Studies with Induced Livers Involving Diazepam
Abstract
Microsomal protein recovery and hepatocellularity have been determined and investigated as scaling factors for interrelating clearance by hepatic microsomes, freshly isolated hepatocytes and whole liver from untreated (UT) rats and rats treated with either the cytochrome P450 inducer phenobarbital (PB) or dexamethasone (DEX). Hepatocellularity in UT rats (1.1 × 108 hepatocytes/g liver) was not significantly different after either PB or DEX induction (1.1 and 1.3 × 108 hepatocytes/g liver, respectively). However the microsomal protein recovery index, which provides a scaling factor that is inversely related to the efficiency of the microsomal preparation procedure, was 47 mg/g liver in both PB and DEX microsomes and differs from UT rats (60 mg/g liver). These contrasting findings are consistent with the interlaboratory trends in the literature, indicating that, although hepatocellularity estimates are in good accord, microsomal recovery can vary 2-fold; this has implications for scaling.
The oxidation of diazepam to its three primary metabolites was measured in PB and DEX microsomes and hepatocytes and the scaling factors were applied to these data and previously reported UT data. Marked changes in kinetics occur on induction resulting in a shift in the major pathway. In particular, 3-hydroxylation is induced over 20-fold by DEX. Diazepam CLint was determined in vivo after administration of a bolus dose into the hepatic portal vein of UT, PB, and DEX rats; values of 127, 191, and 323 ml/min/SRW (where SRW is a standard rat weight of 250 g), respectively, were obtained. Using these scaling factors, the hepatocyte predictions ofCLint were excellent (99, 144, and 297 ml/min/SRW for UT, PB, and DEX, respectively), whereas only the DEX prediction (248 ml/min/SRW) was accurate for the microsomal system, with a substantial underprediction for UT and PB (46 and 68 ml/min/SRW, respectively). Evidence is presented for product inhibition, resulting from accumulation of primary metabolites within the microsomal preparation, as the mechanism responsible for this underprediction.
These results illustrate that the scaling factor approach is applicable to induced livers in which both cytochrome P450 complement and zonal distribution are altered. These data, together with our previous studies, demonstrate that CLint in cells (2.4–297 ml/min/SRW), microsomes (2.7–248 ml/min/SRW), and in vivo (1.5–323 ml/min/SRW) are related in a linear fashion and hence inherently both in vitro systems are of equal value in predicting in vivo CLint.
The use of scaling factors in drug metabolism provides an approach to compare directly kinetic data from two or more in vitrosystems and to predict in vivo behavior from in vitro systems. These factors should not be purely empirical in nature, rather they should originate from the anatomical and/or biochemical characteristics of the in vitro system. For example, hepatic microsomal rates of metabolism are usually expressed per milligram of microsomal protein and may be scaled by recognizing the inefficiency of the microsomal isolation procedure by using a microsomal recovery factor. This recovery factor enables the conversion of rates to units of liver weight and hence to the intact animal. Similarly, hepatocyte suspension data, expressed per million cells, require a hepatocellularity number to render such data comparable with other in vitro systems or predictive of the in vivo situation.
In assessing the utility of scaling factors using a data base of 25 drugs (1), we found excellent in vivo prediction of metabolic clearance from both microsomes and hepatocytes for low clearance drugs. However, whereas the prediction for high clearance drugs from hepatocytes proved to be equally valuable, underestimation of in vivo high clearance from microsomal data was frequently observed. Further comparison of the relationship between microsomal and hepatocellular clearance suggested that a maximum limit for microsomal clearance may exist. However, because the data used in this later analysis originated from a variety of laboratories, no definitive conclusions could be drawn. In contrast, the main analysis clearly demonstrated the value of microsomal protein recovery index and hepatocellularity as microsomal and hepatocyte scaling factors, respectively.
To examine further the scaling factor approach, we have now investigated the scaling of data from hepatic microsomes and hepatocytes from PB2- and DEX-treated Sprague-Dawley rats, together with analogous data from UT animals. Treatment with CYP inducers such as PB and DEX produces a number of effects in the liver, including the induction of members of the CYP2B and CYP3A subfamilies (2, 3). In addition to altering the CYP complement, both PB and the DEX-type inducer pregnenolone-16α-carbonitrile (4) are known to change the zonal CYP distribution (5). To assess the consequences of induction, DZ was used as the substrate because the formation of its three primary metabolites—4′-HDZ, 3-HDZ, and NDZ—is substantially, yet selectively, affected by PB and DEX treatment (6). Several studies have indicated that 3-HDZ formation is mediated by the CYP3A subfamily (6-8), which is inducible by DEX (3) and to a lesser extent by PB (2). There are conflicting reports (6-8) regarding which CYP isoforms are responsible for NDZ formation; however, CYP2B, CYP2C, and CYP3A have been implicated. Also, Neville and coworkers (7) have demonstrated that 4′-HDZ formation is principally mediated by CYP2D1, an isoform believed to be noninducible by either PB or DEX.
The aims of the present studies are therefore 2-fold: 1) to extend the scaling factor approach to livers of varying CYP complement and zonal distribution; and 2) to assess whether there is a progressive increase in both microsomal and hepatocellular clearance of DZ in the induced state, which is predictive of the in vivosituation. In the UT rat the clearance of DZ is high: 174 ml/min/SRW (9). Thus, the use of induced systems extends substantially the range of clearances previously investigated in the authors’ laboratory forin vitro:in vivo comparisons.
Materials and Methods
Chemicals.
3-HDZ and prazepam were generously supplied by Wyeth (Maidenhead, Berks, UK), 4′-HDZ, and 3-HNDZ were gifts from Roche Ltd. (Welwyn Garden City, Herts, UK) and the Warner Co. (Pontypool, Gwent, UK), respectively. NDZ and DZ were purchased from Sigma (Poole, Dorset, UK). All other chemicals were obtained from either BDH (Lutterworth, Leicester, UK) or Sigma.
Animals and Treatment.
Male Sprague-Dawley rats (220–250 g) were obtained from the Biological Sciences Unit at the University of Manchester. They were housed 2–4 per cage on sawdust bedding in rooms maintained at a temperature of 20 ± 2°C and allowed free access to water and Chow Rat Mouse rat diet. Rats were either UT or administered intraperitoneal injections of PB (80 mg/kg in 0.9% saline) or DEX (100 mg/kg in 2% Tween 80) once daily for 3 days, with death on the fourth day. In each case, the animals were divided into three groups for either microsomal (N = 4 for each inducer), hepatocyte (N= 4 for each inducer), or in vivo (N = 5 for PB/DEX). To complement the latter case, an additional group of control rats received corresponding injections of 2% Tween 80 (N = 3).
Microsomal Studies.
PB- and DEX-treated rats were killed by cervical dislocation, and the livers were excised and washed in ice-cold potassium chloride (0.15 M). Livers were subsequently homogenized in ice-cold buffer (0.25 M sucrose, 1 mM EDTA, and 25 mM Trizma base; pH 7.4) to yield a liver homogenate tissue concentration of 0.33 g/ml. Washed microsomes were prepared as described previously (10). Liver homogenate and microsomal CYP and protein contents were determined using the methods of Omura and Sato (11) and Lowry et al. (12), respectively. Liver homogenate DNA content was determined using the method of Downs and Wilfinger (13).
For the determination of Michaelis-Menten parameters, microsomal incubation conditions with DZ were as detailed previously (9), ensuring linearity between rates of formation and time and microsomal protein concentration. The effects of the primary metabolites as product inhibitors of DZ metabolism were investigated by the same methodology at a DZ concentration of 2 μM and an exogenous metabolite concentration (either 4′-HDZ, 3-HDZ, or NDZ) of 100 μM. Total DZ metabolism was assessed by substrate depletion after an incubation time of 15 min.
Hepatocyte Studies.
Hepatocytes were prepared as described elsewhere (10) from PB- and DEX-treated rats. Cell protein and CYP content were estimated as described for microsomes. The DNA content of cells was determined after freshly isolated hepatocytes had been disrupted by the addition of a detergent buffer as detailed by Dunn and coworkers (14). Cell protein content was determined after the digestion of cells in 0.5 M sodium hydroxide; cell CYP content was estimated in cells disrupted by homogenization. Cell numbers in all drug incubations refer to hepatocytes only. The incubation conditions to achieve metabolism and hydrolyze conjugates before analysis were as described previously (9), ensuring linearity of rates of metabolism with respect to time and cell density.
Analysis of DZ Metabolites.
The method of Reilly et al. (6) was used to assay 4′-HDZ, 3-HDZ, and NDZ in the microsomal and hepatocyte incubates as described previously (9).
In Vivo Studies.
Polyethylene cannulae were inserted into the lineal vein (15) and right carotid artery (16) under halothane anesthesia in DEX-treated (N = 5), PB-treated (N = 5), and control (N = 3) rats after the third daily injection of either DEX, PB, or vehicle alone. Animals were subsequently housed one per cage and allowed 1 day to recover from this procedure. DZ [10 mg/kg in propane-1,2-diol, ethanol, and 0.9% saline (40%:10%:50%, v/v/v)] was administered (2 ml/kg) into the hepatic portal veinvia the lineal vein as a 5-min infusion. Blood samples (N = 9) were taken from the carotid artery over a 90-min period, and plasma separated by centrifugation and analyzed for DZ as detailed elsewhere (9).
Data Analysis.
The Michaelis-Menten equation was used to determine theVmax and KM values for each pathway of DZ metabolism using the Siphar nonlinear regression software (Simed, Cretéil, France). Goodness of fit was judged by visual inspection of the line of best fit, consideration of the randomness of the residuals, the standard error of the parameter estimates (<20%), and the Aikaike information criteria. Because the substrate loss in both the microsomal and hepatocyte incubations calculated from the initial rate of metabolite formation was substantial, the logarithmic mean of the initial and final substrate concentrations was used as the independent variable in the curve-fitting procedure.
Hepatocellularity was calculated by three methods based on the ratios of liver/cell protein, DNA, or CYP. The liver comprises ∼65% hepatocytes, which occupy 80% of the cellular volume (5). However, the hepatocyte isolation procedure, involving filtration and low-speed centrifugation, yields a cell suspension of which hepatocytes comprise ∼90% (17, 18). Steinberg et al. (18) demonstrated that hepatocytes contain 97% of the total protein in liver tissue and, that in an isolated cell suspension, 99% of the protein is contained in hepatocytes. Therefore, strictly speaking, the ratio of liver to cell protein should be corrected by these factors (0.97/0.99;i.e. 0.98); however, this was considered unnecessary. Similarly, the calculation of hepatocellularity by CYP content was also based directly on the liver/cell CYP ratio. The use of DNA to calculate hepatocellularity requires consideration not only of the percentage of hepatocytes in both the isolated cell suspension and the whole liver, but also the binuclearity of ∼20% of hepatocytes (17). The percentage of hepatocellular DNA in the whole liver is 69%, whereas the corresponding value in the isolated cell suspension is 92%, therefore the liver/cell DNA ratio was corrected by 0.75 (0.69/0.92).
Microsomal protein recovery ratio (mg protein/g liver) was estimated from the ratio of liver homogenate CYP content (nmol/g liver) to microsomal CYP content (nmol/mg protein). No correction for nuclear or other nonmicrosomal CYP was made because it has been demonstrated that microsomal CYP accounts for >95% of hepatocyte CYP content, even following PB treatment (19, 20).
CLint was calculated from the Vmax/ KM ratio and scaled to in vivo by multiplying by either the microsomal protein recovery ratio or hepatocellularity as the scaling factor. Scaling factors are reported in the text and refer to a SRW of 250 g, with a liver weight of 11 g. Extension of the use of scaling factors to induced livers involves some particular considerations due to the morphological changes known to occur. It is well documented that PB administration to rats causes an increase in liver weight as a result of smooth endoplasmic reticulum proliferation (21). There is also an increase in nuclear DNA, although this is due to increases in ploidy rather than hepatocyte number (21). Therefore, for scaling purposes in the induced state, it seems reasonable to assume that the total number of cells per liver remains unaltered and thus an SRW of 250 g should apply.
In vivo estimates of CLint were obtained from the clearance term after administration viathe intraportal route. A biexponential zero-order input model was used to fit the in vivo plasma concentration data (eq. 1) and the pharmacokinetic parameters MRT, CL , and VSS were calculated from eqs. 2-4, respectively:
CLint was calculated from CL , B/P, and fu , as shown in eq. 5.
Results
In Vitro Hepatocyte Studies.
The production of all three primary DZ metabolites demonstrated monophasic Michaelis-Menten kinetics in PB and DEX hepatocytes as shown in fig. 1. In both induced states, the 3-HDZ pathway dominates in contrast to UT (table1). Substantial formation of NDZ and 4′-HDZ also occurred; however, the secondary metabolite 3-HNDZ was not observed during the 10-min incubation.
PB treatment caused a significant increase in the Vmax values for both the 3-HDZ and NDZ pathways, whereas the corresponding value for the 4′-HDZ pathway was unchanged, compared with UT. There was a concomitant increase in CLint for the former two pathways, whereas the CLint for the 4′-HDZ pathway was reduced as a result of an increase in the KM (table 1). The CLint of the 3-HDZ pathway increased ∼9-fold in DEX hepatocytes as a consequence of a 3-fold change in both Vmax and KM. The NDZ pathway CLint also increased after treatment with this inducer in a comparable manner to PB treatment, whereas 4′-HDZ formation was unaffected by DEX treatment. Figure2 illustrates the qualitative changes in the relative importance of the three pathways that occur on PB and DEX induction. There is a large increase in the contribution of the 3-HDZ pathway to CLint, whereas the 4′-HDZ pathway contribution is reduced and the NDZ pathway is relatively unaffected.
In Vitro Microsomal Studies.
Rates of primary metabolite formation in both PB and DEX microsomes also demonstrated single-site Michaelis-Menten kinetics (fig. 1). Again, only the three primary metabolites were formed in the time course of the study (5 min).
The net effect of these inducers, as shown in table 1, was an increase in the importance of both the 3-HDZ and NDZ routes of DZ metabolism, the largest change occurring for the former pathway. Neither CYP inducer had any effect on the 4′-HDZ pathway. PB microsomes exhibited a 3-fold increase in both the 3-HDZ and NDZ Vmaxvalues, thereby significantly increasing the CLint of these pathways, compared with the UT data. DEX treatment also resulted in increases in the Vmax for 3-HDZ and NDZ in addition to significant decreases in the KM values of both of these pathways, with the major change being a 21-fold increase in the 3-HDZ CLint.
Qualitatively, the effects of induction were similar with respect to each pathway in both in vitro systems (fig. 2). There were no significant differences in KM values between microsomes and hepatocytes for the 4′-HDZ and NDZ pathways, whereas the 3-HDZ pathway KM was significantly lower in hepatocytes of each treatment state. The rank order of KM values was consistently 4′-HDZ < NDZ < 3-HDZ (table 1).
Determination of Hepatocyte and Microsomal Scaling Factors.
To obtain appropriate hepatocyte scaling factors, hepatocellularities based on both the DNA and protein contents of liver homogenate and isolated hepatocytes were determined in UT, PB, and DEX rats (table2). Neither inducer had any significant effect on either the protein or DNA content of liver and isolated cells, although the hepatocellularities determined using the protein methodology were significantly higher than the corresponding DNA determined values. In contrast, the microsomal scaling factor, based on microsomal protein recovery, was altered after PB and DEX treatments (table 3) and was significantly reduced in both populations of induced microsomes compared with UT.
Comparison of Predicted CLint from Hepatocyte and Microsomal Studies.
The in vitro PB and DEX data obtained in the present study and the UT data reported previously (9) were scaled using the above scaling factors to obtain estimates of CLint (as shown in table 4). These results confirm that 3-HDZ formation is the major determinant of DZ disposition in bothin vitro systems after either PB or DEX treatment, whereas 4′-HDZ production is the major route of DZ CLintin UT microsomes and hepatocytes. The hepatocyte/microsomal CLint ratio (table 4) demonstrates that only in the case of NDZ are the ratios consistently close to unity. In all situations, the 4′-HDZ ratio exceeds unity (range: 2.1–3.4), indicating that hepatocytes have greater intrinsic ability to generate this metabolite than microsomes.
In Vivo Studies.
Figure 3 shows plasma concentration-time profiles of DZ in typical DEX, PB, and control rats where the declines in DZ concentration are biexponential in agreement with a number of other studies (22, 23). The dose used was within the region of linear kinetics for DZ (9). A comparison of the control data (N = 3) with the corresponding UT data determined previously in this laboratory (N = 6) (9) showed that the DEX vehicle, 2% Tween 80, had no statistical effect on the terminal half-life, VSS, or plasma clearance; therefore, the parameter values were combined and averaged. Volume of distribution remained unaltered in the induced states when compared with UT rats, as did half-life, whereas plasma clearance and MRT were altered significantly (p < 0.05) in DEX rats as a result of enzyme induction (table5).
Intraportal administration assumes the well-stirred liver model to estimate CLint (24), and using the B/P of DZ of unity (9) and the fu of DZ of 0.22 (9), the CLint of this drug averages 323 and 191 ml/min/SRW in DEX and PB rats, compared with 127 ml/min/SRW in UT rats. Substantial variability is seen between individual rat values (fig.4A) and, although statistical significance is reached between the UT and the DEX groups, the PB data overlap with both other groups.
Prediction of In Vivo CLint from In Vitro Systems.
The sum of the individual CLint values for each pathway provides an estimate of total CLintin vivo (table 6). Neitherin vitro system predicts the consequences of PB induction to be pronounced as the scaled CLint values are <1.5-fold larger than the scaled UT values; however, DEX predictions indicate a 3-fold change in CLint (fig.4B). In the in vivo studies, changes of this magnitude are seen in the mean of the observed CLint values (table 5). The hepatocyte predictions of CLint in UT, PB, and DEX rats are in excellent accord with that observed in vivo, representing 75–92% of CLint, whereas only the DEX microsomal prediction is accurate (77% of the observed value). For both the UT and PB microsomes, the predictions underestimate thein vivo clearance (by 64% and 57%, respectively).
Inhibition of DZ Metabolism by Its Primary Metabolites.
By measuring the extent of DZ depletion in the presence of each of the exogenously added primary metabolites, the effect of product inhibition was investigated (table 7). Inhibition by 4′-HDZ is evident in both UT and induced microsomal preparations. In UT and PB microsomes, 3-HDZ and NDZ also inhibit, but to a lesser degree than 4′-HDZ. For each metabolite, the effect is more pronounced in PB than UT microsomes. In DEX microsomes, however, the effect of each metabolite differs; where 4′-HDZ inhibits, both 3-HDZ and NDZ promote DZ metabolism.
Discussion
The use of induced livers from PB- and DEX-treated rats allows the application of scaling factors to be examined in livers with differing CYP complements and zonal distribution. Thus, the general applicability of this approach to compare in vitro systems and to predictin vivo values of clearance may be challenged. In addition, the use of induced systems allows examination of conditions where particularly high clearance values would be anticipated.
Rat liver hepatocellularity has been estimated using DNA, CYP, and protein methodologies. The DNA contents measured (table 2) compare favorably with the values reported for liver of 2.2–2.5 mg/g liver (13, 17) and of 17 μg/106 cells reported for isolated cells (25). The resulting hepatocellularity of 85 × 106 hepatocytes/g liver in UT rats is somewhat lower than the value reported by Seglen (17) of 128 × 106hepatocytes/g liver who also used data obtained on DNA content to calculate hepatocellularity. This discrepancy probably reflects the polyploid status of the cells, because there are known variations in cellularity and binuclearity of rats of differing ages (26). We have attempted to correct for this complication as detailed inMaterials and Methods. The CYP content of UT isolated cells was determined as 0.31 ± 0.09 nmol/106 cells (N = 4), because rat liver contains 50 nmol of CYP/g liver (table 3); the resulting hepatocellularity (160 × 106 hepatocytes/g liver) was somewhat higher than the DNA determined values. However, inaccuracies in the spectrophotometric determination of samples that are both turbid and of low CYP content compromises this approach to calculate hepatocellularity. The protein methodology yielded a UT hepatocellularity of 109 × 106 hepatocytes/g liver that is more comparable with values reported in the literature [100 × 106 cells/g (25), 95 × 106 cells/g (27), 128 × 106cells/g (17), and 120 × 106 cells/g (28)] using a variety of methods. An advantage of the protein methodology is that no assumptions about hepatocyte characteristics like polyploidy are made and no analytical complications exist; therefore, hepatocellularities estimated by this method were adopted for future scaling of the hepatocyte data to in vivo units.
Microsomal protein recovery index—determined in UT, PB, and DEX microsomes—provides a scaling factor that is inversely related to the efficiency of the microsomal preparation procedure. The use of spectrophotometry to quantitate the CYP content of liver homogenate and microsomal suspensions is more applicable for these tissue samples because they contain appreciably more of the enzyme than hepatocyte suspensions for the same degree of turbidity. PB treatment caused a significant increase in the CYP content of both liver and hepatic microsomes, whereas DEX treatment had no significant effect on overall CYP levels. Because DEX caused a large increase in the rate of DZ metabolism, it is clear that some CYP isoforms were induced; therefore, the levels of other CYP isoforms must have declined. The recovery obtained for UT microsomes (60 mg/g) liver is on the upper limit of the 2-fold range previously reported (25–54 mg/g) (29-32). The scaling factors for PB and DEX of 46.6 and 46.9 mg/g liver, respectively, are lower than the corresponding UT value. Thus, we have observed similar differences within our laboratory for UT and induced livers to that seen between laboratories for the UT case. It has been reported that PB administration to rats causes an increase in hepatocyte size of almost 20% (33). This change may render the cells more susceptible to mechanical disruption, thus improving CYP recovery. These differences in microsomal recovery contrast with the invariant nature of hepatocellularity and highlight a practical issue of scaling microsomal data: the need for individualized scaling factors.
DZ was selected for the present investigation due to its CYP-mediated metabolism to the three primary products 4′-HDZ, 3-HDZ, and NDZ. We have observed marked changes in both hepatic microsomes and freshly isolated hepatocytes from PB and DEX rats, compared with UT that is consistent with the known involvement of several CYP isoforms. The role of the CYP3A subfamily in 3-HDZ formation is evident in the large increase in CLint for this pathway after DEX treatment in both in vitro systems. Interestingly, DEX treatment resulted in a 50% reduction of the 3-HDZ pathway KM in both microsomes and hepatocytes, whereas this parameter remained unaltered after PB treatment. In our study, both inducer treatments resulted in an increase in the NDZ CLint in both in vitro systems, indicating an important role for the CYP2B and CYP3A subfamilies in this pathway. The lack of any induction of the 4′-aromatic hydroxylation of DZ in the present study supports the claim that this pathway is catalyzed by CYP2D1. In the microsomal preparations there is no substantial difference between either of the kinetic parameters for PB and DEX, compared with UT (table 1). A similar effect in hepatocytes for DEX is observed, but for PB, there is a statistically significant decrease in CLint due to an increase in KM. We have no explanation for this observation, but it should be remembered that KM is an apparent value representing the ratio of several rate constants.
Although the consequences of induction on DZ clearance in vivo are quantitatively well predicted from both hepatocytes and microsomes (92% and 77% accuracy, respectively) in the DEX case, only hepatocytes accurately (75%) predict the PB case. This clearly demonstrates the value of the hepatocyte scaling factor determined after the induction treatments and confirms that freshly isolated hepatocytes consistently provide reliable estimates of in vivo CLint. To date, we have demonstrated successful hepatocyte predictions over a range of 2 orders of magnitude, from 2.4 ml/min/SRW for tolbutamide (35) to the value obtained in the present DEX study of 297 ml/min/SRW and including 16 other drugs (1).
The underprediction from PB microsomes is similar to that previously observed for UT microsomes (fig. 4), and we have speculated that this is a consequence of endproduct inhibition (9)—a phenomenon demonstrated for DZ in isolated perfused liver preparations (34). In the present study, we have examined this proposal further and have demonstrated that whereas all three primary metabolites are capable of inhibiting DZ metabolism, 4′-HDZ is the more effective. In UT hepatocytes the 4′-HDZ pathway dominates, representing 70% of CLint, whereas in microsomes the corresponding contribution is only 49%. This effect can be seen in the hepatocyte/microsomal CLint ratio of 2.6 in table 4. PB microsomes also underpredict in vivo clearance and a similar inhibitory effect may occur. Although the rates of formation of the three metabolites are more equally balanced in PB than in UT microsomes, the metabolite inhibitory effects are greater in the former case. In DEX microsomes, 4′-HDZ is also inhibitory, and the hepatocyte/microsomal CLint ratio is 3.4 for the 4′-HDZ pathway. However, this pathway is of minor importance after DEX induction (fig. 2) and hence the phenomenon is of little importance. Interestingly, 3-HDZ and, in particular, NDZ stimulate DZ metabolism. The latter metabolite is known to undergo CYP3A-catalyzed oxidation (6) and there are several examples of substrate activation for this particular CYP isoform (36, 37). However, 3-HDZ, the major metabolite accounting for some 80% of CLint in DEX rats, has only a minor effect in these microsomes. Thus, the modulating effects of these metabolites on predicting the overall DZ clearance would be manifest only for the UT and PB microsomes. The presence of fully functional phase 2 enzymes in the intact hepatocyte would minimize the accumulation of the CYP-formed primary metabolites; thus product inhibition would not be anticipated in this in vitrosystem (9).
The successful prediction from DEX microsomes for DZ is of particular interest. The most accurate microsomal predictions of CLint have previously been confined to compounds with a low CLint such as caffeine (2.7 ml/min/SRW) (10) and tolbutamide (3.1 ml/min/SRW) (35). The predictions for many drugs with a high CLint has been poor (1), and this has led to speculation that an upper limit may exist for microsomal turnover. However the results of the present study unequivocally demonstrate that this widely used in vitrosystem can accurately estimate high in vivo CLint values, because the DEX prediction was 77% of the actual value. The suggestion of a nonlinear relationship between microsomal CL and hepatocyte CL is refuted by the present analysis and thus the reason for previous underpredictions may be isoform/pathway-dependent, rather than a result of the existence of an upper limit of metabolism rates in microsomes.
The results presented herein enable a detailed comparison of the utility of both microsomal and hepatocyte predictions of CLint. These data, together with data previously reported (1, 9, 10, 35, 38), total some 12 CYP substrates with ethoxycoumarin and DZ being studied in various CYP induction states. A number of these compounds are metabolized via two or more primary routes, and thus there is corresponding microsomal and hepatocyte data concerning some 26 pathways. Figure 5 presents the CLint data, scaled to in vivo units using the appropriate scaling factors (tables 2 and 3), withdotted lines representing the line of identity, a 50% underprediction of CLint, and a 100% overprediction of CLint. The data range over 4 orders of magnitude, and the level of similarity in CLint predictions between the two in vitro systems is impressive, given the varied sources of data and the use of CYP inducers that have marked effects on the liver. There are only five pathways that exhibit a marked underprediction in microsomes, compared with hepatocytes, and three of these refer to DZ 4′-hydroxylation in UT, PB, and DEX states, probably due to endproduct inhibition. Therefore, inherently, both microsomes and hepatocytes can be regarded as equally valuable tools to predict in vivopharmacokinetic behavior, because the majority of compounds exhibit similar scaled CLint values in both systems.
The present study exemplifies how both hepatic microsomes and freshly isolated hepatocytes can be used to predict in vivo CLint when easily determined scaling factors are used. The UT microsomal scaling factor determined in this study (660 mg/SRW) is higher than the value of 500 mg/SRW proposed earlier (1), based on literature reports. However, this value is on the upper limit of the 2-fold range of microsomal recovery reported of 275–600 mg/SRW (1). This range in reported values is not surprising, because this parameter is a direct measure of the efficiency of the microsomal preparation procedure that will vary between laboratories according to the details of the methods used. In addition, the estimated hepatocellularity (1.2 × 109/SRW) is slightly lower than the corresponding value of 1.5 × 109/SRW proposed earlier (1), although once again it is within the range of reported hepatocellularities (1.1–1.5 × 109 cells/SRW). The comparatively small range (30%) for this latter scaling factor indicates that, for successful scaling of hepatocyte data, it is not necessary for individual laboratories to determine their own scaling factor, because hepatocellularity is not an experimental variable. In contrast, microsomal protein recovery seems to be laboratory-dependent; therefore, individual laboratories should determine their own scaling factor to maximize the predictive utility of their microsomal data.
Acknowledgments
We gratefully acknowledge the skilled assistance of Beverley Brennan in a portion of this work.
Footnotes
-
Send reprint requests to: Dr. J. Brian Houston, School of Pharmacy and Pharmaceutical Sciences, University of Manchester, Oxford Road, Manchester M13 9PL, UK.
-
↵1 Present address: Department of Anesthesia, Stanford University, Stanford, CA.
-
This study was supported by the Biotechnology and Biological Sciences Research Council and the Wellcome Research Laboratories. A portion of this study was presented at the meeting of the British Pharmacological Society, April 5–7, 1995, Canterbury, UK, and appeared in abstract form in Br. J. Clin. Pharmacol. 40,181P (1995).
- Abbreviations used are::
- PB
- phenobarbital
- DEX
- dexamethasone
- UT
- untreated
- CYP
- cytochrome P450
- DZ
- diazepam
- 4′-HDZ
- 4′-hydroxydiazepam
- 3-HDZ
- 3-hydroxydiazepam
- NDZ
- nordiazepam
- SRW
- standard rat weight
- 3-HNDZ
- 3-hydroxynordiazepam
- CLint
- intrinsic clearance
- MRT
- mean residence time
- CL
- clearance
- VSS
- volume of distribution at steady-state
- CLIP
- intraportal clearance
- B/P
- blood:plasma ratio
- fu
- fraction of unbound drug concentration
- Received May 22, 1996.
- Accepted April 30, 1997.
- The American Society for Pharmacology and Experimental Therapeutics