Prediction of Human Dose

As stated in the Section 6.1, one of the principal purposes of carrying out DMPK studies during the discovery phase is to reduce the failure rate during development. For DMPK this logically means predicting the pharmacokinetics that will be observed and hence the dose that will be required in man when clinical studies are carried out. 

It is possible to predict the steady-state minimum plasma concentration (Fig. 6.4) using the equation  

ss is the minimum plasma concentration at steady state fa is the fraction absorbed in man  

V is the volume of distribution at steady state (in L kg-1) k is the elimination constant (this is given by clearance divided by volume)  

The equation is an approximation, adapted from that for intravenous dosing [81], corrected by addition of a term for absorption. Essentially it assumes instantaneous absorption of the dose, but for compounds with reasonable physico-chemical and PK properties that are expected to be suitable for once-a-day dosing, this approximation makes little difference to the predicted value of Cmin ss. Use of the relationship can provide a simple approach for estimating the required dose in man for a compound in the discovery phase. 

Equation 8.7 can be rearranged to allow the prediction of the dose and dose interval, provided that the following can be estimated human potency, absorption, clearance, and volume. 

Estimation of the potency can be made in several ways and will be highly dependent on the nature of the target. If a purified system is used, it is normal to correct for the effect of plasma protein binding (which can be measured directly in human plasma) as it is usual for the effect to be proportional to the unbound concentration [82]. This can be used to set a value for the minimum plasma concentration at steady state. 

As described above, it will be normal to assume that the dose interval is 24 h, that is, once-a-day dosing. Absorption can be estimated with good confidence in the rat (see Section 8.1). Clearance is the sum of the predicted hepatic, renal, biliary, and extrahepatic clearance. Hepatic clearance can be derived from in vitro studies with the appropriate human system, using either microsomes or hepatocytes. We prefer to use an approach based on that described by Houston and Carlile [83]. Renal clearance can be predicted allometrically (see Section 8.8.1). The other two potential methods of clearance are difficult to predict. To minimize the risks, animal studies can be used to select compounds that show little or no potential for clearance by these routes. As volume can be predicted from that measured in the dog, after correction for human 

---

There are many applications of PK/PD studies in drug discovery. The prediction of human dose and regimen are important to make sure that the dose amount is developable and the dosing regimen meets the target candidate profile. In order to use preclinical PK/PD studies to predict the human dose and dose regimen, the following two issues need to be addressed (1) the exposure parameter that correlates with efficacy and (2) the human relevance of the preclinical model. 

The prediction of human dose from in vitro data can be achieved using the pharmacokinetic model shown in Equation 16.1, which relies on an estimate of Cmin after the first dose as an input and assumes that exposure exceeding the free potency for the entire dosing interval is required to show efficacy and is attained at steady state [59] ... 

It is important to recognize that the in vitro permeability obtained in cell mono-layers (such as Caco-2 models) should be considered as a qualitative rather than quantitative value. Especially poor are predictions of fraction dose absorbed for carrier-mediated drugs with low Caco-2 permeability and predictions of high fraction dose absorbed in humans [7, 20, 42, 48, 51]. However, it is possible to establish a reasonably good IVIVC correlation when passive diffusion is the dominating absorption mechanism. 

Saitoh, R., Sugano, K., Takata, N., Tachibana, T., Higashida, A., Nabuchi, Y., and Aso, Y., Correction of permeability with pore radius of tight junctions in Caco-2 monolayers improves the prediction of the dose fraction of hydrophilic drugs absorbed by humans, Pharm. Res., 21, 749, 2004. 

Holliman, C. L., Buchholz, L. M., McFadden, J. R., and Pace, G. (2005). The prediction of human pharmacokinetics at the therapeutic dose from low sub-therapeutic doses in human. Paper presented at the 11th Annual FDA Science Forum, Washington, DC. 

PB-PK models, sometimes referred to as biologically-based disposition models, allow for accurate extrapolation of rodent data to estimate human dose-response relationships (Paustenbach, 1995). PB-PK models, unlike compartmental models, have the capability of simulating a chemical s behavior in biological systems. The purpose of a PB-PK model is to predict the human dose-response relationship based on animal data by quantitatively estimating the delivered dose of the biologically relevant chemical species in a target tissue (Andersen etal., 1987 Clewell etal., 1994 Leung and Paustenbach, 1995 Ramsey and Andersen, 1984). 

Nonclinical pharmacokinetic investigation is helpful when interpreting the data from safety (pharmacological) studies, and it also provides support for toxicology studies. While nonhuman pharmacokinetic parameters are not perfectly predictive of human pharmacokinetics, they do constitute meaningful quantitative data that improve the chances of selecting the correct range of safe doses to test in humans. This dose selection is critically important to the success of clinical trials. 

Alarie Y. 1981. Dose-response analysis in animal studies Prediction of human responses. Environ Health Perspect 42 9-13. 

At least one well-conducted study must show reproductive or developmental toxicity in a mammalian species. When the study data are insufficient, improper study design or execution, inadequate doses or duration of exposure, poor survival, or too few animals to achieve statistical power are often the cause. At present, no nonmammalian or in vitro systems are considered to be predictive of human responses, and are not accepted by... 

The half-life of a drug is a major contributor to the dosing regimen, and it is a function of the clearance and apparent volume of distribution (VD), each of which can be predicted and combined to predict the half life. Drugs with short half-lives are more likely to be required to be administered more frequently than those with long half-lives. Much attention has been focused on the prediction of human half-life. Good success is attained if the two major components of half-life, clearance and VD, are predicted separately and combined to generate a half-life prediction. 

In the approximately 70 years since the discovery of the toxic G agents and 50 years since the subsequent development of the V agents, humans have only occassionally served as test subjects in laboratory studies designed to determine threshold toxic effects associated with low-level (nonlethal) sarin and VX vapor exposures (2-10 min) (Johns, 1952 Sim, 1956 Bramwell et al., 1963). In addition, although the toxic effects of accidental exposures and nonexperimental exposures from terrorist or military attacks are documented, critical information related to the exposure conditions can only be estimated at best. Thus, estimates of human dose-responses to nerve agent vapor exposures from such sources are often associated with significant uncertainty and are of limited utility in predicting health hazard risks. 

The prediction of human PK from precUnical in vitro and in vivo data remains an important goal in the drug discovery and development process to obtain an early estimate of efficacious human dose, dosing regimen, and safety window prior to first-in-human trials. From a clinical study perspective, under-prediction of human exposure would lead to safety concerns because actual exposures would be greater than expected, and inversely. The primary focus in the literature has been on the estimation of basic PK parameters such... 

---