Pathogenesis of GI and liver manifestations
SARS-CoV-2 uses the spike protein (S) to bind to the ACE-2 receptor in target cells. The ACE-2 receptor is present on type 1 and 2 surface alveolar cells, leading to the predominant respiratory symptoms and the droplet mode of transmission. The ACE-2 receptor is also widely expressed throughout the GI tract (Fig. 1). On immunohistochemical (IHC) staining, Hamming et al.2 demonstrated that the ACE-2 receptor is present in abundance in the vascular endothelium and smooth muscle cells of the vessels supplying the GI tract.
The pathophysiology of liver injury in COVID-19 is not as well established as its intestinal counterpart. In the liver, cholangiocytes and hepatic endothelial cells have been proposed to be the target cells for SARS-CoV-2.3 Cholangiocytes express not only the ACE-2 receptor but also the transmembrane serine protease 2 (TMPRSS2), which cleaves the S protein of the virus prior to its entry into cells, thus providing the basis of cholangiocytes being highly vulnerable to SARS-CoV-2 damage.4 It has also been shown in the liver ductal organoid model that SARS-CoV-2 leads to direct cytopathic changes in cholangiocytes, as hypothesized.4 Histopathologic evaluation of autopsy and post-mortem biopsies reveal mild sinusoidal dilation with increased small lymphocyte infiltration. In addition, steatosis, multifocal hepatic necrosis without inflammatory cellular infiltration, and canalicular cholestasis have all been reported in the liver biopsies of patients with COVID-19 patients. Interestingly, portal tract inflammation was not evident in these biopsies.5 Sinusoidal dilation is attributed to cardiogenic venous outflow slowdown. It is well recognized that hypoxia and impaired cardiovascular function predispose the liver to injury. Both zones 1 and 3 show injury with no cellular infiltrate, ballooning, Mallory hyaline, or fibrosis. Several potential mechanisms have been postulated in the pathophysiology of liver manifestations, such as a direct viral insult, exacerbation of the underlying liver disease, hyperinflammatory states, and drug-induced injury, but evidence to support either mechanism is scanty.6
Liver manifestations of COVID-19
Hepatic injury is common in COVID-19 and is multifactorial. Possible reasons include direct hepatic involvement due to the virus, drug-induced liver injury (DILI) due to various therapeutic agents, hypotension, and the associated underlying liver disease (cirrhosis due to various etiologies, alcoholic steatohepatitis, non-alcoholic fatty liver disease, and viral hepatitis) (Fig. 2). The prevalence of GI and hepatic manifestations of COVID-19 is variable across studies from different regions, as highlighted in the data from meta-analyses (Table 1)7–13 and individual studies (Table 2).14–19 In addition, endemic areas are associated with co-infections, such as acute viral hepatitis, and tropical infections, such as malaria and dengue.20–22
Table 1Prevalence of liver manifestations in patients with COVID-19 as reported in meta-analyses
| AuthorRef | Mao et al.7 | Sultan et al.8 | Parasa et al.9 | Kumar et al.10 | Wan et al.11 | Zarifian et al.12 | Kulkarni et al.13 |
|---|
| Patients included | 6,686 | 10,676 | 4,805 | 4,676 | 15,141 | 13,251 | 20,479 |
| Elevated AST | 21% | 15% | 20% | 25% | 25.4% | 22.8% | 22.5% |
| Elevated ALT | 18% | 15% | 14.6% | 23% | 25.3% | 20.6% | 20.1% |
| Elevated Bilirubin | 6% | 16.7% | NR | 9% | 8.8% | 7.8% | 13.4% |
| Prolonged INR | NR | NR | NR | 7% | NR | 18% | 9.7% |
| Hypoalbuminemia | 6% | NR | NR | 60% | NR | 39.8% | 55.5% |
| ALP | NR | NR | NR | NR | NR | 4.6% | 6.1% |
| GGT | NR | NR | NR | NR | NR | NR | 21.1% |
Table 2Prevalence of GI liver manifestations in patients with COVID-19 infection as reported in individual studies from across the countries to highlight the regional variation
| AuthorRef | Laszkowska et al.14 | Guan et al.15 | Aghemo et al.16 | Moura et al.17 | Docherty et al.18 | Rivera et al.19 |
|---|
| Patients included | 2,804 | 1,099 | 292 | 400 | 20,133 | 76 |
| Country | USA | China | Italy | Brazil | UK | Spain |
| Overall prevalence of GI symptoms | 38.7% | NR | 28.2% | 33.4% | 29% | 59.2% |
| Diarrhea | 23.4% | 3.8% | 27.1% | 17.3% | 20.4% | 40.8% |
| Nausea/vomiting | 23.2% | 5% | 4.0% | 13.8% | 19.8% | 22.4%/9.2% |
| Abdominal pain | 11.9% | NR | NR | 11.5% | 10.2% | 27.6% |
| Anorexia | NR | NR | NR | 6% | NR | 15.8% |
| Elevated AST | NR | 22.2% | 26.7% | NR | NR | NR |
| Elevated ALT | NR | 21.3% | 18.5% | NR | NR | NR |
| Elevated bilirubin | NR | 10.5% | 10.6% | NR | NR | NR |
| Prolonged INR | NR | NR | NR | NR | NR | NR |
| Elevated ALP | NR | NR | 9.6% | NR | NR | NR |
| Hypoalbuminemia | NR | NR | NR | NR | NR | NR |
Liver function test abnormalities
Alanine aminotransferase (ALT) elevations were seen in 4% to 33% of cases, according to China’s initial reports,23–25 and 39% of cases in a large study from New York, USA.26 The prevalence of aspartate aminotransferase (AST) elevation ranged between 4% to 53% in a Chinese cohort and up to 58% in a USA cohort.23,25,26 Both enzymes were mildly elevated in terms of absolute numbers and less than 5-times the upper limit of normal (ULN) in the majority. Kulkarni et al.13 in their meta-analysis, placed the pooled incidence of AST and ALT elevation at 22.5% and 20.1%, respectively.
Elevation in gamma glutamyl transferase (GGT) has been reported in 13% to 54%, whereas elevation in alkaline phosphatase (ALP) is uncommonly elevated, in only 2% to 5% of cases.27,28 In the meta-analysis by Kulkarni et al.,13 the ALP and GGT elevation incidence was 6.1% and 21.1%, respectively. The rise in ALP may be disproportionate to other liver enzymes.
Hyperbilirubinemia may be seen in up to 18% of cases.13,25,28 However, the derangement in the liver function tests can be multifactorial, as highlighted above, and it may be difficult to attribute to SARS-CoV-2-induced hepatic dysfunction alone.
Hypoalbuminemia has been described in severe COVID-19 patients and may not parallel changes in AST and ALT. In a retrospective cohort of 299 patients, 106 (35.5%) patients had low albumin, with significant differences in the albumin levels of survivors and non-survivors (37.6 g/L vs. 30.5 g/L).29 Albumin levels have also been found to be an independent predictive factor for mortality.29 In a meta-analysis of 1,990 patients across 14 studies, hypoalbuminemia was noted in 55.5%.13 An important finding was that only 11% to 45.8% of patients with non-severe infection had hypoalbuminemia. In the severely ill, hypoalbuminemia was seen in up to 72.9%; whereas, among the deceased, hypoalbuminemia was reported in 78–100% cases, making a case for the use of albumin levels as a prognostic marker in these patients.13 Albumin is a negative acute phase reactant, and the clinical relevance of low albumin as a predictor of outcomes must be interpreted with caution.
Coagulation disturbances
Prothrombin time/international normalized ratio
Coagulation disturbances in COVID-19 may be due to either a dysregulated immune response or liver failure, with dysregulated immune response being more commonly encountered.30,31 The cytokine storm syndrome (referred to herein as CSS) associated with COVID-19 leads to excessive pro-inflammatory cytokine release, which eventually results in endothelial injury, which may lead to disseminated intravascular coagulation, microvascular thrombotic angiopathy, and pulmonary embolism.32,33 Several studies have described prolonged prothrombin times and D-dimer levels.30,34,35
Endotheliitis was observed in the liver of patients with COVID-19 and fibrin microthrombi were found in liver sinusoids.36,37 The largest series of liver biopsies taken at autopsy (48 patients) showed massive dilation of portal vein branches, luminal thrombosis, portal tract fibrosis, and microthrombi in the sinusoids.38
The altered liver function tests (LFTs) could be related to CSS leading to shock and coagulopathy, affecting liver perfusion and resulting in cell death.38,39 Klok et al.40 reviewed 184 patients admitted in three intensive care units (ICUs) in the Netherlands and reported the composite incidence of thrombotic events (considering both arterial and/or venous) to be 49 % adjusted for competing risk of mortality. The most common thrombotic event was pulmonary thromboembolism, seen in 87% of patients. Tang et al.41 determined that the administration of low molecular weight heparin (LMWH) for 7 days or longer was associated with lower 28-day mortality in patients with sepsis-induced coagulopathy (SIC) score of ≥ 4 or a D-dimer value of > 6 times the ULN. They used a working definition of SIC as previously defined by the presence of infection-induced organ dysfunction as characterized using a composite score compiled using platelets, international normalized ratio (INR), and sequential organ failure (SOFA) score.42
SARS-CoV-2 and acute liver failure
As noted previously, a non-specific rise in AST and ALT levels up to 5 times the ULN can be seen in COVID-19 patients, along with hyperbilirubinemia. Acute liver failure (ALF) has been reported rarely. Of the five ALF cases reported, three were from the USA and one from Germany and Qatar each.43–47 Two of these were young, aged 24 years and 35 years, while the other three were above 50 years of age. Most of these patients were critically ill and a single etiology could not be identified as a cause except in the patient with hepatitis B co-infection, who had acute fulminant hepatitis B infection but only mild COVID-19 pneumonia. Out of the five ALF patients, two survived, two expired, and one remained critically ill at the time of writing.
SARS-CoV-2 and hepatitis B
Hepatitis B virus (HBV) infection rates among patients with COVID-19 have been reported between 2.1% and 12.2% from China. Zou et al.48 reported their clinical experience of 20 patients with COVID-19 and chronic HBV co-infection in a retrospective analysis, noting its severe illness and poor prognosis compared to 306 patients with only COVID-19 infection. They reported significantly lower prealbumin levels but no difference in levels of liver enzymes, length of hospital stay or discharge rates. Chen et al.,49 in their retrospective analysis of 123 cases, including 15 cases with HBV, reported more severe disease in HBV-COVID-19 coinfection compared to HBV-negative cases (46.7% vs. 24.1%) as well as a higher mortality rate (13.3% vs. 2.8%). Zha et al.,50 in their observational study of 31 cases, had 2 patients with HBV infection and found that they took a longer time to clear COVID-19 infection (mean difference of 10.6 days). Aldhaleei et al.51 reported a case of hepatitis B flare due to COVID-19, although large-scale studies are required to validate these findings. There is a single case report of HBV induced ALF in a patient with mild SARS-CoV-2 infection.47
SARS-CoV-2 and hepatitis C
There is very limited data on hepatitis C virus (HCV) and COVID-19. Wang et al.,52 in their case-control study of over 1 million patients with cirrhosis, included 16,530 with COVID-19 and 820 with COVID-19 and chronic liver disease (CLD) and reported higher odds for patients with HCV in acquiring COVID-19 than for those without [adjusted odds ratio of 12.9]. Thus, although these findings would support the notion that patients with HCV-related CLD are at a greater risk for acquiring COVID-19 infections, there is a dearth of data to validate this finding or identify the impact of COVID-19 on disease course, management and outcome.
SARS-CoV-2 and alcoholic liver disease
No studies have looked exclusively at outcomes of patients with alcoholic liver disease (ALD) with COVID-19. However, a retrospective study from our center reported that the fraction of patients with ALD had decreased in the early part of the pandemic compared to the pre-pandemic era, likely as a result of total lockdown imposed in India and decreased alcohol availability.53,54 However, the outcome of these patients was not different from those with other etiologies.53 Following lifting of the lockdown and increased availability and sale of alcohol, a center from the UK reported doubling of admissions due to ALD and an increase in the proportion of patients with severe alcoholic hepatitis and alcohol-related acute-on-chronic liver failure (ACLF).55
SARS-CoV-2 and autoimmune liver disease
Data on the impact of COVID-19 on primary biliary cirrhosis (PBC), primary sclerosing cholangitis (PSC) or autoimmune hepatitis (AIH) are evolving. Since COVID-19 is associated with transaminitis and hyperbilirubinemia, it may be confused with a flare of AIH. Thus, a liver biopsy may be mandated to confirm the diagnosis prior to initiation of therapy.56 Gerussi et al.,57 in their case series, described 10 patients across seven hospitals in Italy who were undergoing immunosuppression for AIH. Of the 10 patients, 2 had a recent flare for which they were on high dose steroids. Liver enzymes remained normal in all cases and improved in the two acute cases.
A recently published large retrospective study of 70 patients with AIH, 19 with PBC, 19 with PSC and 16 with variant syndromes were compared in a propensity-matched analysis to 862 non-AIH CLD and 769 patients without liver disease.58 The cohort with AIH had no increase in ICU stay or mortality compared to patients with other liver disease etiologies or those without liver disease, although close to 80% of AIH patients were on immunosuppression.58 The only significant factors for AIH mortality were age and baseline liver disease. The authors hypothesized that despite immunosuppression, patients with AIH have preserved immune responses to SARS-CoV-2 and hence are not at a disadvantage. This large study reassures patients and physicians alike and strengthens the already prevalent recommendation of not changing these patients’ immunosuppressive therapy in the pandemic.58
SARS-CoV-2 and non-alcoholic fatty liver disease
Two large retrospective studies show that presence of non-alcoholic fatty liver disease (NAFLD) is a risk factor for development of severe COVID and mortality even after correcting for comorbidities, such as obesity and diabetes.59,60 The authors hypothesized that the increased progression of COVID-19 in patients with NAFLD might be due to either exaggerated hepatic immune response contributing to systemic inflammation or the prothrombotic state in these patients contributing to disease progression.61 However, a third large retrospective study failed to reach similar conclusions, possibly because of different criteria used to define COVID-19 progression and a higher fraction of patients with diabetes (50%) compared to the previous two studies, which might have negated the effect of NAFLD on multivariate analysis.62
SARS-CoV-2 and DILI
Although liver injury might occur in patients infected with SARS-CoV-2 due to many reasons, DILI should be considered among the important differentials of liver injury in these patients. The commonly implicated drugs include those which have been repurposed for use in COVID-19, such as hydroxychloroquine (HCQ), azithromycin, lopinavir/ritonavir, baricitinib, and those which been developed exclusively for COVID-19, such as remdesivir. Idiosyncratic DILI is a well-known but rare adverse effect of HCQ.63 Azithromycin is also associated with rare idiosyncratic cholestatic hepatitis.63 Lopinavir has been shown to cause both hepatocellular and cholestatic liver injury, leading to enzyme elevation up to > 5-times the ULN in 3% to 10% of patients.63 The current information on remdesivir suggests that is an unlikely cause of clinically significant liver injury, as suggested by healthy volunteer studies and controlled studies.64 The safety data on favipiravir, although sparse, appears to be reassuring.65 Tocilizumab and other interleukin (IL)-6 antagonizing therapies, although frequently associated with elevated aminotransferases (10% to 50%), are rarely associated with elevations > 5-times the ULN (1–2%).66 Tocilizumab and other immunosuppressants used in the treatment of COVID-19 are also theoretically associated with the risk of reactivation of viral hepatitis.66 Baricitinib is an unlikely cause of DILI but has been associated with risk of reactivation of hepatitis B.67
Acute decompensation and ACLF
Patients with CLD and cirrhosis have systemic immunodeficiency, which places them at a higher risk for COVID-19. Data available from registries place the number of new decompensations at 45% and the mortality rate in such patients at 40%, which is higher in patients with advanced liver disease.53,68,69 The clinical presentations include acute decompensation-jaundice, ascites, hepatic encephalopathy, and GI bleed.53,68 Severe presentation includes ACLF with organ failure.
In their multicentric study, Bajaj et al.68 reviewed 37 patients with cirrhosis and COVID-19 compared to a cohort of 127 patients with cirrhosis alone and 108 patients with COVID-19 alone. ACLF, as per North American Consortium for the Study of End-Stage Liver Disease (known as the NACSELD) criteria, was seen in 40 patients, with 11 in the cirrhosis-COVID-19 group and 29 in the COVID-19 group alone and with no difference in mortality across both groups (55% vs. 36%, p=0.25). The authors also reported higher mechanical ventilation and non-invasive ventilation use requirements, central line placement and ICU transfer in the cirrhosis and COVID-19 group compared to the cirrhosis only group. A study from our center compared 28 patients with cirrhosis and COVID-19 with 78 historical controls with cirrhosis matched for etiology and model for end-stage liver disease (commonly referred to as MELD) score. The overall mortality rate was higher in the cirrhosis and COVID-19 group, at 42.3% vs. 23.1%, p=0.07. The mortality was even higher in the sicker group with ACLF and COVID-19, 100% vs. 53.3%, p=0.01.53
Variceal bleeding
Although the data on upper gastrointestinal (referred to herein as UGI) bleeding in patients with COVID-19 continues to evolve, there are limited data on variceal bleed in patients with cirrhosis and COVID-19. In a study from our center evaluating the outcomes of cirrhosis in COVID-19 infection, variceal bleeding was the most common form of decompensation present in 11/16 (68%) of the patients.53 In another study from our center, UGI bleeding was present in 24/1,342 (1.8%) of all patients hospitalized with COVID-19.70 The majority (88%) of bleeding episodes represented variceal bleeds in patients with cirrhosis and had encouraging outcomes with no rebleed or death at 5 days with primary conservative management.70
Hepatocellular carcinoma
The presence of hepatocellular carcinoma (referred to herein as HCC) is associated with poor outcomes, with an increased risk of overall and COVID-19-related mortality in patients with CLD and COVID-19 infection.71 In addition, the COVID-19 pandemic has also affected the standard of care of patients with HCC. In a large retrospective study, including more than 600 patients, a lower number of patients were evaluated during the pandemic period compared to the same period prior to the pandemic. More than 20% of patients experienced a treatment delay and 13.1% needed a modification in the treatment strategy, both attributable to the COVID-19 pandemic.72
Liver transplant
A multicentric registry reported outcomes of patients with liver transplants (n=151) compared to controls (n=627).73 GI symptoms (nausea, vomiting, abdominal pain, and diarrhea) were experienced by a greater proportion of patients in the transplant cohort than the comparison cohort- 30% vs. 12%, p<0.001. There was no difference in respiratory symptoms experienced (77% vs. 81%, p=0.248) or hospitalization rates (82% vs. 76%, p=0.106) between the two groups. However, the rates of ICU admission (28% vs. 8%, p<0.001) and the proportion receiving invasive ventilation (20% vs. 5%) were higher, and median hospital stay (11 days vs. 8 days, p=0.046) was longer in the liver transplant group. Surprisingly, the proportion of deaths in the transplant cohort was significantly less than the comparison cohort (19% vs. 27%, p=0.046) with the dominant cause of death being COVID-19 lung disease. The authors also reported no liver-related mortality, rejection, or re-transplant in the transplant group. Similar outcomes have been reported from a prospective study from Spain (18%) and UK national registries (20%).73,74 The Spanish study reported on the prospective follow-up of 111 post-transplant recipients and showed that although chronically immunosuppressed patients are at increased risk of acquiring the infection yet, they are not at increased risk of mortality.74 The analysis also reported no effect on immunosuppression on mortality, particularly calcineurin inhibitors and mammalian target of rapamycin (commonly known as mTOR) inhibitors, except mycophenolate, particularly in doses greater than 1 g per day (relative risk of 3.94).74 The authors hypothesize that this effect might be due to the CD8+ depleting effect of mycophenolate.74
SARS-CoV-2 cholangiopathy
ACE-2 receptor and TMPRSS2 are highly expressed on cholangiocytes, and hepatic organoid models have been used to show the virus’s direct cytopathic effect on cholangiocytes.4 Recently, a small case series described an entity called post-COVID-19 cholangiopathy that is characterized by changes in both extrahepatic and intrahepatic biliary tree with microscopic features of severe vacuolization injury to cholangiocytes, along with microangiopathy and evidence of developing secondary biliary cirrhosis among three patients who initially had severe elevation of liver enzymes and acute hypoxemic respiratory failure, and prolonged hospitalization due to COVID.75 However, the exact contribution of SARS-CoV-2 in the development of cholangiopathy is unclear, since a similar entity has been demonstrated in critically ill hospitalized patients. None of the patients in the above series had immunohistochemical evidence of SARS-CoV-2 infection on liver biopsy samples.76
Gallbladder
Few case reports exist which describe patients presenting with COVID-19 and acute cholecystitis with the disease attributed to the virus on the basis of positive quantitative reverse transcription-polymerase chain reaction (referred to herein as RT-PCR) of tissue samples in one patient, while the other two had positive RT-PCR results for nasopharyngeal swabs.77–79 All three patients had acalculous cholecystitis with positive Murphy’s sign, thickening of gallbladder wall and pericholecystic fluid on ultrasound. However, the significance of these findings in the pathogenesis of COVID-19 needs to be elucidated.
Evaluation of patients with liver manifestations of COVID-19
Patients presenting with acute febrile illness and respiratory symptoms, such as sore throat, nasal stuffiness, dry cough and breathlessness, should undoubtedly be evaluated for COVID-19. Patients presenting with transaminitis, either symptomatic or asymptomatic, should also be offered testing for COVID-19 apart from standard tests for viral hepatitis, autoimmune markers, copper studies and metabolic panel. Similarly, patients with underlying liver disease presenting with new decompensation or ACLF should also be tested for COVID-19.53
Despite the ongoing pandemic, the scourge of tropical diseases, such as dengue, malaria, chikungunya, typhoid, tuberculosis, and scrub typhus, should not be forgotten, as they may share certain symptoms with COVID-19 but treatment and prognosis differ. Moreover, co-infections with COVID-19 and these tropical illnesses have been frequently reported.20–22 Hence, in patients presenting with acute febrile illness, these differentials should also be considered, apart from COVID-19.