1. Methylene Chloride a case study for Dose-Dependent Transitions Raymond M. David, Ph.D. Eastman Kodak Company © Eastman Kodak Company, 2005

2. Overview Methylene chloride (DCM) is a good case study for risk assessment because it is a classic example of dose-dependent transition in carcinogenesis there are human data for metabolism around the inflection point example of species differences in metabolism and genetic polymorphisms, which impact the quantitation of risk PBPK modeling has been used for species extrapolation and risk assessment

3. Carcinogenic Potential National Coffee Association (NCA) study – 0, 60, 125, 185, and 250 mg/kg body in drinking water to F-344 rats and B 6 C 3 F 1 mice for 2 years National Toxicology Program (NTP) study – 0, 2000, or 4000 ppm by inhalation to F-344 rats and B 6 C 3 F 1 mice for 2 years

4. Tumor response Dose mg/kg/d Tumor Incidence OrganSex/Strain/ Species Study 0 60 125 185 250 24/125 51/200 30/100 31/99 35/125 Liver Male B 6 C 3 F 1 NCA; Serota et al. 0 1582 3162 3/50 16/48 40/48 Liver Female B 6 C 3 F 1 NTP 0 1582 3162 3/45 30/46 41/46 Lung Female B 6 C 3 F 1 NTP

5. Liver tumor response

6. Tumor incidence summary Increased incidence of hepatocellular adenomas and carcinomas were observed in mice (one sex) at  125 mg/kg/d. Increased incidence of lung tumors (alveolar/bronchiolar adenomas) were observed in mice exposed to airborne concentrations of  2000 ppm. F-344 rats showed no evidence of increased liver or lung tumors.

7. Mode of action At low concentrations, DCM is metabolized primarily by cytochrome P450 (CYP). Kubic and Anders (1975) and Anders et al. (1977) demonstrated that DCM was metabolized by CYP to carbon monoxide. Kim and Kim (1996) later identified CYP2E1 as the isozyme associated with this pathway. Kubic and Anders (1978) determined the K m (50.1 mM) and V max (5.4 nmol CO/mg prot/min). McKenna et al. (1982) showed that CYP2E1 in laboratory animals was saturated above concentrations of 500 ppm.

8. Mode of action CYP 2E1 catalyzed: CH 2 Cl 2  CHOHCl 2  HCOCl  CO + CO 2 formyl chloride COHb

9. Mode of action At higher concentrations, CYP pathway can be saturated and GST pathway metabolizes DCM. Ahmed and Anders (1976) and Anders et al. (1977) demonstrated that DCM was metabolized via a GST pathway. Gargas et al. (1986) proposed the current metabolic scheme via GST pathway. Blocki et al. (1994) showed that GST 5-5, a  -class GST (also known as T1-1 in humans), had the highest specific activity for DCM (11,000 nmol/min/mg protein) with a Km of 300 µ M.

10. Mode of action GST catalyzed: CH 2 Cl 2  GSCH 2 Cl GSCH 2 OH  HCHO chloromethylglutathione  GSCHO  HCOOH  CO 2

11. MOA – supporting data Reynolds and Yee (1967) and Anders et al. (1977) showed that 14 C-DCM was bound to tissue protein and lipid. Casanova et al. (1992) demonstrated an increase of DNA-protein cross-links (DPX) in the liver and RNA-formaldehyde adducts (RFA) in the lungs. DNA adducts may also be formed directly from the chloromethylglutathione intermediate rather than formaldehyde (Marsch et al., 2001, 2004).

12. MOA – supporting data Graves et al. (1994) and Thier et al. (1993) linked mutations observed only in S. typhymurium strains TA1535 and TA100 to nascent GST activity. Other tests for genetic toxicity generally negative.

13. Constructing the data set Metabolic parameters for different species Developing human data parameters Developing a model Understanding the compartments Physiological parameters in different species

14. Species metabolic parameters SpeciesK m (mM) V max (nmol/min/mg prot) K m (mM) V max (nmol/min/mg prot) mouse 1.84  0.3315.90  1.10137  21118.2  14.4 rat 1.42  0.745.39  0.94 nd human0.92 – 2.82 1.53 – 13.00 43.8 – 44.1 6.04 – 7.05 MFO pathway GST pathway From Reitz et al., 1988. nd = not determined

15. Species metabolic parameters SpeciesTissueMFOGST MouseLiver Lung 1.760 ± 0.115 0.732 ± 0.115 5290 ± 430 727 ± 64 RatLiver Lung 0.814 ± 0.118 0.111 ± 0.035 1380 ± 110 77 ± 5 HumanLiver Lung 0.418 ± 0.157 0.0006 ± 0.0003 1650 ± 480 78 ± 47 Specific activities from Lorenz et al. (1984) in nmol/min/mg protein as reported by Andersen et al. (1987)

16. Species metabolic parameters Glutathione transferase (GST T1-1) catalyzes the conjugation of glutathione and DCM in mice and humans. The gene is polymorphic in humans Non-conjugators: GST T1 ( – / – ) Low conjugators: GST T1 (+/ – ) High conjugators: GST T1 (+/+) Distribution of the null phenotype in humans has been studied.

17. GSTT1 -/- Distribution Group% Population% Homozygous Asian3.962 Caucasian75.519.7 African- American 12.221.8 Mexican- American 11.49.7 From El-Masri et al., 1999.

18. Human data sets Exposure levels (ppm) Duration hrs Number of subjects Reference 100, 35066Andersen et al. (1987) 50, 100, 150, 200 813DiVincenzo et al. (1986) 250, 500, 1000 2.514Astrand et al. (1975)

19. 25 30 35 40 123456789 10111213 A BC D E Clewell (1995) Jonsson and Johanson (2001) Expert Elicitation OSHA Prior OSHA Posterior Individual Values (Sweeney et al., 2004) Individual Values (Jonsson et al. (2001) Population Values Vmaxc/Km (/hr) Human data

20. PBPK modeling First interspecies extrapolation using PBPK modeling was Andersen et al. (1987). Dose metric was blood, tissue, and exhaled DCM. Human metabolic values were mean from subjects exposed to 100 or 350 ppm for 6 hours.

21. Gas Exchange Lung Metabolism Richly perfused Fat Slowly perfused Liver GI tract GSTCYP Blood CYP GST Andersen model

22. PBPK models for DCM assessment CitationRemarks Reitz et al., 1988Deterministic approach. Andersen et al. (1987) model updated with measured MFO and GST rate constants. Andersen et al., 1991Deterministic approach. Blood compartment added to describe carbon monoxide and carboxyhemoglobin kinetics. Dankovic et al., 1994Deterministic approach. Mean values for alveolar ventilation, cardiac ouput, and tissue blood flow increased. Casanova, et al., 1996Deterministic approach. Liver DNA-protein cross-links from formaldehyde used as the dosimeter of effect.

23. PBPK models for DCM assessment CitationRemarks Bois and Smith, 1995 Probabilistic (Bayesian) approach. Bone marrow compartment added, variance in metabolic rate constants increased. Thomas et al., 1996 Probabilistic (Bayesian) approach. Variability from MFO induction, GST inhibition, and tissue solubility included. El-Masri et al., 1999 Probabilistic (Bayesian) approach incorporating GST-T1 polymorphisms and estimating DPX. Jonsson and Johanson, 2001 Probabilistic (Bayesian) approach. New fat and muscle compartments. Includes population estimates of glutathione transfersase T1 gene frequencies

24. Changes in unit risk over time SourceUnit risk (per µg/m 3 ) EPA 1985 1.0  10 -6 EPA 1991 4.7  10 -7 El Masri et al., 1999 1.9  10 -10 Jonsson and Johanson, 2001 1.9  10 -10

25. DCM PBPK model results

28. Gas Exchange Lung Metabolism Richly perfused Fat Slowly perfused Liver GI tract GST CYP Blood CYP GST CYP Sweeney model

29. Updating the risk assessment Do the new human data and the model change the calculated unit risk? Perhaps --- the unit risk is 4.8 x 10 -8 using the Sweeney PBPK model compared with 4.7 x 10 -7 used by the EPA. Using probabilistic methodology and genetic polymorphisms might also impact the unit risk calculation.

30. Summary DCM is a good example for quantitative risk assessment because it demonstrates a dose-dependent transition from non-carcinogenic pathway to carcinogenic pathway human data are available genetic polymorphisms in human populations can be factored into the assessment PBPK models extrapolating from animal to humans are available