Outcome of Surgical Treatment of Spontaneous Spinal Epidural Abscesses for a 10-year Period

Introduction: Spinal epidural abscess is uncommon but potentially debilitating infection. Delay in early diagnosis may be associated with increased morbidity and mortality despite recent advances in medicine. Aim: To present the clinical course and outcome of treatment of spontaneous spinal epidural abscesses. Materials and methods: Thirty-four patients (20 men and 14 women) with clinical, neuroimaging and/or histological data for spinal epidural abscess were treated at the Clinic of Neurosurgery at St George University Hospital, Plovdiv, Bulgaria, for the period 2009-2018. Results: The average age of patients was 62 years (21-76 years) and the ratio of men to women was 1.4:1. All patients (100%) presented with vertebralgia, 13 patients (38.2%) had additional radiculalgia, and 10 patients (29.4%) presented with sensory or motor deficit. The duration of complaints varied from 4 to 180 days. At hospital admission, only 9 patients (26.4%) had intact neurological status. The most common localization of the spinal epidural abscess was in the lumbar and lumbosacral area (52.9%), concomitant spondylodiscitis was present in 31 patients (91.2%). Twenty-four patients (70.6%) underwent emergency surgery within 24 hours, and the rest had planned surgery. Decompressive interlaminotomy or hemilaminectomy was performed in 9 patients (26.5%). The remaining 25 patients (73.5%) underwent laminectomy, in 15 patients (44.1%) it was combined with posterior pedicle screw fixation. After the treatment, 23 patients (67.6%) had a good outcome, the remaining 11 (32.4%) had a poor outcome, and 3 patients died (8.8%). Conclusion: In patients with spinal epidural abscess, emergency surgery is the treatment method of choice. It allows decompression of neural structures, correction of the spinal deformity, segmental stabilization and rapid mobilization of patients.


INTRODUCTION
The spinal epidural abscess (SEA) was first described by G. B. Morgagni in Venice in 1761. This rare but serious infectious disease has an incidence rate of 0.2 to 2.8 cases per 10,000 hospital admissions per year, with a tendency to increase due to population aging, increased surgical activity, and intravenous drug use. [1][2][3] Nowadays, despite the advances of modern medicine and the pharmaceutical industry, the outcome can still be poor or even fatal, with mortality rates remaining relatively high ranging from 15% to 23%. 1,4 Spontaneous or (unrelated to medical manipulations, surgery or spinal injuries) SEAs have an incidence rate of 0.8 cases per 100,000 people. They are classified as primary (PSEA) or secondary spinal epidural abscesses (SSAE). The PSEAs result from hematogenous dissemination from an infectious focus elsewhere in the body, directly into the spinal epidural space, whereas the SSAEs are caused by adjacent spontaneous spondylitis or paravertebral abscess. [5][6][7][8] Risk factors for the onset of SEA are suppressed immunity, due to acquired immune deficiency syndrome (AIDS) or immunosuppressive therapy after organ transplantation, diabetes mellitus, alcohol addiction, neoplastic and systemic inflammatory diseases. 2

AIM
The aim of this study was to examine the epidemiology, risk factors, etiology, clinical manifestation, localization, paraclinical studies, surgical treatment, and outcome in spontaneous SEAs in order to facilitate early diagnosis and adequate treatment.

MATERIALS AND METHODS
We conducted a retrospective analysis of 34 patients (20 males and 14 females), between 21 and 76 years of age [57.647±13.172 CI 95% (53.048-62.246), median 62, p=0.0286], with spontaneous SEA, treated between 2009 and 2018 at the Clinic of Neurosurgery in St George University Hospital, Plovdiv, Bulgaria. All patients were preoperatively diagnosed by magnetic resonance imaging (MRI) and/or computed tomography (CT). The diagnosis was confirmed during the surgery. The patients' case histories, clinical presentation, paraclinic, imaging, and operative protocols were investigated. Attention was paid to the initial symptoms, clinical presentation at admission, concomitant diseases, localization and number of segments affected, laboratory parameters, microbiological results, timing and type of surgical intervention. Fever was considered important when the axial temperature was ≥37.8°C. The severity of pain was evaluated using the Denis scale, the degree of myelopathy was assessed by the Frankel scale, and the degree of disability -by the modified Rankin Scale (mRS). The stage of the disease in each patient at admission was retrospectively classified according to the Heusner criteria. 9 Stage I -presence of vertebralgia only, stage II -vertebralgia and radiculopathy, stage III -motor and sensory deficits and pelvic reservoir dysfunction, and stage IV -paralysis. Regarding the blood tests, attention was paid to the leukocyte count, erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) level. All collected data were analyzed to evaluate the possible correlation with the disease outcome and to determine their prognostic value. The outcome of the disease was evaluated on a 5-point scale suggested by Kumar and Hunter, as follows: 1. Complete recovery of the neurological deficit; 2. Ability for self-care without pelvic-reservoir disorders; 3. Self-care with support with or without sphincter disorders; 4. Quadriplegia/ paraplegia or severe quadriaparesis/paraparesis keeping the patient to bed; 5. Death.
For the purposes of the statistical analysis, Grades 1 and 2 were grouped as a 'good outcome' and grades 3-5 as a 'poor outcome' . 10 A chi-squared test was used to compare categorical data, a non-parametric Mann-Whitney U test was used to compare variables, grouped in rank scales, and a t-test -for normally distributed data. The bilateral significance level was set at p <0.05.

RESULTS
The current study established that the incidence of SEA  was 0.005% or 0.5 per 10,000 hospitalizations, while based on the number of surgically treated patients it was calculated to be 0.3% or 3 per 1,000 hospitalizations.
Only 3 patients (8.8%) from our series were diagnosed with primary SEAs because the infection was restricted only to the spinal epidural space (Fig. 1). The remaining 31 (91.2%) were calssified as secondary SEAs due to the presence of adjacent spondylodiscitis with or without a paravertebral abscess (Fig. 2).
The demographics and clinical data for the patients are presented in Table 1.
In 6 patients (17.6%), the leukocyte count was normal, while in the rest it was slightly increased -with an aver-age of 13.2×109/l (4.50-24.86) ( Table 1). In all patients, the ESR was significantly increased -with an average of 84.50 mm/h (ranging from 10 to 120) ( Table 1). Only one patient had normal CRP but he was prevoisly treated for bronchopneumonia. In all other patients, it was significantly increased -with an average of 97.5 mg/l (ranging from 10 to 267) ( Table 1).
Only 3 patients (8.8%) had positive serological tests. The causative agent of SEA from the extracted materials during surgery (bone, disc, and pus) was isolated in 21 patients (61.8%) ( Table 1). Staphylococcus aureus and Coagulase-negative staphylococci were the most common agents that caused SEA in our group -100% of primary and 38.7% of secondary SEA. The second most frequent were the cases caused by Mycobacterium tuberculosis (11.8%). The microbiological study did not isolate the causative agent in 13 patients (38.3%).
Arterial hypertension and diabetes mellitus were the most common concomitant diseases in 14 (41.2%) and 9 (26.5%) patients, respectively. Other concomitant diseases were inflammatory diseases and tuberculosis in 4 patients each (11.8%), liver disease in 2 patients (5.9%), kidney failure, lung failure and cancer in one patient each.
In 18 patients (52.9%), the cause of the spinal infection remained unknown. In the remaining 16 patients (47.1%), the cause was uncertain, except for 4 patients with proven tuberculosis and 6 patients with inflammatory diseases with an identical microbiological agent.
In all operated patients, we used a posterior or a posterolateral operative access. The purpose of the surgical treatment was to achieve decompression of neural structures, removal of necrotic tissue, and evacuation of the epidural abscess or granulation tissue. In 9 patients (26.5%), in whom the clinical picture was lumbar radiculopathy alone, debridement and decompression of the affected nerve and cauda equine were performed by interlaminotomy on one or more levels. In cases with pronounced narrowing of the spinal canal and compression of medulla spinalis or cauda equina from the present epidural abscess, we performed a hemi-or laminectomy. We performed this intervention in 25 patients (75.8%). In these cases, in order to avoid instability of the affected segments, a posterior pedicle screw fixation was also performed in 15 patients (45.6%) (Fig. 3). The remaining 9 patients refused instrumentation.
The antibiotic therapy was administered in accordance with the results from the microbiological examination and when they were sterile, an intravenous administration of a combination of broad-spectrum antibiotics for 14 days was prescribed, followed by oral antibiotic administration for 4-6 weeks.

DISCUSSION
Spinal epidural abscess can occur in any age group, but it is more common in patients between 50 and 70 years of age and in males. 4,11 In our series, the 50-70 age group was the largest (79.4%), confirming that it is the most vul-nerable age. Although significantly smaller, the group of patients under 50 (20.6%) indicates that the young patients can also be affected by this devastating disease. The distribution of males and females in our series (58.8%:41.2%) is in line with the findings of two extensive meta-analyses by Reihsaus et al. and Arko et al. 4,12 Unlike Chen et al., our analysis found no significant importance of age for treatment outcomes (p=0.646). 13 According to the literature, one of the most common risk factors for the occurrence of SEA is diabetes mellitus, with a frequency ranging from 15.4% to 35.6%. 7,8,13,14 Diabetes mellitus was present in 9 of our patients (26.5%), but  13 Tang et al. reported that the duration of symptoms before hospitalization varies from 1 to 180 days, but most authors consider it to be in the range of 1 day to 2 months. 2,15,16 The delay in diagnosis is due to the lack of specific initial symptoms that resemble degenerative disease, inconsistent presence of fever and attempting conservative treatment without imaging diagnostics. 17 This leads to the progression of SSEA and neurological deficit, as found in 26 of our patients (75.6%), as well as impairment of their functional status and degree of disability. The latter two indicators have a statistically significant effect on treatment outcomes (p=0 .023 and p=0.021).
The time between symptom onset and diagnosis in our patients ranged from 4 to 180 days, with an average of 30 days [mean -49.853±47.845; CI 95% (33.149-66.557); p<0.0001]. It is noteworthy that for the three patients with PSEA, this period was significantly shorter -4, 7, and 12 days, respectively, due to the faster development of a neurological deficit from the directly developing infection into the spinal epidural space.
The classical triad of vertebralgia, temperature ≥38°C and a progressive neurological deficit is often associated with SEA. 8 Unfortunately, it is present in the clinical presentation in 33-40% of the cases, which is confirmed by the 38.2% we found, contributing to the delay of diagnosis. 8,14 Our study confirms that MRI is the diagnostic tool of choice for suspected SEAs. [2][3][4] It allows complete visualization of the spine, neural structures and infectious lesions in all of its anatomical compartments (vertebrae, paravertebral and epidural space). In PSEA, there is no involvement of bony and paravertebral structures. The abscess is mainly localized in the dorsal epidural space, due to the wider posterior spinal epidural space and venous plexus in this area, which is confirmed by our results. 6,11 In SSEA, in addition to a purulent collection, located predominantly ventrally, the classical MRI findings for spondylodiscitis are typical (low signal from the affected vertebral bodies and destruction of their end plates in T1 sequence, and in T2 sequence -high signal from the affected body and disk). 18 During the progression of the disease, the infection spreads throughout the spinal epidural space, both in the affected segments and in the vertical direction, due to the lack of anatomic resistance. 13,14,18 An increase in the number of affected segments shows a statistically significant negative effect on the postoperative outcome (p=0.41).
There is no consensus on the most common localization of SEAs. In the study of Khan et al., PSEAs are located equally in the cervical, thoracic and lumbar spine, whereas in the study of Ptaszynski et al. they are more common in the lumbosacral spine (66.6%). 6,7 The anatomical distribution of PSEAs in our group overlaps the findings of Khan et al. 6 In the meta-analysis by Arko et al., out of a total of 1,099 patients with a SEA, the involvement of the lumbar spine was in 48%, and of the thoracic -in 31%, which is similar to our findings of the localization of SSEA -48.5% in the lumbar spine and 38.7% in the thoracic spine. 12 Our analysis showed that the localization of SEA did not significantly affect treatment outcomes (p=0.391).
The world literature lacks randomized multicenter studies of the results of the different types of SEA treatments, therefore there is no consensus on the therapeutic strategy. In two meta-analyses for the period 1954-1997 and for the period after the year 1999, the advantage of the surgical methods in combination with the administration of antibiotics was found, in 89% and 60%, respectively. 4,12 Most authors accept that emergency surgical intervention is needed to avoid or eliminate debilitating neurological deficit, immediately after the diagnosis of SEA, which in our study is significantly related to postoperative results (p=0.010). 12,19 Like most authors, we also used posterior operative approaches that allowed good decompression and debridement of infected structures. To prevent future instability, we also used instrumentation with titanium implants, which did not increase the recurrence of infection. 20 Our study confirmed that the most common cause of SEA is Staphylococcus aureus, although our findings (30.4%) are significantly less than those reported in literature (63.6%-66.4%). 4,12 The type of infectious agent did not affect the outcome of treatment (p=0.686). The high percentage of patients (34.8%) in whom no causative agent was isolated, in our opinion was due to the chronic phase of the infection in our patients, as well as the inadequate collection of material and relatively long transportation times.
The major problem with SEA is not the treatment but the early diagnosis prior to development of severe neurological symptoms. 4 The duration and severity of the neurological deficit in SEAs complicate the prognosis. 4,18 About 1/3 of the surviving patients have an unfavourable neurological status, which is primarily due to incorrect or delayed diagnosis. 2,3,21 Our study makes no exception to this -32.5% of our patients had poor outcome, which was not significantly correlated to the time to diagnosis (p=0.322), and with the stage of the disease (p=0.190).

CONCLUSION
A large proportion of patients with SEA do not receive early diagnosis and timely treatment due to the nonspecific initial clinical symptoms. Clinicians need to suspect the disease in patients with vertebral pain in all segments of the spine, with or without history of fever of febrile episodes, with paraclinical data of leukocytosis, accelerated ESR and increased C-reactive protein, especially in people suffering from diabetes mellitus or other concomitant diseases. Magnetic resonance imaging is essential for the early diagnosis, before progression of irreversible neurological deficit. The surgical treatment in these cases allows maximal recovery with minimal hospital stay. In cases with instability of the affected segments, spinal instrumentation provides the necessary stabilization and avoids prolonged immobilization.