*3.1. Clinical Study*

A total of 23 subjects, aged 20–34 years, were enrolled in the clinical study. The average plasma concentration–time profiles for tramadol and M1 are shown in Figure 1. The allele (phenotype) frequencies for *CYP2D6* were: Wild-type (EM, 14 subjects), \*10/\*10 (IM, 8 subjects), and \*5/\*5 (PM, 1 subject).

**Figure 1.** *Cont.*

**Figure 1.** The average plasma concentration–time profiles after five times oral administration (τ = 12 h) of 100 mg tramadol for (**a**) tramadol and (**b**) *O*-desmethyltramadol (M1). Solid blue line, average for all subjects (*n* = 23); solid black line, *CYP2D6* wild-type subjects (*n* = 14); short dashed black line, *CYP2D6* \*5/\*5 subject (*n* = 1); long dashed black line, *CYP2D6* \*10/\*10 (*n* = 8).

#### *3.2. Metabolism Assay for O-Desmethyltramadol (M1)*

We conducted metabolic stability study of M1 and confirmed that M1 was metabolized mainly by CYPs and partially by UGTs. In control sample without NADPH or HLM, more than 96% of tramadol and M1 remained during the incubation time, indicating that the disappearance of tramadol and M1 were mainly caused by CYP enzymes. Metabolic stability of M1 by CYP was assessed using results of the disappearance test for M1. The slope of linear regression was calculated and intrinsic clearance (CLint) of the drug in the in vitro microsome system was calculated:

$$\text{CL}\_{\text{int,mic}} \text{ (}\mu\text{L/min/mg protein)} = \mathbf{k} \times \mathbf{V}\_{\text{incubation}} \text{/C}\_{\text{incubation}} \tag{1}$$

where Vincubation is incubation volume, Cincubation is concentration of microsomal protein.

To apply the elimination profile to *O*-desmethyltramadol (M1), an HLM assay was performed. The M1 disappearance test showed an HLM intrinsic clearance (CLint,HLM) of 52.95 μL/min/mg protein (Figure 2). CLint,HLM for M1 was applied to the M1 PBPK model.

**Figure 2.** The plot of remaining rate of *O*-desmethyltramadol (M1) after incubation with human liver microsoms (HLM). Each point (obtained by duplicate measurements) represents the mean value. The intrinsic clearance by HLM (CLint,mic) was calculated as 52.92 μL/min/mg protein.

### *3.3. PAMPA Results*

Results of the PAMPA assay for 12 tramadol samples diluted to 15 μM showed permeability ranges from 9.14 × 10−<sup>6</sup> cm/s to 11.5 × 10−<sup>6</sup> cm/s. The mean ± standard deviation permeability was calculated as 10.4 × 10−<sup>6</sup> ± 0.056 cm/s. The calculated mean PAMPA permeability was applied to the tramadol PBPK model.

#### *3.4. Development of the PBPK Model for Tramadol and M1*

The input parameters for tramadol and M1 and demographic characteristics for virtual population in the PBPK model are presented in Tables 2 and 3. Data for the healthy Korean population were obtained from the Certara repository. Ten virtual trials, including 100 virtual subjects in each virtual trial (total 1000 subjects), were performed for tramadol and M1. In the tramadol model, most of the observation profiles were within the 5th and 95th percentile range, and the predicted mean tramadol concentration in plasma was similar to the observed profile (Figure 3a). The geometric mean ratios of Cmax,ss and AUClast,ss for tramadol were 0.79 and 1.04, respectively (Table 4). Most observed concentration–time profiles were included in the 5th and 95th percentiles of the predicted concentration–time profiles. In addition, the mean predicted plasma tramadol concentration was well fitted to the observed tramadol concentration. In the concentration–time profiles for M1, most of the observations were also within the 5th and 95th percentile range of the predicted profile (Figure 3b). The range of 90% CI for Cmax,ss and AUClast,ss were included in the range of 0.7–1.43 (30% range of geometric mean ratio); however, the geometric mean ratio was predicted to be relatively low (Table 4). Both tramadol and M1 were predominantly distributed in the liver. The second most common distribution sites were the spleen (tramadol) and heart (M1; Table 5).


**Table 2.** Input parameters for tramadol and *O*-desmethyltramadol (M1) in the physiologically-based pharmacokinetic (PBPK) model.

CLint: intrinsic clearance; CLR: renal clearance; CYP: cytochrome P450 superfamily; fup: unbound fraction in plasma; HLM: human liver microsomes; Kp: plasma-tissue partition coefficient; PAMPA: parallel artificial membrane permeability assay; Papp: apparent permeability; Vss: volume of distribution in steady-state; WOMC: whole organ metabolic clearance, n/a: not applicable. \* Metabolite model does not take account of absorption.

**Table 3.** The demographic characteristics of the participated subjects for virtual Korean population (*n* = 1000).


**Figure 3.** The observed (each symbol, *n* = 23) and simulated mean (solid dark green line) plasma concentration–time profiles after administration of 100 mg tramadol extended-release (ER) tablet twice daily (total five times) for (**a**) tramadol, and (**b**) *O*-desmethyltramadol (M1); blue dashed line represents 5th and 95th percentiles.



AUClast: area under the curve from 48 h to 120 h at steady-state; CI: confidence interval; Cmax,ss: maximum drug concentration in plasma at steady-state.


**Table 5.** Maximum simulated concentrations at steady-state for tramadol and *O*-desmethyltramadol (M1) in each organ (Cmax,ss) after oral administration of 100 mg tramadol ER tablet twice daily (five times in total).

*3.5. Prediction of Changes in Concentration–Time Profiles for Tramadol and M1 According to CYP2D6 Genotype and Dosage*

To investigate the e ffect of CYP2D6 genotype and dosage on PK profiles, simulations were performed for the administration of 100 and 200 mg of tramadol every 12 h (total 5 times). The tramadol/M1 concentration–time profiles were captured from the pre-dose (0 h) to 120 h. The di fferences on PK profiles according to CYP2D6 genotypes were assessed in the general Korean population in CYP2D6 groups: PM, IM, EM, and UM. As a result, plasma concentration–time profiles for tramadol were within the therapeutic range in all groups after administration of 100 mg tramadol ER. Predicted plasma M1 concentrations were very low in the PM group (mean Cmax,ss 0.643 ng/mL) compared to the CYP2D6 IM, EM, and UM groups (mean values 40.93, 83.80, and 126.8 ng/mL, respectively).

The plasma concentration–time profiles for tramadol and M1, and changes in PK parameters, in the various CYP2D6 genotype groups following oral administration of 100 and 200 mg tramadol ER tablet twice daily (total five times) are shown in Figure 4 and Table 6 (the plasma concentration–time profiles for each CYP2D6 phenotype after administration of 100 and 200 mg of tramadol were presented in Supplementary Materials Figures S1 and S2). Following tramadol 100 mg administrations, the Cmax,ss of tramadol in CYP2D6 PMs reached to toxic range. For CYP2D6 UMs, the Cmax,ss of M1 exceeded the therapeutic margin (Supplementary Materials Figure S1). Following tramadol 200 mg administrations, the Cmax,ss of tramadol were reached to the toxic range in all CYP2D6 metabolizer groups. For M1, the Cmax,ss exceeded the therapeutic margin in the CYP2D6 IMs, EMs, and UMs (Figure S2). In Table 7, observed and predicted Cmax,ss and AUClast,ss values, and predicted/observed geometric mean ratios are presented. The CYP2D6 UM group was excluded from this table because UM subjects did not exist in the clinical study. In the CYP2D6 EM and IM groups, the predicted/observed geometric mean ratios for Cmax,ss and AUClast,ss for tramadol satisfied the acceptance criteria (0.7–1.43); however, the tramadol AUClast,ss ratio for the CYP2D6 PM group was overestimated at 1.95. The prediction results for M1 showed that AUClast,ss satisfied the acceptance criteria in the CYP2D6 EM group; however, Cmax,ss and AUClast,ss values were underestimated in both the CYP2D6 IM and PM groups, where the predicted values were much lower than observed values.


**Table 6.** Predicted geometric mean Cmax,ss and AUClast,ss values for tramadol and *O*-desmethyltramadol (M1) following oral administration of 100 and 200 mg tramadol ER tablet twice daily (total five times) in various CYP2D6 metabolizer groups. AUClast,ss: area under the curve from 48 h to 120 h at steady-state; Cmax,ss: maximum drug concentration in plasma at steady-state; EM: extensive metabolizer; IM: intermediate metabolizer; PM: poor metabolizer; UM: ultra-rapid metabolizer. \* Above the therapeutic range for tramadol (>800 ng/mL); \*\* in toxic range for tramadol (1000–2000 ng/mL); # above maximum therapeutic range for M1 (>200 ng/mL).

**Table 7.** Predicted and observed geometric mean PK parameters for tramadol and M1 according to CYP2D6 genotype following oral administration of 100 mg tramadol ER tablet twice daily (total five times).


AUClast,ss: area under the curve from 48 h to 120 h at steady-state; CI: confidence interval; Cmax,ss: maximum drug concentration in plasma at steady-state; EM: extensive metabolizer; IM: intermediate metabolizer; PM: poor metabolizer; Ratio = predicted/observed. Since the observed data for the PM group are for 1 subject, the CI value cannot be obtained.

**Figure 4.** The predicted mean concentration–time profiles after administration of 100 mg and 200 mg tramadol ER tablet twice daily (total five times) for tramadol (**a** and **c**), and *O*-desmethyltramadol (**b** and **d**), respectively. Gray areas in (**a**) and (**c**) represent the therapeutic concentration range (100–800 ng/mL); checked gray area in (**c**) represents the toxic range (above 1000 ng/mL) for tramadol; and the gray area in (**b**) and (**d**) represents the maximum therapeutic range for M1 (up to 200 ng/mL).
