Antibiotic-impregnated bone graft decreases time to CRP normalization in pyogenic discitis with vertebral osteomyelitis

Surgical for Although the posterior-only approach has been a safe, effective procedure that minimizes the and complications of anterior or staged parenteral antibiotic treatment for 4-6 weeks postoperatively is still necessary. We hypothesized that antibiotic-impregnated bone graft used in an all posterior approach could result in infection control and shorten the postoperative course of pyogenic discitis and vertebral osteomyelitis. or ototoxicity. We found no signicant change in renal function during follow-up.


Introduction
Surgical treatment of pyogenic discitis with vertebral osteomyelitis (PDVO) is indicated for neurologic de cit, spinal instability, unknown pathogen, poorly controlled infection, or intractable pain. Although the posterior-only approach has been proved a safe, effective procedure that minimizes the risks and complications of anterior or staged surgery, parenteral antibiotic treatment for 4-6 weeks postoperatively is still necessary. We hypothesized that antibiotic-impregnated bone graft used in an all posterior approach could result in infection control and shorten the postoperative course of pyogenic discitis and vertebral osteomyelitis.

Results
Surgical treatment for PDVO resulted in clinical improvement and adequate infection control. There was no difference in pre-op CRP levels between the two groups. Despite shorter post-op IV antibiotic duration (Mean 21.0 d vs 39.8 d), the AIBG group had a stable decline in CRP levels and continued to decrease at 1, 2, 4 & 6 wks, with signi cantly lower CRP levels at 6 weeks compared to bone graft without antibiotics.
VAS scores improved from a mean of 7.2 to 2.3 one month postoperatively. Patients who had increased kyphotic angles had an average angle correction of 7.9° at last follow-up. Eighty-one percent of patients (17/21) had good to excellent functional outcomes.

Conclusion
TIDF combined with AIBG can achieve local infection control with faster reduction in CRP, leading to shorter antibiotic duration for pyogenic discitis and vertebral osteomyelitis Background Pyogenic discitis and vertebral osteomyelitis (PDVO) accounts for the majority of spinal infections, which commonly involve the lumbar region [1]. Although a wide variety of pathogens have been associated with PDVO, Staphylococcus auerus remains the most predominant [2]. Treatment varies depending on infection status, clinical symptoms, and comorbidities [2]. Surgical treatment may be indicated for those who respond poorly to antibiotic treatment or those who have spinal instability, extensive bony destruction, epidural abscess, neurological impairment, or intractable pain, or those whose needle biopsies fail to identify a de nite pathogen [2][3][4][5][6]. Because the lesions are almost always located in the vertebral body and disc space, anterior debridement with bone graft from anterior, anterolateral (transthoracic), or lateral retroperitoneal approaches with or without posterior stabilization have been strongly advocated [7][8][9][10], though anterior approaches are also associated with higher complication rates than posterior approaches [11,12].
Today, an increasing number of studies demonstrate the effectiveness of posterior-only approaches for the treatment of PDVO [5,[13][14][15]. Although the posterior approach has proven to be safe and effective with minimal risk and fewer complications than anterior approaches, there remains a need for 4 to 6 weeks of postoperative IV antibiotic treatment. Transforaminal interbody debridement and fusion (TIDF) is a posterior only approach that has been reported to achieve satisfactory results, though it also requires a long duration of post-operative antibiotic treatment [5,16].
Recently, use of antibiotic-impregnated bone graft (AIBG) has yielded encouraging results [17][18][19][20][21]. The major bene t of this technique is a high local concentration of antibiotics, greatly exceeding concentrations following systemic administration [22]. Most studies discuss the use of AIBG in the surgical treatment of orthopedic trauma or joint arthroplasty infections [17,18,20,21], but few focus on the use of AIBG in spine surgery.
In this study, the effectiveness of TIDF with AIBG for PDVO is assessed, with the hypothesis that this technique would shorten the time needed for post-operative IV antibiotics as well as time to normal CRP levels.
Methods 21 consecutive PVDO patients treated with TIDF with AIBG at a single institution between March 2014 and January 2017 with minimum of 2 years follow-up were retrospectively enrolled. Clinical and radiographic data were collected for all patients with thoracic or lumbar spine infections and who presented with neurologic de cit, spine instability, or intractable pain, and whose pathogens had not been identi ed or whose response to antibiotics had failed. Patients with tuberculosis bacilli infection and post-spinal operation infections were excluded. The study was approved by the institutional review board (CRMH IRB No. 201801075B0).
From the medical record, demographic data, diagnosis, affected levels, placement of screws, imaging results, laboratory data, duration of antibiotic treatment, neurologic status, and functional outcomes were collected. All patients wore Taylor braces for at least 3 months after surgery. Patients were followed-up at two weeks, and then one, three, and six months, then annually until at least 2 years.
Choice of parenteral antibiotics was based opinion of an infectious disease specialist and culture results.
Because some patients initially receiving conservative treatment, antibiotic treatment of some patients began prior to their operations. Serial C-reactive protein (CRP) levels were collected. The normal range of CRP is below 5.0 mg/L. The duration of treatment using parenteral antibiotics was determined based on clinical symptoms and lab parameters. If the patient's CRP declined to below 25 mg/L and they were afebrile without any signs of infection, we discharged the patient on oral antibiotics. All patients were prescribed oral antibiotics for 2 ~ 3 months after discharge.
The clinical outcomes assessed were VAS for back pain and the American Spinal Injury Association (ASIA) impairment scale for neurological status. In addition, local kyphotic angles of the superior and inferior endplates of the affected levels were measured on standing XR before and after surgery and at nal follow-up. Fusion was assessed by two of our authors who didn't participate in the surgery on the nal follow up x-ray and de ned as continuous trabeculae bridging across the interface of the treated segments, whereas the presence of any radiolucent interruption at the treated segments was de ned as pseudarthrosis. Functional outcomes were evaluated using the Kirkaldy-Willis classi cation [23]. Finally, CRP levels and duration of antibiotic administration in the AIBG patients was compared to a control group [5,11] from a cohort of 28 patients receiving an identical TIDF procedure with the same indications but without AIBG.

Statistical analysis
Differences between the "non-antibiotic-impregnated control group" and the "antibiotic-impregnated group" were compared using the Mann-Whitney U test. All statistical operations were performed using SPSS version 22.0 (IBM, Armonk, NY, USA). A p-value < 0.05 was considered signi cant.

Operative procedures
A modi ed transforaminal lumbar interbody fusion (TLIF) surgical approach was performed. This single posterior approach included debridement, fusion, and pedicle screw instrumentation. A midline open approach was used keeping the posterior elements of the spine as intact as possible. The paraspinal muscles were detached subperiosteally. Unilateral facetectomy and limited laminotomy were performed (Fig. 1A), followed by radical debridement to remove abscess, infected disc material, endplates, and sequestrum from the intervertebral space by curved curettes and disc roungeurs. (Fig. 1B). The site of entry was determined based on the expected extend of debridement necessary based on symptom severity, radiographs, CT, and MRI. Posterior stability was maintained by preserving the spinous processes, supraspinous ligaments, and the contralateral facet joint if possible.
Tissue samples were analyzed for aerobic, anaerobic, tuberculous bacilli, and fungi. Residual tissue was sent for histopathologic examination. If iliac autograft was not su cient to ll the residual bone defect after debridement, allograft was added to the autograft (Fig. 3D).
Graft was mixed with antibiotics based on the results of pre-operative cultures if available. If no culture result was available, vancomycin 2 g plus gentamicin 80 mg was added to ensure both gram-positive and gram-negative coverage. After radical debridement, the resulting void was packed with AIBG as solidly as possible (Fig. 1C). Sites of xation were based on bone quality and clinical condition. Pedicle screws were inserted into segments with su cient bone mass to secure the screw and far enough away from the destroyed areas of bone to avoid recurrence of infection.

Results
Thirteen men and eight women with a mean age of 62.7 years old (34-85) underwent TIDF with AIBG. Patients were followed for a mean of 38.6 months . Although there were no infection-related deaths, one patient expired at 36 months as a result of intracranial hemorrhage. Prior to surgery, the most common presenting symptoms were back pain (> 90%) followed by bilateral leg symptoms in ten cases (47.6%) and unilateral symptoms in four cases (19%). Ten patients had epidural abscess and eleven had pre-existing spinal degenerative changes such as hypertrophic ligamentum avum and degenerative disc disease. Fourteen patients (66.7%) received parenteral antibiotics prior to surgery. The affected levels were T10-11 (2 patients), T12-L1 (2 patients), T12-L2 (1 patient), L1-2 (1 patient), L2-3 (3 patients), L3-4 (4 patients), L4-5 (3 patients), and L5-S1 (5 patients). Pedicle screws were used in most cases (19/21, 90%). Two patients receive TIDF without screws. Nine patients received one pedicle screw above the lesion segment and one below, ve patients received two above and two below, three patients received three above and three below, one patient received one above and two below, and one patient received two above and one below.

Neurologic status
Neurologic status was also evaluated. Prior to surgery, 16 patients had a neurologic status of ASIA D and ve were ASIA E. Three months post-operatively, 20 patients were ASIA E. None of the patients suffered neurologic injury during the operation.

Comparison to bone graft without antibiotics
We compared the length of time these 21 AIBG-treated patients received post-operative antibiotic treatment with that of 28 patients (controls) consecutively treated with the same TIDF procedure but without antibiotics in the bone graft in one of our previous studies [5]. As can be seen in Table 1, Table 2, and Fig. 2, both groups had stable CRP decline after surgical treatment. There was no difference in pre-op CRP levels (63.9 vs 56.5, p = 0.669). Even with a shorter mean post-op IV antibiotic duration (21.0 days vs 39.8 days, p < 0.005), the AIBG group had stable decline in mean CRP level that was closer to the normal range at 6 weeks post op (CRP: 13 mg/L vs 5.2 mg/L, p < 0.05). The AIBG group had a greater reduction in CRP than the control group (Tables 1 and 2). Table 1 Comparison between Non-AIBG and AIBG for the treatment of PDVO

Functional Outcome
Mean back pain VAS improved from 7.2 ± 0.9 (range: 4.5-8.8) pre-operatively to 2.3 ± 1.3 (range: 0.7-5.5) at one-month follow-up. Two patients complained of donor site pain at initial follow-up, which was relieved at nal follow-up.
Kirkaldy-Willis functional outcome at the last follow-up showed excellent results in 8 cases; good results in 9 cases; fair results in 3 cases; and poor results in 1 patient who underwent revision surgery due to a deep wound infection.

Revisions
Two patients had revision surgery. One was due to pseudarthrosis with broken S1 screws one year postoperatively. Revision posterolateral fusion performed was with allograft and demineralized bone matrix (DBM). Subsequent standing X-rays showed union with clinical back pain relief. The other patient developed a deep wound infection one-month post-TIDF. Anterior debridement and corpectomy with strut allograft reconstruction was performed for this patient.

Discussion
Parenteral antibiotic treatment is the primary treatment for PDVO; however, surgical debridement may be indicated in patients when conservative treament fails. Some have suggested that infections can be better controlled if the patient receives early surgical debridement with parenteral antibiotics [24,25]. The previous gold-standard surgical approach for PDVO was the anterior approach. Although this approach may enable better debridement, it is accompanied by a higher morbidity rate than the posterior approach [26,27]. Thus, the posterior-only approach has recently become more popular. It is performed with equivalent infection control but with lower surgical risk than the anterior approach [13,28]. Earlier studies of posterior lumbar interbody fusion (PLIF) with instrumentation reported good infection control [13,28]. PLIF requires a laminectomy and retraction of the nerve roots, which may injure the nerve and result in spinal instability. TIDF plus pedicle screw xation was thus invented to treat PDVO [5]. This method prevents postoperative instability and decreases the risk of nerve injury. Previous studies have reported promising results with the use of TIDF combined with posterior instrumentation to treat either nonpyogenic or pyogenic lumbar spine infection [5,16]. This procedure offers immediate stability to prevent further kyphotic change and accelerates infection control. As a result, patients bene t from decreased back pain, decreased use of pain medicine, and improved quality of life [5,16].

Parenteral antibiotic treatment in PDVO
Although TIDF surgery can remove infected disc, endplate, and sequestrum in the treatment of PDVO, post operative antibiotics are always required. There is no consensus as regarding the period of intravenous antibiotics, and treatment is mainly based on personal experience or expert opinion rather than evidence-based recommendations. Bernard et al. suggested that the standard duration of antibiotics treatment for pyogenic vertebral osteomyelitis could be reduced to six weeks in a multicenter noninferiority trial [29]; this reduction in antibiotic treatment period also decreases the risk of side effects, economic burden, and development of antibiotic resistance [8,30]. Park et al. further considered the optimal duration of antibiotics therapy based on the patient's risk of recurrence [31]. In that study, 314 patients were evaluated for recurrence and multivariate analysis revealed methicillin-resistant S. aureus (MRSA) infection, undrained paravertebral/psoas abscesses, and end-stage renal disease to be independent baseline risk factors for recurrence. They concluded that a prolonged duration (≥ 8 weeks) of antibiotics therapy should be used in this high-risk group. In this study, no patients met these criteria. Thus, duration of post-operative antibiotics treatment was decided on based on relief of clinical symptoms and improvement in lab values. Several previous studies suggest post-operative antibiotic treatment for as short as four weeks can be effective [5,16]. In the current study, mean post-operative IV antibiotics duration was only 21 days. (Table 1).

Antibiotic impregnated bone graft (AIBG)
AIBG has been widely used in revision arthroplasty and trauma surgery with promising results [17,20,21,32]. Bio lm-embedded pathogens require an MIC of up to 1000-fold for elimination. Only local application of antibiotics can achieve the high concentrations required to eliminate the bio lm [33]. Although antitiotic impregnated Polymethylmethacrylate (PMMA) cement is frequently used as a temporary spacer when treating infected arthroplasty, it was not used in these cases. Borkhuu et al. reported a signi cant decrease in infection rate after spinal surgery in children with scoliosis due to cerebral palsy after local use of AIBG (4%) as compared with a group in which plain bone graft was inserted (15%) [34]. To best of our knowledge, our study is the rst to employ AIBG with TIDF to treat PDVO. Antibiotics added to the bone graft in this study were based on pre-operative culture results. If the patient had no pre-operative culture result, vancomycin 2 g combined with gentamicin 80 mg were used to cover both gram-positive and gram-negative bacteria. Vancomycin was chosen because it is has been found to be e cacious against gram-positive cocci, the most common cause of PDVO [2]. Gentamycin was chosen because its synergic effect with vancomycin against MSSA [35,36]. Another advantage of vancomycin and gentamicin is better penetration and uptake into the intervertebral disc because they are positively charged [37,38].

Dose of antibiotic in bone graft
There is no consensus on the optimal dose for local antibiotic use [17]. Some concerns about inhibition of osteogenesis by antibiotics have been raised [39], though no signi cant difference in the osteogenic activity was noted in one animal study [40]. We used the same dose that a previous study used in onestage revision arthroplasty, 100 mg vancomycin per 1 cc of bone [21]. In that article, the author mentioned that high tissue and low serum levels of vancomycin could be due to its poor tissue penetration. Although very high concentrations of antibiotics were used, no signi cant adverse effects were noted. Seber et al. measured gentamicin levels in urine after insertion of gentamicin-loaded xenograft in the treatment of osteomyelitis and found the mean level at 24 h to be 4 µg/mL, falling below the effective level of 0.5 µg/mL after 8 days. [41] They found no cases of nephrotoxicity or ototoxicity. We found no signi cant change in renal function during follow-up.

Limitations
One limitation of this study is the small number of patients and heterogeneity of affected levels. Another limitation is use of the ASIA scale to assess neurological status. Since most cases were in the lumbar spine, this scale may not be su ciently sensitive. Still, thirty percent of patients had throracolumber junction lesions, so this assessment was included for comparison purposes. Finally, the positive total positive culture rate was 71.4% and the intra-operative positive culture rate was low (52.4%) compared to other studies [5,16,42]. This lower positive culture rate may be related to the fact that our hospital is a tertiary care center, and some of the referred patients had been receiving parenteral antibiotics for a period of time before cultures were obtained.

Conclusion
Based on our study, TIDF with AIBG is an effective and safe option in the treatment of pyogenic discitis and vertebral osteomyelitis. This technique can be used to achieve superior local infection control as evidenced by faster normalization of CRP levels, possibly shorten the duration of post-operative IV antibiotic use, correct the kyphotic angle, and achieve satisfactory outcomes.

Availability of data and materials
All data generated or analyzed during this study are included in this published article.

Competing interests
The authors declare that they have no competing interests

Funding
There is no funding support in this study.     A 54 y/o female woman with PDVO at T12/L1 level had a pre-operative CRP level of 29.4 mg/L. She also had multiple lumbar degenerative disc disease and degenerative spondylolisthesis with lumbar spinal stenosis at L45 area. Figure 3A: L-spine XR and spinal CT showed T12/L1 PDVO with endplate destruction and the pre-op Cobb angle was 24°. Figure 1C: Necrotic tissue and debris were currated and removed through transforaminal route at T12/L1 level. Figure 1D: Auto-iliac bone graft combined allograft was mixed homogeneously with antibiotics. Figure 1E: Bone graft was rmly impacted into the disc level (white arrow) and the post-op 1 month Cobb angle was 12 degree. Patient received 2 weeks of post-surgery IV antibiotics and the CRP level declined to7.2 mg/L,then the patient was transitioned to oral antibiotics. Figure 1F: Post-op 2 years showed solid fusion at T12/L1 level (black arrow) and the residual Cobb angle was 17°.

Figure 4
A 60-year-old woman with PDVO at T12/L1 level had a pre-operative CRP level of 60.7 mg/L. Figure 4A-B: L-spine XR and MRI showed T12/L1 PDVO with endplate destruction and the pre-op Cobb angle was 15.8°. Figure 4C: Necrotic tissue and debris were removed after transforaminal T12/L1 interbody debridement. Figure 4D: Auto-iliac bone graft was mixed homogeneously with antibiotics. Figure 4E: Bone graft was rmly impacted into the disc level. Patient received 1 week of post-surgery IV antibiotics, the patient was transitioned to oral antibiotics, and CRP level continued to decline to 1.3 mg/L the following week. Figure 4F: Post-op immediately showed the initial kyphotic Cobb angle was corrected to 4.1°. Figure 4G: A solid T12/L1 interbody fusion was observed 1 year postoperatively and the residual Cobb angle was 8.3°