Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Clinical and experimental studies of drug elimination in congestive heart failure have demonstrated that biotransformation of drugs may be impaired (Rissam et al., 1983; Shammas and Dickstein, 1988; Patel et al., 1990; Dunselman and Edgar, 1991; Ng et al., 1995). However, the majority of these studies have concentrated on the effects of heart failure on hepatic oxidative metabolism, and little is known of its effects on the capacity of the liver to eliminate drugs via the conjugation pathways.

Our previous studies have suggested that the impaired oxidative drug metabolism in heart failure is predominantly due to the effects of right heart failure and hepatic congestion. The aim of the current study was to determine whether right heart failure also has significant effects on hepatic drug conjugation, and the mechanisms responsible for any changes observed, using a well characterized experimental model of isolated right ventricular failure (RVF) in the rat (Ng et al., 1995).

We studied the effects of RVF on the elimination by the isolated perfused rat liver (IPRL) of p-nitrophenol (PNP), a drug which is eliminated predominantly via hepatic glucuronidation (Ghabrial et al., 1995). The IPRL model was chosen for these experiments to eliminate the possible effects of other factors that might affect drug clearance in vivo such as changes in extrahepatic clearance, regional blood flow, and protein binding. We also measured the activity and content of drug glucuronidation enzymes in liver microsomes obtained from rats with RVF. Enzyme latency is described as the degree of restriction of UDP-glucuronosyltransferase(s) (UDP-GT) enzyme activity in their natural phospholipid environment (Zakim and Dannenberg, 1992). Because it has been shown that the latency of UDP-GT may change in certain disease states, resulting in changes in the capacity of these enzymes to metabolize drugs (Desmond et al., 1994), we also examined the latency of hepatic UDP-GT in hepatic microsomes.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals

PNP, p-nitrophenyl-sulfate (potassium salt), sodium taurocholate, bovine serum albumin, uridine diphosphoglucuronic acid (UDP-GA), palmitoyl-lysophosphatidylcholine (PLPC), and tetrabutylammonium hydrogen sulfate were purchased from Sigma Chemical Co. (St Louis, MO). p-Nitrophenyl--D-glucuronide (PNPG) was obtained from Roche Diagnostics (Mannheim, Germany). All other chemicals used were of analytical grade. All solvents were HPLC grade (Mallinckrodt, Phillipsburg, NJ).

Right Heart Failure Model

Thirty-one male Sprague-Dawley rats (age 4-5 weeks, weight 90-120 g) were randomized into two groups [pulmonary artery constriction (PAC) or sham]. Rats in the PAC group (n = 22) underwent constriction of the pulmonary artery via a left-sided thoracotomy as previously described (Ng et al., 1995). Briefly, the pulmonary artery was dissected free of the aorta, a blunt needle (18 gauge, 0.5-cm long) was placed against the pulmonary artery. A sterile silk suture was tied around both the needle and vessel, followed by rapid removal of the needle, resulting in restriction of the pulmonary artery to the size of the needle's diameter. The chest was closed with a silk suture under negative pressure, and antiseptic was applied. Rats in the sham group (n = 9) underwent the same surgery but without the PAC. Thirteen PAC rats died within 16 weeks of the surgery leaving 9 for the perfusion study. All sham-operated rats survived for the duration of the experimental period. There was no significant difference (P < .05) between the initial body weight of the sham (n = 9) (mean weight 104 ± 8 g) and the PAC groups (n = 9) (mean weight 101 ± 9 g).

IPRL Study

Experimental Preparation. 15 to 17 weeks after PAC or sham operation, rats of the two experimental groups were anesthetized with pentobarbitone (60 mg/kg). Before removal of the liver, the left jugular vein was cannulated, and the mean central venous pressure was measured using an electronic pressure monitor unit (model 78205A; Hewlett-Packard, Palo Alto, CA). The common bile duct, the portal vein, and the inferior vena cava were cannulated using standard surgical techniques. The liver was dissected free from the abdominal cavity and weighed before it was connected to the perfusion circuit.

Experimental Design. The liver was perfused at a constant flow rate via the portal vein in a closed cabinet maintained at 37°C. Similar flow rates were used in both groups with an average of 1.5 ml/min/g of liver (see Table 2). The perfusion was initially in a recirculating mode (the equilibrium phase), which lasted for about 20 min to allow the liver to stabilize and its viability to be assessed. Then the experiment was commenced by switching the circuit to a single pass mode. The perfusates in the recirculating and the single pass modes were identical except that PNP was added to the perfusate to give a final drug concentration of 10 µg/ml (71.9 µM). The perfusate consisted of Krebs-Henseleit buffer containing 20% (v/v) washed human red blood cells, 1% (w/v) bovine serum albumin, 0.1% (w/v) glucose, and 30 µM sodium taurocholate, had a pH of 7.2 to 7.4, and was maintained at 37°C. Bile was collected throughout the experiment into preweighed vials in two 30-min aliquots. The viability of the liver preparation was assessed by its macroscopic appearance, oxygen consumption, and perfusion pressure. All perfused liver preparations described in this study were homogeneously perfused, as indicated by an even color of the liver lobes during the perfusion, had oxygen consumption of greater than 3 µmol/min/g of liver, had perfusion pressure of no greater than 12 cm H₂0, and had less than 2 cm H₂0 difference in the perfusion pressure between the beginning and the end of the experiment. Inflow perfusate samples (C_in) were collected at 15, 30, 45, and 60 min. Outflow perfusate samples (C_out) were collected at 5-min intervals from 15 min onward. C_in and C_out samples were centrifuged to remove the red cells, and the supernatant was stored at 20°C until drug concentration was analyzed by HPLC. At the completion of the perfusion experiment, the liver was weighed, and a small section of the liver was stored in buffered formalin for subsequent histological study (S. T. Chou, Department of Anatomical Pathology, Austin and Repatriation Medical Center, Melbourne, Victoria, Australia). Histological analysis was performed in the absence of knowledge of hemodynamic and pharmacokinetic parameters. The remaining liver tissue was snap-frozen in liquid nitrogen, stored at 70°C, and later used to prepare hepatic microsomes (Aito and Vainio, 1976). The heart was removed from the thoracic cavity of the rat, and its right ventricle (including the interventricular septum) was isolated and weighed.

Microsomal Study

Preparation of Hepatic Microsomes. Microsomes were prepared from six livers randomly chosen from the sham group and from the livers of all animals that developed RVF (n = 5, see Table 2). Two grams of each liver was homogenized with a Potter-Elvenjem homogenizer in 20 ml of prechilled buffer (pH 7.25) containing 0.25 M phosphate, 0.15 M KCl, and 1 mM EDTA. The 10% (w/v) homogenate was then centrifuged at 9000g for 30 min at 4°C, and the resultant supernatant was centrifuged at 105,000g for 90 min at 4°C. The microsomal pellet obtained after the second centrifugation was suspended in 2 ml of 0.25 M phosphate buffer containing 30% w/v glycerol, pH 7.25, and stored at 70°C until use. Microsomal protein content was measured by standard Lowry assay (Lowry et al., 1951) using bovine serum albumin as the protein standard.

Measurement of UDP-GT Activity in Native and Activated Microsomes. Microsomal UDP-GT activity was assessed by measuring the amount of PNPG formed from PNP. Total UDP-GT activity (i.e., in the activated state) can only be detected after disruption of that restriction with the addition of detergent or introduction of an excess of phospholipid such as PLPC (Yokota and Yuasa, 1992). UDP-GT enzyme latency was determined by comparing the amount of PNPG formed from PNP in native and activated microsomes, at saturating concentrations of both UDP-GA and PNP, which was a modification of a previously described method (Desmond et al., 1994). The incubation mixture (final volume = 0.25 ml) consisted of PNP (4 mM: 80 µl of 12.5 mM), MgCl₂ (5 mM: 12.5 µl of 100 mM), Tris-HCl (0.1 M, pH 7.4: 10 µl of 2.5 M), microsomal protein (0.1 mg: 20 µl of 5 mg/ml), UDP-GA (10 mM: 50 µl of 50 mM), and deionized and distilled water (77.5 µl). UDP-GA was added after 2-min preincubation of all other components under air at 37°C. The reaction was allowed to run for 10 min before it was stopped by addition of 300 µl of acetonitrile. The concentration of PNPG in the incubation medium at the end of the incubation was measured by a sensitive HPLC method as described under Analytical Assays. The formation rate of PNPG was calculated per minute of incubation and per milligram of microsomal protein present. Before normalizing the activity data to time of incubation and amount of microsomal protein, preliminary experiments were conducted to check the linearity of the reaction with respect to time of incubation and microsomal protein concentration. The results showed that the reaction rate was linear for incubation time up to at least 20 min and protein concentration up to 1 mg/ml. Based on these findings, 10 min of incubation and 0.4 mg/ml microsomal protein concentration were the conditions used in the present study. All incubations were carried out in duplicate.

Measurement of UDP-GA V_max Toward PNP. The incubation mixture was the same as used with the activated microsomes except that the microsomes were spiked with 50 µl of varying concentrations of UDP-GA (1.25, 2.5, 7.5, 15, 25, 50, and 100 mM) to give UDP-GA concentrations of 0.25, 0.5, 1.5, 3, 5, 10, and 20 mM, respectively. The concentration of UDP-GA was varied rather than that of PNP, because PNP is known to be a substrate of multiple UDP-GT isoenzymes, hence complicating the kinetics. By choosing a saturating concentration of PNP (4 mM) and varying the UDP-GA concentration, we were able to detect the total UDP-GT activity in the microsomal preparation. The apparent V_max and K_m toward UDP-GA were determined by fitting a Michaelis-Menten equation to the velocity versus concentration data by nonlinear least-squares regression analysis (Minim 3.0.8; R. D. Purves, University of Otago, New Zealand). Utilization of PNP and UDP-GA were less than 10% at the end of the reaction in each experiment (Segal, 1975).

Analytical Assays

Concentration of PNP (C_in and C_out) and its metabolites PNPG and p-nitrophenyl-sulfate in the supernatants of the IPRL perfusate samples were measured simultaneously by a specific and sensitive HPLC method (Ghabrial et al., 1995). The measured concentration of PNP and its metabolites in the red cell free samples (supernatant) was found to be identical with those obtained from the whole blood samples. In the microsomal experiments, after the reaction was stopped with 300 µl of acetonitrile, the mixture (total volume: 0.55 ml) was vortexed, centrifuged, and processed as follows. 400 µl of the supernatant was evaporated under vacuum for 2 h. The residue was reconstituted in 0.4 ml of mobile phase comprising 15% (v/v) acetonitrile, 10% (v/v) MeOH, 75% (v/v) KH₂PO₄, 0.2% (v/v) triethylamine, and 5 mM tetrabutylammonium hydrogen sulfate at pH 6.0 (adjusted with orthophosphoric acid). Microsomal PNPG concentration was then measured using the HPLC method of Ghabrial et al. (1995) with minor modifications. Elution times for UDP-GA, PNPG, and PNP in the assay were approximately 2.3, 3.7, and 10.4 min, respectively. The coefficient of variation for repeated measures (n = 6) of PNPG was 1% for 5 µg/ml and 2% for 10, 80, and 150 µg/ml. Inaccuracy of the assay for these concentrations was 14, 6, 2, and 3%, respectively. PNPG concentration in the microsomes was determined from a seven-point standard curve constructed from PNPG concentrations of 5, 10, 20, 40, 80, 100, and 150 µg/ml.

Calculations and Statistics

Steady-state extraction (E) of PNP in the IPRL experiments was calculated as

(1)

Hepatic clearance (CL) was calculated as

(2)

where Q is the perfusate flow rate.

Metabolic formation clearance (CLf_m) was calculated as

(3)

where A_m,bile,t1t2 is the amount metabolite in bile (molar units) from time t₁ to time t₂, C_m is the metabolite concentration in the hepatic venous outflow at steady state in µM, and Q is the perfusate flow rate. The times t₁ and t₂ were at steady state (20-60 min).

Data in the tables and graphs are presented as mean ± S.D. Statistical comparison between the sham and the RVF groups in the IPRL and the microsomal studies were performed by the Student unpaired t test. All statistical tests were performed using the StatView SE package (v1.4; Abacus Concepts Inc., Berkeley, CA), and P values of less than .05 were accepted as significant.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Right Heart Failure Model

At 15 weeks, five of the nine rats that underwent the PAC had developed RVF as evidenced by a 10-fold increase in the mean central venous pressure (Table 1) and an engorged appearance of the liver at laparotomy. In these animals, there was evidence of cardiac hypertrophy with a mean increase in right ventricular weight of 57% (Table 1). Although the other four rats that underwent the PAC also developed cardiac hypertrophy (mean right ventricular weight: 1.18 ± 0.13 g), there was no evidence of RVF as indicated by the near normal mean central venous pressure (3 ± 1 mm Hg) and the absence of hepatic congestion. The RVF rats did not have increased lung weight to suggest pulmonary edema due to impairment of left ventricular function (Table 1). There was no significant difference in the mean body weight between the sham and the RVF groups (Table 1).

Evidence of Hepatic Congestion. The livers from the five animals with RVF showed macroscopic evidence of hepatic congestion. When examined under the light microscope, livers from these animals showed sinusoidal dilatation and congestion, but hepatic fibrosis was absent. None of the livers from the sham group showed hepatic congestion under light microscopy. The mean liver weight of the RVF rats was not significantly different from those of the sham group (Table 1).

Elimination by the IPRL

Viability of the IPRLs. Physiological parameters in the isolated perfused livers of the sham and RVF rats are summarized in Table 2. The data illustrate that liver preparations in both groups were perfused at similar flow rates and that oxygen delivery and perfusate pH were similar. All preparations were viable, with oxygen extraction and consumption not being different between livers in the two groups. However, the mean perfusion pressure of the livers from the RVF rats was slightly higher from that in shams (11 versus 9 cm H₂O, P = .047).

Extraction and Hepatic Clearance of PNP in Isolated Perfused Livers. Perfusate outflow concentration of PNP in the perfused livers of the RVF animals was significantly elevated compared with those in sham rats. Extraction and hepatic clearance of PNP in the isolated perfused livers are shown in Table 3. Extraction and hepatic clearance of PNP were reduced by approximately 40% in RVF compared with the sham-operated livers. In keeping with the reduced drug clearance, metabolic formation clearance for PNPG was significantly reduced in RVF (Table 3). There was a trend toward reduced metabolic formation clearance of PNPS in the RVF group, but the reduction was not statistically significant (Table 3).

Microsomal UDP-GT Metabolism and Content

Effect of RVF on the Content of PNP UDP-GT. When the microsomal glucuronidation rate of PNP was plotted against the variable concentration of UDP-GA, a typical Michaelis-Menten relationship was found (Fig. 1). Analysis of this plot in the sham and the RVF groups showed that there was no significant difference the V_max of the reaction between the two groups (Fig. 2. However, the Michaelis-Menten constant of the reaction (K_m) was slightly reduced in the RVF group (P < .05, Fig. 2).

View larger version (15K): [in this window] [in a new window]   Fig. 1.   Mean ± S.E. Lineweaver-Burk plot of the reciprocal of velocity (v) and UDP-GA concentration, illustrating a uni-reaction Michaelis-Menten relationship between the rates of PNPG formation and UDP-GA concentration. () RVF group; () control group. The line represents the linear regression fit of the data to a single active site model.

View larger version (36K): [in this window] [in a new window]   Fig. 2.   Mean values of Vmax and Km of UDP-GT toward varying concentrations (0.25-20 mM) of UDP-GA in native microsomes of sham and RVF livers. The hashed columns represent results in shams, and the black columns represent results in RVF. There is no significant difference in the Vmax between the sham and the RVF. Km was significantly reduced in the RVF (*P < .05).

Effects of RVF on the Activity and Latency of the UDP-GT. The mean glucuronidation rates of PNP in native and activated microsomes are summarized in Fig. 3. There was no significant difference between the sham and the RVF groups in mean UDP-GT activity in both the native and activated microsomes. Thus, the percentage of activation of the UDP-GT activity was not significantly different between the two groups (Fig. 3).

View larger version (28K): [in this window] [in a new window]   Fig. 3.   Glucuronidation of PNP in native and activated microsomes. Activation of microsomes was achieved by the addition of PLPC. Percentage of activation is calculated from the ratio between the activity in the activated state and that in the native state. The hashed columns represent results in the sham, and black columns represent results in the RVF. There was no significant difference (P > .5) between the sham and RVF in enzyme activity in either the native or activated state, nor was there a significant difference in the percentage of activation.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

There has been very little study of the effects of congestive cardiac failure on hepatic drug conjugation. A previous study in patients with congestive heart failure demonstrated impaired clearance of paracetamol (Ochs et al., 1983), a process dependent on hepatic glucuronidation and sulfation. However, the mechanisms responsible for this finding were not elucidated. The present study in the isolated liver demonstrates that, in the presence of normal hepatic perfusion and oxygenation, hepatic elimination of PNP is significantly impaired in animals with RVF. This suggests that in congestive heart failure there is impairment of the intrinsic capacity of the liver to conjugate drugs.

Hepatic elimination of PNP is expected to be flow-dependent, because the metabolic capacity of the liver to eliminate the drug is large (extraction being >70%). Thus, if hepatic perfusion was reduced in heart failure, it might be expected to result in impairment of hepatic clearance of the drug (Wilkinson and Shand, 1975). The fact that hepatic clearance of the drug was significantly impaired in perfused livers from animals with RVF, when perfusate flow rates were the same as those in controls, suggests that there is reduced intrinsic clearance of the drug in RVF and that this is likely to affect hepatic clearance of the drug, independent of the effects of changes in blood flow.

Consistent with previous studies, in the current study elimination of PNP was predominantly via the glucuronidation pathway (Table 3) (Ghabrial et al., 1995). The impaired elimination of PNP in RVF in the present study primarily reflects a reduction in the glucuronidation capacity of the liver, because there was a commensurate reduction in metabolic formation clearance of PNPG.

Drug glucuronidation may be impaired if hepatic oxygenation is reduced in heart failure, because oxygen is consumed for the generation of ATP, which is subsequently utilized for the synthesis of cofactors required for glucuronidation such as the UDP-GA (Angus et al., 1987; Wu et al., 1990; Aw et al., 1991). Another factor that may make glucuronidation reactions sensitive to hypoxia is that UDP-GTs are predominantly localized in the centrilobular zone of the acinus (Knapp et al., 1988), where oxygen concentration is lowest (Matsumura et al., 1986). Indeed, previous studies have shown that hepatic elimination of PNP was impaired by minor reductions in hepatic oxygen supply that are likely to occur in vivo (Aw et al., 1991). The present finding of markedly reduced hepatic clearance of PNP in livers from animals with RVF, in which the levels of hepatic oxygenation delivery and consumption were the same as controls, suggests that impairment of enzyme function by hypoxia is not likely to be the sole cause of reduced drug elimination.

The amount and the activity of UDP-GTs may also influence the rate of glucuronidation of PNP, with the latter also influenced by enzyme latency (Zakim and Dannenberg, 1992). The latency of UDP-GT may be altered if the interaction between the enzyme and the phospholipid is perturbed or disrupted, as occurs with exposure of microsomes to PLPC (Yokota and Yuasa, 1992). The microsomal studies in the present investigation indicated that the latency of UDP-GT is not altered in RVF, because pretreatment of microsomes with PLPC did not alter the percentage of activation of the UDP-GT when compared with the sham group. Thus, in contrast to conditions such as liver cirrhosis where changes in latency of UDP-GTs contribute to the preservation of glucuronidation (Desmond et al., 1994), RVF does not alter the latency of the hepatic UDP-GT enzymes.

There are at least three different forms of UDP-GTs that are responsible for the glucuronidation of PNP (Antoine et al., 1993). The nonavailability of antibodies to the various rat forms of UDP-GTs meant that we were unable to measure their protein content directly. The alternative method available to us was to measure UDP-GT V_max in activated microsomes as a reflection of enzyme content, as previously used by Luquita et al. (1994). Our kinetic studies in vitro with UDP-GA demonstrated that there was no significant difference in the V_max of the reaction between the sham and RVF groups. Hence, the reduced glucuronidation of PNP in the isolated perfused livers in RVF could not be explained by reduction of the hepatic content of UDP-GTs. There was a modest, but statistically significant reduction in the K_m observed in the RVF group. The reduction in K_m and V_max in the RVF did not translate in a significant difference in calculated intrinsic clearance (V_max/K_m, 93.8 ± 28.2 ml/min/mg of microsomal protein in the sham group versus 84.3 ± 10.2 ml/min/mg of microsomal protein in the RVF group). Therefore, change in K_m could not explain the difference in the elimination of PNP observed. Thus, it appears that in heart failure changes in factors other than reductions in enzyme content, activity, or latency of the UDP-GTs are involved in impaired glucuronidation.

There has been a number of studies demonstrating that conjugation reactions, including the conjugation of PNP, are sensitive to the level of carbohydrate reserves and cofactor supply (Reinke et al., 1979; Thurman et al., 1981; Qu et al., 1995). Thus, it is possible that the reduced clearance of PNP in RVF in the current study is due to depletion of cofactors within the liver. This may explain why PNP glucuronidation was preserved in microsomes in comparison to the IPRL, because in microsomes cofactors such as UDP-GA are supplemented in excess, whereas no cofactors were supplied to the IPRL.

Histological examination of the perfused livers from the RVF animals showed that there was sinusoidal dilatation and congestion, which was not relieved (hepatic outflow open to atmosphere) after the livers had been removed from the RVF animals and perfused for 80 min. It is possible that this sinusoidal congestion in RVF livers leads to disturbed microcirculation and impaired access of drugs to functional hepatocytes, as has been suggested to occur in cirrhosis (McLean and Morgan, 1991; Gariepy et al., 1993). The higher portal perfusion pressure in the RVF perfused livers than in controls supported the presence of a disturbed microcirculation in RVF livers. Factors such as changes in local oxygen delivery, flow distribution, and shunting in RVF livers might explain in part the decreased clearance of PNP in the IPRL. A previous electron microscopic study demonstrated that chronic hepatic congestion leads to deposition of collagen in the space of Disse and development of a sinusoidal basement membrane (Safran and Schaffner, 1967). It is possible that similar changes occur in RVF and affect hepatic drug uptake. In addition, it has been suggested that congestion of the liver can result in perisinusoidal edema (Brauer et al., 1959; Greenway and Lautt, 1970), which may also lead to impaired drug uptake (Dunn et al., 1973).

In conclusion, the current study shows that right heart failure impairs the ability of the liver to conjugate drugs and that this cannot be totally explained by reduced hepatic oxygen delivery, reductions in hepatic content or activity of UDP-GT, or changes in the latency of the UDP-GT. Our findings provide further evidence that the elimination of a wide range of drugs may be impaired in heart failure increasing the risk of drug accumulation and toxicity.