Abstract
We predicted the drug–drug interaction (DDI) potential of siponimod in presence of cytochrome P450 (CYP)2C9/CYP3A4 inhibitors/inducers in subjects with different CYP2C9 genotypes by physiologically‐based pharmacokinetic (PK) modeling. The model was established using in vitro and clinical PK data and verified by adequately predicting siponimod PK when coadministered with rifampin. With strong and moderate CYP3A4 inhibitors, an increased DDI risk for siponimod was predicted for CYP2C9*3/*3 genotype vs. other genotypes area under the curve ratio (AUCR): 3.03–4.20 vs. ≤ 1.49 for strong; 2.42 vs. 1.14–1.30 for moderate. AUCRs increased with moderate (2.13–2.49) and weak (1.12–1.42) CYP3A4/CYP2C9 inhibitors to the same extent for all genotypes. With strong CYP3A4/moderate CYP2C9 inducers and moderate CYP3A4 inducers, predicted AUCRs were 0.21–0.32 and 0.35–0.71, respectively. This complementary analysis to the clinical PK‐DDI studies confirmed the relevant influence of CYP2C9 polymorphism on the DDI behavior of siponimod and represented the basis for the DDI labeling recommendations.
Study Highlights.
WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?
☑ The clinical relevance of the different cytochrome P450 (CYP)2C9 genotypes is well described. However, there are few examples of dedicated drug–drug interaction (DDI)–physiologically‐based pharmacokinetic (PBPK) simulations published that support the DDI labeling recommendations for the different genotypes.
WHAT QUESTION DID THIS STUDY ADDRESS?
☑ Do CYP3A4/CYP2C9 perpetrators impact the exposure of siponimod in the various CYP2C9 genotypes, and to what extent?
WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE?
☑ PBPK simulations are valuable tools to translate clinical observations to untested perpetrators and/or subpopulations with different genotypes.
HOW MIGHT THIS CHANGE CLINICAL PHARMACOLOGY OR TRANSLATIONAL SCIENCE?
☑ For drugs with polymorphic enzymes as major clearance factors, PBPK simulations can be utilized to give a genotype‐specific dose and dose adjustment recommendation, which is a step in personalized medicine without performing DDI studies for all relevant genotypes.
Siponimod is a potent, oral, selective sphingosine 1‐phosphate receptor subtypes 1 and 5 modulator.1 In a recently published phase III study, siponimod 2 mg (s.d.) reduced the risk of disability progression in patients with secondary progressive multiple sclerosis.2 Siponimod has recently been approved by the US Food and Drug Administration (FDA) for the treatment of relapsing forms of multiple sclerosis, including clinically isolated syndrome, relapsing‐remitting disease, and active secondary progressive disease, in adults.3
Siponimod exhibits time‐independent pharmacokinetics (PKs). The rate and extent of systemic exposure increase in a dose‐proportional manner after single (0.1–75 mg; Table 1) and multiple (0.3–20 mg, Table 1) doses of siponimod.1 The mean elimination half‐life (t1/2) of siponimod is ~ 30 hours (range: 22–38 hours) with a washout period of 6 days.1 Siponimod is eliminated from systemic circulation primarily through hepatic oxidative metabolism with subsequent fecal/biliary excretion (Figure 1).4 Cytochrome P450 (CYP)2C9 is the major hepatic enzyme involved in the metabolism (79.3%) of siponimod, followed by CYP3A4 (18.5%).4, 5 Siponimod did not exhibit clinically relevant inhibition potential on CYP enzymes and transporters based on static‐mechanistic model calculations,6 and it was not identified as transporter substrate in vitro (Table S1).
Table 1.
Observed and simulated siponimod PK parameters
| Dose (mg) and form | N | Cmax (ng/mL) | AUCinf (ng·hour/mL) | Tmax (h) | t1/2 (h) | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| SimCYP | Observed | Predicted | Observed | Predicted | Observed | Predicted | Observed | Predicted | Observed | |
| Summary of siponimod PK after single intravenous and oral administration of 0.25 mga | ||||||||||
| i.v. | 100 | 15 | 3.02 (13; 2.95–3.09) | 3.22 (19; 1.88–3.81) | 82.1 (37; 76.9–87.3) | 80.1 (25.6; 47.7–125) | 3.00 (2.95–3.00) | 2.92 (2.92–2.92) | 25.5 (36; 23.8–27.1) | 27.4 (22.5; 20.1–47.7) |
| p.o. | 100 | 15 | 1.88 (14; 1.83–1.92) | 1.71 (24; 0.920–2.30) | 69.7 (38; 65.1–74.3) | 67.4 (26; 39.8–105) | 3.25 (2.35–5.05) | 8.00 (4.00–8.03) | 25.5 (36; 23.8–27.1) | 26.7 (22; 17.2–43.8) |
| Summary of observed and simulated siponimod PK after single oral administrationb | ||||||||||
| 0.1 Solution | 340 | 0.78 (20; 0.76–0.79) | 29.1 (48; 27.7–30.5) | — | 3.45 (2.31–5.60) | — | 25.6 (49; 24.4–26.8) | — | ||
| 0.3 Solution | 340 | 8 | 2.33 (20; 2.29–2.37) | 2.26 (6; 2.17–2.34) | 87.2 (48; 83.1–91.4) | 89.3 (20; 79.2–99.4) | 3.45 (2.30–5.60) | 5 (4.00–8.00) | 25.6 (49; 24.4–26.8) | 33.21 (28) |
| 1 Solution | 340 | 8 | 7.76 (20; 7.62–7.91) | 7.43 (42; 5.37–9.50) | 291 (48; 277–305) | 349 (28; 292–406) | 3.45 (2.30–5.60) | 5 (4.00–15.7) | 25.6 (49; 24.4–26.8) | 45.7 (26) |
| 2.5 Solution | 340 | 6 | 19.4 (20; 19.1–19.8) | 21.9 (35; 16.7–27.0) | 727 (48; 693–761) | 766 (36; 598–935) | 3.45 (2.30–5.60) | 4 (3.00–8.00) | 25.6 (49; 24.4–26.8) | 27.0 (22) |
| 2.5 Capsule | 340 | 7 | 19.4 (20; 19.1–19.8) | 19.3 (19; 17.1–21.5) | 727 (48; 693–761) | 745 (25; 632–858) | 3.45 (2.30–5.60) | 6 (4.00–8.00) | 25.6 (49; 24.4–26.8) | 29.3 (17) |
| 5 Capsule | 340 | 8 | 38.8 (20; 38.1–39.5) | 38.5 (21; 33.5–41.5) | 1,454 (48; 1,385–1,523) | 1,260 (20; 1,149–1,384) | 3.45 (2.30–5.60) | 3 (2.00–8.00) | 25.6 (49; 24.4–26.8) | 31.3 (19) |
| 10 Capsule | 340 | 8 | 77.7 (20; 76.2–79.1) | 77.3 (25; 64.8–89.7) | 2,908 (48; 2,770–3,045) | 2,710 (14; 2,507–2,915) | 3.45 (2.30–5.60) | 5 (3.82–16.0) | 25.6 (49; 24.4–26.8) | 32.0 (17) |
| 17.5 Capsule | 340 | 8 | 136 (20; 133–138) | 111 (32; 93.7–128.4) | 5,088 (48; 4,847–5,329) | 4,230 (25; 3,666–4,792) | 3.45 (2.30–5.60) | 4 (2.00–8.00) | 25.6 (49; 24.4–26.8) | 42.3 (38) |
| 25 Capsule | 340 | 8 | 194 (20; 191–198) | 217 (29; 180–254) | 7,269 (48; 6,925–7,613) | 8,140 (24; 6,987–9,302) | 3.45 (2.30–5.60) | 3.50 (1.50–12.0) | 25.6 (49; 24.4–26.8) | 47.7 (20) |
| 75 Capsule | 340 | 8 | 582 (20; 572–593) | 491 (51; 346–637) | 21,807 (48; 20,774–22,840) | 18,600 (51: 11,831–25,397) | 3.45 (2.30–5.60) | 6 (2.00–24.00) | 25.6 (49; 24.4–26.8) | 56.7 (12) |
| Dose (mg) and form | N | Cmax (ng/mL) | AUC(0–24 h) (ng·hour/mL) | Cmin (ng/mL) | ||||
|---|---|---|---|---|---|---|---|---|
| SimCYP | Observed | Predicted | Observed | Predicted | Observed | Predicted | Observed | |
| Summary of observed and simulated siponimod PK after multiple oral administration at day 28c | ||||||||
| 0.3 Solution | 340 | 6 | 4.00 (38; 4.63–4.97) | 5.13 (16; 4.76–5.86) | 86.2 (48; 82.2–90.3) | 97.9 (19) | 2.62 (69; 2.46–2.78) | 2.83 (27) |
| 1 Solution | 340 | 6 | 16.0 (38; 15.4–16.6) | 14.9 (10; 14.0–15.9) | 287 (48; 274–301) | 282 (16) | 8.74 (69; 8.20–9.27) | 7.74 (20) |
| 2.5 Capsule | 340 | 7 | 40.0 (38; 38.6–41.5) | 38.3 (37; 29.1–47.4) | 719 (48; 685–752) | 692 (45) | 21.9 (69; 20.5–23.2) | 15.2 (120) |
| 10 Capsule | 340 | 9 | 160 (38; 154–166) | 147 (24; 126–168) | 2,874 (48; 2,740–3,009) | 2,580 (24) | 87.4 (69; 82.0–92.7) | 70.6 (29) |
| 20 Capsule | 340 | 9 | 320 (38; 309–332) | 359 (17; 325–394) | 5,749 (48; 5,480–6,017) | 6,370 (23) | 175 (69; 164–186) | 166 (88) |
Pharmacokinetic data are presented as geometric means with percentage of coefficient of variation geometric mean and/or 90% confidence interval range in parentheses except for Tmax (median with range).
AUC0–24 h, area under the curve from time 0–24 hours; AUCinf, area under the curve from time zero to infinity; Cmax, maximum plasma concentration; Cmin, minimum plasma concentration;h, hour; PK, pharmacokinetic; t1/2, terminal elimination half‐life; Tmax, time to reach maximum systemic concentration.
aObserved values were taken from Novartis data on file. bObserved values were taken from single ascending dose study. cObserved values were taken from multiple ascending dose study.1
Figure 1.
Drug disposition pathways for siponimod. aOral absorption of siponimod was estimated to be ~ 91% based siponimod excreted to feces in the human absorption, distribution, metabolism, and excretion study,4 assuming that the drug and metabolites are stable against intestinal bacterial enzymes. bBased on SimCYP simulations. cFractional contribution of CYP enzymes based on phenotyping data.5 Values represent normalized values scaled to 100%. CYP, cytochrome P450.
CYP2C9 is a polymorphic enzyme with >50 single nucleotide polymorphisms (SNPs). Of these, the CYP2C9*2 and CYP2C9*3 SNPs are the primary causes of clinically relevant reductions in enzyme activity.5 These SNPs lead to six most prominent CYP2C9 genotypes conferring to three functionally different phenotypes: extensive metabolizers, intermediate metabolizers, and poor metabolizers (PMs).5, 7, 8, 9, 10 The prevalence of the six clinically relevant CYP2C9 genotypes in the white population ranges from 62–65% for CYP2C9*1/*1, 20–24% for CYP2C9*1/*2, 9–12% for CYP2C9*1/*3, 1–2% for CYP2C9*2/*2, 1.4–1.7% for CYP2C9*2/*3, and 0.3–0.4% for CYP2C9*3/*3.5, 9, 11 CYP2C9 polymorphism has shown a significant impact on siponimod metabolism.5 Regarding siponimod clearance, population PK analyses (PopPK; Table S2) and a CYP2C9 pharmacogenetic study5 indicated that subjects with CYP2C9*1/*1 and CYP2C9*1/*2 genotypes behave as extensive metabolizers, CYP2C9*2/*2 and CYP2C9*1/*3 genotypes behave as intermediate metabolizers, and CYP2C9*2/*3 and CYP2C9*3/*3 genotypes as PMs. Compared with subjects with the CYP2C9*1/*1 genotype, those with the CYP2C9*2/*2, *1/*3, *2/*3, and *3/*3 genotypes have 20%, 35–38%, 45–48%, and 74% lower total body clearance of the drug from plasma values, respectively (Novartis data on file, Table S2).5 The recommended maintenance dose of siponimod as approved by the FDA is 2 mg taken orally once daily. In patients with a CYP2C9*1/*3 or *2/*3 genotype, a lower maintenance dose of 1 mg taken orally once daily is recommended, whereas siponimod is contraindicated in patients homozygous for CYP2C9*3 (i.e., CYP2C9*3/*3 genotype), because of the substantially elevated siponimod plasma levels.3
In subjects with the CYP2C9*1/*1 genotype, upon coadministration with fluconazole (a moderate CYP3A4/CYP2C9 inhibitor), siponimod area under the curve (AUC) increased approximately two fold and the maximum plasma concentration (Cmax) increased by ~ 10%.5 Coadministration of rifampin (a strong CYP3A4/moderate CYP2C9 inducer) in CYP2C9*1/*1 subjects reduced siponimod Cmax by 45% and AUC by 57%.12 Although an increased siponimod exposure was expected, coadministration of itraconazole (a strong CYP3A4 inhibitor) in CYP2C9*1/*2 and CYP2C9*1/*3 subjects led to a decrease in siponimod AUC by 9–10% and 24%, respectively (unpublished data13).
It is expected that the CYP2C9 genotype influences the effects of CYP3A4 and CYP2C9 inhibitors and inducers on siponimod PK. Physiologically‐based pharmacokinetic (PBPK) modeling can help in predicting in vivo PK changes and drug–drug interactions (DDIs) caused by CYP perpetrators for the six different CYP2C9 genotypes while waiving a large number of additional clinical DDI studies.14, 15 This simulation study aimed to establish a middle‐out siponimod PBPK model reflecting the in vitro fractional enzyme contributions confirmed by clinical DDI data. Subsequently, the DDI potential of siponimod as a substrate in the presence of typical CYP2C9 and CYP3A4 perpetrators was evaluated for the six clinically relevant CYP2C9 genotypes in the white population.
Results
Siponimod model verification
Systemic exposure
The simulated AUC from time 0 to infinity (AUCinf) or from time 0–24 hours (AUC0–24 h), Cmax, and minimum plasma concentration (Cmin) after single i.v. (0.25 mg), single oral (s.d.; 0.1–75 mg), and multiple oral (0.3–20 mg q.d. for 28 days1) administration of siponimod were compared with the observed values (Table 1). Good model predictability of AUC, Cmax, and Cmin was achieved. The calculated percentage of residuals (=[predicted−observed]/observed) were within 30% (Figure S1). The residual plots for the PK parameters were well distributed, indicating that the established siponimod PBPK model was not biased by any arbitrary parameter optimization process. Linear PK observed clinically for siponimod single doses up to 75 mg (p.o., s.d.; Figure 2 a,b) and up to 20 mg for the multiple doses (p.o., q.d. for 28 days; Figure S2) were reflected by the siponimod model.
Figure 2.
SimCYP simulation for siponimod pharmacokinetics following single dose oral administration. (a) Day 1, 0.1 mg. (b) Day 1, 75 mg for the CYP2C9*1/*1 genotype (n = 11 at 0.1 mg and n = 8 at 75 mg) and 0.1 mg (c, d) for the CYP2C9*2/*3 and CYP2C9*3/*3 genotypes (n = 6 [*2/*3]; n = 66 [*3/*3]). The black and gray lines represent simulated mean time‐plasma concentration profiles and the 5th/95th percentile of the total virtual population, respectively (n = 340). Full circles are the measured values after oral administration. CYP, cytochrome P450.
A sensitivity analysis within the absorption rate constant (Ka) range of 0.1–2 showed a moderate change in siponimod (0.1 mg p.o., s.d. at day 1) Cmax from 0.4 to 1.1 ng/mL. Therefore, small changes in Ka would not result in a significant Cmax change. Time to reach the highest systemic drug concentration (Tmax) of siponimod was sensitive to Ka changes below Ka values of 0.5 per hour. A sensitivity analysis within the fraction unbound in the enterocytes (fugut) range from 0.0002–0.01 revealed little impact on siponimod (0.1 mg p.o., single dose at day 1) Cmax with changes from 0.69 to 0.65 ng/mL. Thus, any value within this range could be used. The sensitivity analysis results for the above siponimod model PK parameters are presented in Figure S3.
Impact of different CYP2C9 genotypes
The simulation trial parameters used to demonstrate the predictability of the siponimod model in different CYP2C9 genotypes (*1/*1, *2/*3, and *3/*3) were similar to the actual clinical study5 conditions (e.g., female proportionality and age ranges in populations; Table S3). The simulated siponimod PK after single oral administration to subjects of these three different genotypes was comparable to the observed values (Table 2 , Figure 2 c,d), verifying the PK and the fraction metabolized via CYP2C9 (fm,CYP2C9).
Table 2.
Observed and simulated siponimod PK parameters after single oral administration to subpopulations carrying the different CYP2C9 genotypes
| CYP2C9 genotype | N | Cmax (ng/mL) | AUCinf (ng·hour/mL) | t1/2 (h) | ||||
|---|---|---|---|---|---|---|---|---|
| Predicted | Observed | Predicted | Observed | Predicted | Observed | Predicted | Observed | |
| *1/*1 | 100 | 12 | 1.90 (14) | 2.03 (17.2) | 70.6 (37) | 70.5 (21.2) | 25.5 (37) | 28.1 (18.5) |
| *1/*2 | 100 | NA | 1.92 (14) | NA | 76.2 (37) | NA | 27.4 (40) | NA |
| *1/*3 | 100 | NA | 1.99 (14) | NA | 117 (34) | NA | 41.1 (36) | NA |
| *2/*2 | 100 | NA | 1.97 (15) | NA | 98.7 (34) | NA | 34.9 (36) | NA |
| *2/*3 | 100 | 6 | 2.02 (14) | 2.45 (13.1) | 142 (39) | 144 (15.7) | 49.2 (42) | 50.9 (31.7) |
| *3/*3 | 100 | 6 | 2.10 (15) | 2.35 (29.0) | 348 (45) | 271 (22.4) | 118 (49) | 126 (12.7) |
AUCinf, area under the curve from time zero to infinity; Cmax, maximum plasma concentration; CYP, cytochrome P450; h, hour; NA, not applicable; PK, pharmacokinetic; t1/2, half‐life.
Predicted vs. observed siponimod DDI potential with CYP3A4 and CYP2C9 perpetrators
Moderate CYP3A4/CYP2C9 inhibitor (fluconazole)
The geometric mean AUCinf and Cmax ratios of siponimod (4 mg p.o., s.d. on day 3) following oral administration of fluconazole (200 mg b.i.d. on day 1 and 200 mg q.d. on days 2–19) was predicted to be 2.15 and 1.07, respectively, for the CYP2C9*1/*1 genotype (Table 3). The predicted ratios were in line with the corresponding clinical observation (AUC ratio (AUCR): 1.98; Cmax ratio: 1.10).5 For siponimod and fluconazole at steady state, the predicted AUCRs ranged from 2.18 for CYP2C9*1/*1 to 2.09 for CYP2C9*2/*3 (Table 3). An increase in AUCR was predicted for the CYP2C9*3/*3 genotype (2.49).
Table 3.
Predicted siponimod multiple dose AUC and Cmax ratios in the presence of CYP3A4 and CYP3A4/CYP2C9 inhibitors or inducers in the six clinical relevant CYP2C9 genotypes (simulated single dose DDI ratios and observed DDI ratios in bold)
| CYP2C9 genotype | CYP inhibitor | CYP inducer | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Strong 3A4 inhibitor | Moderate 3A4 inhibitor | Moderate 3A4/2C9 inhibitor | Weak 3A4/2C9 inhibitor | Strong 3A4/moderate 2C9 inducer | Moderate 3A4 inducer | |||||||||
| Itraconazole | Ketoconazole | Erythromycin | Fluconazole | Fluvoxamine | Rifampin | Efavirenz | ||||||||
| AUCi/AUC | Cmaxi/Cmax | AUCi/AUC | Cmaxi/Cmax | AUCi/AUC | Cmaxi/Cmax | AUCi/AUC | Cmaxi/Cmax | AUCi/AUC | Cmaxi/Cmax | AUCi/AUC | Cmaxi/Cmax | AUCi/AUC | Cmaxi/Cmax | |
| *1/*1 | 1.17 (1.16–1.18) | 1.12 (1.12–1.13) | 1.24 (1.23–1.26) | 1.18 (1.17–1.19) | 1.14 (1.13–1.16) | 1.11 (1.10–1.12) | 2.18 (2.14–2.22) Simulated single dose: 2.15 Observed: 1.98 | 1.88 (1.85–1.92) Simulated single dose: 1.07 Observed: 1.10 | 1.42 (1.39–1.46) | 1.33 (1.30–1.35) | 0.32 (0.31–0.33) Observed: 0.43 | 0.50 (0.49–0.51) Observed: 0.55 | 0.71 (0.69–0.73) | 0.79 (0.77–0.80) |
| *1/*2 | 1.18 (1.17–1.19) Simulated single dose: 1.18 Observed: 0.90 | 1.14 (1.13–1.15) Simulated single dose: 1.02 Observed: 1.01 | 1.26 (1.25–1.28) | 1.20 (1.19–1.21) | 1.16 (1.15–1.17) | 1.12 (1.11–1.13) | 2.18 (2.14–2.22) | 1.89 (1.86–1.92) | 1.41 (1.38–1.45) | 1.32 (1.30–1.35) | 0.32 (0.31–0.33) | 0.49 (0.48–0.50) | 0.70 (0.68–0.71) | 0.77 (0.76–0.79) |
| *1/*3 | 1.29 (1.28–1.31) Simulated single dose: 1.30 Observed: 0.76 | 1.24 (1.23–1.26) Simulated single dose: 1.02 Observed: 0.94 | 1.41 (1.38–1.44) | 1.34 (1.31–1.36) | 1.25 (1.23–1.27) | 1.21 (1.19–1.22) | 2.13 (2.09–2.16) | 1.93 (1.90–1.96) | 1.35 (1.32–1.38) | 1.29 (1.27–1.32) | 0.30 (0.29–0.31) | 0.43 (0.42–0.44) | 0.61 (0.59–0.63) | 0.69 (0.67–0.71) |
| *2/*2 | 1.24 (1.22–1.25) | 1.19 (1.18–1.21) | 1.34 (1.32–1.36) | 1.27 (1.25–1.29) | 1.21 (1.19–1.22) | 1.16 (1.15–1.18) | 2.16 (2.12–2.20) | 1.92 (1.89–1.95) | 1.38 (1.34–1.41) | 1.31 (1.28–1.33) | 0.31 (0.30–0.32) | 0.46 (0.45–0.47) | 0.65 (0.63–0.67) | 0.73 (0.71–0.74) |
| *2/*3 | 1.35 (1.33–1.37) | 1.30 (1.28–1.31) | 1.49 (1.45–1.52) | 1.41 (1.38–1.44) | 1.30 (1.27–1.33) | 1.25 (1.23–1.27) | 2.09 (2.05–2.12) | 1.91 (1.88–1.94) | 1.31 (1.28–1.34) | 1.27 (1.24–1.29) | 0.29 (0.28–0.30) | 0.41 (0.40–0.42) | 0.58 (0.56–0.60) | 0.65 (0.63–0.67) |
| *3/*3 | 3.03 (2.83–3.22) | 2.89 (2.71–3.08) | 4.20 (3.94–4.46) | 3.97 (3.74–4.19) | 2.42 (2.26–2.58) | 2.32 (2.17–2.47) | 2.49 (2.43–2.55) | 2.39 (2.33–2.45) | 1.12 (1.11–1.13) | 1.11 (1.10–1.12) | 0.21 (0.20–0.22) | 0.27 (0.27–0.29) | 0.35 (0.33–0.38) | 0.41 (0.39–0.43) |
Data presented in geometric mean (90% confidence interval), for a single dose siponimod DDI simulations with fluconazole CYP2C9*1*1 or itraconazole CYP2C9*1*2 and *1*3 the AUC/Cmax ratios were reported as clinical reference data exists for this study design.
AUC, area under the curve; AUCi, area under the curve for inhibition/induction; Cmax, maximum plasma concentration; Cmaxi. maximum plasma concentration for inhibition/induction; CYP, cytochrome P450; DDI, drug–drug interaction.
The sensitivity analysis for the fluconazole CYP2C9 unbound inhibition constant range from 5–30 μM, predicted an AUCR change from 4.17 to 1.84, respectively, suggesting that the predictions of AUC changes for siponimod in the presence of fluconazole were sensitive to CYP2C9 inhibition constant (Ki) changes for the CYP2C9*1/*1 genotype (Figure S3). However, the higher CYP2C9 Ki value has been published and successfully verified by the SimCYP team.16
Strong CYP3A4/moderate CYP2C9 inducer (rifampin)
The steady‐state geometric mean AUC (AUCτ) and Cmax ratios of siponimod (2 mg p.o., q.d. for 12 days) with coadministration of rifampin (600 mg b.i.d. for 12 days) were predicted to be 0.32 and 0.50, respectively (Table 3), which were comparable to the clinical observation (AUCR: 0.43; Cmax ratio: 0.55).12 For siponimod and rifampin at steady state, the predicted AUCRs ranged from 0.32 for CYP2C9*1/*1 to 0.29 for CYP2C9*2/*3; for the CYP2C9*3/*3 genotype, an AUCR of 0.21 was predicted (Table 3). The induction effect of rifampin (600 mg b.i.d. for 12 days) on the PK of siponimod (2 mg p.o., q.d. for 12 days) in the CYP2C9*1/*1 genotype was also predicted using the rifampin model without CYP2C9 induction parameters. The simulated AUCτ (0.59) and Cmax ratios (0.71) showed a less pronounced DDI effect compared with the clinical observations12 indicating that the contribution of CYP2C9 induction is important and cannot be neglected.
When performing a sensitivity analysis, the AUCR of siponimod in the presence of rifampin within a degradation rate constant (Kdeg), CYP3A4 range of 0.005–0.03 per hour changed from 0.294–0.275, suggesting a weak dependency. Similarly, within a Kdeg, CYP2C9 range of 0.005–0.03 per hour, the AUCR changed from 0.293 to 0.252, indicating a weak dependency for the CYP2C9*1/*1 genotype population (Figure S3).
Strong CYP3A4 inhibitor (itraconazole)
The geometric mean AUCinf and Cmax ratios of siponimod (0.25 mg p.o., s.d. on day 5) with coadministration of itraconazole (100 mg p.o., b.i.d. from day 1 to day 17) were predicted to be 1.18 and 1.02, respectively, for the CYP2C9*1/*2 genotype, and 1.29 and 1.02, respectively, for the CYP2C9*1/*3 genotype. These predictions were different from the clinical observation (AUCR: 0.90; Cmax ratio: 1.01 for the CYP2C9*1/*2 genotype; AUCR: 0.76; Cmax ratio: 0.94 for the CYP2C9*1/*3 genotype). The coadministration of itraconazole and siponimod, the predicted AUCRs ranged from 1.17 for CYP2C9*1/*1 to 1.35 for CYP2C9*2/*3 at steady state, for the CYP2C9*3/*3 genotype, an AUCR of 3.03 was predicted (Table 3).
DDI prediction of siponimod as a victim drug
Steady‐state DDI effects of strong, moderate, and weak inhibition of CYP3A4 and/or CYP2C9 and moderate induction of CYP3A4 on the PK of siponimod were simulated for the six different CYP2C9 genotypes. Due to the longer t1/2 of siponimod in subpopulations carrying CYP2C9*3/*3,5 the simulation time was extended to 90 days for the inhibitors, and to 24 days for the inducer efavirenz, maintaining coadministration of perpetrators. The DDI effects of CYP3A4 and CYP2C9 perpetrators on siponimod exposure in subjects of the six clinically relevant genotypes are presented in Table 3.
Effect of inhibitors
Strong CYP3A4 inhibitor (ketoconazole)
In the presence of the strong CYP3A4 inhibitor ketoconazole, predicted siponimod AUCinf inhibition ratios tended to increase with reduced CYP2C9 metabolic activity in the different genotypes. The AUCRs increased from 1.24 in CYP2C9*1/*1 up to 1.49 in CYP2C9*2/*3. The CYP2C9*3/*3 genotype population showed a significantly higher DDI risk, with a predicted AUCR of 4.20 (Table 3).
Moderate CYP3A4 inhibitor (erythromycin)
In the presence of the moderate CYP3A4 inhibitor erythromycin, an increase in predicted siponimod AUCinf inhibition ratios was observed, increasing from 1.14 in CYP2C9*1/*1 to 1.30 in CYP2C9*2/*3. The AUCR was larger for the CYP2C9*3/*3 genotype (2.42) suggesting a higher DDI risk (Table 3).
Weak CYP3A4/2C9 inhibitors (fluvoxamine)
In the presence of the weak CYP3A4/2C9 inhibitor fluvoxamine, the predicted AUCRs ranged between 1.42 in CYP2C9*1/*1 genotype and 1.31 in CYP2C9*2/*3 genotype. A further decrease in the AUCR was predicted in the CYP2C9*3/*3 genotype (1.12) due to the dual inhibition of the metabolic pathways (Table 3).
Effect of inducers
Strong CYP3A4/moderate CYP2C9 inducers
In presence of strong CYP3A4 and moderate CYP2C9 inducer (rifampin), the predicted AUCR ranged between 0.32 (CYP2C9*1/*1) and 0.21 (CYP2C9*3/*3) and the Cmax ratios ranged between 0.5 and 0.27, respectively. Both metabolic pathways of siponimod were impacted, and only a small difference in the induction effect was predicted between CYP2C9 genotypes (Table 3).
Moderate CYP3A4 inducer (efavirenz)
The predicted exposure change in presence of the moderate CYP3A4 inducer efavirenz was comparatively lower than with rifampin. The predicted AUCRs ranged between 0.35 and 0.71 and the Cmax ratios ranged between 0.41 and 0.79 for the CYP2C9 genotypes (Table 3). The lowest DDI ratios were predicted for the CYP2C9*3/*3 genotype (AUCR: 0.35 and Cmax ratio: 0.41).
Siponimod CYP2C9 genotype‐based DDI management
When considering the siponimod maintenance doses recommended by the FDA (2 mg daily in patients with CYP2C9*1/*1, *1/*2, and *2/*2 genotypes and 1 mg daily in patients with *1/*3 and *2/*3 genotypes) and the contraindication for patients with the CYP2C9*3/*3 genotype, the estimated net effects on siponimod exposure in the presence of CYP2C9/CYP3A4 inhibitors or inducers were estimated after multiple administrations (Table 4). The net effect is the simulated inhibition or induction ratio (Table 3) multiplied by the corresponding CYP2C9‐genotype ratio (AUCss,tau of the genotype vs. AUCss,tau of wild type CYP2C9*1/*1 at the FDA recommended dose for the respective genotypes without coadministration of any perpetrator).
Table 4.
Net inhibition or induction effect ratios (inhibition or induction × CYP2C9 genotype exposure ratios)a
| CYP2C9 genotype | CYP inhibitor | CYP inducer | |||||
|---|---|---|---|---|---|---|---|
| Strong 3A4 inhibitor | Moderate 3A4 inhibitor | Moderate 3A4/2C9 inhibitor | Weak 3A4/2C9 inhibitor | Strong 3A4/moderate 2C9 inducer | Moderate 3A4 inducer | ||
| Itraconazole | Ketoconazole | Erythromycin | Fluconazole | Fluvoxamine | Rifampin | Efavirenz | |
| AUCi/AUC | AUCi/AUC | AUCi/AUC | AUCi/AUC | AUCi/AUC | AUCi/AUC | AUCi/AUC | |
| *1/*1 | 1.18 | 1.24 | 1.14 | 2.20 | 1.44 | 0.29 | 0.68 |
| *1/*2 | 1.29 | 1.36 | 1.25 | 2.31 | 1.55 | 0.30 | 0.72 |
| *2/*2 | 1.73 | 1.83 | 1.66 | 2.91 | 1.90 | 0.37 | 0.85 |
| *1/*3 | 1.06 | 1.14 | 1.01 | 1.72 | 1.11 | 0.22 | 0.48 |
| *2/*3 | 1.32 | 1.42 | 1.25 | 1.98 | 1.28 | 0.26 | 0.53 |
AUC, area under the curve; AUCi, area under the curve for inhibition/induction; CYP, cytochrome P450.
aCalculated using drug–drug interaction ratios from Table 3 multiplied by the genotype‐specific exposure increase, and compared with the wild type (CYP2C9*1/*1, Table 1) where simulated exposures at a 2 mg siponimod dose were used for the genotypes CYP2C9*1/*1, *1/*2, and *2/*2, and a dose of 1 mg for CYP2C9*1/*3 and *2/*3. Calculations are based on geometric mean data.
Discussion
The present siponimod PBPK modeling study was conducted as a complementary analysis to the clinical PK DDI studies to gather complete understanding of the DDI potential of siponimod with respect to the six clinically relevant CYP2C9 genotypes. These modeling results were used to guide the development of drug label recommendations. The study evaluated the DDI effect of CYP3A and CYP2C9 enzyme inducers or inhibitors on systemic exposure of siponimod through PBPK simulations in healthy subjects. The linear PK profiles of siponimod up to doses of 75 mg for the single dose and 20 mg for the multiple dose (q.d. for 28 days) were described well by the PBPK model. The systemic exposure at steady state after multiple oral administration was reached after 6 days, confirming validity of the reported t1/2 of 30 hours for CYP2C9*1/*1 genotypes.1 The PK profiles of the two genotypes CYP2C9*2/*3 and CYP2C9*3/*3 were also well described by the established model.
Predicted effects of fluconazole (AUCR: 2.15; Cmax ratio: 1.07) and rifampin (AUCR: 0.32; Cmax ratio: 0.50) on the siponimod PK were comparable to the respective corresponding clinical data (fluconazole/rifampin; geometric mean AUCinfR: 1.98/0.43 and Cmax ratio: 1.10/0.55)5, 12 for the CYP2C9*1/*1 genotype (Table 3), verifying the siponimod PK and fraction metabolized via CYP3A4 (fm,CYP3A4). The fraction metabolized via CYP2C9 (fm,CYP2C9, 0.79) was based on the in vitro phenotyping data,4 and the selected CYP2C9 intrinsic clearance (CLint) values for the different genotypes based on PopPK (Table S2) were verified by the correct PK estimation for the different genotypes (CYP2C9*1/*1, CYP2C9*2/*3, and CYP2C9*3/*3).
With decreased CYP2C9 metabolic activity, CYP3A4 becomes the dominant elimination pathway for the PMs (CYP2C9*3/*3) with an estimated fraction metabolized via CYP2C9 (fm,CYP3A4) of 0.82; this genotype subpopulation is sensitive to strong CYP3A4 inhibitors (Tables 3 and 5). All other genotypes have <35% CYP3A4 metabolism contribution to total clearance (CLtot). In the presence of strong CYP3A4 inhibitors, an increased DDI risk in subjects with CYP2C9*3/*3 was predicted compared with other genotypes (predicted siponimod AUCR: 3.03–4.20 vs. <1.49), demonstrating the sensitivity, when both pathways CYP2C9 and CYP3A4 are less functional or inhibited.
Table 5.
SimCYP input parameters for the siponimod model
| Input parameters | Siponimod | Fluconazolea |
|---|---|---|
| 1. Physicochemical and binding properties | ||
| Molecular weight (g/mol) | 516.6 | 306.3 |
| Log P: Octanol‐water partition | 1.8 | 0.20 |
| Compound type | Ampholyte | Monoprotic base |
| pKa | 3.1/8.1 | 1.76 |
| Blood to plasma drug concentration ratio | 0.765 | 1 |
| Fraction unbound in plasma | 0.0002 | 0.89 |
| Main plasma binding protein | HSA (assumption) | HSA |
| 2. Absorptionb | ||
| Absorption model | First order | |
| Fraction available from dosage form | 0.91c | 0.988 |
| CV fa (%) | 8.6d | 30 |
| Absorption rate constant (1/hour) | 0.687d | 1.863 |
| CV Ka (%) | 7.8 | 30 |
| Lag time (h) | 1.5 (optimized) | |
| Unbound fraction in enterocytes | 0.0002e | 0.89 |
| Nominal flow in gut model (L/hour) | 9.851 | 14.376 |
| CV Q(gut) (%) | 30 | 30 |
| PAMPA permeability, PAMPA (10−6 cm/s) | 10 | — |
| Permeability A‐B, Caco‐2 (cm/s*10−6) | — | 0.89 |
| Caco‐2 reference (L/hour) | — | 14.376 |
| Permeability scalar | — | 0.885 |
| 3. Distribution | ||
| Distribution model | Full PBPK | Minimal PBPK model |
| Tissue model | Perfusion limited model | Perfusion limited model |
| Volume of distribution at steady state (L/kg) | 1.45f | 0.748 |
| CV Vss (%) | 3.4d | 30 |
| Extent of tissue Kp, Kp scalar | 0.574 (optimized) | — |
| 4. Enzyme/transporter phenotyping | ||
| Human liver microsomes | ||
| In vitro intrinsic clearance (CYP2B6, μL/minutes/pmol) (fmCYP2B6) | 0.733g (0.004) | — |
| CLint (CYP2C8, μL/minutes/pmol) (fmCYP2C8) | 2.941g (0.017) | — |
| CLint (CYP2C19, μL/minutes/pmol) (fmCYP2C19) | 0.593g (0.001) | — |
| CLint (CYP3A4, μL/minutes/pmol) (fmCYP3A4) | 5.607g (0.185) | — |
| CLint (allelic CYP2C9*1/*1, μL/minutes/pmol) (fmCYP2C9/fmCYP3A4) | 45.105h (0.804/0.175) | — |
| CLint (allelic CYP2C9*1/*2, μL/minutes/pmol) (fmCYP2C9/fmCYP3A4) | 45.885h (0.788/0.189) | — |
| CLint (allelic CYP2C9*1/*3, μL/minutes/pmol) (fmCYP2C9/fmCYP3A4) | 24.605h (0.678/0.287) | — |
| CLint (allelic CYP2C9*2/*2, μL/minutes/pmol) (fmCYP2C9/fmCYP3A4) | 33.271h (0.727/0.244) | — |
| CLint (allelic CYP2C9*2/*3, μL/minutes/pmol) (fmCYP2C9/fmCYP3A4) | 18.924h (0.616/0.343) | — |
| CLint (allelic CYP2C9*3/*3, μL/minutes/pmol) (fmCYP2C9/fmCYP3A4) | 2.869h (0.074/0.822) | — |
| 5. Other distribution and elimination property | ||
| In vivo CL | — | — |
| In vivo renal clearance in a 20–30 year healthy man (L/hour) | 0 | 0.7 |
| CL following i.v. administration (L/hour) | — | 1.01 |
| CV CLiv (%) | — | 24 |
| In vitro CL | — | — |
| Hepatic uptake | 1.0j | 1 |
| Overall biliary clearance, CLint(hep) (μL/minutes/106 cells) | 0i | — |
| CV CLint(hep) (%) | 30i | — |
| 6. Interaction | ||
| Inhibition constant (CYP2C9; μM) | — | 20.4 |
| Fraction unbound in human liver microsome (CYP2C9) | — | 1 |
| Ki (CYP2C19; μM) | — | 2 |
| fu(mic) (CYP2C19) | — | 1 |
| Ki (CYP3A4; μM) | — | 10.7 |
| fu(mic) (CYP3A4) | — | 1 |
| Ki (CYP3A5; μM) | — | 84.6 |
| fu(mic) (CYP3A5) | — | 1 |
CL, clearance; CLint,u, unbound intrinsic clearance; CLint(hep), overall biliary clearance; CLiv, clearance following intravenous administration; CV, coefficient of variation; CYP, cytochrome P450; fa, fraction available from dosage form; fm, fraction metabilized; fu(mic), fractions unbound in human liver microsome; HSA, human serum albumin; Ka, absorption rate constant; Ki, inhibition constant; Kp, partition coefficient; PAMPA, parallel artificial membrane permeability assay; PBPK, physiologically‐based pharmacokinetic; Q(gut), nominal flow in gut model; Vss, volume of distribution at steady state.
aAll data taken from the fluconazole SimCYP compound file V16 except for a CYP2C9 Ki.16 This was verified by showing improved predictability of the clinical drug−drug interaction effects on probe CYP2C9 substrates. bNo involvement of intestinal efflux transporters in the absorption process ( Table S4 ). cBased on amount of siponimod excreted to feces.4 dEstimated based on population pharmacokinetic (PopPK) model. eSensitivity analysis performed. fBased on mean Vss observed (Table 1), converted to L/kg based on mean body weight of the study subjects (80.08 kg). gCL used in the retrograde calculator was the observed geometric mean CL = 3.12 L/hour (Table 1); due to the retrograde calculation fu(mic) was set to 1 for all enzymes. hCalculated based on PopPK CL data ( Table S2 ). iBased on human absorption, distribution, metabolism, and excretion study4 and enzyme phenotyping data,4 it was concluded that hepatic metabolism is exclusively driving CL. j Table S4.
For all genotypes, there was a moderate AUC increase (AUCR: 2.13–2.49) predicted with moderate dual CYP3A4/CYP2C9 inhibitors (fluconazole) and a weak AUC increase (AUCR: 1.12–1.42) with weak dual CYP3A4/CYP2C9 inhibitors (fluvoxamine). For the interaction with the strong CYP3A4 and moderate CYP2C9 inducer (rifampin), no relevant difference in AUCRs across all CYP2C9 genotypes was predicted (range: 0.21–0.32). The different CYP2C9 genotypes are equally inducible by a CYP2C9 inducer.17 Therefore, we assumed that no adaptations were required for simulating CYP2C9‐mediated induction effects on siponimod for the six clinically relevant genotypes.
The predicted geometric mean AUCinf ratios of siponimod (0.25 mg, single dose at day 5) with a coadministration of the strong CYP3A4 inhibitor itraconazole (100 mg b.i.d.) in CYP2C9*1/*2 (1.18) and CYP2C9*1/*3 genotypes (1.29) were different from the clinical observations (0.90, 0.76, respectively; unpublished data13), indicating an apparent induction potential of itraconazole on the PK of siponimod rather than an inhibition effect. A potential explanation for the miss‐predictions would be the involvement of another metabolizing enzyme of siponimod (i.e., CYP1A1), which was identified in the in vitro enzyme phenotyping study as a minor pathway.4 CYP1A1 was not considered for the PBPK model due to its low abundance in the noninduced state.18 However, additional investigations indicated that the potential involvement of CYP1A1 was not likely to be the root cause of the decreased siponimod exposure in the presence of itraconazole (unpublished data13), as the itraconazole in vitro CYP1A1 inhibition effect is stronger than its induction potential. The underlying mechanism leading to these clinical observations is currently unknown. However, it is considered that the effect observed was specific to itraconazole. It is concluded that other CYP3A4 inhibitors, without the effect observed with itraconazole (e.g., clarithromycin), should act as strong and selective CYP3A4 inhibitors, resulting in increased AUCRs of siponimod > 1 when coadministered. The basis for this assumption is the good correlation between observed and predicted DDI results of the dual perpetrators (CYP2C9/CYP3A4), fluconazole, and rifampin, verifying the fractional contribution of CYP3A4 to CLtot of siponimod. The fraction metabolized via CYP2C9 (fm,CYP2C9) was confirmed by the good match of the genotype‐specific PK data for CYP2C9*1/*1, CYP2C9*2/*3, and CYP2C9*3/*3. Therefore, it is believed that the established siponimod model for DDI simulations is still valid for typical CYP3A4 and CYP3A4/CYP2C9 perpetrators without CYP1A1 induction potential.
The recommended siponimod maintenance doses in the FDA approved prescribing information are 2 mg daily in patients with CYP2C9*1/*1, *1/*2, and *2/*2 genotypes, and 1 mg daily in patients with CYP2C9*1/*3 and *2/*3 genotypes, and siponimod is contraindicated for patients with a CYP2C9*3/*3 genotype. Considering these doses, the predicted net effect on siponimod exposures ranged between 1.06 and 1.90 across CYP2C9 genotypes in presence of strong and moderate CYP3A4 inhibitors, and weak dual CYP3A4/2C9 inhibitors. A larger effect is predicted in the presence of moderate dual CYP3A4/2C9 inhibitors with ratios ranging between 1.72 and 2.91 (Table 4). Because of this significant increase in exposure of siponimod, the US prescribing information stipulated that concomitant use of siponimod and drugs that cause moderate CYP2C9 and moderate or strong CYP3A4 inhibition are not recommended. In addition, caution should be exercised for concomitant use of siponimod with moderate CYP2C9 inhibitors.3
Dual strong CYP3A4/moderate CYP2C9 inducers are predicted to reduce siponimod exposure by ~ 63% to 78%, and moderate CYP3A4 inducers by ~ 15–52% (Table 4). Therefore, concomitant use of siponimod with drugs that cause moderate CYP2C9 and strong CYP3A4 induction is not recommended for all patients irrespective of their CYP2C9 genotype.3 Caution should be exercised for concomitant use of siponimod with moderate CYP2C9 inducers. In addition, concomitant use of siponimod and moderate (e.g., modafinil and efavirenz) or strong CYP3A4 inducers is not recommended for patients with CYP2C9*1/*3 and *2/*3 genotypes.3
The current PBPK study indicated that the CYP2C9 genotype has a relevant influence on the DDI behavior of siponimod in the presence of different CYP3A4/CYP2C9 inducers and inhibitors. Moreover, this study demonstrated the potential of PBPK modeling to complement clinical DDI information with simulations of untested scenarios. These results served as the basis for the approved label recommendations3 when siponimod is administered in combination with CYP2C9/CYP3A4 perpetrator drugs, to ensure that safety and the efficacy of siponimod are preserved in all patients.
Methods
The population‐based PKs simulator SimCYP version 16 (Certara, Princeton, NJ) was used to predict the DDI victim potential of siponimod in the presence of typical CYP3A4/CYP2C9 inhibitors and inducers.
All simulations within this study were performed with the SimCYP healthy volunteer population, as there was no clinical relevant PK difference between a healthy volunteer and a patient with multiple sclerosis population identified by PopPK.12
The fluconazole SimCYP version 16 model was adapted by changing the CYP2C9 Ki value from 7.92 to 20.4 μM16 to increase predictability of the inhibition effects on probe CYP2C9 substrates (S‐warfarin, tolbutamide, and phenytoin). The efavirenz model was built as a pure CYP3A4 inducer model based on SimCYP version 16, because negligible CYP2C9 induction is expected for a moderate CYP3A4 inducer.19, 20 Input values for the perpetrator drugs itraconazole, ketoconazole, erythromycin, fluvoxamine, and rifampin were provided by SimCYP version 16.
All systemic PK data were based on plasma concentrations. AUC, Cmax, Cmin, t1/2, and Tmax are reported as geometric means or median with range or as 90% confidence intervals. The exposure change in the presence of perpetrators was defined as the AUCR and/or Cmax ratio.
Siponimod PBPK model
Model building
A siponimod PBPK model was established, in alignment with a drug disposition scheme of siponimod (Figure 1), using a mixed approach combining in vitro data, physicochemical parameters, and PK parameters derived from clinical studies. The clinical PK data from the single ascending dose (Table 1), multiple‐ascending dose (Table 1), absolute bioavailability (Table 1), human absorption, distribution, metabolism, and excretion,4 and fluconazole DDI study results5, 12, 21 were used for building of the PBPK model. This model was verified for its PK and the fraction metabolized via CYP2C9 (fm,CYP2C9) by the genotype PK data (Table 2)5 and by clinical siponimod PK and DDI data in the absence and presence of rifampicin verifying the fraction metabolized via CYP3A4 (fm,CYP3A4).12 The final PBPK model was used to predict the siponimod DDI potential as substrate at steady state in the presence of typical CYP3A4/CYP2C9 inhibitors for the six clinically relevant CYP2C9 genotypes (*1/*1, *1/*2, *1/*3, *2/*2, *2/*3, and *3/*3). The model‐building workflow for the siponimod PBPK model is illustrated in Figure S4.
Model input parameters
The siponimod SimCYP model was established based on the physicochemical properties from the in vitro absorption, distribution, metabolism, and excretion data. Fraction absorbed was determined from the human absorption, distribution, metabolism, and excretion study.4 Fraction unbound in the enterocytes (fugut) was set to be equal to unbound fraction in plasma. Absorption rate constant (Ka) was taken from the PopPK analysis results (Novartis data on file). Lag time was optimized based on the clinical Tmax data at steady state of the multiple‐ascending dose study (Table 1). Volume of distribution at steady state and CLtot were used as determined after single i.v. administration of siponimod in the clinical study (Table 1). Siponimod clearance is exclusively driven by CYP‐mediated metabolism. The hepatic uptake of siponimod in human hepatocytes was purely passive with no contribution of active uptake transporters (Table S1). Therefore, the hepatic intrinsic clearance values (CLint,h) for CYP2C9, CYP3A4, CYP2B6, CYP2C8, and CYP2C19 were determined by distributing the relative fractions metabolized (fm) from the in vitro enzyme phenotyping study.4 The respective CLint,h values for the six different CYP2C9 genotypes were calculated based on PopPK total body clearance of the drug from plasma results for each genotype (*1/*1, *1/*2, *1/*3, *2/*2, *2/*3, and *3/*3; Table S3). Total clearance (CLtot) values were derived by multiplying CLtot/F by F. There is no excretion of siponimod via urine and no biliary or intestinal secretion.4 The siponimod model input parameters and all assumptions are summarized in Table 5.
DDI simulations used for model verification and prospective DDI predictions
Effects of CYP3A4 and CYP2C9 inhibitors on siponimod
Simulated DDI potential of fluconazole on the systemic exposure of siponimod after oral administration was compared with the respective clinical study results, matching the study design. This was used to verify estimations of fm by CYP3A4 (fm,CYP3A4) for siponimod. Fraction metabolized by CYP2C9 (fm,CYP2C9) based on in vitro data was verified in model building based on the PK of the different genotypes CYP2C9*1/*1, CYP2C9*2/*3 and CYP2C9*3/*3.4 The model was further verified by the clinical rifampicin DDI study data (described in the next section). Using the siponimod model with the verified fm,CYP3A4 value, the effects of itraconazole, ketoconazole, erythromycin, and fluvoxamine on the siponimod PK were simulated (Table S4).
Effects of CYP3A4 and CYP2C9 inducers on siponimod
Simulated DDI potential of rifampin on the systemic exposure of siponimod after oral administration was compared with the respective clinical study results,12 matching the study design. The result supported validity of fm,CYP3A4 for siponimod (i.e., estimated based on the clinical PK results and DDI data with fluconazole)5 and the PK of the genotypes CYP2C9*2/*3 and CYP2C9*3/*3. Subsequently, the effect of efavirenz on the siponimod PK was simulated (Table S4).
Sensitivity analysis
To investigate the impact of Ka changes on siponimod AUC, a sensitivity analysis within the range of 0.1–2 was performed. To investigate the impact of fugut changes on siponimod Cmax, a sensitivity analysis was performed within the range of 0.0002–1. The fluconazole CYP2C9 Ki was revised by the SimCYP team. Therefore, to investigate the impact of CYP2C9 Ki changes on siponimod AUC in the presence of fluconazole, a sensitivity analysis was performed for the CYP2C9*1/*1 genotype population. The CYP2C9 and CYP3A4 Kdeg values were defined in the SimCYP version 16. A sensitivity analysis within the range of 0.005–0.03 per hour was performed to investigate the impact of Kdeg,CYP2C9 and Kdeg,CYP3A4 changes on siponimod AUCRs when coadministered with rifampin.
Funding
The study was funded by Novartis Pharma AG, Basel, Switzerland.
Conflict of Interest
F.H., A.G., and H.H. are employees of Novartis. K.U. was an employee of Novartis during the conduct of the study and development of the manuscript.
Author Contributions
F.H., A.G., K.U., and H.H. wrote the manuscript. F.H. and A.G. designed the research. F.H. performed the research. F.H., A.G., K.U., and H.H. analyzed the data.
Supporting information
Figure S1. Residual plot showing the predictability of Cmax, Cmin, and AUC; (A) Cmax and AUCinf after single oral administration of siponimod; (B) Cmin, Cmax, and AUC0‐24 h at days 28 after multiple oral administration of siponimod. Area0‐24h, area under the curve between 0‐24 hours; AUC, area under the curve; AUCinf : area under the curve extrapolated to infinity; Cmax, maximum plasma concentration; Cmin, minimum plasma concentration.
Figure S2. Siponimod pharmacokinetics: SimCYP simulation for multiple dose oral administration.
Figure S3. (A) Effect of Ka on predicted Tmax of siponimod. (B) Effect of fugut on predicted AUC and (C) Cmax of siponimod. (D) Effect of Ki,CYP2C9 on the systemic AUC and (E) Cmax ratio prediction of siponimod in the presence and absence of fluconazole. (F) Effect of mean turnover CYP2C9 and CYP3A4 on AUC and Cmax ratios of siponimod in the presence of the CYP3A4/CYP2C9 inducer rifampin; mean turnover CYP2C9 on AUC and (G) Cmax ratios for CYP2C9*1/*1 genotype; mean turnover CYP3A4 on (H, I) AUC and (J, K) Cmax ratio for CYP2C9*1/*1 and CYP2C9*3/*3 genotypes, respectively. AUC, area under the curve; Cmax, maximum plasma concentrations; CYP2C9, Cytochrome P450 2C9; CYP3A4, Cytochrome P450 3A4; fugut, fraction unbound in enterocytes; Ka, absorption rate constant; Ki,CYP2C9, inhibition constant for CYP2C9; Tmax, time at maximum plasma concentration.
Figure S4. Model building process for siponimod.
Table S1. In vitro investigation on siponimod as cytochrome P450 enzyme or transporter inhibitor and transporter substrate.
Table S2. Intrinsic clearance values of allelic cytochrome P450 2C9 genotypes calculated based on genotype specific clearance data from population pharmacokinetic analysis.†
Table S3. Pharmacokinetic simulation design parameters for the siponimod model verification and for predictions of subpopulations with the different cytochrome P450 2C9 genotypes.
Table S4. Drug–drug interaction simulation design parameters for predictions of siponimod as a victim drug with cytochrome P450 enzyme inhibitors and inducers.
Acknowledgments
The authors would like to acknowledge Jitendriya Mishra and Anuja Shah (Novartis Healthcare Pvt. Ltd., Hyderabad, India) for providing medical writing support, which encompassed manuscript preparation, formatting, referencing, preparing tables and figures, incorporating the authors’ revisions, finalizing, and submission all under the direction of the authors. In keeping with the guidelines of the International Committee of Medical Journal Editors, all authors have contributed significantly to the study and have been thoroughly involved in the critical review of the manuscript for important intellectual content.
References
- 1. Gergely, P. et al The selective sphingosine 1‐phosphate receptor modulator BAF312 redirects lymphocyte distribution and has species‐specific effects on heart rate. Br. J. Pharmacol. 167, 1035–1047 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Kappos, L. et al Siponimod versus placebo in secondary progressive multiple sclerosis (EXPAND): a double‐blind, randomised, phase 3 study. Lancet 391, 1263–1273 (2018). [DOI] [PubMed] [Google Scholar]
- 3. Mayzent (siponimod) [prescribing information] (Novartis Pharmaceuticals Corporation, East Hanover, NJ, 2019). [Google Scholar]
- 4. Glaenzel, U. et al Metabolism and disposition of siponimod, a novel selective S1P1/S1P5 agonist, in healthy volunteers and in vitro identification of human cytochrome P450 enzymes involved in its oxidative metabolism. Drug Metab. Dispos. 46, 1001–1013 (2018). [DOI] [PubMed] [Google Scholar]
- 5. Gardin, A. et al Effect of fluconazole coadministration and CYP2C9 genetic polymorphism on siponimod pharmacokinetics in healthy subjects. Clin. Pharmacokinet. 58, 349–361 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Fahmi, O.A. , Maurer, T.S. , Kish, M. , Cardenas, E. , Boldt, S. & Nettleton, D. A combined model for predicting CYP3A4 clinical net drug‐drug interaction based on CYP3A4 inhibition, inactivation, and induction determined in vitro. Drug Metab. Dispos. 36, 1698–1708 (2008). [DOI] [PubMed] [Google Scholar]
- 7. Human Cytochrome P450 (CYP) Allele Nomenclature Committee <https://www.pharmvar.org/htdocs/archive/index_original.htm> (2016). Accessed April 1, 2019.
- 8. Caudle, K.E. et al Clinical pharmacogenetics implementation consortium guidelines for CYP2C9 and HLA‐B genotypes and phenytoin dosing. Clin. Pharmacol. Ther. 96, 542–548 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Scott, S.A. , Khasawneh, R. , Peter, I. , Kornreich, R. & Desnick, R.J. Combined CYP2C9, VKORC1 and CYP4F2 frequencies among racial and ethnic groups. Pharmacogenomics 11, 781–791 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Van Booven, D. et al Cytochrome P450 2C9‐CYP2C9. Pharmacogenet. Genom. 20, 277–281 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Kirchheiner, J. & Brockmoller, J. Clinical consequences of cytochrome P450 2C9 polymorphisms. Clin. Pharmacol. Ther. 77, 1–16 (2005). [DOI] [PubMed] [Google Scholar]
- 12. Gardin, A. et al Siponimod pharmacokinetics, safety, and tolerability in combination with rifampin, a CYP2C9/3A4 inducer, in healthy subjects. Eur. J. Clin. Pharmacol. 74, 1593–1604 (2018). [DOI] [PubMed] [Google Scholar]
- 13. Gardin, A. , Shakeri‐Nejad, K. , Feller, A. , Huth, F. , Neelakantham, S. & Dumitras, S. Siponimod pharmacokinetics, safety and tolerability in combination with the potent CYP3A4 inhibitor itraconazole in healthy subjects with different CYP2C9 genotypes. Eur. J. Clin. Pharmacol. (in press). [DOI] [PubMed] [Google Scholar]
- 14. European Medicines Agency (EMA) . Guideline on the investigation of drug interactions <http://www.ema.europa.eu/docs/en_gb/document_library/scientific_guideline/2012/07/wc500129606.Pdf> (2012).
- 15. US Food and Drug Administration (FDA) . Clinical drug interaction studies — study design, data analysis, and clinical implications. Guidance for industry <https://www.fda.gov/downloads/drugs/guidances/ucm292362.pdf> (2017).
- 16. Neal, J.M. , Kunze, K.L. , Levy, R.H. , O'Reilly, R.A. & Trager, W.F. Kiiv, an in vivo parameter for predicting the magnitude of a drug interaction arising from competitive enzyme inhibition. Drug Metab. Disp. 31, 1043–1048 (2003). [DOI] [PubMed] [Google Scholar]
- 17. Vormfelde, S.V. et al Relative impact of genotype and enzyme induction on the metabolic capacity of CYP2C9 in healthy volunteers. Clin. Pharmacol. Ther. 86, 54–61 (2009). [DOI] [PubMed] [Google Scholar]
- 18. Kim, K. , Johnson, J.A. & Derendorf, H. Differences in drug pharmacokinetics between East Asians and caucasians and the role of genetic polymorphisms. J. Clin. Pharmacol. 44, 1083–1105 (2004). [DOI] [PubMed] [Google Scholar]
- 19. Fahmi, O.A. , Kish, M. , Boldt, S. & Obach, R.S. Cytochrome P450 3A4 mRNA is a more reliable marker than CYP3A4 activity for detecting pregnane x receptor‐activated induction of drug‐metabolizing enzymes. Drug Metab. Dispos. 38, 1605–1611 (2010). [DOI] [PubMed] [Google Scholar]
- 20. European Commission . Eurl‐ecvam. Multi‐study validation trial for cytochrome P450 induction providing a reliable human metabolically competent standard model or method using the human cryopreserved primary hepatocytes and the human cryopreserved Heparg cell line. CYP induction Eurl‐ecvam validation project report <https://www.oecd.org/chemicalsafety/testing/cyp-validation-project-report.pdf>. Accessed April 5, 2019.
- 21. Jin, Y. , Borell, H. , Gardin, A. , Ufer, M. , Huth, F. & Camenisch, G. In vitro studies and in silico predictions of fluconazole and CYP2C9 genetic polymorphism impact on siponimod metabolism and pharmacokinetics. Eur. J. Clin. Pharmacol. 74, 455–464 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1. Residual plot showing the predictability of Cmax, Cmin, and AUC; (A) Cmax and AUCinf after single oral administration of siponimod; (B) Cmin, Cmax, and AUC0‐24 h at days 28 after multiple oral administration of siponimod. Area0‐24h, area under the curve between 0‐24 hours; AUC, area under the curve; AUCinf : area under the curve extrapolated to infinity; Cmax, maximum plasma concentration; Cmin, minimum plasma concentration.
Figure S2. Siponimod pharmacokinetics: SimCYP simulation for multiple dose oral administration.
Figure S3. (A) Effect of Ka on predicted Tmax of siponimod. (B) Effect of fugut on predicted AUC and (C) Cmax of siponimod. (D) Effect of Ki,CYP2C9 on the systemic AUC and (E) Cmax ratio prediction of siponimod in the presence and absence of fluconazole. (F) Effect of mean turnover CYP2C9 and CYP3A4 on AUC and Cmax ratios of siponimod in the presence of the CYP3A4/CYP2C9 inducer rifampin; mean turnover CYP2C9 on AUC and (G) Cmax ratios for CYP2C9*1/*1 genotype; mean turnover CYP3A4 on (H, I) AUC and (J, K) Cmax ratio for CYP2C9*1/*1 and CYP2C9*3/*3 genotypes, respectively. AUC, area under the curve; Cmax, maximum plasma concentrations; CYP2C9, Cytochrome P450 2C9; CYP3A4, Cytochrome P450 3A4; fugut, fraction unbound in enterocytes; Ka, absorption rate constant; Ki,CYP2C9, inhibition constant for CYP2C9; Tmax, time at maximum plasma concentration.
Figure S4. Model building process for siponimod.
Table S1. In vitro investigation on siponimod as cytochrome P450 enzyme or transporter inhibitor and transporter substrate.
Table S2. Intrinsic clearance values of allelic cytochrome P450 2C9 genotypes calculated based on genotype specific clearance data from population pharmacokinetic analysis.†
Table S3. Pharmacokinetic simulation design parameters for the siponimod model verification and for predictions of subpopulations with the different cytochrome P450 2C9 genotypes.
Table S4. Drug–drug interaction simulation design parameters for predictions of siponimod as a victim drug with cytochrome P450 enzyme inhibitors and inducers.