Liver dysfunction in adults with COVID-19 infection: A longitudinal study with transient elastography evaluation.

BACKGROUND AND AIM

Abnormal liver biochemistry (ALB) is common among patients with COVID-19 infection due to various factors. It is uncertain if it persists after the acute infection. We aimed to investigate this.

METHODS

A multicenter study of adult patients hospitalized for COVID-19 infection, with at least a single abnormal liver function test, was conducted. Detailed laboratory and imaging tests, including transabdominal ultrasound and FibroScan, were performed at assessment and at 6-month follow-up after hospital discharge.

RESULTS

From an initial cohort of 1246 patients who were hospitalized, 731 (58.7%) had ALB. A total of 174/731 patients fulfilled the inclusion criteria with the following characteristics: 48.9% patients had severe COVID-19; 62.1% had chronic liver disease (CLD); and 56.9% had metabolic-associated fatty liver disease (MAFLD). ALB was predominantly of a mixed pattern (67.8%). Among those (55.2%) who had liver injury (aspartate aminotransferase/alanine aminotransferase >3 times the upper limit of normal, or alkaline phosphatase/γ-glutamyl transferase/bilirubin >2 times the upper limit of normal), a mixed pattern was similarly predominant. Approximately 52.3% had normalization of the liver lunction test in the 6-month period post discharge. Patients with persistent ALB had significantly higher mean body mass index (BMI) and serum low-density lipoprotein (LDL), higher rates of MAFLD and CLD, higher mean liver stiffness measurement and continuous attenuated parameter score on FibroScan, and higher rates of liver injury on univariate analysis. Multivariate analysis was not statistically significant.

CONCLUSIONS

Approximately 47.7% of COVID-19 patients were found to have persistent ALB up to 6 months following the acute infection, and it was associated with raised BMI, elevated serum LDL, increased rates of MAFLD and CLD, and higher rates of liver injury on univariate analysis, but not on multivariate analysis.