Characteristics of the liver parenchyma according to the native CT examinations data at the stages of COVID-19 treatment

Purpose. To assess the liver density according to the data of native CT studies in patients with COVID-19, depending on the severity of the pulmonary parenchyma damage and the prescribed treatment, to compare the data with biochemical indicators, and also to demonstrate changes in density indicators over time. Material and methods. Lung CT data from 200 patients with COVID-19 were retrospectively analyzed. The density of the liver, spleen, and subcutaneous fat tissue was measured in all patients on the images of the upper abdominal cavity that entered the scan area. The ratio of the density of the liver to the spleen and to the density of the fat tissue was assesed. These indicators were compared with each other in two groups of lung tissue damage: CT 1–2 and CT 3–4. The CT 3–4 group was assessed in detail: the density indicators of the liver were studied in dynamics, and their relationship with biochemical indicators – during the initial study. A comparison was also made between two subgroups: patients taking tocilizumab and those without tocilizumab. Results. A decrease in liver density and the ratio of liver density to spleen density was observed in 35.5% and 47.5% of patients respectively. Liver density and the ratio of liver density to spleen density were lower in

Introduction respiratory distress syndrome, and multiple organ failure develop [3].
Literature data show, that, except for the pulmonary manifestations, COVID-19 may lead to liver damage, caused by a combination of various mechanisms, including the administration of potentially hepatotoxic drugs, systemic inflammatory response, hypoxia, caused by respiratory distress syndrome, and multiple organ failure [4].
Covid-19 diagnostics is performed with a combined assessment of the epidemiological anamnesis, clinical presentation, and results of radiology and laboratory examinations [5].
Computed tomography (CT) of the chest organs showed up to 98% of sensitivity in detecting changes in the lungs of the patients with COVID-19 pneumonia and a negative PCR result, in addition, CT allows us to evaluate these changes in dynamics [6,7].
From April 17 to June 10, 2020 the Vishnevsky National Medical Research Center of Surgery was redesignated as a Center for providing medical care to patients with the new coronavirus infection. At admission, all patients underwent chest CT. It is natural, that the upper sections of the liver and spleen were also examined. Based on the percentage of lung parenchyma lesion, patients were assigned with a degree of severity of disease from CT1 to CT4 according to the recommendations of the "Scientific and practical clinical center for diagnostics and telemedical healthcare technologies of the Moscow Healthcare Department" [8]. Except for the primary CT-scans, follow-up examinations were carried out every 3-4 days or when clinical presentation changed. We have previously analyzed CT-signs of lung parenchyma damage with COVID-19 [9]. This research is devoted to extra-pulmonary manifestations of the disease, particularly, changes in the liver parenchyma.
Analyzing CT scans of the patients admitted to our COVID-19 Center, we show co-morbid conditions. Among them, signs of hepatic steatosis in the form of a diffuse decrease in the liver parenchyma CT density less than 45 HU were quite frequent. Assessment of liver CT density in dynamics, as a rule, showed an increase in its density during time of hospitalization. In addition, biochemical analysis of blood serum of many patients showed an increase in levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), C-reactive protein (CRP), lactate dehydrogenase (LDH), D-dimer and ferritin above normal, and a decrease in albumin levels.
During hospitalization, most patients received standard three-component therapy in accordance with the "Temporary guidelines. Prevention, diagnosis and treatment of new coronavirus infection " of the Ministry of Health of the Russian Federation, which included hydroxychloroquine, lopinavir/ritonavir and azithromycin [10]. During treatment, other antibiotics of various groups were also prescribed (fluoroquinolones, carbapen-ems, aminoglycosides, penicillins and cephalosporins in combination with a beta-lactamase inhibitor, etc.) after cessation of the three-component therapy. In addition, in case of appropriate indications in clinical and laboratory studies, some patients were treated by tocilizumab -a recombinant humanized monoclonal antibody to human interleukin-6 receptor (IL-6) from the IgG 1 immunoglobulin subclass [11].
The research goal is evaluation of the liver CT density, depending on the severity of pulmonary parenchyma damage, biochemical parameters and prescribed treatment, and demonstration of liver CT density changes in dynamics, based on native CT studies in COVID-19 patients.

Materials and methods
We have analyzed chest CT scans of 200 patients, including 87 women aged from 23 to 93 years (the average age is 62 years) and 113 men aged from 21 to 89 years (the average age is 57 years). The study was performed using a standard Protocol for chest CT and high-resolution reconstruction on a Philips Ingenuity CT 64 multi-detector scanner with the patient lying in prone position with his hands thrown back behind his head. The following scanning parameters were used for the standard protocol: 64 × 0.625 collimation, 1 mm reconstruction, 0.5 mm increment. A scan area including the chest was planned by the plan scan. Assessment of the scans was carried out in the DICOM-images viewing module of medical hardware-software complex "ArchiMed" (Med-Ray, Russia, 2004).
Patients were divided into two groups, depending on the maximum severity of the pulmonary parenchyma lesion on CT scans, observed during hospitalization. The CT 1-2 group included 53 patients (27 with CT1 and 26 with CT2), and the CT 3-4 group included 147 patients (81 with CT3 and 66 with CT4). A criterion for inclusion in the CT 3-4 group was also presence of at least three CT scans series in dynamics. All studies included the upper sections of liver and spleen in the scanning area. According to CT images, the following measurements were made in all patients: the CT density of the right and left lobes of the liver and the CT density of the spleen. Based on the results of two measurements, the average liver CT density was calculated and the ratio of liver CT density to spleen CT density was calculated, based on the received data. Taking into account the fact, that some patients had edema of subcutaneous fat, which means that tissue and interstitial fluid could influence the CT density of the liver and spleen, the density of subcutaneous fat was also measured on the anterior abdominal wall and on the lateral surface near the mid-axillary line. Based on the results of the two measurements, the average CT density of the subcutaneous fat was calculated. The ratio of the liver CT density to the subcutaneous fat CT density was calculated, using the received data АННАЛЫ ХИРУРГИЧЕСКОЙ ГЕПАТОЛОГИИ, 2020, том 25, №3 ANNALS OF HPB SURGERY, 2020, Vol. 25, N 3 (figure). These indicators were compared in two groups to determine, whether the decrease in liver CT density depends on the severity of the pulmonary parenchyma lesion according to the CT examinations. At the second stage of this study, the CT 3-4 group was analyzed in more details. For all patients, data from biochemical tests, performed on the day of the initial CT study, were collected. Values of ALT, AST, CRP, LDH, D-dimer and ferritin were obtained to assess, whether there is a link between liver CT density and laboratory data.
At the third stage of the study, we evaluated, how the above-described CT density changes in dynamics during subsequent CT examinations in patients from the CT 3-4 group.
At the fourth stage, patients with the severity of CT 3 or CT 4 were divided into two groups. The first group included 91 patients, who were treated by three-component therapy and antibiotics. The second group included 30 patients, who additionally were treated by tocilizumab. Decision for its administration were based on a combination of CT data: a significant volume of compacted lung tissue and (or) prevalence of 50-75% of lung damage (CT3 or more) with 2 or more clinical symptoms: a decrease in SpO 2 , an increase in CRP> 60 mg/l or an increase in the level of CRP in 3 times on 8-14 days of the disease, fever >38 °C for 5 days, leukocytopenia <3,0-3,5 × 10 9 /l, lymphocytopenia <1 × 10 9 /l and/or <15% [12]. The above-mentioned CT density, measured both at the initial CT study and in dynamics, were compared in the two groups.
Statistical analysis was performed using the STATISTICA package for statistical data analysis (Statistica for Windows, Copyringht © by StatSoft, 1994 licensed), using descriptive statistics and Student's t-test methods to compare the average values in two subgroups. The differences were considered statistically significant at p < 0.05. To analyze the dependence of liver density indicators on the values of biochemical tests, the Pearson correlation coefficient was used.

Results
Among 200 patients at the primary CT study, 71 patients (35.5 %) showed a decrease in liver CT density of less than 45 HU. Among them, the maximum severity of CT 1 was observed in 5 patients (7%), CT 2 -in 9 patients (12.7%), CT 3 -in 30 patients (42.3%) and CT4 -in 27 patients (38%). A decrease in the liver to spleen density ratio less than 1 was observed in 95 patients (47.5%), including: 5 patients with CT 1 (5.2%), 12 patients with CT 2 (12.6%), 40 patients with CT 3 (42.1%) and 38 patients with CT 4 (40%). When comparing the average values of CT density, the liver CT density in the CT 3-4 group was significantly lower than in the CT 1-2 group, 43.9 HU vs. 49.3 HU, respectively (p < 0.008). The liver to spleen density ratio was also lower in the CT 3-4 group than in the CT 1-2 group, and was 0.9 vs. 1.0, respectively (p < 0.014). At the same time, the average subcutaneous fat density in the two groups was almost the same (−111.8 HU in the CT 1-2 group and −111.6 HU in the CT 3-4 group), and the ratio of liver CT density to subcutaneous fat CT density was slightly lower in the CT 3-4 group (table 1).  The results of biochemical tests of the vast majority of the patients showed an increase in the level of CRP, LDH and ferritin. An increase of ALT and AST levels in the serum was detected in 23% and 34% of patients, respectively. A decrease in albumin levels was observed in 30% of patients. It is important to note, that not all 147 patients of the CT 3-4 group had the results of the entire list of biochemical parameters at the time of the initial CT examination. The number of CT scan series, as well as the average values, medians, and percentage of patients with deviations in biochemical parameters are presented in table 2.
Data analysis to identify possible correlation between liver CT density and biochemical tests at initial examination, showed moderate negative correlation between the liver CT density and albumin level in serum (r = −0,30; p < 0.05). There was also noticed a weak negative correlation between the liver to spleen density ratio and albumin level in serum (r = −0,26; p < 0.05) was also noticed. There was no correlation with other biochemical parameters (table 3).
The dynamics of CT density in patients of the CT 3-4 group show an increase the liver parenchyma CT density and the liver to spleen density ratio in each subsequent study. The difference between the average CT densities of the liver in the first and in the fourth CT scans series was 11.85 HU. At the same time, the maximum "increase" in liver CT density was observed at the third CT examination (6.14 HU). At the fourth CT examination, the liver CT density increased by 3.98 HU. The average value of the liver to spleen density ratio increased from 0.93 to 1.15 between the first and fourth examinations. It is noteworthy, that four consecutive CT examinations were performed only in 63 patients out of 147, assigned to the CT 3-4 group. The change in the dynamics of spleen and subcutaneous fat densities did not ex- Table 3. Correlation of liver density values and data of biochemical analyzes in the primary study (p < 0.05)

Parameters
Liver CT density (r) Liver density/spleen density ratio (r)  2 HU, but there was a gradual increase in the absolute values of the ratio of liver density to subcutaneous fat density (table 4).
When assessing the dynamics of CT density in groups with and without treatment by tocilizumab, the first three CT examinations in the group with tocilizumab the liver CT density was slightly higher. The liver to spleen density ratio was also higher in the tocilizumab group in the first two CT examinations. However, no significant differences were found in any of the CT examinations, performed during hospitalization. At the same time, four consecutive CT examinations were performed only in 38 patients in group without tocilizumab, and in 19 patients from the tocilizumab group (table 5).

Discussion
According to our results, 35.5% of patients with COVID-19 showed a decrease of liver CT density at the primary CT examination, and 47.5% had liver to spleen density ratio less than one. These indicators were significantly lower in the CT 3-4 group compared to the CT 1-2 group. In other words, COVID-19 encourages hepatitis, characterized by a decrease of liver CT density.
Our results confirm Ping Lei and co-authors' data. They analyzed 115 patients with COVID-19, among which a decrease of liver CT density was observed in 26.09%, and it prevailed in patients with severe disease course (58.82%). The authors have shown, that severity of the disease correlates not only with the values of AST, CRP and the percentage of lung parenchyma damage, but also with a decrease in the liver to spleen density ratio to less than one [13].
Over the course of the disease, control CT scans showed an increase in the average values of liver CT density from 43.9 HU to 55.75 HU, and the liver to spleen density ratio from 0.93 to 1.15, This does not allow us to consider the changes in liver parenchyma solely as a premorbid pathology, and they may imply a transient nature of the changes.
Various possible mechanisms of liver damage in COVID-19, including administration of potentially hepatotoxic medications, systemic inflammatory response , hypoxia, caused by respiratory distress syndrome , and multiple organ failure were described [4,14].
The assumption that increasing in liver CT density may be caused by exposure of tissue and interstitial fluid, similar to subcutaneous fat edema, was rejected. Density values of subcutaneous fat and increase the absolute values of the ratio of the liver density to the subcutaneous fat density were almost constant when evaluating in dynamics. Nanshan Chen and co-authors analyzed clinical and laboratory data from 99 patients with COVID-19. They observed an increase in AST in 35% of patients, ALT in 28%, LDH in 76%, CRP in 86%, D-dimer in 36%, ferritin in 63%, and a decrease in albumin was observed in 98% of patients [15]. According to Zhang C and co-authors' data, 14-53% of patients with COVID-19 were registered with increased AST and ALT, and in 2-11% of the cases, the infection developed in the setting of chronic liver diseases [16]. Results of the study of a group of 1099 patients from 552 hospitals showed that the increase in ALT and AST was more pronounced in patients with severe course of disease [17].
Similar to the literature data, in our study, many patients with biochemical analysis of blood serum showed a deviation of the above-mentioned parameters. At the same time, most of the biochemical parameters were not associated with changes in CT density during the primary CT examination. Nevertheless, a moderate and weak negative correlation between serum albumin level, and CT-liver density, and the liver to spleen density ratio was obtained.
Ji D. and co-authors analyzed data from biochemical tests of 202 patients in order to study manifestations of non-alcoholic hepatic steatosis in COVID-19. According to the results, liver damage was observed in 101 (50%) and 152 (75.2%) patients at admission and during hospitalization, respectively. In 67 (33.2%) cases, persistent liver disorders were noted from the time of admission until the last study. In 39 (19.3%) cases, progressive changes were revealed, and in 163 (80.7%) patients, stabilization was АННАЛЫ ХИРУРГИЧЕСКОЙ ГЕПАТОЛОГИИ, 2020, том 25, №3 ANNALS OF HPB SURGERY, 2020, Vol. 25, N 3 Most of the patients in our study administered standard three-component therapy, which included hydroxychloroquine, lopinavir/ritonavir, and azithromycin. After cessation of three-component therapy, antibacterial therapy was prescribed, usually a combination of drugs of different groups. For this reason, it is impossible to assess, which of the drugs made a greater contribution to the reduction of CT liver density. In addition, many patients with COVID-19 received antipyretics during treatment, paracetamol in particular, which could cause liver damage with a rise in ALT and AST at a dose of more than 7.5-10 g [23]. According to the literature, taking a number of antiviral medications, such as oseltamivir, arbidol, or lopinavir, before or during hospitalization, may also cause some hepatotoxic effects. Thus, among patients with liver damage in COVID-19, the proportion of those taking lopinavir/ ritonavir was higher (57.8%) than among patients with normal liver function (31.3%) [24].
According to the results of our study, no significant differences between CT density, measured in dynamics in groups with and without tocilizumab, were found in any of the examinations, performed during the hospitalization.
Recent studies have shown, that 15.7% of patients with COVID-19 develop severe pneumonia and cytokine storm, which is accompanied by increased levels of cytokines, interleukin 6 (IL-6) especially. That is an important factor, leading to rapid disease progression [17]. This process can cause multiple injuries not only in the lungs, but also in the liver, heart, and kidneys. However, considering the data we obtained, when comparing groups with and without tocilizumab, this mechanism is, probably, not dominant.
In addition to the above-mentioned mechanisms of liver damage in COVID-19, respiratory distress syndrome and multiple organ failure can cause hypoxia and shock. Thus, they can cause ischemia and liver dysfunction. Reduced oxygen content and accumulation of lipids in hepatocytes during shock and hypoxic conditions can lead to cell death. A subsequent increase in the amount of reactive oxygen species and their peroxidation products can act as a secondary intermediary, further enhancing the release of many pro-inflammatory factors and liver damage [25].

Conclusion
According to our results, the values of liver CT density were lower in patients with COVID-19 with a higher degree of CT pulmonary parenchyma damage. Over the disease course, the liver density increased, which indicates the reversible nature of such changes. Assessment and monitoring of the dynamics of liver CT density on CT scans of the upper abdominal cavity included in the lung scanning area does not require additional CT examination and can become a useful parameter in determining the severity of the disease.
Literature suggests considering liver damage in COVID-19 as a result of secondary liver damage caused by a number of factors, including the use of hepatotoxic drugs and systemic inflammatory response. However, a comparison of groups with and without tocilizumab administration demonstrated that treatment with tocilizumab does not affect liver density.
In addition, many patients with COVID-19 had deviations in a number of biochemical parameters, which indicated some liver damage. In the study, no strong correlations was found between the CT density in primary CT scan and any biochemical parameter, a more detailed analysis in dynamics. However, a more detailed dynamic analysis is needed, which may suggest the prevailing mechanism of liver damage in COVID-19.