Introduction

Despite huge efforts in the treatment of cardiogenic shock (CS), it remains the most life-threatening complication of acute myocardial infarction (AMI) [1, 2]. Key features of CS are hypotension and global tissue hypoxia leading to an impaired microcirculation [3], vascular leakage, increased platelet and leucocyte adhesion to endothelial cells as well as endothelial dysfunction [4], and an activation of the sympathoadrenal system [5]. All these changes ultimately lead to multiorgan dysfunction syndrome (MODS), centrally determining the patients’ prognosis. Therefore, current treatment in CS patients aims at optimal prevention and therapy of MODS [6].

Different organ systems can be affected including the liver. Hepatic dysfunction is associated with poor outcome in critically ill patients [7, 8]. This disease state is referred to as hypoxic hepatitis (HH), ischemic hepatitis or shock liver [9]. It is characterized by centrilobular liver cell necrosis and distinctly elevated serum aminotransferase levels and occurs in the clinical setting of cardiac, circulatory or respiratory failure [10]. To date, HH is the most frequent cause of acute liver injury and has been reported with a prevalence of up to 10% of critically ill patients [11]. Although there is no specific treatment available aiming at improving hepatic function in HH, it is still of clinical relevance due to complications caused by HH. This includes alterations of glucose metabolism including spontaneous hypoglycemia, respiratory insufficiency caused by hepatopulmonary syndrome and hyperammonemia [12].

To date, the most frequent used diagnostic method is the determination of serum aminotransferase levels. Although other diagnostic modalities are available, such as determination of hepatic blood flow, liver biopsy, the invasive assessment of hepatic hemodynamics and the use of imaging techniques such as computed tomography [10], in the majority of patients the evaluation of the typical time pattern of transaminases during the course of HH is used. However, also new laboratory markers have been proposed. This includes argininosuccinate synthase 1 (ASS1) and the sulfotransferase isoform SULT2A1 (SUL), which is a major catalyst of the sulfation of dehydroepiandrosterone. Both are hepatic proteins that are degraded in the liver and rapidly released into circulation during liver ischemia [13]. Prima and coworkers were able to show in different animal models of liver injury that both new markers had higher sensitivity and specificity regarding liver injury determination compared to standard laboratory values and suggested them for clinical evaluation [14]. Therefore, the aims of the study were to assess the incidence of HH in CS complicating AMI, its prognostic implication and to evaluate the diagnostic value of new liver parameters in this clinical scenario.

Methods

The present study is a predefined substudy of the Intra-aortic Balloon Pump in Cardiogenic Shock II (IABP-SHOCK II) trial (ClinicalTrials.gov Identifier: NCT00491036) that investigated the use of Intra-aortic Balloon Pump (IABP) counterpulsation vs. control in patients with CS complicating AMI and showed no significant difference between the two treatment groups regarding the primary outcome 30-day mortality. The detailed design and main results of the trial have been published previously [15–18]. In summary, 600 patients were enrolled in 37 centers in Germany and underwent randomization to IABP support or to control in a 1:1 fashion. CS was defined as systemic hypotension, pulmonary congestion, and signs of impaired end-organ perfusion. Exclusion criteria were: duration of CS >12 h, cardiopulmonary resuscitation >30 min, severe cerebral deficit, mechanical causes of cardiogenic shock, age >90 years, contraindications against IABP insertion, shock of other cause, or severe concomitant disease with limited life expectancy. Of the total 600-patient study population, 218 were enrolled at the University of Leipzig-Heart Center with planned prospective blood sampling in the catheterization laboratory during the initial invasive procedure. Most importantly, cardiac catheterization was performed as soon as possible in all patients after hospital admission. In resuscitated patients, cooling was initiated after percutaneous coronary intervention (PCI) and blood sample drawing. The study was approved by an institutional review committee and the subjects gave informed consent. The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the institution’s human research committee.

Assessment of biomarker serum levels

For the evaluation of different serum markers blood samples were drawn at the end of PCI from the catheterization sheath (day 1) and on the next day (day 2) on intensive care unit. The EDTA-plasma was immediately separated by centrifugation (2400×g for 10 min) and aliquots were stored at −80 °C until assayed. Aspartate-aminotransferase (ASAT) and alanine-aminotransferase (ALAT) and glutamate-dehydrogenase (GLDH) were measured using techniques established in clinical routine. Therefore, a fully automated system (Architect ci16200, Abbott, IL, USA) has been used allowing standardized and validated measurements using established immunoassays. Regarding the definition of HH, several authors proposed that ALAT or ASAT should be at least 20 times the upper normal level [10] which is 14.8 µmol/l*s (normal range <0.74 µmol/l*s) for ALAT and 11.6 µmol/l*s (normal range <0.58 µmol/l*s) for ASAT. In the current study, the diagnosis of HH was met if at least one measurement of ALAT or ASAT was above these thresholds. The reference value of GLDH is <120 nmol/l*s.

ASS1 and SUL levels were measured with standard enzyme-linked immunosorbent assay kits (ASS1: Hoelzel Diagnostika, Germany; SUL: Acris Antibodies, Germany). Samples were assayed in duplicate. Serum lactate probes were taken at the same time point and analyzed by conventional blood gas analysis. Serum creatinine, troponin T and serum creatine kinase, with blood also drawn simultaneously, were determined via standard institutional laboratory measurements.

Statistical analysis

Categorical data are presented as counts or proportions with the corresponding percentages. Most continuous variables had non-normal distribution. For reasons of uniformity, summary statistics for all continuous variables are, therefore, presented as medians with interquartile range (IQR). For primary outcome analysis, all-cause mortality at 30-day was used. Patients were stratified according to the 30-day outcome. Logistic regression modeling was used to identify predictors of death at 30 days. All admission variables with an association (p value <0.1) to 30-day mortality in univariable analysis entered a multivariable stepwise logistic regression model consisting of forward inclusion and subsequent backward removal of parameters not needed within the model. This results in a robust model with high statistic power. Statistical analysis was performed using commercially available software (MedCalc for Windows, version 16.4.3; MedCalc Software, Ostend, Belgium). A two-tailed p value <0.05 was considered statistically significant.

Results

In total, 172 patients with blood available for the present analysis (79% of 218 patients randomized in Leipzig) were included in this predefined substudy. Table 1 shows the admission characteristics of the study population. The overall mortality rate after 30 days was 40%. Following the common definition of HH with an increase of ALAT or ASAT to be 20 times above the upper normal level on day 1 or day 2, the incidence of HH was 18%. Important differences between patients having HH or not were increased creatinine levels, increased serum lactate levels, increased creatine kinase levels, increased troponin levels, prior myocardial infarction and impaired flow following revascularization in patients with HH. Since only patients with available frozen plasma samples were included in the substudy, it is important to note that there were no important differences regarding baseline characteristics between patients within the substudy and the general study population [19]. To determine factors associated with the occurrence of HH, uni- and multivariable analyses were performed revealing serum lactate and impaired flow following revascularization as independent predictors of HH (Table 2).

Table 1 Characteristics of the study population
Full size table
Table 2 Logistic regression analysis for prediction of the occurrence of hypoxic hepatitis in cardiogenic shock
Full size table

Values regarding the different laboratory markers of liver injury are given in Table 3. All classical markers including ALAT, ASAT and GLDH were higher in non-survivors compared to survivors, whereas no differences could be observed for ASS1 and SUL. ALAT, ASAT, GLDH served as diagnostic parameters and were, therefore, by definition increased in HH. ASAT correlated strongly with ALAT (r = 0.939, p < 0.001) but showed only a weak correlation to serum lactate (r = 0.276, p < 0.001) and creatinine (r = 0.325, p < 0.001). Furthermore, on day 1, SUL was increased in patients with HH. In addition, an increase could be observed for ASAT between day 1 and 2 (p < 0.001), whereas a decrease for ASS1 was observed (p < 0.001). Patients with HH had a mortality rate of 68%, which was substantially higher compared to patients without HH (Fig. 1). There was no significant difference in the incidence of HH between patients randomized to IABP and the control group (21 vs. 15%, p = 0.36).

Table 3 Liver serum markers of comparing survivors and non-survivors on day 1 and day 2 of the treatment as well as serum markers comparing patients with and without hypoxic hepatitis
Full size table
Fig. 1
figure 1

Kaplan–Meier curves indicating the prognostic relevance of the incidence of hypoxic hepatitis in cardiogenic shock as defined by an increase of aminotransferase levels (p < 0.001)

Full size image

To determine predictors of outcome, uni- and multivariate Cox regression analyses were performed including the different laboratory liver values. Univariable logistic regression analysis identified baseline ASAT and baseline ALAT among others to be of relevance for prediction of 30-day mortality. After multivariable adjustment baseline ASAT remained an independent predictor of 30-day mortality together with serum lactate and serum creatinine (Table 4).

Table 4 Logistic regression analysis for prediction of the primary endpoint (30-day mortality)
Full size table

Then, the different laboratory values were compared using ROC analysis and the area under the curves (Fig. 2; Table 5). Using this model, classical liver function tests were confirmed to be superior to the new laboratory values. Comparing ALAT, ASAT and GLDH to each other, there was no statistical difference regarding the prediction of mortality. However, the highest area under the curve was calculated for ASAT.

Fig. 2
figure 2

Receiver operator characteristic curves comparing classical and new parameters regarding the prediction of the 30-day mortality endpoint at baseline. ALAT alanine-aminotransferase, ASAT aspartate-aminotransferase, GLDH glutamate-dehydrogenase, ASS1 argininosuccinate synthase 1, SUL sulfotransferase isoform SULT2A1

Full size image
Table 5 Comparison of the area under the curve (AUC) using ROC analysis to predict 30-day mortality regarding different liver laboratory values. Conventional biomarkers yield non-significant findings
Full size table

Discussion

The major findings of our study can be summarized as follows: (1) HH occurs in approximately every fifth patient in CS and is associated with particularly poor outcome, (2) as conventional biomarker, ASAT is the strongest laboratory predictor of outcome while newer proposed biomarkers failed to predict outcome. Of note, after multivariable adjustment ASAT, serum lactate and serum creatinine revealed to be independent predictors of 30-day mortality.

The central dilemma of CS is the fact that it is not simply a decrease in cardiac contractile function, but also a MODS as consequence of peripheral hypoperfusion with microcirculatory dysfunction [20, 21], often complicated by a systemic inflammatory response syndrome (SIRS). Dysfunctional organs contribute to a further impairment of renal function, a dysfunctional intestinal barrier or respiratory failure [22]. Therefore, the central treatment aim of pharmacological therapy as well as mechanical support is to maintain adequate perfusion and to prevent irreversible end-organ failure [23, 24].

An organ which is also prone to fail in CS is the liver, resulting in HH. This organ failure may be induced by hemodynamic instability or arterial hypoxemia in critically ill patients. In recent registries, the incidence of HH has been estimated to be at 1–11% of all patients admitted to intensive care units [11, 25, 26]. The most common cause of HH is cardiogenic or septic shock and most frequently these patients are characterized by other organ failures. Most importantly, HH represents an independent predictor of poor outcome as indicator of markedly compromised patients, but also by mediating several complications such as hepatopulmonary syndrome and hypoglycemia [27]. Of note, the current literature lacks a comprehensive description of acute heart failure and CS effects on the liver. Van Deursen was able to show that elevated liver function tests mainly indicate higher central venous pressure. However, ASAT showed also the strongest association with low cardiac index [28]. In a review, Samsky and coworkers summarized the pathophysiological liver changes in HF: venous congestion may increase the susceptibility of the liver leading in acute settings such as acute decompensated HF and CS to HH as second hit. The main reasons for hepatic dysfunction are increased hepatic venous pressure, decreased hepatic blood flow and decreased arterial oxygen saturation, all exaggerated in CS [29]. However, abdominal congestion might lead to increased abdominal pressure subsequently leading to impaired liver perfusion as well as dysregulated splanchnic vessel regulation [30].

The overall mortality in a mixed cohort of critically ill patients with HH has been estimated at 50–60% within 1 month, but no recent data are available in current practice in CS.

Our new data now comprehensively show that in a large contemporary study, the incidence of HH was high, namely 18%. These patients were characterized by a distinctly elevated mortality rate (68%). Considering the above-mentioned high mortality rates, this finding is consistent with the current literature on critically ill patients. Contemporary data on patients with liver failure in CS are lacking, older, smaller registries estimate the short-term mortality in acute decompensated heart failure and CS to be around 50% [31]. In addition, it has been shown that an increase in transaminases is associated with poor outcome in these patients [32].

The central diagnostic modality in HH is the determination of aminotransferase levels. In the present analysis, baseline ASAT levels had the highest diagnostic accuracy and remained an independent predictor of outcome even after multivariable adjustment. Therefore, baseline ASAT levels might serve to identify patients characterized by an excessively increased mortality risk. Of note, the proposed new parameters ASS1 and SUL failed to distinguish survivors and non-survivors yielding no prognostic information, although SUL was also increased in patients with HH on day 1. While SUL did not change between 1 and 2, there was a marked decline of ASS1 between 1 and 2. This might indicate a parameter with rapid increase and clearance following hepatic ischemia. Since baseline ASS1 was of no prognostic relevance this does not warrant further investigation in this clinical setting. Preclinical data also indicate rapid dynamics of ASS1 levels [13]; however, this might be of interest in other clinical scenarios. In the current study, ASS1 and SUL provided no additional information, although the AUC are rather low also for the conventional liver markers. Although HH might be one important contributor to excess mortality, patients’ prognosis is dependent on a number of different pathophysiological processes in CS.

A major problem in HH is that no specific treatment is available. The overall aim in these patients is to restore cardiac output and hemodynamic stability although this is the case also for patients with CS without HH. Still HH, should be taken into consideration while planning fluid balancing since too aggressive diuretic therapy might further decrease hepatic perfusion. An interesting observation has been described by Drolz and coworkers in patients on statin treatment prior to the index admission. Critically ill patients with HH and statin pretreatment had improved 28-day survival compared to patients without statin treatment. However, long-term survival was not positively influenced. This study might suggest that statins mediate protective effects [33], although this needs to be confirmed in additional studies. Another treatment option in case of severe HH is the use of a Molecular Adsorbent Recirculating System (MARS) as extracorporeal liver support device. However, again no larger trials are available and only case reports exist [27]. Nevertheless, this device might help to temporarily bridge the hepatic detoxification function in HH complicated CS. The detoxification process might also help to reduce rates of hepatopulmonary syndrome. Although the exact mechanisms in hepatopulmonary syndrome remain elusive, pulmonary vasodilatation is present due to reduced liver detoxification processes [34]. Another complication in critically ill patients has been brought into connection by Warkentin and Pai. A small minority of critically ill patients with CS, multiorgan failure, and disseminated intravascular coagulation develop symmetrical acral limb loss due to microvascular thrombosis. This has recently been linked with HH [35]. It has been speculated that the profoundly disturbed procoagulant–anticoagulant balance results in uncontrolled generation of thrombin due to the failure of the liver in protein C and antithrombin synthesis as natural anticoagulants. However, as HH precedes the onset of limb ischemia by several days, early therapeutic intervention may be possible [35]. In addition, it is tempting to speculate that also liver function is of prognostic relevance in another context. PCI is a central feature in the treatment of patients with CS and the patency of the revascularized vessel is warranted by dual antiplatelet therapy. Two of the three frequently used substances (clopidogrel and prasugrel) need hepatic activation which is possibly limited due to HH. The use of ticagrelor or the bridging of critical episodes with cangrelor might be useful to bypass liver activity-dependent processes [36].

There are certain limitations of our study that need to be discussed. First, only blood samples on day 1 and day 2 are available in our cohort and the onset of shock until hospital arrival differs as well as pre-hospital treatment. Although it can be expected that the increase in aminotransferase levels occurs until day 2 we cannot exclude that these levels might have peaked later and led to an underestimation of HH rates. However, baseline levels are available in all patients and indicate the prognostic relevance of liver function tests and help the clinician to identify patients at risk. Second, we did not collect data on baseline statin treatment which might have prognostic relevance in this cardiovascular patient cohort. Third, the investigated parameters mainly reflect hepatocellular injury and other parameters reflecting liver synthesis function and excretory function have not been investigated but might provide additional information [37]. In addition, no data on bilirubin, alkaline phosphatase, cholinesterase and antithrombin were recorded but might also provide additional information. Fourth, no biopsy has been taken as gold standard to diagnose HH. This should be taken into account in the evaluation of new biomarkers. Fifth, ASAT is also released by cardiac tissue and this might influence the potential to predict outcome in the setting of AMI.

In summary, HH occurs frequently in CS and is associated with particularly poor outcome. As conventional liver biomarker, ASAT is the strongest laboratory predictor of outcome while newer proposed parameters failed in the clinical scenario. Baseline liver function tests might serve as important parameters to identify patients at high risk. Future studies need to investigate the role of HH for several complications including the hepatopulmonary syndrome and the potential therapeutic role of liver detoxification.