Arterio-venous differences in glutamate concentration predict clinical outcomes in acute respiratory distress syndrome

We aimed to explore whether the arterio-venous (A-V) differences in amino acids are useful for diagnosing and predicting acute respiratory distress syndrome (ARDS) outcomes, thereby guiding clinical therapy. The enrolled subjects included 36 adult patients with ARDS admitted in the intensive care unit (ICU) and 38 age- and sex-matched controls. Patients with ARDS were further divided into mild, moderate, and severe categories, and into survivor and non-survivor subgroups. Serum samples from the radial artery and jugular vein were collected on the rst day following ICU admission, and the serum concentrations of 15 amino acids from the arterial and venous blood were measured by high performance liquid chromatography (HPLC) on the LC-10DVP system. in were in other amino acids between controls ARDS, surprisingly, greater A-V differences in Glu concentration than non-survivors. Particularly, non-survivors who died within 2 weeks following ICU admission showed signicantly less A-V differences in Glu concentration compared to those of survivors.


Introduction
Acute respiratory distress syndrome (ARDS), characterized by tachypnea, severe hypoxemia, and decreased respiratory compliance, is a common cause of respiratory failure [1]. Despite signi cant improvements in its clinical management and treatment, about 10% of all patients still require intensive care unit (ICU) admission worldwide, and the mortality rate remains high, at 30-40%, in most studies [1]. Currently, ARDS diagnosis is based on "the Berlin de nition," which includes clinical criteria. According to the oxygenation level (PaO 2 /FiO 2 ), the degree of severity of patients with ARDS are diagnosed as mild (PaO 2 /FiO 2 : 200-300 mm Hg), moderate (PaO 2 /FiO 2 : 100-200 mm Hg), and severe (PaO 2 /FiO 2 ≤ 100 mm Hg) ARDS [2]. Even after the development and improvement of the Berlin diagnostic criteria, it is still a challenge to accurately diagnose ARDS or predict its prognosis. The best prediction method may be to combine clinical predictors with a variety of biomarkers [3]. Thus far, a large number of studies have found several biomarkers with the highest sensitivity and speci city useful in the diagnosis and prognosis of ARDS [4]. However, no biomarker has yet been con rmed to diagnose or predict the outcome of ARDS, and none of those found has su cient reliability for the clinical diagnosis of ARDS or the prediction of its prognosis. The lack of a speci c biomarker for ARDS is undoubtedly one of the most important obstacles in the development of new treatment methods or strategies for ARDS. Consequently, researchers have called for the identi cation of speci c markers that could lead to the early diagnosis of ARDS and accurately predict its outcome.
Blood is the main carrier of metabolites, including various nutrients (lipids and amino acids), hormones, and organic waste [5]. The circulatory system of the human body includes the arterial blood that delivers nutrition to the tissues and the venous blood that clears the byproducts of metabolism [5]. The blood circulation in the arteries and veins forms the basic part of the human anatomy, which plays an important role in maintaining the internal balance in all tissue stroma [5]. It is easier to obtain the venous blood, which can re ect the metabolism in speci c organs[6]. However, the arterial blood, as a uid, better re ects the average metabolic state of the body, before it is processed by any particular organ [7].
Moreover, the pulmonary blood ow (PBF) constitutes the entire output of the right ventricle and supplies the lung with mixed venous blood draining all the tissues of the body[8]. To some extent, the venous blood draining the whole body changes to arterial blood after gas and substance exchange in the alveoli in pulmonary capillaries. Therefore, comparing the differences between the venous blood ow into the lung and the arterial blood out of the lung will help to reveal how substances are released or consumed in the lung tissue. Generally, there is a difference in the metabolites in the arterial and venous blood [9][10][11][12][13]. However, the comprehensive biochemical characteristics of the difference between arterial and venous blood have not been reported. In critically ill patients like those with ARDS, either arterial or venous blood is usually available for sampling and measurement. Therefore, the question regarding the arterio-venous differences (A-V differences) in ARDS patients can be addressed.
Amino acids play an important role in the physiological and biochemical processes involved in the development of ARDS [14]. However, previous studies have not determined the exact relationship between amino acid levels and ARDS severity and outcomes. Hence, we performed global pro ling of arterial and venous blood to assess the arterio-venous differences (A-V differences) in amino acids between control and ARDS patients. Therefore, we aimed to determine whether the arterio-venous differences in amino acids are useful predictors of ARDS diagnosis and outcome. (2) bilateral pulmonary in ltrates consistent with pulmonary edema; (3) impaired oxygenation PaO2/FIO2 ratio < 300 mm Hg; (4) Impaired oxygenation not fully explained by cardiac failure; these ARDS patients were also strati ed into mild (7 cases), moderate (8 cases), and severe (21 cases) groups according to the Berlin de nition [2]. The exclusion criteria were as follows: (1)  (2)Amino acid determination Blood samples for the determination of serum amino acids were collected from the radial artery and jugular vein of control and ARDS patients within 24 h after admission. The blood was centrifuged at 3,000 rpm for 15 min. The supernatants were stored at -80°C prior to analysis. The concentrations of 15 different kinds of amino acids (phenylalanine, alanine, methionine, glycine, glutamate, arginine, lysine, tyrosine, leucine, serine, threonine, aspartate, valine, isoleucine, and histidine) in serum were measured by high performance liquid chromatography (HPLC) on the LC-10DVP system (Shimadzu, Kyoto, Japan).

(3)Data collection and statistical analysis
All data were entered into the Microsoft Excel worksheet and statistical analyses were performed using SPSS version 23.0 software (IBM Corp., Armonk, N.Y., USA). The results for continuous variables with normal distributions were reported as the means ± standard deviations (SD). Qualitative variables were analyzed using the Chi-squared test, and quantitative variables were compared using the Student's t-test after being tested for normality with Levene's test. A two-sided p value < 0.05 was considered statistically signi cant.

Results
(1)General information Between August 2018 and May 2019, a total of 36 patients with ARDS and 38 controls patients were selected following the relevant criteria. The ARDS group was further divided into mild (7 cases), moderate (8 cases), and severe ARDS (21 cases) subgroups, as well as into survivor (19 cases) and non-survivor (17 cases) subgroups. The non-survivor subgroup was again divided into 2 weeks (9 cases) and 2 weeks (8 cases) death subgroups (Fig. 1). As shown in Table 1, no signi cant differences were detected in age, sex, Hb level, and serum creatinine level between the controls and ARDS groups. WBC count, serum CRP, PCT, BUN, AST, ALT, TB, and DB levels were higher in the ARDS group than in the control group (p < 0.05). Additionally, in the ARDS group, the PaO2/FiO2, APACHE II, and SOFA scores were 123.5 ± 51.0; 18.6 ± 4.8; and 9.0 ± 2.5, respectively. (2)The A-V differences in amino acid pro les between control and ARDS patients on ICU admission We detected the levels of 15 different kinds of amino acids in the arterial and venous serum of control and ARDS patients. As shown in Table 2, there were no differences in the levels of all the 15 different kinds of amino acids detected in the arterial serum between control and ARDS patients. The level of phenylalanine (p = 0.0495) and methionine (p = 0.0396) was increased in the venous serum of ARDS patients, while the levels of the remaining 13 amino acids in the venous serum showed no signi cant differences between control and ARDS patients. Since differences in amino acid levels between the venous and arterial blood could help to reveal how substances are released or consumed in the lung tissue, we further compared the A-V differences between control and ARDS patients. Table 3 shows that the A-V differences in Glu concentration in ARDS patients were signi cantly higher than those in controls (p = 0.002). We did not nd any signi cant changes in the A-V differences of the remaining amino acid concentrations, comparing between control and ARDS patients. Since the venous blood draining the whole-body changes to the arterial blood after gas and substance exchange in the alveolar pulmonary capillaries, comparing the differences between venous and arterial blood could help to reveal how substances are released or consumed in the lung tissue. In ARDS patients, only the A-V differences in Glu concentration was greater when compared with controls.

(3)Correlations between A-V differences in Glu concentration and the severity of clinical outcomes of ARDS patients
ARDS patients were divided into mild, moderate, and severe groups according to the Berlin de nition. We found no signi cant changes in A-V differences in Glu concentration between mild, moderate, and severe patient groups ( Fig. 2A). To explore the relationship between A-V differences in Glu concentration and outcome of patients, we compared the A-V differences in Glu concentration between survivors and nonsurvivors based on the 28-day survival period. A-V differences in Glu concentration in survivors (A-V = 0.615 ± 0.391, p < 0.001) and non-survivors (A-V = 0.3378 ± 0.306, p = 0.026) were both signi cantly higher than in control patients (A-V= -0.011 ± 0.581) (Fig. 2B). Surprisingly, we found that the A-V difference in Glu concentration in non-survivors was much lower than that in survivors (p = 0.029) (Fig. 2B). Further, to determine the relationship between Glu A-V differences and outcome of patients, we compared Glu A-V differences between survivors and non-survivors (patients who died within 2 weeks or after 2 weeks of ICU admission). Surprisingly, we found that patients who died within 2 weeks of ICU admission (A-V = 0.143 ± 0.259) had much lower Glu A-V differences than survivors (A-V = 0.615 ± 0.391, p = 0.020) (Fig. 2C). However, there was no signi cant difference between survivors and non-survivors who died after 2 weeks of ICU admission. Our results showed that ARDS patients had a greater A-V difference in Glu concentration than controls, whereas in non-survivors, the difference was not as high as that in survivors. Thus, the A-V difference in Glu concentration showed a signi cant association with mortality in ARDS patients.

Discussion
Using HPLC, our study rst showed the A-V differences in amino acid concentrations in controls and ARDS patients. In the 15 different amino acids that we detected in the arterial and venous blood, the A-V differences in only Glu concentration were signi cantly greater in ARDS patients than in controls. However, A-V differences in Glu concentration in ARDS patients showed no signi cant correlation with disease severity (grouped according to the ARDS Berlin de nition). Furthermore, the result revealed that A-V differences in Glu concentration in ARDS survivors were considerably higher than in non-survivors, especially in non-survivors who died within 2 weeks of ICU admission. These results indicate that A-V differences in Glu concentration may be a biomarker for ARDS patients, especially to predict the outcome of ARDS.
Clinically, acute lung injury (ALI)/ARDS manifests as decreased lung compliance, severe hypoxemia, and bilateral pulmonary in ltrates, and is associated with a high mortality [1]. Diagnosis of ARDS are mainly based on clinical characterization. Numerous studies have focused on the identi cation of biomarkers, which can re ect pathophysiological mechanisms, to improve diagnosis, determining disease severity and prediction for outcome of ARDS. Researches already identi ed several biomarkers that are associated with the highest sensitivity and speci city for the diagnosis or outcome prediction of ARDS that have been investigated in blood, pulmonary edema uid, and exhaled air [4]. In general, these are cellspeci c for epithelial or endothelial injury or involved in the in ammatory or infectious response. But currently they are not reliable enough for clinical diagnosis of ARDS or prediction of its prognosis or have not yet been con rmed [4]. The lack of a speci c biomarker to de ne, diagnose, monitor responsiveness to therapy or predict prognosis of ARDS has limited progress in developing novel treatments for ARDS. Thus, the discovery of a speci c biological marker with early diagnostic signi cance is desirable.
Currently, it is an emerging area to identify metabolites including amino acid as clinically relevant biomarkers and potential therapeutic targets in several diseases, such as sepsis, diabetes, cancers and so on [14][15][16][17][18]. One study showed elevated levels of several metabolites, such as hippurate, L-phenylalanine, creatine, methionine, L-glutamate, and L-proline in bronchoalveolar lavage uid (BALF) of patients with sepsis-induced ARDS compared to healthy controls [19]. Another study also showed that amino acid levels in the plasma differed in patients with acute exacerbation of COPD (AE-COPD), depending on the presence of bacterial infection. They claimed that speci c amino acids (i.e., asparagine, citrulline, glutamine, histidine, serine, and threonine) have a potential utility as diagnostic markers to distinguish between bacterial and nonbacterial AE-COPD [20]. In another study, signi cant differences in the metabolomics and metabolic pathways between control and ARDS groups were observed. This study suggests the potential utility of metabolomics to identify biomarkers that predict early ARDS onset, progression, severity, and prognosis [21] Another study focused on the plasma amino acid levels during the acute phase of endotoxin-induced lung injury in eight sheep, they found that norepinephrine, epinephrine, and alanine levels increased whereas that of isoleucine decreased. However, they did not determine the relationship between amino acid levels and ARDS severity and outcomes [14] Till now, the dynamics and clinically signi cant changes in amino acid levels in patients with ARDS are largely unknown.
To the best of our knowledge, to date, A-V differences in amino acid levels have not been extensively studied. In our study, we detected the A-V differences in amino acid levels between controls and ARDS patients. We found that A-V differences in Glu concentration in ARDS patients were signi cantly greater than in controls; however, the differences were not signi cant for other amino acids we study. Importantly, survivors showed greater A-V differences in Glu concentration than non-survivors, while non-survivors that died within 2 weeks following ICU admission showed signi cantly smaller A-V differences in Glu concentration than survivors. These results indicate that the A-V differences in Glu concentration can be a biological marker for ARDS and for outcome prediction. We tried to understand the higher A-V differences in Glu concentration in ARDS patients than in non-survivors. A-V difference in Glu concentration represents the arteriovenous difference in the levels of Glu (arterial level of Glu-venous level of Glu). Generally, venous blood becomes arterial blood in the lungs. Thus, to some extent, the A-V differences in substances can reveal how substances are released or consumed in the lung tissue. Our previous research detected 17 kinds of amino acids (phenylalanine, alanine, methionine, glycine, glutamate, arginine, lysine, tyrosine, leucine, serine, threonine, aspartate, valine, isoleucine, and histidine) in BALF of mice. The results showed that only Glu and glycine were increased in BALF of BLM-induced lung injury mice compared to the saline control group [22]. This research indicated that endogenous Glu and glycine may be selectively released from the lungs at the beginning of BLM-induced acute lung injury. If Glu is released from the injured lungs, it will also increase the Glu level in the arteries. Thus, we hypothesized that when the lung gets injured, cells selectively release Glu from the lungs to the blood or alveolar, which is consistent with the increased level of Glu in BALF of injured lungs as well as in arteries. System x(c)-is a cysteine/Glu transporter, with two subunits-4F2hc and the function unit of x (c)-system (xCT). At physiological conditions, system x(c)-mediates the release of Glu while it uptakes cysteine, which is essential for maintaining intracellular glutathione levels and cellular defenses against oxidative stress [23]. Previous researches suggested that xCT is upregulated in injured lung tissues induced by paraquat and plays a protective role against oxidative stress and lung injury [24]. As we detected in this study, ARDS patients had higher A-V differences in Glu concentration while that in non-survivors was much lower than for the survivors. We hypothesized that System x(c)-is upregulated in injured lung tissue to promote the uptake of cysteine and the release of Glu. This is important for the synthesis of glutathione, protecting the patients from oxidative stress-induced injury. System x(c)-function in nonsurvivors may be damaged, resulting in decreased release of Glu from the lung tissue, leading to the lower level of Glu in the arterial blood; which is consistent with the lower A-V differences in Glu concentration in non-survivors. A-V differences in Glu concentration in non-survivors were much lower than those of survivors. This may indicate a reduction in the uptake of cysteine and glutathione synthesis, causing insu cient capacity of antioxidants in the lung cells of non-survivors.
Our study has several limitations that should be addressed. First, the results were obtained from a study population with a relatively small sample size; hence, the results will need to be con rmed using a larger sample size. Second, due to the limitation of clinical sampling, it was really di cult to obtain pulmonary arterial and venous blood. The blood from the radial artery and jugular vein that we collected could not fully represent the blood owing in and out of the lung completely; thus, our results may not have fully determined the relationship between lung injury and A-V differences in Glu concentration. Third, we hypothesized that System x(c)-plays an essential role in the A-V differences in Glu concentration between controls and ARDS patients. However, we did not detect the level of xCT and glutathione in the patients. Further research is needed to clarify the mechanism and signi cance of releasing Glu from the lung. Furthermore, we did not show the relationship between long-term consequences in ARDS survivors and A-V differences in amino acid levels.

Conclusion
This is the rst study to demonstrate that the A-V differences in Glu concentration in ARDS patients are signi cantly greater than those in controls. These differences can be used as a biological marker to predict the ARDS outcome. This can facilitate improved ARDS management and outcome prediction.
Moreover, we hypothesized that greater A-V differences in Glu concentration in patients is an important biomarker which can indicate an endogenous protective factor preventing mortality or prolonging survival. The decreased A-V differences in Glu concentration in patients with poor prognosis may result from damaged System x(c)-or decreased xCT level in the lung, which needs further investigation.