This retrospective cohort study was designed to identify the diagnostic yield of CT-guided bone biopsy and to evaluate the clinical outcomes in patients with spinal infection from March 1996 to March 2008 in a 1,900-bed tertiary care university hospital. To search patients who have underwent CT-guided bone biopsy and diagnosed to have osteomyelitis, we screened cases in which “bone” specimen was obtained by “CT-guided procedure.” Specimen code for “bone tissue” and procedure codes for “bone needle biopsy procedure” and “imaging guidance procedure” were used for screening cases. Electronic medical records were reviewed among screened individuals. From the medical records, we selected patients who had cultural and pathologic reports and radiologic findings compatible with those of pyogenic and tuberculous spondylitis. Based on histopathologic reports, cases proven to have malignancy or non-septic inflammatory discitis were excluded. Finally adults patients, who had underwent CT-guided bone biopsy and been diagnosed to have infectious spondylitis by the evidence of cultural, pathologic, and radiological data and been treated with prolonged antimicrobial therapy, were enrolled in the study.

Patients with tuberculous spondylitis were categorized into definite, probable and possible groups, whereas those with pyogenic one were categorized into definite and possible. The definite group referred to cases with microbiologic evidence of bacteria, fungus or tuberculosis. The presence of caseating granulomatosis or acid-fast bacilli in Ziehl-Neelsen stained slides on histopathology defined probable tuberculosis in cases without mycobacterial growing. Detection of tuberculosis by polymerase chain reaction (PCR) was also used to aid the probable diagnosis when microbiologic evidence was not given [12]. The possible group contained cases that were clinically diagnosed and treated with conventional antibacterial and anti-tuberculous agents by attending physicians, based on clinical presentation and manifestation when test results from cultures, histopathology, and tuberculosis PCR failed to prove presumptive pathogen. Radiologic findings such as the patterns of destruction of vertebral end plates and signal-intensity changes of involved vertebrae, and presence of epidural, paravertebral, and dis-space abscess were also used to support a clinical diagnosis.

The following patient data were collected using electronic medical and laboratory records: initial clinical presentation such as fever, constitutional symptoms, and back pain; underlying comorbidity; laboratory data including white blood cell count (WBC), erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP) level; involved vertebra; radiologic findings suggesting tuberculosis or pyogenic spinal infections, the presence of paravertebral or epidural abscess; results of histopathology; antibiotic regimen and duration of treatment; and performance of surgical intervention

Clinical improvement was assessed at 4 weeks during the course of therapy. Lack of improvement referred to the cases with persistent or aggravated back pain, necessity of abscess drainage, or development of neurologic deficit [13]. The patients were followed up for 6 months after completion of treatment. Clinical success was defined as cases without recurrence at 6 months after finishing treatment, whereas clinical failure was as those with death, relapse or repetitive surgery during the follow-up period.

The time of surgery was defined as days between the date of CT-guided bone biopsies and the date of surgical intervention. Types of surgical intervention were further classified into early and late surgery. The late cases were designated as those performed 1 month after diagnosis. The number and percentage of late surgeries was also calculated to evaluate the efficacy of initial medical treatment. The decision to perform early or late surgery was made by the attending surgeon. The reasons for late surgical approach were divided into two categories whether it was initially planned at the time of diagnosis when the early surgery was indicated or it was abruptly performed as the results of lack of clinical improvement.

Student t test was used to compare continuous variables, and Pearson chi-square test or Fisher exact test was used to compare categorical variable. Values of p < 0.05 were considered statistically significant. SPSS statistics for Windows, PASW version 17.0 (SPSS Inc., Chicago, IL, USA), was used for all statistical analyses.

During the study period between 1996 and 2008, a total of 1,027 cases were screened to have had undergone CT-guided bone biopsy. Of them, 927 were excluded; 898 malignancy and non-septic discitis; 29 incomplete medical records. Only one hundred patients have been diagnosed and treated as infectious spondylitis among recipients of CT-guided bone biopsy over the 13 year-study period.

Of 100 patients, 67 were identified to have pyogenic spondylitis. Definite diagnosis for pyogenic spondylitis were made in 36 patients because bacterial and fungal growing was confirmed in those (32 and 4), whereas possible diagnosis were made in 31 patients who had been treated with antibacterial agents based on clinical and laboratory supports without microbiologic evidence. Thirty-three patients were diagnosed to have tuberculous spondylitis. Definite growth of Mycobacterium tuberculosis was confirmed in 13 patients. Histopathological analyses of vertebral marrow showed granuloma with caseating necrosis in nine patients, suggesting probable tuberculosis. Eleven patients were treated with antituberculous agents based on clinical diagnosis and referred to possible group.

Microbiologic and pathologic yield of CT-guided bone biopsy for diagnosing definite and probable infectious spondylitis was shown in Table 1. In pyogenic spondylitis group, of 36 confirmed cases, 22 micro-organisms were recovered from bone specimens obtained by biopsies and remaining 14 organisms were from blood and other tissues, which were not obtained from CT-guided biopsy. This resulted in 33% of diagnostic yield of CT-guided bone biopsy for pyogenic spondylitis. Definite growth of M. tuberculosis in 13 patients was identified respectively from 11 bone specimens and two sputum cultures. Nine pathologic results from bone marrows had assisted to make a diagnosis for probable tuberculous spondylitis. Among a total of 33 tuberculosis cases, 20 were identified by test results from bone specimens, which had yielded 61% of diagnosis of tuberculosis.

Clinical characteristics of patients with tuberculous and pyogenic spondylitis are compared in Table 2. The median age was 56 years and males were predominant in the pyogenic group (68.7%) compared with the tuberculous group (45.5%, p = 0.044). Patients presented with spinal disease (26%), diabetes (13%), and hypertension (13%). The most commonly involved vertebrae were lumbar (56%), followed by thoracic (19%) and thoracolumbar (9%). Cervical thoracolumbar vertebrae were more commonly involved in the tuberculous group (51.5%) than in the pyogenic group (17.9%), with statistical significance (p = 0.002). More than three ranges of spines were extensively involved in only in one case, which was caused by Pseudomonas aeruginosa. An abscess in the psoas muscles or in the epidural or paravertebral space was observed in 49% of patients. Patients with pyogenic spondylitis more frequently presented with fever accompanied by an increase in inflammatory markers than did those with tuberculosis. A total of 11 cases were identified to have surcgical site infection which occurred within 3 months after spine surgery in pyogenic spondylitis group.

Surgical intervention was performed in 46% of spondylitis cases. Sixteen patients (16%) underwent surgery 1 month after diagnosis despite medical treatment: these included nine cases of tuberculous spondylitis and seven cases of pyogenic one (56.3% vs. 23.3%, p = 0.026). There were no differences between the two groups with respect to rates of 30-day mortality and repetitive surgery. Patients with pyogenic spondylitis had stayed longer in hospital than had those with tuberculous spondylitis (38 days vs. 21 days, p = 0.003).

Thirty-eight isolates were recovered in 36 patients of definite pyogenic groups. Mixed organisms were found in two patients with methicillin-susceptible Staphylococcus aureus and Streptococcus viridans and methicillin-resistant S. aureus and Enterococcus faecalis. S. aureus was the most commonly recovered isolate, accounting for 47% of cases with definite pyogenic spondylitis. Gram-negative organisms were cultivated in eight patients (21%). Fungus was recovered in four patients (11%), and all of these cases were related to surgical site infection. Detailed microbiological results are provided in Table 3.

Characteristics of patients with positive culture results in definite bacterial infection group were compared with those who had negative culture results in possible group (Table 4). Four fungal cases were excluded in this analysis. The rates of prior antimicrobial treatment were not different between the two groups. An increase in inflammatory markers was observed in the culture-positive group, in particular the WBC and CRP levels. Clinical outcomes were not different between the two groups.

A late surgical approach was attempted in 16 patients with nine cases of tuberculosis and seven cases of pyogenic spondylitis, including probable and possible diagnoses (Table 5). Five patients with spinal tuberculosis had accompanying pulmonary infection. When we analysed the reasons for late surgery, most cases were related to lack of clinical improvement related to aggravation of back pain, progression of abscess and neurologic deficit. In four tuberculous spondylitis cases, delayed surgery was planned intentionally at the time of diagnosis even though patients presented with epidural and paravertebral abscesses. In these cases, attending physicians have decided to observe the response of antituberculosis treatment and delay the surgery for at least 1 month, expecting the improvement of spondylitis by medical therapy. The size of the abscess in four patients, at the time of assessment of clinical response, was not improved despite antituberculous therapy, which had resulted in the late surgery.