The present study demonstrated that different liver metabolic phenotypes result from exposure to different types of anesthetic agents. In particular, the liver metabolic phenotype after exposure to propofol was markedly different from that after exposure to other anesthetic agents. However, the differences in liver metabolic phenotype between anesthetics decreased over time.
In this study, experiments were conducted using intravenous anesthetics (propofol and dexmedetomidine) and inhaled anesthetics (isoflurane and sevoflurane). These anesthetics are extensively used in everyday clinical practice. Metabolomic changes in the livers of rats anesthetized with one minimum alveolar concentration of inhaled anesthetics (sevoflurane at 2.4%, isoflurane at 1.5%) [10, 11] or with continuous infusion of intravenous anesthetics (propofol at 600 μg/kg/min, dexmedetomidine at 1 μg/kg/min) [12, 13] were investigated. Infusion of propofol at 600 μg/kg/min is a median effective dose (ED50) for rats. These doses were chosen because lower doses may not induce adequate anesthesia and because higher doses can produce hemodynamic changes that can independently alter liver metabolism. In a previous study of changes in brain metabolism over time in response to anesthetics, rats were anesthetized for 2 h and at 6 h ; the 6-h anesthesia groups exhibited clear metabolic changes. Therefore, in the present study, it was thought that 6 h was a sufficient time period to produce metabolic changes, but sufficiently short to prevent any toxicity.
Separation of PCA scores between the different groups was related to differences in 3-D-HB, choline, phosphocholine, GPC, TMAO, taurine, mannitol, betaine and glucose levels (Figure 3C). 3-D-HB is a ketone body derived from fatty acids metabolism in liver mitochondria (lipid β-oxidation). In the context of hepatic insufficiency, lipid β-oxidation and gluconeogenesis accelerates, and acetyl-coenzyme A (acetyl-CoA) derived from glycolysis and lipid β-oxidation exceeds the capacity of the Krebs cycle [18, 19]. Normally, ketone bodies are transported from the liver to other tissues, where they can be reconverted to acetyl-CoA to produce energy; impairment of the Krebs cycle leads to increased release of ketone bodies from the liver as fuel for other tissues .
A change in the level of hepatic glucose in anesthetized rats suggests an alteration in the rate of glycogenolysis and glycolysis that is consistent with mitochondrial impairment, and can lead to an inability to use pyruvate in the Krebs cycle and to enhancement of anaerobic metabolism . Thus, the changes of 3-D-HB and glucose likely reflect a general decrease in energy metabolism.
In addition, rats were not fed during this study. Thus, it is possible that changes in the level of glucose, mannitol and taurine may have resulted from the fasting state. Among the four anesthetics administered in this study, only propofol can provide calories due to its fat content. Indeed, propofol is a 1% fat preparation, which corresponds with 1.1 kcal/mL and a total of 7.3 kcal administered to rats during this study. Mannitol is a sugar alcohol, while taurine is related to bile acids and digestion. Thus, the fasting state may also affect levels of those substances.
Choline and phosphocholine are metabolic products of phosphatidylcholine, which is a major membrane constituent. GPC is also a membrane constituent, while TMAO is a product of choline degradation. Betaine helps to maintain cellular osmotic pressure when the cell membrane damaged. Increased levels of choline, phosphocholine, GPC, TMAO and betaine are associated with cell membrane disruption .
Changes of endogenous metabolites in hepatic tissue suggest that propofol was the most influential anesthetic on hepatic energy metabolism among the four anesthetics administered in the present study. PC scores showed that TMAO was the most effective metabolite that separated the propofol group from the other groups in the PC1 direction (Figure 3A, 3B,3C). The dexmedetomidine group was separately clustered from the other groups, but the extent of the difference on the PCA plot was smaller than the difference between the propofol group and the other groups. PC scores showed that GPC was effective in separation of the dexmedetomidine group from the other groups in the PC2 direction (Figure 3A,3B,3C). It is possible that the intravenous anesthetics affected the metabolism in the liver by different mechanisms. There was little difference in hepatic metabolites when comparing the inhalational anesthetic groups to the control group. However, the effect of inhalational anesthetics persisted at the 24-h time point while the effect of intravenous anesthetics on the liver resolved over time (Figure 4B), suggesting that propofol and dexmedetomidine were metabolized rapidly when compared with sevoflurane and isoflurane. The loading plot included in Figure 4A-4C remained static over time, and at the 48-h time-point, PCA plots were mixed, and the difference between groups became unclear (Figure 4C).
Sevoflurane and isoflurane are metabolized by hepatic cytochrome P450 CYP2E1 , and hepatotoxicity is rarely seen with these two agents [24–27]. In a previous study, there were no significant differences in liver function or total hepatic blood flow when comparing sevoflurane and isoflurane anesthesia . This finding is consistent with observations from the present study, in which PCA scores were relatively similar when comparing the inhalational anesthetic groups and the control group. Indeed, investigators have suggested that liver function is relatively preserved with inhalational anesthetics when compared with intravenous anesthetics [29, 30].
Propofol is rapidly metabolized in the liver by conjugation to glucuronide and sulfate to produce water-soluble compounds, and 68.3% of these compounds are excreted in the urine within 24 h . Dexmedetomidine is rapidly distributed and extensively metabolized in the liver. It undergoes conjugation (41%), N-methylation (21%) or hydroxylation followed by conjugation, and 85% of the resulting compounds are excreted in the urine within 24 h [32, 33]. Some reports suggest that dexmedetomidine preserves blood flow to the most vital organs (e.g., brain, heart, liver, kidneys)  and that there is no difference in clearance of indocyanine green or hepatic blood flow when comparing propofol and dexmedetomidine [32, 35]. By contrast, the PCA scores in the present study suggest that propofol has a larger effect on hepatic metabolism when compared with dexmedetomidine.
In previous studies, propofol administration did not affect hepatic arterial blood flow , but increase total liver blood flow, portal tributary blood flow, and hepatic oxygen consumption . The increase in oxygen consumption may reflect an increase in hepatic metabolic activity during the clearance of propofol and reflect propofol-induced changes in hepatic metabolism. Previous studies reported that the expression ratios of genes encoding DMEs were elevated under anesthesia and that the expression ratio of Ugt1a7 (a gene encoding DME) was elevated under propofol anesthesia. Ugt1a7 is biotransformation enzyme of the glucuronidation pathway that transforms small lipophilic molecules into water-soluble excretable metabolites , and propofol is mainly metabolized by glucuronidation in the liver. In addition, propofol is administered as an emulsion consisting of soybean oil, glycerol, and purified egg phosphatide, which may also affect hepatic metabolism. Further, propofol delivers calories in the form of fat, which may result in differences from other anesthetics and control groups on PCA in this study.
To investigate the utility of 1H-NMR for clinical applications, the hydrophilic fraction of the rat liver extracts were analyzed in this study. When the metabolic state is investigated for clinical purposes, the major target is the hydrophilic fraction. The extracted endogenous metabolites in the present study may have utility as a marker of liver metabolism, thereby enhancing the clinical utility of 1H-NMR.
Metabolomics is an exhaustive analysis, and we did not attempt to accurately quantitate individual NMR peaks in this study. Our main purpose was to compare the patterns of spectra using chemometric methods to visualize the metabolic phenotypes of the samples. Further investigations that target individual metabolites with refined analysis may be of benefit.
In the present study, metabolite profiling of the liver showed that anesthesia with propofol, dexmedetomidine, sevoflurane and isoflurane exerted different effects on rat liver metabolism. Although analysis of metabolic substances from the liver after exposure to various drugs has been conducted , an exhaustive analysis of liver metabolites after exposure to anesthetics has not previously been reported. Thus, the results from this study may enhance the clinical use of anesthetics.