We retrospectively evaluated 30 consecutive patients with PVDO who underwent TIDF with AIBG performed by a single surgeon between March 2014 and January 2022 (AIBG group). Patients had a minimum 2-year follow-up. For comparative analysis, we included 28 PDVO patients treated by another surgeon (the third author) without AIBG between January 2009 and June 2011 (non-AIBG group) [3]. The decision to adopt AIBG reflected evolving practice patterns over time. Both groups had identical clinical characteristics and surgical indications, with AIBG being the sole variable. CRP levels and the postoperative antibiotic courses were compared between the two groups.

Clinical and radiographic data were collected for all patients with thoracic or lumbar spine infections who also presented with neurological deficits, spine instability, intractable pain, unidentified pathogens, and failed antibiotic treatment. Patients with tuberculosis bacilli infection and post-spinal operation infections were excluded. The requirement for informed consent from individual patients was omitted because of the retrospective design with chart review of this study. The study was approved by the institutional review board (IRB) of Kaohsiung Chang Gung Memorial Hospital (IRB no., 202401296B0).

Demographic data, diagnosis, affected levels, implant positions, radiographs, laboratory data, duration of antibiotic treatment, and neurologic and functional outcomes were collected. All patients wore Taylor braces for at least 3 months after surgery. Patients underwent follow-up at 2 weeks, 1 month, 3 months, and 6 months, and then annually.

The choice of antibiotics was determined by an infectious disease specialist based on the culture results. Some patients had already been on antibiotics before surgery as part of conservative treatment. The normal reference range of CRP is <5.0 mg/dL. Serial CRP levels were collected to monitor the patients’ progress. The duration of parenteral antibiotic therapy was based on symptoms and laboratory parameters. The decision to switch from intravenous to oral antibiotics was made by the infectious disease specialist based on individual clinical progress. This decision was influenced by various factors such as CRP levels <25 mg/dL, an afebrile status, and the absence of clinical signs of infection. Notably, in the AIBG group, the duration of oral antibiotic treatment varied for each patient, depending on their recovery and response to treatment. To ensure that there was no recurrence of infection, CRP levels were monitored at the 3-month postoperative follow-up.

The clinical assessments included the Visual Analog Scale (VAS) scores for back pain and the American Spinal Injury Association (ASIA) impairment scale for neurological status. The local kyphotic angles of the superior and inferior endplates of the affected levels were measured on standing radiographs before and after surgery. Spine fusion was defined as continuous trabeculae bridging across the interface of the treated segments. Radiolucent interruption at the treated segments was defined as pseudarthrosis. At the final follow-up, fusion status was assessed by two authors who did not participate in the surgery. Functional outcomes were evaluated using the Kirkaldy-Willis classification [14]. Finally, the CRP levels and duration of antibiotic administration were compared between the AIBG group and non-AIBG group [3,5].

Differences between the AIBG and non-AIBG groups were assessed for statistical significance using the Mann-Whitney U test. All statistical analyses were performed using IBM SPSS ver. 22.0 (IBM Corp., Armonk, NY, USA). A p-value <0.05 was considered significant.

A modified TLIF surgical approach was performed, involving debridement, fusion, and pedicle screw instrumentation. In the AIBG group, 26 patients underwent the traditional open method, while four patients underwent unilateral biportal endoscopic surgery. Unilateral facetectomy and limited laminotomy were performed (Fig. 1A), followed by radical debridement to remove the abscess, infected disc material, endplates, and sequestrum from the intervertebral space using curved curettes and disc rongeurs (Fig. 1B). Unilateral biportal endoscopy is a novel, minimally invasive technique typically used to treat lumbar spinal stenosis and lumbar disc herniations [15]. However, we successfully employed this technique for treating PDVO in four cases (Fig. 2).

The choice of entry point depended on the extent of debridement required, dictated by symptom severity and radiographic findings. To maintain posterior stability, we preserved the spinous processes, supraspinous ligaments, and the contralateral facet joint whenever possible.

Tissue samples were examined for aerobic, anaerobic, and tuberculous bacilli and fungi. The residual tissue was sent for histopathological examination. If the iliac autograft was not sufficient to fill the residual bone defect, an allograft was added and merged with the autograft.

Antibiotic selection was based on preoperative culture reports, if available. Otherwise, 2 g vancomycin and 80 mg gentamicin were administered to ensure broad-spectrum coverage against both Gram-positive and Gram-negative pathogens. Following radical debridement, the void was densely packed with AIBG (Fig. 1C). Pedicle screw fixation sites were determined based on bone quality and clinical considerations. Screws were inserted into segments with sufficient bone mass, adequately distant from compromised bone areas to minimize the risk of infection recurrence.

The most common pathogens were methicillin-sensitive S. aureus (MSSA) (30%) and Escherichia coli (14.3%). Two cases (6.7%) had multiple pathogens. The positive intraoperative culture rate was 76.7% (23/30). The mean CRP level improved from 64.9 mg/dL preoperatively to 35.7 mg/dL at 1 week postoperatively, 23.9 mg/dL at 2 weeks, 12.7 mg/L at 4 weeks, 5.7 mg/dL at 6 weeks, and 5.5 mg/dL at 12 weeks (Table 1). Prior to surgery, 14 patients (46.7%) received antibiotics for a mean duration of 11.3 days. Among patients with positive cultures, the sensitivity rate of the antibiotics used in the bone graft was 78% (18/23).

Prior to surgery, 23 patients were classified as ASIA D and seven were classified as ASIA E. Three months postoperatively, 28 patients were classified as ASIA E. None of the patients sustained any intraoperative neurological injury.

The duration of the postoperative antibiotic course was compared between 30 patients in the AIBG group and 28 patients in the non-AIBG group. Patients in the non-AIBG group underwent the same TIDF procedure presented in our previous study [3]. CRP levels declined in both groups after surgical treatment. There was no significant between-group difference in terms of preoperative CRP levels (63.9 mg/dL versus 64.9 mg/dL, p=0.524). Even with a shorter mean postoperative intravenous antibiotic duration (19.0 days versus 39.8 days, p<0.05), the AIBG group showed a greater reduction in CRP levels compared with the non-AIBG group (5.7 mg/dL versus 13 mg/dL, p<0.05) at postoperative 6 weeks. Both groups had similar CRP levels at postoperative 3 months (5.5 mg/dL versus 6.3 mg/dL, p=0.456) (Tables 1, 2). These findings indicate that despite a shorter course of intravenous antibiotics, the AIBG group showed comparable infection control to the non-AIBG group.

The duration of oral antibiotic therapy varied among patients, based on the infectious disease specialist’s clinical judgment, considering individual recovery, clinical symptoms, and CRP levels. In the AIBG group, the mean duration of oral antibiotic treatment was 43.6±14.3 days, shorter than that in the non-AIBG group; however, the between-group difference in this respect was not statistically significant (p=0.938).

Fusion status was evaluated at the last follow-up, revealing solid unions in 26 patients (86.7%) in the AIBG group. Eighteen patients showed improvement in the lordotic angle over the infected levels after surgery. The mean preoperative kyphotic angle (14.1°±6.8°) showed significant improvement immediately postoperatively (2.5°±6.5°, p<0.001). At the final follow-up, the mean kyphotic angle measured 5.9°±10.1° (p<0.05) with a mean kyphotic angle correction of 8.2°.

Mean back pain VAS scores significantly improved from 7.2±0.9 (range, 4.5–8.8) preoperatively to 2.1±1.3 (range, 0.7–5.5) at the 6-week follow-up. Four patients initially reported donor site pain, which was resolved by the final follow-up. Kirkaldy-Willis functional outcome at the last follow-up showed excellent results in 14 cases, good results in 10 cases, fair results in five cases, and poor results in one patient who underwent revision surgery due to a deep wound infection.

Two patients underwent revision surgery. One required revision due to pseudarthrosis with broken S1 screws 1 year postoperatively. Revision posterolateral fusion was performed using an allograft and demineralized bone matrix, resulting in successful bony union and resolution of back pain. The other patient developed a deep wound infection 1-month post-TIDF. This patient underwent anterior debridement and corpectomy with strut allograft reconstruction.

There is no clear consensus on the optimal duration of intravenous antibiotics, with treatment often guided by personal experience or expert opinion. However, a multicenter non-inferiority trial by Bernard et al. [18] suggests that 6 weeks of antibiotics may be sufficient for treating pyogenic vertebral osteomyelitis. Shortening the treatment duration reduces the risk of side effects, financial burden, and development of antibiotic resistance [4,19]. Park et al. [20] suggested that the optimal antibiotic therapy duration depends on the patient’s risk of recurrence. Their multivariate analysis identified methicillin-resistant S. aureus infection, undrained paravertebral/psoas abscesses, and end-stage renal disease as independent risk factors for recurrence. They recommended prolonged antibiotic therapy (≥8 weeks) for high-risk patients with these conditions. None of the patients in our study met these criteria. A 2024 review article suggests that pyogenic spondylodiscitis treatment typically entails 6–12 weeks of antibiotic therapy, guided by culture results [21]. In our study, the mean postoperative intravenous antibiotic duration was 21 days, followed by a mean 6-week oral antibiotic course (Table 1).

No consensus exists on the optimal dose for local antibiotic use [9]. Some have expressed concerns regarding the inhibition of osteogenesis by antibiotics [22]. A 2020 study explored local antibiotic use in spinal osteomyelitis, utilizing antibiotic-impregnated polymethylmethacrylate (PMMA) strut grafts. Mixing 1 g of infused tobramycin with 1 g of vancomycin powder in bone cement, researchers found that a PMMA spacers increase local antibiotic concentration, reduce the dead space, and ensure adequate stability [23]. We used 100 mg of vancomycin per 1 mL of bone, consistent with a previous study on one-stage revision arthroplasty [12]. In that study, the authors noted that vancomycin’s poor tissue penetration likely caused high tissue and low serum levels. Despite using comparatively high concentrations of antibiotics, no significant adverse effects were observed.

AIBGs have shown promising results in revision arthroplasty and trauma surgery [9,11,12,24]. Biofilm-embedded pathogens require a minimum inhibitory concentration of up to 1,000-fold for elimination. Local application of antibiotics can achieve the high concentrations required for biofilm elimination [25]. Borkhuu et al. [26] reported a significant decrease in postoperative infection rates in children undergoing cerebral palsy-related scoliosis surgery using AIBG (4%) versus plain bone grafts (15%). To the best of our knowledge, the present study is the first to employ AIBG with TIDF to treat PDVO. In this study, the added antibiotics were based on preoperative culture results; if not available, a combination of 2 g vancomycin and 80 mg gentamicin were used. Vancomycin was chosen for its efficacy against Gram-positive cocci, the most common pathogen in PDVO [2]. Gentamicin was chosen because of its synergistic effect with vancomycin against MSSA [27]. Additionally, vancomycin and gentamicin have better penetration and uptake into the intervertebral disc because of their positive charge [28]. Our selection of antibiotics is consistent with recent literature recommendations, advocating for pathogen-specific antimicrobial therapy, with vancomycin and gentamicin being commonly used [29].

Recent studies suggest that both percutaneous endoscopic lumbar debridement (PELD) and PLIF achieve comparable infection control outcomes for pyogenic spondylodiscitis treatment. However, PELD can result in shorter hospital stays, faster pain relief, and quicker functional recovery [30].

Some limitations of this study should be considered. First, the AIBG and non-AIBG groups were not propensity score-matched due to the retrospective study design and small sample sizes. However, the consistency in surgical indications and postoperative management between the two groups helped mitigate potential bias, ensuring a meaningful comparison. Second, we did not measure local or blood antibiotic concentrations. Measurement of antibiotic concentrations can help determine the causal relationship between each factor. This would be a key area of improvement in the next phase of our research. Third, the retrospective design might lead to selection bias and survivorship bias. Despite these limitations, this is the first study to compare the effectiveness of AIBG in this disease.