Low Lactic Acid and Hypercholesterolemia Reduce 90-day Mortality in Patients Suffering From Septic Shock According to the Sepsis-3 Denition

Background: Dyslipidemia and lactic acid levels are associated with poor prognosis of septic shock. After the revised denition of sepsis and septic shock (Sepsis-3) in 2016, data on the prognostic value of lactic acid levels and hypocholesterolemia were lacked. This study aimed to evaluate whether lactic acid and cholesterol can be used to predict mortality in ICU patients suffering from septic shock. Methods: Prospective observational study, 349 patients suffering from septic shock as dened by Sepsis-3. The COX model and the binary logistic regression model evaluate the correlation between lactic acid or total cholesterol and death from septic shock, and are adjusted according to the demographics, chronic diseases, and biomarkers of the participants. Results: Total cholesterol, total protein, and albumin levels were signicantly lower, and lactic acid, acute physiology, and chronic health assessment (APACHE II) were signicantly higher in the surviving group compared to those in the non-surviving group. Lactate levels < 4.4 mmol/L and total cholesterol levels < 2.42 mmol/L were associated with mortality after adjusting for confounders (odd ratio, 3.06 [95% CI, 1.78-5.27] and 1.99 [95% CI, 1.15-3.46]). In multivariate analysis, urogenital origin, albumin levels < 25 g/L, age < 60 years, and APACHE II < 20 were independent risk factors for death from septic shock. Conclusions: In those suffering from septic shock, increased blood lactate levels and decreased total cholesterol levels were associated with higher mortality. Furthermore, it was found that lactic acid and total cholesterol are sensitive markers of mortality in an ICU setting. (95% CI 1.00-2.99, P = 0.049); the lactate odds ratio, 3.06 (95% CI 1.78–5.27, P < 0.001); the total cholesterol odds ratio, 1.99 (95% CI 1.15– 3.46, P = 0.014); the odds ratio associated with age, 0.74 (95% CI 1.02–2.98, P = 0.042); and the odds ratio associated with urogenital origin was 5.36 (95 % CI 2.06–13.97, P = 0.001). The Hosmer-Lemeshow goodness-of-t test conrms that the model was well calibrated (P = .094).


Background
Septic shock, which is one of the most serious form of sepsis, has a high mortality rate [1,2] and an increasing incidence [3,4]. The hemodynamic management of septic shock aims to correct and prevent insu cient tissue perfusion in order to reduce the risk of multiple organ dysfunction and death [5]. The 2016 "Surviving Sepsis Campaign Guidelines" has identi ed several indicators of sepsis, such as central venous pressure and central venous oxygen saturation, redundant [7]. Increased serum lactate levels are indicative of insu cient tissue perfusion which can occur as a result of severe sepsis [6]. While serum lactic acid levels are not a direct indication of tissue perfusion, the absence of speci c sepsis indicators may increase the clinical importance of this biochemical measurement in patients undergoing early uid resuscitation [8]. It should be noted that the prognostic value of serum lactate levels in patients with septic shock is poorly understood due to the lack of data.
Cholesterol plays an important role in septic resistance and the prevention of dysfunctional responses to microbial infection [9]. Previous studies have found that high-density lipoproteins (HDL) have multiple protective mechanisms in patients with sepsis, such as binding and neutralizing bacterial toxins, inhibiting the expression of adhesion molecules, stimulating the production of endothelial nitric oxide synthase, and protecting low-density lipid proteins (LDL), which help to remove bacterial toxins from patients with sepsis [11], from peroxidative damage [10]. In the absence of HDLs, these processes can lead to in ammatory disorders, endothelial damage, organ damage, and death. Lagrost et al. demonstrated that patients who contracted sepsis after cardiac surgery have lower total cholesterol levels [12].
It remains unclear whether a decrease in serum lipoprotein levels re ects the severity of sepsis, as well as whether low levels of serum lipoprotein can be regarded as a risk factor for septic shock. Therefore, using the latest de nitions and guidelines for sepsis and septic shock, this study aimed to evaluate the role of lactic acid and cholesterol as predictors of mortality in patients with septic shock.

Data collection
During the patient's rst day in the ICU, baseline characteristics, such as sex, age, and Acute Physiology and Chronic Health Evaluation (APACHE II) score were recorded along with the presence of comorbidities, such as hypertension, chronic obstructive pulmonary disease (COPD), diabetes, and cancer. The patient's white blood cell count, procalcitonin levels, creatinine levels, alanine aminotransferase levels, aspartate aminotransferase levels, and lactic acid levels were also noted. The mortality of each patient was recorded and evaluated after 90 days.

Statistical analyses
Continuous variables were expressed as the mean ± standard deviation, and categorical variables were expressed as percentages. For univariate analysis, the subgroups were de ned as survivors and nonsurvivors. The signi cance of continuous variables were determined using the Mann-Whitney U-test or the Student's t-test. Categorical variables were determined using the Pearson chi-square test or the bilateral Fisher's exact test which identi ed any signi cant differences between the subgroups. Multivariate analysis was performed using Cox regression models to identify variables that independently predicted survival. To perform a multivariate analysis, the median continuous variable was halved, while the qualitative variable remained unchanged. In order to estimate the proportional hazard ratio and adjust for related confounding factors, multivariate Cox regression analysis was used to analyze variables whose P value was < 0.05 in univariate analysis. In order to approximate the life expectancy of each subgroup, a Kaplan-Meier survival curve was drawn using variables from the Cox analysis. The log-rank test was used to obtain the signi cant differences in survival rates between subgroups. Finally, a multiple logistic regression model was used to identify variables that were signi cantly related to death. This analysis was completed by estimating the "dying" odds ratio and the 95% con dence intervals (CI). Statistically different variables (age, lactate, cholesterol, APACHE II, total protein, albumin, and urogenital origin of infection), which were identi ed using univariate analysis, were then used in logistic regression models. The Hosmer-Lemeshow goodness-of-t test was used to evaluate the calibration of the logistic regression model and examine the differences between observed and expected deaths. The positive stepwise selection procedure was assessed using a likelihood ratio test and was used in two regression models (logical model and Cox model) to select variables that were signi cantly related to death and survival. SPSS 17.0 (SPSS, Inc, IBM, Chicago, IL) and GraphPad Prism 5.0 software (GraphPad Software, La Jolla, CA, USA) were used for statistical analysis. For all tests, P < 0.05 was considered statistically signi cant.

Patients characteristics
A total of 349 patients (mean age, 58.6 ± 15.9 years; range, 20-93 years) participated in this study; 45% of the patients were men and 49.9% suffered from comorbidities. Lung infection was the main source of infection in patients with septic shock (38.7%), followed by blood-borne infection (34.4%). The urogenital tract was found to be the least common source of infection (6.9%). The APACHE II score indicated critically ill patients. The overall 90-day fatality rate was 71.9% (Table 1). All patients demonstrated a decrease in total protein levels, albumin levels, and blood lipids (total cholesterol, triglycerides, HDL, and LDL) and an increase in liver transaminase levels, blood creatinine levels, white blood cell count, and procalcitonin (PCT) following their admission to the ICU (Table 1). ªValues are mean ± SD unless otherwise indicated.
Comparison between survival and non-survival patients

Nutritional parameters
The survival and non-survival groups were similar in terms of HDL, LDL, triglycerides, and white blood cells. In the non-surviving group, total cholesterol, high density lipoprotein, low density lipoprotein, total protein, and albumin levels were signi cantly reduced. It was found that 78.7% of patients with a median total cholesterol level < 2.42 mmol/L, and 65.1% of patients with a median total cholesterol level ≥ 2.42 mmol/L (P = 0.006) died. Compared to the survival group, the total protein and albumin levels of the nonsurvival group were signi cantly reduced (P = 0.004, P = 0.001). The fatality rate of patients with a median total protein level < 52.3 g/L was 78.3%. The fatality rate of patients with a median total protein level ≥ 52.3 g/L (P = 0.009) and a median albumin level of < 25 g/L (median value) was 79.3%. Additionally, the fatality rate of patients with a median albumin level ≥ 25 g/L was 64.6% (Table 2).

Other variables
In patients suffering from septic shock, older age, APACHE II scores, and blood lactic acid levels were found to be associated with a high mortality. Lactic acid levels were found to be signi cantly higher in the non-surviving group (P < 0.001) compared to the surviving group. The fatality rate of patients with median lactic acid levels ≥ 4.4 mmol/L was 83.6%, while the fatality rate of patients with median lactic acid levels < 4.4 mmol/L was 59.9%. The non-surviving group had signi cantly higher APACHE II scores (P < 0.001) compared to the surviving group. The fatality rate of patients with a median APACHE II value ≥ 20 was 84.1%, while the fatality rate of patients with a median APACHE II value < 20 was 58.7%. (P < 0.001).
The age distribution was similar between the two groups. The fatality rate of patients ≥ 60 years old was 77.2%, and the fatality rate of patients < 60 years old was 66.3% (P = 0.024). Furthermore, the fatality rate associated with infections originating from the urogenital system was 37.5%, which was lower than the fatality rate associated with infections originating from other regions (74.5%) (P < 0.001). The surviving and non-surviving groups were similar in terms of sex, comorbidities, blood creatinine, and PCT (Table 3).

Multivariate Cox regression analysis and Kaplan-Meyer survival curves
Variables that were found to be statistically signi cant during univariate analysis (APACHE II score, total protein, albumin, lactic acid, and total cholesterol urogenital origin and, age) were evaluated using the Cox proportional hazards regression model. Independent predictors of overall patient survival rate were total cholesterol < 2.42 mmol/L, APACHE II < 20, age < 60 years, urogenital source, and lactate < 4.4 mmol/L (Table 4). Figure 1 demonstrates the Kaplan-Meier survival curve grouped according to the median value of total serum cholesterol. Patients with low serum cholesterol had signi cantly lower survival rates compared to those with high cholesterol levels (log-rank test, P = 0.002). Figure 2 demonstrates the Kaplan-Meier survival curve grouped according to median total serum lactate levels.
The survival rate of patients with low lactate levels was signi cantly higher than that of patients with high lactate levels (log-rank test, P < 0.001). Figure 3 demonstrates the relationship between lactic acid and cholesterol (r 2 = 0.09, insigni cant). Please note that the survival rate of patients in the upper left quadrant (total cholesterol < 2.42 mmol/L and lactate ≥ 4.4 mmol/L) is extremely low.

Multivariate analysis for independent factors associated with death
Multivariate analysis was used to evaluate APACHE II, albumin, lactic acid, and total cholesterol (divided into two parts according to the median), age and urogenital source of infection. All of the aforementioned factors are independent and signi cantly correlated with mortality (

Discussion
This study evaluated whether serum lactic acid and cholesterol could predict mortality in patients with septic shock. The study cohort consisted of patients suffering from septic shock, as de ned by the International Consensus on Sepsis-3. This study found that initial lactate levels are associated with 90day mortality in patients with septic shock. This nding is consistent with other recent data [8,13]. Based on the latest de nition of septic shock, previous studies which evaluated the effect of cholesterol on the prognosis of patients with septic shock are limited and inconsistent. This study found that higher serum total cholesterol levels are bene cial to patients suffering from septic shock.
It is generally believed that low serum lipid levels are bene cial as the risk of cardiovascular disease is reduced. However, in recent years, studies have shown that low serum lipid levels are signi cantly correlated with patient mortality [14,15]. In the early stages of a critical illness, the concentration and composition of lipids and lipoproteins in ICU patients change profoundly [16,17]. Studies have shown that infection or in ammatory response syndrome can cause a signi cant decrease in total cholesterol levels along with LDL and HDL levels [18,19]. This study found that the content of total cholesterol and the concentrations of LDLs and HDLs were signi cantly lower than the normal values in patients who met the de nition of septic shock, according to Sepsis-3 [7]. In patients with sepsis and septic shock, rapid changes in serum lipid levels are closely related to clinical prognosis [20,21]. Furthermore, HDL and LDL play an important role in removing bacterial toxins. In vitro studies have shown that lipopolysaccharides of gram-negative bacteria combine with HDL (60%) and LDL (25%). Therefore, low HDL and LDL concentrations increase the risk of a poor prognosis in sepsis patients due to the body's inability to remove bacterial toxins from the bloodstream [9].
Interestingly, this study did not demonstrate that a low HDL concentration is associated with a poor 90day prognosis in patients with septic shock. Contrary to the results of a previous meta-analysis, lower HDL levels are associated with higher mortality in adult patients suffering from sepsis [22]. Such results are not entirely surprising; a randomized controlled study demonstrated that raised HDL levels have no bene t in protecting patients with myocardial infarction [23]. At the same time, studies have shown that under some chronic in ammatory conditions, such as metabolic syndrome and rheumatoid arthritis, and acute some acute in ammatory conditions, such as sepsis and myocardial infarction, HDL function is dysfunctional and pro-in ammatory [24][25][26][27][28]. Current evidence suggests that HDL function may be more important than quantity. Daniel et al. found that hypocholesterolemia is an independent risk factor for death in critically ill patients [29]. Sérgio et al. also proved that total cholesterol concentrations ≤ 2.48 mmol/L are an independent risk factor for a poor prognosis in patients with acute kidney injuries [30].
The results of this study indicate that a reduction in total cholesterol levels is signi cantly associated with an increase in mortality in patients with septic shock. To the best of our knowledge, this study is the rst to suggest that a total cholesterol level < 2.42 mmol/L is an independent risk factor for 90-day mortality in patients with septic shock.
Lactic acid levels in patients who died as a result of septic shock were signi cantly increased. Blood lactic acid levels were therefore determined as a predictor of mortality in patients with septic shock. Blood lactic acid levels ≥ 4.4 mmol/L were associated with a higher mortality rate in patients suffering from septic shock [13]. Studies have shown that septic shock, as de ned by Sepsis-3, reduces the sample size by 50% and increases the 28-day mortality rate compared to the previous de nition [31]. Therefore, the prognostic value of blood lactic acid levels in patients with septic shock, as de ned by Sepsis-3, will be of great signi cance to clinicians in lactic acid-led resuscitation [8]. Previous research found that early, goaldirected therapy can reduce the mortality of adults with severe sepsis and septic shock, and lactateguided therapy is more effective than early target-oriented therapy [32]. In fact, in the current study, when used alongside cholesterol to predict prognosis, patients with a lactate level above the median and a cholesterol level below the median accounted for only 10% of surviving patients.
In this study, serum albumin levels of non-surviving patients with septic shock were signi cantly reduced. Serum albumin levels below 25 g/L were found to be associated with a higher mortality rate in patients suffering from septic shock. Previous studies have shown that serum albumin levels are related to the prognosis of critical illness, and the relationship between hypoproteinemia, in ammation, and malnutrition is indissoluble [33,34]. Furthermore, it is known that serum albumin levels can be reduced by stress. Clinical studies have shown that albumin replacement does not improve the 28-day and 90-day survival rates in patients with severe sepsis compared to crystalloids alone [35]. Therefore, serum albumin may not be an indicator that accurately re ects the prognosis of patients with septic shock.
A survey of sepsis severity demonstrated that patients with septic shock had an APACHE II score of ≥ 20.
Furthermore, deaths in patients over 60 years of age were signi cantly higher than in those under the age of 60 with an APACHE II score of < 20. These ndings are consistent with previous research [36,37]; however, this study further quanti ed the severity of the disease and the risk of death in patients suffering from septic shock. Previous studies have demonstrated that the majority of sepsis cases originate from primary lung infections [38,39]. Furthermore, that patients are more likely to develop septic shock if their infection originated from the urogenital tract, abdomen, or blood [40]. The mortality rate of those suffering from a primary bloodstream infection is higher than that of patients whose primary infections did not originate from the blood [41]. This study shows that the 90-day prognosis of patients with septic shock originating from a urogenital tract infection is better than that of patients with infections which originated from other regions.

Conclusions
High serum lactate levels and hypocholesterolemia at the time of admission to the ICU have been associated with higher mortality in patients with septic shock. The combined use of these prognostic factors may facilitate the early prediction of mortality in patients suffering from septic shock. Abbreviations HDL, High-density lipoproteins; LDL, Low-density lipid proteins; ICU, intensive care unit; APACHE, Acute Physiology and Chronic Health Evaluation; COPD, Chronic obstructive pulmonary disease; CI, con dence intervals Declarations Ethics approval and consent to participate The study was approved by the Scienti c and Clinical Ethics Committee of the First A liated Hospital of Zhengzhou University, and all patients signed written informed consent.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Con ict of Interests
The authors declare that they have no competing interests.  Kaplan-Meier survival curves according to the median value of total cholesterol serum levels in patients with septic shock. Patients with low cholesterol levels had a signi cantly lower survival rate than did those with high cholesterol levels (log-rank test, p = 0.007).

Figure 2
Kaplan-Meier survival curves according to the median value of serum lactic acid levels in patients with septic shock. Patients with high lactic acid levels have a signi cantly lower survival rate than did those with low lactic acid levels (log rank test, p < 0.001).