Reduced Serum Cholinesterase Activity Distinguishes Hepatic Encephalopathy From 48 Types of Human Diseases

Background: Hepatic encephalopathy is a complication of central nervous systems due to liver failure-related brain inammation. Less than half of patients suffering from liver failure develop hepatic encephalopathy, which suggests other factors beyond liver failure might contribute to hepatic encephalopathy. Indeed, we reported previously that the levels of serum direct bilirubin, a liver cell-made product, are counter-intuitively highest in hepatic encephalopathy patients among 72 clinically dened diseases. In current study, we tested if cholinesterase could serve as a biomarker for hepatic encephalopathy by comparing serum cholinesterase activities among 48 different types of human diseases. Methods: The activity of serum cholinesterase was determined by the standard “continuous monitoring method” in the clinical laboratory of the hospital where the serum cholinesterase activities from 137,305 independent tests with 48 clinically dened diseases and 3,387 independent tests from healthy individuals who came to the hospital for physical examination during the past 5 years were retrieved. All data were analyzed with RStudio V.1.3.1073 and python libraries 3.8. Results: We found that all 48 types of diseases had decreased cholinesterase activity compared to control based on either mean or median values. Remarkably, hepatic encephalopathy had the lowest cholinesterase activity and the serum cholinesterase activity was the best biomarker for hepatic encephalopathy (AUC 0.99, sensitivity 100%, and specicity 99%) among all diseases. Moreover, two component analysis of cholinesterase activity distributions revealed hepatic encephalopathy resembled preeclampsia and uremia whereas cirrhosis resembled multiple myeloma, leukemia, myeloproliferative disorder, and liver cancer. Conclusions: Decreased cholinesterase activity was an almost perfect serum biomarker for patients suffering hepatic encephalopathy at all stages. The resemblance of hepatic encephalopathy to preeclampsia and uremia based on cholinesterase activities provided new insight in understanding hepatic encephalopathy etiology beyond liver failure.


Background
Hepatic encephalopathy (HE) is known as a spectrum of neuropsychiatric abnormalities in patients suffering from liver dysfunction when other known brain disease has been excluded [1,2]. HE is not only associated with systemic in ammation, but also with neuroin ammation [3][4][5] or endotoxemia [6,7]. Its onset may be gradual or sudden manifested as neuropsychiatric disorders and motor dysfunctions [8].
The one-year mortality of severe HE is 54% [9]. HE is divided into covert type including minimal HE and grade I HE, and overt type ranging from grade II to grade IV HE [10]. Increased serum ammonia [11], γaminobutyric acid [12], manganese [13], or oxidative/nitrosative stress, etc. are correlated to the occurrence and development of HE [14]. Lactone, rifaximin, probiotics or molecular adsorbent recirculating system that have anti-in ammatory effects can signi cantly improve the neuropsychiatric symptoms of patients.
Even though HE is associated with liver failure, less than half of patients suffering liver failure develop HE, suggesting other factors contributed to HE development. Indeed, we found that the levels of serum direct bilirubin, a liver cell-made product, are counter-intuitively highest in HE patients among 72 clinically de ned diseases [15]. In order to monitor therapeutic effects for patients suffering from HE, blood-based biomarkers are desirable. High levels of blood ammonia are usually used to monitor the therapeutic effects of HE. However, recent studies have shown that concentrations of blood ammonia cannot always re ect the severity of HE [16][17][18]. Therefore, it is necessary to explore other serum biomarkers with better sensitivity and speci city for monitoring the therapeutic effects of HE and revealing more etiology of HE.
Serum CHE is a serine hydrolase mainly synthesized in liver, which is also expressed in pancreas, glia, and the endothelial cells of the central nervous system. CHE consists of four identical subunits with 574 amino acids each and each monomer contains ten potential N-glycosylation sequences in which nine are glycosylated [19,20]. CHE is centered on a proline-rich polypeptide and the four identical subunits interweave at the C-terminal to form the mature tetramer [21]. It comes to realize that CHE plays a vital role in the hydrolysis of different esters, including butyrylcholine, succinylcholine and ghrelin [22], and is also involved in the metabolism of certain drugs and toxic substance, such as procaine, cocaine, heroin [23,24], aspirin, and organophosphorus poisons [25,26]. In addition, CHE is associated with in ammation and parasympathetic dysfunction [27,28]. However, the serum CHE activities and their dynamic ranges have never been studied and compared systematically in different types of diseases.
In current study, the serum CHE levels from 137,305 patients with 48 clinically de ned diseases and 3,387 healthy individuals during the same period were retrieved from the clinical laboratory of the A liated Hospital of Qingdao University during the past 5 years and different statistical methods were employed to perform data analysis.

Study participants
After obtaining approval from the Hospital Ethics Review Board of Qingdao University, we were allowed to retrieve the electronic medical records and lab data of serum CHE activities of both healthy individuals and patients with clinically de ned diseases from the clinical laboratory of the A liated Hospital of Qingdao University during the past 5 years. All research was performed in accordance with relevant guidelines/regulations and informed consent was obtained from all participants and/or their legal guardians.

Clinical Data Collection
All diseases with more than 84 independent serum CHE activity tests over past 5 years were chosen. All patients with primary diagnosis of speci c disease were included in current study without exclusions. Thus, all diseases include all patients at different stages of disease development with or without medical interventions. However, we excluded the 2.5% highest and the 2.5% lowest CHE activities for each disease. As results, 137,305 CHE activities from patients suffering from 48 types of diseases and 3,387 CHE activities from healthy controls were used for data analysis.

Measurements Of Serum Che Activities
Various methods have been used to determine the serum CHE activity during the past [29], but the use is limited due to their limitations. Classical spectrophotometric Ellman's method is used widely to determine activity of CHE due to its advantages of being fast, simple, cheap and accurate. The method involves two steps. Firstly, butylthiocholine is converted into butyric acid and thiocholine under the action of serum CHE. Thiocholine is then reacted with 5,5'-dithiobis-(2-nitrobenzoic) spontaneously, which produces 5thio-2-nitrobenzoic acid and 2-nitrobenzoic acid-5-thiocholine [30,31]. The activity of CHE in the sample is calculated by measuring the increased absorbance of 5-thio-2-nitrobenzoic acid at 410 nm. The serum CHE assay used by the clinical lab in our hospital is "the continuous monitoring method" with the same principle of the spectrometric Ellman's method.
ROC analysis. ROC curves were plotted using SPSS v26 (IBM, Armonk, US). Youden's indices were calculated using ROC curve coordinates in order to determine AUCs, accuracy, sensitivity and speci city at the point where test performance is optimal.

Statistical analysis
All data retrieved was analyzed with RStudio V.1.3.1073 (RStudio, Boston, USA) and python libraries 3.8 (Anaconda Software Distribution). The results were demonstrated as median values and means ± standard deviation (SD). Standard t-test was used to compare the clinical characteristics of subjects in the speci c disease and control groups. Median levels of serum CHE activities between groups were compared by means of the Mann-Whitney U-test. Groups were compared using the Kruskal-Wallis test (a non-parametric one-way ANOVA). Logistic regression was used to test the interactive effects of other variables on the observed association. P < 0.05 was considered as being statistically signi cant.

Results
Based on the lab data of serum CHE activities retrieved from the clinical lab of our hospital over the past 5 years, the number of cases, mean (standard deviation,SD), median ( interquartile ranges) and P values for each of the 48 diseases in comparison to healthy controls were summarized in Table 1. To evaluate the diagnostic properties of serum CHE activities as serum biomarkers, the receiving operator curve (ROC) analysis was performed for 48 different types of diseases (Supplemental Fig. 1). Based on the ROC analysis, the area under the curve (AUC), accuracy, sensitivity and speci city for all diseases were listed in Fig. 1. Among the 48 types of diseases studied, 39 of them had the AUCs over 0.60 ranging from 0.60 to 1.00. Remarkably, the serum CHE activity was the best biomarker for HE among all 48 diseases with the AUC of 1.00, sensitivity 100%, and speci city 99%, which were followed by sepsis (AUC = 0.97), liver cancer (AUC = 0.92), myeloproliferative disorder (AUC = 0.92), and pancreatitis (AUC = 0.88).
Interestingly, the serum CHE activities had the lowest AUCs for breast cancer and nephrotic syndrome, being 0.51 and 0.52, respectively.
We further noticed that all kidney-related diseases including diabetic nephropathy, gout, kidney cancer, nephritis, lupus erythematous, and nephrotic syndrome and all female-related cancers including cervical, ovarian, and breast cancers had relatively low sensitivities (0.14-0.43) and high speci cities (0.79-0.95) when CHE activities were calculated as serum biomarkers (Fig. 1).
To comprehend the observations, we divided the 48 diseases into 6 major categories as cancers, autoimmune diseases, cardio-and cerebrovascular diseases, blood-related diseases (including blood cancers), kidney diseases, and others with different color-code. To visualize the results, we made boxed plots of serum CHE with lower quartile (25%), median (50%), and upper quartile (75%) ranges, and 95% con dent intervals marked for 48 clinically de ned human diseases and healthy controls and shown in HE had not only the lowest median level (50%) but also the lowest 2.5%, 25%, 75%, and 95% of serum CHE activities among all diseases. In contrast, nephrotic syndrome was the only disease that had 75% of serum CHE activities higher than that of control. In addition, diabetic nephropathy, hepatitis, and nephritis were the three diseases that had 97.5% of serum CHE activities higher than that of control. These observations led to speculate that the extreme serum CHE activities were the characteristics of speci c diseases. We then assumed that further studying such a relationship might reveal their relationship with HE and cirrhosis.
Thus, we analyzed the statistical features of serum CHE activities in six classes of diseases against HE and cirrhosis according to mean values, standard deviation, 2.5, 25, 50, 75, and 97.5 percentiles. The results of two component analysis of all 48 diseases were shown in Fig. 3.
The apparent clustering of the same category of diseases, such as cancers (at upper side of chart), autoimmune disease (at left side of chart), blood-related diseases (at low right of chart) based on the statistical analysis. Except for HE, diseases directly related to the kidney, such as nephritis, diabetic nephropathy and nephrotic syndrome, were clustered at the low left side of chart. Most impotantly, the two component analysis revealed HE resembled preeclampsia and uremia whereas cirrhosis resembled multiple myeloma, leukemia, myeloproliferative disorder, and liver cancer as shown in Fig. 3.

Discussion
All 48 different types of diseases studied in the current study had the lower median serum CHE activities than that of healthy controls ( Table 1), implicating that decreased serum CHE might be a common feature of human diseases. Most of publications reported the dynamic changes of serum CHE activities in one speci c disease [32][33][34][35][36][37][38][39][40], we performed the systematical comparison of serum CHE activities (Table 1) and CHE as biomarkers (Fig. 1) in 48 different types of diseases for the rst time. Moreover, serum CHE had almost perfect AUC, accuracy, sensitivity and speci city for HE diagnosis (Fig. 2). Meanwhile, serum CHE activities were also decent serum biomarkers for liver cirrhosis, sepsis and pancreatitis (Fig. 2). Furthermore, the two component analysis results shown in supplemental Fig. 3 and Fig. 3 revealed the resemblance of HE to preeclampsia and uremia were more than that to cirrhosis, which provided new insight in understanding HE etiology beyond liver failure. Thus, understanding the molecular mechanism behind the data presented in current study may be helpful in discovering new molecular targets for HE prevention and treatment.
Increased blood ammonia levels have been considered as the main pathophysiological mechanism of HE [41,42] by causing swelling astrocytes, activation of microglia and neuroin ammation [43][44][45][46][47][48][49][50]. Microglia activation can promote the proliferation and release of pro-in ammatory cytokines [50], and the in ammation state can eventually lead to neuronal death [51]. In addition, astrocytes cultivated with proin ammatory cytokines and interferon gamma (INF-γ) becomes edematous,which indicates the potential function of pro-in ammatory cytokines in the process [52,53]. Several studies have shown that severe systemic in ammation can induce [54][55][56] and aggravate HE in cirrhosis [57]. Coltart et al [58] demonstrated that sepsis can induce HE by affecting blood ammonia metabolism and promoting the release of pro-in ammatory mediators. The current study showed that the resenblance of HE to sepsis (Fig. 3), which is consistent with their conclusion.
In ammation plays a vital role in neurodegenerative diseases [51,59], such as Alzheimer's disease [60], dementia with Lewy bodies [61], Wilson's disease [62], Parkinson's disease [63], multiple sclerosis [64]. Serum CHE activities is closely related to the in ammation and is involved in many pathophysiological processes of nervous system diseases [65]. Darreh-Shori et al [66] have shown that serum CHE plays an important role in regulating intrathecal cytokine and activity of cholinoceptive glial cells in Alzheimer's disease. Macdonald et al [67] showed that CHE has good sensitivity and speci city in the diagnosis of Alzheimer's disease. However, the retrospective CHE activities retrieved in our current study did not provide enough independent test results for other neurodegenerative diseases. Further prospective studies should be useful in addresing these issues.
Chronic hepatitis, especially chronic hepatitis B, is the most common cause of cirrhosis and hepatic cancer in China. Hepatic diseases usually progress from hepatitis to cirrhosis and hepatic cancer. Previous studies have shown that CHE levels of hepatitis and cirrhosis are lower than that in healthy controls, which is consistent with the result of the study. Our data showed that serum CHE levels in HE was the lowest not only in liver-related diseases [68]but also in all other diseases included in current study.

Conclusions
The current study shows that serum CHE was an almost perfect serum biomarker for patients suffering HE at all stages. The resemblance of HE to preeclampsia and uremia based on CHE activities provided new insight in understanding HE beyond liver failure. The etiology of HE deserves further exploration. The Hospital Ethics Review Board of Qingdao University approved current study. All research was performed in accordance with relevant guidelines/regulations and informed consent was obtained from all participants and/or their legal guardians.

Consent for publication
All authors read and approved the manuscript and the nal submission.
Availability of data and materials. All data les are available upon request. Correspondence and requests for the data les should be addressed to H.P and L.Z.
Data analysis. Correspondence and requests for detailed data analysis should be addressed to H.P, Y.G., and L.Z.