Introduction
Pyogenic spondylodiscitis is a rare disease with high morbidity and consecutive long-term sequelae [
1]. The bacterial infection is usually of hematogenic origin and transmitted from the oral cavity, the respiratory tract or the skin [
2,
3]. The primary site of infection is the avascular disc, the infection may then spread continually into the vertebral body and up into the subligamentous paravertebral area, epidural space and contiguous vertebral bodies. In consequence, back pain and limitation of spine movement are the predominant signs in patients with spondylodiscitis [
4,
5]. Treatment concepts of pyogenic spondylodiscitis include either a conservative or a surgical management and should be evaluated individually. The currently available therapeutic guidelines are not standardised and based on individual preferences resulting in a high variability of outcome with conflicting results [
6‐
8].
The conservative treatment consists of germ-compatible long-term antibiotic treatment and immobilisation of the spine [
5,
9]. In patients with progressive neurologic deficits, large abscesses, progressive deformities or involvement of at least two adjacent vertebral bodies, surgical management should be favoured [
10]. Surgical strategies are versatile and include extensive debridement of the disc and vertebral bodies, stabilisation and titanium mesh cages [
10,
11]. Hence, patients treated for pyogenic spondylodiscitis harbour a numerous amount of well-established risk factors for pulmonary embolism such as immobilisation and prolonged bed rest, infection and bacteremia, operation and neurological deficits such as paralysis or paraplegia [
12,
13]. The currently lacking literature on patients with pyogenic spondylodiscitis and pulmonary embolism in the scientific literature prompted this study. Per our hypothesis, patients with pulmonary embolism may accommodate risk factors that are currently unknown to the literature. The identification of such risk factors may contribute to future concepts for thromboprophylaxis in this demanding cohort.
Methods
Study design
The present analysis is a retrospective, single centre observational study of patients with pyogenic spondylodiscitis. The hypothesis of the study was that several, to date largely unknown, risk factors contribute to the development of pulmonary embolism in patients with pyogenic spondylodiscitis. The identification of these could lead to a better risk stratification and adjusted anticoagulation regimes in this vulnerable cohort and potentially improve outcome.
Patients and data collection
For this retrospective analysis, an ethical approval was obtained by the local ethics committee (identification number: 20-683). As a non-interventional single-centre study no patient consent was necessary. Patients over 18-years old who were surgically or conservatively treated for pyogenic spondylodiscitis from 2012 to 2017 were identified retrospectively. Diagnosis of pyogenic spondylodiscitis was made upon clinical evaluation and spinal MRI with contrast agent. Indication for surgery was based upon individual case by case discussion including patients will, neurologic deficit, large abscesses, or involvement of at least two adjacent vertebral bodies. Surgery and postoperative management was performed as described before [
14,
15]. Indication for thoracic CT scan was acute onset of one, or the combination of the following symptoms: collapse upon mobilisation, shock, hypotonia, tachycardia, dyspnoea, chest pain or dip in oxygen saturation [
16]. Patients with a pre-existing haematological disorder (e.g. factor V Leiden, prothrombin mutation, protein C/S deficiency, leukaemia, lymphoma, systemic amyloidosis) were excluded. Further exclusion criteria were treatment of patient for less than 48 h. For venous thromboembolism (VTE) prophylaxis, all patients received s.c. low molecular-weight heparin (LMWH) within 48 h of admission. In patients with elective surgical procedures, phenprocoumon therapy was paused at least 2 weeks prior surgery and switched to therapeutic LMWH which was paused on the day of surgery. The primary outcome was in hospital mortality. All patients received 40 mg of LMWH (Clexane
®) subcutaneously starting on the first postoperative day. On the day of operation, 20 mg of LMWH (Clexane
®) were administered s.c. 10 h post-surgery. This anticoagulation concept was recommended by our department of hemostaseology and applied in our department for over 20 years. Mobilisation was anticipated as soon as possible and realised with the assistance of a physiotherapist. All patients were urged to wear compression stockings. Patients were seen in a follow-up examination 3-months post-surgery in our out-patient department.
Computed tomography imaging
Thoracic CT scans were performed in the department of neuroradiology at a multidetector Philips CT Scanner. Ultravist® 300 was administered intravenously (80 ml/kg, 4.0 ml/s) and imaging started after the contrasting of the pulmonary artery.
Spinal MRI
All patients received spinal MRI in the department of neuroradiology, Goethe University Hospital Frankfurt at a 3 Tesla Siemens Verio scanner. Gd‐DO3A‐butrol (Gadovist®, Bayer Vital GmbH) was administered intravenously (0.1 ml/kg; max. 10 ml). T1w sagittal post contrast imaging started directly, axial T1w post contrast imaging was preformed between 3 and 7 min after administration of the contrast agent.
Statistics
Data analysis was performed with IBM SPSS Statistics Version 23.0 (SPSS Inc., IBM Corp., Armonk, NY, USA). For patient characteristics, descriptive statistics were used. Fisher’s exact test was used for the comparison of categorical variables between the cohorts. For continuous parameters, the Wilcoxon–Mann–Whitney test was used. To assess the impact of the variables, odds ratio (OR) with 95% confidence intervals (CI) was calculated. Results with p ≤ 0.05 were considered statistically relevant.
Discussion
To the best of our knowledge the present study is the first to describe the remarkably high incidence of PE in patients with pyogenic spondylodiscitis compared to published numbers of other surgical disciplines [
17‐
19]. Furthermore, we identified female sex, higher age, coronary heart disease, anticoagulation at admission and non-O blood type as significant risk factors for PE. PE was associated with DVT and resuscitation. Surprisingly, neither surgical intervention itself nor the chosen approach were associated with elevated PE risk compared to conservatively treated patients.
In our cohort, 12% of the included patients with pyogenic spondylodiscitis suffered from PE. This incidence appears to be significantly higher than described in the scientific literature. As for example studies on PE after orthopaedic surgeries, including a large series of spinal surgeries, vary from 2 to 6% [
16,
20,
21]. However, the published variety of PE manifestation among spinal surgery is extremely high as other studies report an incidence of up to 31% (i.e. patients with major spinal reconstruction) [
17,
18]. Our finding therefore amends to the published knowledge and raises the question of preventive scopes of action and anticoagulation management in patients with surgical interventions for pyogenic spondylodiscitis and those with conservative management.
The finding of significantly overrepresented female patients in the PE cohort is intriguing as data on whether patient sex is a risk factor for pulmonary embolism is conflicting. An analysis of national mortality data found that death rates from PE were up to 30% higher among men than among women [
21]. Although female sex was not found to be a significant risk factor in large unselected observational studies, female patients with PE had higher in-hospital mortality, a higher need for transfusion and occurrence of shock [
22].
Further, all patients in our PE cohort had non-O blood type. The difference in hemostaseological properties of ABO antigens led to the identification of non-O blood type as a significant risk factor for postoperative thrombosis and embolism [
23,
24]. Our finding is in line with this observation and demonstrates a significant association of patient blood type and for postoperative PE in patients with pyogenic spondylodiscitis.
Several publications claim that patients with high BMI are at elevated risk for PE [
19]. In specific, excessive visceral adipose tissue causes hypoxia and increases free fatty acids (FFA) to the liver where coagulation factors are synthesised [
25]. However, our study found no association between patients BMI and PE. A possible explanatory approach is the fact that in our analysis the investigated BMI was obtained at admission. Since pyogenic spondylodiscitis is a disease of mid- to long-term duration, patient BMI was subject to fluctuations after admission and thus not detected in our analysis.
In our cohort, the presence of anticoagulation at admission was associated with PE. This finding is challenging. All patients on anticoagulation were converted to LMWH prior to neurosurgery as described in the methods section. In most cases, the rational for anticoagulation was a positive history for DVT or PE. The underlying coagulation disorder that led to DVT or PE in the first place could therefore be accountable for PE in our cohort. Furthermore, a reactive hypercoagulability state or an increased hemostaseological response could also be made accountable for this association. The American College of Chest Physicians (ACCP) and the Eastern Association for the Surgery of Trauma (EAST) have recommended the use of low‐dose unfractionated heparin (UFH) or low molecular weight heparin (LMWH) with or without mechanical prophylaxis for the prevention of VTE complications but they are unclear on timing its initiation [
26,
27]. However, we here described our individual approach of administering 20 mg LMWH 10 h postoperatively and report a satisfactory result in postoperative management since our surgically treated patients did not show an increased PE development compared to the conservatively treated.
We found coronary heart disease to be significantly associated with PE in our cohort. Although, we are not aware of other studies that explored the association between pyogenic spondylodiscitis and PE, the increased risk of PE after the diagnosis of coronary heart diseases is known in literature and explained by the fact that coronary heart diseases induce venous stasis and elevated systemic venous pressure [
28,
29]. In our cohort, this phenomenon could be aggravated by the present bacteremia, with often infective endocarditis and consecutive further deterioration [
30].
In our cohort, patients with PE had a significant association with DVT and resuscitation. The complex situation of a hazardous cardiovascular event with often inevitable long-term anticoagulation after a recent neurosurgical procedure is a known risk factor for poor outcome and coherent with the scientific literature [
20,
23].
Furthermore, we found surgical intervention for pyogenic spondylodiscitis to be a non-significant risk factor for PE when compared to a conservative treated cohort. This finding is surprising since major surgery is a well-established risk factor for PE development in the scientific literature [
31‐
33]. The complexity and long duration of both conditions could be accountable for this finding. Hypothetically, surgery could be accountable for a periodic hypercoagulability state with increased thrombotic potential but the benefit of early mobilisation could counterbalance this finding when compared to prolonged bed rest in conservative managed patients.
The obvious limitation is that this investigation was a single centre study and of retrospective design. As this study is of observational character, confounding, selection bias, reverse causation and uncontrolled statistical error risk cannot be excluded. Further, the small sample size of our PE cohort is a possible confounder which should be addressed in future prospective multicentre studies. Despite the relative high incidence of PE in our surgical cohort, when compared to conservative treated patients the sequela is non-hazardous. Only one patient died in the PE cohort. In contrast, data from patients with spinal cord injury display a DVT incidence of almost 100% [
34,
35]. The benefits of surgical treatment in patients with pyogenic spondylodiscitis (bacterial detection, reduced bed rest and early ambulation) outweighs the incidence of PE. Because of the small sample a size specific interaction between the variables measured by a multivariate analysis was not possible. Ultimately, a randomised clinical trial is necessary to verify our findings.