Introduction

Since the coronavirus disease 2019 (COVID-19) outbreak began in December 2019, the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has quickly spread worldwide¹. COVID-19 presentation is heterogeneous and ranges from asymptomatic disease to severe illness characterized by pneumonia, acute respiratory distress syndrome (ARDS), and sepsis that may eventually lead to death². Despite the concerted efforts of health systems around the globe, mortality attributable to COVID-19 remains considerably high with case fatality rates as high as 9-11%³. Early medical intervention in seriously and critically ill COVID-19 patients may reduce fatal outcomes⁴. Therefore, there is an urgent need for novel biomarkers that can help clinicians identify patients with poorer prognosis. Early implementation of more aggressive drug regimens in these patients could not only increase survival rates but also optimize intensive care unit (ICU) resources.

Due to the extensive extrapulmonary manifestations of COVID-19, evaluation of patients typically includes comprehensive blood panels that allow the clinician to simultaneously estimate the effects of the disease on multiple organ systems⁵. Liver function tests (LFT) are commonly included in laboratory analyses and measure a variety of molecules such as transaminases, alkaline phosphatase, gamma-glutamyl transferase, bilirubin, and albumin⁶ that indicate the global function and integrity of the liver.

Large-scale descriptive studies have been conducted to examine the relationship between altered liver function and SARS-CoV-2 infection, but only a few of them describe factors that might be used to predict the most adverse outcomes of COVID-19, including ARDS development, ICU admission, and/or death⁷-⁹. Herein, we summarize the most promising liver biomarkers for COVID-19 prognosis in an effort to help clinicians rapidly, easily, and inexpensively predict the risk of ARDS, ICU admission, and death at hospital admission.

Serum albumin

Serum albumin (SA) is the most abundant protein in human blood (35 to 50 g/L)¹⁰. SA is synthesized in hepatocytes as preproalbumin and then released to the rough endoplasmic reticulum, where an N-terminal peptide is removed to facilitate the conversion of the protein to proalbumin. Proalbumin is in turn processed in the trans-Golgi apparatus to form mature albumin that is subsequently released into circulation¹¹. SA plays multiple physiological roles, including regulation of plasma colloid oncotic pressure, transportation of endogenous and exogenous molecules, anticoagulation, and antioxidation responses¹⁰. In clinical practice, SA level is commonly used as an indicator of mortality risk in severely ill patients such as those experiencing sepsis, trauma, or cancer¹².

Numerous studies support the use of SA as a biomarker of poor prognosis in COVID-19 patients, especially those at much higher risk of ICU admission or death. Three meta-analyses that evaluated several laboratory parameters reported decreased SA at admission in patients with severe COVID-19 (3.50 g/dL, 95% CI 3.26–3.74 g/dL) as compared to non-severe patients (4.05 g/dL 95% CI 3.82–4.27 g/dL)¹³-¹⁶. Hypoalbuminemia (SA < 30 g/L) also correlates with the severity of COVID-19 (p < 0.0001) in patients from Italy¹⁷. Two additional meta-analyses showed that SA levels correlate with disease severity but fail to find associations among other liver parameters⁹,¹⁶.

When considering these liver markers in the context of critical care, it is noteworthy that ICU patients with COVID-19 generally have lower SA values at admission than patients admitted to regular hospital floors (31.3 ± 5.2 g/L versus 36.5 ± 5.7 g/L; p = 0.001). In fact, SA predicts ICU admission with a HR score of 1.87, 95% CI 1.05–3.32, p = 0.034¹⁸. A study conducted in 63 COVID-19 patients from Turkey revealed that SA at admission predicts ICU admission with a cutoff point of ≤ 36 g/L, AUC of 0.989 (95% CI 0.924–1.00; p < 0.001), sensitivity of 96.7%, and specificity of 93.9%¹⁹. On the other hand, a study of 128 COVID-19 patients from Dubai showed that SA failed to predict ICU admission (AUC 0.256, 95% CI 0.146-0.366)²⁰. Thus, the use of SA as a predictor of ICU admission in seriously ill patients with COVID-19 should be performed with caution, and clinicians must take into account the genetic background and variability among the patient populations with which they work.

Regarding mortality, a meta-analysis evaluating eleven studies predominantly conducted in China indicated that SA values tend to be lower in non-survivors than survivors (-3.7 g/L, 95% CI -5.3 – -2.1; p < 0.00001)²¹. A separate study conducted on 207 patients from Italy showed that hypoalbuminemia correlates with death in COVID-19 patients (p = 0.003)¹⁷. Another study in 319 patients from Italy concluded that SA < 32 g/L is a risk factor for mortality, with HR of 2.48 (95% CI 1.44–4.26; p = 0.001)¹⁸. Concurrently, two additional studies conducted in COVID-19 patients from China revealed that decreased SA at admission was an independent risk factor for mortality (OR = 1.929, 95% CI 1.199–3.104, p = 0.007, and OR, 6.394; 95% CI 1.315-31.092,p = 0.021, respectively)²². This evidence supports the idea that SA is an independent risk factor for ICU admission and death in patients that meet hospitalization criteria for COVID-19. As we have shown, critical consideration of SA levels may help clinicians identify patients with SARS-CoV-2 infection who are at much higher risk of developing adverse outcomes than other patients. Having a reliable and easy-to-detect biomarker at their disposal would enable clinicians to provide these high-risk patients with more aggressive medical treatments that will increase their probability of survival.

Bilirubin

Bilirubin derives from the breakdown of the heme group in hemoglobin, which occurs in the spleen, bone marrow, and liver²³. In these tissues, heme oxygenase, an enzyme found in macrophages, can catalyze iron release to form carbon monoxide and biliverdin. Biliverdin is in turn reduced to form unconjugated bilirubin, also referred to as indirect bilirubin, which is poorly water soluble²⁴. The 200-300 mg of bilirubin produced by the human body per day is typically excreted. Excretion of bilirubin occurs when indirect bilirubin (IBIL) is released into the plasma, where it binds albumin that is in turn transported to the liver²⁵. Then, bilirubin is conjugated to uridine diphosphate sugars (UDP) via UDP-glucuronyltransferase. This increases IBIL solubility and allows its excretion through the bile. Conjugated bilirubin, also referred to as direct bilirubin (DBIL), is broken down into urobilinogens by intestinal bacteria. At this point, a small quantity of DIBIL can be deconjugated and re-absorbed by intestinal epithelial cells²³. Any increase in the rate of release of the heme group or failure to completely excrete bilirubin leads to a concomitant elevation of total bilirubin, direct bilirubin, or indirect bilirubin depending on the step of bilirubin metabolism that is impaired²⁵.

Common clinical laboratory tests include measures of total bilirubin (TBIL), IBIL, and/or DBIL, which can help establish the cause of hyperbilirubinemia²⁴. Elevation of bilirubin in patients with SARS-CoV-2 infection is not common. However, among COVID-19 patients who do show abnormal liver function test results during hospitalization, TBIL levels have been found to be elevated as high as 18%⁸,²⁶,²⁷. A large study conducted in 5771 patients from China suggested that TBIL values are independent of the severity of COVID-19²⁸. Indeed, the average values of TBIL, DBIL, and IBIL fell within normal ranges of values in all patients enrolled in this study (10.4 μmol/L, IQR 7.9-14.1, 3.0 μmol/L, IQR 2.1-4.4, and 7.4 μmol/L, IQR 5.2-10.3, respectively)²⁸.

Conversely, emerging evidence now supports the use of bilirubin values at hospital admission to estimate a patient’s risk of developing severe COVID-19. Seriously ill patients have higher TBIL and DBIL levels at admission than non-severe patients (for TBIL, 10.6 μmol/L, IQR 7.9-15.0 versus 10.3 μmol/L, IQR 7.9-14.0, p = 0.053, respectively; for DBIL, 3.3 μmol/L, IQR 2.2-5.2 0 versus 2.9 μmol/L, IQR 2.0-4.20, p < 0.001, respectively)²⁸. In addition, the peak of TBIL correlates with mortality risk, where TBIL maximum values of 21-63 μmol/L confer a 3-fold increased mortality risk (HR 3.28 95% CI 2.47-4.35; p < 0.001), while patients with TBIL greater than 63 μmol/L show 8-fold increased mortality risk (HR 7.98 95% CI 3.88-16.41; p < 0.001)²⁸. Higher TBIL levels also confer ~3-fold increased risk of developing severe COVID-19 (OR 2.94, 95% CI 2.18-3.97)²⁹. Likewise, TBIL values are elevated in ICU patients relative to non-ICU patients (15.8 ± 6.8 μmol/L versus 9.1 ± 3.6 μmol/L, p = 0.0082, respectively)³⁰. In a population of pediatric COVID-19 patients, Wang and collaborators also reported higher TBIL values in seriously ill patients than in non-severe individuals (10.13 μmol/L versus 6.40 μmol/L, p = 0.05, respectively) with OR of 1.316, 95% CI 0.575-3.014, p = 0.516)⁴. These studies suggest that bilirubin can be used to estimate the risk of mortality or ICU admission in adult and pediatric patients with COVID-19, especially TBIL and DBIL³¹.

Lactate dehydrogenase

Lactate dehydrogenase (LDH) is a ubiquitous enzyme in human cells. LDH consists of four polypeptide chains that constitute five different isoforms located in the cytosol and mitochondria³². In the cytosol, LDH catalyzes the conversion of lactate to pyruvate by transferring a hydride group from NAD+ to NADH³³. In the mitochondria, LDH catalyzes the conversion of lactate to pyruvate by transferring a hydride group from ferricytochrome c to ferrocytochrome c. Release of LDH into cells occurs primarily as a result of necrosis³²,³³. Serum LDH is elevated in a variety of clinical scenarios including hemolysis, infection, sepsis, infarction, liver and kidney diseases, pancreatitis, bone fracture, rhabdomyolysis or myositis, hypoxia, shock, and cancer³⁴.

LDH elevation is one of the most commonly reported laboratory anomalies in COVID-19 patients around the world. In fact, 50-80 percent of patients with SARS-CoV-2 infection show abnormally high LDH levels¹³. Numerous studies report a strong positive correlation between LDH levels and COVID-19 severity, with higher levels present in severe or critical patients. A study conducted in 548 patients reported 250 U/L and > 445 U/L as two cutoff values for LDH that can help distinguish non-severe patients from critically ill patients who are at a much higher risk of death. Patients with LDH > 445 U/L at admission were at a 4.4-fold increased risk of developing severe COVID-19 (OR, 4.4, 95% CI 2.6-7.6) and a 2-fold increased risk of death (HR 2.0, 95% CI, 1.2-3.3; p = 0.007)³⁵. In a meta-analysis of 19 papers that gathered clinical information from more than 3000 patients, LDH elevation at admission carried an 8-fold increased risk of developing severe COVID-19 (OR 8.28, 95% CI 4.75 - 14.46). LDH values were higher in patients who required ICU admission, with an OR of 5.78 (95% CI 1.65 – 20.28; p < 0.001)³⁶. Numerous studies and meta-analyses report significantly increased LDH levels in non-survivors with respect to survivors (p < 0.001)³⁶. This increase in LDH also associates with mortality risk, with OR values ranging from 4.09 to 10.88 depending on the study (p < 0.001)³⁷. In line with this, a study conducted in 375 COVID-19 patients from China reported that LDH > 365 U/L predicts mortality with an AUC of 0.943. Similarly, Feng and coworkers found that LDH > 303 U/L predicts death with an AUC of 0.829. Finally, another study conducted in 119 severe-to-critical COVID-19 patients estimated that LDH > 353.5 U/L predicts mortality with an AUC of 0.949, sensitivity of 94.4%, and specificity of 89.2%³⁸. As we have outlined here, LDH appears to be a reliable independent risk factor for mortality in patients who meet hospitalization criteria concurrent with severe-to-critical COVID-19.

Ratios and clinical scores based on measures of liver function

Liang and coworkers proposed the use of the COVID-GRAM score to predict the development of critical COVID-19. The assessment takes into account such factors as lactate dehydrogenase, direct bilirubin, chest radiography, age, hemoptysis, dyspnea, unconsciousness, cancer history, neutrophil-to-lymphocyte ratio, and number of comorbidities ^39]. This score has an AUC of 0.88 (95% CI 0.85-0.91) for predicting the proportion of patients at higher risk of needing critical care for COVID-19³⁹.

Notably, Yan and collaborators proposed a score based on LDH > 365 U/L, C reactive protein < 41.2 mg/L, and lymphocyte count > 14.7% that was shown to reliably predict mortality in COVID-19 patients, with an AUC of 0.978. In fact, use of this score permitted 100% accurate prediction of a patient’s outcome ten days prior its occurrence³⁸.

Another score based on LDH, procalcitonin, smoking history, oxygen saturation, and lymphocyte count predicts ICU admission of COVID-19 patients with an AUC of 0.761 (95% CI 0.71–0.81; p < 0.001), sensitivity of 10.5%, and specificity of 99.2%. Likewise, the combined use of LDH, procalcitonin, history of chronic obstructive pulmonary disease (COPD), oxygen saturation, heart rate, and age predicts mortality in COVID-19 patients with an AUC of 0.87 (95% CI 0.83–0.92; p < 0.001), sensitivity of 7.1%, and specificity of 100%³⁶.

Finally, the Sequential Organ Failure Assessment (SOFA) is commonly used to predict mortality in seriously ill patients with sepsis⁴⁰. The SOFA score includes oxygenation index, mean arterial pressure, the Glasgow coma scale, creatinine or urine volume, platelets, and TBIL⁴⁰. Liu and collaborators found that SOFA score at admission predicts mortality of COVID-19 patients with a cutoff point ≥ 3 points, AUC of 0.890 (95% CI 0.826–0.955), sensitivity of 90.00%, and specificity of 83.81%⁴⁰. This information supports the idea that scores that take parameters of liver function into account are better predictors of pneumonia, ICU admission, and death in patients with COVID-19.

Conclusion

As we have outlined here, liver impairment is commonly reported in COVID-19 patients, especially those with severe or critical disease who frequently experience the worst outcomes²⁶. The cause of abnormal liver function in these cases remains unclear, although several hypotheses have been proposed. Firstly, hepatocytes, cholangiocytes, and endothelial cells express the angiotensin-converting enzyme 2 (ACE2) receptor that acts as the main SARS-CoV-2 entry point⁴¹ into cells. Upon SARS-CoV-2 infection, liver cells may undergo necrosis and/or apoptosis. This can, in turn, contribute to inflammation and lead to hepatic damage. It is also believed that ARDS, hypoxia, and coagulation disorders might contribute to liver damage by increasing hepatic isquemia⁴². Further, exacerbation of the inflammatory response directly contributes to the cytokine storm often observed in severe COVID-19 cases, which in turn promotes multiple organ failure, including failure of the liver⁴³. Finally, the use of hepatotoxic drugs such as acetaminophen might also alter liver function tests in patients with COVID-19⁴²,⁴⁴.

In terms of liver function tests, serum albumin and LDH are better predictors of ICU admission and death than levels of transaminases, ALP, bilirubin, and GGT⁴⁵. Clinical scores that include indicators of liver function such as albumin, LDH, and total bilirubin are also accurate predictors of the most adverse outcomes, including pneumonia, ICU admission, and death, in patients with SARS-CoV-2 infections. Therefore, prognosis in COVID-19 should be assessed using scores that take liver function tests such as albumin and LDH into account, together with other demographic or clinical parameters. Liver function tests greatly enhance the ability of clinicians to identify patients at higher risk of developing the most severe cases of COVID-19, which could help increase survival rates and optimize hospital resources throughout the current pandemic.