Acute kidney injury risk in orthopaedic trauma patients pre and post surgery using a biomarker algorithm and clinical risk score

Acute kidney injury (AKI) after major trauma is associated with increased mortality. The aim of this study was to assess if measurement of blood biomarkers in combination with clinical characteristics could be used to develop a tool to assist clinicians in identifying which orthopaedic trauma patients are at risk of AKI. This is a prospective study of 237 orthopaedic trauma patients who were consecutively scheduled for open reduction and internal fixation of their fracture between May 2012 and August 2013. Clinical characteristics were recorded, and 28 biomarkers were analysed in patient blood samples. Post operatively a combination of H-FABP, sTNFR1 and MK had the highest predictive ability to identify patients at risk of developing AKI (AUROC 0.885). Three clinical characteristics; age, dementia and hypertension were identified in the orthopaedic trauma patients as potential risks for the development of AKI. Combining biomarker data with clinical characteristics allowed us to develop a proactive AKI clinical tool, which grouped patients into four risk categories that were associated with a clinical management regime that impacted patient care, management, length of hospital stay, and efficient use of hospital resources.


Scientific Reports
| (2020) 10:20005 | https://doi.org/10.1038/s41598-020-76929-y www.nature.com/scientificreports/ biomarker combinations together with clinical risk factors have not been identified for orthopaedic trauma patients. Therefore, we considered (1) could measurement of blood biomarkers pre and post surgery be used to stratify risk of AKI in orthopaedic trauma patients? and (2) could biomarker data combined with clinical characteristics be used to develop a tool to assist clinicians in identifying orthopaedic trauma patients at risk of AKI and guide patient management?

Methods
Study population. This prospective study of 237 patients was performed within the Fracture Unit of the Royal Victoria Hospital, Belfast, UK between May 2012 and August 2013. The study complied with the Declaration of Helsinki, was approved by the Office for Research Ethics Committee Northern Ireland, the Royal Victoria Hospital Research Office Research Governance Committee and written informed consent was obtained from all participating patients. Orthopaedic trauma patients who were consecutively scheduled for open reduction and internal fixation (ORIF) of their fracture, were recruited into the study. Patients were excluded if they were < 18 years of age, had preoperative or pre-trauma dialysis-dependent renal failure or had a history of significant renal disease prior to recruitment. Of the n = 237 patients recruited to the study, pre and post operative samples were available for 201/237 (84.8%) patients. Patient samples were not available for 36/237 (15.2%) and these patients were excluded from the study (Fig. 1).
Clinical data collection. Clinical data was recorded for each patient from medical records that included baseline demographic characteristics, comorbidities and current medications.
Sampling and laboratory methods. Patient blood samples (10 ml) were collected preoperatively and on day 1 post operatively. Patient blood samples were centrifuged, and serum and plasma were aliquoted within 30 min of collection and stored at − 80 °C.
Outcome definition. Patients did not have a baseline eGFR measurement prior to trauma but were assumed to have a normal renal function with a baseline eGFR of at least 60 ml/min/1.73m 218-20 . A value of < 45 ml/ min/1.73m 2 was used to define a patient as AKI positive on any of the recorded pre and post operative sampling days, in accordance with the RIFLE classification 21 ; any patient with an eGFR result at any time (day 0, 1, 2, and 5) > 25% of 60 ml/min/1.73m 2 (45 ml/min/1.73m 2 ) were determined to have AKI. Statistical analysis. Statistical analyses were performed using R 22 . Wilcoxon rank sum test was used to identify differentially expressed biomarkers. Biomarkers with a p < 0.05 were considered significant. The ability of the biomarkers to predict AKI was further investigated using logistic regression (Lasso regression). For each biomarker and biomarker combination, areas under the receiver operator characteristic (AUROC), sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were generated pre and post operatively to identify models that differentiated between the two diagnostic groups (non AKI vs. AKI).

Results
Clinical characteristics for patients involved in the study are presented in Table 1. Of the 28 biomarkers that were investigated, sTNFR1 and H-FABP had the highest AUROC pre surgery to stratify risk of AKI in orthopaedic  Fig. 2A,B)). LASSO regression identified a combination of 3 biomarkers post operatively to stratify risk of AKI, namely H-FABP, sTNFR1 and MK (Table 2) (H-FABP, sTNFR1 and MK combined sensitivity 80.5%; specificity 86.0%; AUROC 0.885 (CI 0.825-0.944) (Fig. 3A,B)).
To develop a tool to assist clinicians in identifying orthopaedic trauma patients at risk of AKI and guide patient management, the clinical characteristics between non AKI and AKI patients were investigated. We used biomarker combinations to determine a biomarker risk score (BRS) pre and post surgery (based on AUROC determined by LASSO regression) that could be used to identify patients at risk of AKI. Establishing a biomarker set point (cut-off) pre and post surgery categorised patients either positive or negative for AKI i.e. if a post surgery patient had a BRS above the set point they would be predicted to be positive for AKI (Table 3).
Three clinical characteristics were identified for patients at potential risk of AKI pre and post operatively ( Table 4). Each clinical characteristic was given a score 0 or 1 (0 = no risk, 1 = risk). Each clinical characteristic was then added to give a cumulative risk score (CRS). For example, pre surgery patients who score ≥ 1 e.g. an ≥ 80-year-old patient with dementia and hypertension would have a cumulative CRS of 3 and would therefore be categorized high risk for AKI. The cut-off for age was based on significance, where patients > 80 years were at greater risk of AKI, based on our patient cohort.
To translate the results of the BRS and CRS into a proactive clinical AKI tool, the BRS and CRS were combined. Combining BRS with CRS either pre or post surgery identified 4 risk categories for patient management (Table 5). Categories 1 and 2 = low risk; Categories 3 and 4 = high risk. Two worked examples for a non AKI and AKI patient, are shown in the Supplementary Notes S1-S3 and Supplementary Tables S1-S6. The distribution of non AKI and AKI within the patient cohort, for each risk category, is shown in Supplementary Note S4.

Discussion
Hip fracture is the most common serious injury reported in the elderly resulting in long hospital stays, high post-operative morbidity and mortality, and reduced quality of life 4 . Furthermore, AKI after trauma, such as hip fracture is associated with a poor prognosis.
Diagnosis of AKI using sCr and urine output can often result in misdiagnosis. The aim of this study was to further investigate if blood biomarkers and clinical risk factors could be used to identify AKI risk in orthopaedic trauma patients pre and post ORIF surgery in a similar fashion to those identified in patients undergoing cardiac surgery 11 . Interestingly, the same blood biomarkers, H-FABP, Midkine, sTNFR1 or sTNFR2, that predicted AKI  A total of 28 blood biomarkers were investigated. However only two biomarkers, sTNFR1 or H-FABP, were identified as predictive of AKI pre surgery and a combination of three biomarkers, sTNFR1, H-FABP and MK, were predictive for AKI post surgery. Remarkably these biomarkers represent three main pathological processes of AKI. Mechanisms contributing to AKI include (1) perioperative episodes of under perfusion, followed by (2) ischemia reperfusion injury during restoration of normal blood pressure. Inflammatory mediators (3) contribute to and augment the renal injurious effects of this twofold process. Accordingly, an additional separate inflammatory insult arising from other perioperative factors such as coagulation disturbance (which is an important proinflammatory mechanism), can augment the renal injurious effect of hypotension and ischemia reperfusion. Biomarkers have been associated with identification of underlying processes of hypotension (VEFG and H-FABP), IRI (MK) and inflammation (sTNFR1 and 2), which as anti-inflammatory biomarkers are taken as surrogates for the underlying proinflammatory response which drives them 11 . Elevated sTNFR1 levels have been identified in many clinical conditions e.g. kidney disease 23 , neuropathy, cardiovascular disease and diabetes 24 , and circulating levels of sTNFR1 have been shown to be an independent predictor of CKD progression in elderly patients 25 . Tumour necrosis factor alpha (TNFα) and TNFR2 are almost undetectable in the kidneys of healthy subjects unlike TNFR1 which is expressed within the trans-golgi network of the glomerular endothelium 26 . An increase in the level of sTNFRs in CKD patients has been implicated in declining eGFR [27][28][29] . Moreover, TNFα acting through TNFR1 has a damaging effect on renal endothelial cells 30 , possibly through iNOS, which would generate intratubular toxic levels of NO, as demonstrated by increased urinary nitrate levels in a porcine model of ischaemia reperfusion-mediated AKI 31 . The elevated anti-inflammatory sTNFR1 response in blood may be driven by an underlying proinflammatory response which includes TNFα 31 . Since monomeric TNFα is much smaller than sTNFR1 and 2, it is more readily filtered by the glomerulus. Accordingly, TNFα is able to cause glomerular injury once it escapes from the moderating biological effect of sTNFR1 or 2. This is consistent with orthopaedic trauma patients who develop AKI having elevated levels of sTNFR1 when compared to non AKI patients.
H-FABP was also predictive of AKI pre operatively, and in combination with sTNFR1 and MK, post operatively. H-FABP, associated with cardiac injury, is released into the bloodstream 30 min after an ischaemic event and peaks at 6 h before returning to normal levels after 24 h 32 . H-FABP has been reported to predict AKI pre and post cardiac surgery 7,11,33 however, this is the first time that H-FABP has been demonstrated to predict AKI in patients pre and post ORIF surgery.
H-FABP is predominantly expressed in the heart but also at lower levels in skeletal muscle, kidney, stomach, brain and testis 34,35 . The levels of H-FABP in skeletal muscle have been shown to be almost half that found in the heart. Moreover, kidney H-FABP levels are almost two-thirds that found in skeletal muscle 36 . While it is known that H-FABP levels increase in the blood, this may arise from skeletal muscle or renal sources, but it is more likely to be from the heart, which is the largest reservoir of H-FABP in the body. In elderly patients, acute coronary insufficiency is common and would be reflected in elevated H-FABP. Any transient hypoperfusion, which such an event would provoke, could result in a significantly heightened risk of AKI. This is the most likely reason why H-FABP was predictive in this orthopaedic trauma patient cohort.
In addition to sTNFR1 and H-FABP, MK was also identified in the biomarker combination to predict AKI post operatively. The pathophysiological roles of MK are diverse, ranging from AKI to progression of CKD, accompanied by hypertension, renal ischaemia and diabetic nephropathy 37,38 . After ischaemic reperfusion MK is immediately induced in the proximal tubules, leading to the upregulation of macrophage inflammatory protein-2 for neutrophils and monocyte chemotactic protein-1 for macrophages 38 . Eventually, infiltrated inflammatory cells cause severe tubulointerstitial injury. Silencing renal MK expression with anti-sense oligos prevents kidney damage and increases osteogenic activity 39 . Midkine is also involved in chondrogenesis and fracture healing 39 . Interestingly, MK-deficient mice have been shown to display increased bone formation rate and volume 39 . This is the first study, to our knowledge which has identified MK as a biomarker for stratifying patients at risk of AKI following orthopaedic trauma and ORIF surgery.
Risk factors that have previously been reported for AKI include age, pre-existing CKD, male gender, diabetes, heart failure and surgery [15][16][17] . In this study, three clinical risk factors were identified for patients who were at potential risk for the development of AKI, pre and post operatively; age, dementia and hypertension (Table 4). Using biomarker data and clinical factors we developed a BRS and a CRS, respectively. Combining BRS (Table 3) Table 5. Clinical management of patients using a combination of BRS and CRS either pre or post surgery. Combining BRS and CRS assigns a patient to a risk category. BRS biomarker risk score, CRS clinical risk score.  (Table 4) grouped patients into four risk categories, each of which is associated with a clinical management regime ( Table 5). Deployment of this proactive clinical AKI tool would allow clinicians to stratify patients at risk of AKI enabling early intervention and improving patient outcomes. Use of a cardiac proactive clinical AKI tool has been described previously 11 . In the study cohort 63/201 (31.3%) orthopaedic trauma patients developed AKI post ORIF surgery. The incidence of AKI in this patient cohort is higher than previously reported [40][41][42][43] . Patients who developed AKI were significantly older and more likely to have hypertension and/or dementia. Advanced age is frequently reported as a risk factor for AKI 1 however, to our knowledge, this is the first report that identified dementia as a potential risk factor in the development of AKI. Recently, an association of heightened proinflammatory activity in patients with dementia has been reported 44 . Our results are consistent with these findings. Interestingly, patients surviving AKI have a higher probability of developing dementia in the long-term compared to patients who did not develop AKI 45 .
The time between presentation and surgery was 2 days and was not significant between non AKI and AKI patients. An optimal operation time of between 24-48 h after orthopaedic trauma has been identified for lower extremity fracture fixation to reduce complications. Operations performed outside of this timeframe are associated with increased morbidity and mortality 46 .
AKI patients stayed an additional two days in hospital compared to non AKI patients (12.0 (3.7-20.3) days for AKI vs. 9.8 (1.9-17.7) days for non AKI patients), consistent with previous findings 47 . Patients that develop AKI following elective total joint arthroplasty also have increased hospital stay 48 . The management of patients with AKI is a significant burden to the healthcare service 49 . Earlier diagnosis and management of patients at risk of AKI will potentially reduce the financial burden on healthcare systems in addition to improving patient outcomes and welfare.
Surprisingly previous work has failed to identify hypotension as a serious risk factor in AKI 50 . Since intraoperative blood pressure modulation is a readily available strategy for anaestheologists, inability to show a link between hypotension and AKI at orthopaedic surgery could arguably generate a false sense of complacency. However, the reason for lack of the relationship between perioperative blood pressure and subsequent AKI could be because most patients have non-invasive blood pressure measurements where blood pressure readings are obtained by an arm cuff measurement every 5 min whereas more critically ill patients have continuous arterial blood pressure measurements which detect and record all hypotensive episodes. In Braüner's study 50 they recorded lowest blood pressure measurement intraoperatively. Their work suggested that this was not a useful marker in terms of AKI prediction. However, we argue that clinically significant hypotensive episodes may have been missed in this study if they happened in between measurements. This means that transient, albeit clinically significant, hypotension could be missed in between these times. In summary, the use of clinical data alone (including perioperative hypotensive events) to predict perioperative AKI is of limited usefulness in hip fracture surgery. It has already been shown in cardiac surgery that biomarkers of ischaemia reperfusion (MK) or hypotension (VEGF or HFABP) and inflammation augmented clinical parameters 11 . This present work suggests that this principle is also applicable to hip fracture surgery.  11 . AKI acute kidney injury, BP blood pressure, H-FABP heart-type fatty acid-binding protein, MK midkine, sTNFR soluble tumour necrosis factor receptor, TNFα tumour necrosis factor alpha.

Scientific Reports
| (2020) 10:20005 | https://doi.org/10.1038/s41598-020-76929-y www.nature.com/scientificreports/ In a meta-analysis it was demonstrated that perioperative hemodynamic optimization in surgery patients, reduces post-operative acute renal injury 51 . Preoperative prediction would allow for enhanced perioperative hemodynamic optimization i.e. provision of Level 2 care provided post operatively, and invasive hemodynamic monitoring intraoperatively rather than blood pressure measurements every 5 min, as is routine for such cases in many centres. It could also be taken as a contraindication to non steroidal anti-inflammatory use post operation.
Biomarkers are not a substitute to the classical approach to using low-cost information-but add to the information available to the clinician. However, it must be noted that a clear clinical history in these elderly patients can sometimes be unreliable. Hence the need for the objective information that biomarkers provide.
Limitations of the study. Clinical characteristics were not reliably available for everyone in this patient group including a guaranteed history of normal renal function pre trauma; patients were assumed, based on available clinical history, to have a normal renal function prior to their trauma and a baseline eGFR of at least 60 ml/min/1.73m 2 . Therefore, patients who had undiagnosed pre-trauma renal dysfunction could have been included in the study. Nevertheless, subsequent fluctuations in renal function were still detectable using our proactive clinical AKI tool, demonstrating the clinical utility of our proposed method in this patient cohort, where obtaining clinical history is sometimes challenging and unreliable.

Conclusion
In conclusion, serum H-FABP and sTNFR1 measured pre operatively and serum H-FABP, MK and sTNFR1 measured post operatively, identified orthopaedic trauma patients at risk of developing AKI during ORIF surgery. Utilisation of the proactive clinical AKI tool, which combines BRS with CRS, would allow clinicians to stratify patients into one of four AKI risk categories with related treatment regimens that could impact patient care and management, length of hospital stay, and the efficient use of hospital resources.

Data availability
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
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/.