one patients who recovered had rapid regression of symptoms. Treatments for 9 patients (26%; 95% CI, 12%–43%) were considered failures, and 5 (14%; 95% CI, 5%–30%) had relapse. All patients for whom therapy failed continued to have moderate or intense pain that interfered with work or impeded their daily tasks after 6 weeks of therapy. The initial diagnosis for 5 of these 9 patients was brucellosis, but spondylitis was not suspected; therefore, the duration of initial therapy was only 45 days. Of the 9 patients for whom therapy failed, 1 received maintenance treatment with doxycycline over a longer period, and 8 had their treatments switched to another combination (table 4).
The total duration of treatment for the patients for whom therapy failed was longer than that for those who recovered (mean duration of treatment: 266 and 94 days, respectively; P = .0001). These patients were also on sick leave longer (or they could not carry out their daily chores) than were patients who recovered (mean sick leave: 10.7 vs. 4.8 months, respectively; P = .02). The overall long-term clinical response was favorable, but 1 patient could not continue to work at his previous job.
Of the 35 patients included in this study, 5 (14%) had relapse after completion of therapy. Of these 5 patients, all had clinical relapse with characteristic clinical findings, and 3 (60%) had associated brucellar bacteremia. All patients had focal disease in the same location during relapse. Relapse occurred between 1 and 4 months after completion of therapy.
Characterization of prognostic factors. To identify important prognostic variables, we correlated the presence or absence of certain features with outcome. Outcome was categorized as favorable (recovered with no or mild sequelae) or unfavorable (therapeutic failure, relapse, or moderate or severe sequelae). No clinical or demographic variables were correlated with outcome. One laboratory variable did predict outcome: the GGT/alanine aminotransferase (ALT) ratio. Nine (90%) of 10 patients with a GGT/ALT ratio of <1.00 had favorable outcomes, compared with 12 (48%) of 25 with a GGT/ALT ratio of at least 1.00 (RR, 1.88; 95% CI, 1.19–2.96; P = .028). On the other hand, patients for whom therapy failed tended to have larger epidural masses on MRIs than did those who recovered (mean maximum diameter: 9.3 vs. 5.4 mm, respectively; P = .02).
Brucellar spondylitis is unusual in developed countries, since brucellosis has practically been eradicated in animals . However, cases have been described in patients from other countries where the disease is endemic or in people who have traveled to these areas and have consumed unpasteurized dairy products [8, 9, 13, 29–34]. Brucellosis is a relatively common cause of vertebral osteomyelitis in geographic areas of the world in which B. melitensis is endemic (e.g., the Mediterranean basin [15–17, 35–37], the Middle East [10, 14, 38–40], and Latin America ). However, only a few series have been reported in sufficient detail to allow for analysis. In this discussion, we will compare our findings with those of 5 major case studies performed between 1974 and 1997 (table 5) [14–17, 35].
The frequency of spondylitis in brucellosis ranges from 2% to 53% [2–7, 20], depending on the species of Brucella involved, the selection criteria, and the population of patients studied. Overall, we identified 1 case of spondylitis for every 8 cases of brucellosis seen in our hospital and 1 case for every 16 cases reported to the health authorities .
The demographic and clinical characteristics of the patients in this study were similar to those of previously described groups of patients with brucellar spondylitis (table 5). As observed in pyogenic vertebral osteomyelitis [17, 36, 37], brucellar spondylitis occurs most commonly in adults aged between 50 and 60 years. Because of the demographic features of human brucellosis in Spain, most patients in this cohort were male, lived in rural areas, and had occupational risk factors, whereas in the studies from Portugal  and Israel , the disease predominated in females and was mainly caused by the ingestion of unpasteurized dairy products.
Within the spine, osteomyelitis occurs most frequently in the lumbar region, followed by cervical and thoracic locations. B. melitensis accounts for most cases in most studies [14–17], but cases caused by Brucella abortus and Brucella suis are also frequent [7, 34]. Cases caused by Brucella canis have been described in dogs , but, as far as we know, none have been described in humans.
Diagnosis. The major obstacle to establishing a clinical diagnosis of spondylitis early in the course of brucellosis is the nonspecific and subtle nature of the symptoms and signs. Abnormal results of blood tests are usually mild and nonspecific. The hemoglobin level may be lowered as a result of prolonged infection, and a moderately elevated ESR is found in most cases. Liver function tests disclose a mild to moderate increase in serum hepatic transaminase levels. An increase in GGT and alkaline phosphatase levels with a normal bilirubin value is frequently found.
The finding of Brucella organisms in blood culture is diagnostic, and several specimens for culture should always be taken. Continuous bacteremia and a high frequency of positive blood cultures are typical of infection due to Brucella species [15, 21–24]. Failure to culture the organism in brucellosis may result from inadequate microbiological techniques or, more importantly, the administration of antimicrobial agents before blood specimens for culture are obtained [15, 17]. In addition to blood cultures, culture of disk or bone tissue specimens that have been removed surgically or by needle aspiration may reveal the presence of the organism [7, 13, 45].
The presumptive diagnosis of brucellar spondylitis can be made serologically . Positive results (titers of antibodies to Brucella of ≥1:160 [standard tube agglutination test] or ≥1:320 [Coombs' test]) are common (and are usually high titers), although low titers determined by standard tube agglutination testing have been reported . In rare cases, patients may have brucellosis with positive blood cultures but negative serology [47, 48].
MRI plays an important role in the diagnosis, assessment, and management of patients with spondylitis. Brucellar spondylitis may be unifocal or multifocal. Predilection for the lower lumbar spine, intact vertebral architecture despite evidence of diffuse vertebral osteomyelitis, and minimal associated para-spinal soft-tissue involvement are all features that suggest brucellar infection over other infectious diseases, including granulomatous diseases such as tuberculosis . In our study, the presence of soft-tissue masses was very common, and these masses were found early. Most of the soft-tissue masses are minimal or moderate in size, and abscesses are rare. A soft-tissue mass does not necessarily mean that drainage is needed in the absence of neurological deterioration. The decrease of these masses is correlated with a good clinical outcome.
Plain radiograms of the spine are rapid, inexpensive, and accessible, but their sensitivity is low in the first weeks of the disease . Because brucellar spondylitis is a slow process, changes in plain radiographs can be difficult to differentiate from those of degenerative diseases . Radionuclide scinti-graphy is a very sensitive but poorly specific means of detecting inflammatory alterations in bone. Although it is well suited for total-body assessment of the extent and distribution of musculoskeletal involvement , the limited tissue resolution of scintigraphy may necessitate the use of additional imaging modalities such as MRI.
Treatment. According to Spink , “brucella spondylitis is one of the most incapacitating and painful maladies that can afflict man.” Spontaneous remission was reported in the preantibiotic era; however, the course of the illness was long and painful, and the presence of sequelae was frequent [50, 51]. Currently, a combination of antibiotics is the treatment of choice. However, management of brucellar spondylitis remains controversial regarding the selection of antibiotics, the duration of treatment, and the role of surgery. The most widely used antibiotic combination for therapy is tetracycline (particularly doxycycline) and aminoglycoside; this combination provides sustained improvement in the conditions of 60%–90% of patients [15–17]. Treatments with trimethoprim-sulfamethoxazole , ciprofloxacin  or doxycycline, and rifampin [21, 53] are associated with the worst outcomes, as was shown in our series.
The duration of antibiotic therapy for brucellar spondylitis is longer than that for systemic brucellosis without spondylitis . Although the duration varied considerably in different studies [14, 15, 17, 39], treatment includes antibiotics administered for a period between 6 and 12 weeks, combined with analgesics and immobilization for pain reduction. Even though most patients with brucellar spondylitis respond to treatment with an antibiotic combination, some respond only partially or temporarily and require several courses of treatment as in 14 of our cases. A lack of response to a tetracycline/aminoglycoside combination seems to be encountered more frequently with certain conditions, such as large paravertebral abscesses . Failures of antimicrobial treatment rarely represent actual resistance of bacteria to antibiotics [54–56]. In the present study, there were 5 documented cases of relapse, resulting in an incidence of brucellar spondylitis of 14%. This rate is comparable with the 10% relapse rate reported by Ariza et al.  but is higher than the 4% rate reported by Colmenero et al.  and lower than the 55% rate reported by Lifeso et al. .
Unless the diagnostic process requires tissue sampling, surgical intervention is rarely required . However, surgical intervention can be indicated for patients with neurological deficits [7, 14–17, 39, 57]. No mortality or severe sequelae were present in this study. Patients who died had a diagnosis of cervical spondylitis and epidural abscess with medullar compression and tetraplegia [17, 58] or extension of an inflammatory process to the abdominal aorta . When medullar compression by an epidural abscess was diagnosed early and treated appropriately, patients with complicated brucellar spondylitis responded well [14, 15, 39]. Percutaneous drainage or aspiration of epidural  or paravertebral abscesses  has been performed successfully, and this procedure may be an alternative to surgery, especially for the patient who is a poor surgical candidate.
We are indebted to Una O'Connor, Walter Bini, and María Melero for editorial assistance. We also thank Lourdes Saez, Fernando Mateo, José Javier Blanch, Emilio Serna, and Pilar Paterna of the Unit of Infectious Diseases; and the many staff members of the Clinical Microbiology Laboratory and the Departments of Internal Medicine, Neurosurgery, and Orthopedics, Albacete General Hospital, who collected the study data.