Back HIV Populations Women & HIV CROI 2015: Levonorgestrel Contraception May Be Less Effective for Women Taking Efavirenz

CROI 2015: Levonorgestrel Contraception May Be Less Effective for Women Taking Efavirenz


African women who took efavirenz as part of a combination antiretroviral regimen were more likely to become pregnant while using the levonorgestrel implant, likely due to a drug interaction that lowers levels of the hormonal contraceptive, according to a presentation at the recent 2015 Conference on Retroviruses and Opportunistic Infections (CROI) in Seattle.

Efavirenz (Sustiva or Stocrin, also in the Atripla single-tablet regimen) is a commonly used antiretroviral therapy (ART) component worldwide and is generally safe and highly effective. The levonorgestrel subdermal (inserted under the skin) implant is widely recommended for long-acting (4-5 years) reversible contraception, and it has an expected failure rate of less than 1%.

Kimberly Scarsi from the University of Nebraska Medical Center and colleagues conducted a pharmacokinetic analysis of efavirenz-based ART in women using the levonorgestrel implant.

Efavirenz may decrease levonorgestrel concentrations due to a drug interaction mediated by CYP3A4, an enzyme that metabolizes drugs in the liver, the researchers noted as background. They hypothesized that over 1 year of combined use efavirenz would reduce levonorgestrel exposure, but not below the 180 pg/mL threshold thought to be necessary for effective contraception.

This longitudinal, parallel group, pharmacokinetic analysis enrolled 40 women with HIV in Uganda who desired an implant for contraception. All were black African and the median age was 30 years.

Participants in this non-randomized, open-label study were classified into 2 groups: 20 were still in good health and were not yet eligible for ART according to national guidelines (median CD4 T-cell count 758 cells/mm3), while 20 were taking ART consisting of efavirenz plus tenofovir/emtricitabine and had undetectable HIV viral load (median CD4 count 568 cells/mm3). Women in the ART group had taken efavirenz for at least 1 month, with a median of 10 months. The ART-treated group had a significantly lower average body weight compared to the untreated group (59 vs 73 kg).

All participants had a standard-dose levonorgestrel implant inserted into their upper arm at study entry. At each follow-up visit they received a pregnancy test. Blood was drawn at 1, 4, 12, 24, 36, and 48 weeks after study entry and plasma levonorgestrel concentrations were measured.

There were 3 women in the untreated control group who were excluded from the analysis due to a pregnancy detected at week 1, ART initiation, and self-removal of the levonorgestrel implant.


  • Pharmacokinetic analysis showed that levonorgestrel concentrations in the efavirenz group were 45%-57% lower than those in the untreated control group starting at week 1 and persisting through week 48.
  • The levonorgestrel area under the curve -- a measure of concentration over time -- was 48% lower in women taking efavirenz compared to women in the control group (11.60 vs 22.24 ng*wk/mL).
  • 3 women -- or 15% -- in the efavirenz group became pregnant between weeks 43 and 48, while no pregnancies occurred in the untreated control group.
  • Levonorgestrel concentrations in these 3 women were 122 pg/mL at 2 weeks prior, 299 pg/mL at 2 weeks prior, and 303 pg/mL at 10 weeks prior to conception -- all above the level thought to be necessary for contraceptive effectiveness.
  • Based on these findings, the efavirenz study arm was halted.
  • At the last study visit, 15 women (75%) in the efavirenz group -- but none in the untreated control group -- had levonorgestrel concentrations below 303 pg/mL, the highest concentration at which pregnancy occurred.
  • Efavirenz concentrations in the pregnant women were similar to population norms and none of these women had a high body weight, which might have led to inadequate dosing.

A "high rate of unintended pregnancy" was observed following 48 weeks of combined levonorgestrel and efavirenz use, and levonorgestrel exposure was reduced by 45%-57%, the researchers concluded. In addition, the proposed threshold for levonorgestrel efficacy (180 pg/mL) was inadequate in this study population.

"The immediate clinical priority is frank discussion of potential increased risk [of pregnancy] and discussion of the pros and cons of other forms of contraception," Scarsi said.

Limited data support the use of subdermal implants with alternative antiretrovirals including nevirapine (Viramine) and lopinavir/ritonavir (Kaletra), but further investigation of novel subdermal implant dosing strategies and alternative ART options is urgently needed, the study authors emphasized.



KK Scarsi, KMDarin, S Nakalema, et al. Levonorgestrel Implant + EFV-Based ART: Unintended Pregnancies and Associated PK Data. 2015 Conference on Retroviruses and Opportunistic Infections. Seattle, February 23-24, 2015. Abstract 85LB.