Secondary hepatic injury (HI) without underlying liver disease is the most common form of hepatic dysfunction in intensive care unit patients and increases morbidity and mortality.1–4 Hypoxic liver injury, cholestasis, and drug-induced liver injury (DILI) are the most frequent forms of secondary HI in critically ill patients.5–7

While the definition of HI influences the incidence and mortality rates, there is no standard definition for critically ill patients. Many studies use the definition of bilirubin >2 mg/dl in the SOFA score.8 On the other hand, previous studies have shown that elevated alanine aminotransferase (ALT) or aspartate transaminase (AST) above the upper limit of normal (ULN) is associated with increased mortality.9 For this reason, other studies use 2 to 3-fold elevation of AST, ALT, alkaline phosphatase (ALP), and gamma-glutamyl transpeptidase (GGT) to define it.10

Earlier studies used the definition of cholestasis instead of the definition of hepatic injury, and this definition is quite variable in the literature. While some studies have used total bilirubin levels greater than 2 mg/dL as a cutoff, others prefer a 2 to 3-fold elevation of ALP and/or GGT to define cholestasis.11

The importance of this problem is evident from the heterogeneity of definitions used in studies conducted during the COVID-19 pandemic and the difficulty in interpreting and generalising their results. The reported prevalence of liver injury in COVID-19 patients varied widely across studies, ranging from 5% to 78%. For this reason, it has generally been concluded that the results of these studies cannot be adapted to clinical practice.12–14

On the other hand, while having a standardised definition of HI is essential, no study compares HI definitions and investigates the ideal description in the ICU patient population. This study compares HI definitions and their relationship with incidence and mortality rates in the same ICU patient population with secondary HI and investigates the most reasonable definition.

This single-centre retrospective observational cohort study was performed in a University Hospital Medical ICU between 2015 and 2019. The research protocol was accepted by the Gazi University ethics committee (number 869/2020), and informed consent was waived due to its retrospective design. Adult (>18 years old) patients were included in the study. For patients with multiple admissions, only the first one was considered. Exclusion criteria: Patients with acute liver failure, chronic liver disease, and preexisting liver failure. Patients with bone metastases or using drugs such as filgrastim that increase leukocyte alkaline phosphatase. Demographics, admission diagnosis, usage of vasopressors, renal replacement therapy, mechanical ventilation during ICU stay, ICU infections, cardiac arrest history, medications, main laboratory results on ICU admission and during the stay, and hospital survival rates of the patients were recorded. The presence of any shock (septic shock, cardiogenic, obstructive, hypovolemic), acute respiratory failure, acute kidney injury, neutropenia, or cardiac arrest was also recorded.

Secondary HI was defined based on three different definitions: occurrence during ICU admission and ICU stay. ALT, AST, ALP, GGT, and total bilirubin (TBL) levels were recorded during hospital admission.8,15,16

In this study, we found that the incidence of HI was between 10 and 45% according to the definition used, and the mortality rate was between 33% and 57% (except SOFA-6 and SOFA12 criteria). The majority of the patients had borderline or mild HI, with more than 60% presenting as a cholestatic pattern. When SOFA-1.2 was accepted as the gold standard, HI, defined according to ACG-1, ACG-2 and EASL definitions, was an independent risk factor for mortality. At the same time, ACG-1 had the highest sensitivity, specificity, ppv, npv, accuracy, agreement and correlation values. Considering ACG1 for the definition of HI, although it had a similar mortality rate to the gold standard definition, it was an independent risk factor for mortality and showed higher incidence rates. As a result, the definition with the highest diagnostic value compared to the gold standard and best predicts mortality is ALT AST elevation < 2xULN and TBL > 1.2 mg/dL. That is, adding TBL > 1.2 to the borderline criteria of ACG. This definition includes more patients with HI with the same mortality rate as the gold standard definition and supports the study results, showing that even a tiny increase in ALT and AST increases mortality.19,20

No specific study compares secondary HI incidence and mortality rates according to the definition in the same ICU patient population in the literature. There is no consensus on the definition of HI used in critical care patients, neither the criteria included in the definition nor their cutoff points. Since liver dysfunction is an important prognostic marker in ICU patients, bilirubin levels are included in most scoring systems, such as SOFA and SAPS. In critically ill patients without underlying liver disease, hepatic dysfunction is a predictor of ICU and hospital mortality independent of the severity of illness at presentation.

In this study, SOFA-1.2 was accepted as the gold standard test for diagnosing HI because it was part of an intensive care scoring system and had reasonable incidence and mortality rates. The SOFA score uses TBL to evaluate HI and scores from TBL > 1.2 mg/dL. The most commonly used threshold value in the literature is TBL > 2 m/dL.11 In our study, when HI was defined as TBL > 2 mg/dL, the incidence was very low, and mortality was high. Therefore, we chose TBL > 1.2 mg/dL as the gold standard threshold value. While survival rates worsen with higher bilirubin levels, the peak time of bilirubin is later; it may increase during hemolysis and is elevated in only one-third of patients with hypoxic liver injury. So, it might not be a reliable marker alone.8,21–23

The ACG definition includes AST, ALT, ALP, and TBL (without any cut-off value in the original definition) criteria and lower cutoff levels. In contrast, the EASL definition has AST, ALP and TBL parameters and recommends higher cutoff values. We preferred to compare these three definitions because we wanted to compare definitions containing different parameters and threshold points and because they are the definitions of important institutions in this regard. In our study, more than 60% of the patients had cholestatic HI. Since almost 40% of the patients have hepatocellular and mixed patterns, it suggests that the standard definition to be used in ICU should include both transaminases and cholestasis indicators. While SOFA, one of the criteria used in our study, was a scoring system related to ICU, the others (ACG, EASL) were not developed for ICU patients. On the other hand, it is well known that the smallest increase in injury markers (AST, ALT) correlates very well with mortality.19,20 For this reason, we wanted to investigate a lower cut of values in our study. When choosing parameters, the definition of the cutoff point may also change incidence. Some researchers defined HI as any increase in liver enzymes above the ULN, while others described it as an increase at least two or three times above the ULN.24–29 To see this effect on ICU patients, we evaluated the ACG definition as ACG-1, ACG-2, ACG-3 and ACG-4.

Another chaotic issue related to this problem is that previous studies frequently used the definition of cholestasis for the definition of HI. Currently, there is no uniform definition of cholestasis in critically ill patients. While the most commonly used and recommended by the guidelines (ACG, EASL) is the RUCAM classification, many studies used total bilirubin levels greater than 2 mg/dL criteria or 2-fold to 3-fold elevation of ALP and/or GGT to define cholestasis.30–32 In the literature, the incidence of cholestatic liver injury in ICU patients varies between 11–36%, and mortality has been reported between 27–48%.18,21

Although ALP and AST are considered to indicate hepatocellular damage, ALP, TBL, and GGT are accepted as cholestasis markers; on the other hand, any of these substances may increase in any HI type. Meanwhile, although ALP, GGT, and TBL have high sensitivity for the diagnosis of cholestasis, their specificity is low as they can be elevated in many other conditions. For example, ALP may be of bone and leukocyte origin, while TBL may be elevated by non-hepatic causes like hemolysis.33

When the EASL definition is used in our study, it is seen that the incidence of HI was lower, and the mortality was high. The likely reasons for this are as follows: The definition of EASL was developed especially for diagnosing DILI, and this definition uses ALT, ALP, and TBL criteria.

Although most of the patients in our study group had borderline and mild HI, the emergence of HI as an independent risk factor that increased mortality by four times suggested that a more sensitive diagnosis should be used for ICU patients.

In our study group, HI developed in 29% of the patients after ICU admission. Similar to our study’s results, HI rates detected during hospital admission increase during hospitalisation studies published on this subject in the literature.25,26,34

Having a clear and consistent definition is essential for several reasons. It allows us to evaluate how interventions impact injury incidence and mortality rates. It ensures that injury cases are reported consistently across healthcare institutions and research studies. This consistency is beneficial for healthcare planning and resource allocation purposes. Moreover, a standardised definition of injury is crucial for quality improvement efforts within healthcare institutions. Furthermore, medical societies and organisations can develop guidelines and recommendations for managing injury based on this standardised definition. On the other hand, the sensitivity and specificity of the definition greatly influence the accuracy of disease incidence. A more sensitive definition may include milder injury cases and increasing reported incidence. On the contrary, a specific definition may only identify severe cases leading to a lower reported incidence but potentially higher mortality risk. It’s important to emphasise that any definition changes should be based on clinical evidence to ensure accurate diagnosis and improved patient care.

These results showed that defining HI with only TBL may cause lower incidence and higher mortality rates, and adding other hepatic injury markers may improve the diagnostic value of the definition in ICU patients. According to this study’s results, the potential best definition would be ALT AST elevation < 2xULN and TBL > 1.2 mg/dL. Further studies are necessary to validate these results.

Gül Gürsel: Conceptualization, methodology, formal analysis, writing-review editing.

Ayshan Mammadova: Data curation, investigation, project administration, supervision.

Eda Macit Aydın: Visualization, project administration, supervision.

Zeynep Çınar: Project administration, supervision.

Nurgül Naurzvai. Data curation, investigation.

Sümeyye Kodalak: Data curation, investigation.

None were declared.

There is no financial support for the study