Total SOFA and SOFA Kidney are Associated with Short Term and Long term Outcoming of Sepsis-Associated Liver Injury

Background: Liver injury is considered as a common complication of sepsis. However, there are still few studies on short-term and long-term prognostic factors of sepsis-associated liver injury (SALI). The objective of our study is to conduct a large sample data cohort study to explore the risk factors for short-term and long-term prognosis of SALI. Methods: Data from a public, US-based, critical-care database (Medical Information Mart for Intensive Care-III [MIMIC-III]) was used. Septic patients who met the denition of acute liver injury were enrolled. Variables extracted from MIMIC-III were used to evaluate patient demographics, clinical characteristics on Day 1 of intensive care unit admission, and clinical outcomes. The Logistic regression models were used to calculate risk ratio (RR) and 95% condence intervals (CIs) after adjusting for potential factors. Results: Among the 14687 participants in our study, there were 3140 (21.38%) with SALI. SALI was signicantly positively associated with ICU mortality (RR, 1.54; 95% CI, 1.32, 1.79), 28-day mortality (RR, 1.27; 95% CI, 1.11, 1.45) and 1-year mortality (RR, 1.19; 95% CI, 1.06, 1.34) after adjusting confounding factors. Stratied by SOFA, there was a positive association between SALI and ICU mortality (RR, 2.15; 95% CI, 1.64, 2.80), 28-day mortality (RR, 1.60; 95% CI, 1.28, 1.99), 1-year mortality (RR, 1.24; 95% CI, 1.04, 1.48) after adjusting confounding factors among people with sofa score ≤ 5. Similar results were also obtained between SALI and ICU mortality (RR,1.40; 95% CI, 1.17, 1.67), 28-day mortality (RR, 1.17; 95% CI, 0.99, 1.38), 1-year mortality (RR, 1.19; 95% CI, 1.02, 1.38) after adjusting confounding factors among people with sofa score> 5. Compared with SOFA renal> 1, SALI had a stronger positive correlation with ICU mortality (RR, 1.36; 95% CI, 1.01, 1.84), 28-day mortality (RR, 1.19; 95% CI, 0.91, 1.56), 1-year mortality (RR, 1.11; 95% CI, 0.88, 1.41) after adjusted confounding factors among people with SOFA renal ≤ 1. The values of polytomous variables may not sum to 100% due to rounding.


Background
Sepsis is an organ dysfunction syndrome resulting from dysregulated host response to infection and can develop into severe sepsis and septic shock. It is high morbidity and mortality [1,2] . Sepsis can cause lifethreatening multiple organ dysfunction, including the liver, kidneys, lungs, gastrointestinal tract, and circulation [2] . Traditionally, liver injury has been considered as a late manifestation of sepsis-induced multiple organ dysfunction syndrome. Recently, it has been shown that liver injury is an early event in sepsis [3][4][5] .
Studies have shown that liver injury was considered as an independent risk factor of poor prognosis in sepsis and was a powerful independent predictor of ICU mortality [6][7][8] . Based on a few small-sample studies, the incidence of SALI is estimated to be in the range of 30-34.7% and the mortality of SALI is estimated to be in the range of 30-60% [3,9−13] . Kobashi, H et al [3] demonstrated that the poor prognosis ratio(the proportion of the patients who died or whose condition had worsened or was unchanged)of SALI was 50.6% in a retrospective cohort study. Dou, J et al [11] demonstrated that hospital mortality of pediatric patients with SALI was 23.81%. Y. R. KANG et al [9] demonstrated that ICU mortality in septic shock patients with hepatic dysfunction was 36% and in-hospital mortality was 48% in a retrospective observational study. It is therefore essential to nd factors associated with high mortality in sepsisassociated liver injury.
At present, there are few studies on the factors associated with short-term and long-term prognosis of liver injury in sepsis. SOFA score has been demonstrated to be an effective predictor of ICU mortality in sepsis [14,15] . The Third International Consensus De nitions for Sepsis and Septic Shock (Sepsis-3) was released in 2016. The consensus de nition included the Sequential (Sepsis-related) Organ Failure Assessment (SOFA) score [2] . The SOFA score consists of six different subscores: respiratory, cardiovascular, liver, coagulation, renal, and neurological [14,16] . It re ects the number of dysfunctional organs as well as the degree of damage [17,18] .Therefore the SOFA score is commonly used to predict ICU mortality in critically ill patients, especially those with sepsis [19,20] . Gupta, T et al demonstrated that elevated hepatic SOFA scores were most predictive of in-hospital death [21] . In addition, serum T-BIL level lactate level, lactate clearance rate, increased alanine aminotransferase (ALT) and aspartate aminotransferase (AST), and decreased C3 and C4 were all independent risk factors for the prognosis of liver injury [22][23][24][25] . However, there is still a lack of cohort studies with large samples, and there are few studies on the long-term prognosis of SALI [26] . There are also fewer studies exploring the interaction between organs in sepsis.
Thus, the purpose of our research is to investigate the association between SALI and ICU mortality, 28day mortality, 1-year mortality in the cohort study. Also, we further evaluated the relationship between SALI and mortality with short term and long term strati ed by SOFA component.  [27] . This study undertook an analysis of the third-party database, which is anonymized and publicly available with preexisting institutional review board (IRB) approval.

Participants
Adult patients (> 18 years) who satis ed the third Sepsis de nition (Sepsis 3.0: sepsis as a condition with life-threatening organ dysfunction caused by a dysregulated host response to infection) at ICU admission were included (Fig. 1) [2] . We screened patients with documented or suspected infection, plus an acute change in the total SOFA score of ≥ 2 points at ICU admission [2] . Infection was diagnosed by the International Classi cation of Diseases (Nine version, ICD-9) diagnostic code in the MIMIC-III that was implemented by Angus et al. [28] . Patients with preexisting liver diseases, as identi ed by Quan et al. [29] or biliary disease were excluded based on the ICD-9 diagnostic code.
In the absence of standardized diagnostic criteria for SALI, septic patients who met the de nition of acute liver injury were eligible for study inclusion. Acute liver injury was de ned as a state wherein the patient's blood reports met at least one of these four conditions: (i) ALT ≥ 80 U/L; (ii) AST ≥ 80 U/L; (iii) T-Bil ≥ 3.0 mg/dL; and (iv) serum direct bilirubin (D-Bil) ≥ 0.6 mg/L. In patients with multiple ICU admissions for acute liver injury, only the rst session was included for analysis.
In the MIMIC-III cohort, 18,328 adult admissions satis ed the criteria of Sepsis 3.0 at ICU admission. After excluding patients with preexisting liver and biliary diseases, 14,687 patients remained, of whom 3,140 (21%) were diagnosed with SALI ( Fig. 1).

Data collection
The following variables were extracted from the MIMIC-III database to evaluate the patient demographics and clinical characteristics: age, sex, SOFA score, Acute Physiology and Chronic Health Evaluation (APACHE) III score, mechanical ventilation, renal replacement (RRT), vasopressor use in the rst 24 h of ICU admission, and other comorbidities including solid tumor, diabetes. The APACHE III and SOFA scores were calculated within the rst 24 h following ICU admission by using the code of the MIMIC-III repository [30] . Other variables, including platelet count (PLT), partial thromboplastin time (PTT), Urine output, lactate measured more than once in the rst 24 h of ICU admission were expressed as mean values.

statistical Analysis
The normal variables were expressed as the mean ± SD and the nonnormal variables were median (IQR).

Results
Among the 14687 participants in our study, there were 11547 people without SALI (78.62%) and 3140 (21.38%) with SALI. Compared with non-SALI, people with SALI were younger. Compared with males, females were more likely to have SALI. People with SALI were higher sore of sofa than people without SALI. (Table 1  Values were median (IQR) or mean ± SD or n (%).
The values of polytomous variables may not sum to 100% due to rounding.
The values of polytomous variables may not sum to 100% due to rounding.

Discussion
We found that SALI was an independent risk factor for ICU mortality, 28-day mortality and 1-year mortality. Also, strati ed by SOFA (5 points), there was a positive association between SALI and ICU mortality, 28-day mortality, 1-year mortality. Interestingly, compared with SOFA renal > 1, SALI had a stronger positive correlation with ICU mortality, 28-day mortality, 1-year mortality among people with SOFA renal ≤ 1.
Liver injury is a risk factor for the prognosis of sepsis and an important predictor of poor prognosis in septic patients [3,8,13,31,32] . This study showed that ICU mortality in patients with SALI was 11. 68%, 28day mortality was 19. 33, the one-year mortality rate was 39. 82%. This study had a slightly lower short-term mortality compared to other studies. This may be related to the different inclusion and exclusion criteria and the number of samples. In the study by Y.R.KANG et al [9] , the study population was 188 septic shock patients with hepatic dysfunction in ICU. The population included in this study was patients satis ed the criteria of Sepsis 3.0. And there is a lack of a uniform de nition of liver injury in sepsis,the standard de nition of liver injury used in this study was based on the previous literature [3,4,10,[33][34][35][36] . A prospective, observational, multicentre cohort study has demonstrated that liver failure is a factor associated with early mortality in septic patients [37] . Brun-Buisson.C. et al demonstrated that liver failure is strongly associated with mortality in the SOFA score [38] . This conclusion is consistent with the conclusions of previous studies. Although the study has demonstrated a close relationship between the occurrence and deterioration of liver injury during septic shock and long-term mortality (180-day mortality) [39] , one-year mortality of SALI is still lacking to be reported.
After adjusting confounding factors, the RRs among people with sofa score < 5 were higher compared with sofa score ≥ 5. Patients with sepsis-associated liver injury had higher short-term and long-term mortality even if SOFA scores were lower. This further con rms the poor prognosis of patients with sepsis-associated liver injury. Moreover, our present study showed that in short-term outcome studies, the liver interacts with the kidney, respiratory, and cardiovascular systems, and for long-term outcome, only the kidney interacts with the liver. These results indicate that the liver and kidney might be most closely related in multiple organ failure in sepsis. It has been shown that renal failure is a common complication in cirrhotic patients with sepsis [40] . At the same time, renal injury can also cause dysfunction of extrarenal distant organs (such as liver, intestines, lung, brain, etc.), which is also called "renal and extrarenal organs crosstalk" [41] . Renal and liver crosstalk during acute kidney injury may be caused by a complex combination of soluble in ammatory mediators and cellular immunity [42] . Interestingly, septic patients without renal injury have a worse short-term or long-term prognosis if they present with liver injury. The mechanism is still unclear and needs further exploration.
This study has some limitations. Although the sample size of this study is larger than that of previous studies, the MIMIC-III database is only from a single center. Therefore, some conclusions of this study will need to be veri ed by multicenter studies. The diagnostic criteria for SALI used in our study were based on previous studies. At present, there is no gold standard for the diagnosis of hepatic injury of sepsis. Our study shows that, owing to the high incidence and mortality of sepsis liver injury, further clinical studies are needed to develop diagnostic criteria for liver injury in sepsis. Simultaneously, attention toward the prevention and treatment of SALI is needed. This is a retrospective study, and we extracted data from patients only for their ICU stay. As some patients may have developed sepsis before ICU admission, the exact time between sepsis and liver injury cannot be accurately calculated. This may cause bias in the results for the time of liver injury. The study included only data on the day of ICU admission and dynamic observation of continuous clinical data was absent. Our study found that total SOFA and SOFA kidney are associated with short term and long term outcoming of SALI. Also, further research was needed to clarify the potential speci c mechanisms.
RYD, XCM and YNM designed the study and were guarantor of the paper. HM and HL prepared the draft and nished the manuscript. HL and YNM contributed to the data analysis and interpretation of the results. YTM and CT were involved in data collection. All authors read and approved the nal manuscript.