Neutrophil Gelatinase-Associated Lipocalin (NGAL) in Patients with ST-Elevation Myocardial Infarction and Its Association with Acute Kidney Injury and Mortality

Neutrophil gelatinase-associated lipocalin (NGAL) is an inflammatory biomarker related to acute kidney injury (AKI). Including 1892 consecutive patients with ST-elevation myocardial infarction (STEMI), in which NGAL was measured in 1624 (86%) on admission and in a consecutive subgroup at 6–12 h (n = 163) and 12–24 h (n = 222) after admission, this study aimed to evaluate the prognostic value of NGAL in predicting AKI and mortality. Patients were stratified based on whether their admission NGAL plasma concentration was greater than or equal to/less than the median. The primary endpoint was a composite of the first occurrence of AKI or all-cause death within 30 days. AKI was classified by the maximal plasma creatinine increase from baseline during index admission as KDIGO1 (<200% increase) or KDIGO23 (≥200% increase) according to the Kidney Disease Improving Global Outcomes (KDIGO) system. Admission NGAL > the median was independently associated with a higher risk of severe AKI (KDIGO2-3) and 30-day all-cause mortality when adjusted for age, admission systolic blood pressure and high-sensitivity C-reactive protein, left-ventricular ejection fraction, known kidney dysfunction, and cardiogenic shock with an odds ratio (95% confidence interval) of 2.26 (1.18–4.51), p = 0.014. Finally, we observed increasing predictive values in a subgroup during the first day of hospitalization suggesting that assessment of NGAL should be delayed for optimal prognostic purposes.


Introduction
Ischemic heart disease is the most frequent cause of death globally [1]. Of surviving patients with acute myocardial infarction, up to 26% develop acute kidney injury (AKI) during hospitalization [2]. AKI is associated with increased mortality in patients with ST-elevation myocardial infarction (STEMI) [3].
NGAL is a biomarker secreted from neutrophils of the innate immune system in response to inflammation and renal ischemia [4,5]. High blood concentrations of NGAL have previously been associated with AKI [3,6] and all-cause mortality in small cohorts of STEMI patients [7][8][9][10].
NGAL is a biomarker secreted from neutrophils of the innate immune system response to inflammation and renal ischemia [4,5]. High blood concentrations of NG have previously been associated with AKI [3,6] and all-cause mortality in small cohort STEMI patients [7][8][9][10].
Due to the early accumulation and detectability of circulating NGAL compared w creatinine [11,12], NGAL is a promising biomarker in early prediction of AKI. W creatinine is the all-time gold standard for detecting kidney injury in real time, NG may predict future AKI, allowing us to limit kidney injury and improve the gene prognosis of STEMI patients.
In this study, we assessed the association between NGAL plasma concentration admission and the development of AKI and all-cause death within 30 days in STE patients. Additionally, the study evaluated the temporal changes of NGAL plas concentration after admission and the prognostic value hereof.

Materials and Methods
This study population ( Figure 1) has previously been described in detail [13]. Brie this prospective, observational study aimed to include a consecutive cohort of patie with STEMI from two tertiary heart centers in Denmark (Copenhagen Univer Hospital, Rigshospitalet, and Odense University Hospital) over one year in 2015 and 20 Figure 1. Flow diagram depicting inclusion of patients in the cohort and subgroup. Patients w included in the study if they had verified STEMI and available plasma creatinine levels admission. Outcomes of patients were based on the level of AKI according to the KDI classification system and/or death within 30 days. In a subgroup of included STEMI patie biomarkers were measured again at 6-12 and 12-24 h after admission. CAG = coron angiography. STEMI = ST-elevation myocardial infarction. KDIGO = Kidney Disease Improv Global Outcomes (classification system). NGAL = Neutrophil gelatinase-associated lipocalin. CRP = High-sensitivity C-reactive protein.

Study Population
On arrival in the catheterization laboratory, patient characteristics, hemodyna parameters, and time from symptom debut to acute coronary angiography (CAG) w Outcomes of patients were based on the level of AKI according to the KDIGO classification system and/or death within 30 days. In a subgroup of included STEMI patients, biomarkers were measured again at 6-12 and 12-24 h after admission. CAG = coronary angiography. STEMI = ST-elevation myocardial infarction. KDIGO = Kidney Disease Improving Global Outcomes (classification system). NGAL = Neutrophil gelatinase-associated lipocalin. Hs-CRP = High-sensitivity C-reactive protein.

Study Population
On arrival in the catheterization laboratory, patient characteristics, hemodynamic parameters, and time from symptom debut to acute coronary angiography (CAG) were registered. During a 3-month period from January to March 2016, repeated plasma samples and biochemical measurements were obtained in a consecutive subgroup of patients with STEMI admitted at Rigshospitalet (Figure 1).

Kidney Disease Improving Global Outcomes (KDIGO) Criteria and 30-Day All-Cause Mortality
The definition and level of AKI were based on the serum creatinine criteria of the KDIGO classification system [14]. In this study, patients were stratified according to the maximal increase in plasma creatinine from baseline during index admission:

Blood Sampling and Biomarker Analysis
On arrival in the catheterization laboratory, arterial blood samples were taken immediately after sheath insertion before the CAG. Blood samples were collected in lithium-heparin tubes, centrifuged at 2000 G, and plasma was stored at -80 • C.
Time points for temporal biomarker measurements in the subgroup were 0 h (baseline), 6-12 h, and 12-24 h after admission.

Statistical Analysis
Categorical variables are presented as numbers (n) and percentages (%). Pearson's Chi-squared test or Fisher's exact test were used accordingly to assess differences between categorical variables. Continuous variables are presented as the mean with standard deviation (SD) or the median with 25th and 75th percentile (25th; 75th). Differences between continuous variables were tested using the Wilcoxon rank test or Kruskal-Wallis rank sum test for >2 groups.
Patients were stratified into > versus ≤ the median of admission NGAL plasma concentration. The cumulation of KDIGO2-3 and 30-day all-cause mortality was used as the composite endpoint.
The association between admission NGAL plasma concentration and composite endpoint was assessed using logistic regression in a univariable and multivariable model (adjusted for age, admission systolic blood pressure, left ventricular ejection fraction (LVEF), known kidney dysfunction, baseline hs-CRP, and cardiogenic shock (CS)). The logistic models were assessed for both NGAL concentration as a continued variable (log 2 transformed to obtain a normal distribution) and for NGAL concentration dichotomized into > versus ≤ the median on admission. Finally, the logistic regression models were reassessed without patients in CS.
The association between the cumulative incidence of KDIGO2-3 and 30-day all-cause mortality as a competing event in patients with admission NGAL plasma concentration > versus ≤ to the median was assessed using Gray's test for competing risk. The log-rank test was used to compare Kaplan-Meier survival curves between NGAL concentration > and ≤ to the median for the composite endpoint.
The predictive value of admission biomarker plasma concentrations and the composite endpoint were assessed as the area under the receiver operating characteristics curve (AUC ROC ). The biomarker concentration with the highest combined sensitivity and specificity comprised the optimal concentration for predicting the composite endpoint.
In the subgroup, temporal changes in biomarker plasma concentrations were assessed using a linear mixed model of covariance. The predictive value of NGAL concentration at each time point (baseline, 6-12, and 12-24 h after admission) and composite endpoint was assessed as AUC ROC .
A causal mediation analysis was performed to determine the relative impact of baseline NGAL and hs-CRP plasma concentration on the composite endpoint.
Statistical analyses were performed using RStudio version 2022.07.1 (The R Foundation, Vienna, Austria) and SAS Enterprise software version 7.4 (SAS Institute, Cary, NC, USA). A two-sided p-value < 0.05 was considered statistically significant.
NGAL concentration was measured in 1624 patients (86%) on admission and in a consecutive subsample at 6-12 h (n = 163) and 12-24 h (n = 222) after admission. NGAL concentrations on admission were not available for the remaining 253 patients (14%) due to early death, early transfer to local hospital, or assay failure.
When dichotomizing patients into NGAL concentration on admission > versus ≤ the median of 79 (56; 118) ng/mL (Table 1), patients with NGAL concentration > the median were older and more frequently found to have hypertension, peripheral artery disease (PAD), IHD, known kidney dysfunction (creatinine levels > 100 mmol/L and/or dialysis treatment), heart failure, and previous stroke. These patients were in worse hemodynamical conditions with lower systolic blood pressure, lower LVEF, and higher heart rate, more likely to have anterior STEMI, multi-vessel disease, and to be comatose after cardiac arrest, and in CS. The composite endpoint occurred in 107 (13%) patients with NGAL > the median compared with 22 patients (2.7%) with NGAL ≤ the median, p < 0.001 ( Table 1). The proportion of patients alive 30 days after admission but who developed KDIGO2-3 (n = 36) was similarly higher in patients with NGAL > the median (31 (3.9%) vs. 5 (0.6%), p < 0.001).
In the logistic regression model, both dichotomized admission NGAL plasma concentration (odds ratio (OR) (95% confidence interval (CI) 5.64 (3.60-9.25), p < 0.001 (univariable) and 2.26 (1.18-4.51), p = 0.014 (multivariable)) and log 2 transformed NGAL concentration (3.16 (2.59-3.90), p < 0.001 (univariable) and 1.75 (1.24-2.45), p = 0.002 (multivariable)) were associated with an increased risk of the composite endpoint ( Table 2). The logistic regression models were reassessed excluding patients in CS, with similar results. In a cumulative incidence plot (Figure 2), we found cumulative events and incidence to be associated with patients with NGAL > the median.   Considering the two endpoints (KDIGO2-3 and 30-day mortality) separately found both log2 transformed and dichotomized NGAL concentrations on admission associated with 30-day all-cause mortality in both the univariable and the multivar model. Regarding KDIGO2-3, both log2 transformed and dichotomized N concentrations on admission were associated with KDIGO2-3 in the univariable m but not in the multivariable model: p = 0.12 (log2 transformed) and p = 0.31 (dichotom (Table 3).  Considering the two endpoints (KDIGO2-3 and 30-day mortality) separately, we found both log 2 transformed and dichotomized NGAL concentrations on admission to be associated with 30-day all-cause mortality in both the univariable and the multivariable model. Regarding KDIGO2-3, both log 2 transformed and dichotomized NGAL concentrations on admission were associated with KDIGO2-3 in the univariable model but not in the multivariable model: p = 0.12 (log 2 transformed) and p = 0.31 (dichotomized) ( Table 3).    Temporal changes in biomarkers for patients by outcome are depicted in Figure 4. Regarding NGAL, we observed immediately increasing concentrations of NGAL in the subsequent twelve hours for patients with the primary composite endpoint, whereas no increase was observed in patients without a primary composite endpoint event. The predictive values for composite endpoint increased during the first day of hospitalization for all biomarkers. However, the number of cases in the subgroup is low (n = 163; 6-12 h after admission, n = 222; 12-24 h after admission), limiting the interpretation (Figure 3).  Through causal mediation analysis, the direct effect of baseline NGAL on composite endpoint (ADE: average direct effect) was estimated to 0.00017 (95% 0.00012-0.00022) (p < 0.001) and the indirect effect of baseline NGAL plas concentrations through baseline hs-CRP (ACME: average causal mediation effect) w estimated to 0.000015 (0.0000039; 0.000030) (p = 0.01), leaving the relative effect of mediator hs-CRP through NGAL to be 0.088 (0.033; 0.14) (p = 0.006). The causal mediat effect is visualized in Figure 5.

Discussion
In a cohort of more than 1600 consecutive STEMI patients, we found admiss NGAL plasma concentrations > the median to be independently associated with a hig risk of severe AKI (KDIGO2-3) and death from all causes within 30 days. Finally, our d Through causal mediation analysis, the direct effect of baseline NGAL on the composite endpoint (ADE: average direct effect) was estimated to 0.00017 (95% CI: 0.00012-0.00022) (p < 0.001) and the indirect effect of baseline NGAL plasma concentrations through baseline hs-CRP (ACME: average causal mediation effect) was estimated to 0.000015 (0.0000039; 0.000030) (p = 0.01), leaving the relative effect of the mediator hs-CRP through NGAL to be 0.088 (0.033; 0.14) (p = 0.006). The causal mediation effect is visualized in Figure 5. Through causal mediation analysis, the direct effect of baseline NGAL on the composite endpoint (ADE: average direct effect) was estimated to 0.00017 (95% CI: 0.00012-0.00022) (p < 0.001) and the indirect effect of baseline NGAL plasma concentrations through baseline hs-CRP (ACME: average causal mediation effect) was estimated to 0.000015 (0.0000039; 0.000030) (p = 0.01), leaving the relative effect of the mediator hs-CRP through NGAL to be 0.088 (0.033; 0.14) (p = 0.006). The causal mediation effect is visualized in Figure 5.

Discussion
In a cohort of more than 1600 consecutive STEMI patients, we found admission NGAL plasma concentrations > the median to be independently associated with a higher

Discussion
In a cohort of more than 1600 consecutive STEMI patients, we found admission NGAL plasma concentrations > the median to be independently associated with a higher risk of severe AKI (KDIGO2-3) and death from all causes within 30 days. Finally, our data indicate the predictive value of NGAL plasma concentration to be higher 6-24 h after index admission.
NGAL is a biomarker released from mature neutrophils during diapedesis and from nephrons during renal injury and ischemia [4,5]. NGAL is accumulated and detected earlier after kidney injury compared with plasma creatinine [5] and can differentiate between structural and functional kidney injuries such as ischemia [16]. In this study, 403 patients had levels of NGAL on admission between 100 and 200 ng/mL, representing a cardiac surgery-associated NGAL (CSA-NGAL) score of 1 and possible tubular damage, while 144 patients had NGAL levels on admission > 200 ng/mL, indicating a CSA-NGAL score of 2-3 and significant or severe tubular damage [17].
Patients with increased concentrations of NGAL (>the median) had higher baseline hs-CRP, which corresponded with previous studies [18,19]. Hs-CRP is known to be associated with AKI and ongoing inflammation. NGAL may reflect a combination of tubular injury and inflammatory responses. To minimize the effect of inflammation on the relationship between NGAL and the composite endpoint (KDIGO2-3 and 30-day all-cause mortality), we included hs-CRP in the multivariable analyses and found NGAL > the median to be independently correlated to the composite endpoint. In the causal mediation analysis, it appears that most of the total effect is explained by the direct effect from baseline NGAL on the composite endpoint; however, there is also a minor indirect effect of hs-CRP on the composite endpoint.
In a study by Cosentino et al., CS was independently correlated with both mortality and AKI [20]. In this study, 9.4% of the total cohort had CS; 3.2% in the group with NGAL ≤ the median and 16% in the group > the median (p < 0.001). To eliminate the fact that cardiogenic shock contributes particularly to the development of AKI and death, we adjusted for cardiogenic shock in the multivariable analyses. Furthermore, we assessed the logistic regression analyses excluding patients with CS and found similar OR estimates and p values.
In a previous study, 30% of surviving AMI patients developed AKI during hospitalization [2]. In this study, only 10% of surviving STEMI patients developed AKI during hospitalization, perhaps due to fewer comorbidities, younger age, more fortunate geography, and newer techniques of treatment.
In a cohort study of more than 100,000 AMI patients, mild renal dysfunction (defined as glomerular filtration rate (GFR) < 67 mL/min per 1.73 m 2 ) increased the 1-year mortality risk by 9%, while severe renal dysfunction (GFR < 31 mL/min per 1.73 m 2 ) more than doubled the 1-year mortality risk [21]. Increased NGAL concentrations have previously been predictive of AKI in smaller cohort studies of AMI patients (n < 1050) [6,[22][23][24] and increased mortality in STEMI patients [7][8][9][25][26][27]. This study correlates with previous studies of smaller STEMI cohort sizes. To the best of our knowledge, the STEMI population in this study is the largest of its kind.
Since both inflammation and kidney dysfunction are associated with NGAL and unfavorable outcomes in STEMI patients, NGAL plasma concentration may improve the prognostic prediction and identify high-risk patients for whom close follow-ups are essential for survival and nephrotoxic treatment should be considered with great care.
In this study, the predictive values of NGAL plasma concentrations for the composite endpoint trends higher compared with hs-CRP and creatinine both on admission and 6-12 and 12-24 h after admission, which may suggest that NGAL is a better predictor of severe AKI (KDIGO2-3) and 30-day all-cause mortality during index admission. Previously, it has been proposed that NGAL is detected within a few hours of tubular damage before glomerular dysfunction leads to the accumulation of creatinine in the bloodstream [28]. This may explain the lower predictive values of creatinine compared with NGAL. The late accumulation of serum creatinine may cause delayed identification of decreasing GFR, often calculated as Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) in clinical practice, possibly delaying the initiation of preventive measures of AKI. NGAL has previously been found to be strongly correlated with estimated GFR [29,30].
Additionally, we found increasing predictive values when NGAL concentrations were measured 6-24 h after admission compared with the baseline in a smaller subgroup, suggesting that assessment of NGAL should be delayed for optimal prognostic purposes. However, the size of the subgroup is small, limiting the interpretation; nonetheless, the result invites future studies of the matter.

Limitations
The primary limitation of this study is the fact that NGAL values at admission were unavailable for 14% of the cohort. Furthermore, we did not adjust for creatinine concentration in the multivariable analyses since we assumed creatinine to be closely correlated to the plasma concentration of NGAL, possibly removing the prognostic signal of NGAL. These limitations may cause an over-or underestimation of the prognostic value in the initial risk stratification. However, our findings match previous results from minor studies.

Conclusions
NGAL plasma concentration on admission > the median was independently associated with increased risk of severe AKI (KDIGO2-3) and 30-day all-cause mortality in STEMI patients. Furthermore, we observed increasing predictive values of NGAL plasma concentrations in a subgroup during the first day of hospitalization, suggesting that assessment of NGAL should be delayed for optimal prognostic purposes. However, the number of cases in the subgroup were few, limiting the interpretation, and should be verified in future studies.

Informed Consent Statement:
Patients were asked to give written, informed consent as soon as possible after blood sampling. If patients were unable to give informed consent, we obtained consent by proxy from the patient's next of kin and, eventually, from the patient's general practitioner according to national guidelines at the time. Patients younger than 18 years old and patients without informed consent were excluded. Data Availability Statement: Not applicable.