Pharmacokinetics (PK) of ethinylestradiol/levonorgestrel co-administered with atazanavir/cobicistat

Abstract

Background and objectives: Access to safe and reliable contraception in the context of ARVs is essential. This study aimed to investigate the steady-state pharmacokinetics (PK) of ethinylestradiol/levonorgestrel (EE/LNG) 30/150 μg (Microgynon^®) and atazanavir/cobicistat (ATV/COBI) 300/150 mg (Evotaz^®), co-administered in HIV negative female volunteers, and assess its safety and tolerability.

Methods: This phase 1, open label, 57-day, cross over, PK study randomized participants to one of two groups: (i) group 1 received EE/LNG alone on days 1–21, EE/LNG (21 days) + ATV/COBI (14 days) in the co-administration phase (days 22–42) and ATV/COBI alone on days 43–56; (ii) group 2 followed the same sequence but started with ATV/COBI and concluding with EE/LNG. Each group underwent intensive PK sampling on days 14, 35, and 56. EE/LNG and ATV/COBI concentrations were measured using validated LC-MS/MS methods.

Results: Of 14 healthy female volunteers screened, 11 attended baseline and six completed all PK phases (five withdrew secondary to side effects). Paired data were available for analysis in six subjects for EE/LNG and eight for ATV/COBI. Geometric mean ratios (GMR, with versus without ATV/COBI) and 90% confidence intervals (CI) for LNG C_max, AUC_0-24, C₂₄ were 0.83 (0.68–1.02), 0.92 (0.71–1.18), 1.01 (0.73–1.38). GMR and 90% CI for EE C_max, AUC_0-24, C₂₄ were 1.05 (0.92–1.19), 1.01 (0.83–1.22), 0.75 (0.60–0.93). No grade 3 or 4 adverse events or laboratory abnormalities were observed in the women who completed the study.

Conclusions: Our findings showed no significant changes in LNG concentrations and a 25% decrease in EE C₂₄ when EE/LNG was co-administered with ATV/COBI.

Keywords:

• Atazanavir
• cobicistat
• oral contraceptive
• levonogestrel
• ethinylestradiol

Introduction

Women account for slightly more than half of the world’s 36.7 million people living with HIV/AIDS and the majority are of childbearing age.¹ Early and sustained HIV viral load suppression with antiretroviral (ARV) therapy now enables longer, healthier lives and improved fertility in women living with HIV (WLWH).^2–4 A significant number do, however, report requiring the flexibility to plan or prevent pregnancy,^3,4 meaning that access to safe and reliable contraception in the context of ARVs is critical.^5,6 The combined oral contraceptive pill (COCP) is a preferred method of contraception for many women worldwide.^7,8 Nevertheless, WLWH have often been unable to use the COCP due to drug–drug interactions (DDIs) with some ARVs leading to altered drug levels and adverse events.⁹ For instance, a reduction in the progestogen component of the COCP potentially runs the risk of contraceptive failure and unwanted pregnancy⁹ whilst overexposure can lead to a number of side effects such as increased appetite, fluid retention, acne, and headaches.¹⁰ Meanwhile, changes in oestrogen pharmacokinetics (PK) impact tolerability (e.g. breakthrough bleeding with underexposure, which can impact adherence) and toxicity (e.g. thrombotic events with increased concentrations).^11–13 There is now good evidence that an undetectable viral load equates to HIV being un-transmittable¹⁴ and, hence, many HIV serodifferent couples choose to rely on hormonal contraception alone without condoms.^3,15 It is therefore essential to define the pharmacokinetics and pharmacodynamics (PD) of hormonal contraception co-administered with individual ARVs in order to inform guidelines and ensure that the efficacy and safety of both are maintained.¹¹

Furthermore, exogenous oestrogens are also used by transgender women (TGW) as feminizing therapy.¹⁶ TGW carry a disproportionate burden of HIV infection (estimated global prevalence 19.6% and incidence 3.4–7.8 per 100 person-years worldwide)^17,18 and are less likely to engage into care or being compliant with ART.^19,20 Qualitative studies have attributed this, in parts, to a limited understanding of the interactions between some ARVs and gender affirming hormones and to fears amongst TGW that ARVs will impede on hormonal efficacy, further underscoring the importance of having evidence-based data on DDIs.^16,21 Ethinylestrodiol (EE), specifically, is not recommended for feminization therapy in guidelines;²² however, it is the most studied oestrogen and the reference agent for drug interaction guidance¹⁶ and it is often used by TGW outside of healthcare settings for medical transition.²³

Despite very recently being removed from some major guidelines,^24–26 boosted atazanavir (ATV/b) remains the preferred second-line ARV agent in the World Health organization (WHO) guidelines and large numbers of patients, stable on treatment, continue therapy with ATV/b worldwide.²⁷ With its long-standing experience history, high genetic barrier and once daily dosing (OD), it remains an instrumental option in the management of HIV.²⁸ For over a decade, ATV was mainly co-administered with the only pharmacological booster available, ritonavir (RTV), a potent CYP3A4 inhibitor. Hormonal contraceptives are extensively metabolized by CYP enzymes (CYP3A and CYP2C9/19) and drug interactions between the COCP and ATV (with or without RTV) have been demonstrated.²⁹ The progestogens studied to date, norgestimate (NGM) and norethisterone (NET), are both increased in exposure with unboosted and with ritonavir-boosted ATV, through the inhibition of CYP3A4-mediated progestogen metabolism by both agents, thereby potentially leading to the side effects described above.^30,31 Similarly ethinylestradiol concentrations also increase with unboosted ATV (48% rise in area under the curve, AUC) but, in contrast, decrease with ATV/RTV (16% decrease in EE maximum concentration (C_max), 19% in AUC, and 37% in minimum concentration (C_min)),³¹ which is thought to result from RTV’s concomitant induction of CYP2C9 and of the glucuronidation responsible for EE clearance (via uridine-diphosphate glucuronosyltransferase 1A1–UGT1A1).²⁹ Therefore, according to guidelines and the atazanavir summary of product characteristics, if a combined oral contraceptive is administered with ATV/RTV, it must contain at least 35 μg of EE, strict compliance is necessary and a second method of contraception is recommended, considering the unknown effect of drug interactions.^31–33

Cobicistat (COBI), is the more recently approved alternative pharmacological booster to RTV. Its molecular properties offer the opportunity for co-formulation with PIs into fixed dose combinations³⁴ and it is available co-formulated with ATV as Evotaz^®.³⁵ It has no ARV activity and is a potent CYP3A4 inhibitor but, unlike ritonavir, it is neither a UGT1A1 nor a CYP2C9 inducer³⁴ and it may be better tolerated in some patients than RTV.³⁴ The product insert for Evotaz^®35 and the University of Liverpool’s HIV drug interactions website³⁶ state that no dosing recommendations can be made for COCP co-administration with ATV/COBI due to a lack of published data. It is currently suggested that additional or alternative (non-hormonal) forms of contraception should be considered.

ATV/COBI co-administration with the COCP has been investigated in an unpublished, phase I drug interaction study (Majeed et al, IWCPAT, Chicago 2017), which only assessed a single dose of drospirenone (DRSP)/EE 3 mg/20 μg administered prior to and at the end of a 14-day course of standard dose ATV/COBI in healthy volunteers (n = 14). The investigators found a 130% increase in DRSP AUC_0-∞ and a 12% increase in C_max. There was a nonclinically significant decrease in EE (22% reduction in AUC_0-∞ and 18% in C_max). The minimum concentration within the standard dosing interval of 24 h (C₂₄) was not reported.³⁷ No COCP C₂₄ or “steady state” (real life use) PK data are available; the latter are important since, progestogen serum levels are 2 to 3 fold higher in the steady state compared to a single administration, (after approximately 8–10 days of treatment) and EE steady state concentrations increase by a 30–40% rise in plasma level (5–6 days post initiation).^13,38

EE/levonorgestrel (LNG) (Microgynon^®) is the leading COCP prescribed in the UK.³⁹ The aim of this study was therefore to investigate the steady state PK of EE/LGN 30/150 μg and ATV/COBI 300/150 mg (Evotaz^®) when co-administered in HIV negative female healthy volunteers and to assess the safety and tolerability of co-administration.

Participants and methods

Participants

Written informed consent was obtained from nonpregnant and nonlactating female healthy volunteers aged between 18 and 35 years with a body mass index (BMI) between 18 and 35 kg/m². Participants were excluded if they had any significant acute or chronic medical illness; abnormal physical examination, ECG or clinical laboratory determinations; positive screens for HIV, hepatitis B or C; current or recent (within three months) gastrointestinal disease; clinically relevant alcohol or drug use that the investigator felt would adversely affect compliance with trial procedures; exposure to any investigational drug or placebo within three months of the first dose of the study drug; use of any other drugs, including over the counter medications and herbal preparations, within two weeks of the first dose of the study drug; and previous allergy to any of the constituents of the pharmaceuticals administered during the trial. Women of childbearing potential required a negative pregnancy test at screening and baseline.

Study design

This was an open-label, crossover, 57-day (excluding screening and follow-up) phase 1 PK trial carried out at the Clinical Trial Unit of the St. Stephen’s Centre, Chelsea and Westminster Hospital, London, UK.

At screening, clinical assessment and routine laboratory investigations were performed in all participants. The safety and tolerability of study medications were evaluated throughout the trial (on days 7, 28, 49, on PK days and at follow-up) using the National Institute of Allergy and Infectious Diseases (NIAID) Division of AIDS table for grading the severity of adult and pediatric adverse events (2004), that characterizes abnormal findings, vital signs, physical examinations, and clinical laboratory investigations.

The study design is illustrated in Figure 1. After successful screening, volunteers were randomized to either (i) group 1, which received EE/LNG 30/150 μg alone on days 1–21, EE/LNG (21 days) + ATV/COBI 300/150 mg (14 days) in the co-administration period (days 22–42) and ATV/COBI alone on days 43–56 (14 days) or (ii) group 2, which followed the same structured sequence but started with ATV/COBI alone and finished with EE/LNG alone (14 or 21 days, patient choice). Each group underwent intensive PK sampling on study days 14, 35, and 56 to measure plasma concentrations of EE/LNG and/or ATV/COBI at 0 (predose), 2, 4, 8, 12, and 24 h postdose. On the PK days, study staff witnessed study medication intake with a standardized breakfast (626 kcal) and 240 mL of water.

Figure 1 Study design.

Adherence

Adherence was assessed through direct questioning of dosing schedules, missed and late doses at each safety and PK visit. A pill count was also carried out at each PK visit.

Analytical and PK methods

Blood samples were collected into lithium heparin-containing blood tubes (12 mL) at each time-point, immediately inverted several times and then kept on ice or refrigerated until centrifugation. Within 30 min of blood collection, each blood sample was centrifuged for 10 min at 2000g at 4 °C. Plasma was then aliquoted equally into three 2 mL tubes (Sarstedt, Germany) and stored at –20 °C.

Samples were shipped on dry ice to the Liverpool Bioanalytical Facility for analysis. The laboratory is Good Clinical Laboratory Practice-accredited and participates in an external quality assurance scheme (KKGT, the Netherlands).^40,41

Quantification of levonorgestrel, ethinylestradiol, atazanavir, and cobicistat

Concentrations of LNG, EE, ATV, and COBI in plasma were measured using validated high-pressure liquid chromatography–tandem mass spectrometry methods (LC-MS/MS).^42–44 The lower limits of quantification (LLQ) for the plasma analyses were 0.240 ng/mL for LNG, 5 pg/mL for EE, 10 ng/mL for ATV, and 5 ng/mL for COBI. For concentrations below the assay limit of quantification, a value of one-half of the quantification limit was used.

Accuracy (percentage bias) was between –0.42% and 1.5% (LNG), 0.61% and 3.28% (EE), 4.70% and 6.36% (ATV), and 6.45% and 8.07% (COBI) and precision was less than 3.1% (LNG), 8.0% (EE), 6.3% (ATV), and 8% (COBI).

Data analysis

The calculated PK parameters for plasma LNG, EE, ATV and COBI were plasma C₂₄, C_max and AUC from 0 to 24 h (AUC_0–24). All PK parameters were calculated using actual blood sampling time and noncompartmental modeling techniques (WinNonlin Phoenix, version 6.1; Pharsight, Mountain View, CA). Descriptive statistics, including geometric mean (GM) and 95% confidence intervals (95% CI) were calculated for LNG, EE, ATV, and COBI plasma PK parameters. Each drug PK parameter during the co-administration period was compared to the unaccompanied drug PK parameter by calculating GM ratios (GMR) and 90% CI (co-administered/alone).

Inter-individual variability in drug PK parameters was expressed as a percentage coefficient of variation [CV, (standard deviation/mean)×100].

Statistical power

This is an exploratory study and, as such, no formal sample size calculation was performed. Sixteen participants completing the study were deemed appropriate to allow for relevant conclusions, and is standard for PK studies.

Ethics

The study protocol was approved by the Westminster Research Ethics Committee, London, United Kingdom, as well as by the Medicines and Healthcare products Regulatory Agency (MHRA) in the United Kingdom. The study was conducted according to Good Clinical Practice and the Declaration of Helsinki (NCT02697851).

Results

Study population

Fourteen healthy female volunteers were screened, 13 were enrolled and 11 attended baseline (two withdrew prior to baseline for personal reasons). Nine subjects completed the intensive PK day 14 (five in group 1, four in group 2), eight completed day 35 (four in each group) and six completed day 56 (four in group 1, two in group 2). Overall, seven participants withdrew consent; five because of adverse events and two before starting the study medications. Of the six participants who completed all PK phases (up to day 56), the median (range) age and BMI were 31 (19–35) years, and 24 (19–29) kg/m², respectively. Of the eleven participants who attended the baseline visit, four participants described themselves as Caucasian, six as black African, and one as Hispanic. Participant demographics, withdrawals and withdrawal reasons are summarized in Table 1.

Table 1 Participant demographics, withdrawals and withdrawal reasons

ID	Age	Ethnicity	BMI	Group	Study stage reached	Reason for withdrawal
101	31	Hispanic	21.1	G1	Completed	N/A
102	27	White	23.7	G2	Between D7 and D14	Rash
103	30	African	29.3	G1	Completed	N/A
104	33	White	22.1	G2	Completed	N/A
105	30	African	26.8	G1	Between D28 and D35	Deranged LFTs
106	28	Turkish	23.1	G1	Before baseline	First dose not given
107	35	White	26.9	G1	Completed	N/A
108	23	White & black Caribbean	27.4	G2	Between D35 and D43	Moderate nausea, vomiting, orthostatic hypotension, diarrhoea
109	18	White	20.7	G2	Between Baseline and D7	Stopped taking study medication due to side effects, not specified
110	27	White	25.3	G2	Completed	N/A
111	19	White	19.0	G1	Completed	N/A
112	25	White & black Caribbean	22.4	G2	Between D35 and D43	No cause given
113	29	White	27.6	G1	Before baseline	First dose not given

ID = study identification, BMI = body mass index, G = group, D = day, n/a = not available.

Note: N/A: not applicable; G1: Group 1; G2: Group 2; D: day. Highlighted rows in gray = participants who completed the study. Drug sequences:

Group 1: EE/LNG (Microgynon) alone on days 1–21, EE/LNG (21 days) + ATV/COBI (14 days) in the co-administration phase (days 22–43) and ATV/COBI alone on days 43–56.

Group 2: ATV/COBI alone on days 1–21, EE/LNG (21 days) + ATV/COBI (14 days) in the co-administration phase (days 22–43) and EE/LNG alone on days 43–56.

Pharmacokinetic results

Pharmacokinetic data for all four drugs in the two separate groups are detailed in Table 2. For each drug, only paired data from participants who completed the first period (either EE/LNG or ATV/COBI alone) and the co-administration period were used, so as to be able to assess and summarize intra-individual PK changes between the two periods. Groups 1 and 2 (G1 and G2) are grouped together in the results described below.

Table 2 Summary of pharmacokinetic data for all four drugs

	GM Cmax (90%CI) (ng/mL)					GM AUC0-24 (90% CI) (ng*h/mL)					GM C24 (90% CI) (ng/mL)
	Alone	CV%	Combined	CV%	GMR	Alone	CV%	Combined	CV%	GMR	Alone	CV%	Combined	CV%	GMR
LNG (n = 6)
Total	6.5 (3.5–9.6)	60	5.4 (3.7–7.2)	43	0.83 (0.68–1.02)	92 (41–143)	67	84 (57–111)	43	0.92 (0.71–1.18)	2.83 (1.2–4.5)	67	2.85 (1.9–3.8)	45	1.01 (0.73–1.38)
Group 1 (n = 4)	8.8 (5.9–12)	46	7.2 (6.1–8.3)	23	0.82 (0.61–1.11)	130 (81–180)	50	111 (93–129)	23	0.85 (0.64–1.12)	4.2 (2.7–5.7)	47	3.8 (3.2–4.5)	25	0.90 (0.64–1.27
Group 2 (n = 2)	3.6 (1.5–5.7)	47	3.09 (2.1–4.1)	26	0.86 (0.66–1.12)	45 (3.2–87)	69	48 (36–61)	23	1.07 (1.74–2.69)	1.3 (–0.05 to 2.6)	75	2.6 (1.0–4.2)	59	0.9 (0.64–1.27)
EE (n = 6)
Total	40 (28–53) × 10^–3	43	42 (32–52) × 10^–3	34	1.05 (0.92–1.19)	407 (259–555) × 10^–3	49	410 (318–501) × 10^–3	32	1.01 (0.83–1.22)	7.4 (3–12) × 10^–3	72	5.6 (3.9–7.3) × 10^–3	42	0.75 (0.6–0.93)
Group 1 (n = 4)	43 (28–58) × 10^–3	47	46 (32–60) × 10^–3	35	1.07 (0.88–1.3)	446 (281–612) × 10^–3	50	440 (335–544) × 10^–3	33	0.98 (0.77–1.26)	8.2 (3–13) × 10^–3	76	5.5 (3.7–7.3) × 10^–3	45	0.67 (0.50–0.89)
Group 2 (n = 2)	36 (29–42) × 10^–3	26	36 (32–39) × 10^–3	15	1 (0.88–1.13)	338 (202–475) × 10^–3	55	356 (302–411) × 10^–3	22	1.05 (0.7–1.59)	6.1 (3.2–9) × 10^–3	63	5.7 (3.5–8) × 10^–3	53	0.94 (0.82–1.08)
ATV (n = 4)
Total	4157 (3151–5163)	39	3133 (2472–3795)	34	0.75 (0.60–0.95)	48998 (37387–60608)	38	38293 (27206–49380)	45	0.78 (0.64–0.96)	1061 (614–1507)	62	949 (527–1371)	66	0.89 (0.72–1.11)
Group 1 (n = 4)	4447 (2692–6201)	45	2861 (2158–3565)	29	0.64 (0.46–0.90)	49455 (35411–634500)	33	33680 (24048–43312)	32	0.68 (0.51–0.91)	930 (348–1510)	65	787 (543–1031)	35	0.85 (0.59–1.21)
Group 2 (n = 4)	3886 (2692–5080)	36	3432 (2286–4577)	38	0.88 (0.65–1.21)	48544 (27777–69312)	48	43539 (23393–63684)	50	0.90 (0.69–1.17)	1211 (467–1954)	65	1145 (341–1949)	71	0.95 (0.71–1.27)
COBI (n = 4)
Total	1760 (1604–1917)	15	1554 (1325–1782)	25	0.88 (0.8–0.97)	16456 (13863–19049)	26	14054 (10703–17405)	39	0.85 (0.77–0.95)	70 (27–113)	84	62 (13–112)	101	0.89 (0.66–1.21)
Group 1 (n = 4)	1810 (1669–1952)	9	1414 (1173–1655)	20	0.78 (0.7–0.87)	15834 (13044–18624)	21	11820 (9572–14069)	23	0.75 (0.65–0.86)	49 (25–73)	53	44 (26–61)	44	0.89 (0.47–1.67)
Group 2 (n = 4)	1712 (1409–2014)	21	1707 (1332–2082)	26	1 (0.95–1.05)	17102 (12363–21842)	32	16710 (10942–22479)	40	0.98 (0.91–1.05)	99 (21–177)	78	89 (–4.7 to 182)	94	0.90 (0.74–1.09)

EE = ethinylestradiol, LNG = levonorgestrel, ATV = atazanavir, COBI = cobicistat, CV = coefficient of variation, GM = geometric mean, GMR = geometric mean ratio, CI = confidence interval, C_max = maximum concentration, AUC_0-24h = area under the curve from 0 to 24 h, C_24h = concentration at 24 h postdose.

Levonogestrel plasma pharmacokinetics

Six participants provided data for LNG intake with and without ATV/COBI (G1 n = 4, G2 n = 2). LNG GM plasma concentrations versus time are shown in Figure 2. Geometric mean ratios (with ATV/COBI versus without) and 90% CI for LNG C_max, AUC_0-24, and C₂₄ were 0.83 (0.68–1.02), 0.92 (0.71–1.18), and 1.01 (0.73–1.38) respectively.

Figure 2 Levonogestrel (LNG) GM (95% CI) plasma concentration versus time curves, with and without ATV/COBI, GM (95%CI) n = 6.

Ethinylestradiol plasma pharmacokinetics

Six participants provided data for EE intake with and without ATV/COBI (G1 n = 4, G2 n = 2). GM plasma concentrations versus time curves are shown in Figure 3. GMR (90% CI) for EE C_max, AUC_0-24, and C₂₄ were 1.05 (0.92–1.19), 1.01 (0.88–1.22), and 0.75 (0.60–0.93) respectively.

Figure 3 Ethinylestradiol EE GM (95% CI) plasma concentration versus time curves, with and without ATV/COBI, GM (95%CI) n = 6.

Atazanavir plasma pharmacokinetics

Eight participants provided data for ATV intake with and without EE/LNG (G1 n = 4, G2 n = 4). GMR (90% CI) of ATV C_max, AUC_0-24, and C₂₄ were 0.75 (0.60–0.95), 0.78 (0.64–0.96), and 0.89 (0.72–1.11) respectively.

Cobicistat plasma pharmacokinetics

Eight participants provided data for COBI intake with and without EE/LNG (G1 n = 4, G2 n = 4). GMR (90% CI) of COBI C_max, AUC_0-24, and C₂₄ were 0.88 (0.8–0.97), 0.85 (0.77–0.95), and 0.89 (0.66–1.21) respectively.

Table 3 summarizes our results together with currently reported findings in COCP interaction studies involving ATV/RTV and/or COBI.

Table 3 Summary of COCP drug interactions studies with atazanavir, ritonavir, and cobicistat

		ATV/COBI + EE/LNG (This study)	PI/RTV + NET^a (n = 16)	ATV/RTV + EE/NGM^b (n = 13)	TDF/FTC/EVG/COBI + EE/LNG^c (n = 20)	ATV/COBI + EE/DRSP^d (n = 14)
Progestogen	AUC	8%	50%	85%	110%	130% (AUC0-∞)
	Cmax	17%	33%	68%	60%	12%
	Cmin	1%	26%	102%	Not available	Not available
EE	AUC	1%	N/A	19%	20%	22% (AUC0-∞)
	Cmax	5%	N/A	16%	10%	18%
	Cmin	25%	N/A	37%	Not available	Not available

AUC = area under the curve; C_max = maximum concentration reached; C_min = trough concentration; ATV = atazanavir; COBI = cobicistat; EE = ethinylestradiol; LNG = levonogestrel; PI = protease inhibitor; RTV = ritonavir; NET: norethindrone; NGM = norgestimate; TDF = tenofovir; FTC = emtracitabine; EVG = elvitegravir; DRSP = drospirenone; N/A: not applicable.

Safety and tolerability

Five participants withdrew consent from the study secondary to side effects; of those, data on the reason are available for three and are listed in Table 1. Full adherence was confirmed through regular direct questioning and pill counts in the women who completed the study and no grade 3 or 4 adverse events or laboratory abnormalities were observed in this group.

Discussion

To our knowledge, this is the first study to investigate the steady state PK of Microgynon^® co-administered with ATV/COBI (Evotaz^®).

The hormonal components of the COCP undergo extensive first-pass metabolism by phase I and II microsomal enzymes in the small intestinal mucosa and the liver before reaching the systemic circulation, meaning that they are highly susceptible to DDI.^45,46

LNG is first hydroxylated in the liver, mainly by CYP3A4, and its metabolites are then excreted as glucuronide conjugates.⁴⁷ Our results showed only a small reduction in steady state LNG C_max (17%) and no changes in C₂₄ or AUC₀-₂₄ when it is administered with ATV/COBI (GMR range 0.92–1.01). Majeed et al. had also reported little interaction with progestogens, with only a 12% increase seen in DRSP C_max when co-administered as a single dose with ATV/COBI. Both studies therefore suggest that ATV/COBI has a lesser impact on progestogen peak concentrations than ATV/RTV (co-administered with NGM or NET, GMR 1.33–1.68)^30,31,37 and when NGM is co-administered with COBI as part of Stribild^® (tenofovir/emtricitabine/elvitegravir/COBI; GMR 2.08; Polina et al. 12th Clin Pharm of HIV Therapy Workshop 2011). Whilst the number of participants completing the relevant phases in our study (n = 6) was small and our data cannot be conclusive, a lack of clinically significant decrease in LNG minimum concentration and exposure is cautiously reassuring. This is because the progestogen-mediated suppression of the luteinizing hormone (LH) surge is one of the main contraceptive mechanisms of the COCP^48,49 and LNG is the progestogen contained in the most commonly prescribed form of the emergency contraceptive pill.⁴⁷

Additionally, the lack of meaningful rise in progestogen C₂₄ and AUC₀-₂₄ in our study differs to the increases seen with ATV/RTV co-administration (GMR 1.85 and 2.01 respectively) or with COBI co-administration when combined with DRV or EVG (GMR range 1.58–2.67). This finding is important since progestogen overexposure can lead to a number of significant side effects, as described above, including nausea, weight gain, and acne amongst others, that may impact tolerability and adherence.¹⁰

Our study showed that EE C₂₄ decreased by 25% with ATV/COBI compared to a 37% decrease reported with ATV/RTV. Furthermore, we found no clinically significant changes in C_max or AUC_0-24 (5% and 1% increases respectively), compared to the decreases seen with ATV/RTV (19% and 16% respectively).³¹ This may be explained by the fact that, unlike RTV, COBI does not induce UGT1A1 or CYP2C9, both of which are involved in EE clearance.³⁴ As the EE component of the COCP is mainly responsible for endometrial stability and a significant reduction in C₂₄ can lead to breakthrough bleeding potentially impacting adherence to contraception,⁹ smaller decreases in PK parameters are likely to optimize tolerability and adherence. Our findings also compared well to EE co-administration with DRV/COBI in the Majeed et al. study, which lead to a 30% and a 14% decrease in EE AUC_0-∞ and C_max respectively, secondary to DRV-mediated induction of CYP2C9 and CYP2C19.^29,37 In contrast, the induction of CYP2C9 and UGT through boosted EVG appears to overcome general COBI-medicated enzyme inhibition and leads to a decrease in EE PK parameters when it is co-administered with Stribild (GMR range 0.66–0.94; Polina et al. 12th Clin Pharm of HIV Therapy Workshop 2011).

As previously discussed, EE concentrations increase with unboosted ATV (AUC, C_max, and C_min 48%, 15%, and 91%, respectively), as do progestogen concentrations, through ATV mediated CYP3A4 and UGT1A1 inhibition. Yet, when co-administered with ATV/COBI in our study, EE AUC_0-24, EE C_max, LNG AUC_0-24, and LNG C₂₄ did not increase, whilst EE C₂₄ and LNG C_max even decreased. This is despite the lack of known COBI related enzyme or transporter induction (in vitro) to counteract ATV and COBI-mediated enzyme inhibition. This highlights that COBI’s in vivo metabolic effects may not yet be fully elucidated and warrant further investigation.

Of note, EE C_max levels were consistently lower in our study (36–46 pg/mL regardless of the study phase), than in a majority of other EE PK studies (mean levels 80 pg/mL following a 30ug EE dose).⁵⁰ This could be explained by the sampling schedule used in this study – EE T_max is usually reached at an average of 1.5 h; it is therefore possible that we missed the maximum concentration reached for some subjects by sampling 1- and 2-h postdose. Additionally, levels of EE have been shown to steadily increase within the first treatment cycle,³⁸ so concentrations measured at the end of the cycle (day 21) may be higher than mid-cycle (day 14), when our subjects underwent intensive PK sampling. It may also be higher again in subsequent cycles. This is however offset by our study design, which allowed for the intra-individual effect of co-administration to be measured. LNG PK parameters in our study were comparable to previous data.^10,13

ATV and COBI concentrations were slightly reduced during the co-administration period (Table 2). For ATV, the changes in both groups combined were small and within the no-effect boundary (GMR C_max was 0.79 and GMR C₂₄ 0.89). Interestingly, there was a significant decrease in ATV/COBI C_max and AUC_0-24 when combined with EE/LNG (GMR 0.64 and 0.68 respectively) in group 1 subjects only (i.e. those who received Microgynon30 alone for 21 days prior to combining with ATV/COBI) whereas no interaction was present with the group 2 subjects. Whilst this could be interpreted as being due to enzyme induction by EE, it is an observation and it is not possible to come to any conclusions based on the very small number of subjects in a subgroup (n = 4); C₂₄ remained unchanged. Importantly, ATV concentrations in both groups remained reassuringly above the in vivo suggested minimum effective concentration for wild type HIV (MEC = 150 ng/mL).²⁸

The target recruitment was not reached and five/11 participants (45%) withdrew from the study secondary to side effects, three of which disclosed information on the reason (rash, deranged liver function tests, and gastrointestinal symptoms). This is a higher dropout rate than seen in Majeed et al. (22%), although participants in that study had only received one dose of COCP. Whilst high, it is in keeping with the combined published discontinuation rates secondary to any AE for boosted atazanavir (15%) and the COCP (29%).^49,51 Both are drugs with relatively common side effects in the initial period and it is important that patients are counseled accordingly.

There are limitations to this study. The subjects were HIV negative healthy volunteers. As such, PK or pharmacodynamic comparisons with HIV positive women must be made cautiously and the practical implications of these PK observations are unknown. Clinical outcome data are required in large cohorts of HIV infected participants, and studies investigating pharmacodynamic endpoints (such as failure of viral suppression, HIV-related clinical disease progression, or unintended pregnancy) are needed in order to draw definite conclusions on how likely a contraceptive or an ARV is to fail in the context of co-prescription. It is also important to remember that efficacy rates of user-dependent contraception differ between perfect use (as seen in clinical studies) and real-life use.

The fact that our study involved healthy volunteers may also explain the low number of participants completing the study. In real-life settings, mild side effects associated with the initiation of COCP and/or of ARVs do not normally persist beyond three to six months, at which point an alternative would usually be offered. The option to persist and reassess was not available in the context of this study.

Nonetheless, our study provides the first steady state PK data on EE/LNG co-administered with ATV/COBI, demonstrating minimal changes in LNG concentrations and a smaller decrease in EE than seen with ATV/RTV. Whilst preliminary, these data are important in informing physicians, who need to discuss and choose safe and reliable contraception with their female patients living with HIV. Whether this minor difference between ATV/RTV and ATV/COBI will be clinically significant warrant further characterization in future studies.

Disclosure

ERE has received speaking and travel grants from Janssen, ViiV, Bristol-Myers Squibb, Merck Sharp & Dohme, and Gilead. EB, LE, and SDP have no conflicts of interest to declare. SK has received support from ViiV Healthcare, Merck, Janssen, Gilead and Bristol Myers Squibb for the HIV drug interactions website, and research grants from Merck, Janssen and ViiV Healthcare. GM has received speaker's and advisor's fees from Gilead Sciences, MSD, Janssen, BMS and has served as a member of the board of directors and on the scientific advisory board of Tobira Therapeutics. NN has received the following: Gilead Sciences (speaker fees, travel grants, conference registration, institutional research grant), BMS (speaker fees), Janssen (speaker and advisor fees, travel grants, conference registration), ViiV/GSK (speaker and advisor fees). MB had received travel and research grants from and has been advisor for Janssen, Roche, ViiV, Bristol-Myers Squibb, Merck Sharp & Dohme, Gilead, Mylan, Cipla, Teva.

Acknowledgments

The authors would like to thank the St Stephen’s Clinical Research team for their hard work and the volunteers who took part in the study. Part of this data was presented at the Conference on Retroviruses and Opportunistic Infections (CROI), 3 - 7 March 2018, Boston, MA, USA.

Additional information

Funding

This work was supported in part by a research grant from Bristol-Myers Squibb. Funding support was also provided by the St. Stephen’s AIDS Trust.

Notes on contributors

Emilie R Elliot

Dr Emilie R Elliot is an HIV Physician at Chelsea and Westminster Hospital, London, UK with a special interest in ARV pharmacokinetics, pharmacodynamics and pharmacogenomics. She is currently finalising a PhD in ARV pharmacology at the Institute of Translational Medicine, University of Liverpool, UK.

Elisa Bisdomini

Elisa Bisdomini, RN. Previously Senior Clinical Trial Coordinator and currently Inflammatory Bowel Disease Clinical Nurse Specialist at Chelsea and Westminster Hospital. Currently pursuing MRes in Clinical Research at King's College London.

Sujan Dilly Penchala

Dr Sujan Dilly Penchala PhD is a postdoctoral research associate at University of Liverpool in Bioanalytical facility with extensive experience in the field of regulated quantitative bioanalysis of drugs and metabolites.

Saye Khoo

Prof Saye Khoo, FRCP is Professor in the Institute of Translational Medicine, University of Liverpool with an interest in Pharmacology of HIV treatment His research focuses on the pharmacology of HIV treatment failure and how therapy may be improved through individualised pharmacology.

Nneka Nwokolo

Dr Nneka Nwokolo, FRCP, is a Consultant Physician in HIV Medicine with an interest in the sexual and reproductive health of women.

Marta Boffito

Dr Marta Boffito, MD, PhD, FRCP is a Consultant Physician, Clinical Research Lead and HIV Service Director at Chelsea and Westminster Hospital and Reader at Imperial College, London, UK. Her clinical and research focus is on complex pharmacological issues in HIV and ARV pharmacology in older age HIV. She contributes to the educational, scientific, and guideline-formulation activities of national bodies including the British HIV Association and is involved in capacity building programmes for resource-limited settings (e.g. Uganda).