FormalPara Key Points

Liver cirrhosis decreases the activity of most UDP-glucuronosyltransferases (UGTs) of the UGT1A and UGT2B subfamilies.

The effect on activity is quantitatively smaller than expected from the hepatic UGT protein content and is related to the severity of liver cirrhosis.

Predictions regarding systemic UGT activity based on in vitro data are difficult because of the limited substrate selectivity and extrahepatic expression of UGTs.

1 Introduction

It is well established that phase I reactions or the functionalization of many drugs are impaired in patients with liver cirrhosis [1, 2], possibly causing dose-dependent adverse reactions if the dose is not adjusted [3]. In comparison to phase I reactions, much less information is available about the activity of conjugation or phase II reactions in patients with liver cirrhosis. The most important conjugation reaction is glucuronidation, which is responsible for > 90% of the hepatic conjugation reactions of drugs [4]. Humans have 22 UDP-glucuronosyltransferases (UGTs), which are grouped into four families (1, 2, 3, 8) and five subfamilies (1A, 2A, 2B, 3A, 8A) [4, 5]. Out of the five subfamilies, only the two largest, subfamilies 1A and 2B, contribute relevantly to the glucuronidation of xenobiotics such as drugs. While the liver has the highest UGT content, UGTs are also expressed in the intestinal tract and in the kidney. UDP-glucuronosyltransferases are considered to be less substrate specific than cytochrome P450 enzymes (CYPs), but for many drugs or drug metabolites that are glucuronidated, one or more UGTs have been identified as the principal glucuronosyltransferases [4, 6].

In patients with liver cirrhosis, the assessment of glucuronide formation from drugs yielded variable results. In early studies, it was shown that the glucuronidation of oxazepam and lorazepam was maintained in patients with liver cirrhosis compared to healthy subjects [7, 8]. In contrast, the glucuronidation of zomepirac, a non-steroidal anti-inflammatory drug carrying a carboxyl group, was impaired in patients with liver cirrhosis [9]. Therefore, the authors of this study speculated that glucuronidation of hydroxyl groups (in oxazepam and lorazepam) is maintained, whereas the glucuronidation of carboxyl groups (in zomepirac) is impaired in patients with liver cirrhosis. This hypothesis could be rejected by the findings that the hepatic extraction of morphine, which is metabolized by glucuronidation of the hydroxyl groups in position 3 and 6, is impaired [10] and, more directly, that the glucuronidation of morphine is reduced in patients with liver cirrhosis [11]. Similarly, the metabolism of paracetamol, which is metabolized mainly by glucuronidation of the hydroxyl group, is impaired in patients with liver cirrhosis [12,13,14].

A focus on the UGT enzymes involved in drug metabolism, rather than drug chemistry, is therefore suggested. Because zomepirac is glucuronidated by UGT2B7 [15], oxazepam, lorazepam, and temazepam by UGT2B15 [4], morphine by UGT2B7 [4], and paracetamol by UGT1A1, 1A6, 1A9, and 2B15 [4, 16], it is possible that liver cirrhosis affects the individual UGTs differently, similar to the effect of liver cirrhosis on CYP activity [17, 18]. A possible approach to study this hypothesis is to assess the glucuronidation of additional substrates of these and other UGTs and to compare the results with published data regarding UGT involvement, if available.

We have recently conducted a large study in patients with liver cirrhosis to assess the activity of the six most important CYPs using the Basel phenotyping cocktail [17], which contains six substrates in a subtherapeutic dosage in a single capsule [19] (see Table 1 of the Electronic Supplementary Material [ESM] for pharmacokinetic data about these substrates). In this project, we determined the plasma concentrations of the parent drugs and of a specific metabolite with and without deglucuronidation of the plasma samples before analysis by liquid chromatography tandem mass spectrometry. This approach allowed us to obtain individual plasma concentration–time profiles of the parent drug and the specific primary metabolite, and to derive plasma concentration–time profiles of glucuronides of the parent drug and of the primary metabolite. These data enabled us to not only characterize the pharmacokinetics of the glucuronides but also to calculate the metabolic ratio (MR) for glucuronide formation (AUCglucuronide parent or metabolite/AUCparent or metabolite). Metabolic ratios are considered to reflect the activity of the enzyme performing the formation of the metabolite [20, 21]. We have validated this approach for the formation of the primary metabolites of the substrates of the Basel phenotyping cocktail in healthy subjects [22] and provided data that the approach also works for the assessment of glucuronide formation for the same substrates and their primary metabolites [23]. Specific aims of the study were to gain more information about the effect of liver cirrhosis on the glucuronidation of parent compounds and/or specific metabolites of the substrates in the Basel phenotyping cocktail and to investigate the possibility that glucuronides accumulate in patients with liver cirrhosis.

2 Materials and Methods

2.1 Clinical Study

We have described the clinical study in detail in a previous publication [17]. In brief, this was a single-center clinical study (ClinicalTrials.gov, ID: NCT03337945) in 36 patients with liver cirrhosis and 12 matched control subjects. The study was approved by the local ethics committee EKNZ (Ethikkommission Nordwestschweiz/Zentralschweiz, BASEC-Nr: 2017-01329) and was conducted in accordance with good clinical practice guidelines and the current version of the Declaration of Helsinki.

The 36 patients with liver cirrhosis had been hospitalized in the University Hospital of Basel and were studied after discharge, when they were in a stable condition. At the day of the study, the study subjects ingested a capsule containing the “Basel phenotyping cocktail” consisting of 10 mg of caffeine (CYP1A2 substrate), 50 mg of efavirenz (CYP2B6 substrate), 12.5 mg of flurbiprofen (CYP2C9 substrate), 10 mg of omeprazole (CYP2C19 substrate), 12.5 mg of metoprolol tartrate (CYP2D6 substrate), and 2 mg of midazolam (CYP3A substrate). Venous blood samples were obtained 5 min, 15 min, 30 min, 45 min, 1 h, 2 h, 3 h, 4 h, 6 h, 8 h, 12 h, and 24 h after cocktail administration into EDTA-coated tubes. Plasma samples were centrifuged at 1500 g for 10 min at 4 °C and the plasma stored at −80 °C until analysis.

The study subjects were examined clinically and biochemically to assess hepatic and renal function as shown in Table 2 of the ESM. This allowed us to calculate the Child score [24], which was used to categorize the patients with liver cirrhosis. The glomerular filtration rate was estimated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation [25]. We also determined the metabolic capacity of the liver in all subjects included by measuring the elimination of galactose (galactose elimination capacity) [26, 27]. For that, we infused a galactose solution (40% w/v) at a dose of 0.5 mg/kg body weight. We determined the galactose concentration in serial plasma and urine samples and calculated the galactose elimination capacity as described in a previous investigation [17].

2.2 Study Drug

The “Basel phenotyping cocktail” capsules were prepared under Good Manufacturing Practice conditions by Dr. Hysek Apotheke (Biel, Switzerland) as described previously [28].

2.3 Bioanalysis

The substrates of the “Basel phenotyping cocktail” were quantified in plasma via high-performance liquid chromatography tandem mass spectrometry as described previously [20, 23]. The plasma concentrations of all analytes (caffeine, efavirenz, flurbiprofen, omeprazole, metoprolol, and midazolam and their metabolites) were determined prior to deglucuronidation and after treatment with β-glucuronidase from Helix pomatia (≥ 100,000 units/mL; Sigma-Aldrich, Buchs, Switzerland). The analyses were conducted on a Shimadzu (Kyoto, Japan) ultra-high pressure liquid chromatography system, which was connected to an API 5500 QTRAP mass spectrometer (Sciex, Concord, ON, Canada). The concentrations of the respective aglucones were calculated as the difference after and before deglucuronidation. From the aglucone concentration, glucuronide concentrations were calculated assuming the attachment of one glucuronic acid moiety per aglucone using the molecular weights shown in Table 3 of the ESM. In the tables and figures, we provide the calculated glucuronide concentrations.

2.4 Pharmacokinetic Analysis and Statistics

The primary endpoint of the study was the effect of liver cirrhosis on the glucuronidation of the parent compound and/or of their specific metabolites. Therefore, we analyzed the samples before and after deglucuronidation. This allowed us to determine the area under the curve (AUC) of the non-glucuronidated precursors and their specific glucuronide and to calculate the MRs AUC0–inf glucuronide/AUC0–inf non-glucuronidated precursor. The MR was calculated both using AUCs expressed as µg/L × h or as nmol/L × h. The molecular weights used for the conversion of the measured µg/L to nmol/L are given in Table 3 of the ESM. The MR is considered to reflect the activity of the enzyme performing this reaction [20, 21].

We performed non-compartmental pharmacokinetic analyses of the plasma concentration–time curves of the study drugs and their metabolites using PKanalix (version 2019R1; Lixoft SAS, Abtony, France). We obtained maximal plasma concentration directly from the plotted plasma concentration–time profiles and the terminal elimination rates (elimination rate constant) from the terminal linear part of the semilogarithmic plasma concentration–time plots. The elimination half-lives were calculated as ln2/elimination rate constant. The AUC was calculated using the linear trapezoidal method.

Differences between the study groups (controls and patients with Child A, Child B, and Child C liver cirrhosis) were analyzed by a one-way analysis of variance followed by the Holm–Sidack method versus the control values to localize differences. All statistical analyses were conducted with GraphPad Prism 8 (Graphpad Software, La Jolla, CA, USA). Statistical significance was defined as p < 0.05.

3 Results

The subjects included in the study are characterized in Table 2 of the ESM. Although we tried to match the study subjects, patients with liver cirrhosis tended to be older than control subjects and to have a lower body weight and body mass index. As expected, patients with liver cirrhosis had higher transaminase and alkaline phosphatase activities, higher serum bilirubin, and lower serum albumin concentrations as well as lower prothrombin ratios and factor V concentrations than control subjects. The galactose elimination capacity, a marker of the metabolic activity of the liver [26, 27], was clearly reduced in patients with liver cirrhosis. Patients with Child C cirrhosis tended to have higher serum creatinine concentrations and a reduced glomerular filtration rate compared with control subjects.

Liver cirrhosis affected the glucuronidation differently, depending on the extent of liver damage and the substrate investigated. As expected, the xanthines caffeine and paraxanthine were almost not glucuronidated (Fig. 1, Fig. 1 of the ESM and Table 1) [29]. Nevertheless, our data suggest that a small portion of caffeine and, possibly, also of paraxanthine can be glucuronidated (Fig. 1, Fig. 1 of the ESM). The abundance (AUC0–inf) of caffeine-glucuronide increased in parallel to the abundance of caffeine in patients with liver cirrhosis, leaving the MR unchanged (Fig. 2A and Table 1). In contrast, the abundance of paraxanthine-glucuronide decreased in patients with liver cirrhosis, which was associated with a decrease in the corresponding MR (Fig. 3A and Table 1).

Fig. 1
figure 1

Bland–Altman plots (difference plots) of plasma samples before and after deglucuronidation. Thirty-six patients with liver cirrhosis (n = 16 patients with Child A cirrhosis, n = 15 patients with Child B cirrhosis, n = 5 patients with Child C cirrhosis) and 12 matched control subjects were treated with the Basel phenotyping cocktail containing six substrates. Subjects abstained from caffeine ingestion for the 24 h preceding and during the study day. Serial blood samples were obtained and analyzed before and after deglucuronidation. The figure shows the difference between the concentrations after and before deglucuronidation of the substrates and their primary metabolites, which is plotted against their average (Bland–Altman plot). Negative values are plotted in red. The figure shows that caffeine, paraxanthine, metoprolol, midazolam, 8-hydroxyefavirenz, 4-hydroxyflurbiprofen, 5-hydroxyomeprazole, and 1′-hydroxymidazolam are partially glucuronidated

Full size image
Fig. 2
figure 2

Plasma concentration–time plots (1A–C) and area under the curve (AUC) and metabolic ratios (2A–C) of the glucuronides of the substrates of the Basel phenotyping cocktail showing significant glucuronidation. Thirty-six patients with liver cirrhosis (n = 16 patients with Child A cirrhosis, n = 15 patients with Child B cirrhosis, n = 5 patients with Child C cirrhosis) and 12 matched control subjects were treated with the Basel phenotyping cocktail containing six substrates. Serial blood samples were obtained, and the plasma concentration of the substrates determined before and after deglucuronidation. Glucuronide concentrations were calculated as the difference between the values after and before deglucuronidation. Values are shown as the mean ± standard error of the mean. h h

Full size image
Table 1 Pharmacokinetic variables of caffeine, paraxanthine, and their glucuronides
Full size table
Fig. 3
figure 3

Plasma concentration–time plots (1A–E) and area under the curve (AUC) and metabolic ratios (2A–E) of the glucuronides of the primary metabolites of the substrates of the Basel phenotyping cocktail showing significant glucuronidation. Thirty-six patients with liver cirrhosis (n = 16 patients with Child A cirrhosis, n = 15 patients with Child B cirrhosis, n = 5 patients with Child C cirrhosis) and 12 matched control subjects were treated with the Basel phenotyping cocktail containing six substrates. Serial blood samples were obtained, and the plasma concentration of the primary metabolites of the substrates determined before and after deglucuronidation. Glucuronide concentrations were calculated as the difference between the values after and before deglucuronidation. Values are shown as the mean ± standard error of the mean. h hours

Full size image

Concerning efavirenz, we found no clear evidence for glucuronidation of the parent compound in the plasma samples, although N-glucuronidation of efavirenz has been reported [30, 31] (Fig. 1 and Fig. 2 of the ESM). In comparison, 8-hydroxyefavirenz, which was barely detectable in plasma, was rapidly glucuronidated (Fig. 1, Fig. 2 of the ESM and Table 2). The abundance of 8-hydroxyefavirenz-glucuronide (AUC0–inf OHefa-g) showed a tendency to increase with the severity of liver cirrhosis (Table 2 and Fig. 3B). When related to the AUC of 8-hydroxyefavirenz (AUC0–inf OHefa), we observed a significant increase in the AUC0–inf OHefa-g/AUC0–inf OHefa ratio, reaching a factor of 3 for patients with Child C cirrhosis compared with control subjects (Fig. 3B). The ratio AUC0–inf OHefa-g/AUC0–inf OHefa showed a weak, but significant, negative correlation with the estimated glomerular filtration rate (r = − 0.458, p < 0.05), but no significant correlation with markers of cholestasis such as alkaline phosphatase (r = 0.120, p > 0.05) or serum bilirubin (r = 0.124, p > 0.05). These findings suggest that impaired renal function may have contributed to the increase in the AUC0–inf OHefa-g/AUC0–inf Ohefa ratio in patients with liver cirrhosis.

Table 2 Pharmacokinetic variables of efavirenz, 8-hydroxyefavirenz, and 8-hydroxyefavirenz glucuronide
Full size table

Similar to efavirenz, we detected no relevant glucuronidation of flurbiprofen (Fig. 1, Fig. 3 of the ESM and Table 3). This was surprising because flurbiprofen contains a carboxyl group that is glucuronidated in structurally related compounds [9]. In addition, flurbiprofen carboxyl-glucuronide formation by human liver microsomes has been reported [32] and flurbiprofen carboxyl-glucuronides have been detected in the urine of humans ingesting flurbiprofen [33]. Possible reasons for the lack of flurbiprofen-glucuronide detection could be cleavage during the analytical procedure and/or the indirect measurement applied in this study, which requires that the sum of the drug and glucuronide is measurably higher than that of the drug alone. In comparison, 4-hydroxyflurbiprofen was efficiently glucuronidated (Fig. 3C, Fig. 3 of the ESM and Table 3), whereby the abundance of 4-hydroxyflurbiprofen-glucuronide (AUC0–inf OHflu-g) closely paralleled the abundance of 4-hydroxyflurbiprofen (AUC0–inf OHflu). Accordingly, the ratio AUC0–inf OHflu-g/AUC0–inf OHflu was not affected by liver cirrhosis (Fig. 3C and Table 3).

Table 3 Pharmacokinetic variables of flurbiprofen, 4-hydroxyflurbiprofen, and 4-hydroxyflurbiprofen glucuronide
Full size table

Similar to efavirenz and flurbiprofen, we detected no relevant glucuronidation of omeprazole (Fig. 1, Fig. 4 of the ESM and Table 4). However, 5-hydroxyomeprazole was partially glucuronidated (Fig. 3D, Fig. 4 of the ESM and Table 4). Like the glucuronidation of 4-hydroxyflurbiprofen, the abundance of 5-hydroxyomeprazole-glucuronide (AUC0–inf OHome-g) was parallel to the abundance of 5-hydroxyomeprazole (AUC0–inf OHome). Accordingly, the ratio AUC0–inf OHome-g/AUC0–inf OHome was not affected by the presence of liver cirrhosis (Fig. 3D and Table 4).

Table 4 Pharmacokinetic variables of omeprazole, 5-hydroxyomeprazole, and 5-hydroxyomeprazole glucuronide
Full size table

For metoprolol, we found evidence that the parent compound is partially glucuronidated (Figs. 1 and 2B, Fig. 5 of the ESM, and Table 5), whereas the primary metabolite α-hydroxymetoprolol was not glucuronidated (Fig. 1, Fig. 5 of the ESM). The finding that the parent compound (metoprolol) can be glucuronidated is in agreement with a previous study of ours, where we obtained comparable results [23]. As shown in Fig. 2B and Table 5, the ratio AUC0–inf met-g/AUC0–inf met was significantly decreased in Child C patients compared with control subjects.

Table 5 Pharmacokinetic variables of metoprolol, metoprolol glucuronide, and α-hydroxymetoprolol
Full size table

For midazolam, we detected glucuronidation of both the parent compound (Fig. 2C and Fig. 6 of the ESM and Table 6) and the primary metabolite 1′-hydroxymidazolam (Fig. 3E, Fig. 6 of the ESM, and Table 6). The AUC0–inf for 1′-hydroxymidazolam-glucuronide increased with the severity of liver cirrhosis (Fig. 3E). The AUC0–inf of the glucuronides of midazolam and 1′-hydroxymidazolam normalized to the corresponding AUC0–inf of midazolam and 1′-hydroxymidazolam, respectively decreased with the severity of liver cirrhosis because of a higher abundance of midazolam and 1′-hydroxymidazolam than the corresponding glucuronides (Figs. 2C and 3E and Table 6).

Table 6 Pharmacokinetic variables of midazolam, 1′-hydroxymidazolam, and their glucuronides
Full size table

4 Discussion

The aims of the current study were to gain more information about the effect of liver cirrhosis on the systemic glucuronidation capacity of specific compounds and to detect possible accumulation of glucuronides. For that, we obtained full pharmacokinetic profiles of the six substrates and their primary metabolites of the Basel phenotyping cocktail in patients with liver cirrhosis and matched control subjects before and after deglucuronidation of the corresponding plasma samples. The results show that liver cirrhosis can affect the systemic glucuronidation capacity depending on the substrate considered and on the extent of liver damage. The dependency on the severity of liver disease is illustrated by the finding that impaired glucuronidation was only observed in patients with Child B or C cirrhosis, but not in patients with Child A cirrhosis.

The effect of liver cirrhosis on the glucuronidation of different substrates of the Basel phenotyping cocktail and of their specific primary metabolite is summarized in Table 7. Regarding the MR (AUCparent glucuronide/AUCparent or AUCmetabolite glucuronide/AUCmetabolite), which we considered to reflect the activity of the UGT(s) involved, liver cirrhosis did either not affect (caffeine, 4-hydroxyflurbiprofen, 5-hydroxyomeprazole) or decrease the glucuronidation (paraxanthine, metoprolol, midazolam, 1′-hydroxymidazolam). 8-Hydroxyefavirenz was an exception because the MR increased with the severity of liver cirrhosis. Regarding the negative correlation with the glomerular filtration rate and the lack of a significant correlation with markers of cholestasis, this increase could possibly be explained with impaired renal function in patients with liver cirrhosis. Considering the limited strength of this correlation, it must be realized that the estimated glomerular filtration rate may overestimate the glomerular filtration rate in patients with liver cirrhosis [1, 34], potentially weakening the correlation. Because a significant correlation cannot be taken as a proof for an association, the relationship between the elimination of 8-hydroxyefavirenz-glucuronide and renal function would have to be confirmed in a clinical study.

Table 7 Effect of liver cirrhosis on the glucuronidation of the parent compounds and specific primary metabolites of the Basel phenotyping cocktail
Full size table

As shown in Table 7, the specific UGTs performing the glucuronidation are known for some of the reactions monitored in the current study. Glucuronidation of 8-hydroxyefavirenz is performed by members of the UGT subfamily 1A (1A1, 1A3, 1A7, 1A8, 1A9, 1A10) and subfamily 2B (2B7) [30]. While UGT2B7 produces the carboxyl-glucuronide of flurbiprofen [32], the UGTs responsible for 4-hydroxyflurbiprofen glucuronidation are currently not known. Interestingly, 4-hydroxyflurbiprofen is mainly glucuronidated at the carboxyl and not at the 4-hydroxy-group [33], suggesting that UGT2B7 may be responsible for 4-hydroxyflurbiprofen glucuronidation. Midazolam is N-glucuronidated by UGT1A4 [35, 36] and 1′-hydroxymidazolam is N-glucuronidated by UGT1A4 and 1′-hydroxy-glucuronidated by UGT2B4 and 2B7 [36, 37]. Glucuronidation of 5-hydroxyomeprazole has been described in dogs [38], but, to the best of our knowledge, not in humans. Accordingly, the specific UGT performing this reaction is not known. Similarly, the formation of a glucuronide of metoprolol has been suspected in a previous investigation [23] but the UGT performing this reaction is not known. For the xanthines, N-glucuronide formation has been described in mice but not in humans [29] and the responsible UGT is not known.

The data in Table 7 show that the effect of liver cirrhosis on the exposure (AUC) of a specific glucuronide can be different from the effect on the MR. Considering the definition of the MR, this must be because of changes in the AUC of the parent compound (substrate of the phenotyping cocktail or primary metabolite), which can be affected by liver cirrhosis [17, 18]. To judge the effect on glucuronide formation, it is, therefore, important to consider the MR and not only the AUC of the glucuronides produced.

Table 7 illustrates that liver cirrhosis may affect the activity of UGTs of subfamily 1A and subfamily 2B. Regarding the UGT1A subfamily, the results suggested a decreased activity of UGT1A4 (N-glucuronidation of midazolam and 1′-hydroxymidazolam). This finding agrees with that for the UGT1A4 substrate lamotrigine [4], whose elimination is impaired in patients with liver cirrhosis [39]. Considering also the glucuronidation of paracetamol, which is realized by UGT1A1, 1A6, and 1A9 [4, 16] and is impaired in patients with liver cirrhosis [12,13,14], liver cirrhosis appears to have a broad negative impact on the activity of the members of the UGT1A subfamily, despite the fact that UGT1A1, 1A4, 1A6, and 1A9 have an important extrahepatic expression [4, 40].

Considering the effect of liver cirrhosis on the glucuronidation activity of members of the UGT2B subfamily, the current study suggests a decrease in the activity of UGT2B4 and 2B7 (impaired glucuronidation of 1′-hydroxymidazolam). UGT2B4 is mainly expressed in the liver and has a high activity [4]. However, although UGT2B4 shows an overlap regarding endogenous substrates with UGT2B7 [41], UGT2B7 is considered to be more important for the glucuronidation of xenobiotics than UGT2B4 [4]. The suggested decrease in the glucuronidation of 1′-OH-midazolam may, therefore, mainly result from a decreased activity of UGT2B7. However, if UGT2B7 were responsible for the glucuronidation of 4-OH-flurbiprofen, this would be in contradiction to this statement because the MR of 4-OH-flurbiprofen was not affected by liver cirrhosis. In contrast, the glucuronidation of morphine, which is a UGT2B7 substrate [4], is impaired in patients with liver cirrhosis [11]. Similarly, the glucuronidation of the UGT2B7 substrate zidovudine [4] has been shown to be impaired by hepatic microsomes from patients with liver cirrhosis [42] and in a clinical study in patients with liver cirrhosis [43]. In addition, the glucuronidation of the UGT2B7 substrate zomepirac [15] is reduced in patients with liver cirrhosis [9]. Regarding UGT2B7, it has also to be considered that there is a large extrahepatic activity, mainly in the gut and in the kidneys [4]. Total UGT2B7 expression in the kidneys has been reported to correspond to approximately 25% of total liver expression [40] and is, therefore, relevant for systemic UGT2B7 activity. Assuming that the extrahepatic UGT2B7 activity is not negatively affected by liver cirrhosis, extrahepatic glucuronidation may at least partially compensate for impaired hepatic activity. In addition, extrahepatic activity may explain variable effects of liver cirrhosis on UGT2B7 substrates, possibly because of different tissue concentrations reached and/or different affinities for extrahepatic UGTs. This implies that predictions regarding the effect of liver cirrhosis on the glucuronidation of specific drugs are difficult and that in vitro findings should be confirmed by clinical studies. Interestingly, the metabolism of the benzodiazepines lorazepam, oxazepam, and temazepam, which have a free hydroxyl group that is glucuronidated by UGT2B15 [4], was not or only minimally affected by liver cirrhosis [7, 8, 44]. These findings suggest that the activity of UGT2B15 is not severely affected by liver cirrhosis, despite its preferential hepatic localization [4]. Based on these considerations, liver cirrhosis appears to affect the members of the UGT2B subfamily differently.

Recent studies describe the protein expression of UGTs in hepatic microsomes isolated from patients with liver cirrhosis [45] and from patients with non-alcoholic steatohepatitis [46]. While El-Khateeb et al. reported a 70–90% decrease in the protein expression of UGT1A4, 1A6, 1A9, 2B4, 2B7, and 2B15 in hepatic microsomes from patients with liver cirrhosis [45], Hardwick et al. found a decrease in the protein expression of UGT1A6, but not of 1A1, 1A9, and 2B10 in hepatic microsomes from patients with non-alcoholic steatohepatitis-associated fibrosis or cirrhosis [46]. Considering the study of El-Khateeb et al., the data would predict a massive decrease in the activity of most UGTs in patients with liver cirrhosis. This is obviously in contradiction with the current study and other clinical studies in this population [7, 8, 44]. Reasons for this discrepancy may be the extrahepatic expression of UGTs and non-hepatocyte contributions in the microsomal preparations used. This underscores the necessity to confirm results obtained in vitro by appropriate in vivo studies.

Liver cirrhosis did not significantly affect the elimination half-lives of the glucuronides of theophylline, 8-hydroxyefavirenz, 4-hydroxyflurbiprofen, midazolam, and 1′-hydroxymidazolam and increased the elimination half-lives of the glucuronides of caffeine, 5-hydroxyomeprazole, and metoprolol by a factor of approximately 3. For the glucuronides of caffeine, 5-hydroxyomeprazole, and metoprolol, we observed a similar increase in the elimination half-life of the direct precursors, suggesting that the increase in the elimination half-life of the glucuronides was due to a decrease in the elimination of the respective precursor. Based on these considerations, we assume that liver cirrhosis did most probably not impair the elimination of the glucuronides determined in this study.

The current study has important limitations. Although we have validated the MR approach for the substrates of the Basel cocktail in healthy volunteers [22], this may not be true for patients with liver cirrhosis and for the formation of glucuronides. Regarding glucuronide formation, we could demonstrate an increase in the MR for all glucuronides determined in the current study in healthy volunteers treated with the constitutive androstane receptor (CAR) activator metamizole [23], suggesting that the MR approach is also working for glucuronidation. While this may be true for healthy subjects, this may not be the case for patients with liver cirrhosis. Because the AUC of a drug or metabolite is not only a function of the dose (or metabolite formation), but also of the volume of distribution and elimination, the effect of liver cirrhosis on these factors must be considered. As discussed above, we assume that liver cirrhosis did not affect the elimination of the glucuronides determined. Assuming mainly extracellular localization of the glucuronides formed and considering the increase in the extracellular water in patients with liver cirrhosis (approximately by a factor of 1.2 and 1.8 compared with controls in patients with Child A/B cirrhosis and Child C cirrhosis, respectively [47]), the AUC would decrease by approximately 20% and 45% in patients with Child A/B cirrhosis and Child C cirrhosis, respectively, which would not change the interpretation of the data provided in Tables 1, 2, 3, 4, 5 and 6. Furthermore, compartmental population pharmacokinetic modeling of the data could possibly have provided more accurate results regarding glucuronide formation than the current approach based on MRs. However, the Basel phenotyping cocktail has been validated using MRs wherefore we decided to analyze the data in this manner. Last, we determined the glucuronide concentrations indirectly via deglucuronidation. The precision of this method may be limited at very low glucuronide concentrations and double glucuronidation of compounds will be missed.

5 Conclusions

The results of the current study suggest that liver cirrhosis decreases the activity of most UGTs of the UGT1A and UGT2B families. The effect is quantitatively smaller than expected from the reported reductions in the hepatic UGT protein content and is related to the severity of liver cirrhosis with patients with Child A cirrhosis not being affected. A relevant accumulation of glucuronides in patients with liver cirrhosis was not observed in the current study but this may be different in patients with a higher degree of cholestasis and/or renal dysfunction. Predictions regarding the effect of liver cirrhosis on the pharmacokinetics of drugs glucuronidated based on in vitro data are difficult because of the limited selectivity and extrahepatic expression of UGTs.