Correlation between kidney function and mortality in pyogenic spondylodiscitis: the glomerular filtration rate (GFR) as new predictive parameter?

Pyogenic spondylodiscitis is a severe medical condition, often requiring surgical intervention. Numerous risk factors are known, such as obesity, neurological impairment and old age. In-hospital mortality remains high, therefore other factors may be contributing to the increased mortality. To evaluate kidney function as a risk factor for increased morbidity of pyogenic spondylodiscitis, the glomerular filtration rate (GFR) was correlated with the patients' clinical course. We retrospectively reviewed the cases of 366 patients and 255 were included for analysis. Clinical, laboratory and surgical data were recorded with a minimum follow-up of three months. For clinical outcome measurement, mortality, length of stay and perioperative complications were analysed. The study included 255 patients (173 men, 82 women; mean age 66.3 years). Patients with a GFR < 59 mL/min spent an average of 5 days longer in the hospital than those with a GFR ≥ 60 mL/min (p = 0.071). The mortality rate increased significantly with a decrease in GFR: A GFR of 30–59 mL/min had a mortality rate of 17.6%, whereas a GFR of < 29 mL/min had one of 30.4% (p = 0.003). Patients with impaired GFR showed an increased rate of postoperative complications (OR 4.7 p = 0.002) and higher rate of intensive care unit (ICU) stay (OR 8.7 p =  < 0.001). Preoperative GFR values showed a significant correlation with in-hospital mortality in patients with spondylodiscitis, when graded according to the KDIGO stages. Furthermore, a GFR of < 29 ml/mL contributes to a longer ICU stay, postoperative complications and a longer total hospital stay. Therefore, the preoperative GFR could be a marker of kidney function and as a valuable predictive risk factor regarding the clinical in-hospital course of patients suffering from pyogenic spondylodiscitis.


Introduction
Pyogenic spondylodiscitis is a challenging disease in spinal surgery, as the all-cause mortality rate is reported to be 5-20%, which is remarkably higher than the general mortality rate after spine surgery, which is reported to be 0.3-1.4% [1]. With an incidence rate of 0.2-2.4/100.000 per year, spondylodiscitis is the most common spinal infection, and the incidence rate is double for male patients [2,3]. Incidence levels are rising mainly due to the ageing population, higher numbers of immunosuppressed patients and, in particular, improved diagnostic capabilities [4,5].
Typically, bacteraemia, caused by a distant infectious process results in variable clinical presentations, and can cause spinal infections. Therefore, delays in diagnosis are 1 3 common. The treatment of pyogenic spondylodiscitis is challenging, as conservative treatment with long-term antibiotics usually require verification of the pathogen in blood culture or intervention sample taking. However, as the disc lacks significant vascularisation, the optimal duration of intravenous and oral antibiotic treatment is an ongoing debate. Moreover, numerous cases require surgical intervention, mainly indicated by neurological deficits, mechanical instability and spinal deformity [6]. Even with accurate therapy, all-cause mortality and morbidity remain high due to perioperative complications caused by either the infection or the intensive treatment [7][8][9].
According to previously published studies, several risk factors for a severe case of spondylodiscitis are known, such as obesity (using body mass index-BMI), diabetes, age and epidural empyema [10][11][12]. Little is known about the role of renal function in spondylodiscitis. In a study suggesting an important role of kidney function, Chikuda et al. reported that dialysis in patients with spondylarthropathy significantly increased perioperative complications and mortality [13]. Moreover, chronic kidney diseases (CKD), which are increasing in incidence, were observed as significant risk factors for higher mortality and morbidity, especially in endstage renal disease (ESRD) [14,15]. Next to CKD, which is rising in prevalence in an elderly society, acute kidney injury defined by KIDGO as a rise of > 0.5 mg/dL of creatinine levels or limited urine output may contribute to a worse outcome in critically ill patients [16]. The severity of AKI even was significantly associated with higher mortality [17]. While AKI can have numerous reasons such as ischemia, direct nephrotoxic injury or inflammatory insults, sepsis due to severe infection is commonly seen [18]. Sepsis is the leading course of AKI in ICU patients and AKI is associated with 45-70% of all cases with sepsis [19,20]. A possible reason for the elevated mortality could be distant organ injury induced by AKI, as cardiac-insufficiency, encephalic disorders and respiratory failure is seen regularly [21,22]. The systemic inflammation in sepsis determines the clinical picture [23]. In the literature, the clinical presentation of spondylodiscitis varies from rare asymptomatic cases to patients in septic-shock requiring life-saving surgical intervention [8,11].
Therefore, a key question in pyogenic spondylodiscitis research is the influence of AKI due to the infection and septic constellation. The glomerular filtration rate (GFR) is a parameter for kidney function. The clinical GFR value is determined by different calculations based on creatinine levels, ethnicity or age, and is measured in mL/min/1.73m 2 body surface area. Furthermore, the GFR levels are age dependent, with levels decreasing with ageing. According to the "Kidney Disease-Improving Global Outcomes" (KDIGO) guidelines, GFR values allow kidney function to be classified into several grades, ranging from G1 to G5. Grades G1 and G2 show GFR levels > 60 mL/ min/1.73m 2 and concomitant proteinuria (group 1 in our present study). Grade G3 (30-59 mL/min), representing moderate impairment, is subdivided into grades G3a and G3b. For severe disease, grade G4 represents GFR values of 15-29 mL/min/1.73m 2 and grade G5 represents GFR values of < 15 mL/min/1.73m 2 .
Therefore, the underlying aim of this study was to validate the preoperative GFR as a marker of kidney function and as a valuable predictive risk factor regarding the clinical in-hospital course of patients suffering from spondylodiscitis. Factors including mortality, revision surgery and postoperative complications, as well as length of stay in the ICU and hospital, were studied.

Study design
We performed a monocentre retrospective study, including patients undertaking surgical treatment of spondylodiscitis in a level I spine centre, which were investigated with respect to potential prognostic criteria. Any surgery including fusion surgery with either dorsal, ventral or combined dorsal-ventral fusion surgery, was included. The criteria for exclusion were any form of malignant disease that would modify the laboratory results, especially leukocytes and high CRP levels, as well as tuberculous spondylodiscitis. In addition, 19 patients were excluded as insufficient data were available, resulting in 255 eligible patients being included in analysis.

Data collection
Data were extracted from the hospital information system (HIS), such as gender and age, as well as admission, length of stay and peri-or postoperative complications. Mortality, revision surgery as well as pathogen levels were documented. Regarding laboratory results, pre-and postoperative GFR, creatinine and CRP levels were recorded over a two-week period. Patients were assessed until six weeks postoperative, although some patients presented symptoms later on and therefore had a longer follow-up. The BMI values, the perioperative antibiotics used and the American Society of Anesthesiologists (ASA) scoring were obtained from the anaesthesiological documentation. Surgical data, such as operation time, type of fusion surgery, extent of levels included in surgery and spinal location, were recorded. Microbiological testing from blood culture and intraoperative samples were recorded.

Statistics
Statistical analysis was performed using SPSS V.27 software (SPSS, Chicago, Illinois). We used descriptive statistics to summarise the means and standard deviations. A Pearson test was performed for detecting significant correlations. Two-sided t-tests were performed to detect any statistically significant differences. Binary logistic regression analysis was performed to analyse risk and likelihood (odds ratio). The level of significance was defined as a p-value of p = < 0.05.

Ethics
The analysis was revised by our institutional ethics committee and was approved by the Institutional Review Board of the Faculty of Medicine (approval no. 20-1631).

Results
A total of 255 patients were included in the study (173 men, 82 women). The mean age was 66.3 years (± 13.3). The mean BMI level was 23.5 ± 5. An overall in-hospital mortality of 10.2% was observed with a mean of 32.2 days of hospital stay. 85.1% of all spondylodiscitis was found in lumbosacral segments. Surgery included dorsal fusion surgery in 82% of all cases, with another 35% undergoing an additional ventral procedure for spinal stabilisation. Cervical spondylodiscitis was predominantly treated by an anterior approach (88% vs 12% posterior).
Regarding preexisting comorbidities, only 15.9% had no previous medical condition. The most common were cardiovascular diseases (hypertension, cardiac arrhythmia and heart valve defects) in 39.2%, of patients followed by diabetes in 23.5% and endocrinological diseases in 17.3% of patients. A preexisting chronic kidney disease was recorded in 14.1% of patients, and a chronic liver disease in 6.3% of all patients. Regarding further risk factors, obesity (BMI > 30) was observed in 18.8% of patients and another 18% were actively smoking. The main pathogen found was Staphylococcus aureus (25%), followed by other Staphylococcus species (15.3%) and Enterobacteriacae (10.2%). In 25% of all cases, no pathogen was detected despite blood culture and intraoperative probe analysis.
Based on the KDIGO guidelines, the cohort was subdivided into three groups: Group 1 with GFR values > 60 mL/min, group 2 with GFR values of 30-59 mL/ min and group 3 with GFR < 29 mL/min.
The cohorts' group characteristics are described in detail in Table 1. As expected, GFR decreases as age increases and the mean age of group 1 was 64 (± 13.1), the mean age of group 2 was 72.1 (± 10.4) and mean age of group 3 was 74 (± 10.9).
The BMI levels are nearly equally distributed among the three categories, as well as the male/female ratio. However, the American Society of Anesthesiologists (ASA) classification score, which have been an established score for measuring perioperative risk since 1941 and regularly updated, do not seem to reflect the deteriorating kidney function, as measured by GFR. Moreover, a documented kidney impairment, chronic or acute, was only present in 14.1% of the cases, while the others presented acutely elevated retention parameters and decreased GFR levels.
The mean length of total hospital stay was 31.4 days for group 1, 32.7 for group 2 and 43.8 for group 3, and showed significant correlation with the GFR values (p = 0.012) (Fig. 2). Group 1 had a mean stay of 10 days, group 2 had 7.6 days and group 3 had 12.5 days, respectively (Fig. 2). The total length of stay in hospital showed no significant correlation to the GFR levels (p = 0.22).

Discussion
We have provided a detailed analysis of relationship between complications arising in pyogenic spondylodiscitis vs, kidney function, as measured by GFR levels. The results demonstrate an increased complication rate in patients with lower GFR levels. The complications include higher in-hospital all-cause mortality rates, longer hospital stays and higher postoperative complication rates.  Compared to previously published studies, which are in accordance with our findings, we provide a larger cohort of surgically treated patients. A high mortality rate of spondylodiscitis of up to 20% is previously reported, while a mean mortality rate of 10% was observed for our cohort, which is comparable to previous data [24,25]. The high rate of in-hospital mortality is often described as multi-organ failure, owing to septic shock and progressive circulatory instability, with the need for increased support. Supporting these data, S. aureus is seen as the most deadly pathogen detected in spondylodiscitis.
However, the high mortality rate may instead be linked to the handful of risk factors known so far. Detailed knowledge of these could potentially decrease mortality. Detecting more significant risk factors and predictive factors in the outcome in pyogenic spondylodiscitis would help to assign patients to the most appropriate treatment option. Moreover, a preoperative improvement of kidney function by electrolyte control, the adaption of accurate volume status and the discontinuation of all nephrotoxic medication might be helpful. Consultation of nephrologists in clinical settings might improve the outcome, as seen in antibiotic stewardship by infectiologists. However, this hypothesis must be analysed in randomised controlled clinical trials and cannot be answered by these data.
Analysing the role of kidney function in pyogenic spondylodiscitis, our results showed significant predictive factors in outcome and mortality. In our case series, only 14.2% of all patients had recorded a known kidney disease. We see this as the main reason to believe that the acute kidney injury resulting in lower GFR levels is related to the infectious spondylodiscitis. Most of the patients are presented to our hospital only after a medical consultation either by their general-practitioner or another hospital. Therefore, we assume fundamentally elaborated medical findings and a comprehensible medical history. Hence, only 14.2% had a previously diagnosed CKD, while 29.4% had compromised GFR levels of < 60 mL/min. This fact corroborates the association to an acute infectious process resulting in sepsis and kidney dysfunction.
However, GFR levels of < 60 ml/min result in a higher rate of complications and mortality. The findings of this study concur with previously published data regarding mortality and complications in patients with spondylodiscitis, the need for haemodialysis, or diabetes [13]. In a wide multicentre analysis by Kushioka, dialysis-related spondyloarthropathy and infectious spondylodiscitis showed highest mortality rates [26]. Here, diabetic nephropathy could be the main impact of the disease, which is associated with higher mortality and morbidity. A compromised kidney function in ESRD has already been shown to increase mortality in any lumbar spinal surgery [14,15].
While a trias of inflammation, direct toxic injury of the kidney and ischemia are causes of AKI, the infection and sepsis caused by severe pyogenic spondylodiscitis which is often caused by bacteriaemia, is thought to result in AKI [18]. In a recent systematic review, Lee et al. stated distant organ dysfunction caused by AKI including lung, heart, brain and even liver [22]. These organs seem to be harmed by several mechanisms, including the accumulation of uremic toxins accumulation and change in circulation including ischemia leading to oxidative stress, inflammation, electrolyte imbalance as well as acid/base imbalances [22]. Molecular mechanisms found so far, include inflammatory mediators such as Interleukin-6 (IL-6), plasminogen activator inhibitor-1 and soluble tumour necrosis factor receptors (TNF-alpha) [27].
The current epidemiologic outline the increased mortality and morbidity in patients with acute renal failure [28]. This poses a heavy burden with high costs for the healthcare Fig. 2 Boxplot of Length of stay ICU (in days) of the three groups system due to intensive-care treatment [29]. Understanding the complexity of interactions between the kidney and distant organs could lead to new diagnostics and therapies to improve the outcomes in patients with AKI. This could lead to improved therapy and management of sepsis in case of severe spondylodiscitis, thus resulting in lower mortality rates.
Of course, the GFR level itself does not represent the complete risk-stratification in patients, and additional parameters, such as retention parameters, or albuminuria and hypoalbuminemia, can play an important role. A new scoring system that implements the KDIGO guidelines, as well as including other risk factors for spondylodiscitis, may help to make targeted treatment decisions. Not every patient receives full urine diagnostics on admission, making GFR, as well as other retention parameters, an easy marker to determine for risk stratification. This is particularly important as the GFR can be seen as parameter for the multimorbidity of patients. Kidney function is lowered by numerous influences resulting from several diseases, with cardiac, pulmonary, endocrinological and malignant diseases all impairing the function of the kidneys [30]. This is also elucidated by our results, as the GFR was more accurate in predicting complications and mortality than the ASA score.
The role of kidney function and GFR might become more evident in further clinical trials, which might alter the preoperative decisions for these patients. A higher tendency for conservative therapy with intravenous antibiotics, or minimally invasive sample taking by radiologists using CT-controlled puncture, might minimise morbidity and mortality. For accurate and successful antibiotic treatment, it is extremely important to identify the pathogens before starting antibiotics among patients with spondylodiscitis as the clinical presentation can differ. Empiric antibiotic treatment can even lead to a delay in diagnosis of spondylodiscitis causing further progress of the infection [31]. Lora-Tamayo et al. showed a significant delay in patients without positive microbiological cultures [32].
The aim is to generate a new severity scoring in pyogenic spondylodiscitis, to perform a more accurate risk stratification for patients who need surgical treatment, and to minimise complications and mortality. These factors known so far include age, obesity (BMI), diabetes, renal dysfunction (GFR), the pathogen detected and any preoperative antibiotic treatment, as well as neurologic impairment [33]. Based on an automatic intelligence (AI) algorithm, an individual risk profile can be established, eventually determining the adequate treatment option in the future.

Limitations
Limitations of this study are its retrospective, non-randomised nature; the inhomogeneous patient population; and the limited follow-up after the hospital course (minimum 6 weeks maximum 12 weeks postoperative). Furthermore, the GFR grade alone is usually not used as stand-alone parameter to evaluate kidney function; other parameters, such as creatinine, albuminuria and serum albumin levels are also considered. The modification in the KDIGO grading system was used to gain higher statistical power and to facilitate application in standard clinical evaluation. Nevertheless, in our opinion, the results from the in-patient course, in addition to the epidemiological data obtained, provide interesting findings that should be further investigated, especially the GFR value as predictive factor in pyogenic spondylodiscitis. Compared to other studies, we provide a large cohort of analysed cases. Consideration of the comorbidities with regard to the chronic and acute courses is desirable for further studies, as well as a longer follow-up of at least 12 months. Ideally, the studies are in the sense of a prospectively designed multicentre study. Furthermore, a multicentre prospective analysis could elucidate therapeutic consequences in pyogenic spondylodiscitis.

Conclusion
In conclusion, our study provides significant evidence that kidney function measured by GFR levels can be a significant predictive factor for estimating the length of hospital stays, mortality and the risk of complications throughout the course. Other risk factors, such as age, BMI, diabetes, pathogen, neurologic deficits and now GFR, should be used to create a new scoring system for the severity of spondylodiscitis, which considers the mortality and morbidity even after radical surgical procedures.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.