The neutrophil percentage-to-albumin ratio is associated with all-cause mortality in critically ill patients with acute myocardial infarction

Aim In this study, we evaluated the utility of neutrophil percentage-to-albumin ratio (NPAR) in predicting in critically ill patients with acute myocardial infarction (AMI). Methods The information of patients were collected from Medical Information Mart for Intensive Care III database. Admission NPAR was calculated as neutrophil percentage divided by serum albumin. The endpoints of this study were 30-day, 90-day, 180-day, and 365-day all-cause mortality. Cox proportional hazards models and subgroup analyses were used to determine the relationship between admission NPAR and these endpoints. Results 798 critically ill patients with AMI were enrolled in. After adjustments for age, race and gender, higher admission NPAR was associated with increased risk of 30-day, 90-day, 180-day, and 365-day all-cause mortality in critically ill patients with AMI. And after adjusting for possible confounding variables, two different trends have emerged. Stratified by tertiles, high admission NPAR was independently associated with 180-day and 365-day all-cause mortality in critically ill patients with AMI (tertile 3 vs. tertile 1: adjusted HR, 95% CI 1.71, 1.10–2.66, p < 0.05; 1.66, 1.10–2.51, p < 0.05). In other hand, stratified by quartiles, highest admission NPAR levels were independently associated with 90-day, 180-day and 365-day all-cause mortality (quartile 4 vs. quartile 1: adjusted HR, 95% CI 2.36, 1.32–4.23, p < 0.05; 2.58, 1.49–4.47, p < 0.05; 2.61, 1.56–4.37, p < 0.05). ROC test showed that admission NPAR had a moderate ability to predict all-cause mortality of critically ill patients with AMI. No obvious interaction was found by subgroup analysis in most subgroups. Conclusions Admission NPAR was an independent predictor for 180-day and 365-day all-cause mortality in critically ill patients with AMI.


Introduction
As it is well known, acute myocardial infarction (AMI) as a common cardiovascular disease continues to be the leading cause of hospital admission and mortality rate worldwide in the past years [1]. Although potent antiplatelet agents and early revascularization has greatly reduced the occurrence of major adverse cardiovascular events in AMI patients [2], the prognosis of AMI patients is still an apparent problem worthy of our attention.
The inflammatory response participates in myocardial infarction (MI) area and left ventricular (LV) remodeling [3,4]. Neutrophil, the major participant in the inflammatory burst, mediates the inflammatory response to myocardial injury [3,5]. Albumin acts through its multiple binding sites and free radical-capture properties, involved in antioxidant activities [6]. It has been confirmed that low albumin level has a significant effect on the mortality and prognosis of AMI [7][8][9].
According to the previous studies, admission neutrophil percentage-to-albumin ratio (NPAR) was an independent predictor of in-hospital mortality in patients with acute ST-segment elevation myocardial infarction (STEMI) [10]. In the another study, in critically ill patients with coronary artery disease (CAD), the higher NPAR level was closely correlated with the higher rate of 30-day, 90-day, and 365-day all-cause death [11]. However, there is no study yet that reported the association between admission NPAR level and the all-cause mortality in critically ill patients with AMI. For all the above reasons, in this study we hypothesized that admission NPAR level could be a prognostic predictor of all-cause mortality in critically ill patients with AMI.

Source of data
We conducted a retrospective cohort study, where the data was collected from a large, single-center critical care database called Medical Information Mart for Intensive Care III (MIMIC III) v1.4 [12]. MIMIC III is a public and freely available database and integrates comprehensive clinical data of patients admitted to the intensive care units (ICU) at the Beth Israel Deaconess Medical Center between 2001 and 2012. This database was approved by the Institutional Review Boards (IRB) of the Massachusetts Institute of Technology (MIT). After successfully completing the National Institutes of Health (NIH) Web-based training course and the Protecting Human Research Participants examination (no. 40683764), we were given the permission to extract data from MIMIC III.

Population selection criteria
All ICU inpatients with AMI diagnosed by ICD-9 diagnosis code were enrolled in this study. Exclusion criteria were as follows: (1) multiple ICU admissions; (2) aged < 18 years old; (3) had died before admission; (4) lack of the information of albumin and neutrophil percentage during ICU stay.

Data extraction and definition of NPAR
Data extraction from MIMIC III was accomplished using Structured Query Language (SQL) with the PostgreSQL tool (version 9.6). We extracted demographics, vital signs, comorbidities and medical history, laboratory parameters, scoring system, medication use, and clinical survival information. Age, gender and race were contained in demographics, and vital signs included heart rate, respiratory rate and mean blood pressure (MBP). Comorbidities and medical history included CAD, prior MI, atrial fibrillation (AF), chronic heart failure (CHF), chronic kidney disease (CDK), chronic obstructive pulmonary disease (COPD) and percutaneous transluminal coronary angioplasty/percutaneous coronary intervention (PTCA/ PCI). Laboratory parameters were consisted of neutrophils, albumin, hemoglobin, white blood cell (WBC), prothrombin time (PT), platelet, red cell distribution width (RDW), potassium, sodium, creatinine, blood urea nitrogen (BUN), alanine aminotransferase (ALT), aspartate aminotransferase (AST), creatine kinase-peak (CK), MB isoenzyme of creatine kinase (CK-MB), and glucose. Scoring systems contains two important scoring scalessequential organ failure assessment (SOFA) score [13] and simplified acute physiology score II (SAPS II) [14]. In addition, the medication use of patients in this study would be shown in the baseline characteristics, including whether they have used aspirin, clopidogrel, metoprolol, angiotensin-converting enzyme inhibitors/angiotensin receptor blockers (ACEI/ARBs) or statins. The other extracted data have weight and urine output within 24 h. All the laboratory parameters were firstly-measured data after admission to the ICU. The endpoints of this study were 30-day, 90-day, 180-day, and 365-day all-cause mortality. We divided neutrophil percentage by albumin to get admission NPAR [10].

Statistical analysis
All the patients with AMI in the study were stratified in term of admission NPAR tertiles. Data distribution of all continuous variables were tested using the Shapiro-Wilk test, and they were nonnormally distributed, and manifested as median and interquartile range (Q1-Q3). All categorical data was expressed as number and percentage. Kruskal-Wallis or Fisher's exact test was performed to evaluate statistical differences among different groups of NPAR.
By using Log-rank tests, survival rates of different groups were compared, and the Kaplan-Meier curves were built.
In order to evaluate the independent effect of admission NPAR on 30-day, 90-day, 180-day, and 365-day allcause mortality, cox proportional hazard models were developed. The first tertile and quartile groups of admission NPAR were treated as the reference group, and the results were summarized as hazard ratios (HR) with 95% confidence intervals (CI). In model I, age, race and gender were incorporated into adjustment. In model II, we further adjusted for age, gender, race, respiratory rate, MBP, heart rate, ALT, AST, CK-peak, CK-MB-peak, glucose, PT, hemoglobin, RDW, creatinine, potassium, sodium, BUN, WBC, platelet, CAD, AF, COPD, hypertension, diabetes, prior MI, CHF, CKD, stroke, SOFA and SAPS II. And P for trend was calculated. Subgroup analysis was conducted to estimate the effect of admission NPAR on 180-day all-cause mortality. And we got the P value for interaction.
Receiver-operating characteristic (ROC) curve was performed to measure the sensitivity and specificity of admission NPAR, as well as SOFA score. Moreover, the area under the curve (AUC) was calculated to estimate the quality of admission NPAR as a predictor of 365-day all-cause mortality. Statistical analyses were performed using EmpowerStats version 2.0 (http:// www. empow ersta ts. com/ cn/, X&Y solutions, Inc., Boston, MA) and R software version 3.4.3; P value of < 0.05 was considered to be statistically significant.

Baseline characteristics of patients
After reviewing the data of 61,532 critically ill patients, a total of 798 patients with AMI were enrolled in our study (Fig. 1). Based on tertiles of admission NPAR level, participants were categorized into three groups (tertile 1: < 21.58; tertile 2: ≥ 21.58, < 26.77; and tertile 3: ≥ 26.77), and each group included 266 AMI patients. The baseline characteristics were displayed in Table 1. Patients in the highest tertile of admission NPAR level were older than other groups, and most of them were white. In addition, they reported more medical history of AF, but less comorbidities of CAD, hypertension and CHF. Moreover, patients in the highest tertile of admission NPAR level were less likely to use aspirin, clopidogrel, metoprolol, ACEI/ARBs and statin, and to receive PTCA or PCI. Finally, they had lower MBP, weight, albumin, hemoglobin, urine output in 24 h, and higher values of heart rate, neutrophils, WBC, PT, RDW, creatinine, BUN, ALT, AST, SAPS II and SOFA.

Admission NPAR and outcome
As it had been shown in Table 2, the overall length of ICU stay (LOS) was 3.69 days, and the overall in-hospital, 30-day, 90-day, 180-day and 365-day all-cause mortality were 17.92%, 19.67%, 25.44%, 29.82% and 33.21%, respectively. Furthermore, as admission NPAR levels increased, the all-cause death rate of in-hospital, 30-day, 90-day, 180-day and 365-day were distinctly raised.

Admission NPAR as a predictor of the clinical endpoints
In cox regression models, admission NPAR levels were stratified by tertiles and quartiles, to appraise whether admission NPAR was related to 30-day, 90-day, 180-day and 365-day all-cause mortality ( The ROC test was employed to measure the sensitivity and specificity of admission NPAR with an AUC of 0.6421 (95% CI 0.6016-0.6826, p < 0.0001). Then the AUC area of admission NPAR was compared with SAPS II and SOFA score. There was no difference between NPAR and SOFA. Thus, it ascertained the quality of NPAR as a reliable predictor of 365-day all-cause mortality (Fig. 3).
Then admission NPAR was integrate into SAPS II and SOFA score. The ROC test was performed on the two combined models, and the results obtained were compared with the original models. For a more comprehensive evaluation of the effects of the new models, integrated discrimination improvement (IDI) was introduced. According to Table 4, after adding NPAR, the AUC of the combined models of SOFA score was increased to 0.714 (p < 0.001) and the integrated improvement was 4.72%. However, the discrimination between SAPS II and its integrated model is not obvious.

Subgroup analysis
In most subgroups, no significant interaction between admission NPAR and 180-day all-cause mortality was observed (Table 5). Patients with high values of heart rate SAPS II, ICU LOS and age had higher risks of all-cause mortality for high admission NPAR.

Discussion
The study indicated that admission NPAR was an independent factor of 30-day, 90-day, 180-day, and 365-day all-cause mortality in critically ill patients with AMI, despite adjustment for age, race and gender. However, after adjustments for more potential confounders, admission NPAR was merely associated with 180-day, and 365day all-cause mortality. Based on ROC curves, NPAR had a moderate ability to predict all-cause mortality of COPD chronic obstructive pulmonary disease, PTCA/PCI percutaneous transluminal coronary angioplasty/percutaneous coronary intervention, WBC white blood cell, PT prothrombin time, RDW red cell distribution width, BUN blood urea nitrogen, ALT alanine aminotransferase, AST aspartate aminotransferase, CK creatine kinase, CK-MB MB isoenzyme of creatine kinase, SAPS II simplified acute physiology score II, SOFA sequential organ failure assessment score, ACEI/ARBs angiotensin-converting enzyme inhibitor/angiotensin receptor blockers  critically ill patients with AMI. Furthermore, subgroup analysis illustrated that there was no evident interaction in most subgroups. Inflammation is a hallmark of atherosclerosis where immune cells, especially monocytes and white blood cells, together with cytokines and phospholipids contribute to trigger the inflammatory reaction [5,15]. The pervious study indicated that neutrophils, as an important member of white blood cells, played a significant role in coronary atherosclerosis and the formation of AMI [16,17]. Meissner et al. [18] found that high neutrophil count was associated with increased risk for AMI in patients presenting to the emergence department with chest pain. The findings hinted that neutrophil accumulation was a strong predictor of all-cause mortality in patients with AMI [18,19].
All the time, albumin has been considered as an indicator of nutritional status. However, the current study  showed that there was a certain correlation between albumin and inflammation [7,20,21]. Serum albumin levels had been affirmed to be inversely related to occurrence of ischemic heart disease [21]. Low serum albumin level was an independent predictor of in-hospital mortality in patients with acute coronary syndrome (ACS) [7,22]. Furthermore, a significant interaction was found between low serum albumin level and first incident AMI [23] and there was a evident correlation between low serum albumin level long-term mortality in patients with STEMI undergoing PCI [24] and patients with unstable angina pectoris (UAP) or non-ST elevation myocardial infarction (NSTEMI) [25].
The previous studies has demonstrated the prognosis value of NPAR in other clinical events including severe sepsis or septic shock [26], acute kidney injury [27] and cardiogenic shock [28]. Sun et al. [11] documented that NPAR was an independent risk factor in critically ill patients with CAD. In term of the above study, we were curious about that NPAR would have same outcomes in AMI as it did in chronic CAD. The previous study identified that admission NPAR was an independent predictor of in-hospital mortality in patients with STEMI. Based on this, we investigated the impact of admission NPAR on the short-term and long-term risk of death in AMI patients. Our study showed that higher admission NPAR was only associated with increased risks of 180-day and 365-day all-cause mortality and may be an independent marker for long-term all-cause mortality in critically ill patients with AMI. According to the AUC area, admission NPAR had a moderate predictive ability in critically ill patients with AMI.
Compared with SOFA and SAPS II, the capacity of admission NPAR in predicting the risk of death was as well as SOFA though its effectiveness was not so adequate than SAPS II. However, by comparing SOFA score with the integrated model, we could speculate Adjust II model adjust for: Age; gender; race; respiratory rate; MBP; heart rate; ALT; AST; CK-peak; CK-MB-peak; glucose; PT; hemoglobin; RDW; creatinine; potassium; sodium; BUN; WBC; Platelet; CAD; AF; COPD; hypertension; diabetes; prior MI; CHF; CKD; STROKE; SOFA; SAPS II (a) (b) Fig. 3 ROC curves for the prediction of 365-day all-cause mortality. a About NPAR. b About NPAR, SAPS II score, and SOFA score. NPAR neutrophil percentage-albumin ratio, SAPS II simplified acute physiology score II, SOFA sequential organ failure assessment score   Heart rate (beats/minute) 0.0023 that admission NPAR was added to SOFA score may played a guiding role in predicting the all-cause mortality.
On account of the test methods for neutrophil percentage and albumin are economical and practical, admission NPAR would permit it possible to quickly evaluate the risk of death in critically ill patients with AMI. Especially, it has precious clinical utility for areas with underdeveloped economy and poor medical conditions.

Limitation
This study was a single-center retrospective study, the selection bias was inevitable. All data came from a publicly open clinical database, so it was difficult to extract some important variables, such as the history of smoking and drinking. Since both neutrophil percentage and albumin change dynamically, the study just chose admission NPAR measured at the first time after admission.
Random error maybe inevitable. Due to missing values of more than 20%, it was difficult to obtain information about some important clinical or laboratory variables.

Conclusion
Our study suggests that higher admission NPAR was independently associated with 180-day and 365-day allcause mortality in critically ill patients with AMI. NPAR may be a clinical maker to predict risk stratification in patients with AMI and further to offer the Individualized treatment services.  HR hazard ratio, CI confidence interval, AF atrial fibrillation, CHF chronic heart failure, CDK chronic kidney disease, COPD chronic obstructive pulmonary disease, CAD coronary artery disease, prior MI prior myocardial infarction, ACEI/ARBs angiotensin-converting enzyme inhibitor/angiotensin receptor blockers, PTCA/PCI percutaneous transluminal coronary angioplasty/percutaneous coronary intervention, ALT alanine aminotransferase, AST aspartate aminotransferase, CK creatine kinase, CK-MB MB isoenzyme of creatine kinase, PT prothrombin time, RDW red cell distribution width, BUN blood urea nitrogen, WBC white blood cell, MBP mean blood pressure, ICU LOS length of ICU stay, SAPS II simplified acute physiology score II, SOFA sequential organ failure assessment score