(TIMP2) x (IGFBP7) as early renal biomarker for the prediction of acute kidney injury in aortic surgery (TIGER). A single center observational study

Objective Postoperative acute kidney injury (po-AKI) is frequently observed after major vascular surgery and impacts on mortality rates. Early identification of po-AKI patients using the novel urinary biomarkers insulin-like growth factor-binding-protein 7 (IGFBP7) and tissue inhibitor of metalloproteinases-2 (TIMP-2) might help in early identification of individuals at risk of AKI and enable timely introduction of preventative or therapeutic interventions with the aim of reducing the incidence of po-AKI. We investigated whether biomarker-based monitoring would allow for early detection of po-AKI in patients undergoing abdominal aortic interventions. Methods In an investigator-initiated prospective single-center observational study in a tertiary care academic center, adult patients with emergency/ elective abdominal aortic repair were included. Patients were tested for concentrations of urinary (TIMP-2) x (IGFBP7) at baseline, after surgical interventions (PO), and in the mornings of the first postoperative day (POD1). The primary endpoint was a difference in urinary (TIMP-2) x (IGFBP7) levels at POD1 in patients with/ without po-AKI (all KDIGO stages, po-AKI until seven days after surgery). Secondary endpoints included sensitivity/ specificity analyses of previously proposed cut-off levels and clinical outcome measures (e.g. need for renal replacement therapy). Results 93 patients (n = 71 open surgery) were included. Po-AKI was observed in 33% (31/93) of patients. Urinary (TIMP-2) x (IGFBP7) levels at POD1 did not differ between patients with/ without AKI (median 0.39, interquartile range [IQR] 0.13–1.05 and median 0.23, IQR 0.14–0.53, p = .11, respectively) and PO (median 0.2, IQR 0.08–0.42, 0.18, IQR 0.09–0.46; p = .79). Higher median (TIMP-2) x (IGFBP7) levels were noted in KDIGO stage 3 pAKI patients at POD1 (3.75, IQR 1.97–6.92; p = .003). Previously proposed cutoff levels (0.3, 2) showed moderate sensitivity/ specificity (0.58/0.58 and 0.16/0.98, respectively). Conclusion In a prospective monocentric observational study in patients after abdominal aortic repair, early assessment of urinary (TIMP-2) x (IGFBP7) did not appear to have adequate sensitivity/ specificity to identify patients that later developed postoperative AKI. Clinicaltrials.gov NCT03469765, registered March 19, 2018.

practice as e.g. specificity of investigated patient cohorts, potential co-medications, and other factors could affect sensitivity and specificity [27,28].
Currently, data on biomarkers for timely prediction of postoperative renal outcomes are not available for open abdominal aortic repair and EVAR patients. We therefore performed an investigator-initiated prospective observational study to investigate whether early assessment of the biomarker (TIMP-2) x (IGFBP7) would predict po-AKI in patients with abdominal aortic repair.

Study characteristics and study patients
The (Timp-2) x (IGfbp7) as early renal biomarker for the prediction of acute kidney injury in aortic surgery (TIGER) study was an investigator-initiated, single center, prospective, observational study. Patients with emergency or elective abdominal aortic surgery (open repair with infra-or suprarenal clamping, or EVAR) referred to our university-based cardiovascular center were eligible for study inclusion. The study was performed from June 2018 to September 2019. Patients were excluded if any of the following criteria were met: 1) patient unable to provide informed consent (e.g. due to pre-existing severe neurological or psychiatric illness) or institutionalized patients, 2) Age < 18 years and/or 3) patients not speaking German or French. Besides additional biomarker monitoring, all patients received best standard-of-care treatment according to local standard operating procedures (SOP) without interference by the research team. The study was approved by the local Ethics Committee on Human Research (Kantonale Ethikkomission, KEK, Bern, Nr. 2018-00331). Written informed consent was obtained from all patients or respective representatives. The study was registered on Clinicaltrials.gov (NCT 03469765, registered March 19, 2018). Data quality visits (in-/exlusion criteria, informed consent, primary endpoint data) were performed by an internal qualified person.

Biomarker assessment and patient data
Urinary and blood samples were obtained after anesthesia induction (baseline), after end of anesthesia (before IMC/ICU-admission, PO), and in the morning of the day after surgery (minimum of 8 hours up to 24 hours post-operatively, POD1). Urinary concentrations of TIMP-2 and IGFBP7 were measured using a commercially available standardized assay (NephroCheck, Astute Medical, San Diego, CA, USA). Test results are given as product of both markers in (ng/ml) 2 / 1000. Blood and urinary creatinine and sodium concentrations were assessed to calculate fractional sodium excretion. Other post-operative laboratory parameters were assessed at the discretion of the independent treating physician.
The following pre-operative data were recorded: age; sex; BMI; American Society of Anesthesiologists (ASA) physical status classification system; comorbidities and routine medications. Laboratory parameters including serum creatinine, urea, sodium, eGFR (Chronic Kidney Disease Epidemiology Collaboration [CKD-EPI] formula) [30]. Diagnostic procedures using contrast agents (if applicable) were recorded in the 24 hours proceeding operations. The following perioperative data were recorded: type of surgery (infra-vs. suprarenal clamping, EVAR and emergency vs. elective); duration of clamping (if applicable); use/ duration of cold renal perfusion; use/ quantity of contrast agents (if applicable); duration of surgery; episodes of hypotension (mean arterial blood pressure < 20% of baseline for more than 5 minutes); cumulative dose/ type of fluids applied including autologous and allogenic packed red blood cells (RBC), fresh frozen plasma (FFP), and platelet transfusions; blood loss; use/ dose of mannitol; use/ cumulative dose of vasopressors or inotropes, and urine output during surgery. Data recorded post-operatively included routine laboratory parameters until hospital discharge (if available: sodium, potassium, serum creatinine, urea, eGFR, hemoglobin); cumulative urinary output while on IMC/ ICU; clinically overt complications after surgery (sepsis, colon ischemia, wound infection, pneumonia, delirium, others); need for and duration of RRT; length-of stay in IMC and/or ICU, length of hospital stay. ICU-/IMC-, in-hospital-and 28-day-mortality was recorded.

Sample size calculation
An incidence of 20% po-AKI was expected in the cohort under investigation. Based on (TIMP-2) x (IGFBP7) data of previous studies in cardiac surgery patients [22,26], we assumed to reach a mean (TIMP-2) x (IGFBP7) in patients without AKI of 0.4 (ng/ml) 2 / 1000 and in patients with AKI (i.e. KDIGO stages 1-3) a mean of 1.6 (ng/ml) 2 / 1000 with a standard deviation (SD) of 1.4. If the true difference in the experimental and control means is 1.2, n = 18 experimental subjects and 72 control subjects are required to reject the null hypothesis that the population means of the experimental and control groups are equal with a probability (power) of 0.9. The Type I error probability associated with this test of this null hypothesis is 0.05. Assuming a drop-out rate greater than 10%, a total of 100 patients are intended to be included.

Statistical analysis
For statistical analysis, Student's t-test for normally distributed data and the Mann-Whitney U test for non-normally distributed data was used to compare quantitative variables between groups, ANOVA for multiple groups. Chi square test or Fisher's exact test were used for qualitative variables, as appropriate. Statistical significance was assumed for p<0.05. Sensitivity-/ specificity-analysis were performed using receiver-operating characteristics (ROC) curves using the R software package cutpointr. Diagnostic value of a biomarker is defined as excellent with Area Under the Receiver Operating Characteristics curve (AUROC) >0.9, as good with a AUROC between 0.75-0.9, as poor with a AUROC between 0.50-0.75 and as without any diagnostic value with AUROC values <0.5 [31]. Optimal cut-off values were estimated using the Youden index [31]. No optimal cut-off point was calculated for curves with poor or lower diagnostic value and for exploratory analyses. All statistical analyses were performed using the softwares SPSS for Windows (version 25;SPSS Inc., Chicago, IL, USA) and R, version 3.6.1, and Rstudio version 1.2.5001.

Clinical outcomes and fractional excretion of sodium (FeNa)
In patients with po-AKI, an early post-operative increase in serum creatinine levels after surgery was observed. No differences were noted in regard to early fractional sodium excretions between patients with vs. without AKI (Table 3). AUROC for FeNa-values immediately after surgery and on POD1 were 0.57 (95%CI 0.47-0.66) and 0.47 (95%CI 0.37-0.57), respectively. One (3.2%) patient died in the AKI group (Table 3). Three (10%) patients required RRT postoperatively, with one (3.2%) patient in need for RRT at day 28. Longer ICU-, IMC-, and hospital stay was noted in patients with AKI (all p < .05) ( Table 3). Increased rates of clinically overt postsurgical complications were noted in the AKI group (71% vs 26% without AKI; p< .0001) ( Table 3).

Discussion
In an investigator-initiated prospective monocentric observational study, we did not observe differences in early postoperative (TIMP-2) x (IGFBP7) levels in abdominal aortic surgery patients with versus without po-AKI which developed in the first seven days after surgery. Higher median (TIMP-2) x (IGFBP7) levels were noted in cases of severe AKI (KDIGO stage 3). These findings were consistent along the investigated subgroups. Our findings are in contrast to previous studies demonstrating good accuracy in prediction of all-stage AKI [18,22] or KDIGO-AKI-stages 2 and 3 [13,18,19,24]. However, other authors showed that (TIMP-2) x (IGFBP7) may distinguish between patients with vs. without AKI, but with limited diagnostic accuracy following cardiac surgery [25,26]. We deliberately aimed to investigate (TIMP-2) x (IGFBP7) in patients undergoing abdominal aortic procedures as this patient group is considered at particular high risk for post-operative AKI [3]. Therefore, our patient cohort may include some heterogeneity with regard to the surgical intervention chosen (open vs. endovascular repair, emergency vs. scheduled procedures) and po-AKI etiology (inflammation, ischemia-reperfusion injury, [micro-]embolization, contrast media application, hemodynamic perturbations, and others). However, the recent literature suggests that the biomarker under investigation reflects "renal stress" induced following various renal injuries (inflammation, toxins, drugs, oxidative stress) [12]. Others have previously tested TIMP2xIGFBP7 in a variety of pathologies leading to AKI including e.g. cardiac surgery, major non-cardiac surgery, transcatheter aortic valve implantation (TAVI), critical illness, heart failure and platinum-based chemotherapy and have showed rather consistent results [17,18,20,22,23,32]. Thus, current recommendations suggest (TIMP-2) x (IGFBP7) testing in different patient cohorts irrespective of po-AKI-etiology [16], as performed in this study.
The po-AKI incidence observed (33% in overall cohort, 39% in patients receiving open aortic interventions) might be comparable to previous studies including cardiac and major noncardiac surgery [22,24]. Of note, the subgroup with moderate to severe AKI according KDIGO (12% in our study) appeared rather comparable to other studies [22,26] or the rate is even higher [24,25]. Further, demographics and course of respective patients (including length of stay and number of postoperative complications) was also considered rather typical for AKI patients.
Recommendations for (TIMP-2) x (IGFBP7) assessment vary regarding the time of assessment [26]. We thus chose to measure (TIMP-2) x (IGFBP7) immediately after surgery and in the morning of the first postoperative day. This was performed deliberately for pragmatic reasons rather than artificial study conditions. On many if not most ICUs, respective markers would be tested at the timepoints investigated. However, earlier studies in cardiac and noncardiac major surgery differed with regard to timing of (TIMP-2) x (IGFBP7) testing and chosen time points ranged from intra-operative testing [24], to 0 to 6 hours post-operatively [18,[24][25][26], until post-operative day 3 [24]. Some authors tested urinary (TIMP-2) x (IGFBP7) immediately after surgery and observed decreasing (TIMP-2) x (IGFBP7) values (when compared to baseline) without differences in (TIMP-2) x (IGFBP7) levels between patients with vs. without AKI [24,25], which is in line with our findings.
Earlier studies in healthy subjects and in patients with stable chronic comorbidities showed a (TIMP-2) x (IGFBP7) reference intervals of 0.04 to 2.22 (ng/ml) 2 / 1000 with a median at about 0.3 (and 75% <0.75 (ng/ml) 2 / 1000) [33]. One might thus argue whether the observed increased baseline values in our study would reflect this range. Nevertheless, although we are unable to conclude on the underlying mechanisms, respective baseline levels may theoretically be related to pre-operative fasting. AKI stage 3 patients were mostly emergency patients and might theoretically have been pre-hydrated to a larger extent.
Further, timing of (TIMP-2) x (IGFBP7) may be crucial. Measurements 4-6 hours after surgery could be regarded the earliest time point for assessment [22,24]. In addition, underlying intraoperative mechanisms for AKI may differ in cardiac and vascular surgery patients (e.g. altered renal blood flow, inflammation, and haemolysis) [11], which could theoretically explain some of the differences between previous cardiac surgery studies and the current presented data. Nevertheless, in our study, the typical post-operative course of plasma creatinine and/ or decline in urinary output at 12 and 24 hours post-operatively indicated most patients with moderate to severe AKI within 14 hours. Further, 2 out of 3 patients received RRT within 24 hours. A theoretical "ideal" renal biomarker would likely diagnose moderate to severe AKI early after the impact, with good discriminatory power superior or additive to established renal biomarkers. In accordance with other studies [25,26], our data might thus challenge the early use of urinary (TIMP-2) x (IGFBP7) in patients with abdominal aortic procedures. Previously, patients with negative urinary (TIMP-2) x (IGFBP7) (<0,3[ng/ml] 2 /1000) were proposed as potential "fast track" patients [16]. Here, 50% of patients with KDIGO AKI stage 2 had levels <0. 3. Thus, respective recommendations should likely be limited to cardiac surgery patients in which more data is available. However, as course of creatinine and/or urinary output indicated development of po-AKI in nearly all patients with in the first 14hours postsurgery, we theoretically might have missed maximum TIMP2xIFGBP7 concentrations. Therefore, further studies on biomarker kinetics such as the study of Cummings et al. [34] seem warranted.
Our study has important further limitations that deserve discussion. First, even if adequately powered (assumed vs. observed AKI incidence: 20% vs. 33%; assumed vs. observed drop-out rate 10% vs. 7%) to detect a difference in patients with AKI according to all KDIGO stages, the sample size of our study is limited and our results warrant larger confirmatory investigations. Second, we present data from a monocentric study with all inherent limitations. Third, pre-operative risk prediction and selection of patients (at higher risk) might have (theoretically) influenced biomarker performance [22,35]. Fourth, we assessed (TIMP-2) x (IGFBP7) at baseline and twice post-operatively. As kinetics appear not yet exactly understood, exact timing may be crucial, and more measurements might have potentially revealed a larger (temporary) rise in affected patients. Fifth, we did not test (TIMP-2) x (IGFBP7) in combination with other biomarkers e.g. urinary neutrophil gelatinase-associated lipocalin (NGAL) or peak serum cystatin C. Combining several biomarkers may improve predictive power [36]. Sixth, building of ROC curves and estimating AUROC in the subgroup of KDIGO-AKI stage 3 patients was based on a limited number of study patients. However, we aimed to account for this by bootstrapping with 1000 iterations. Seventh, in line with previous studies [22,26], this study was designed to potentially predict AKI at any AKI stage, whereas the biomarker is only validated to detect moderate to severe AKI [13].

Conclusions
In an investigator-initiated prospective monocentric observational study, we did not observe differences in early postoperative (TIMP-2) x (IGFBP7) levels in abdominal aortic surgery patients with versus without postoperative AKI that developed in the first seven days after surgery. Subgroup analysis showed higher median (TIMP-2) x (IGFBP7) levels in patients with severe postoperative AKI (KDIGO stage 3). Further studies are required regarding the specific type of interventions and the kinetics of the biomarker.