The importance of the bacterial spectrum in the clinical diagnostics and management of patients with spontaneous pyogenic spondylodiscitis and isolated spinal epidural empyema: a 20-year cohort study at a single spine center

Background Personalized clinical management of spondylodiscitis (SD) and isolated spinal epidural empyema (ISEE) is challenging due to limited evidence of microbiologic findings and their clinical impact during the clinical course of the disease. We aimed to characterize clinico-microbiological and imaging phenotypes of SD and ISEE to provide useful insights that could improve outcomes and potentially modify guidelines. Methods We performed chart review and collected data on the following parameters: bacterial antibiogram-resistogram, type of primary spinal infection, location of spinal infection, source of infection, method of detection, clinical complications (sepsis, septic embolism, and endocarditis), length of hospital and intensive care unit (ICU) stay, relapse rate, and disease-related mortality in patients with proven pyogenic SD and ISEE treated surgically in a university hospital in Germany between 2002 and 2022. Results We included data from 187 patients (125 SD, 66.8% and 62 ISEE, 33.2%). Gram-positive bacteria (GPB) were overall more frequently detected than gram-negative bacteria (GNB) (GPB: 162, 86.6% vs. GNB: 25, 13.4%, p < 0.001). Infective endocarditis was caused only by GPB (GPB: 23, 16.5% vs. GNB: 0, 0.0%, p = 0.046). Methicillin-susceptible Staphylococcus aureus was the most frequently isolated strain (MSSA: n = 100, 53.5%), occurred more frequently in the cervical spine compared to other bacteria (OB) (MSSA: 41, 41.0% vs. OB: 18, 20.7%, p = 0.004) and was most frequently detected in patients with skin infection as the primary source of infection (MSSA: 26, 40.6% vs. OB: 11, 16.7%, p = 0.002). Streptococcus spp. and Enterococcus spp. (SE: n = 31, 16.6%) were more often regarded as the cause of endocarditis (SE: 8, 27.6% vs. OB: 15, 11.4%, p = 0.037) and were less frequently detected in intraoperative specimens (SE: 19, 61.3% vs. OB: 138, 88.5%, p < 0.001). Enterobacterales (E: n = 20, 10.7%) were identified more frequently in urinary tract infections (E: 9, 50.0% vs. OB: 4, 3.6%, p < 0.001). Coagulase-negative Staphylococci (CoNS: n = 20, 10.7%) were characterized by a lower prevalence of sepsis (CoNS: 4, 20.0% vs. OB: 90, 53.9%, p = 0.004) and were more frequently detected in intraoperative specimens (CoNS: 20, 100. 0% vs. OB: 137, 82.0%, p = 0.048). Moreover, CoNS-associated cases showed a shorter length of ICU stay (CoNS: 2 [1–18] days vs. OB: 6 [1–53] days, median [interquartile range], p = 0.037), and occurred more frequently due to foreign body-associated infections (CoNS: 8, 61.5% vs. OB: 15, 12.8%, p = 0.008). The presence of methicillin-resistant Staphylococcus aureus (MRSA) prolonged hospital stay by 56 [24–58] days and ICU stay by 16 [1–44] days, whereas patients with Pseudomonas aeruginosa spent only 20 [18–29] days in the hospital and no day in the ICU 0 [0–5] days. Conclusions Our retrospective cohort study identified distinct bacterial-specific manifestations in pyogenic SD and ISEE regarding clinical course, neuroanatomic targets, method of pathogen detection, and sources of infection. The clinico-microbiological patterns varied depending on the specific pathogens.

Despite diagnostic and therapeutic improvements, PSI remains a challenging disease with a high mortality rate [12].Microbiological analysis of the causative bacteria plays a key role in the diagnostic and treatment of PSI [13], thus effective antibiotic treatment depends on the correct identification of the underlying bacterial species and their antimicrobial susceptibility.
A better understanding of the clinico-microbiological differences between the various bacterial species would improve the treatment and outcome, thus affecting the complication rate and length of stay in the hospital and intensive care unit (ICU).We analyzed the bacterial spectrum in our cohort of 187 surgical patients and assessed its impact on clinical disease course to determine a reproducible phenotyping of a bacterial population in SD and ISEE.

Study design and patient data
We performed a retrospective analysis of consecutive spontaneous PSI cases from 2002 to 2022 treated surgically at our Neurosurgery University Spine Center in Dresden, Germany.All patients with proven PSI who underwent surgical treatment were included.We excluded five cases with non-pyogenic pathogens, sixteen cases without any evidence of pathogens, eight patients with only conservative treatment, and twelve patients with intradural infection.A total of 187 patients were included in the analysis (Fig. 1).
The study was approved by the local ethics committee of the Carl Gustav Carus University Hospital in Dresden (Ref: BO-EK-17,012,022).Patients' data were collected via the ORBIS system (ORBIS, Dedalus, Bonn, Germany) and neuroimaging studies through the IMPAX system (IMPAX, Impax Asset Management Group plc, London, UK).
pneumonia.If the spinal infection clearly emerged from one of these three routes and had the same pathogen, we called it the source of infection.

Diagnosis and treatment
SD and ISEE were diagnosed on the basis of clinical, laboratory, microbiological, and radiological findings (magnetic resonance imaging (MRI) and CT) according to the guidelines of the Infectious Diseases Society of America (IDSA) [14].Clinical and laboratory parameters included back or neck pain, fever, leukocyte count, C-reactive protein (CrP), and Procalcitonin.
Depending on the clinical condition, at least two blood cultures were taken for microbiological diagnosis before antibiotic therapy was initiated, while some patients were initially treated with antibiotics in a peripheral hospital due to their severe clinical condition.In addition, tissue samples taken during open surgery and samples obtained by CT-guided biopsies were also used for microbiological analysis.We treated all patients surgically in different procedures depending on the location of the infection, clinical and laboratory findings, and severity of the bone defects.
Patients with hemodynamic instability, sepsis, septic shock, or progressive neurological symptoms underwent empiric antibiotic therapy (EAT) immediately after collection of blood cultures.On the other hand, patients with stable hemodynamic and neurological symptoms received targeted antibiotic therapy (TAT) after pathogen detection in the blood culture, image-guided biopsy, or open surgery.In the case of culture-negative spondylodiscitis (blood culture, image-guided biopsy, and surgery), we initiated an EAT.EAT was based on the suspected pathogen, suspected source of infection, clinical condition, epidemiologic risk, and local historical in vitro susceptibility data.In most cases, our therapy was a combination of vancomycin with ceftriaxone or vancomycin with piperacillin/tazobactam.
In patients with foreign body infections or osteosynthesis material (fixator as therapy), a combination such as flucloxacillin or vancomycin with rifampicin was used.Depending on the clinical condition, infection parameters and MRI findings, the IV-therapy was switched to an oral therapy for another 4 weeks.

Microbiological assessment
Blood culture Both the aerobic and anaerobic blood culture flasks (BACTEC Plus Aerobic/F and Anerobic/F) were incubated for a maximum of 14 days in the BAC-TECTM FX blood culture machine (BD, Heidelberg, Germany).After cultural growth was indicated, a Gram staining was performed.Furthermore, an agar diffusion test was carried out using a Müller-Hinton agar (bioMérieux, Nürtingen, Germany) and test platelets (bestbion dx, Köln, Germany) for specific antibiotics to provide a provisional resistogram.Cultivation from aerobic culture was performed by plating sample on a Columbia blood Fig. 1 Study design: One hundred eighty-seven patients with proven pyogenic spondylodiscitis (SD: 125) and isolated spinal epidural empyema (ISEE: 62) were included in this study.Forty-one patients were excluded based on exclusion criteria agar (bioMérieux, Nürtingen, Germany), on chocolate agar (bioMérieux, Nürtingen, Germany), bile Chryosidine Glycerol Agar (Becton Dickinson, Heidelberg, Germany), Sabauroud glucose agar (Becton Dickinson, Heidelberg, Germany) and on Columbia CNA agar with 5% sheep blood (Becton Dickinson, Heidelberg, Germany).HCG agar (Sifin Diagnostics GmbH, Berlin, Germany) was used to detect anaerobic bacterial.It was incubated in a culture pot "AnaeroJar" (Becton Dickinson, Heidelberg, Germany) under absorption of oxygen using AnaeroGen (Oxoid, Wesel, Germany).The aerobic cultures were incubated at 37 degrees for 24 h.The HCG agar was incubated for 48 h.As soon as bacterial colonies were detectable, the species were identified using MALDI TOF MS (Bruker Daltonics GmbH, Bremen, Germany).Tissue from CT-guided or intraoperative sampling Tissue obtained intraoperatively or CT-guided was placed directly in the operating theatre in Schaedler boullions (bioMérieux, Nürtingen, Germany) and these were subsequently sent to the Institute of Medical Microbiology and Virology, Carl Gustav Carus University Hospital in Dresden for analysis.There, the boullions were first incubated at 37 degrees and examined for turbidity after 48 h.Once turbidity was detected, the culture suspension was plated on Columbia Blood Agar (bioMérieux, Nürtin-gen, Germany) and HCB Agar (bioMérieux, Nürtingen, Germany).Aerobic culture growth was checked for the first time after 24 h and anaerobic culture growth after another 48 h.

Antimicrobial susceptibility
The final antimicrobial susceptibility testing for aerobic bacterial species was performed using the VITEK 2 system (bioMérieux, Nürtingen, Germany) and the respective antimicrobial susceptibility test (AST) cards (bioMérieux, Nürtingen, Germany).The analysis was performed according to the guidelines of the manufacturer.Antimicrobial susceptibility profiles of anaerobic bacteria were generated using the ATB ANA System (bioMérieux, Nürtingen, Germany) or Gradient diffusion tests (bestbion, Cologne, Germany).
Bacterial species that are known to be of definite human pathogenic relevance have been included in the evaluation if they occur in already one sample.In contrast, low-pathogenic species (e.g., facultative pathogens) were only included in the evaluation if they were detected in at least two independent clinical samples.The results of the resistograms were then compared to assess whether two isolates were of the same origin or not.

Formation of groups
The detected bacteria were grouped in gram-positive bacteria (GPB) and gram-negative bacteria (GNB) and further subgrouped according to the taxonomic classification, specific resistance phenotypes or characteristic properties (MSSA, Streptococcus spp.and Enterococcus spp.(SE), CoNS, methicillin-resistant Staphylococcus aureus (MRSA), bacillus spp., Pseudomonas aeruginosa, Enterobacterales (E), and anaerobic bacteria).Furthermore, Enterococcus spp.and infections showing a polymicrobial spectrum were analyzed separately.

Statistical analysis
Data were statistically analysed with the SPSS software package (SPSS Statistics 28, IBM, Armonk, New York, USA).Descriptive statistics were used, and categorical variables were adjusted by Fisher exact tests or chi-square tests.Numerical variables were analysed with Mann-Whitney U tests.All statistical tests were two-sided, and a p value p < 0.05 was considered statistically significant.
Multiple blood cultures and intraoperative samples were taken from all patients, whereas 65 tappable  2).

Anaerobic bacteria
Anaerobic bacteria were equally detected in SD and ISEE patients (SD: 3, ISEE: 3).Women were more frequently affected than men (women: 5, men: 1) and the median age was 80 [78-82] years.Two patients had septic embolism, and sepsis, but none had endocarditis or died, whereas one patient suffered a relapse.Patients spent 24.5 [13.5-34] days in the hospital and 1.5 [1-7.5]days on the ICU.All bacteria were identified only by intraoperative specimens.Three bacteria were found in the CS, one in the TS, and 2 in the LS.Retropharyngeal, and odontogenic infections were the primary of infections.

MRSA
The "MRSA subgroup" of 2 male and 3 female patients aged 81 [80-83] years was observed in 4 SD patients and one patient with ISEE.Sepsis was observed in 3 patients and endocarditis in 2 patients, but septic emboli did not occur.We found two relapses, but no in this subgroup.Prolonged hospital stay of 56  days and ICU stay of 16  days were recorded.In blood cultures and intraoperative specimens, four bacteria were detected, while two were detected in CT-guided biopsies.MRSA was diagnosed twice in the CS, once in the TS, and two times in the LS.Respiratory, gastrointestinal, and skin infections were the primary sources.

Pseudomonas aeruginosa
In three men and one woman aged 72 [71-75] years, P. aeruginosa caused three SD and one ISEE.Two patients had sepsis and one endocarditis, while no patient had a relapse or died.Patients spent 20 [18][19][20][21][22][23][24][25][26][27][28][29] days in the hospital and no day in the ICU 0 [0-5].All bacteria were identified in intraoperative specimens and 2 of them in blood culture.Two patients underwent CT-guided biopsy of the psoas muscles, which was successful in one of them.Three bacteria were found in the TS and one in the CS, whereas LS was not affected.Respiratory and retropharyngeal infections were the source of the infections.

Polymicrobial Infections
In our cohort polymicrobial infections caused only spondylodiscitis and were documented in 8 patients.Candida albicans accounted for half of the polymicrobial infections and it was always associated with S. aureus (2 MSSA, 2 MRSA).B. fragilis was founded in infections with P. aeruginosa.C. acnes was present as a co-infection in MSSA infection.S. epidermidis was detected in a S. anginosus infection.
Therapy was adjusted according to the antibiogram.One patient died during treatment, and relapse was observed in 3 patients.The source of infection was

Discussion
Our study provides insights into the bacterial spectrum and related clinical courses in SD and ISEE and identifies several bacterial group-related profiles.GPB were more frequently detected than GNB and exclusively caused infective endocarditis in our study.Previous studies showed that GPB was more common than GNB in ISEE and SD [8].
MSSA was the most common bacterium in SD and ISEE in our cohort, representing more than half of cases, which is consistent with the literature [12,[17][18][19][20][21][22].In our study, MSSA occurred more frequently in ISEE than in SD, which was also observed by Stangenberg et al. [12].MSSA occurred more frequently in CS than OB in our cohort, which previous studies have confirmed [12,[23][24][25].Skin infections and epidural administrations such as infiltration therapies were the most source of infection in the "MSSA subgroup".This is in line with studies showing that MSSA was common source of skin infections with spinal involvement [26][27][28].
The "SE subgroup" were the second most frequent bacterial subgroup in our study, causing 16.6% of infections, as reported in most studies [11,[29][30][31].Nonetheless, some authors observed S. epidermidis being the second common causative bacteria [12,22].Endocarditis was more frequently found in the "SE subgroup", in accordance with previous studies [11,29].Streptococcus spp.and Enterococcus spp.were detected more frequently in SD and less in ISEE, which has not been reported in literature yet.In accordance with the European diagnostic guidelines, we recommend first collecting blood cultures and then image-guided biopsy or open surgery, whereby the diagnostic sensitivity of intraoperative specimens in our study was lower in the "SE subgroup" (61.3%) and higher in the "CoNS subgroup" (100.0%)[Lazzeri, 2019 #48].Other methods such as blood culture and CTguided biopsy of psoas abscess showed no difference between the pathogen groups.
Infections with CoNS have increased recently due to the potential ability to biofilm formation and colonization in various surfaces, as well as the increasing use of medical implants [32].One possible other cause would be the increasing sonication examination of foreign bodies.CoNS was associated with a shorter ICU stay and less sepsis and could always be isolated by intraoperative specimens, which was due to the detection of infected foreign bodies.In line with other studies, the source of infection in CoNS in our collective was frequently a foreign body-associated infection such as port infection [11,[33][34][35].S. epidermidis is the second most common Bold values are significant results (p < 0.05) as indicated in the methods, MSSA: Methicillin-susceptible Staphylococcus aureus, CT: computed tomography, (1) Chi-squared test, (2) Fisher's exact test, (3) Mann-Whitney U test Bold values are significant results (p < 0.05) as indicated in the methods, CT: computed tomography, (1) : Chi-squared test, (2) : Fisher's exact test, (3) : Mann-Whitney U test Bold values are significant results (p < 0.05) as indicated in the methods, CT: computed tomography, (1) Chi-squared test, (2) Fisher's exact test, (3) Mann-Whitney U test single bacteria after MSSA in our cohort (n = 18, 9.6%) and forms the fourth largest subgroup of SD and ISEE.Th previous studies reported similar results with a few exceptions [12,22,35,36].Enterobacterales were the main representatives of GNB and caused 10.7% of SD and ISEE in our collective.The literature data on the frequency of this subgroup in SD and ISEE varies widely [12,22,35].Endocarditis was associated less frequently with Enterobacterales, in accordance with the literature data showing that endocarditis occurs in more than 80% of patients with GPB [11,23].Consistent with reported literature, the main source of infection in this subgroup was urinary tract infections [11,35].
The "Anaerobic subgroup" could only be detected in intraoperative specimens, which showed the importance of this procedure.MRSA increased hospital length of stay to 56 days and ICU length of stay to 16 days, demonstrating the complexity of managing this infection, whereas patients with P. aeruginosa spent only 20 days in the hospital and no day in the ICU.Polymicrobial infections showed a complicated course and were half associated with C. albicans, which has not been previously investigated.

Limitation and strengths of this study
Our study is limited by its retrospective design and a possible selection bias toward the more severe cases due to the high degree of specialization at our university center.Nevertheless, our data derived from detailed clinical, imaging, and microbiological state-of-the-art diagnostic assessment with high internal validity and a meaningful sample size Lastly, the generalizability of our observations is limited by the monocentric design of our study.The study spanned 20 years, and the number of patients per year is not large enough to make a conclusion about the distribution and transformation of the pathogens over a period of time.

Conclusions
Our study shows that the bacterial groups in SD and ISEE have distinct bacterial-related patterns, which may help modify guidelines towards more tailored management.
GPB dominated the bacterial spectrum of SD and ISEE and were solely responsible for endocarditis.The most frequently isolated strain was MSSA, which was found more often in patients with skin infections and in the CS compared to other bacteria.Streptococci and enterococci were more frequently isolated in intraoperative specimens, were more frequently considered the cause of endocarditis, and were more often found in SD than in ISEE.CoNS resulted in less sepsis, shorter ICU stay, and was frequently associated with foreign body infection.Enterobacterales were the main representatives of GNB Bold values are significant results (p < 0.05) as indicated in the methods, CT: computed tomography, (1) Chi-squared test, (2) Fisher's exact test, (3) Mann-Whitney U test that did not cause endocarditis and were associated with urinary tract infections.Anaerobic bacteria could only be detected in intraoperative specimens in our study.In addition, MRSA increased hospital and ICU length of stay.

Table 2
Gram-positive and Gram-negative bacteria in spinal infections Bold values are significant results (p < 0.05) as indicated in the methods, SD: spondylodiscitis, ISEE: isolated spinal epidural empyema, *Fisher exact test

Table 3
Clinical characteristics of methicillin-susceptible Staphylococcus aureus

Table 4
Clinical characteristics of Streptococcus spp.und Enterococcus spp

Table 5
Clinical characteristics of Enterobacterales

Table 6
Clinical characteristics of Coagulase-negative Staphylococci