Prediction of Human Drug Clearance

For those compounds predominantly cleared by metabolism, human blood clearance can be pre dicted using simple enzyme kinetic data (Houston 1994; lwatsubo etal, 1996 Obach 1996a; Ashforth et al., 1995). These predictions may be strengthened by comparing preclinical in vivo data with the predictions made from in vitro data using tissues from the same preclinical species. Rane et al 1977 As an illustration, consider a novel compound currently in development, identified as Compound X. This compound has a molecular weight less than 400, and a log D7,4 value of approximately 0.5, suggesting that it could undergo both renal and hepatic clearance. Preclinical in vivo studies indicate that Compound X is eliminated largely unchanged in the urine in the rat 90%). Several oxidative biotransformation pathways have nonetheless been identified. In common with studies of Compound X clearance in humans, simple in vitro enzyme kinetic studies were used in conjunction with knowledge from rat in vivo data. The general strategy for prediction of kinetic studies, is shown in Figure 10.2.

Using liver microsomes from different species, the intrinsic clearance (Cl(nt) for each species can be determined, and then scaled to hepatic clearance. This is typically done by first determining in vitro Km (the Michaelis-Menten constant) and Vmax (the maximal rate of metabolism) for each metabolic reaction, using substrate saturation plots (using the familiar algebra and, because of enzyme saturation, finding that Cl(nt = Vmax/Km). However, for Compound X, the situation is more complicated because we know that the Cl(nt (drug disappearance) actually is due to several combined biotransformation pathways (i.e. Cl(nt (total) = Cljnt j + Cl(nt 2 + Cl[nt 3 + ...), thus complicating any Km and Vmax determinations from a simple substrate saturation plot.

To determine the Cl'int of compound X, we are able to use the in vitro half-life method, which is simpler than finding all the component Cl(nt values. When the substrate concentration is much smaller

Km, vmax for metabolic pathway

In vitro CL


In vivo CL

Blood flow t1/2 for drug loss

Serum protein binding Microsomal protein binding

Hepatic clearance

Figure 10.2 Strategy for the in vitro-in vivo scaling of hepatic clearance (see for example Iwatsubo et al., 1996)

or than the Km, the Michaelis-Menten equation simplifies from velocity ( V ) = Vmax([S])/(Km + [S ])to V = Vmax/Km, because ([S] is substrate concentration) becomes negligible. Furthermore, under these conditions the in vitro half-life (T\/2 = 0.693/Kel) can be measured, and this, in turn, is related to the Michaelis-Menten equation through the relationship velocity (V) = volume x Kei (where volume is standardized for the volume containing \ mg of microsomal protein). When both V and Vmax are known, then the Km is also found. Although simpler than finding a complicated Cint, one caveat of the in vitro half-life method is that one assumes that the substrate concentration is much smaller than the Km. It may be necessary to repeat the half-life determinations at several substrate concentrations, and even model the asymptote of this relationship, because very low substrate concentrations that are beneath biochemical detection may be needed to fulfill the assumptions needed to simplify the Michaelis-Menten equation.

Note that in this in vitro application, intrinsic clearance, like all conventional mathematical evaluation of clearances, has units of volume • time~\. It is obtained from Vmax and Km measurements, where Vmax has units of mass • time~\. The definition of intrinsic clearance as Vmax • K^ should not be confused with the historically prevalent calculation of kel (the first-order rate constant of decay of concentration in plasma), calculated from kel = Vmax • Km\, where Vmax is the zero order rate of plasma concentration decay seen at high concentrations, and Kmax is the concentration is plasma at half-maximal rate of plasma level decay.

Once the in vitro intrinsic clearance has been determined, the next step, scaling in vitro intrinsic clearance to the whole liver, proceeds as follows:

In vivo Cl'nt — in vitro Cl[nt x weight microsomal protein/g liver x weight liver/kg body weight

The amount of microsomal protein/g liver is constant across species (45mg/g liver). Thus, the only species-dependent variable is the weight of liver tissue/kg body weight.

In vivo, hepatic clearance is determined by factoring in the hepatic blood flow (0, the fraction of drug unbound in the blood (fu), and the fraction of drug unbound in the microsomal incubations (/u(inc)), against the intrinsic clearance of the drug by the whole liver (the in vivo Cl[ni). The fu and /u(inc) are included when the drug shows considerable plasma or microsomal protein binding (Obach 1996b). Several models are available for scaling in vivo intrinsic clearance to hepatic clearance, including the parallel tube model or sinusoidal perfusion model, the well-stirred model or venous equilibration model, and the distributed sinusoidal perfusion model (Wilkinson 1987).

Thus far, for Compound X, we have obtained good results in this context with the simplest of these, the well-stirred model (see Table 10.1 for the equations, with and without significant plasma and/ or microsomal protein binding). Using this well-stirred model, it has proved possible to predict the hepatic clearance from in vitro intrinsic clearance rates in the rat, dog and human (Table 10.2). The hepatic clearance value for the rat (0.972ml-min-1 - mg-1 protein) was approximately one-tenth the actual clearance found in vivo; well in agreement with the observation that in vivo Compound X was eliminated by the rat, largely unchanged, by the kidneys 90%).

To predict hepatic clearance of Compound X in man, human in vitro intrinsic clearance could then be scaled to hepatic clearance, using a technique that had been validated in the rat. Ashfortt et al 1995Renal clearance is subject to an allometric relationship and can generally be scaled across species (see below). The predicted in vivo renal Cl for the rat (estimated by multiplying the predicted hepatic Cl by 9) may be scaled allometrically to

Table 10.1 Equations for predicting hepatic clearance using the well-stirred model

In the absence of serum or microsomal protein binding

In the presence of significant serum protein binding

In the presence of both serum and microsomal protein binding

Table 10.2 Comparison of the predicted in vivo hepatic clearance and the actual clearance values for Compund X

Predicted in vivo hepatic Predicted in vivo renal Predicted in vivo total Actual in vivo CL (ml min-1 kg-1) CL (mlmin-1 kg-1) CL (mlmin-1 kg-1) CL (mlmin-1 kg-1)

Predicted values were scaled from in vitro half-life data using liver microsomes and the well-stirred model of hepatic extraction. Hepatic CL predictions were corrected for plasma and microsomal protein binding. Predicted total CL was obtained by adding in renal CL estimates which were, in turn, scaled allometrically ( Y = aW0-75).

obtain a prediction for human in vivo renal clearance. Total or systemic CI in man can then be estimated by adding the two clearance parameters (hepatic and renal) together; in practice, for Compound X, later first-in-man data revealed an actual in vivo CI nearly identical to the predicted total CI (2.15 vs. 1.87-2.45 ml min-1mg-1, respectively; Table 10.2). Here, then, is a real-world example of, first, how rat in vitro and in vivo preclinical data were used to develop and validate a scaling method for Compound X in the rat; and second, how the scaling method successfully predicted in vivo overall drug clearance in man.

However, if the same methods are used for Compound X in the dog, things initially appear to be different. Scaling the in vitro intrinsic clearance to hepatic CI using the rat-validated method, in conjunction with allometric scaling of renal CI, resulted in a five-fold underprediction of total or systemic clearance in vivo. However, further metabolism studies in the dog in vivo revealed that Compound X undergoes significant additional biotransformation, particularly N-methylation, which is unique (as far as we are aware) to this species, and invalidates some of our in vitro assumptions. This canine biotransformation pathway was not detected by our initial microsomal studies because there are no N-methyl transferases in microsomes. Thus, although we did not successfully predict dog systemic clearance for Compound X, our scaling tactics did eventually teach us about a new clearance mechanism, and how important this was for the systemic clearance of Compound X in the dog.

This is an example of how in vitro studies can be combined with in vivo preclinical data, leading to useful prediction of human systemic drug clearance. Nonetheless, several caveats are encountered in such scaling exercises, which warrant restating.

The first caveat is that all clearance pathways (hepatic, renal, biliary, or other) must be taken into consideration. If a compound undergoes a high level of hepatic clearance, then in vitro-in vivo scaling may be used to predict the fraction of systemic clearance expected from this pathway. If a compound undergoes a high level of renal elimination, allometric scaling may be also used to predict the clearance attributed to this pathway.

The second caveat is that, in order to accurately predict hepatic clearance, the correct in vitro system must be chosen. If the candidate drug is primarily oxidatively metabolized, then liver microsomes will be sufficient. However, if the potential for non-microsomal biotransformation exists, then a different in vitro system, such as hepatocyte suspensions, should be used. In the illustration above, it turned out, as far as clearance of Compound X is concerned, man is specifically like a rat, and unlike a dog.

The third caveat is that one must consider the variability in the expression of metabolizing enzymes between individuals. Oxidative metabolism (seen in vivo and in microsomal enzymes), and especially cytochrome P450s, vary tremendously between human individuals (Meyer 1994; Shimada et al 1994). Had we used a single donor microsomal sample, rather than pooled liver microsomes (a pool consisting of at least eight individual donors), to scale in vitro data to in vivo hepatic clearance, we might have made greatly misleading predictions (note that oxidative, initial drug metabolism is sometimes called 'Phase I metabolism' in the literature, causing ambiguity with the stage of drug development or type of clinical trial).

0 0

Post a comment