No previous clinical studies evaluated the use of HDL2b combined with SOFA score for predicting 28-day and overall in-hospital mortality of sepsis patients. Our study had three major findings. First, among patients admitted to an ICU, patients with sepsis had a significantly lower level of HDL2b than non-septic controls. Second, multivariate analysis of sepsis patients indicated that the HDL2b level was independently and significantly associated with 28-day mortality. Third, an elevated HDL2b level combined with the SOFA score was associated with increased 28-day mortality and overall in-hospital mortality. These results suggest that HDL2b level has potential use as a predictor of short-term and long-term mortality in sepsis patients.
Traditionally, the main function of HDL was thought to be transportation of cholesterol from surrounding tissues into the liver for recirculation or excretion as bile acids. However, recent studies found that HDL also binds to endotoxins, has anti-inflammatory antioxidant effects, regulates serum complement levels, has antithrombotic effects, and improves endothelial function [11]. Some clinical studies showed that the HDL level decreased by approximately 50% during hospitalization of patients with sepsis, and a greater decrease was associated with more severe sepsis and worse prognosis [12–16]. In agreement, our analysis of ICU patients showed that the HDL-C level was notably lower in those who had sepsis (0.57 vs. 1.13 mmol/L, p < 0.001). These findings therefore confirmed that a lower HDL level is associated with sepsis and the prognosis of patients with sepsis.
HDL2 and HDL3 are the two main types of HDL, and there is more known about HDL2 than HDL3 in terms of metabolic pathway analysis [17]. HDL2b is a subtype of HDL2, and gradient-gel electrophoresis indicated it has a greater density and size than the other known subtypes (9.7–12.0 nm) [18]. Li et al. [6] studied a mouse model of sepsis and reported that the plasma HDL2b level in the sepsis group was significantly lower than in the control group (p < 0.01). Another study by Tang et al. [19] showed that the HDL2b level was negatively correlated with inflammatory factors (IL-1β and TNF-α), suggesting that it might have anti-inflammatory effects. These previous findings are consistent with our finding that the HDL2b level was significantly lower in patients with sepsis than in other ICU patients (10.95% vs. 23.78%, p < 0.001). Our multivariate analysis also showed that HDL2b (aOR: 0.780 [0.621, 0.978], p = 0.031) was significantly and independently associated with 28-day mortality in sepsis patients. HDL2b can be easily measured in clinical settings, suggesting it may be useful in real-world settings as a prognostic indicator for sepsis patients. Our multivariate analysis also found that hemoglobin and lactate levels measured within 24 h of ICU admission, and the presence of complications of acute liver failure were also reliable prognostic indicators in patients with sepsis.
Our ROC analysis of 28-day mortality indicated the AUC for HDL2b was 0.755, lower than the AUC for the SOFA score (0.782) but higher than the AUC for the APACHE II score (0.716). SOFA is a reliable prognostic indicator because it is based on assessment of multiple organ systems. In particular, the SOFA score assesses function of the respiratory, hematological, hepatic, cardiovascular, neurological, and renal systems, and the resulting score correlates significantly with in-hospital morbidity and mortality [20]. Other studies of sepsis patients indicated that a 2-point increase in the SOFA score significantly increased the risk of overall mortality [21–23]. Toker et al. showed that the SOFA score provided better predictions of in-hospital mortality than the SIRS score or the qSOFA score in patients with sepsis [24]. Therefore, we examined the combined use of HDL2b level with SOFA score as a prognostic indicator. Our ROC analysis of 28-day mortality indicated the HDL-2b combined with the SOFA score (AUC = 0.806, sensitivity: 72.73%, specificity: 83.33%) was a valuable indicator of patient risk, and had a better AUC value than HDL2b or the SOFA score alone. Our survival curve analysis demonstrated that a low HDL2b level combined with a high SOFA score was associated with significantly increased risk of 28-day mortality and overall in-hospital mortality. Based on our review of the literature, the combined HDL2b + SOFA score described herein may be the best tool for predicting the short-term outcome of sepsis patients.
Previous researchers have used many tools to predict 28-day and total in-hospital deaths in sepsis patients, including the Nutrition Risk In Critically ill (NUTRIC) score [25], Mortality in Emergency Department Sepsis (MEDS) score [26], Sepsis Patient Evaluation Emergency Department (SPEED) score [27], APACHE II score [28], and SOFA score [29]. In contrast, we showed that HDL2b combined with the SOFA score better described the onset and progression to multiple organ dysfunction syndrome (MODS) and morbidity in sepsis patients, and was objective, simple, easy to measure, and reliable. We believe that use of HDL2b combined with SOFA score will greatly assist clinicians in assessing the prognosis of sepsis patients using measurements performed at admission. Clinicians can also use the HDL2b + SOFA score to better inform family members of a patient's current condition and to describe the evolution of a patient's condition.
There is still a need for further research on this topic due to several limitations of our study. First, our sample size was small and all patients were from a single center. Second, although all samples were collected from patients within 24 h after admission for sepsis, there were some differences among patients in time since the onset of symptoms to sample collection. Additionally, we did not monitor the dynamics of HDL2b in sepsis patients and did not examine the possible mechanism by which it promoted the onset or progression of sepsis.