Efficacy and safety of ulipristal compared with levonorgestrel for emergency oral contraception

26 Jul 2024 Efficacy and safety of ulipristal compared with levonorgestrel for emergency oral contraception

Emergency contraception (EC) helps prevent pregnancy after unprotected sexual intercourse or a failed contraceptive. Methods of EC are copper-bearing intrauterine devices (IUDs) and emergency contraceptive pills (ECPs) (WHO, 2021, Emergency Contraception). In Canada, ECPs either contain levonorgestrel (Plan B® or generic equivalent) or ulipristal acetate (ella® or generic equivalent) (HealthLinkBC, 2012, Emergency Contraception).

In Canada, levonorgestrel EC is provided as 1.5 mg tablets to be taken orally as a single dose, as soon as possible, up to 5 days (120 hours) after last sexual contact or possible exposure. Levonorgestrel (LNG) is available at local pharmacies and some community health clinics without a prescription. The approved regimen for ulipristal (UPA) EC is one oral dose of 30 mg up to 5 days after unprotected intercourse. In Canada, UPA currently requires a prescription.

Pharmaceutical, Laboratory and Blood Services Division, BC Ministry of Health (PLBSD) requested the Therapeutics Initiative to conduct a systematic review of randomized controlled trial (RCT) evidence to compare the efficacy and safety of ulipristal acetate with levonorgestrel for use as emergency contraception.

Methods and results

A 2019 Cochrane review (Shen 2019) identified 2 randomized controlled trials (RCTs) (Creinin 2006; Glasier 2010) comparing the efficacy and safety of UPA with LNG. We used standard Cochrane Collaboration systematic review methodology to update the Cochrane review. We searched the Cochrane Database of Systematic Reviews (CDSR), Cochrane Central Register of Controlled Trials (CENTRAL), Ovid MEDLINE, Ovid Embase, CINHAHL, PsycINFO and Google Scholar from the end date of Shen 2019 until May 3, 2022 for any RCTs published since the completion of Shen 2019. We also searched ClinicalTrials.gov and WHO ICTRP for all relevant RCT reports. Risk of bias was assessed for included RCTs using the Cochrane Risk of Bias tool. Outcomes are presented according to the UBC Therapeutics Initiative’s (TI) hierarchy: All-cause mortality; total serious adverse events (SAEs) including complications of pregnancy (e.g., molar pregnancy); rate of pregnancy; quality of life (QOL); withdrawal due to adverse events (WDAEs); and total adverse events (AEs) such as nausea, vomiting, spotting or bleeding after treatment and effects on menses.

No new comparative studies were identified in our updated literature search. Only two RCTs (Creinin 2006; Glasier 2010) comparing the efficacy and safety of UPA with LNG were included in our review. Creinin 2006 was supported by federal funds from the National Institutes of Health and Glasier 2010 was funded by HRA Pharma.

Both Creinin 2006 and Glasier 2010 were non-inferiority trials. Creinin 2006 was a double-blind study that enrolled 1672 healthy women seeking EC within 72 hours of unprotected intercourse. Participants were randomly assigned to receive a single dose of 50 mg of ulipristal plus a placebo 12 hours later or two doses of 0.75 mg of levonorgestrel taken 12 hours apart. Glasier 2010 was a single-blind study that enrolled 2221 women with regular menstrual cycles (24–35 days) seeking EC within 120 h of unprotected sexual intercourse. Women were randomly assigned to receive a single dose of ulipristal acetate 30 mg or levonorgestrel 1.5 mg given orally.

Creinin 2006 considered a non-inferiority margin of 2% difference between groups for pregnancy rate, whereas Glasier 2010 considered a non-inferiority margin of 1% difference between groups (limit of 1·6 for odds ratio) for the same outcome. Modified intent-to-treat (mITT) analysis in the Creinin 2006 study included all women who were randomly assigned and for whom a pregnancy outcome was known (792 UPA, 786 LNG). A total of 94 (5.6%; 40 UPA, 54 LNG) who were lost to follow-up and pregnancy outcome could not be evaluated were excluded from the mITT analysis. An additional 29 women (5 pre-treatment pregnancy; 24 additional EC during the treatment cycle) were also excluded. Intention-to-treat (ITT) analysis in Glasier 2010 included all randomized patients (1104 UPA, 1117 LNG). A total of 88 women (48 UPA, 40 LNG), estimated as 4.4% and 3.6%, were lost to follow up among UPA and LNG users, respectively.

No deaths were reported in either of the two RCTs (Table I). There were 2 and 7 SAEs in Creinin 2006 and Glasier 2010 studies, respectively. Two SAEs in UPA-treated subjects in Creinin 2006 included kidney infection (2 months after UPA intake) and pelvic inflammatory disease (1 month after UPA intake). Neither SAE was considered by investigators to be treatment-related. In Glasier 2010, 7 SAEs were reported; 3 in the UPA arm (urinary tract infection, corneal ulcer, and dizziness) and 4 in the LNG arm (vomiting blood-stained fluid, molar pregnancy, ruptured ovarian cyst and kidney stones). Only dizziness (UPA) and molar pregnancy (LNG) were considered by the investigators to be possibly related to treatment with study drug (SAEs in Glasier 2010 study were reported in the FDA Center for Drug Evaluation and Research (CDER) Medical Review 2010 for Ella (ulipristal acetate 30 mg)). SAEs in the Creinin study were reported from: https://www.fda.gov/files/drugs/published/22474-Ella-Clinical-PREA.pdf)

Table I: Hierarchy of outcomes

* Efficacy-evaluable population who received EC within 72 h of sexual intercourse; NR not reported

In Creinin 2006 the estimated pregnancy rates in the efficacy evaluable population of 1549 women were 0.9% (95% CI 0.2–1.6%) and 1.7% (95% CI 0.8–2.6%) among UPA and LNG users, respectively. The difference between treatments of −0.8%, with an upper limit of the 97.5% one-tailed CI of 0.77%, was statistically noninferior (P<.001). In the mITT population (n=1578), pregnancy rates were estimated as 1.5%, (95% CI 0.7–2.4%) and 1.8%, (95% CI 0.9–2.7%) among UPA and LNG users, respectively. This difference between treatments of −0.3%, with an upper limit of the 97.5% one-tailed CI of 1.42%, was considered statistically non-inferior (P=0.003).

In Glasier 2010, 2221 eligible women were randomized but only 1899 women comprised the efficacy-evaluable population, defined as women aged 35 years or younger (as recommended by the US FDA) who were enrolled for the first time in the study, and whose pregnancy status after treatment was known. Those excluded from the efficacy-evaluable were: 88 (4.0%) lost to follow-up with unknown pregnancy status; 145 (6.5%) age >35 years; 46 (2.1%j with unknown follow-up pregnancy status; 36 (1.6%); and 7 (0.3%) pregnancies judged by the data safety monitoring board to have occurred before EC was taken (n=4) or at least 10 days after treatment (n=3).

In the efficacy-evaluable population, 1696 women received EC (a single dose of 30 mg UPA or 1·5 mg LNG orally) within 72 h of sexual intercourse. The authors explained that micronization of the drug had allowed a reduction of the UPA dose from 50 mg (studied in Creinin 2006) to 30 mg when used in tablet form, and evidence from an indirect pharmacokinetic comparison indicated the similarity of the two formulations. In women who received emergency contraception within 72 h after sexual intercourse, the pregnancy rates in the efficacy-evaluable population were 1·8% (95% CI 1·0–3·0) and 2·6% (95% CI 1·7–3·9) among UPA users and LNG users, respectively [OR 0·68, 95% CI 0·35–1·31); RR 0.69, 95% CI 0.36-1.32)]. The authors explained that efficacy results were consistent with those obtained for the ITT population but the data were not shown.

In 1899 women who received emergency contraception within 120 h after sexual intercourse, pregnancy rates were estimated as 1·6% and 2·6% in the UPA and LNG groups, respectively. The difference between groups was not statistically significant; [OR 0.57, 95% CI (0.29–1.09); RR 0.61, 95% CI (0.32,1.15)]. Comparison of the efficacy of UPA with LNG in women who received EC between 72 h and 120 h of sexual intercourse (OR 0.26, 95% CI 0.03-2.39) is limited by the small sample size (i.e., only 10.7% of the efficacy evaluable population), which contributed to the wide 95% Cis. (Note: (OR 0.26, 95% CI 0.03-2.39) was accessed from the ClinicalTrials.gov website in 2022).

No data were available for quality of life outcomes in both studies. WDAEs were not reported in the Creinin study. There were 2 WDAEs in the UPA arm in Glasier 2010 (reported in the FDA CDER Statistical Review 2010 for Ella (ulipristal acetate 30 mg)).

Although adverse events were generally similar in both treatment groups, more UPA users in Creinin 2006 experienced nausea compared with LNG users (29% compared with 24%, P=0.03). Women in both groups experienced variation in menstrual cycle length compared with their reported individual normal cycle lengths. In Glasier 2010, adverse events were reported by approximately half of the ITT population (54% and 56% of UPA and LNG users, respectively). The most frequently reported adverse events were similar in both groups.

Meta-analysis of the 2 comparative studies revealed pregnancy rates of 1.4% and 2.2% among women in UPA and LNG groups, respectively. There was no difference between UPA and LNG users when taken up to 72 hours after sexual intercourse (RR 0·63, 95% CI 0·37–1.07; p=0·09).

Based on 2 RCTs (Creinin 2006 and Glasier 2010), total AEs were similar for both treatment groups (1342 ULA, 1370 LNG). UPA users and LNG users did not differ in rates of nausea, vomiting, headache, dizziness, fatigue, breast tenderness, diarrhea, spotting/bleeding after treatment, overall abdominal pain, lower abdominal pain, upper abdominal pain, back pain, or dysmenorrhoea.

Meta-analysis of the 2 comparative studies (Creinin 2006 and Glasier 2010) found that women who took UPA were less likely to have early return (before the expected date) of menses, compared with those who received LNG (RR 0.43, 95% CI 0.37 to 0.50, 2 RCTs, n = 3593), and UPA users were more likely to have delayed return (after the expected date) of menses than those who received LNG (RR 1.65, 95% CI 1.42 to 1.92, 2 RCTs, n = 3593).

Efficacy of UPA and LNG up to 120 hr
Shen 2019 stated that UPA can be used up to 120 hours after unprotected intercourse based on evidence from a non-randomized study (Fine et al., 2010). This review provides a summary of the findings from that study related to the efficacy of UPA up to 120 hours after unprotected intercourse.

Fine 2010 study
Fine 2010 study was a nonrandomized, open-label study supported by HRA Pharma. It was conducted in the US and evaluated the efficacy of UPA among 1241 women with regular cycles during 48 to 120 hours after unprotected sexual intercourse. The primary efficacy outcome was the pregnancy rate. The main secondary efficacy analysis compared the upper bound of the 95% CI of the observed pregnancy rate with a clinical irrelevance threshold of 4%. This threshold corresponds to a reduction by half of the expected 8% pregnancy rate in the absence of contraception as observed in previous international studies. Other secondary outcomes included the prevented fraction of pregnancies and the trend in pregnancy rates over time (by 24-hour intervals).

The observed pregnancy rate was estimated at 2.1% (95% CI 1.4–3.1%). The authors concluded that the study was successful because the pregnancy rate was lower than both the estimated expected pregnancy rate and a predefined clinical irrelevance threshold (4%). The estimated pregnancy rates were reported as 2.3% (95% CI 1.4–3.8%), 2.1% (95% CI 1.0–4.1%), and 1.3% (95% CI 0.1–4.8%) for intervals of 48 to 72 hours, more than 72 to 96 hours, and more than 96 to 120 hours, respectively. According to the authors, changes in pregnancy rates over the different time intervals were not statistically significant (P=0.44).

The evidence from this study is limited due to a lack of a control or a comparison group of LNG users in the study, and only approximately 44% (548/1241) of participants received EC treatment within 72 and 120 hours after unprotected intercourse. Notably, the primary efficacy population experienced over 10% dropout rate, including women lost to follow-up with unknown pregnancy outcome data and those with repeat enrollment. The authors did not clarify whether they accounted for an appropriate dropout rate in their sample size calculation.

Piaggio 2011 study
An analysis of data from 4 United Nations/World Health-funded RCTs compared levonorgestrel with alternative treatments for emergency contraception. Efficacy of levonorgestrel was evaluated in 6,794 women, including 5,489 in the 2 RCTs that received EC treatment up to 120 hours after unprotected intercourse. The authors reported that delaying levonorgestrel administration until the fifth day after unprotected intercourse increased the risk of pregnancy over five times compared with its administration within 24 hours (OR=5.81, 95% CI 2.87- 11.76). However, only 3.4% (230/6794) of women took LNG during the final interval of 97 to 120 hours. The odds ratio of becoming pregnant among levonorgestrel users did not differ significantly on the second, third, and fourth day (Piaggo et al., 2011)

Association between BMI and the efficacy of emergency contraceptives
Glasier 2011 evaluated the impact of BMI on the risk of pregnancy after using LNG or UPA as EC. The authors reported that compared with women with a normal BMI (<25 kg/m²) the odds of pregnancy was more than 3 times greater (OR 3.60; 95% CI 1.96-6.53; P< 0.0001) for obese women (BMI > 30 kg/m²) regardless of which method of EC was used. Among obese women the odds of pregnancy was greater with the use of LNG (OR 4.41; 95% CI 2.05-9.44; P< 0.0002) than with the use of UPA (OR 2.62; 95% CI 0.89-7.00). However, the difference between treatment groups was not statistically significant.

European Medicines Agency (EMA) report on the association between the clinical effectiveness of hormonal contraceptives and high body weight/high BMI

The Committee for Medicinal Products for Human Use (CHMP) concluded that there was limited and inconsistent data on the effect of high body weight/high BMI on contraceptive efficacy among levonorgestrel users. As well, limited and inconclusive data suggested a reduced efficacy of UPA with increased body weight in women. EMA concluded that all women should take EC as soon as possible after unprotected intercourse (Scientific conclusion by EMA, n.d.).

Analysis of data from WHO studies by Festin 2017

Festin 2017 provided an analysis of the data for the LNG arms from the four studies supported by WHO Development and Research Training in Human Reproduction (WHO HRP). Evidence from an adjusted regression model (n=6783 in LNG arms) suggested that a BMI over 30 kg/m² or obesity had a negative impact on efficacy OR = 8.27, (95% CI 2.70–25.37). Festin 2017 performed a sensitivity analysis on the pooled subset without the Nigerian study data and the sensitivity analysis failed to show that BMI affected efficacy (Festin 2017).

Overall Conclusion

• The evidence from RCTs found no difference in the short-term safety and tolerability profiles of ulipristal acetate and levonorgestrel.
• The relative efficacy of levonorgestrel and ulipristal acetate within 72 hours of unprotected intercourse is not established conclusively.
• The relative efficacy of levonorgestrel and ulipristal acetate within 72 to 120 hours of unprotected intercourse is not well established due to limited evidence.
• The evidence of an association between the efficacy of ulipristal acetate or levonorgestrel and high body weight/high BMI is inconclusive.

Read the full report…