Pathophysiological mechanisms of liver injury in COVID‐19

Abstract The recent outbreak of coronavirus disease 2019 (COVID‐19), caused by the Severe Acute Respiratory Syndrome Coronavirus‐2 (SARS‐CoV‐2) has resulted in a world‐wide pandemic. Disseminated lung injury with the development of acute respiratory distress syndrome (ARDS) is the main cause of mortality in COVID‐19. Although liver failure does not seem to occur in the absence of pre‐existing liver disease, hepatic involvement in COVID‐19 may correlate with overall disease severity and serve as a prognostic factor for the development of ARDS. The spectrum of liver injury in COVID‐19 may range from direct infection by SARS‐CoV‐2, indirect involvement by systemic inflammation, hypoxic changes, iatrogenic causes such as drugs and ventilation to exacerbation of underlying liver disease. This concise review discusses the potential pathophysiological mechanisms for SARS‐CoV‐2 hepatic tropism as well as acute and possibly long‐term liver injury in COVID‐19.


| THE S PEC TRUM OF LIVER INVOLVEMENT IN COVID -19
COVID-19 associated liver injury is defined as any liver damage occurring during disease course and treatment of COVID-19 patients, with or without pre-existing liver disease. 4,[34][35][36][37][38][39] This includes a broad spectrum of potential pathomechanisms including direct cytotoxicity from active viral replication of SARS-CoV-2 in the liver, 40,41 immunemediated liver damage due to the severe inflammatory response/ systemic inflammatory response syndrome (SIRS) in COVID- 19,42 hypoxic changes induced by respiratory failure, vascular changes due to coagulopathy, endothelitis or cardiac congestion from right heart failure, drug-induced liver injury and exacerbation of underlying liver disease ( Figure 2). The incidence of elevated liver transaminases (ALT and AST) in COVID-19 patients ranges from 2.5% to 76.3%. 35,38,43,44 In a recent meta-analysis, the pooled rate for AST and ALT outside the reference range was 20%-22.5% and 14.6%-20.1% respectively. 35,45 These abnormalities can be accompanied by slightly increased total bilirubin levels in up to 35% of cases. 35,38,43,44 While elevations of cholestatic liver enzymes [alkaline phosphatase (ALP) and gamma glutamyl transferase (γGT)] were initially considered rather rare, 4,22,23,46 recent systemic reviews highlight elevations of ALP and γGT in 6.1% and 21.1% of COVID-19 patients respectively. 35,45 Moreover, a biphasic pattern with initial transaminase elevations followed by cholestatic liver enzymes has been reported, which could reflect SIRS-induced cholestasis at the hepatocellular/canalicular level or more severe bile duct injury in the later stage of the disease. 47 Although COVID-19-associated liver injury has been reported to be mild, it may affect a significant proportion of patients, especially those with a more severe disease course. In F I G U R E 2 Proposed pathophysiology for liver injury upon SARS-CoV-2 infection. COVID-19-associated hepatocellular damage is mainly characterized by moderate steatosis, lobular and portal inflammation, apoptotic/necrotic foci and elevation of plasma ALT and AST (upper left panel). Preliminary observations suggest that the injury might be caused by hepatocellular infection with direct cytopathic effects of SARS-CoV-2, which could induce mitochondrial dysfunction and ER stress contributing to steatosis. Furthermore, SARS-CoV-2 infection might also activate mTOR, which eventually inhibits autophagy (as a mechanism of viral degradation) and facilitates viral escape from the immune system. In addition, cytokine storm, hypoxic conditions due to ARDS and drug-induced liver injury (DILI) may contribute. COVID-19-associated cholangiocellular injury has also been observed and is mainly characterized by bile duct proliferation, occasionally bile plug formation and elevation of plasma γGT and ALP (lower left panel). From a hepatological perspective, COVID-19-positive patients may be divided into three categories: patients without pre-existing chronic liver disease, patients with early stage chronic liver disease and patients with advanced chronic liver disease/cirrhosis. COVID-19-associated liver injury may have a more severe outcome in patients with preexisting liver disease, such as non-alcoholic fatty liver disease (NAFLD) and associated metabolic comorbidity. Moreover, COVID-19 may induce hepatic decompensation with increased mortality in cirrhotic patients (right panel) the light of the central role of the liver for the production of albumin, acute phase reactants and coagulation factors, hepatic dysfunction may impact on the multisystem manifestations of COVID-19 such as ARDS, coagulopathy and multiorgan failure. [2][3][4][5][6][7]48 Moreover, the liver is the primary metabolic and detoxifying organ in the human organism, and even a moderate loss of hepatic function could alter the safety profile and therapeutic efficacy of antiviral drugs metabolized in the liver. Hence, it is crucial to understand the causes of COVID-19-associated liver injury in more detail.
So far, systematic information on underlying histopathological alterations is scarce. Hepatic steatosis (in part microvesicular) and Kupffer cell activation appear to be commonly encountered in livers of SARS-CoV-2-infected deceased, together with vascular alterations including derangement of intrahepatic portal vein branches, usually mild lobular and portal inflammation, ductular proliferation and liver cell necrosis. 40,46,[49][50][51] Of note, examination of liver biopsies from a cohort of 48 deceased COVID-19 patients revealed extensive luminal thrombosis at the portal and sinusoidal level, together with portal fibrosis accompanied by significant pericyte activation. 51

| P OTENTIAL MOLECUL AR MECHANIS MS FOR SAR S -COV-2 TROPIS M OF THE LIVER
The presence of SARS-CoV-2 viral RNA has recently been demonstrated by qRT-PCR in liver among various other organs outside the respiratory tract, 52 although the exact cellular site of replication remained unspecified since nucleic acids have been isolated by wholetissue homogenization. However, in situ hybridization analyses revealed SARS-CoV-2 virions in vessel lumens and endothelial cells of portal veins of COVID-19 liver specimens. 51 Moreover, electron microscopic analyses on liver samples from two deceased COVID-19 patients with elevated liver enzymes demonstrated the presence of intact viral particles in the cytoplasm of hepatocytes. 40 Given recent, although still limited, discoveries, 40,51,52 hepatic tropism for SARS-CoV-2 and direct cytopathic effects should be considered as potential mechanism of COVID-19 associated liver injury, although a classic hepatitic picture has not been reported. 40,46,[49][50][51] The availability of viral receptors at the host cell surface is a major determinant of viral tropism for a specific tissue. 53 As such, SARS-CoV-2 cell entry is mediated by the S protein of the virus, which specifically interacts with host ACE2 and TMPRSS2 ( Figure 1). In order to understand whether SARS-CoV-2 might be able to infect liver cells, we explored the expression pattern of the human ACE2 and TMPRSS2 proteins using the Human Protein Atlas (data available at https://www.prote inatl as.org/ENSG0 00001 30234 -ACE2/tissue and https://www.prote inatl as.org/ENSG0 00001 84012 -TMPRS S2/ tissue). Interestingly, the expression levels of the two proteins is highest in intestine and gall bladder, but it appears to be virtually absent in the liver. These data might be incomplete or lack sensitivity, since in the Human Protein Atlas ACE2 expression also seems to be absent in the lungs, where infection is definitely known to occur.
In a recent study, Chai and colleagues applied single-cell RNAseq to healthy human liver samples and found that ACE2 expression levels in bile duct epithelium (cholangiocytes) is comparable to that of alveolar cells in the lungs, whereas hepatocellular ACE2 expression is low but still detectable. 54 Further confirmation of significant ACE2 and TMPRSS2 expression in liver parenchymal cells comes from bio-informatics analyses from the single-cell transcriptome database Single Cell Portal. 55 Interestingly, sinusoidal endothelial cells appear to be ACE2-negative, in line with previous observations. 56 This finding may be important considering recent reports on endothelitis of large intrahepatic vessels caused by SARS-CoV-2 48,57 and high ACE2 expression in other endothelia, including central and portal veins, which also can become infected by the virus. 51 Of note, studies in both mice and humans revealed increased hepatic ACE2 expression in hepatocytes upon liver fibrotic/cirrhotic conditions 58,59 (and our own unpublished observations). This finding may be of great relevance since pre-existing liver injury could thereby exacerbate SARS-CoV-2 hepatic tropism. Moreover, hypoxia, which is a typical feature in severe COVID-19 cases, has been shown to be a main regulator of hepatocellular ACE2 expression. 58 This might explain why extra-pulmonary SARS-CoV-2 dissemination is mainly observed in patients manifesting ARDS and other hypoxic conditions. Importantly, inflammatory conditions/diseases in the liver, as shown for other organs, 60,61 could also upregulate ACE2 expression. Since drug-induced liver injury (DILI) may contribute to liver injury in COVID-19 patients, 62 it might be of interest to explore whether DILI or certain drugs induce hepatic ACE2 over-expression.
In vitro experiments also showed that the S protein of lineage B beta-coronaviruses significantly increases the affinity for its receptor when it is pre-incubated with trypsin, that is when it is proteolytically activated. 63 Since liver epithelial cells express trypsin 64 and a plethora of other serine-proteases which constantly remodel the extracellular matrix, 65 ACE2 expression required for SARS-CoV-2 target and recognition in the liver might be lower than in other tissues with reduced extracellular proteolytic activity. 66 In line with these findings, it has been recently discovered that the S protein of SARS-CoV-2 bears a furin-like proteolytic site never observed before in other coronaviruses of the same lineage. 67 Interestingly, furin is predominantly expressed in organs that have been proposed as permissive for SARS-CoV-2 infection, such as salivary glands, kidney, pancreas (data for The Human Protein Atlas, available at https:// www.prote inatl as.org/ENSG0 00001 40564 -FURIN/ tissue) and the liver. 55 Finally, other factors, as for example ganglioside (GM1), 68 might influence S protein-ACE2 interaction. Therefore, research should also explore more deeply the S protein-ACE2 interactome to achieve new molecular and therapeutic insights.
In a recent report, Ou and colleagues tested pseudovirions containing the SARS-CoV-2 S protein for their ability to infect different cell lines. Interestingly, HuH7 cells, a hepatocyte cell line, as well as Calu3 cells, a human lung carcinoma cell line, were more efficiently transfected by viral vectors carrying the SARS-CoV-2 S protein than control pseudovirions. 69 Moreover, these studies revealed that viral entry might depend on the PIKfyve-TCP2 endocytotic pathway. A crosscheck in the Human Protein Atlas revealed that both PIKfyve and TPC2 are expressed in liver and gall bladder at comparable levels as in the lung (data available at https://www.prote inatl as.org/ ENSG0 00001 15020 -PIKFY VE/tissue and https://www.prote inatl as.org/ENSG0 00001 62341 -TPCN2/ tissue), highlighting the potential relevance of this pathway for hepatic tropism, which therefore expands from simple targeting and recognition to support of intracellular viral replication.
In an effort to establish a new and effective functional viromics A reliable source of information comes from recent work by Yang and colleagues, who demonstrated SARS-CoV-2 tropism for hepatocytes using organoids obtained from human pluripotent stem cell (hPSC)-derived hepatocyte and primary adult human hepatocytes. 73 In these systems, pseudovirions expressing SARS-Cov-2 S protein were able to infect human hepatocytes, while SARS-CoV-2 infection resulted in robust viral replication. 73 Gene expression analyses also showed that SARS-CoV-2-infected primary hepatocytes over-express pro-inflammatory cytokines, while downregulating key metabolic processes, as reflected by the inhibition of CYP7A1, CYP2A6, CYP1A2 and CYP2D6 expression. 73 Finally, Wang and colleagues applied electron microscopy imaging to liver samples of two deceased COVID-19 patients, and identified viral structures in hepatocytes which distinctively resemble SARS-CoV-2 virions. 40 This raises the possibility that the histopathological alterations seen in these patients may be caused by direct cytopathic effects of SARS-CoV-2 40 although a typical hepatitis pattern appears to be lacking. 40,46,[49][50][51] However, further studies with larger biopsy/autopsy cohorts and the combined imaging (including immune electron microscopy) may be necessary to confirm these preliminary observations of hepatocellular SARS-CoV-2 presence.

Bile duct epithelial cells (cholangiocytes) participate in bile
production and flow as well in immune response. 74 Single-cell sequencing of human long-term liver ductal organoid cultures showed preservation of ACE2 and TMPRSS2 expression. 75  whereas bile from two other small sample series tested negative. 24,49 These discrepancies might rely on the fact that the positive-tested bile sample has been obtained during surgical resolution of bile duct obstruction, 76 whereas the negatively tested bile was obtained from 48h post-mortem autopsies. 24,49 Tight junctions allow cholangiocytes to act as a protective barrier for parenchymal liver cells from toxic bile components. Viral infec-  78 and is induced by pro-inflammatory and pro-fibrotic cues, such as Angiotensin II, generated by the catalytic action of ACE as part of the pro-fibrotic branch of the renin-angiotensin system. 79 Of note, ACE2 counteracts ACE function by producing the anti-inflammatory and anti-fibrotic Angiotensin-(1-7) and thereby decreasing the Angiotensin II/Angiotensin-(1-7) ratio. 79 However, ACE2 expression has neither been detected in quiescent, nor in fibrogenic/ activated hepatic stellate cells. 58,[80][81][82][83] These findings suggest that these cells may be a rather non-permissive host for SARS-CoV-2.
Nevertheless, the pro-inflammatory milieu generated by direct or

| SAR S -CoV2 AND HEPATI C S TE ATOS IS
Microvesicular and macrovesicular steatosis have been observed in liver autopsies of COVID-19 patients who presented with SARS-CoV-2 infection as the only risk factor for liver injury, and in some cases, SARS-CoV-2 hepatocellular infection has been proven. 40,49 Importantly, hepatic lipid accumulation as a result of SARS-CoV-2 infection must be differentiated from pre-existing NAFLD, which has been shown to increase the risk for poor outcome in COVID-19 patients. 50 Deregulated in host lipid metabolism and mitochondrial activity as a result of potential direct SARS-CoV-2 cytopathic effects and/or immunopathology induced by cytokine storm, as well as drug side effects (eg corticosteroids) may be important contributors to the development of hepatic steatosis in COVID-19 ( Figure 2). Microvesicular steatosis is typically caused by genetic or acquired mitochondrial β-oxidation defects. 94 Preliminary observations suggest that SARSR-CoV-2 affects mitochondrial activity. 95 Furthermore, Wang et al also identified mitochondrial crista abnormalities in liver specimen of COVID-19 patients. 40 Interestingly, impaired mitochondrial activity has also been implicated in the pathogenesis of NAFLD/NASH. 96 Thus, SARS-CoV-2 infection might even worsen the metabolic state and aggravate pre-existing NAFLD by these mechanisms.
Endoplasmic reticulum (ER) stress is known to induce de novo lipogenesis in hepatocytes. 97 Several studies have implicated SARS-CoV infection in the induction of ER stress. For instance, significant up-regulation of ER stress markers glucose-regulated protein 78 (GRP78) and GRP94 has been observed upon SARS-CoV infection in several cell lines. [98][99][100] The coronavirus S protein seems to be a major burden for the host ER and might play a key role in ER stress induction. 98,99 Rearrangement of intracellular membranes by extensive depletion of lipid components from the ER during SARS-CoV-2 infection may also contribute to ER stress. 20 Moreover, the ER stress-related PERK-eIF2-α pathway is over-activated upon SARS-CoV infection in vitro. 101 Finally, electron microscopy examinations, which proved SARS-CoV-2 hepatocellular infection, reported a pathological ER dilatation in infected hepatocytes, 40 which most probably will cause ER stress. Collectively, these data could indicate that SARS-CoV-2, as other coronaviruses, induces ER stress upon infection, and that the ER stress-induced de novo lipogenesis could also contribute to the development of steatosis in COVID-19 patients ( Figure 2).
De novo lipogenesis is also induced by the mammalian target of rapamycin (mTOR), 102 which is also the cardinal regulator of autophagy. 103 SARS-CoV has been previously shown to hijack the autophagy pathway through processes that rely on the viral non-structural protein 6 (nsp6), highly conserved in SARS-CoV-2. 104-106 Furthermore, mTOR hyper-activation has been observed in MERS-CoV-infected HuH7 cells, and inhibition of mTOR signalling pathway by rapamycin inhibits viral replication. 107 Given the recent observations that SARS-CoV-2 infection restricts autophagy, 108 it is tempting to speculate that SARS-CoV-2, SARS-CoV and MERS-CoV share a similar mTOR-dependent mechanism of infection. Furthermore, significantly increased mTOR activity has been revealed upon IL-6 stimulation. 109

| S IR S -INDUCED CHOLE S TA S IS AND B ILE DUC T ALTER ATI ON S IN COVID -19
Cholestatic features such as bile duct proliferation, portal inflammatory infiltrates, and in some cases, canalicular/ductular bile plugs have been reported in post-mortem evaluations on COVID-19 patients. 49 In addition to hepatocellular features, bile duct changes, such as ductular proliferation have been observed in postmortem studies. 49 Notably, IL-6 is a strong cholangiocellular mitogen factor 128 and induces a proliferative and pro-inflammatory phenotype. 74,129 Bile ducts from patients with COVID-19 could therefore be ex- Therefore, hepatic long-term follow-up for COVID-19 survivors who experienced a severe disease course, such as ARDS with ECMO and prolonged ICU admission might be considered. Early diagnosis is paramount to best manage symptoms and disease progression of SSC-CIP, which could be counteracted with anti-cholestatic, cholangio-protective drugs such as UDCA or more recently norUDCA. 132-134

| SAR S -COV-2 AND HYP OXI C HEPATITIS
Causes for hypoxic hepatitis are multifactorial. In general, cardiac failure, sepsis and respiratory failure account for more than 90% of all cases. [135][136][137][138] Additionally, right-sided heart failure was found to aggravate liver injury by liver congestion as a result of elevated central venous pressure. 122,[135][136][137][138][139][140] In cases of long-lasting hemodynamic and/or respiratory failure, hypoxia results in hepatic cell death, histopathologically defined as centrilobular necrosis. 141 COVID-19-associated ARDS remains the most common complication requiring critical care management including invasive ventilation, high levels of positive end-expiratory pressure (PEEP) and vasoconstrictor therapy in case of hemodynamic instability. [142][143][144][145] These factors may be accompanied by right ventricular dysfunction caused by high pulmonary vascular resistance as a result of hypoxaemia and hypercapnia during ARDS. 146,147 Furthermore, COVID-19 causes a hyper-coagulate state with a significant incidence of pulmonary thrombotic complications aggravating acute right-sided heart failure and consequently liver congestion. 148 However, in the majority of cases, SARS-CoV-2 associated liver injury was generally mild and did not exceed >5 times the upper reference limit, therefore not fulfilling the diagnostic criteria for hypoxic hepaitis. 35 These findings were also obtained in critically ill patients referred to the ICU, suggesting that even in cases of severe respiratory failure during SARS-CoV-2 infection, the adequate oxygen supply to the liver is ensured by compensatory mechanisms. 35,36,39,[149][150][151][152][153][154] 155 is also clearly associated with steatosis or glycogenosis. 156 Recently, the first case of DILI associated with tocilizumab use in a COVID-19 patient has been reported. 62 Tocilizumab undergoes minimal hepatic metabolism, and the most probable etiology for its hepatotoxic effect is the interference with the IL-6 pathway, which plays a key role in hepatic regeneration. 157  RNA and nucleocapsid protein detection in gastric, duodenal and rectal epithelia. 160 Interestingly, gastrointestinal symptoms may appear before or even in the absence of manifestations in the respiratory tract. 165 This suggests that the GI tract might be a primary site of COVID-19 infection, and therefore that oral-fecal transmission could be an alternative route of infection for SARS-CoV-2 (this has been extensively reviewed). 162,170 We would like to propose the following putative way of SARS-

| CON CLUS I ON S AND PER S PEC TIVE S
Over the last months, several studies have highlighted the potential role of liver involvement in COVID-19 infection and pathology.
In this review, we analysed the published experimental and clinical findings concerning SARS-CoV-2 and previous coronavirus pandemics and proposed mechanisms concerning a putative SARS-CoV-2 hepatic tropism and the interplay between cytopathic and systemic effects in hepatic COVID-19 pathophysiology.
Elevated liver enzymes reflecting hepatic injury are common in COVID-19 patients both with and without chronic liver diseases. 35,38,43,44 Interestingly, while early clinical studies identified significant raises exclusively in serum ALT and AST upon SARS-CoV-2 infection, which reflect hepatocellular damage, recent investigations and metanalyses also highlighted significant increases in ALP and γ-GT and therefore cholangiocellular injury. 35,45 However, it is still not clear whether elevated serum liver biochemistries are causative for the worse outcome, or a consequence of the severe disease course.
In COVID-19 patients without pre-existing hepatic conditions who experienced liver damage, the injury is mostly mild. However, given the central role of the liver in endo-and xenobiotic/drug metabolism, coagulation, albumin and acute phase reactant production, hepatic dysfunction may impact on systemic disease pathophysiology of COVID-19. Long-term follow-up studies are required to explore potential long-term sequels of SARS-CoV-2 infection such as fibrosis.
Crucial questions remain open and need to be answered by future research: Which specific hepatic cells are infected by SARS-CoV-2?
Which molecular processes are dysregulated by the infection? What is the real contribution of direct cytopathic effects, cytokine storm, DILI or hypoxia in hepatic dysfunction? By which means could liver injury promote respiratory failure and predispose to a severe course of COVID-19?
The establishment of international registries collecting clinical reports of patients with liver diseases also tested positive for COVID-19, such as the COVID-Hep 175 and the SECURE-Cirrhosis, 176 together with molecular and translational research will surely help us shed some light on these intriguing questions and to set up more effective hepatoprotective programs for future pandemics.