Pharmacodynamic evaluation of the etonogestrel contraceptive implant initiated midcycle with and without ulipristal acetate: an exploratory study

Abstract

Objective:

To estimate the incidence of ovulation suppression within five days of etonogestrel 68 mg implant insertion in the presence of a dominant follicle with and without same-day ulipristal acetate.

Study Design:

This single site non-masked, exploratory randomized trial recruited people age 18–35 years with regular menstrual cycles, no pregnancy risk, and confirmed ovulatory function. We initiated transvaginal ultrasound examinations on menstrual day 7–9 and randomized participants 1:1 to etonogestrel implant alone or with concomitant ulipristal acetate 30 mg oral when a dominant follicle reached ≥14mm in diameter. We completed daily sonography and serum hormone levels for up to 7 days or transitioned to labs alone if sonographic follicular rupture occurred. We defined ovulation as follicular rupture followed by progesterone >3ng/mL. We calculated point estimates, risk ratios and 95% confidence intervals for ovulation for each group. Ovulation suppression of >=44% in either group (the follicular rupture suppression rate with oral levonorgestrel emergency contraception), would prompt future method testing.

Results:

From October 2020 - October 2022, we enrolled 40 people and 39 completed primary outcome assessments: 20 with etonogestrel implant alone (mean follicular size at randomization: 15.2mm ± 0.9mm) and 19 with etonogestrel implant + ulipristal acetate (mean follicular size at randomization: 15.4mm ± 1.2mm, p=0.6). Ovulation suppression occurred in 13 (65%) of etonogestrel implant-alone participants (Risk ratio 0.6 (95% CI: 0.3, 1.1), P=0.08) and 7 (37%) of implant + ulipristal acetate participants.

Conclusions:

Ovulation suppression of the etonogestrel implant alone exceeds threshold testing for future research while the implant + ulipristal acetate does not.

Implications:

Data are lacking on midcycle ovulation suppression for the etonogestrel implant with and without oral ulipristal acetate. In this exploratory study, ovulation suppression occurred in 65% of implant participants and 37% of implant + ulipristal acetate participants. Ovulation suppression of the implant alone exceeds threshold testing for future emergency contraception research.

1. Introduction

Ever-use of oral emergency contraception (EC) increased from 4.2% of reproductive age respondents in 2002 to 25.1% in the 2017–2019 National Survey of Family Growth [1]. Options for oral EC include levonorgestrel (LNG) 1.5 mg, a synthetic progestogen, or ulipristal acetate 30 mg (UPA), a progesterone receptor modulator. Oral LNG is over the counter and more accessible than UPA, which requires a prescription. UPA has several advantages over oral LNG. It can be used up to five days after unprotected intercourse, compared to three for oral LNG, and is more effective overall, especially for higher BMI users [2,3].

Oral EC has rigorous data on safety and risk of pregnancy, but does not provide ongoing contraception [4]. Both the copper T380A and LNG 52 mg intrauterine devices provide ongoing contraception and have EC efficacy data, which leaves the etonogestrel (ENG) implant as the only long-acting reversible contraception method lacking EC data [5,6]. Clinical guidelines support ENG implant insertion at the same time as oral LNG EC, but providers may delay insertion if oral LNG is not available in office and reschedule the insertion after use and next menses. If UPA is chosen for EC, recommended ENG implant insertion is delayed five days due to concern for drug-drug interactions [7]. This recommendation is based on pharmacodynamics data showing increased ovulation frequency when UPA is taken immediately before starting an oral progestogen contraceptive [8–10]. Initiation of the ENG implant at time of medication abortion with mifepristone, also a progesterone receptor modulator, did not increase medication abortion failure risk, but attenuation of effectiveness was noted with depo medroxyprogesterone acetate initiation [11–12]. As the ENG implant is a systemic progestogen with rapid rise of serum levels similar to oral LNG, it could plausibly serve as stand-alone EC [13]. In order to overcome initiation barriers to the ENG implant and evaluate its potential as an independent method of EC, evidence is needed to support clinical guidelines.

Here we address the lack of evidence assessing ovulatory suppression when the ENG implant is initiated in the presence of a dominant follicle, but prior to ovulation, and whether co-administration of UPA interferes with this effect. As has been standard in other EC studies, a pharmacodynamic assessment of ovarian activity is necessary [14–16]. An exploratory study design was chosen to help define the research question by providing baseline data with pre-determined go/no-go criteria to decide whether future inquiries for either method should proceed [17,18]. Thus, the objective of this study was to estimate point estimates and confidence intervals for the incidence of ovulation suppression within five days of ENG implant insertion in the presence of a dominant follicle with and without concomitant oral UPA and report a risk ratio comparing the groups.

2. Materials and Methods

We conducted an exploratory randomized, non-blinded clinical trial at a single site from October 2020 to October 2022. The University of Utah Institutional Review Board approved the study protocol and the ClinicalTrials.gov identifier is NCT04291001. We recruited healthy individuals through targeted social media ads who were aged 18–35 years, female sex, had self-report of regular menstrual cycles (24–35 days), a BMI < 30 kg/m2, and were interested in using the ENG implant but not currently at risk for pregnancy [19]. Exclusion criteria included current pregnancy or breastfeeding, use of hormonal contraception or exogenous hormones or glucocorticoids in the last month or planned use during the study, use of depo medroxyprogesterone acetate in the past six months, vaginal bleeding of unknown etiology, allergy to UPA or ENG, or current or planned use of any exogenous hormone or medication that could potentially interfere with sex steroid metabolism. We did not collect gender identity if participants met inclusion criteria, and thus, we do not use gender labels when referring to participants. We instructed heterosexually-active participants to use barrier methods during the screening and through seven days post ENG implant insertion.

We completed a phone screen with potential participants. The subsequent in-person screening visit included informed consent, history review, and a transvaginal ultrasound examination (TVUS) to confirm normal pelvic anatomy and ability to image ovaries. We collected a serum progesterone level during cycle days 20–24 to confirm ovulatory function (>3 ng/mL) [20]. In a subsequent cycle starting on day seven (+/− 2 days), we began monitoring participants with TVUS examination of their ovaries three times a week and then switched to daily monitoring when a lead follicle measured ≥13 mm in diameter. At the visit when the lead follicle was ≥14 mm in two perpendicular dimensions in the same plane, we randomized participants to one of two treatment groups: 1) ENG 68 mg implant alone or 2) ENG 68 mg implant with UPA 30 mg oral tablet [20]. We collected daily serum progesterone, luteinizing hormone (LH), and estradiol specimens via venipuncture with immediate transfer to ARUP laboratories (Salt Lake City, UT) for processing. The lab assay measured estradiol and progesterone concentrations via quantitative high-performance liquid chromatography- tandem mass spectrometry [21,22]. The lab measured LH by quantitative electrochemiluminescent immunoassay [23].

Participants returned daily for up to seven days for TVUS examination to measure ovarian follicle size and serum specimen collection. Daily TVUS examination continued until the exam demonstrated follicle rupture (≥50% decrease in mean follicular diameter compared to the previous days imaging in a follicle previously measuring ≥15 mm), the follicle was <12 mm on two consecutive visits, or we reached seven days after implant insertion, whichever occurred first. If we identified follicular rupture via TVUS examination, we stopped daily ultrasonography and continued to collect blood to measure serum hormone levels for the remainder of the seven days [15,16,24,25]. Participants completed an exit visit 14 days after randomization (+/− 2 days) for TVUS examination of any persistent follicles and collection of final serum hormone assessments to identify post-monitoring progesterone elevation. In this study, we aimed to randomize 40 participants 1:1 across the two study arms. One of the authors (A.G) generated the randomization scheme in block size of four and uploaded into the web-based access to the University of Utah Clinical Translational Science Institute supported Research Enterprise Data Capture (REDCap), which clinical staff then accessed at the time of participant enrollment. The sequence allocation was concealed from clinical staff. We did not conceal the treatment arm to participants or providers.

We report the primary outcome of ovulation suppression through five days post-randomization between the ENG implant alone and the ENG implant with oral UPA. We obtained three potential assessments of ovulation, including the day of follicle collapse by ultrasonography, day and value of highest LH, and day and value of highest progesterone level. If the date of ovulation was unclear by ultrasonography alone, we used serum hormone levels to adjudicate day of rupture based on previously established outcome definitions for ovulation by Croxatto, et al [15,16,25]. We assessed risk as the proportion of subject who ovulated by five days, our primary dichotomous outcome measure, by calculating the point estimates, risk difference, risk ratios and 95% confidence intervals for implant alone and with concomitant oral ulipristal acetate. We compared demographics between the two groups using the Wilcoxon signed-rank test or t-test for continuous variables, and chi square for categorical variables.

We modified our definition of ovulation to be more conservative and rely on follicular rupture followed by a progesterone rise, regardless of whether the LH level crossed the surge threshold of ≥21 IU/L defined by Croxato, et al [15,16,25]. For the primary outcome, there were five possible outcomes.

We report the proportion of each study group with each of the five outcomes. We completed univariate analyses to assess associations between cycle timing, follicle size, and hormonal levels with ovulation suppression by group using two-sample t-tests. Finally, we used logistic regression to calculate the odds of ovulation suppression by these aforementioned continuous variables, as well as conducted a survival analysis of time to ovulation by treatment day. We used Stata 16.0 or higher (StataCorp LP, College Station, TX) for all analyses.

Consistent with guidelines regarding sample size for pilot studies, baseline determination of ovulation suppression in these two previously untested situations will determine if either or both study groups warrant future testing for EC [26]. Fixed resources available for this study resulted in the total number of 40 participants. Our planned sample size of 20 in each study group is in the range of previous studies which used available participants from prior studies (n=27) or enrolled small sample sizes (n= 26 and 30) and then further stratified by follicular size at enrollment [15,20,25]. The threshold ovulation suppression rate we chose is >=44% [16]. Due to lack of direct comparison data and differences in prior study definitions of ovulation, we chose the threshold rate based on the suppression of follicular rupture for all follicle sizes using the original standard LNG regimen (0.75 mg X2, with a 12 hour interval between doses). We opted not to include a study group of the ENG implant with oral LNG as there is no reason to suspect a lower rate of ovulation suppression with the implant and oral LNG relative to oral LNG alone. We balanced the potential of three study groups and the reduced sample size of each with a more precise estimate of two options. We selected the latter to inform a future study comparing the implant and oral LNG for EC with the alternative(s) that met exceeded our threshold criteria. No power calculation was done to calculate sample size and thus post-hoc power calculation was not performed.

3. Results

We randomized 20 participants to the ENG implant alone and 20 to the ENG implant with oral UPA (Figure 1). The study groups included non-distinct differences in demographic characteristics. The mean follicular size at randomization was similar at 15.2 mm ± 0.9 mm in the ENG implant-alone group compared to 15.4 mm ± 1.2 mm (p=0.6). Mean estradiol levels on the day of randomization were similar between groups (161.3 ± 86.3 vs 173.9 ± 84.1; p = 0.64), while the peak estradiol level in the ENG implant-alone group (269.9 ± 111.2) was higher compared to the ENG implant with oral UPA (238.9 ± 78.4) (P=0.03). (Table 1)

Figure 1.

CONSORT (Consolidated Standards Reporting Trials) flow diagram for participants randomized to the etonogestrel implant alone or the etonogestrel implant with oral ulipristal in the setting of a dominant follicle

Table 1.

Characteristics of study participants randomized to the etonogestrel implant alone or the etonogestrel implant with ulipristal acetate 30 mg oral in the setting of a dominant follicle

	All (n=40)	ENG Implant only (n=20)	ENG Implant + oral UPA (n=20)	P-value
Age	23.7 (3.7)	23.2 (3.9)	24.3 (3.5)	0.37
BMI	24.0 (2.9)	24.2 (2.7)	23.7 (3.1)	0.60
Ever pregnant	1 (2.5)	1 (5.0)	0 (0.0)	0.31
Follicular Diameter at Randomization (mm)	15.3	15.2 (0.9)	15.4 (1.2)	0.52
Estradiol Level at Randomization (pg/mL)	167.6 (84.4)	161.3 (86.3)	173.9 (84.1)	0.64
Max Estradiol Level (pg/mL)	254.6 (97.4)	269.9 (111.2)	238.9 (78.4)	0.03*

Data reported as mean (standard deviation); ENG= etonogestrel; UPA= ulipristal acetate; BMI= body mass index; mm= millimeter; pg= picogram; mL= milliliter; study location Salt Lake City, UT, USA; Years of research 2020–2022

One participant in the ENG implant with oral UPA group requested removal of the ENG implant on day two due to pain at the insertion site, thus 39 participants completed primary outcome assessments. Ovulation suppression occurred in seven (37%) of ENG implant + UPA participants and in 13 (65%) of ENG implant-alone participants (Risk ratio 0.6 (95% CI: 0.3, 1.1), P=0.08) (Figure 2). The five ovarian outcome possibilities through day five after randomization are listed in Table 2 by study group. Five of the seven participants (71%) who ovulated in the ENG implant-alone group and eight of the 12 ENG implant + UPA participants (67%) who ovulated had a follicle size ≥15 mm at randomization. Only one ovulation dysfunction occurred in an ENG implant-alone participant with a follicle <15 mm at time of randomization. For those without follicular rupture, one outcome of follicular atresia occurred in each group in participants with follicles <15 mm at randomization. Persistently enlarged follicles occurred in nine of the 20 ENG implant-alone participants (45%) and six of the 19 ENG implant + UPA participants (32%). Luteinized unruptured follicles only occurred in two of the ENG implant-alone participants.

Figure 2.

Kaplan-Meier curves for time to ovulation after participant randomization to the etonogestrel implant alone and the etonogestrel implant with concomitant ulipristal acetate 30mg oral. Time to ovulation was censored at day five.

Table 2.

Outcomes of ruptured and unruptured dominant follicles during the five days following randomization to the etonorgestrel implant with or without ulipristal acetate 30mg oral stratified by follicle size on day of randomization.

	Implant Only (n=20)			Implant + UPA (n=19)
	<15mm	≥15mm	All	<15mm	≥15mm	All
Follicular rupture			8			12
Ovulation1	2	5	7	4	8	12
Ovulation dysfunction2	1		1			0
No Follicular Rupture			12			7
Follicular atresia3	1		1	1		1
Persistently enlarged follicle4	7	2	9	1	5	6
Luteinized unruptured follicle5	1	1	2			0
			35% ovulation			63% ovulation

¹Ovulation= a follicle ≥15mm that disappears or decreases in size by >50% and serum progesterone >3 ng/mL. It MAY be preceded 24–48 h by an LH surge (≥21 IU/L)

²Ovulation dysfunction= a follicle ≥15mm that disappears or decreases in size by >50% but not proceeded by an LH surge or preceded by a blunted (<21 IU/L) LH surge and without any progesterone >3 ng/mL

³Follicular atresia = a follicle <15 mm with arrest in growth or decrease in size of <50% without any progesterone >3 ng/mL

⁴Persistently enlarged follicle = a follicle measuring ≥15 mm that persists for at least one week without any progesterone >3 ng/mL

⁵Luteinized unruptured follicle= a follicle ≥15 mm that persists for at least one week with at least one progesterone >3 ng/mL

We completed univariate analyses to further understand the effects of clinical factors on ovarian activity. An increase in estradiol level on the day of randomization remained associated with ovulation status in the ENG implant only group but not in the ENG implant with oral UPA participants (Supplemental Table 1). Ovulation suppression was 64% less likely among those assigned the implant + UPA than those assigned the implant alone (Supplemental Table 2). For each 1 pg/mL increase in estradiol level measured on the day of randomization decreased the likelihood of ovulation suppression by 1.5% (P=0.03), controlling for other factors. Additionally, each 1mm increase in mean follicular diameter measured on the day of randomization decreased the odds that ovulation was suppressed by 33% (P=0.37).

4. Discussion

The ovulation suppression of the ENG implant alone when inserted in the setting of a dominant follicle surpassed the pre-set threshold for this exploratory study, while the ENG implant with oral UPA did not. These findings support future inquiry of the ENG implant alone to evaluate whether it is equivalent or superior to the ENG implant plus oral LNG for pregnancy prevention. The low incidence of ovulation suppression with combined use of the ENG implant with oral UPA supports concerns for drug-drug interactions that previously lacked direct data. Previous researchers found a negative influence of oral progestogens on UPA efficacy, as well as concern for decreased ovulation suppression by combination oral contraceptives when UPA is co-administered [8–10]. This may be due to UPA’s competitive inhibition of hypothalamic progesterone receptors. While we followed participants in this study through seven days post randomization, we are limited in defining follicular rupture outcomes on days six or seven as we did not continue progesterone levels after day seven to confirm true ovulation vs. other outcomes. Based on our findings, concomitant ENG and UPA is not recommended for individuals with recent unprotected intercourse who do not wish to become pregnant. If individual circumstances prompt concomitant use, the clinical guidelines which recommend use of back-up contraception should be followed, due to lack of evidence on ovulation suppression in the 5–7-day post-insertion timeframe [7].

This study’s rigor and reproducibility is based on established methodology using the definitions of ovulatory outcomes established by Croxatto et al., which allow for comparisons to numerous studies of emergency and ongoing contraceptive method suppression [15,16,20,24,25]. A recent publication by Westhoff et al. highlights the limitations in this approach due to potential misclassification and challenges differentiating normal ovulation, which carries a risk of pregnancy from luteinized unruptured follicles, and ovulation dysfunction, which does not [27]. This exploratory study was designed to refine research questions for progression to future studies. As such, the sample size was based on guidelines for pilot studies, threshold testing, and available resources rather than power calculations for group comparisons [16]. The sample size is comparable to other pharmacodynamic studies of EC and continuous methods. While we considered other study designs, such as use of oral LNG alone as a comparator, we opted to use the two methods with the least available data to address knowledge gaps and guide future research. We simplified baseline ovulation by testing only luteal-phase progesterone. Not including serum LH and estradiol assessments prior to randomization limited our ability to use LH as part of the outcome definitions for the early follicular ruptures. We ended progesterone measurement seven days after randomization; if a rise in progesterone was delayed, we would have misclassified the outcome. Prior research on hormone assessments after a normal ovulation showed progesterone rise is delayed ~2 days and should be identifiable on day seven if ovulation occurred by day five [28]. All participants ovulated prior to day five except one participant with ovulation on day five in the ENG implant with oral UPA arm, who had a quick progesterone rise and was classified as ovulation. If we opted to use Westhoff et al.’s criteria of defining ovulation only by progesterone rise, we would need to reclassify the luteinized unruptured follicles in the ENG implant-alone arm [27]. If we reclassified these as ovulation, the ovulation suppression rate to decreases to 55%, which still exceeds our predetermined threshold for future investigation. Finally, we excluded people with a BMI > 30 kg/m² who may have different pharmacokinetic and pharmacodynamic outcomes. The ENG implant alone will require future study in higher BMI categories, consistent with other oral EC options [3, 29].

In conclusion, these pharmacodynamic findings are encouraging to support future testing of the ENG implant as both an EC option and ongoing contraceptive method for a population at high risk of unintended pregnancy. The low rate of ovulation suppression in the ENG implant plus oral UPA group supports current clinical recommendations to delay ENG implant initiation if oral UPA is chosen for EC. Following the current clinical guidelines of concomitant oral LNG EC and ENG implant insertion is not harmful, but may be unnecessary, and serves as an initiation barrier for providers who lack in-office oral LNG. An ongoing, adequately powered randomized clinical trial will assess EC pregnancy risk with the ENG implant with and without oral LNG (Clinical Trials registry NCT06162611) to address knowledge gaps and associated barriers to method initiation in current clinical guidelines.

Five outcome measures stratified by sonographic evidence of follicular rupture.

• In the setting of sonographic follicular rupture, outcomes included:

  • (1)“Ovulation” defined as a follicle ≥15mm that disappears or decreases in size by >50% and serum progesterone >3 ng/mL. It MAY be preceded 24-48 h by an LH surge (≥21 IU/L), or
  • (2)“Ovulatory Dysfunction” defined as a follicle ≥15mm that disappears or decreases in size by >50% but not proceeded by an LH surge or preceded by a blunted (<21 IU/L) LH surge and without any progesterone >3 ng/mL.

• When sonographic follicular rupture did not occur, we defined categories of “Ovulation Suppression”, which included:

  • (3)“Follicular Atresia” defined as a follicle <15 mm with arrest in growth or decrease in size of <50% without any progesterone >3 ng/mL, or
  • (4)“Persistently Enlarged Follicle” defined as a follicle measuring ≥15 mm that persists for at least one week without any progesterone >3 ng/mL, or
  • (5)“Luteinized Unruptured Follicle” defined as a follicle ≥15 mm that persists for at least one week with at least one progesterone >3 ng/mL.