Association between Changes in Plasma Metabolism and Clinical Outcomes of Sepsis

Current prognostic biomarkers for sepsis have limited sensitivity and specificity. This study aimed to investigate dynamic lipid metabolomics and their association with septic immune response and clinical outcomes of sepsis. This prospective cohort study included patients with sepsis who met the Sepsis 3.0 criteria. On hospitalization days 1 (D1) and 7 (D7), plasma samples were collected, and patients underwent liquid chromatography with tandem mass spectrometry. A total of 40 patients were enrolled in the study, 24 (60%) of whom were men. The median age of the enrolled patients was 81 (68–84) years. Thirty-one (77.5%) patients had a primary infection site of the lung. Participants were allocated to the survivor (25 cases) and nonsurvivor (15 cases) groups based on their 28-day survival status. Ultimately, a total of 113 lipids were detected in plasma samples on D 1 and D 7, of which 42 lipids were most abundant in plasma samples. The nonsurvival group had significantly lower lipid expression levels in lysophosphatidylcholine (LysoPC) (16 : 0, 17 : 0,18 : 0) and 18 : 1 SM than those in the survival group (p <  0.05) on D7–D1. The correlation analysis showed that D7–D1 16 : 0 LysoPC (r = 0.367, p = 0.036),17 : 0 LysoPC (r = 0.389, p = 0.025) and 18 : 0 LysoPC(r = 0.472, p = 0.006) levels were positively correlated with the percentage of CD3+ T cell in the D7–D1. Plasma LysoPC and SM changes may serve as prognostic biomarkers for sepsis, and lipid metabolism may play a role in septic immune disturbances.


Introduction
Sepsis is an infection-induced systemic infammatory response involving multiple mediators and cytokines. It is a commonly encountered life-threatening condition in intensive care units (ICUs) worldwide [1]. Despite medical advances, sepsis remains a serious global health problem with an estimated 48.9 million cases and 11 million deaths annually [2]. Martin et al. [3] found that over a period of 24 years (1979-2002), the rate of increase in sepsis incidence was 20.4% higher in those aged ≥65 years than that in their counterparts. Te sepsis risk in 65-year-old individuals grew by an average of 9.5% per year over 24 years, and the incidence rate for people aged 65 years and older increased by 11.5% annually. Currently, the Acute Physiology and Chronic Health Evaluation (APACHE) II score is commonly used in the ICU to assess the severity of hospitalized patients [4]. However, this method has some limitations and fails to refect prognosis in patients whose disease peak is not observed within the frst 24 hours after admission.
Research interest in the association between lipid metabolism and sepsis is increasing [5,6]. Recent studies suggest that poor prognosis is related to increased free fatty acid levels, changes in polyunsaturated fatty acid metabolism [7], decreased lysophosphatidylcholine (LysoPC) levels [8], and elevated plasma ceramide levels [9]. In addition, Park et al. [10] conducted a serum lipid analysis of 74 ICU patients and found that LysoPC levels in the survival group were higher than those in the mortality group and that LysoPC monitoring helped predict the 28-day mortality rate of patients with sepsis and septic shock. Tese studies suggest that lipidomic analysis may help develop prognostic markers for sepsis.
To further examine the role of lipid metabolism in sepsis, this study aimed to investigate dynamic lipid metabolomics and their association with T cells and the septic immune response and clinical outcomes of sepsis.

Study Design and Period.
Tis prospective cohort study was conducted at the medical ICU (MICU) of a universityafliated urban teaching hospital from December 2019 to February 2022. Te study protocol was approved by the medical science research ethics committee of our hospital (approval no. M2019396). All patients or their legally authorized representatives provided written informed consent to participate in the study.

Study Population.
A total of 40 patients with sepsis admitted to the MICU were included and subsequently divided into the survivor and nonsurvivor groups based on their 28-day survival status. Te inclusion criteria were as follows: (1) patients who met the Sepsis 3.0 diagnostic criteria [11] and (2) were aged ≥18 years. Te exclusion criteria were as follows: (1) pregnancy, (2) human immunodefciency virus or tuberculosis infection, and (3) a "do-not-resuscitate" or discontinuation of tracheal intubation order.

Clinical and Laboratory
Data. Data on the following variables were collected: age, sex, body weight, height, body mass index (BMI), days of mechanical ventilation, and length of stay at discharge or death. Finally, plasma (4 mL) samples were collected on day 1 (D1) and day 7 (D7) of hospitalization. Te collected samples were aliquoted into multiple tubes to avoid repeated thawing and freezing. Samples were stored in a −80°C freezer until use.
Samples stored at −80°C were thawed in a 4°C refrigerator. Subsequently, a 10 µL sample aliquot was sequentially supplemented with 10 µL of an internal standard mix solution, 10 µL of water, and 100 µL of chloroform: methanol (2 : 1, v/v) extracting solution. Te resulting mixture was vortexed for 20 s, placed in a 4°C refrigerator for 30 min, and then centrifuged at 7,800 × g for 3 min. Subsequently, the supernatant was collected using a 1-mL syringe and placed into a 0.5-mL EP tube. Tereafter, 40 µL of the supernatant was immediately placed into a clean EP tube and evaporated until dry via nitrogen gas blowdown. Te tube was sealed and stored in a −20°C freezer until use. Before injection, the sample was dissolved in 50 µL of acetonitrile and isopropyl alcohol (1 : 1, v/v) and vortexed for 60 s.

Statistical Analyses.
Clinical data were compared between the survivor and nonsurvivor groups. SPSS software v.21.0 (IBM Corp., Armonk, NY, USA) was used for statistical analysis. Continuous variables are expressed as medians (range), while categorical variables are expressed as frequencies (%). Continuous variables were compared using the nonparametric Mann-Whitney U-test, and categorical variables were compared using the χ 2 test. Te degree of correlation was expressed using Pearson's correlation coefcient "r." A p < 0.05 indicated statistical signifcance.

Patient Characteristics.
We included 40 patients with sepsis, 24 (60%) of whom were men. Participants were allocated to the survivor (n � 25) and nonsurvivor (n � 15) groups based on their 28-day survival status. Te median age of the enrolled patients was 81 (68-84) years. Tirty-one (77.5%) patients had a pulmonary infection. Table 1 shows the detailed characteristics of the participants.

Lipidomics Analysis.
In total, 113 lipids were detected in plasma samples on day 1 and day 7, of which 42 lipids were most abundant in plasma samples (Table S1 shows Rt and m/z of 42 kinds of lipids). Table 2 shows all lipid results in D1, D7, and D7-D1 survival and nonsurvival groups. We found that the diference in lipid concentration between groups was signifcant for D7-D1. Figure 1 shows the scatter plot of D7-D1(changes between D7 and D1) diferential metabolites between the survival and nonsurvival groups.

Relationship between T Lymphocyte Subsets and Lipids.
In order to reveal the correlation between changes of lipid metabolism with lymphocyte disturbance in sepsis, we analyzed the lymphocyte subsets on D1 and D7 in survivor and nonsurvivor groups, and found that there were signifcant diferences in CD3 + T lymphocytes in the blood on D1 and D7 in both survivor and nonsurvivor groups (Table 3). According to the results in Table 2, the between-group diference in plasma lipid levels was greater for D7-D1 than for D1 or D7. We analyzed the correlation between D7-D1 plasma lipid levels and D7-D1 lymphocyte subsets, and found that D7-D1 plasma 16 : 0 LysoPC (r � 0.367, p � 0.036),17 : 0 LysoPC(r � 0.389, p � 0.025) and 18 : 0 LysoPC(r � 0.472, p � 0.006) levels were positively correlated with the percentage of CD3 + T cell in the D7-D1 ( Figure 2).

Discussion
In this study, we investigated diferences in lipid metabolism between surviving and nonsurviving patients with sepsis based on their 28-day survival status. Tere are 42 lipids in plasma, including LysoPC PC, PE, LysoPE LysoPI, SM, and ceramide. Scatter plots were used to represent the abovegiven diferential metabolites (Figure 1), and we found that LysoPC and SM of the D7-D1 survival group were higher than those of the nonsurvival group, and the diferences were statistically signifcant.
LysoPC, a lipid mediator and major phospholipid component, is involved in immune cell recruitment, stimulation, and infection [12][13][14]. Kamisoglu et al. [9] reported that fve LysoPC type levels were signifcantly lower in patients with sepsis than in healthy controls. In our study, the LysoPC concentrations in the nonsurvivor group decreased signifcantly by D7. Drobnik et al. [12] demonstrated that the LysoPC level was signifcantly lower in patients with sepsis than those in healthy people. Moreover, samples obtained from patients with sepsis on days 4 and 7 revealed an association between LysoPC concentrations and sepsisrelated mortality. Park et al. [10] noted that serial measurements of LysoPC concentrations predicted 28-day mortality in ICU patients with severe sepsis and septic shock.
Te results of this study were largely consistent with those of previous studies. However, previous studies only analyzed on D1 and D7, and our study emphasized the signifcance of dynamic monitoring. Repeated measurements indicated that the decrease in LysoPC concentration was more evident in the nonsurvivor than that in the survivor group. Tis decrease may be attributed to the reduced expression of lipopolysaccharide-induced tissue factor in monocytes owing to changes in LysoPC [15], which can attenuate the release of proinfammatory cytokines and promote the release of anti-infammatory cytokines [16]. Te persistently decreased LysoPC levels in patients with sepsis result primarily from disruptions to metabolic homeostasis [17]. In our study, the decreased LysoPC levels in the nonsurvivor group were associated with sepsis-induced    immune overreaction, possibly leading to LysoPC consumption. A study [18] using 1 H nuclear magnetic resonance to evaluate sepsis-induced acute lung injury (ALI) in 13 patients reported that sphingolipid levels were lower in patients with sepsis-induced ALI than in healthy controls. Although the specifc mechanisms remain unclear, our results indicate that LysoPC could be a prognostic sepsis biomarker. SM lipids, including sphingosines, fatty acids, phosphoric acid, and nitrogenous bases, are major membrane constituents [19]. Moreover, SM lipids and their metabolites are essential signaling molecules that regulate key signal transduction processes, including cell growth, diferentiation, senescence, and apoptosis [20]. Previous studies have reported that SM is involved in sepsis-induced lung injury [21,22]. Mecatti et al. [5] performed a lipidomics analysis of plasma and erythrocytes in 20 patients with sepsis, and lower levels of SM, PC, and LysoPC were found in these patients than in healthy controls. Arshad et al. [23] also observed low SM, PC, and LysoPC levels and high acid sphingomyelinase (aSMAse) activity in patients with community-acquired pneumonia, with lipid levels gradually returning to the normal range with clinical improvement. Sepsis involves increases in aSMAse plasma levels and activity, which are correlated with disease severity [24]. Te aforementioned studies suggest an association between SM and sepsis severity, possibly due to SM involvement in the infammatory response. Similarly, we observed a decrease in plasma SM levels in patients with sepsis, which was more apparent in the nonsurvivor than in the survivor group. Tis diference may be attributed to SM hydrolysis to ceramide, which exerts anti-infammatory efects by inhibiting reactive oxygen species, mitogen-activated protein kinases, phosphatidylinositol-3-kinase, protein kinase B, Janus kinase/signal transducer, and activators of transcription pathways, while upregulating protein kinase A and heme oxygenase-1 expression [25].
Sepsis is a host response disorder to infection that leads to life-threatening organ dysfunction [11] and refects the complex response of the host immune system to pathogens [26]. During sepsis, CD4 + T cells are activated in response to antigen presentation by dendritic cells or monocytes, releasing immunomodulators and coordinating cytotoxic CD8 + T cells [27]. Moreover, low lymphocyte levels and nonrecovery are associated with sepsis mortality risk [28]. LysoPC is produced by the action of phospholipase A2 on PC and promotes infammatory efects, including endothelial cell adhesion, monocyte chemotactic activation, and    Emergency Medicine International macrophage activation [29]. Lin et al. [30] demonstrated that the most abundant plasma LysoPC species (16 : 0, 18 : 0, and 18 : 1) inhibited the production of reactive oxygen species and activation of neutrophils, and LysoPC prevented neutrophil-mediated pulmonary vascular injury in an in vitro lung perfusion model. Asaoka et al. [31] found that LysoPC specifcally enhanced T lymphocyte activation. Furthermore, LysoPC enhances interferon-c secretion, activates CD4 + and CD8 + T cells, and increases CD40L and C-X-C chemokine receptor type 4 expression in CD4 + T cells [32]. In addition, Ni et al. [33] observed that plasmalogen lysophosphatidylethanolamine was a type of autoantigen that stimulated natural killer T cell production and activation. In our preliminary analysis, the D7-D1 LysoPC (16 : 0, 17 : 0,18 : 0) plasma concentration was positively correlated with the percentage of CD3 + T cell in the D7-D1. Tese results suggest that lipids play a role in the disturbance of lymphocyte homeostasis during sepsis, although further investigation is required. Te aging process is characterized by decreased immune function and increased stress response. Oxidative and metabolic stress may lead to changes in sphingolipid metabolism and increase the risk of age-related diseases. However, an association was reported between age and 18: 0-22 : 6 PC lipids in women and between age and eicosapentaenoic acid-containing lipids (e.g., 16:0-20 : 5 PC and 20 : 5 cholesteryl ester) and 18:1-22 : 0 ceramide in men; however, no other lipid values were correlated with age [34]. A study involving 800 healthy volunteers reported no correlation between total LysoPC levels and age [35]. Another study reported that the in-hospital mortality risk of sepsis was signifcantly associated with advanced age [36]; furthermore, the survivor group had higher BMIs than the nonsurvivor group, although the diference was nonsignifcant. While obesity is associated with an infammatory response, the underlying mechanisms remain unclear. High BMI was previously reported as a protective factor against sepsis [37]. Recently, an inverse association between LysoPCs (17 : 0, 18 : 1, and 18 : 2) and BMI was reported [38]. Bagheri et al. demonstrated that LysoPC (18 : 1 and 18 : 2) values, although not SM values, were negatively associated with obesity in Iranian adults [39]. No such correlations were observed in our study, likely because of diferences in the study populations. Further studies are required to clarify whether sepsis diferentially afects lipid metabolism in patients with and without obesity.
Tis study had limitations. It did not include a healthy control group since previous studies have repeatedly demonstrated lipid variations in patients with sepsis. Moreover, given the lack of an external validation cohort and small sample size, our fndings require validation.

Conclusions
In conclusion, our fndings indicate that repeated monitoring of lipids in plasma LysoPC and SM may help predict the prognosis of patients with sepsis. Furthermore, the correlation between changes in lipids and lymphocyte subsets suggests that lipid metabolism plays a role in the immune disturbance that occurs during sepsis.

Data Availability
Te datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.