Role of CYP3A in Oral Contraceptives Clearance

INTRODUCTION

We evaluated the relative contribution of CYP3A in the overall clearance of commonly used combined oral contraceptives (COCs) based on the results of clinical DDI studies in the literature and new drug applications (NDAs). The results revealed a limited role of CYP3A4 in the metabolism of COC components. Characterization of inhibition or induction spectrum of perpetrators on non‐CYP3A pathways might also be crucial in predicting drug interaction potential of an investigational new drug with COCs.

BACKGROUND

COC is the most commonly used contraceptive method in the United States.1 Approximately 9.7 million women at reproductive age are COC users.1 COCs usually contain two synthetic steroid hormones, an estrogen, typically ethinyl estradiol (EE), and a progestin. The commonly used progestins include norethindrone (NET), levonorgestrel (LNG), drospirenone (DRSP), norgestimate (NGM), desogestrel, and gestodene.2 Potential drug‐drug interactions (DDIs) should be considered when a medication or an herbal supplement is taken with a hormonal contraceptive (HC). The decreased or increased concentrations of estrogen and progestins due to concomitant medications may lead to unintended pregnancy (loss of efficacy) or increased incidence of adverse events (e.g., increased risk of venous thromboembolism, a rare but severe adverse event). Given the high prevalence of COC use in women and possible consequences of unwanted pregnancy, clinical studies to evaluate the DDI potential between an investigational new drug (that is intended to be used in women with childbearing potential) and COCs have been conducted routinely during the drug development stage and sometimes after drug approval.

The metabolic pathways of progestins and EE are not completely understood, as many of these steroids were developed >50 years ago with a limited number of studies. The general consensus is that CYP3A is the major enzyme for oxidative metabolism of EE and the commonly used progestins, including NET, LNG, NGM, and DRSP.3, 4, 5 Therefore, most DDI studies assessing the effect of other drugs on the exposure of COCs have been conducted based on the possible interaction via CYP3A. However, there is a certain level of variation among progestins in terms of chemical structures, metabolic pathways, and pharmacokinetic (PK) characteristics. In addition, the significance of CYP3A in the overall disposition of these hormones remains unclear. The objective of this study was to assess the relative contribution of CYP3A in the metabolism of steroid hormone components of COCs using publically available clinical DDI study results. The results from the current assessment might give a new insight on the significance of CYP3A‐mediated drug interactions with COCs.

SELECTION OF COC DDI STUDIES FOR THE SURVEY

The results of DDI studies with COCs were collected via (i) literature search using the electronic databases MEDLINE and PubMed from 1996 to November 2014; (ii) publically available US Food and Drug Administration (FDA) review for new drug applications (Drugs@FDA, http://www.accessdata.fda.gov/scripts/cder/daf/) from 1996 to November 2014; and (iii) the University of Washington Metabolism and Transport Drug Interaction Database (http://www.druginteractioninfo.org).6 The following search terms were applied: drug interactions, contraception and oral contraceptives, EE, DRSP, LNG, NET, and NGM. Four progestins (DRSP, NET, LNG, and NGM) containing COCs were selected for this survey, as they are the most commonly used COCs in the United States and are frequently studied in COC DDI studies.2

The following major criteria were applied for the selection of DDI studies for further analyses: (i) prospective clinical studies conducted in healthy subjects or patients to assess a DDI potential with COCs; and (ii) there is sufficient treatment duration of perpetrators for their inhibitory or induction effect on CYP3A (e.g., inhibitors: treatment duration to achieve steady state or shorter duration if it is consistent with clinical use; inducers: longer than 1 week). In addition, to have a relatively clean data set for further analyses, we excluded some DDI studies based on the following criteria: (i) DDI studies that were conducted based on the likelihood of coadministration without a clear mechanism of drug interactions; (ii) case reports; (iii) perpetrators that have mixed CYP3A4 DDI potential (i.e., both inhibition and induction of CYP3A4); (iv) combination drug products containing multiple CYP3A4 perpetrators, which potentially would result in combined DDI effects (i.e., for two‐drug combination, inhibition plus inhibition, induction plus induction, and inhibition plus induction); and (v) perpetrators that did not show inhibition or induction effect on sensitive CYP3A substrates in clinical DDI studies. The sample size of the study and its statistical power were not considered for the selection because there were a limited number of studies that provided its rationale based on the power or sensitivity. If there were multiple studies evaluating the same perpetrators, each of them was included in the final analysis to support the conclusion from each other unless it is/they are not representative of clinical scenarios. If differences were found in the DDI study results, causes would be explored.

Based on the selection criteria listed above, selected studies were grouped based on the inhibitory and induction potency of perpetrators on CYP3A (i.e., perpetrators are classified as strong, moderate, or weak inhibitors or inducers of CYP3A based on the criteria described in the FDA's draft DDI guidance).7, 8 Specifically, drugs or herbal supplements that increase the area under the curve (AUC) of a sensitive index CYP3A substrate by greater than or equal to fivefold are considered as strong CYP3A inhibitors. Drugs or herbal supplements that increase AUC of a sensitive index CYP3A substrate by twofold to fivefold or less than twofold are classified as moderate and weak CYP3A inhibitors, respectively. Similarly, strong, moderate, and weak CYP3A inducers should decrease AUC of a sensitive index CYP3A substrate by ≥80%, 50–80%, and 20–50%, respectively. The geometric mean ratio (GMR) of AUC for COCs with and without perpetrators are presented using forest plots to illustrate the effect of CYP3A inhibitors and CYP3A inducers on oral contraceptive exposures (Figures 2, 3, 4, 5). The plot illustrates the fold‐change and 90% confidence intervals (CIs) for AUC of COCs with and without perpetrators observed in the clinical DDI studies. The shaded area shows the GMR between 0.80 and 1.25, the default no‐effect boundary, as specified in the DDI guidance. The DDI data for peak concentration (C_max) are not discussed in the current assessment because the trend of changes in C_max are, in general, either similar to or lesser than that of AUC for the DDI studies included in the current assessment. In addition, the underlying mechanisms for the changes in C_max are more complicated to interpret toward the impact on the systemic clearance than that for AUC.

Figure 2.

Effect of various cytochrome P450 3A (CYP3A) inhibitors on the area under the curve (AUC) of ethinyl estradiol. Circles in black, grey, and white represent strong, moderate, and weak CYP3A inhibitors, respectively. Symbols and bars indicate geometric mean ratio and 90% confidence intervals, respectively. Shaded area indicates the default no‐effect boundary of 0.8 and 1.25.

Figure 3.

Effect of various cytochrome P450 3A (CYP3A) inhibitors on the area under the curve (AUC) of progestins. Circles in black, grey, and white represent strong, moderate, and weak CYP3A inhibitors, respectively. Symbols and bars indicate geometric mean ratio and 90% confidence intervals, respectively. Shaded area indicates the default no‐effect boundary of 0.8 and 1.25. DRSP, drospirenone; LNG, levonorgestrel; NET, norethindrone.

Figure 4.

Effect of various cytochrome P450 3A (CYP3A) inducers on the area under the curve (AUC) of ethinyl estradiol. Circles in black, grey, and white represent strong, moderate, and weak CYP3A inducers, respectively. Symbols and bars indicate geometric mean ratio and 90% confidence intervals, respectively. Shaded area indicates the default no‐effect boundary of 0.8 and 1.25.

Figure 5.

Effect of various cytochrome P450 3A (CYP3A) inducers on the area under the curve (AUC) of progestins. Circles in black, grey, and white represent strong, moderate, and weak CYP3A inducers, respectively. Symbols and bars indicate geometric mean ratio and 90% confidence intervals, respectively. Shaded area indicates the default no‐effect boundary of 0.8 and 1.25. LNG, levonorgestrel; NET, norethindrone; NGMN, norelgestromin.

FINDINGS ON THE EFFECT OF CYP3A ON THE CLEARANCE OF EE AND PROGESTINS

In total, 141 clinical drug interaction studies for these four COC products (EE + NET, EE + LNG, EE + NGM, and EE + DRSP) were collected, among which some were conducted with the same perpetrator‐COC drug pair. Of these studies, 33 studies were conducted only based on the likelihood of coadministration without any in vitro or in vivo data to specify underlying mechanisms of interaction. Therefore, they were not included in the data analysis. Six studies were excluded for one or more of the following reasons: (i) they were case reports; (ii) they were perpetrators that are thought to have a mixed effect (both induction and inhibition) on enzymes involved in the metabolism of EE and progestins (e.g., aprepitant and ritonavir (inhibitor and inducer of CYP3A))9, 10, 11, 12; and (iii) they were DDI studies with combination drug products, such as Stribild (elvitegravir, cobicistat, emtricitabine, and tenofovir disoproxil fumarate). In addition, 72 studies were excluded from the current review because the perpetrators of these studies had no inhibition or induction effect on sensitive CYP3A substrates, as demonstrated in clinical DDI studies. After reviewing all the studies, 30 studies were selected for further analysis (Figure 1).

Figure 1.

Flow chart of selecting combined oral contraceptive (COC) drug‐drug interaction (DDI) studies to assess the role of cytochrome P450 3A (CYP3A) in COC clearance.

Effect of CYP3A inhibitors on the exposure of EE and progestins

The magnitude of increase in the exposure (AUC) of EE was not large in the presence of CYP3A inhibitors (Figure 2 and Table 1, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26). Strong CYP3A inhibitors, such as voriconazole and ketoconazole, increased EE exposure by 60% and 40%, respectively.13, 14 Telithromycin did not affect the exposure of EE.19 Interestingly, two protease inhibitors, telaprevir and boceprevir, which are known strong CYP3A inhibitors, seemed to decrease the AUC of EE by 26% and 28%, respectively.16, 17 Moderate CYP3A inhibitors, including fluconazole, atazanavir, and faldaprevir, resulted in a 38% to 48% increase in the AUC of EE.20, 21, 22, 27 No significant effect on the exposure of EE was observed with other moderate (netupitant23) or weak CYP3A inhibitors.24, 25, 26

Table 1.

Effect of CYP3A inhibitors on the AUC of EE, NET, LNG, or DRSP

CYP inhibitor		COC			AUC ratio of EE or progestin (GMR, 90% CI)
Drug	Dose	Drug(s)	Dose on the PK study day	No. of subjects	EE	Progestin
Voriconazole13	400 mg b.i.d. on the first day, then 200 mg b.i.d. for 3 days	EE/NET multiple doses for 21 days	EE 0.035 mg/NET 1 mg	16	1.61 (1.50–1.72)	1.53 (1.44–1.64)
Ketoconazole14	200 mg b.i.d. for 10 days	EE/DRSP multiple doses for 38 days	EE 0.020 mg/DRSP 3 mg	20	1.40 (1.31–1.49)	2.68 (2.44–2.95)
Indinavir15	800 mg t.i.d. for 8 days	EE/NET multiple doses for 8 days	EE 0.035 mg/NET 1 mg	18	1.22 (1.15–1.30)	1.26 (1.20–1.31)
Telaprevir16	750 mg t.i.d. for 28 days	EE/NET multiple doses for 21 days	EE 0.035 mg/NET 0.5 mg	24	0.72 (0.69–0.75)	0.89 (0.86–0.93)
Boceprevir17	800 mg t.i.d. for 28 days	EE/NET multiple doses for 21 days	EE 0.035 mg/NET 1 mg	20	0.74 (0.68–0.80)	0.96 (0.87–1.06)
Boceprevir18	800 mg t.i.d. for 7 days	EE/DRSP multiple doses for 7 days	EE 0.020 mg/DRSP 3 mg	20	0.76 (0.73–0.79)	1.99 (1.87–2.11)
Telithromycin19	800 mg q.d. for 10 days	EE/LNG multiple doses for 21 days	EE 0.040 mg/LNG 0.075 mg	30	1.02 (0.98–1.07)	1.50 (1.44–1.57)
Faldprevir20	240 mg b.i.d. for 1 day, then 240 mg q.d. for 7 days	EE/LNG multiple doses for 13 & 8 days	EE 0.030 mg/LNG 0.150 mg	16	1.41 (1.34–1.48)	1.40 (1.36–1.45)
Fluconazole21	200 mg q.d. for 10 days	EE/LNG single dose	Unknown	25	1.38 (0.89–2.01)	1.25 (0.88–1.82)
Atazanavir22	400 mg q.d. for 14 days	EE/NET multiple doses for 21 & 14 days	EE 0.035 mg/NET 1 mg or 0.75 mg	19	1.48 (1.31–1.68)	2.10 (1.68–2.62)
Netupitant23	300 mg single dosea	EE/LNG single dose	EE 0.060 mg/LNG 0.3 mg	24	1.12 (1.02–1.22)	1.40 (1.24–1.58)
Ivacaftor24	150 mg b.i.d. for 28 days	EE/NET multiple doses for 21 days	EE 0.035 mg/NET 0.5 mg	22	1.07 (1.00–1.14)	1.05 (1.00–1.12)
Simeprevir25	150 mg q.d. for 10 days	EE/NET multiple doses for 21 days	EE 0.035 mg/NET 1 mg	18	1.12 (1.05–1.20)	1.15 (1.08–1.22)
Ticagrelor26	90 mg b.i.d. for 21 days	EE/LNG multiple doses for 21 days	EE 0.030 mg/ LNG 0.15 mg	22	1.20 (1.03–1.40)	1.03 (0.97–1.10)

AUC, area under the curve; CI, confidence interval; COC, combined oral contraceptives; CYP3A, cytochrome P450 3A; DRSP, drospirenone; EE, ethinyl estradiol; GMR, geometric mean ratio; LNG, levonorgestrel; NET, norethindrone; PK, pharmacokinetic.

^aNetupitant is defined as a moderate CYP3A4 inhibitor based on a clinical drug‐drug interaction study of midazolam with a 300 mg single dose of netupitant. In addition, this study was conducted with a fixed combination drug of netupitant and palonosetron, but palonosetron has been known to have no inhibitory or inducible potential of CYP3A.²⁰

The effects of CYP3A inhibitors on the systemic exposure of progestins are presented in Figure 3 and Table 1.13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26 Strong CYP3A inhibitors, voriconazole and indinavir, increased the AUC of NET by 53% and 26%, respectively, whereas two strong CYP3A inhibitors, telaprevir and boceprevir, did not significantly affect the exposure of NET.13, 15, 16, 17 Coadministration of telithromycin, increased the AUC of LNG by 50%.28 The exposure of DRSP was increased by 100% and 170% when coadministered with boceprevir and ketoconazole, respectively.14, 17, 18, 29

Moderate CYP3A inhibitors showed a modest impact on the exposure of LNG and NET, as demonstrated by <50% increase in the AUCs. Fluconazole, netupitant, and faldaprevir, moderate CYP3A inhibitors, increased the AUC of LNG by ∼25% to up to 40%.21, 23, 27 Unexpectedly, atazanavir, a moderate CYP3A inhibitor, increased the AUC of NET by 110%.22 Weak CYP3A inhibitors seemed to have a negligible effect on the exposure of NET, LNG, and DRSP.24, 25, 26

Effect of CYP3A inducers on the exposure of EE and progestins

Strong CYP3A inducers, including rifampin, carbamazepine, and phenytoin, reduced the exposure of EE by 38–66% (Figure 4 and Table 2, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45). Bosentan and eslicarbazepine, moderate CYP3A inducers, decreased the AUC of EE by 31% and 42%, respectively.30, 31 Efavirenz (a moderate CYP3A inducer) had only 10% reduction in EE exposure, as opposed to a marked effect on progestins, LNG and NGM (Figure 5).32, 33 Oxcarbazepine, which is known as a weak CYP3A inducer, decreased EE exposure by 47%34, 35; nevirapine, a weak inducer, decreased EE exposure by 27%46; perampanel, also a weak inducer, at 4–12 mg daily treatment, did not significantly affect the AUC of EE.36

Table 2.

Effect of CYP3A inducers on the AUC of EE, NET, LNG, or NGM

CYP3A inducer		COC			AUC ratio of EE or progestins (GMR, 90% CI)
Drug	Dose	Drug(s)	Dose on the PK study day	No. of subjects	EE	Progestin
Rifampin (1st)39	300 mg q.d. for 10 days	EE/NET multiple doses for 21 days	EE 0.035 mg/NET 1 mg	22	0.36a	0.4a
Rifampin (2nd)40	600 mg q.d. for 14 days	EE/NET multiple doses for 21 days	EE 0.035 mg/NET 1 mg	8	0.34a	0.49a
Carbamazepine (1st)41	600 mg q.d. for 21 days	EE/NET multiple doses for 21 days	EE 0.035 mg/NET 1 mg	7	0.62 (0.54–0.71)	0.5 (0.4–0.61)
Carbamazepine (2nd)44	600 mg q.d. for 2 months	EE/LNG multiple doses for 2 months	EE 0.020 mg/LNG 0.1 mg	10	0.55a	0.56a
St John's wort (1st)38	300 mg t.i.d. for 28 days	EE/NET multiple doses for 21 days	EE 0.020 mg/NET 1 mg	16	0.86a	0.88a
St John's wort (2nd)37	300 mg t.i.d. for 28 days	EE/NET multiple doses for 21 days	EE 0.035 mg/NET 1 mg	12	0.68 (0.14–1.23)	0.88 (0.76–1.0)
Phenytoin42	200–300 mg for 8–12 weeks	EE/LNG single dose	EE 0.050 mg/LNG 0.250 mg	6	0.51a	0.58a
Efavirenz33	600 mg q.d. for 14 days	EE/NGM multiple doses for 21 days	EE 0.035 mg/NGM 0.25 mg	28	0.90 (0.80–1.01)	0.36 (0.33–0.38)
Bosentan31	125 mg b.i.d. for 7 days	EE/NET single dose	EE 0.035 mg/NET 1 mg	17	0.69 (0.60–0.80)	0.86 (0.76–0.97)
Efavirenz32	600 mg q.d. for 14 days	LNG only 2 doses	LNG 0.75 mg	17	NA	0.42 (0.36–0.48)
Eslicarbazepine30	1,200 mg q.d. for 15 days	EE/LNG single dose	EE 0.030 mg/LNG 0.15 mg	19	0.58 (0.55–0.62)	0.64 (0.56–0.72)
Oxcarbazepine(1st)34	600 mg q.d. for 2 days, then 900 mg q.d. for 2 days, then 1,200 mg q.d. for 18 days, then 600 mg q.d. for 2 days, then 300 mg q.d. for 2 days	EE/LNG multiple doses for 21 days	EE 0.050 mg/LNG 0.25 mg	16	0.53	0.53
Oxcarbazepine (2nd)35	300 mg q.d. for 1 day, then 300 mg b.i.d. for 1 day, then 300 mg t.i.d. for 28 days	EE/LNG multiple doses for 21 days	EE 0.030 mg/LNG 0.125 mg	13	0.53a	0.64a
Rifabutin40	300 mg q.d. for 14 days	EE/NET multiple doses for 21 days	EE 0.035 mg/NET 1 mg	12	0.65a	0.87a
Nevirapine45	200 mg q.d. for 14 days, then 200 mg b.i.d. for 14 days	EE/NET single dose	EE 0.035 mg/NET 1 mg	10	0.73 (0.54–1.02)	0.79 (0.67–0.96)
Perampanel (1st)36	2 mg q.d. for 7 days, then 4 mg q.d. for 21 days	EE/LNG multiple doses for 21 days	EE 0.030 mg/LNG 0.15 mg	20	0.96 (0.91–1.02)	0.93 (0.87–1.00)
Perampanel (2nd)36, b	4 mg q.d. for 7 days, then 8 mg q.d. for 7 days, then 8 mg q.d. for 21 days	EE/LNG single dose	EE 0.030 mg/LNG 0.15 mg	12	0.99 (0.93–1.06)	0.91 (0.82–1.01)
Perampanel (3rd)36, b	4 mg q.d. for 7 days, then 8 mg q.d. for 7 days, then 12 mg q.d. for 21 days	EE/LNG single dose	EE 0.030 mg/LNG 0.15 mg	8	1.05 (0.97–1.14)	0.60 (0.52–0.68)
Lersivirine43	1000 mg q.d. for 10 days	EE/LNG multiple doses for 10 days	EE 0.030 mg/LNG 0.15 mg	14	1.10 (0.92–1.31)	0.87 (0.78–0.97)

AUC, area under the curve; CI, confidence interval; COC, combined oral contraceptives; CYP3A, cytochrome P450 3A; EE, ethinyl estradiol; GMR, geometric mean ratio; LNG, levonorgestrel; NA, not available; NET, norethindrone; NGM, norgestimate; PK, pharmacokinetic.

^aThe ratio was calculated using the arithmetic mean or arithmetic median of the estrogen and progestins reported in the study result. ^bCOC drug‐drug interaction study with perampanel was split into two different dosing regimens due to the tolerability of perampanel (4 mg q.d. for 7 days, 8 mg q.d. for 7 days, then 12 or 8 mg q.d. for 21 days).

St John's wort extract, a dietary supplement that exhibited moderate to strong induction on CYP3A4, decreased EE exposure by 14–32%.37, 38

As shown in Figure 5 and Table 2,30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45 strong CYP3A inducers, rifampin and carbamazepine, decreased the AUC of NET by 50–60%.39, 40, 41 Carbamazepine and phenytoin decreased the AUC of LNG by 44% and 42%, respectively.42 Moderate CYP3A inducers, efavirenz and eslicarbazepine, decreased LNG exposure by 58% and 37%, respectively.30, 32 Efavirenz also significantly reduced the exposure of norelgestromin (NGMN), a major active metabolite of NGM, by 64%.33 Bosentan, a moderate CYP3A inducer, did not show a significant effect on the exposure of NET.31 The effect of weak CYP3A inducers on progestins varied. Two studies demonstrated that oxcarbazepine significantly decreased the exposure of LNG by 47% and 36%, respectively. Rifabutin and nevirapine showed a small reduction in the AUC of NET (∼10–20%).40, 46, 47 Perampanel demonstrated a dose‐dependent effect on the change of the exposure of LNG.48 At the dose of 12 mg per day for 21 days, perampanel decreased the AUC of LNG by 40%, whereas the doses at 8 mg and 4 mg per day had no significant impact on LNG exposure. Lersivirine, another weak CYP3A inducer, slightly decreased the AUC of LNG.43

Two studies demonstrated that St John's wort extract resulted in a 12% reduction on the exposure of NET.37, 38

DISCUSSION

CYP3A inhibition showed limited impact on the systemic exposure of EE and progestins

The current assessments showed that EE and progestins, such as NET and LNG, are minimally sensitive to CYP3A inhibition. The AUC changes of EE and progestins in the presence of a strong CYP3A inhibitor were less than twofold, except for DRSP. Mild inhibitory effect of CYP3A inhibitors on the PK of EE and progestins suggests that CYP3A‐mediated oxidation may have limited contribution to the overall disposition of these steroid hormones.

EE is extensively metabolized, primarily through intestinal sulfation and hepatic oxidation, glucuronidation and sulfation.3 The oxidative metabolism accounts for the elimination of 30% of EE dose and is catalyzed mainly by CYP3A (67%) and to a minor extent by CYP2C9 (23%).49, 50, 51 In addition to CYP‐mediated biotransformation, other alternate elimination pathways are glucuronidation by UDP‐glucuronosyltransferase 1A1 (UGT1A1) and sulfation by sulfotransferase 1E1 (SULT1E1; Table 3, 5, 14, 17, 29, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58). Strong CYP3A inhibitors resulted in no (<1.25‐fold) or small increases (<twofold) in EE exposure, suggesting that CYP3A does not contribute significantly to the elimination of EE. Interestingly, two strong CYP3A inhibitors, boceprevir and telaprevir, decreased the exposure of EE. The mechanisms contributing to the decreased exposure of EE remain unclear. The in vitro data indicated that boceprevir and telaprevir have very low or no induction potential for CYP3A.16, 18, 59 Therefore, the reduced EE concentrations are unlikely to be explained by CYP3A induction from boceprevir or telaprevir. The terminal half‐life of EE was not changed much when boceprevir was coadministered (14 h and 15 h for EE with and without boceprevir, respectively).17 Based on this observation, it seems that the drug interaction with boceprevir mainly affects the intestine metabolism/efflux (fg) of EE. The EE is a highly permeable drug and its cellular uptake is primarily driven by passive diffusion.3 Although uptake transporters seem to play a minimal role in EE cellular transport, it is possible that EE interacts with efflux transporters, as indicated by the Caco‐2 results. In particular, in vitro data showed that EE is a substrate of P‐glycoprotein (P‐gp), breast cancer resistance protein (BCRP), and multidrug resistance‐associated protein (MRP)‐2.3 Boceprevir has been reported as an in vitro P‐gp inhibitor with limited in vivo inhibition on digoxin clearance.18 Nonetheless, transporter‐mediated drug interactions still cannot explain the findings from the DDI study with boceprevir, as reduced intestine efflux of EE arising from boceprevir‐mediated inhibition of P‐gp would be anticipated to increase EE exposure, rather than decrease EE exposure.

Table 3.

Metabolic pathways of EE, NET, LNG, DRSP, and NGM^a

	Phase I metabolism	Phase II metabolism
EE	CYP3A (major), CYP2C9 (minor)49, 50, 51	Sulfation (SULT1E1),52 glucuronidation (UGT1A1)53
NET	CYP3A (major), CYP2C19 (minor), reduction4, 5, 54	Sulfation, glucuronidation5
LNG	CYP3A, reduction5, 55	Sulfation, glucuronidation5, 56, 57
DRSP	CYP3A,14, 17 reduction29	Sulfation, glucuronidation29
NGM (NGMN)b	CYP3A58	Glucuronidation (UGT1A1)58

CYP3A, cytochrome P450 3A; DRSP, drospirenone; EE, ethinyl estradiol; LNG, levonorgestrel; NET, norethindrone; NGM, norgestimate; NGMN, norelgestromin; SULT1E1, sulfotransferase 1E1; UGT1A1, UDP‐glucuronosyltransferase 1A1.

^aInformation on metabolic pathway of EE and progestins is limited and this table is based on the current knowledge.

^bNGM is rapidly converted to active metabolite NGMN via esterase in the gut and liver. Metabolism of NGMN is not well characterized, with some literature evidence suggesting the involvement of UGT1A1 or CYP3A or both.

The effect of moderate CYP3A inhibitors on the exposure of EE varied. Interestingly, the effect of atazanavir, a moderate CYP3A inhibitor, as specified by the FDA's draft DDI guidance, on the exposure of EE seemed to be unexpectedly stronger than that of strong CYP3A inhibitors except for voriconazole. It has been reported that the glucuronidation clearance pathway is around 35% of the total clearance of EE and EE 3‐O‐glucuronidation via UGT1A1 is considered to be the major hepatic UGT pathway.3, 60 Considering atazanavir is known to be a UGT1A1 inhibitor,8, 22 the increase in EE exposure in the presence of atazanavir may be partially due to inhibition of UGT1A1.

Metabolic characteristics of progestins have not been well addressed. It was proposed that NET and LNG undergo extensive reduction of α, β‐unsaturated ketone of ring A in their steroidal structures forming reduced, and, to a lesser extent, hydroxylated metabolites. The parent drugs and their metabolites can be conjugated, forming sulfated and glucuronidated products, which are excreted primarily in urine, and also in feces5, 61 (Table 3, 5, 14, 17, 29, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58). In vitro studies showed that NET is a substrate of CYP3A and the oxidation of NET was inhibited in the presence of ketoconazole, a strong inhibitor of CYP3A.4 Although there is no dedicated study suggesting whether CYP‐mediated biotransformation reactions may be of minor relevance compared with reduction and conjugation, the data from current assessment seem to indicate that CYP3A does not play a major role in NET metabolism. In particular, strong CYP3A inhibitors exhibited mild (<50% increase) or no impact on NET exposure. The greatest increase in the exposure of NET was observed (twofold increase) when concomitantly treated with atazanavir, a moderate CYP3A inhibitor. In addition to CYP3A inhibition, atazanavir‐mediated inhibitory effect on UGT1A1 may contribute to this interaction.62, 63, 64 Thus, observation of more pronounced effects of atazanavir on the systemic exposure of NET may be explained by dual inhibition of CYP3A and UGT1A1.

Interestingly, boceprevir and telaprevir also slightly decreased NET exposure, even though they exert a strong inhibitory effect on the metabolism of other CYP3A substrates, such as midazolam. The variable and inconsistent inhibition effect of strong and moderate CYP3A inhibitors on NET may suggest that CYP3A‐mediated metabolism is unlikely to be the only major pathway in the clearance of NET and other pathways are likely to also be involved.

Mild impact of telithromycin, a strong CYP3A inhibitor, on the exposure of LNG also indicates a limited contribution of CYP3A on the metabolism of LNG. This is consistent with the findings from in vitro studies, in which CYP‐medicated biotransformation seemed to be less important compared with reduction and conjugation.65

An in vitro study showed that the biotransformation of DRSP is mainly mediated by reductases, sulfotransferases,29 and CYP 3A contributes only to a minor extent to the metabolism of DRSP (<10%).14, 29 When considering this metabolic fate of DRSP, it is anticipated that an inhibitory effect on CYP3A would have little influence on the exposure of DRSP. However, the multiple dosing of a strong CYP3A inhibitor, ketoconazole, increased the AUC of DRSP by 2.7‐fold. The half‐life of DRSP was also prolonged when ketoconazole was coadministered.14 Another strong CYP3A inhibitor, boceprevir, increased DRSP exposure by twofold. In vitro data showed that boceprevir is a reversible time‐dependent inhibitor of CYP3A, but not an inhibitor of other major CYPs, or of UGT1A1 and UGT2B7.66, 67 In addition, there has been no clinical evidence that ketoconazole or boceprevir inhibit other metabolic enzymes of DRSP, including reductases and sulfotransferases based on our literature search. Therefore, contradictory to the in vitro data, the results of clinical DDI studies with strong CYP3A4 inhibitors ketoconazole and boceprevir indicated the contribution of CYP3A to the metabolism of DRSP. Interestingly, in a similar DDI study design, concomitant dosing of boceprevir showed no impact on NET exposure. Compared with NET and LNG, the significant and consistent increase in the systemic exposure of DRSP by strong CYP3A inhibitors may imply that CYP3A plays a more important role in the metabolism of DRSP than that of NET and LNG.

CYP3A inducers showed pronounced effect on the systemic exposure of EE and progestins

Although the impact of CYP3A inhibition on the metabolism of EE, LNG, and NET is limited, the current survey demonstrated that the exposure of LNG, NET, and EE was significantly influenced by concomitant dosing of CYP3A inducers. In particular, strong CYP3A inducers, such as carbamazepine, rifampin, and phenytoin, and moderate inducer, efavirenz, led to a >50% reduction in the exposure of these COC components. It is noted that these CYP3A inducers have the induction potential of multiple drug metabolizing enzymes, including phase I (CYPs) and phase II (UGT1A1 and SULT1A) enzymes by activating the pregnane X receptor and/or the constitutive androstane receptor.68, 69, 70, 71, 72, 73, 74 In addition, nuclear receptors pregnane X receptor and constitutive androstane receptor can mediate the regulation of some aldo‐keto reductase, which may play an important role in the biotransformation of NET, LNG, and DRSP.75 Therefore, significant decrease in the exposure of EE and progestins in the presence of these inducers is likely attributed to induction on multiple metabolizing enzymes of steroid hormones including CYP3A, aldo‐keto reductase, UGT, and SULT.

St John's wort has been known to be a moderate to strong inducer of CYP450 enzymes (particularly CYP3A) and/or transporter proteins, such as P‐gp.76 It was reported that consumption of St John's wort may lead to a contraception failure.77 The AUC decrease following pretreatment with St John's wort for both NET and EE in two studies was no more than 32%,37, 38 suggesting a minor change. It is noted that the induction effect of St. John's wort on CYP enzymes varies widely and could be preparation/formulation‐dependent.8, 76, 78

LIMITATIONS OF THE SURVEY

The current survey may have several limitations in interpreting the results. The DDI findings were collected from various studies, which had different study designs and treatment dosages of HC and perpetrators. The dose and dosing regimen (single dose vs. multiple doses) of perpetrators may play an important role in determining the magnitude of induction or inhibition effect.48 In addition, some studies included in this survey might have limitations for generalization, such as inadequate sample size and insufficient treatment period. Furthermore, this study did not consider the perpetrators’ inhibition or induction potential on drug transporters that may also be involved in the disposition of EE or certain progestins. For example, in vitro data showed that efflux transporters, including P‐gp, BCRP, and MRP‐2, play a role in EE efflux and EE 3‐O‐glucuronide (a major metabolite of EE) is a substrate of MRP2 and BCRP.3 However, there have been no dedicated studies examining the impact of these transporters on the overall disposition of EE and progestins.

SUMMARY OF THE SURVEY

This assessment showed that coadministration of strong and moderate CYP3A inhibitors led to a modest increase in the systemic exposure of EE, LNG, and NET with the exception of DRSP, suggesting that the contribution of CYP3A in the metabolism of EE, NET, and LNG is not predominant. DRSP seemed to be more sensitive to CYP3A inhibition compared with NET and LNG. In contrast, strong CYP3A inducers seemed to impose a marked reduction in the systemic exposure of EE, LNG, NET, and NGM (>40% decrease). It may be explained by the possibility that CYP3A inducers used for those studies had induction potential of multiple drug metabolizing enzymes including phase II (UGT1A1 and SULT 1A) in addition to phase I (CYPs) enzymes and/or transporters involved in the disposition of EE and progestins. Characteristics of inhibition or induction spectrum of perpetrators on non‐CYP metabolic pathways (e.g., glucuronidation and sulfation) and/or transporters should be considered in predicting and interpreting the overall DDI potential with COCs.

As observed from this survey, the same perpetrator had different effects on different progestins (in combination with EE), which pose a challenge in extrapolating the study findings of one specific progestin containing hormonal contraception to the other products. The results may also shed a light on the selection of progestins for DDI assessment based on perpetrators’ characteristics. More research is needed to understand the metabolic and transporter pathways of EE and progestins as well as relative contribution of each pathway in their clearance. Such information could enable us to utilize physiologically based PK modelling to predict the DDI potential with various HCs and routes of administration.8

Conflict of interest

The authors declared no conflict of interest.