Elevated Liver Biochemistries in Hospitalized Chinese Patients With Severe COVID‐19: Systematic Review and Meta‐analysis

Abstract

Background and Aims

Several recent studies have reported an abnormal liver chemistry profile among patients with coronavirus disease 2019 (COVID‐19), although its clinical significance remains unknown.

Approach and Results

This systematic review and meta‐analysis identified six studies of 586 patients delineating liver chemistries among patients with severe/critical illness versus mild cases of COVID‐19 infection. Patients with severe/critical illness with COVID‐19 infection have increased prevalence of coronary artery disease, cerebrovascular disease, and chronic obstructive pulmonary disease as compared with mild cases. A significant association between severe/critical COVID‐19 infections with elevations in aspartate aminotransferase (pooled mean difference [MD], 11.70 U/L; 95% confidence interval [CI], 2.97, 20.43; P = 0.009), elevated total bilirubin (pooled MD, 0.14 mg/dL; 95% CI, 0.06, 0.22; P = 0.0005), and decreased albumin (pooled MD, −0.68 g/L; 95% CI, −0.81, −0.55; P < 0.00001) was noted. There was also a trend toward elevated alanine aminotransferase levels among these severe cases (pooled MD, 8.84 U/L; 95% CI, −2.28, 19.97; P = 0.12); however, this did not reach statistical significance. More severe/critically ill cases were associated with leukocytosis, neutrophilia, lymphopenia, elevated creatinine kinase, elevated lactate dehydrogenase (LDH), and elevated prothrombin time (PT).

Conclusions

Comorbidities, including coronary artery disease, cerebrovascular disease and chronic obstructive pulmonary disease, are more prevalent in hospitalized Chinese patients with severe/critical illness from COVID‐19, and these patients are more likely to manifest with abnormal liver chemistries. Further prospective studies are crucial to understand the pathophysiologic mechanisms underlying the hepatic manifestations of the novel COVID‐19 infection and its clinical significance.

Abbreviations

ALT

alanine aminotransferase

AST

aspartate aminotransferase

CI

confidence interval

COVID‐19

coronavirus disease 2019

df

degrees of freedom

ICU

intensive care unit

LDH

lactate dehydrogenase

M‐H

Mantel‐Haenszel

MD

mean difference

OR

odds ratio

PT

prothrombin time

SARS

severe acute respiratory syndrome

WBC

white blood count

After recently being declared a global pandemic, the novel severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2), or coronavirus disease 2019 (COVID‐19), has spread to more than 200 countries with close to 2 million confirmed cases in the United States alone.^{( 1 )} Although this infection typically presents with an incubation period followed by pulmonary manifestations, the presence of concomitant liver injury among patients with COVID‐19 has recently been described.^{( 2 )} However, the significance of liver injury remains unknown.

It is still relatively early in the evolution of the COVID‐19 pandemic. The exact pathophysiologic mechanism remains uncertain, especially regarding the hepatobiliary involvement of this disease. The primary aim of this systematic review and meta‐analysis is to further characterize liver injury and other unique clinical characteristics across COVID‐19 severity.

Materials and Methods

Literature Search

We searched three major databases—MEDLINE/PubMed, Embase, and Cochrane Central Register of Controlled Trials—for clinical studies dated from inception to March 22, 2020. In an effort to broadly identify studies detailing liver chemistry testing among patients with COVID‐19, the following search criteria were used: “coronavirus,” “cov2,” “cov 2,” “ncov,” “sars” and “cov2,” “sars cov 2,” “covid” and “liver,” “ast,” “aspartate aminotransferase,” “alt,” or “alanine aminotransferase.” This meta‐analysis was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta‐analyses^{( 3 )} and Meta‐analyses of Observational Studies in Epidemiology.^{( 4 )} Identification of studies reporting liver chemistries were prioritized to ascertain an association among patients who had severe or critical illness versus mild cases of COVID‐19 infection. Severe cases were defined as significant respiratory distress from the COVID‐19 infection based on work of breathing or degree of hypoxia (diagnosed by either pulse oximetry or arterial blood gas). Critical cases were defined as patients requiring admission to the intensive care unit (ICU) for either mechanical ventilation, shock, or multisystem organ failure. Mild cases were defined as patients who tested positive for COVID‐19, but did not meet criteria for severe or critical cases and were either asymptomatic or the symptoms from their infection were self‐limiting.

Inclusion Criteria

Articles and clinical studies that met the following inclusion criteria were eligible for this meta‐analysis: (1) studies performed in adult, human subjects; (2) studies written in English; (3) studies measuring liver chemistries among patients with COVID‐19 (aspartate aminotransferase [AST], alanine aminotransferase [ALT], total bilirubin); and (4) studies comparing patients with severe or critical illness versus mild cases.

Exclusion Criteria

Studies with the following characteristics were excluded from this meta‐analysis: (1) studies in nonhuman subjects; (2) studies written in a language other than English; (3) studies that were not a clinical trial, such as a review paper or letter; (4) studies that were out of scope of the study question detailed previously; (5) studies that lacked proper controls; (6) studies that did not provide raw data to perform quantitative meta‐analysis; (7) studies conducted in pediatric subjects; (8) studies that did not have an available manuscript; (9) studies that were duplicates; and (10) studies there were ongoing or not completed.

Outcomes and Endpoints

The primary outcomes for this meta‐analysis were as follows: (1) liver chemistries (AST, ALT, total bilirubin, and albumin); (2) demographic data (age, gender, and comorbidities); (3) laboratory parameters from complete blood count, comprehensive metabolic panel, or coagulation testing (white blood count [WBC], neutrophils, lymphocytes, platelets, creatinine, creatinine kinase, lactate dehydrogenase [LDH], prothrombin time [PT], and activated partial thromboplastin time [aPTT]); and (4) mortality.

Data Extraction

The authors extracted data from the literature search, and any disagreements were resolved by consensus. Ambiguity in the reported data was attempted to be resolved by e‐mailing the corresponding author of the study where appropriate. Author names, dates, study type, setting of study, number and characteristics of patients, and treatment outcomes were gathered for all included studies.

Risk of Bias

Nonrandomized studies were assessed by the Newcastle‐Ottawa Scale (NOS).^{( 5 )} The authors determined the quality of selection, comparability, and exposure/outcome in the NOS. Each form of bias was awarded either a low, high, or unclear risk of bias. A funnel plot was calculated to determine the presence of publication bias.

Statistical Analysis

This meta‐analysis was performed using Review Manager (RevMan, version 5.3; The Cochrane Collaboration, Copenhagen, Denmark). Odds ratios (ORs) were used as a summary measure of efficacy for dichotomous data, and mean differences (MDs) were used between groups for continuous variables; 95% confidence intervals (CIs) were reported for both measures. If medians and interquartile ranges were provided in the included studies instead of MDs and SDs, the MDs and SDs were imputed as described in other studies^{( 6} , ⁷ , ^{8 )} as well as the Cochrane Handbook for Systematic Reviews of Interventions.^{( 9 )} Data were considered statistically insignificant if P > 0.05, OR includes 1.00, or MD includes 0. Statistical heterogeneity was assessed using the I ² statistic. I ² values of 0% to 25%, 25% to 50%, 50% to 75%, and more than 75% were awarded values of homogeneity, mild heterogeneity, moderate heterogeneity, and high heterogeneity, respectively. If significant heterogeneity was present (I ² ≥ 50%), the random effects model was used to pool the effect sizes of included studies and subgroup analyses; if no significant heterogeneity was found (I ² < 50%), the fixed‐effect model was used.^{( 10} , ^{11 )} Publication bias was determined by Egger’s regression asymmetry test.^{( 12 )} Funnel plots were qualitatively assessed by visual inspection based on logarithmic ORs plotted versus their standard errors.^{( 13 )} Asymmetric funnel plots represented studies with high risk of publication bias.

Results

Study Selection

The data search, literature review, and study selection are outlined in Fig. 1. Overall, 455 clinical studies were identified based on the predefined inclusion and exclusion criteria. Studies that did not have an appropriate clinical comparison between groups or present enough data to perform a quantitative meta‐analysis were excluded. In the end, six studies with 586 patients were finalized that satisfied all of these criteria.^{( 14} , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ^{19 )} Also of note, all of these studies present data from Chinese patients who tested positive for COVID‐19. Five are from Wuhan and the province of Hubei. The final study details a patient cohort from the neighboring province of Chongqing.^{( 16 )} All data were included from hospitalized patients, with laboratory data collected and analyzed upon presentation. Quality assessment of the included studies is provided in Supporting Table S1.

Fig. 1.

Study flow diagram.

Higher Burden of Increased Age and Comorbidities Among Patients With Severe or Critical COVID‐19 Infection

The characteristics of patients with severe/critical illness versus mild cases of COVID‐19 are shown in Fig. 2 and Supporting Fig. S1. Comorbidities were not uniformly reported in all six studies. Overall, severe/critical cases of COVID‐19 infection were significantly associated with patients who had coronary artery disease (CAD) (pooled OR, 2.68; 95% CI, 1.43, 5.01; P = 0.002), cerebrovascular disease (pooled OR, 20.20; 95% CI, 2.34, 174.44; P = 0.006), and chronic obstructive pulmonary disease (COPD) (pooled OR, 12.22; 95% CI, 2.57, 58.09; P = 0.002) at baseline as compared to patients with mild cases. Increased age (pooled MD, 11.15; 95% CI, −1.75, 24.0; P < 0.09) and diabetes mellitus (pooled OR, 2.34; 95% CI, 0.66, 8.26; P = 0.19) demonstrated a trend toward association with severe/critical COVID‐19 cases; however, these differences did not reach statistical significance. These studies failed to show a significant link of severe cases with gender, hypertension, active smoking or tobacco use, chronic liver disease, chronic kidney disease, malignancy, or current treatment with angiotensin‐converting enzyme inhibitor/angiotensin 2 receptor blocker. However, these studies did not delineate the definition of “chronic liver disease” into the underlying etiology, such as cirrhosis or viral hepatitis.

Fig. 2.

Forest plot comparison of demographic data with respect to gender, hypertension, diabetes mellitus, coronary artery disease, and cerebrovascular disease [Figure 2A] as well as smoking, chronic obstructive pulmonary disease, chronic liver disease, chronic kidney disease, malignancy, and ACEi/ARB treatment [Figure 2B] among patients with severe/critical illness versus mild COVID‐19 infection.

Elevated Liver Chemistries Are More Pronounced Among Patients With COVID‐19 With Severe or Critical Illness

The comparison of liver chemistries among patients with severe/critical illness versus mild COVID‐19 cases is provided in Fig. 3. Overall, there was a significant correlation of severe or critical COVID‐19 infections with elevations in AST (pooled MD, 11.70; 95% CI, 2.97, 20.43; P = 0.009), elevated total bilirubin (pooled MD, 0.14; 95% CI, 0.06, 0.22; P = 0.0005), and decreased albumin (pooled MD, −0.68; 95% CI, −0.81, −0.55; P < 0.00001) upon presentation. There was also a trend toward association of severe/critical cases with elevated ALT (pooled MD, 8.84; 95% CI, −2.28, 19.97; P = 0.12). However, this did not reach statistical significance.

Fig. 3.

Forest plot comparison of liver chemistries among patients with severe/critical illness versus mild COVID‐19 infection. Abbreviation; IV, inverse variance.

Other Laboratory Aberrancies Associated With Severe or Critical COVID‐19 Infection

A more comprehensive battery of laboratory testing provided among these studies is further detailed in Supporting Fig. S2. Overall, more severe or critical illness from COVID‐19 was statistically associated with leukocytosis (pooled MD of WBC, 2.14; 95% CI, 0.20, 4.08; P = 0.03), neutrophilia (pooled MD, 1.68; 95% CI, 0.38, 2.97; P = 0.01), lymphopenia (pooled MD, −0.40; 95% CI, −0.58, −0.22; P < 0.001), elevated creatinine kinase (pooled MD, 39.95; 95% CI, 20.94, 58.95; P < 0.0001), elevated LDH (pooled MD, 96.39; 95% CI, 65.92, 126.87; P < 0.0001), and elevated PT (pooled MD, 0.48; 95% CI, 0.25, 0.72; P < 0.0001). Platelet count appeared to be decreased among more severe infections; however, this did not achieve statistical significance (pooled MD, −18.36; 95% CI, −50.22, 13.51; P = 0.26). No significant relationship was established between serum creatinine or aPTT among patients with severe/critical illness versus mild cases.

Mortality Data Among Severe or Critical Illness in Patients With COVID‐19 Infection

Mortality data are presented in Fig. 4. Overall, only three studies reported mortality data. Although there was no outpatient analysis for these patients, the mortality data included follow‐up throughout the duration of their inpatient stay. Unsurprisingly, there was a significantly increased rate of mortality (pooled OR, 21.70; 95% CI, 7.01, 67.24; P < 0.00001) in studies reporting survival among patients with severe versus mild COVID‐19 infections.

Fig. 4.

Forest plot comparison of mortality reported among patients with severe/critical illness versus mild COVID‐19 infection.

Publication Bias

Funnel plots were generated for the outcomes of this meta‐analysis and are demonstrated in Supporting Figs. S3‐S7. Where appropriate, 95% CIs were included for fixed‐effect model analysis. Overall, there was no asymmetry noted in the funnel plots to suggest a large degree of publication bias.

Discussion

As the current pandemic continues to sweep the globe, there will inevitably be an increasing amount of clinical data from Europe and North American regarding liver injury among patients with COVID‐19; however, at the time of this meta‐analysis, the data have come exclusively from China. There remains a paucity of data on liver histopathology, and limited autopsy case reports reveal microvesicular steatosis in addition to lobular and portal inflammation, although not as exaggerated as inflammation seen in the lungs.^{( 20 )} Our study shows that abnormalities of liver chemistries are more common in severe COVID‐19 disease and more likely to be seen in those with other serious comorbidities. Based on circumstantial evidence, the Centers for Disease Control and Prevention and the American Association for the Study of Liver Diseases Expert Panel has suggested that there will be a heightened health care burden among patients with chronic liver diseases, especially those requiring liver transplantation.^{( 21} , ^{22 )} Our study appears to support this statement, as patients with nonalcoholic steatohepatitis cirrhosis or those waiting liver transplant are likely to have significant comorbidities.

Several studies from Europe and North America have emerged recently.^{( 23} , ^{24 )} Among 5,700 hospitalized patients in New York, only 19 total patients (0.4%) had cirrhosis, with only 11 patients (0.2%) described as having either hepatitis B or C virus.^{( 23 )} Another Italian study reports chronic liver disease only among 28 of 1,591 patients admitted to the ICU.^{( 24 )} Mildly increased liver chemistries is a common finding, particularly among severe COVID‐19 infections, although it has been previously demonstrated to lack significant, standalone impact on mortality.^{( 25} , ^{26 )} In contrast, a large propensity‐matched study has reported that patients with chronic liver disease and COVID‐19 have a 3‐fold higher mortality risk as compared to patients without liver disease.^{( 27 )} Conversely, two other meta‐analyses have recently failed to demonstrate a link between chronic liver disease and either mortality or severity of COVID‐19 infection.^{( 28} , ^{29 )}

Mechanism of liver injury and its relationship to severity of COVID‐19 infection is largely speculative. Similar to previous SARS‐CoV infections, the binding and uptake by the angiotensin‐converting enzyme 2 (ACE2) cholangiocyte receptor has been demonstrated to mediate liver damage among patients with COVID‐19.^{( 30 )} Not only does the degree of AST or ALT elevation appear to correlate with the severity of COVID‐19 infection, but this mechanism for liver injury appears to be mediated by an up‐regulation of ACE2 also among hepatocytes in a mouse model.^{( 31 )} Other etiologies for liver injury include underlying liver disease, medications including experimental drugs, and in severe cases, the cytokine storm (through tumor necrosis factor α, IL‐6, or IL‐18) as well as the ischemic hypoxia‐reperfusion that accompanies more severe COVID‐19 infections.^{( 32 )}

Although this manuscript reports findings in this area, it is not without limitations. First, our analysis is limited to hospitalized patients from China, and hence cannot be generalizable to other geographical areas. Studies detailing liver chemistries and other clinical characteristics remain extremely limited for the outpatient population. In our study, significant heterogeneity was found with respect to outcomes of age (I ² = 95%), history of hypertension (I ² = 82%), history of diabetes mellitus (I ² = 68%), AST (I ² = 78%), ALT (I ² = 81%), WBC (I ² = 86%), neutrophil count (I ² = 70%), lymphocyte count (I ² = 77%), platelet count (I ² = 69%), and serum creatinine (I ² = 63%). This phenomenon is most likely related to small sample size, limited study data, and skewed representation of patient populations. Another potential limitation is the definition used to stratify severe/critical COVID‐19 disease. Some studies included only patients requiring ICU admission, whereas others included severe respiratory distress not requiring intubation in their severe patient cohort. Laboratory parameters were all collected from patients upon presentation for their inpatient admission to the hospital. A more comprehensive trend of liver chemistries may be a future focus, to capture the peak of these laboratory parameters and compare them based on clinical outcomes. Furthermore, characterization of baseline “chronic liver disease” remains ill‐defined among these studies and warrants further investigation. Many of these laboratory findings, such as aberrancies in liver chemistries, low platelet count, decreased albumin, or prolonged PT, could be explained by an acute COVID‐19 infection, but also may be influenced by underlying cirrhosis or viral hepatitis that is not clearly reported among these included studies.

The characterization of liver injury among COVID‐19 infection remains paramount with respect to potential therapeutic options. Hydroxychloroquine has been used in many countries as a potential therapeutic option against COVID‐19,^{( 33 )} and although hepatotoxicity is relatively uncommon, this drug is metabolized by the liver and has been reported to cause acute liver injury in select cases.^{( 34 )} Furthermore, the use of antiviral therapies has undergone increased investigation. Remdesivir, a novel antiviral agent that acts as an RNA‐chain terminator, has been demonstrated to be effective in vitro, and is in early clinical trials in the United States.^{( 35 )} Recent published studies on remdesivir have shown infrequent liver‐test abnormalities among other side effects.^{( 36} , ³⁷ , ^{38 )} However, elevation of AST or ALT greater than 5 times the upper limit of normal remains an exclusion criterion for initiation of remdesivir among patients with COVID‐19, and this may exclude many patients with preexisting liver disease from receiving remdesivir. The true incidence of remdesivir hepatotoxicity is an area for further research. Preliminary studies of IL‐6 inhibitor, tocilizumab, in patients with COVID‐19 also suggest that this drug also may be associated with uncommon hepatotoxicity.^{( 39 )} Among preliminary data from a study reporting the first hospitalized cases of COVID‐19 in the United States, all patients treated with remdesivir developed transaminitis during their clinical course.^{( 40 )}

As demonstrated in this systematic review and meta‐analysis, the patients suffering from more severe or critical COVID‐19 infections have significantly increased liver chemistries. This clinical finding poses quite a dilemma for future patients infected with COVID‐19, because many of the current investigational or ongoing randomized controlled trials are excluding patients with values of AST or ALT greater than 5 times the upper limit of normal.^{( 41} , ⁴² , ^{43 )} Therefore, further characterization and recognition of liver injury is vital among patients harboring severe/critical COVID‐19 infections, particularly if one of the few therapeutic options is predicated on the status of underlying liver chemistries.

As the COVID‐19 pandemic continues to afflict more individuals across the globe, both the incidence and clinical significance of liver injury among this patient population will be important to analyze. In this systematic review and meta‐analysis, the characterization of severe or critically ill patients, as compared with individuals suffering from mild cases of COVID‐19, is described in detail. There appears to be a significant correlation of aberrant liver chemistries, in addition to CAD, cerebrovascular disease, COPD, leukocytosis, neutrophilia, lymphopenia, elevated creatinine kinase, elevated LDH, and increased PT, among patients with severe or critical illness from COVID‐19 as compared with mild cases. Current knowledge of contributing factors, such as underlying liver disease or use of concomitant medications, remains unknown. It is imperative for future prospective studies to continue to examine the role of multiple confounders in the pathogenesis of liver injury in COVID‐19 infection.

Author Contributions

Alexander Kovalic provided substantial contributions to conception and design of the study, acquisition of data or analysis and interpretation of data, drafting the article or revising it critically for important intellectual content, and final approval of the version to be published. Glen Huang provided substantial contributions to acquisition of data or analysis and interpretation of data, drafting the article or revising it critically for important intellectual content, and final approval of the version to be published. Paul Thuluvath and Sanjaya Satapathy provided substantial contributions to conception and design of the study, analysis and interpretation of data, revising the article critically for important intellectual content, and final approval of the version to be published.

Supporting information

Supporting Information