**5.1 Sampling excreta to estimate drug and metabolites in tissues**

The concept of sampling excreta to estimate drug and metabolites in human tissues is still evolving. The importance of understanding absolute abundance of metabolites from sampling excreta was highlighted by the need to understand the importance of metabolites in safety testing or MIST (Baillie et al., 2002; Smith & Obach, 2005). Smith and Obach concluded that the risk assessment of metabolites would seem more prudent if it was based on absolute mass and not proportion of drug-related material (Smith & Obach, 2005); therefore, sampling excreta and analyzing total amount of metabolite excreted would be more useful than sampling plasma (especially at higher dose of the drug). The recommendation for sampling excreta was based on determining the entire body burden of the metabolites for this MIST guidance and less about sampling excreta to estimate drug and metabolites in tissues.

In animals, the concept of sampling excreta to estimate drug and metabolites in tissues has been applied in a limited fashion. For example in beef steers treated with gentamicin, a small residue remains bound to the kidney cortex tissue for many months (this residue is unacceptable at the time of slaughter). Interestingly, plasma levels of gentamicin declined rapidly to no detectable levels within 3 days after intramuscular administration of gentamicin, while measurable amounts in urine persisted for 75 days before the concentration of gentamicin declined to levels too low to quantitate by the available liquid chromatography tandem mass spectrometry (LC/MS/MS) technique (Chiesa et al., 2006). An estimated correlation between an extrapolation of urine gentamicin concentration to the corresponding kidney tissue sample suggested a urine to kidney tissue relationship of 1:100. A test system sufficiently sensitive to a urine gentamicin concentration of 1 ng/mL correlated with the estimated 100 ng/g gentamicin limit applied to the fresh kidney of the recently slaughtered bovine (Chiesa et al., 2006). This example highlights the utility of measuring excreta (e.g., urine) to better estimate concentrations of drug in tissue (e.g., kidney).

The challenge of excreta being a surrogate model to assess concentrations of drug and metabolites in human tissues is the limited understanding of how concentrations of drug and metabolites in the excreta will relate to the concentrations in the respective tissue. This challenge can be minimized by establishing a relationship between the concentration of drug and metabolites for excreta and tissues in animals (as illustrated by the above example with gentamicin in beef steers). In addition, translating that relationship from animal to human with in silico tools (e.g., PBPK modelling) and in vitro and in vivo human models (e.g., primary in vitro human cell models and humanized mice) will increase the confidence in including safety margins from exposure of drug and its metabolites in the tissues (in addition to plasma) where organ specific toxicity is observed. Below is a case example where the utility of semen as a surrogate model to assess the concentrations of drug and metabolites in dog testes was investigated.

#### **5.1.1 Case example: Utility of semen as a potential matrix to estimate drug and metabolites in testes**

In this case example, the potential of semen was evaluated as a matrix to determine the concentration of Drug A (same drug candidate described in the section for Toxicokinetics and Testicular Toxicity) and its metabolite (M1) in dog testes (for potential extrapolation to human). For this study, dogs were given a single oral dose of Drug A and then at different

For the best extrapolation, the mechanism of interaction leading to toxicity would be known; for example, a known biological process that is disturbed by a known entity, parent, and/or metabolite (Andersen, 1995). However in many cases, this mechanism is not known and PBPK models can assist in possibly identifying these mechanisms. Especially when modelling efforts address the appropriate questions, the systematic discovery of these mechanisms is possible. The key is to develop models with appropriate measures of tissue concentrations in animals and possibly excreta concentrations in animals and humans. To strengthen this extrapolation, in vitro systems, such as primary in vitro human cell models (e.g., hepatocyte sandwich-cultured cell model and proximal tubule cell monolayers), and humanized mice, will also provide vital parameters (e.g., pharmacokinetics rate constants) for the PBPK modelling in order to extrapolate tissues concentrations from animal to human. In 2001, a consensus building workshop sponsored by the Society of Toxicology concluded that the human in vitro systems, through quantitative measurements and PBPK modelling, can play an important role in dose-response assessment (MacGregor et al., 2001). Therefore in the near future, the combination of these technologies may allow researchers

the ability to estimate drug and metabolites concentrations in human tissues.

the amount of models for in vitro human tissue models.

**tissues** 

**5.3 Utility of supplementary human models to estimate drug and metabolites in** 

The primary challenge in calculating safety margins in tissues where organ-specific toxicity is observed is the access to human tissue samples for the measurement of drug and its metabolites. One method to address this challenge is to simulate the distribution of drug and its metabolites in a human in vitro model. For example, development of valid and reliable techniques to quantify biliary excretion of drugs in healthy human volunteers is difficult. Measurements of drug concentrations in bile can only be obtained from patients diagnosed with diseases of the gallbladder and biliary tract who require medical procedures that allow this measurement (Ghibellini et al., 2006). However, there is a promising, recent technique to estimate bile in healthy human volunteers with an oroenteric catheter to aspirate duodenal secretions, and gamma scintigraphy to determine gallbladder contraction. This technique allowed the comparison of the biliary clearance of three compounds estimated with sandwichcultured human hepatocytes (a human in vitro model). The rank order of biliary clearance predicted from in vitro corresponded well with the in vivo biliary clearance values in mL/min/kg for Tc-99m mebrofenin (7.44 vs 16.1), Tc-99m sestamibi (1.20 vs 5.51), and Tc-99m piperacillin (0.028 vs 0.032) (Ghibellini et al., 2007). Since sandwich-cultured human hepatocytes need to uptake drug across their sinusoidal membrane in order to excrete the drug across their canalicular membrane for the in vitro measurement of biliary excretion, this verification of a good prediction of this human in vitro model from the clinical study suggests that the intracellular concentration within these sandwich-cultured human hepatocytes can also estimate concentrations of drug in the human hepatocyte in vivo. Therefore, in vitro models have the potential to supplement costly and difficult sampling in healthy human volunteers to estimate drug and metabolite concentrations in tissues and excreta. However, significantly more research is needed to realize this potential in existing models and to expand

Another possible human model to estimate drug and metabolites in organs and/or tissues is mice with humanized organs and/or tissues. To create this model, a severe combined immunodeficient (SCID) mouse line is injected with human cells from the human tissue into

time points dogs were ejaculated to collect semen and their testes were sampled. The toxicokinetic profile of M1 in semen and testes was similar (Table 7). Furthermore, the exposure of parent in testes also approximated the exposure of parent in semen where the exposure of Drug A in semen was approximately 2.5-fold higher than the exposure in testes (Table 7). These results suggest that semen approximated the exposure of Drug A and M1 in testes. Therefore, excreta may be a possible surrogate matrix to estimate tissue concentrations of drug candidate and its metabolites; however, supplementary systems like primary in vitro human cell models and humanized mice, combined with PBPK modelling, will be needed to extrapolate these results to human.


Table 7. Toxicokinetic Profile of Drug A and its Metabolites in Dog Testes and Semen.

Before dosing, Beagle dogs (n=9) were trained for ejaculation 2 times/week. Dogs were administered a single oral dose of 15 mg/kg Drug A. Testes (n = 1/time point) were collected at 1, 4, 7, 24, 48, 72, 96, 168, and 336 hours post dose from dogs at each time point. Semen was also collected in the period between dosing and sacrifice. Bioanalysis of semen and testes for Drug A and its metabolites, M1 and M2, was performed. Toxicokinetic parameters were then determined.

#### **5.2 Utility of PBPK to estimate drug and metabolites in tissues**

PBPK models aid in the understanding of the disposition of chemicals in the body in different animal species, including humans. In toxicological research, PBPK modelling was initiated approximately 30 years or so, and mainly from an environmental toxicology perspective. For example, PBPK models were developed for polychlorinated biphenyls, methylene chloride, and other persistent lipophilic compounds starting in the mid 1980s (Andersen, 1995). In the past, the utilization of PBPK models in safety assessment departments within the pharmaceutical industry was not common, although the utilization of PBPK models is gaining momentum.

The utility of PBPK models is to extrapolate from one environment to another; for example, PBPK models extrapolate from high to low dose, different routes of administration, interspecies, and different durations of exposure. All of these extrapolations are potentially needed to bridge knowledge of drug and metabolites concentrations in the tissues of safety assessment species (e.g., rat, dog, and monkey) to human tissues (Thompson et al., 2007).

time points dogs were ejaculated to collect semen and their testes were sampled. The toxicokinetic profile of M1 in semen and testes was similar (Table 7). Furthermore, the exposure of parent in testes also approximated the exposure of parent in semen where the exposure of Drug A in semen was approximately 2.5-fold higher than the exposure in testes (Table 7). These results suggest that semen approximated the exposure of Drug A and M1 in testes. Therefore, excreta may be a possible surrogate matrix to estimate tissue concentrations of drug candidate and its metabolites; however, supplementary systems like primary in vitro human cell models and humanized mice, combined with PBPK modelling,

> Cmax (ng/mL or g)

Testes M1 13 7 2890 80787 81831

Testes M2 3 1 28 87 108

Testes Parent 18 7 12000 232701 233251

Semen M1 16 7 3680 75766 75867

Semen M2 ND ND ND ND ND

Semen Parent 5 4 42700 593334 593391

Before dosing, Beagle dogs (n=9) were trained for ejaculation 2 times/week. Dogs were administered a single oral dose of 15 mg/kg Drug A. Testes (n = 1/time point) were collected at 1, 4, 7, 24, 48, 72, 96, 168, and 336 hours post dose from dogs at each time point. Semen was also collected in the period between dosing and sacrifice. Bioanalysis of semen and testes for Drug A and its metabolites, M1 and M2, was performed. Toxicokinetic

PBPK models aid in the understanding of the disposition of chemicals in the body in different animal species, including humans. In toxicological research, PBPK modelling was initiated approximately 30 years or so, and mainly from an environmental toxicology perspective. For example, PBPK models were developed for polychlorinated biphenyls, methylene chloride, and other persistent lipophilic compounds starting in the mid 1980s (Andersen, 1995). In the past, the utilization of PBPK models in safety assessment departments within the pharmaceutical industry was not common, although the utilization

The utility of PBPK models is to extrapolate from one environment to another; for example, PBPK models extrapolate from high to low dose, different routes of administration, interspecies, and different durations of exposure. All of these extrapolations are potentially needed to bridge knowledge of drug and metabolites concentrations in the tissues of safety assessment species (e.g., rat, dog, and monkey) to human tissues (Thompson et al., 2007).

Table 7. Toxicokinetic Profile of Drug A and its Metabolites in Dog Testes and Semen.

**5.2 Utility of PBPK to estimate drug and metabolites in tissues** 

AUClast (ng\*hr/mL or g)

AUCinf (ng\*hr/mL or g)

will be needed to extrapolate these results to human.

(hr)

Tmax (hr)

Half Life

parameters were then determined.

of PBPK models is gaining momentum.

For the best extrapolation, the mechanism of interaction leading to toxicity would be known; for example, a known biological process that is disturbed by a known entity, parent, and/or metabolite (Andersen, 1995). However in many cases, this mechanism is not known and PBPK models can assist in possibly identifying these mechanisms. Especially when modelling efforts address the appropriate questions, the systematic discovery of these mechanisms is possible. The key is to develop models with appropriate measures of tissue concentrations in animals and possibly excreta concentrations in animals and humans. To strengthen this extrapolation, in vitro systems, such as primary in vitro human cell models (e.g., hepatocyte sandwich-cultured cell model and proximal tubule cell monolayers), and humanized mice, will also provide vital parameters (e.g., pharmacokinetics rate constants) for the PBPK modelling in order to extrapolate tissues concentrations from animal to human. In 2001, a consensus building workshop sponsored by the Society of Toxicology concluded that the human in vitro systems, through quantitative measurements and PBPK modelling, can play an important role in dose-response assessment (MacGregor et al., 2001). Therefore in the near future, the combination of these technologies may allow researchers the ability to estimate drug and metabolites concentrations in human tissues.

#### **5.3 Utility of supplementary human models to estimate drug and metabolites in tissues**

The primary challenge in calculating safety margins in tissues where organ-specific toxicity is observed is the access to human tissue samples for the measurement of drug and its metabolites. One method to address this challenge is to simulate the distribution of drug and its metabolites in a human in vitro model. For example, development of valid and reliable techniques to quantify biliary excretion of drugs in healthy human volunteers is difficult. Measurements of drug concentrations in bile can only be obtained from patients diagnosed with diseases of the gallbladder and biliary tract who require medical procedures that allow this measurement (Ghibellini et al., 2006). However, there is a promising, recent technique to estimate bile in healthy human volunteers with an oroenteric catheter to aspirate duodenal secretions, and gamma scintigraphy to determine gallbladder contraction. This technique allowed the comparison of the biliary clearance of three compounds estimated with sandwichcultured human hepatocytes (a human in vitro model). The rank order of biliary clearance predicted from in vitro corresponded well with the in vivo biliary clearance values in mL/min/kg for Tc-99m mebrofenin (7.44 vs 16.1), Tc-99m sestamibi (1.20 vs 5.51), and Tc-99m piperacillin (0.028 vs 0.032) (Ghibellini et al., 2007). Since sandwich-cultured human hepatocytes need to uptake drug across their sinusoidal membrane in order to excrete the drug across their canalicular membrane for the in vitro measurement of biliary excretion, this verification of a good prediction of this human in vitro model from the clinical study suggests that the intracellular concentration within these sandwich-cultured human hepatocytes can also estimate concentrations of drug in the human hepatocyte in vivo. Therefore, in vitro models have the potential to supplement costly and difficult sampling in healthy human volunteers to estimate drug and metabolite concentrations in tissues and excreta. However, significantly more research is needed to realize this potential in existing models and to expand the amount of models for in vitro human tissue models.

Another possible human model to estimate drug and metabolites in organs and/or tissues is mice with humanized organs and/or tissues. To create this model, a severe combined immunodeficient (SCID) mouse line is injected with human cells from the human tissue into

I would like to thank 1) Rita Geerts, Wenying Jian, Rick Edom, and David La for their contribution towards the rat testicular toxicity section; 2) Gregory Reich and Freddy Schoetens for their contribution towards the dog liver toxicity section; 3) David La for his contribution towards the monkey CNS toxicity section; and 4) Rob Thurmond, David Evans,

Andersen, M.E. (1995). Development of physiologically based pharmacokinetic and

testing. *Toxicol Appl Pharmacol,* Vol.182, No.3, pp. 188-196, ISSN 0041-008X Chandra, P., & Brouwer, K.L. (2004). The complexities of hepatic drug transport: current

Chiesa, O.A., von Bredow, J., Heller, D., Nochetto, C., Smith, M., Moulton, K., & Thomas, M.

Cordon-Cardo, C., O'Brien, J.P., Casals, D., Rittman-Grauer, L., Biedler, J.L., Melamed, M.R.,

Dixit, R., & Ward, P.D. (2007). Use of Classical Pharmacokinetic Evaluations in Drug

Ghibellini, G., Leslie, E.M., & Brouwer, K.L. (2006). Methods to evaluate biliary excretion of

Ghibellini, G., Vasist, L.S., Leslie, E.M., Heizer, W.D., Kowalsky, R.J., Calvo, B.F., &

Jiang, X.L., Gonzalez, F.J., & Yu, A.M. (2011). Drug-metabolizing enzyme, transporter, and

humans. *Clin Pharmacol Ther,* Vol.81, No.3, pp. 406-413, ISSN 0009-9236 Ito, K., Uchida, Y., Ohtsuki, S., Aizawa, S., Kawakami, H., Katsukura, Y., Kamiie, J., &

risk assessment. *Toxicol Lett,* Vol.79, No.1-3, pp. 35-44, ISSN 0378-4274 Baillie, T.A., Cayen, M.N., Fouda, H., Gerson, R.J., Green, J.D., Grossman, S.J., Klunk, L.J.,

physiologically based pharmacodynamic models for applications in toxicology and

LeBlanc, B., Perkins, D.G., & Shipley, L.A. (2002). Drug metabolites in safety

knowledge and emerging concepts. *Pharm Res,* Vol.21, No.5, pp. 719-735, ISSN

(2006). Use of tissue-fluid correlations to estimate gentamicin residues in kidney tissue of Holstein steers. *J Vet Pharmacol Ther,* Vol.29, No.2, pp. 99-106, ISSN 0140-

& Bertino, J.R. (1989). Multidrug-resistance gene (P-glycoprotein) is expressed by endothelial cells at blood-brain barrier sites. *Proc Natl Acad Sci U S A,* Vol.86, No.2,

Development and Safety Assessment, In: *Toxicokinetics and Risk Assessment*, Lipscomb, J.C., & Ohanian, E.V. (Ed.), pp. 95-122, Informa Healthcare USA Inc.,

drugs in humans: an updated review. *Mol Pharm,* Vol.3, No.3, pp. 198-211, ISSN

Brouwer, K.L. (2007). In vitro-in vivo correlation of hepatobiliary drug clearance in

Terasaki, T. (2011). Quantitative membrane protein expression at the blood-brain barrier of adult and younger cynomolgus monkeys. *J Pharm Sci,* Vol.100, No.9, pp.

nuclear receptor genetically modified mouse models. *Drug Metab Rev,* Vol.43, No.1,

Sandra Snook, Jan de Jong, and David La for reviewing this chapter.

**7. Acknowledgment** 

**8. References** 

0724-8741

1543-8384

pp. 695-698, ISSN 0027-8424

3939-3950, ISSN 1520-6017

pp. 27-40, ISSN 1097-9883

ISBN 978-0-8493-3722-2, New York, NY

7783

the respective mouse tissue. For example, injection of cryopreserved human hepatocytes through a small, left flank incision into the inferior splenic pole in a SCID mouse created a mouse with humanized liver that was replaced by more than 80% of human hepatocytes (Okumura et al., 2007). In this chimeric mice model, cefmetazole (CMZ) excretions in urine and feces were 81.0 and 5.9% of the dose, respectively; however, excretions in urine and feces in control SCID mice were 23.7 and 59.4% of the dose, respectively (Okumura et al., 2007). Because CMZ is mainly excreted in urine in humans, the excretory profile in chimeric mice was demonstrated to be similar to humans. Interestingly in the chimeric mice, the hepatic mRNA expression of human drug transporters (e.g., MDR1, BSEP, MRP2, BCRP, OCT1, and OATP1B1/1B3) were detectable; whereas, the hepatic mRNA expression of mouse drug transporters in the chimeric mice was significantly lower than in the control SCID mice (Okumura et al., 2007). In conclusion, chimeric mice exhibited a humanized profile of drug excretion, suggesting that this chimeric mouse line would be a useful animal model to predict human ADME. Most studies have focused on humanized liver models; however, the potential for humanization of other organs and/or tissues in the mouse is evident in the near future. These new potential models will markedly improve the ability to estimate drug and metabolite concentrations in human organs and/or tissues.
