High-density lipoprotein-cholesterol and triglyceride levels and their prognosis in critically ill septic patients

Abstract

BACKGROUND

Lipoproteins, particularly high-density lipoproteins (HDLs), have the ability to bind to lipopolysaccharides, the primary component of the extracellular membrane of gram-negative bacteria, as well as to lipoteichoic acid, an immunostimulant derived from gram-positive bacteria. Triglyceride (TG) levels are also indicators of acute inflammation. Altered levels of both are prevalent in sepsis, rendering them potential biomarkers that are cost-effective and readily applicable in clinical practice for monitoring these patients. It is crucial to compare the levels of these molecules with those of more traditional and costly biomarkers.

AIM

To evaluate the association between HDL and TG levels in the diagnosis of sepsis (HDL/TG day 1), as well as the association between the ratio of HDL and TG levels on days 3 and day 1 (HDL/TG ratio) and 28-day mortality in patients with sepsis. We also compared the serum levels of HDL and TG levels with traditional inflammatory markers used in the prognostic assessment of sepsis: Interleukin (IL)-1β, IL-6, and IL-10, measured serially in patients with sepsis.

METHODS

This was a prospective cohort study. Adult patients (> 18 years of age) admitted to the intensive care unit with sepsis and receiving vasopressors were included. We recorded the epidemiological and clinical characteristics, as well as the severity scores, at the time of sepsis diagnosis and on day 3 of sepsis management. We measured HDL 1, TG 1, IL-1β, IL-6, IL-10, and C-reactive protein (CRP) levels upon the diagnosis of sepsis and on day 3 and calculated the HDL and TG ratios. The main outcome was 28-days mortality.

RESULTS

Seventy-five patients had measurements on day 1 (male:female ratio 58%:42%), and 50 patients had measurements on day 3. Patients who died had lower HDL 1 [9 mg/dL (5-17) vs 18.5 mg/dL (8-31), P = 0.02] but did not have lower TG levels on day 1 [142 mg/dL (83-224) vs 148 mg/dL (97-196), P = 0.97]. Additionally, patients who succumbed exhibited a non-statistically significant increase in the HDL ratio [1.26 (1-1.79) vs 1.16 (1.01-1.39), P = 0.29] and a non-statistically significant increase in the TG ratio [1.16 (0.85-1.36) vs 1.02 (0.67-1.36), P = 0.69]. HDL levels on day 1 were negatively associated with IL-10 on day 1 (Pearson’s r = -0.37; P < 0.01) and with CRP on day 1 (Pearson’s r = -0.52; P < 0.01), but not with IL-6 (P = 0.19) or IL-1β (P = 0.62). TG 1 was positively associated with CRP on day 1 (Pearson’s r = 0.57, P < 0.01) but not with IL-6 (P = 0.21), IL-10 (P = 0.12), or IL-1β (P = 0.09).

CONCLUSION

HDL levels are a promising biomarker for the evaluation of patients with sepsis, and their levels correlate with traditional biomarkers in this field, such as CRP and IL-10. Further multicenter studies with larger cohorts should be conducted to confirm or refute this hypothesis.

Key Words: Sepsis; High-density lipoprotein; Triglycerides; Immunologic response; Sequential organ failure assessment; Interleukins; Mitochondrial metabolism

Core Tip: High-density lipoprotein (HDL) and triglyceride (TG) should be considered inflammatory biomarkers in sepsis, yet they exhibit distinct characteristics in the early response to the condition. TG levels correlate with patient severity upon intensive care unit admission, while lower HDL levels at the time of sepsis diagnosis are slightly linked to higher mortality. However, an early improvement in HDL does not correlate with better outcomes. Nonetheless, both biomarkers are associated with levels of significant inflammatory markers in sepsis, such as interleukin-10 and C-reactive protein.

INTRODUCTION

A wide range of biomarkers, measured by a host of different technologies, are being investigated to discriminate a systemic inflammatory response syndrome rapidly, or early identification of infection-triggered organ dysfunction (sepsis)[1]. Biomarkers, especially in sepsis, could help clinicians identify a population with this condition and indicate a dysregulated response, aiding in diagnosis, risk stratification, prognostication, and patient management[2].

Common biomarkers include procalcitonin and C-reactive protein (CRP), which reflect bacterial infection and inflammation, respectively. The rise in CRP levels is primarily induced by interleukin (IL)-6 and IL-1β, which act on the gene responsible for CRP transcription during the acute phase of an inflammatory process, activating the immune response in monocytes[3]. Other important biomarkers include lactate (indicating tissue hypoperfusion and energetic dysfunction) and cytokines such as IL-6, IL-10, and IL-1β, which are involved in the inflammatory cascade and are also associated with the prognosis of this population[4].

The response to sepsis is influenced not only by the inflammatory response but also by the expression of pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns, which function as molecular signals that activate the innate immune system. In the pathophysiology of sepsis, high-density lipoprotein (HDL) possesses the ability to remove various PAMPs, such as lipopolysaccharides and lipoproteic acid, originating from infections by gram-negative or gram-positive bacteria[5]. Moreover, pre-sepsis levels of triglycerides (TG) are correlated with disease severity and mortality in the context of sepsis, indicating that the lipid profile plays a significant role in patient survival[6]. These phenomena are well described in sepsis, especially regarding decreased levels of HDL[5,7]. Additionally, increased hepatic production of TGs occurs due to the availability of free fatty acids, released by lipolysis of peripheral and visceral adipose tissues[8]. This occurs as a result of stimulation by catecholamines, endotoxins, and cytokines, such as tumor necrosis factor (TNF)-α, IL-1, interferon (IFN)-α, IFN-β, and IFN-γ and the synthesis of fatty acids in the liver, which is stimulated by inflammatory cytokines (TNF-α, TNF-β, IL-1, IL-6, and IFN-α)[9,10]. This metabolic and inflammatory response has an important interplay with mitochondrial metabolism, which is especially altered in the mononuclear cells of patients with sepsis[11]. Due to the elevated hepatic production of TGs, there is a reduction in their clearance, contributing to an increase in plasma TGs[8]. Consequently, a potential interaction exists between TG metabolism and mitochondrial impairment[11]. This impairment is most effectively assessed through the biochemical coupling efficiency (BCE), an integrative measure of mitochondrial metabolism that can be readily evaluated in immune cells[12].

HDL interacts with different pathways in the sepsis response. HDL has been shown to bind and neutralize endotoxins, such as lipopolysaccharides and lipoteichoic acid, and inhibit cytokine responses in animal models of sepsis, making it especially beneficial in reducing their ability to trigger an innate immune response[7,13]. Thus, HDL cholesterol levels are significantly decreased in patients who develop organic failure[14,15]. As a potentially relevant biomarker, HDL has a superior predictive ability compared with all routine clinical markers for both the development of multiorganic failure and 28-day mortality[7].

Although the literature indicates a role for both TG and HDL levels as potential biomarkers in sepsis, there are still unresolved gaps regarding them. We do not adequately know the association between lipid profiles and the incidence of outcomes beyond mortality, such as the need for hemodialysis or mechanical ventilation, as well as its association with the improvement or worsening of organ failures throughout the course of sepsis, is not well understood. Furthermore, it is imperative to assess novel potential biomarkers in sepsis beyond their mere correlation with clinical variables, aiming to elucidate their role in the pathophysiological process, as has been demonstrated in other disciplines such as immunology and oncology[16]. Therefore, it is relevant to know the association between serum TG and HDL levels and levels of inflammatory biomarkers, such as CRP, IL-6, or IL-1, or with IL-10 levels, which are associated with a counter-regulatory response to inflammatory activity in sepsis. The main objective of this study was to evaluate the association between HDL and TG levels and clinical and inflammatory outcomes in sepsis.

MATERIALS AND METHODS

We performed a cohort study that prospectively evaluated consecutive patients who had been admitted to four different intensive care unit (ICU) in a tertiary academic hospital (Grupo Hospitalar Conceição, Porto Alegre, Brazil). This study was approved by the local ethics committee (Plataforma Brazil number 66240017.0.0000.5530). Written informed consent was obtained from the patient or their next of kin.

Adult patients (> 18 years of age) admitted to the ICU with sepsis and persistent hypotension were enrolled in this study. Sepsis was defined according to the current guidelines[17], and persistent hypotension was defined as the requirement of vasopressors to maintain a MAP of > 65 mmHg after initial fluid administration. Patients were excluded if they were pregnant, refused to provide informed consent, were in imminent death, or had withheld or withdrawn treatment.

The epidemiological characteristics and treatments of the patients were prospectively recorded, including the Simplified Acute Physiology Score III (SAPS III), sequential organ failure assessment (SOFA) score at admission to the ICU, and SOFA score on day 3. Clinical and laboratory data were also obtained. We measured HDL, TG, IL-1β, IL-6, IL-10, and CRP levels at the time of sepsis diagnosis (day 1) and on day 3. We defined this time interval for sequential assessment because it is a traditional period in which the response to sepsis management is reassessed[18]. HDL and TG levels were measured using a colorimetric assay (Roche Diagnostics^©, Mannheim, Germany) and expressed in mg/dL. IL-1β, IL-6, and IL-10 levels were measured using an enzyme-linked immunosorbent assay (Invitrogen, Waltham, MA, United States) and were measured in pg/mL. CRP was examined using a quantitative analyzer by turbidimetric immunoassay (Cobas c311, Roche Diagnostics^©, Mannheim, Germany), and the measurements were in mg/L, according to the manufacturer. As a marker of adequate mitochondrial metabolism, we measured the BCE, which is a measure of the mitochondrial oxygen flow coupled to adenosine triphosphate (ATP0 production[12]. BCE was measured using high-resolution respirometry measurements performed using oxygraphy (Oxygraph-2 k; Oroboros Instruments, Innsbruck, Austria) according to the appropriate methodology[19].

The primary outcome of the study was to evaluate the 28-days mortality with HDL levels on day 1 (HDL 1) and TG levels on day 1 (TG 1), as well as the association between HDL ratio [ratio between HDL on day 3 (HDL 3) and HDL 1] and TG ratio [ratio between TG on day 3 (TG 3) and TG 1] with the outcome. The secondary outcomes were the correlation between HDL 1, TG 1, HDL ratio, and TG ratio with clinical outcomes [need for hemodialysis (HD), mechanical ventilation (MV) along the sepsis course, reinfection along the hospitalar stay], association with SAPS III score and SOFA score improvement at day 3, association with hemodynamic response [maximum norepinephrine (NE) dose in the first 24 hours of sepsis management], with nutritional status, evaluated by albumin levels at sepsis diagnosis, and the association with levels of inflammatory variables (IL-1β, IL-6, IL-10, and CRP). Furthermore, we assessed TG 1 levels in conjunction with BCE on day 1, which is currently identified as an additional inflammatory biomarker. The relationship between these secondary outcomes and HDL and TG levels is considered exploratory due to the limited sample size of this study.

Descriptive statistics included n (%) for categorical variables and means, standard deviations, confidence intervals, medians, and interquartile ranges for continuous variables. To compare continuous and independent variables, we used the Student’s t-test or the Mann-Whitney U test, as appropriate. To compare HDL 3 and HDL 1 levels, as well as TG 3 and TG 1 levels, we used the Wilcoxon rank test. We used the χ² test or Fisher’s exact test to analyze categorical variables. The association between continuous variables was analyzed using Pearson’s r correlation test. The differences in TG 1 and HDL 1 levels in the different primary foci of sepsis were analyzed using ANOVA with Bonferroni correction. To assess the impact of HDL 1 on the main outcome, we performed a binomial logistic regression in a model with SOFA score on day 1, IL-10, and IL-1 as potential confounders, because these variables presented a P value < 0.05 in the univariate analysis. The 28-days mortality was the dependent variable. Statistical tests were two-tailed, with significance defined as P value less than 0.05. All P values were two-tailed. We used Jamovi software 2.6.44.0 and BlueSky Statistics 10.3.4 (BlueSky Statistics LLC, Chicago, IL, United States) for all analyses.

RESULTS

Main characteristics of the population and association with severity score

Seventy-five patients had measurements on day 1 (male:female ratio 58%:42%), and 50 patients had measurements on day 3. There was an increase in HDL 3 when compared with HDL 1:21 mg/dL (9-31) vs 14 mg/dL (6-30), P < 0.01. There was no difference in TG 3 when compared with TG 1: 139 mg/dL (104-207) vs 139 (88-223), P = 0.55. The main clinical characteristics dichotomized by survivors and non-survivors are shown in Table 1. All patients received enteral feeding along the study period, and none of them received parenteral nutrition. HDL 1 was correlated with albumin levels on day 1 (Pearson’s r = 0.434, 95%CI: 0.222 to 0.604; P < 0.01), but not TG 1 (Pearson’s r = -0.109, 95%CI: -0.33 to 0.125; P = 0.36). HDL 1 were not associated with SAPS III score (Pearson’s r = -0.187, 95%CI: -0.397 to 0.042; P = 0.11), but TG 1 were negatively correlated (Pearson’s r = -0.29, 95%CI: -0.485 to -0.07, P = 0.01). Patients who used statins before the sepsis episode (n = 14) did not have different HDL 1 levels compared with those who did not use statins (n = 58): [28 mg/dL (9-32) vs 13 mg/dL (6-25), respectively, P = 0.08]. Moreover, patients who used statins did not have different TG 1 levels as compared with those who did not use statins: 118 mg/dL (87-238) vs 152 mg/dL (100-214), P = 0.92. There was no difference in HDL 1 levels between different foci of infection (P = 0.14), but there was a difference in TG 1 levels (P = 0.02). In a post-hoc analysis, there was a difference between abdominal and urinary sepsis (P = 0.01) and between pulmonary and urinary sepsis (P = 0.01).

Table 1 Association between clinical and laboratory variables and 28-day mortality.

Variables	Non-survivors (n = 31), mean (SD) or median (IQR) or proportion	Survivors (n = 44), mean (SD) or median (IQR) or proportion	P value
Age	67 (14.3)	67 (15.8)	0.47
SAPS III score	77.2 (13.6)	72.7 (12.6)	0.15
SOFA score at sepsis diagnosis	9 (7-11)	7 (5-9)	< 0.01
SOFA improvement at day 3	14/31	37/44	< 0.01
Sepsis foci			0.22
Abdominal	14/31	14/44
Pulmonary	13/31	24/44
Primary bloodstream infection	3/31	1/44
Urinary	0/31	2/44
Comorbidities
Neoplasy	5/31	8/44	0.82
Cirrhosis	4/31	2/44	0.19
COPD	4/31	7/44	0.72
Diabetes	6/31	15/44	0.16
Hypertension	11/31	17/44	0.78
Inflammatory variables at sepsis diagnosis
IL-1 (pg/mL)	116 (34.2-658)	24.4 (16-374)	0.03
IL-10 (pg/mL)	200 (192-389)	177 (127-239)	0.05
IL-6 (pg/mL)	123 (27.9-213)	57.8 (30.6-191)	0.62
IL-1 ratio (pg/mL)	1.05 (0.68-1.62)	0.82 (0.42-1.27)	0.24
IL-10 ratio (pg/mL)	1 (0.91-1.72)	0.99 (0.76-1.4)	0.38
IL-6 ratio (pg/mL)	0.59 (0.26-2.38)	0.96 (0.34-1.71)	0.74
CRP	186 (112-272)	162 (82-215)	0.14
CRP ratio1	0.61 (0.42-0.82)	0.59 (0.39-0.82)	0.87
New-onset HD	16/31	6/44	0.01
Mechanical ventilation	29/31	33/44	0.04
Reinfection during hospital stay	15/31	21/44	0.96

¹CRP ratio: Ratio of C-reactive protein (CRP) levels on day 3 to CRP levels on day 1.

COPD: Chronic obstructive pulmonary disease; CRP: C-reactive protein; HD: Hemodialysis; IL: Interleukin; SAPS: Simplified acute physiology score; SOFA: Sequential organ failure assessment.

Association between HDL and TG levels with 28-days mortality

Patients who died had lower HDL 1 levels [9 mg/dL (5-17) vs 18.5 mg/dL (8-31), P = 0.02] but not lower TG 1 levels [142 mg/dL (83-224) vs 148 mg/dL (97-196), P = 0.97]. In addition, patients who died had a non-statistically significant increased HDL ratio [1.26 (1-1.79) vs 1.16 (1.01-1.39), P = 0.29] and had a non-statistically significant increased TG ratio [1.16 (0.85-1.36) vs 1.02 (0.67-1.36), P = 0.69]. In a multivariate analysis adjusted for potential confounders (SOFA score on day 1, IL-10, IL-1, and HDL 1), HDL 1 was not associated with 28-days hospital mortality (OR: 0.96, 95%CI: 0.91-1.02, P = 0.29), but IL-1 (OR: 1.01, 95%CI: 1.01-1.02, P = 0.03) on day 1 was associated with mortality.

Association between HDL and TG levels with clinical outcomes

Patients who had improvement in SOFA score at day 3 had increased levels of HDL 1 when compared with those who had not improved: 17 mg/dL (8-31) vs 7 mg/dL (5-17), P = 0.02. However, patients who had an improvement in SOFA score on day 3 did not have increased levels of TG 1 when compared with those who had not improved: 139 mg/dL (93-183) vs 166 mg/dL (101-226), P = 0.37. We evaluated the maximum NE dose (in µg/kg/minute) as a marker of hemodynamic instability. HDL 1 had a statistically significant negative correlation with the maximum NE dose (Pearson’s r = -0.241, 95%CI: -0.015 to -0.444; P = 0.03), but not TG 1 (Pearson’s r = 0.04, 95%CI: -0.188 to 0.265; P = 0.73).

A total of 36 patients developed reinfection during their hospital course. These patients did not have lower HDL 1 levels when compared with those who did not develop reinfection: 14 mg/dL (6.7 to 25) vs 14 mg/dL (6 to 30.5), respectively, P = 0.84; as well as did not have lower HDL ratio: 1.26 (1.07 to 1.75) vs 1.15 (1 to 1.38), respectively, P = 0.26. In addition, patients who developed reinfection during the hospital course did not have statistically different TG 1 levels when compared with those who did not develop reinfection: 133 mg/dL (104 to 208) vs 149 mg/dL (82 to 223), respectively, P = 0.94; as well as did not have statistically different TG ratio: 1.27 (0.77 to 1.39) vs 1.01 (0.7 to 1.28), respectively, P = 0.28.

Twenty-two patients developed acute kidney injury requiring HD. Patients who required HD along the sepsis course had a non-statistically significant lower HDL 1 levels when compared with those who did not need HD: 11 mg/dL (6-26.5) vs 14 mg/dL (7-30), P = 0.64. In addition, patients who needed HD had higher TG 1 levels than those who did not need HD: 178 mg/dL (99-249) vs 127 mg/dL (88-192), respectively, but this difference did not reach statistical significance (P = 0.18). There was no difference in the HDL ratio between patients who needed HD and those who did not: 1.26 (1.01-1.63) vs 1.16 (1-1.52), P = 0.83. In addition, there was no difference in the TG ratio between groups: 1.07 (0.8-1.3) in those who needed HD and 1.11 (0.7-1.37), P = 0.67; in those who did not.

A total of 62 patients required invasive MV during the course of sepsis. These patients had non-statistically significant lower levels of HDL 1 when compared to those that did not need invasive MV: 13 mg/dL (6-30) vs 14 mg/dL (11-22), respectively (P = 0.65). Patients who required invasive MV had a non-statistically significant higher level of TG 1 when compared to those who did not require invasive MV: 141 mg/dL (89-223) vs 139 mg/dL (89-219), respectively, P = 0.89. There was no difference in the HDL ratio in patients who needed MV compared to those who did not need MV: 1.19 (1.01-1.5) vs 1.15 (0.97-1.6), P = 0.94. In addition, there was no difference in the TG ratio between the groups: 1.12 (0.78-1.36) vs 0.98 (0.63-1.36), P = 0.81.

Association between HDL and TG levels with inflammatory biomarkers

Figure 1 shows the correlation between HDL 1 and TG 1 levels and the serum levels of interleukins and CRP using a correlation matrix. HDL 1 was negatively and statistically significantly associated with CRP levels (P < 0.001), as well as with IL-10 levels, but not with IL-1β or IL-6 levels. In contrast, TG 1 was positively and statistically significantly associated with CRP level (P < 0.01). In Figure 2, we evaluate the association between TG 1 and BCE at the time of sepsis diagnosis, and we did not find a statistically significant correlation between the variables [Pearson’s r = 0.05 (95%CI: -0.18 to 0.28); P = 0.667].

Figure 1 Correlation matrix between high-density lipoprotein, triglyceride and inflammatory variables. HDL: High-density lipoprotein; TG: Triglyceride; CRP: C-reactive protein; IL: Interleukin.

Figure 2 Correlation between triglyceride levels and lymphocytic mitochondrial metabolism. BCE: Biochemical coupling efficiency; TG 1: Triglyceride levels at sepsis diagnosis.

DISCUSSION

We identified distinct HDL and TG response profiles in the early phase of sepsis. Lower HDL 1 levels were associated with higher mortality, as previously described[7,14]. However, this association was not confirmed in the multivariate analysis when corrected for other risk variables. This fact may suggest a small discriminatory impact of HDL in predicting negative outcomes and therefore does not indicate an additional role beyond other commonly measured variables. Patients with higher HDL 1 levels had a greater improvement in SOFA score on day 3, an indicative marker of early clinical response in sepsis management[18]. This pattern of HDL levels and its association with improvement in organic failure has also been previously reported[15]. Nonetheless, lower levels of TG 1 were not associated with increased mortality, as previously reported in the literature[11], but were associated with greater severity of critical illness when measured using the SAPS III score.

HDL appears to emerge as a relevant player in both innate and adaptive immunity due to its pleiotropic properties, including anti-inflammatory, anti-apoptotic, and antioxidant functions[15]. It is well known that HDL levels can be reduced in the early phase of sepsis and are associated with mortality and adverse clinical outcomes[9]. Lipid-associated pathways exert distinct effects on immune cell behavior[20]. In the context of inflammatory disease, HDL inhibits the production of pro-inflammatory cytokines and chemokines by macrophages[21], thereby preventing activation and modulating the response of immune cells[22]. However, the linear evolution of HDL, as evaluated in this study, has been less studied[14]. Our results do not support an association between the improvement in the HDL (HDL ratio) and better clinical outcomes. However, our sample size prevents us from drawing definitive conclusions on this topic, and tracking the evolution at only two points (days 1 and 3) may not fully capture the biomarker’s response.

Hypertriglyceridemia may be caused by increased hepatic production of TGs, due to the availability of free fatty acids released by lipolysis of peripheral and visceral adipose tissue, that occurs as a result of stimulation by cathecolamines, endotoxins and cytokines (IL-1β, IL-6, TNFα)[8], classically described in animal models[10]. The immune response in sepsis is tightly associated with mitochondrial metabolism, especially in mononuclear cells[12]. One of the most relevant methodologies for this evaluation is the measurement of BCE in lymphocytes, which assesses the efficiency of oxygen consumption for ATP production[12]. There is an association between a dysregulated immune response and an increase in TG levels, as evidenced in hemophagocytic lymphohistiocytosis, which is a scenario of an extremely dysregulated and harmful immune response for the patient[23]. However, we did not observe an interaction between mitochondrial metabolism efficiency and TG levels in the initial phase of sepsis.

Although the association between serum levels of pro-inflammatory mediators (IL-6 and CRP) and TG levels has already been documented in pediatric patients[24], we were unable to identify this type of association with IL-6 in our cohort. In fact, there is little definitive data in the literature regarding this association; however, our small sample size may explain the negative result in this study. Our study has some limitations. We measured the biomarkers at only two time points, reflecting the early response to sepsis. We also measured short-term mortality, without understanding the potential role of these variables in predicting long-term outcomes and clinically relevant outcomes (quality of life and functional capacity). Moreover, the limited sample size and the single-center design of the study restrict the generalizability of the findings, thereby reducing its external validity.

CONCLUSION

Serum HDL and TG levels exhibit dynamic alterations during the management of sepsis. These parameters are correlated with inflammatory markers that are frequently modified in this context, suggesting their potential utility as biomarkers in this clinical settingx. Nonetheless, HDL and TG display distinct characteristics in the early response to sepsis. While TG levels are associated with patient severity upon ICU admission, lower HDL levels at the time of sepsis diagnosis are marginally linked to higher mortality. However, this association was not present when HDL was evaluated as an independent predictor, and an early improvement in HDL was not associated with better outcomes.