Chronic Liver Disease and COVID-19: A Perspective From Singapore

Background: Chronic liver diseases including non-alcoholic fatty liver disease (NAFLD) are associated with COVID-19 progression. With no signs of the pandemic abating, thousands worldwide are at risk of a severe disease course. Objective: We aimed to describe the epidemiological features, clinical course, treatment and outcomes of our patients with COVID-19 and chronic liver disease. Methods: A retrospective descriptive study of COVID-19 patients with chronic liver disease admitted to the National Centre for Infectious Diseases in Singapore between 29 th February and 2 nd May 2020 was performed. Results: 16 patients had chronic liver disease – 9 had NAFLD. In the NAFLD group, peak ALT was higher (median: 84 U/L vs 38 U/L; P = 0.042) and more patients had hyperlipidaemia (88.9% vs 28.6%; p=0.035), but median Body Mass Index (BMI) was not signicantly higher (24.3 kg/m 2 vs 24.2 kg/m 2 ). NAFLD patients had a poorer clinical course: more required anti-viral medications (66.7% vs 0%; p=0.011), and time to negative swab was longer (24 days vs 13 days; p=0.008). 3 patients had liver cirrhosis (all non-NAFLD). 1 decompensated but none required intensive care unit admission or died. Conclusion: Our results show that in COVID-19 patients with chronic liver disease, those with NAFLD experience a more severe clinical course. In Asians, NAFLD can have poorer prognostic implications, despite them having only mildly raised BMI. We advocate that patients with NAFLD and liver cirrhosis should be closely monitored for COVID-19 disease progression.


Introduction
Millions of people around the world have been infected by the Coronavirus disease (COVID-19) 1 . Many countries are facing new waves of infection with no signs of the pandemic abating. While initial studies suggested that the impact of COVID-19 on the liver was mild, there are now ndings that COVID-19 may be more severe in patients with non-alcoholic fatty liver disease (NAFLD) and other underlying chronic liver diseases (CLD).
Between 1% and 11% of patients with COVID-19 may have underlying CLD 2 . While the SECURE-Cirrhosis Registry and the COVID-HEP registry are collecting data on pre-existing liver disease and COVID-19, our current knowledge of these patients with CLD continues to evolve. NAFLD has an increasing global prevalence of about 25% 3 and was found to be associated with COVID-19 disease progression [4][5][6] . The subset of cirrhotic patients was reported to have potentially increased risk for COVID-19, higher risk for severe disease, and increased risk for hepatic decompensation and death 7 . However, no studies in South-East Asia have directly compared NAFLD patients against CLD patients of other aetiologies (non-NAFLD).
We aimed to describe the epidemiological features, clinical course, treatment and outcomes of our patients with COVID-19 and CLD in Singapore.

Materials And Methods
All patients con rmed to have COVID-19 by a positive SARS-CoV-2 real-time reverse transcriptasepolymerase chain reaction (RT-PCR), who were admitted to the National Centre for Infectious Diseases in Singapore between 29th February and 2nd May 2020, were considered for data source. Only patients with an existing diagnosis of CLD or the presence of steatosis on ultrasound liver imaging were included in the nal analysis [8][9][10] . We identi ed a total of 16 patients with CLD and aimed to retrospectively analyse them according to two groups -NAFLD versus non-NAFLD. Waiver of informed consent for collection of clinical data from infected individuals was granted by the Singapore Ministry of Health, under the Infectious Diseases Act as part of the COVID-19 outbreak investigations.
Patient demographics, co-morbid conditions, presenting symptoms, serial laboratory data and hospitalization outcomes (including need for oxygen, intensive care unit [ICU] admission, time to negative COVID-19 swab and in-hospital mortality) were obtained using a standardized data collection form. The negative swab was either taken as the rst out of two consecutive negative swabs (available investigations in hospital) or from the discharge date from community isolation facilities (where actual dates of negative swab results are unavailable). The R ratio was used to describe the pattern of liver injury and was calculated as (Alanine aminotransferase [ALT] ÷ ALT upper limit normal) ÷ (Alkaline phosphatase [ALP] ÷ ALP upper limit normal) 11 . The peak (highest value obtained during the admission) ALT and ALP values were used, and the upper limit of normal for ALT and ALP was 40 U/L and 120 U/L respectively. Child-Turcotte-Pugh (CTP) status and Model for End-Stage Liver Disease (MELD) were calculated for the known cirrhotic patients at and during admission.
Numbers and proportions were presented for categorical variables, with median values and range for continuous variables. Fisher's exact or Chi-square test was used to compare categorical variables, and Mann-Whitney U test to compare continuous variables between NAFLD and non-NAFLD patients. All statistical tests were two-sided, and statistical signi cance was taken as p < 0.05. Statistical analyses were performed using IBM SPSS Statistics for Windows, V.24.0.

Results
Out of a total of 16 patients with CLD -9 had NAFLD, 6 had Hepatitis B and 1 had alcoholic liver disease. Table 1 shows the patient demographics and baseline characteristics. The majority were Chinese ethnicity (75%), one was Indian, two were Filipino and one was Bangladeshi. The median age was 61 years (range 28-80 years) and nine (56.3%) were female. More patients in the NAFLD group as compared to the non- Serial laboratory data from each patient obtained during their hospitalization stay was analysed. 62.5% of patients had abnormal liver function tests (LFT) during the admission. The median peak ALT was 61 U/L (range 19 U/L − 518 U/L) and the median peak ALP was 86 U/L (range 45 U/L − 332 U/L). The median day of illness corresponding to the peak ALT was day 5 (range day 1-23), and this was similar for both the NAFLD and non-NAFLD groups. The difference in median peak ALT values between the two groups was statistically signi cant (84 U/L vs 38 U/L; p = 0.042). Derangement in LFT was transient with an improving trend as the illness resolved. The temporal patterns of selected laboratory investigations over a 30-day period between the two groups were mostly similar (Fig. 1). The median peak R ratio for all patients in our study was 2.1, corresponding to a mixed picture of liver injury (de ned as R ratio 2 to 5) 11 . Our analysis showed that patients in the non-NAFLD group had a cholestatic pattern of liver injury (de ned as R ratio < 2) 11 with a median peak R ratio of 1.3 versus a mixed picture (2.4) in the NAFLD group (p = 0.055).
Overall, our study found that patients in the NAFLD group developed a more severe clinical course, as compared to those in the non-NAFLD group. 6 patients in the NAFLD group (66.7%) required one or more anti-viral medications versus none in the non-NAFLD group (p = 0.011). Patients were started on anti-viral medications based on the attending physician's clinical discretion. Kaletra was the most common antiviral used (4 patients) followed by Remdesivir (2 patients). Other drugs administered included Hydroxychloroquine and Interferon beta 1b. 43.8% of all patients required supplemental oxygen therapy with similar numbers in both groups. Only two patients required intubation and ICU admission; both of them belonged to the NAFLD group. There were no deaths. In terms of biochemical markers of disease severity, patients in the NAFLD group had higher levels of Lactate Dehydrogenase (LDH) and C-Reactive Protein (CRP) levels as compared to patients in the non-NAFLD group, though this was not statistically signi cant. Those in the NAFLD group also took a longer time to achieve a negative swab (median 24 days vs 13 days; p = 0.008) and had a longer hospitalization stay (median 21 days vs 12 days; p = 0.210) as compared to patients in the non-NAFLD group.
Our cohort included three patients with liver cirrhosis, all of whom were in the non-NAFLD group. The rst patient had hepatitis B and primary biliary cholangitis with CTP-A(5) class, MELD 6 score at baseline. This patient remained well and had normal LFT. The second patient had hepatitis B with underlying chronic kidney disease, and was CTP-A(6) class MELD 12 score at baseline. Despite having a slight increase in the CTP class to B(7) on day 7 of illness due to a drop in Albumin levels, this patient did not decompensate and remained well. The third patient was CTP-C(11) class MELD 17 score at baseline and decompensated on day 7 of illness with an increase in international normalised ratio (INR) that peaked at 2.5, ascites, variceal bleed requiring endoscopic ligation and a newly diagnosed hepatocellular carcinoma. However, this patient did not require anti-viral medications or ICU admission and was discharged on day 14 of illness.

Discussion
Our study showed that patients in the NAFLD group had a worse clinical course as compared to those in the non-NAFLD group. A higher proportion of NAFLD patients required the administration of anti-viral medications, had higher levels of LDH and CRP, required ICU admission, and had prolonged hospitalization. This supports the current evidence that patients with NAFLD do worse with COVID-19 as compared to non-NAFLD patients 4,12 . SARS-CoV-2 enters host cells via angiotensin I converting enzyme 2 (ACE2) receptors present in abundance in adipose tissues which can serve as a viral reservoir 13 . In addition, NAFLD patients may be more vulnerable to increased cytokine production associated with COVID-19 14 . We note that despite a poorer clinical course, patients in the NAFLD group had mild transaminitis with no persistent deterioration in liver function. Studies looking at LFT in COVID-19 patients have also found that the majority had mild LFT abnormalities that were self-limiting [15][16][17] . Thus, other factors such as associated metabolic co-morbidities may contribute to the increased risk of severe COVID-19 infection in NAFLD patients 18 . We found that more patients in the NAFLD group had hyperlipidaemia as compared to those in the non-NAFLD group.
Ji et al 4 . reported that COVID-19 patients who had illness progression had signi cantly higher BMI levels (26.6 ± 2.2 kg/m 2 ), with 87% of these patients having NAFLD. However, we found that the median BMI in our NAFLD group who had a more severe clinical course, was lower than expected (24.3 kg/m 2 ). Thus, it is unclear if weight or the underlying aetiology of fatty liver disease plays a bigger role in determining a patient's clinical course in COVID-19. NAFLD is traditionally associated with central obesity but can also occur in lean subjects, especially in the Asian population. Body fat distribution is more important than total body fat content in the development of non-obese NAFLD [19][20][21][22][23] . Though Asians with NAFLD may have a lower BMI, they are still at an increased risk for adverse cardiometabolic outcomes. This has been attributed primarily to alterations in body fat distribution, in particular visceral adiposity which can be measured by waist circumference or waist-to-hip ratio 24 . Recently, sagittal abdominal diameter (SAD)measured by the anteroposterior diameter of the abdomen in the supine position, has emerged as the preferred anthropometric indicator for visceral adiposity. SAD is clinically feasible and is an independent measurement with high reproducibility and correlation to adverse metabolic outcomes [25][26] . Genetic susceptibility of Asians to non-obese NAFLD has been postulated to be via predisposition to Type 2 diabetes mellitus, patatin-like phospholipase domain-containing-3 (PNPLA3) gene / Sterol regulatory element-binding factor 2 (SREBF) single nucleotide polymorphisms (SNP) and polymorphisms in apolipoprotein 25,27−29 . More studies will be needed to identify the factors that causally drive COVID-19 progression in individuals with obese and non-obese NAFLD.
One out of three of our cirrhotic patients had worsening LFT and decompensation, though none required ICU admission or died. Multiple mechanisms involved in liver injury in COVID-19 have been described.
These include immune-mediated damage, direct cytotoxicity, anoxia due to respiratory failure, druginduced liver injury, and reactivation of pre-existing liver disease 30 . In addition, the presence of cirrhosis associated immune dysfunction syndrome (CAIDS) can predispose cirrhotic patients to severe infection 31 and result in more severe liver injury 32 . A recent study looking at cirrhotic patients with COVID-19 infection showed that mortality correlated strongly with baseline CTP class and MELD score, with a greater proportion of patients with CTP-C liver cirrhosis dying 7 . Other authors have found that patients with advanced liver cirrhosis did poorer with COVID-19 infection [33][34] . We postulate that our 3 cirrhotic patients ran a milder COVID-19 clinical course possibly from the absence of NAFLD and that 2 of 3 were CTP-A at baseline. This may explain why they had better outcomes in terms of not requiring ICU admission and not dying.
Our study had some limitations during the process of data acquisition due to missing clinical information, resulting in a smaller sample size of identi able CLD patients. The majority of patients who were admitted and found to be overweight or have abnormal LFT, did not undergo liver imaging which could have picked up more patients with fatty liver. We also did not have baseline laboratory values of most patients prior to their admission. These values could have been used to calculate brosis scores as elevated Fibrosis-4 (FIB-4) scores have been shown to increase the likelihood of severe COVID-19 illness 5,35 .

Conclusion
Our results demonstrate that in COVID-19 patients with underlying CLD, those with NAFLD have modest ALT elevations and a benign liver outcome. However, these patients with NAFLD have an overall poorer clinical course -requiring anti-viral medications and ICU admission, compared to those of other aetiologies. In Asians, NAFLD can have poor prognostic implications, even when BMI is only mildly raised.
Our group of NAFLD patients had a lower than expected BMI, which suggests that non-obese NAFLD patients with COVID-19 can also run a more severe clinical course regardless of weight. Clinically feasible and non-invasive measurements such as waist circumference / sagittal abdominal diameter, may have prognostic value in these patients. In addition, although one out of three of our cirrhotic patients decompensated when infected by COVID-19, none of them required ICU admission or died.
Clinicians should be cognisant of NAFLD as a poor prognostic indicator in COVID-19 and duly prioritize their management. However, the role of pro-active surveillance for NAFLD amongst those infected with COVID-19 has yet to be determined and further studies are needed to better understand the interplay between these two diseases. ALT, alanine aminotransferase; AST, aspartate aminotransferase; ALP, alkaline phosphatase; BIL, bilirubin; BMI, body mass index; CRP, c-reactive protein; GGT, gamma-glutamyl transferase; ICU, intensive care unit; LDH, lactate dehydrogenase; NAFLD, non-alcoholic fatty liver disease; PLT, platelet; WBC, white blood cell.