Characterization of Variability in CYP2E1

Read Also

Other research teams have also chosen to characterize overall inter-individual variability in CYP2E1 rather than to attempt to attribute variability separately to genetic polymorphisms or to dietary or other factors.  Four recent papers have utilized   different methodologies to characterize variability (Iyer and Sinz, 1999; Johnsrud et al., 2003; Lipscomb et al., 2003; and Dorne et al., 2004.)  All four produced similar estimates of variability in CYP2E1 in adults (GSDs ranging from 1.2-1.7).  These studies are summarized in Table 4-1.

Iyer and Sinz, (1999) characterized the activity of CYP2E1 (nmol/min per mg microsomal protein) in 21 human liver samples obtained from a liver bank using p-nitrophenol as probe substrate.   Although chlorzoxazone is a more common substrate, the authors reported that the specificity of p-nitrophenol for CYP2E1 is shown by studies of CYP2E1 cDNA expression systems in which approximately 85% of p-nitrophenol hydroxylation was catalyzed by this human CYP (Tassaneekyakul et al., 1993).)  The donors ranged in age from 6 to 52 years.  They varied in characteristics associated with CYP2E1 induction/inhibition; the data set included smokers, non-smokers, those who reported alcohol consumption and those who reported none. One individual was characterized as obese.

Variation in CYP2E1 activity in the Iyer and Sinz (1999) samples was consistent with a lognormal distribution:  the geometric mean (GM) and geometric standard deviation (GSD) for their data appear in the first line of Table 4-1.   The activity levels spanned a 4.8 fold range.  This degree of variation is consistent with several other studies of CYP2E1 catalytic activity using different probe substrates in other liver banks alluded to by Iyer and Sinz (1999) where 2 to 11 fold ranges were reported (no other distribution parameters were reported for these data).  The GSD was also a similar magnitude as that reported in other studies discussed below.

Johnsrud et al., 2003 studied developmental changes in CYP2E1 protein expression in human hepatic microsomes.  The investigators analyzed the level of hepatic expression of CYP2E1 in a total of 238 healthy human livers obtained from a liver bank.  Donors ranged in age from 8 weeks gestation to 18 years; the postnatal group (N= 167) ranged in age from less than one day to 18 years.  Of the 219 samples for which gender information was available, 136 were male and 83 female.  Of the 201 samples for which ethnic information was available, 93 were Northern European-American, 88 were African-American, and 20 were Hispanic American.

Johnsrud et al. (2003) used analysis of variance techniques to focus on the key factors explaining CYP2E1 content.  They found no significant differences in CYP2E1 protein expression by ethnic group but did find significant differences in expression by age group.  Regression tree analysis was used to identify age categories between which there were statistically significant differences in CYP2E1 expression; the postnatal groups they identified were 0-30 days, 31-90 days, and 91 days to 18 years.  Table 4-1 shows the GM and GSD for the oldest age group in the study (91 days to 18 years).[1]  For purposes of comparison with the other studies in the table that focused primarily on adults, the values in parentheses show the GM and GSD for the 10 to 18 year old group. Although Johnsrud et al. (2003) found no statistical basis for breaking the 91 day to 18 year age group into subgroups, we found that variability in CYP2E1 protein expression was greater (GSD=1.6) for the full age range than for the 10-18 year old group (GSD=1.35).  This difference may reflect a progression toward more stability in mature levels of CYP2E1 protein expression.

Lipscomb et al., (2003) provide the most directly relevant estimates of variability of CYP2E1 activity for use in PBTK modeling.  They derived estimates of variability in Vmax for CYP2E1 mediated oxidation of trichloroethylene (TCE) per unit liver mass from several sets of in vitro data.   They first determined the CYP2E1 content per unit microsomal protein (pmol CYP2E1/MSP) from a total of 60 adult human livers using enzyme-linked immunosorbent assay (ELISA) following the method of Snawder and Lipscomb (2000).  Of these 60 livers, 40 had been analyzed by Snawder and Lipscomb (2000) previously and included adults aged 22-65 (mean 42 years), 23 males and 17 females, were mostly Caucasian, and reported variable alcohol use.[2]  The additional 20 donors ranged in age from 19-77 years (mean 44 years), included 11 males and 9 females, were primarily Caucasian with a few Black and Hispanic individuals, and reported varying degrees of alcohol use.  The latter set of 20 donor livers were also used to measure the CYP2E1 content of intact liver (pmol CYP2E1/gram liver) using ELISA.  With the data on CYP2E1 content per unit microsomal protein measured previously (pmol CYP2E1/MSP),  the CYP2E1 content of intact liver data were used to derive estimates of the content of microsomal protein per gram of intact liver (mg MSP/gram intact liver).   A third  set of data, measured from 15 adult liver samples,  was used to characterize a distribution for the rate of TCE oxidized per unit microsomal protein (pmol TCE/min/mg MSP) (Lipscomb et al., 1997; Snawder and Lipscomb, 2000).

These three data sets were combined to characterize the distribution of Vmax in units (pmol TCE oxidized/min/gram liver) suitable for pharmacokinetic modeling of TCE metabolism in vivo.  Lipscomb et al. (2003) found Vmax for TCE oxidation to be well-characterized by a log-normal distribution with a geometric standard deviation of about 1.73.  This corresponds to about a 6-fold factor between the 95^th and 5^th percentile.

Dorne et al. (2004) provide estimates of variability in CYP2E1 and other metabolizing systems based on a meta-analysis of in vivo data obtained from a literature search up through October 2002.  Probe substrates for each metabolic route were selected on basis that their oral absorption was high (>90%) and that 60-100% of oral dose was excreted as metabolite of that pathway.  Chlorozoxazone and trimethadione were the two probe substrates chosen to characterize CYP2E1 activity.  Dorne et al. (2004) summarized and compared data from studies involving healthy adults, the elderly, and patients with liver or renal disease in order to evaluate the potential for susceptible subpopulations.  Individuals aged 16 to 70 years were classified as adults.  Those greater than 70 years old were classified as elderly.

For CYP2E1, the authors obtained kinetic in terms of plasma clearance and Cmax, in plasma from several studies.  They combined the data for particular kinetic parameters, compounds and subgroups of the population using a weighted mean method described in an earlier study (Dorne et al., 2001). 

The GSD estimates were similar in magnitude for all the CYP2E1 kinetic parameters described: GSDs for measures of clearance ranged from 1.2-1.4 while GSDs for measures of Cmax ranged from 1.1-1.4 (Dorne et al., 2001).   For the acrylamide analysis, we are interested in the two estimates of clearance adjusted for body weight which had GSDs of 1.3 and 1.2 for the substrates chlorzoxazone and trimethadione, respectively (Table 4-1). 

The estimates of variability in CYP2E1 expression and activity shown in Table 4-1 are generally consistent despite the different types of studies and methodologies.   The Lipscomb et al. (2003) GSD is the largest and reflects the combination of several data sets needed to scale from in vitro to in vivo to estimate Vmax.  In vivo GSDs may be smaller because of compensating factors that would tend to offset inter-individual variability in 2E1 content of liver – e.g., blood flow limitations and other metabolic pathways that can contribute to clearance of substrate (note that Dorne et al. (2004) indicate that only 60-100% of substrate is cleared via the ascribed pathway).  For the purpose of the PBTK modeling in this study, we are most concerned with the in vitro data because it is most direct measurement of the system we are modeling and the other in vivo factors such as blood flow limitation will be automatically accounted for in the model developed for acrylamide.

---

[1] The data set from this study was generously supplied to us by the authors.

[2] Snawder and Lipscomb, 2000 reported no significant differences in the fractional composition of cytochrome P450 with respect to CYP2E1 in human hepatic microsomes.  They also reported no significant differences in CYP2E1 specific activity between drinkers and non-drinkers.