ellaOne for London
ellaOne for London Dr Paul O’Brien, Westside Contraceptive Services, London 23-5-10
Introduction Around 6% to 8% of women who attend for emergency contraception in clinical trials following unprotected intercourse would become pregnant in the absence of emergency contraception. This drops to less than 2% with levonorgestrel, the standard emergency contraceptive used in UK. ellaOne had been developed to improve on this, and has recently been licensed in the UK for use up to 5 days after sexual intercourse, two days later than the licence for levonorgestrel. The active substance of ellaOne is ulipristal acetate, a synthetic selective progesterone receptor modulator with antagonistic and partial agonistic effects at the progesterone receptor. Ulipristal acetate acts by interrupting and delaying ovulation and may have some direct effects on endometrium.
Efficacy Ulipristal has been studied in two randomised trials against levonorgestrel and in one non-comparative clinical study. In the first study Creinin (2006)1 randomised 1,672 women presenting within the first 3 days following unprotected intercourse to ulipristal or levonorgestrel and in the second, recent study, Glasier (2010)2 randomised 2,221 women to the same treatments but included women up to five days after intercourse. Creinin
Day 1-3
Glasier
Ulipristal
Levonorgestrel
Ulipristal
Levonorgestrel
Pregnant/Treated
Pregnant/Treated
Pregnant/Treated
Pregnant/Treated
7/775 (0.9%)
13/774 (1.7%)
15/844 (1.8%)
22/852 (2.6%)
0/97 (0%)
3/106 (2.8%)
7/775 (0.9%)
13/774 (1.7%)
15/941 (1.6%)
25/958 (2.6%)
Day 4-5 Total Adjusted odds ratio (95% CI), ulipristal versus levonorgestrel
0.50 (0.18-1.24) p=0.14
0.57 (0.29-1.09) p=0.09
There was no significant difference in the pregnancy rates in either the Creinin or Glasier trials, but the trials were powered to show non-inferiority within a margin of 2% and 1% respectively, and noninferiority was demonstrated. To increase the power of the comparison Glasier et al performed a metaanalysis of the two studies and found a significant superior efficacy overall (odds ratio 0.55, 95% confidence intervals 0.32-0.93). Glasier et al adjusted, using logistic regression, the pregnancy rates for a number of confounding factors, which is unusual in large randomised trials in which the potential confounding factors are balanced across the two treatments. The adjustment itself can be confounded because it included a notional risk of pregnancy for different days in the cycle in the absence of emergency contraception, which is frequently wrong. In the Fine 2010 study3 the majority of the pregnancies (16/29) occurred when the predicted risk was zero. The adjustments pushed some of the meta-analyses across the 5% significance level. Use of emergency contraception within 3 days following intercourse As levonorgestrel is licensed in the UK for use up to 72 hours after intercourse, the meta-analysis by Glasier et al of the two randomised studies is of particular interest. Unadjusted OR
Adjusted OR
Risk difference (%)
Creinin
0.53 (0.21-1.34)
0.50 (0.18-1.24)
-0.78 (-1.90-0.35)
Glasier
0.68 (0.35-1.33)
0.68 (0.35-1.31)
-0.80 (-2.19-0.11)
Meta-analysis
0.63 (0.37-1.07) p=0.09
0.58 (0.33-0.99) p=0.46
-0.79 (-1.69-0.11) p=0.09
The meta-analyses show a 40% reduction in the pregnancy odds ratios of borderline statistical significance. Significance is achieved through post-hoc adjustment of the earlier trial considered inappropriate by the authors at the time.1 The Cochrane review uses unadjusted rates.4 The difference in pregnancy rates of 0.79% is less than the difference considered clinically relevant by Glasier et al prior to the trial2 and would require treatment of over 120 women to prevent one pregnancy. Glasier et al excluded 14% of the pregnancies occurring in the trial because they were already pregnant or became pregnant soon after the treatment, as they were not treatment failures. In
both trials lower risk women, such as those with irregular cycles, had missed oral contraceptive pills or were over 35 year were excluded. Both of these factors will increase the numbers needed-to-treat for benefit with ulipristal in routine practice. Use of emergency contraction on days 4 or 5 after intercourse The only comparative data on use of ulipristal 4 or 5 days after intercourse comes from the Glasier trial in which 63 women were treated with ulipristal on day 4, and 34 on day 5. The corresponding numbers for levonorgestrel were 73 and 33. There were no pregnancies with ulipristal and 3 pregnancies with levonorgestrel. The difference is not statistically significant (p=0.25). The only other experience with ulipristal at 4 and 5 days after intercourse come from the non-comparative Fine (2010) study.3 There were 8 pregnancies in 390 women (2.1%) treated at day 4, and 2 in 158 women (1.3%) treated on day 5. In the WHO (2002) trial5 the pregnancy rate with levonorgestrel was 1.1% (2/188) for treatment on day 4, and 4.8% (6/126) on day 5. It has been suggested that there is a trend in loss of efficacy the longer the delay in treatment with levonorgestrel that is not seen with ulipristal. However, only one study, WHO 1998,6 has shown a statistically significant trend in pregnancy rate with delay in treatment with levonorgestrel, and there was no trend seen in the Creinin or Glasier trials. If there is a loss of efficacy with delay in treatment with levonorgestrel it is likely to be small. The absence of pregnancies in women treated with ulipristal at days 4 and 5 in the Glasier trial is likely to be due to the small numbers and chance. There is insufficient evidence that ulipristal has superior efficacy to levonorgestrel when treatment is given 4 or 5 days after intercourse.
Safety Adverse effects are in general similar to those reported for levonorgestrel. There is very limited information on the potential effects of treatment with ulipristal in pregnancy that may already be existing and unidentified (4 of 50 in the Glasier trial) or a results of treatment failure,7 and users should be informed on this uncertainty. Although the use of ulipristal does not contraindicate the continued use of regular hormonal contraception, ulipristal may reduce its contraceptive action and the SPC recommends that subsequent acts of intercourse be protected by a reliable barrier method until the next menstrual period starts.8 The use of ulipristal or levonorgestrel on the pregnancy risk from subsequent unprotected intercourse in the same cycle is unknown, but a higher dose of levonorgestrel may be required following ulipristal use; repeat use of ulipristal is not recommended.8
Conclusion The changes in emergency from the Yuzpe regime to levonorgestrel and then from two to one doses, providing a more effective treatment with fewer side effects and easier to take, were based on strong evidence. The evidence supporting a switch to ulipristal is much weaker. The wholesale replacement of levonorgestrel with ulipristal would seem to be inappropriate based on the equivocal superior efficacy, the small size of any benefit, the known safety of levonorgestrel, the unknown effect of ulipristal when given inadvertently in early pregnancy, and the possible impact on hormonal contraception started after ulipristal. Restricting ulipristal to women presenting at days 4 and 5 after intercourse is also inappropriate as the comparative evidence for ulipristal is extremely limited. Efficacy of levonorgestrel when used in days 4 and 5 has been demonstrated on the same basis as for ulipristal7 in the WHO 2002 trial5 and an application for a licence for use on days 4 and 5 would be likely to succeed. Levonorgestrel should be offered to these women, and nurse and pharmacy PGDs should reflect this. Women judged to be at higher risk of pregnancy might benefit from ulipristal, as the absolute benefit could be greater, although the judgment of risk is frequently wrong, making it difficult to select the women for ulipristal. The suggestion that non-use of ulipristal risks litigation2 is unfounded, because of the weakness of the evidence, and the small size of any potential benefit. As the odds ratio is greater than 0.5, any pregnancy that follows levonorgestrel use is as likely as not to have occurred with ulipristal. ellaOne is a new option in emergency contraception that is at least as effective as levonorgestrel. However, in levonorgestrel we have a last chance contraceptive that is of proven efficacy and safety. The problem with levonorgestrel is less its efficacy than its non-use. If the levels of unprotected intercourse are to continue, available, safety and price are key elements in the reduction in unwanted pregnancies.
References 1. Creinin MD, Schlaff W, Archer DF, Wan L, Frezieres R, Thomas M, et al. Progesterone receptor modulator for emergency contraception: a randomized controlled trial. Obstetrics & Gynecology 2006;108(5):1089-97. 2. Glasier AF, Cameron ST, Fine PM, Logan SJ, Casale W, Van Horn J, et al. Ulipristal acetate versus levonorgestrel for emergency contraception: a randomised non-inferiority trial and metaanalysis. Lancet 2010;375(9714):555-62. 3. Fine P, Mathe H, Ginde S, Cullins V, Morfesis J, Gainer E, et al. Ulipristal acetate taken 48-120 hours after intercourse for emergency contraception. Obstetrics & Gynecology 2010;115(2 Pt 1):257-63. 4. Cheng L, Gulmezoglu AM, Piaggio G, Ezcurra E, Van Look PF. Interventions for emergency contraception. Cochrane Database Syst Rev 2008(2):CD001324. 5. von Hertzen H, Piaggio G, Ding J, Chen J, Song S, Bartfai G, et al. Low dose mifepristone and two regimens of levonorgestrel for emergency contraception: a WHO multicentre randomised trial. Lancet 2002;360(9348):1803-10. 6. Task Force on Postovulatory Methods of Fertility Regulation. Randomised controlled trial of levonorgestrel versus the Yuzpe regimen of combined oral contraceptives for emergency contraception. Lancet 1998;352(9126):428-33. 7. EMEA. CHMP Assessment Report for Ellaone, 2009. 8. HRA Pharma. Summary of Product Characteristics for ellaOne, 2009.
Introduction Around 6% to 8% of women who attend for emergency contraception in clinical trials following unprotected intercourse would become pregnant in the absence of emergency contraception. This drops to less than 2% with levonorgestrel, the standard emergency contraceptive used in UK. ellaOne had been developed to improve on this, and has recently been licensed in the UK for use up to 5 days after sexual intercourse, two days later than the licence for levonorgestrel. The active substance of ellaOne is ulipristal acetate, a synthetic selective progesterone receptor modulator with antagonistic and partial agonistic effects at the progesterone receptor. Ulipristal acetate acts by interrupting and delaying ovulation and may have some direct effects on endometrium.
Efficacy Ulipristal has been studied in two randomised trials against levonorgestrel and in one non-comparative clinical study. In the first study Creinin (2006)1 randomised 1,672 women presenting within the first 3 days following unprotected intercourse to ulipristal or levonorgestrel and in the second, recent study, Glasier (2010)2 randomised 2,221 women to the same treatments but included women up to five days after intercourse. Creinin
Day 1-3
Glasier
Ulipristal
Levonorgestrel
Ulipristal
Levonorgestrel
Pregnant/Treated
Pregnant/Treated
Pregnant/Treated
Pregnant/Treated
7/775 (0.9%)
13/774 (1.7%)
15/844 (1.8%)
22/852 (2.6%)
0/97 (0%)
3/106 (2.8%)
7/775 (0.9%)
13/774 (1.7%)
15/941 (1.6%)
25/958 (2.6%)
Day 4-5 Total Adjusted odds ratio (95% CI), ulipristal versus levonorgestrel
0.50 (0.18-1.24) p=0.14
0.57 (0.29-1.09) p=0.09
There was no significant difference in the pregnancy rates in either the Creinin or Glasier trials, but the trials were powered to show non-inferiority within a margin of 2% and 1% respectively, and noninferiority was demonstrated. To increase the power of the comparison Glasier et al performed a metaanalysis of the two studies and found a significant superior efficacy overall (odds ratio 0.55, 95% confidence intervals 0.32-0.93). Glasier et al adjusted, using logistic regression, the pregnancy rates for a number of confounding factors, which is unusual in large randomised trials in which the potential confounding factors are balanced across the two treatments. The adjustment itself can be confounded because it included a notional risk of pregnancy for different days in the cycle in the absence of emergency contraception, which is frequently wrong. In the Fine 2010 study3 the majority of the pregnancies (16/29) occurred when the predicted risk was zero. The adjustments pushed some of the meta-analyses across the 5% significance level. Use of emergency contraception within 3 days following intercourse As levonorgestrel is licensed in the UK for use up to 72 hours after intercourse, the meta-analysis by Glasier et al of the two randomised studies is of particular interest. Unadjusted OR
Adjusted OR
Risk difference (%)
Creinin
0.53 (0.21-1.34)
0.50 (0.18-1.24)
-0.78 (-1.90-0.35)
Glasier
0.68 (0.35-1.33)
0.68 (0.35-1.31)
-0.80 (-2.19-0.11)
Meta-analysis
0.63 (0.37-1.07) p=0.09
0.58 (0.33-0.99) p=0.46
-0.79 (-1.69-0.11) p=0.09
The meta-analyses show a 40% reduction in the pregnancy odds ratios of borderline statistical significance. Significance is achieved through post-hoc adjustment of the earlier trial considered inappropriate by the authors at the time.1 The Cochrane review uses unadjusted rates.4 The difference in pregnancy rates of 0.79% is less than the difference considered clinically relevant by Glasier et al prior to the trial2 and would require treatment of over 120 women to prevent one pregnancy. Glasier et al excluded 14% of the pregnancies occurring in the trial because they were already pregnant or became pregnant soon after the treatment, as they were not treatment failures. In
both trials lower risk women, such as those with irregular cycles, had missed oral contraceptive pills or were over 35 year were excluded. Both of these factors will increase the numbers needed-to-treat for benefit with ulipristal in routine practice. Use of emergency contraction on days 4 or 5 after intercourse The only comparative data on use of ulipristal 4 or 5 days after intercourse comes from the Glasier trial in which 63 women were treated with ulipristal on day 4, and 34 on day 5. The corresponding numbers for levonorgestrel were 73 and 33. There were no pregnancies with ulipristal and 3 pregnancies with levonorgestrel. The difference is not statistically significant (p=0.25). The only other experience with ulipristal at 4 and 5 days after intercourse come from the non-comparative Fine (2010) study.3 There were 8 pregnancies in 390 women (2.1%) treated at day 4, and 2 in 158 women (1.3%) treated on day 5. In the WHO (2002) trial5 the pregnancy rate with levonorgestrel was 1.1% (2/188) for treatment on day 4, and 4.8% (6/126) on day 5. It has been suggested that there is a trend in loss of efficacy the longer the delay in treatment with levonorgestrel that is not seen with ulipristal. However, only one study, WHO 1998,6 has shown a statistically significant trend in pregnancy rate with delay in treatment with levonorgestrel, and there was no trend seen in the Creinin or Glasier trials. If there is a loss of efficacy with delay in treatment with levonorgestrel it is likely to be small. The absence of pregnancies in women treated with ulipristal at days 4 and 5 in the Glasier trial is likely to be due to the small numbers and chance. There is insufficient evidence that ulipristal has superior efficacy to levonorgestrel when treatment is given 4 or 5 days after intercourse.
Safety Adverse effects are in general similar to those reported for levonorgestrel. There is very limited information on the potential effects of treatment with ulipristal in pregnancy that may already be existing and unidentified (4 of 50 in the Glasier trial) or a results of treatment failure,7 and users should be informed on this uncertainty. Although the use of ulipristal does not contraindicate the continued use of regular hormonal contraception, ulipristal may reduce its contraceptive action and the SPC recommends that subsequent acts of intercourse be protected by a reliable barrier method until the next menstrual period starts.8 The use of ulipristal or levonorgestrel on the pregnancy risk from subsequent unprotected intercourse in the same cycle is unknown, but a higher dose of levonorgestrel may be required following ulipristal use; repeat use of ulipristal is not recommended.8
Conclusion The changes in emergency from the Yuzpe regime to levonorgestrel and then from two to one doses, providing a more effective treatment with fewer side effects and easier to take, were based on strong evidence. The evidence supporting a switch to ulipristal is much weaker. The wholesale replacement of levonorgestrel with ulipristal would seem to be inappropriate based on the equivocal superior efficacy, the small size of any benefit, the known safety of levonorgestrel, the unknown effect of ulipristal when given inadvertently in early pregnancy, and the possible impact on hormonal contraception started after ulipristal. Restricting ulipristal to women presenting at days 4 and 5 after intercourse is also inappropriate as the comparative evidence for ulipristal is extremely limited. Efficacy of levonorgestrel when used in days 4 and 5 has been demonstrated on the same basis as for ulipristal7 in the WHO 2002 trial5 and an application for a licence for use on days 4 and 5 would be likely to succeed. Levonorgestrel should be offered to these women, and nurse and pharmacy PGDs should reflect this. Women judged to be at higher risk of pregnancy might benefit from ulipristal, as the absolute benefit could be greater, although the judgment of risk is frequently wrong, making it difficult to select the women for ulipristal. The suggestion that non-use of ulipristal risks litigation2 is unfounded, because of the weakness of the evidence, and the small size of any potential benefit. As the odds ratio is greater than 0.5, any pregnancy that follows levonorgestrel use is as likely as not to have occurred with ulipristal. ellaOne is a new option in emergency contraception that is at least as effective as levonorgestrel. However, in levonorgestrel we have a last chance contraceptive that is of proven efficacy and safety. The problem with levonorgestrel is less its efficacy than its non-use. If the levels of unprotected intercourse are to continue, available, safety and price are key elements in the reduction in unwanted pregnancies.
References 1. Creinin MD, Schlaff W, Archer DF, Wan L, Frezieres R, Thomas M, et al. Progesterone receptor modulator for emergency contraception: a randomized controlled trial. Obstetrics & Gynecology 2006;108(5):1089-97. 2. Glasier AF, Cameron ST, Fine PM, Logan SJ, Casale W, Van Horn J, et al. Ulipristal acetate versus levonorgestrel for emergency contraception: a randomised non-inferiority trial and metaanalysis. Lancet 2010;375(9714):555-62. 3. Fine P, Mathe H, Ginde S, Cullins V, Morfesis J, Gainer E, et al. Ulipristal acetate taken 48-120 hours after intercourse for emergency contraception. Obstetrics & Gynecology 2010;115(2 Pt 1):257-63. 4. Cheng L, Gulmezoglu AM, Piaggio G, Ezcurra E, Van Look PF. Interventions for emergency contraception. Cochrane Database Syst Rev 2008(2):CD001324. 5. von Hertzen H, Piaggio G, Ding J, Chen J, Song S, Bartfai G, et al. Low dose mifepristone and two regimens of levonorgestrel for emergency contraception: a WHO multicentre randomised trial. Lancet 2002;360(9348):1803-10. 6. Task Force on Postovulatory Methods of Fertility Regulation. Randomised controlled trial of levonorgestrel versus the Yuzpe regimen of combined oral contraceptives for emergency contraception. Lancet 1998;352(9126):428-33. 7. EMEA. CHMP Assessment Report for Ellaone, 2009. 8. HRA Pharma. Summary of Product Characteristics for ellaOne, 2009.
