Progesterone prepares the endometrium for pregnancy by stimulating proliferation in response to human chorionic gonadotropin (hCG) produced by the corpus luteum. This occurs in the luteal phase of the menstrual cycle. In assisted reproduction techniques (ART), progesterone and/or hCG levels are low, so the luteal phase is supported with progesterone, hCG or gonadotropin-releasing hormone (GnRH) agonists to improve implantation and pregnancy rates. The ideal method of luteal support remains unclear.
To determine the relative effectiveness and safety of methods of luteal phase support provided to subfertile women undergoing assisted reproduction.
We searched databases including the Cochrane Menstrual Disorders and Subfertility Group (MDSG) Specialised Register, the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE, PsycINFO and trial registers. We conducted the final search in November 2014.
Randomised controlled trials (RCTs) of luteal phase support using progesterone, hCG or GnRH agonist supplementation in ART cycles.
Three review authors independently selected trials, extracted data and assessed risk of bias. We calculated odds ratios (ORs) and 95% confidence intervals (CIs) for each comparison and combined data when appropriate using a fixed-effect model. Our primary outcome was live birth or ongoing pregnancy. The overall quality of the evidence was assessed using GRADE (Grades of Recommendation, Assessment, Development and Evaluation) methods.
A total of 93 RCTs (25,471 women) were included. Most studies had unclear or high risk of bias in most domains. The main limitations in the evidence were poor reporting of study methods and imprecision due to small sample sizes. hCG vs placebo/no treatment (five RCTs, 746 women)No conclusive evidence was found of differences between groups in live birth or ongoing pregnancy; findings suggest benefit for the hCG group with a fixed-effect model (OR 1.76, 95% CI 1.08 to 2.86, three RCTs, 527 women, I(2) = 24%, very low-quality evidence) but not with a random-effects model (OR 1.67, 95% CI 0.90 to 3.12). hCG increased the risk of ovarian hyperstimulation syndrome (OHSS) (OR 4.28, 95% CI 1.91 to 9.6, low-quality evidence). Progesterone vs placebo/no treatment (eight RCTs, 875 women)Evidence suggests a higher rate of live birth or ongoing pregnancy in the progesterone group (OR 1.77, 95% CI 1.09 to 2.86, five RCTs, 642 women, I(2) = 35%, very low-quality evidence). OHSS was not reported. Progesterone vs hCG regimens (16 RCTs, 2162 women) hCG regimens included comparisons of progesterone versus hCG and progesterone versus progesterone + hCG. No evidence showed differences between groups in live birth or ongoing pregnancy (OR 0.95, 95% CI 0.65 to 1.38, five RCTs, 833 women, I(2) = 0%, low-quality evidence). Progesterone appeared to reduce the risk of OHSS (OR 0.5, 95% CI 0.34 to 0.76, five RCTs, 1349 women, low-quality evidence). Progesterone vs progesterone with oestrogen (16 RCTs, 2577 women)No evidence was found of differences between groups in live birth or ongoing pregnancy (OR 1.12, 95% CI 0.91 to 1.38, nine RCTs, 1651 women, I(2) = 0%, low-quality evidence) or OHSS (OR 0.56, 95% CI 0.2 to 1.63, two RCTs, 461 women, I(2) = 0%, low-quality evidence). Progesterone vs progesterone + GnRH agonist (seven RCTs, 1708 women)Live birth or ongoing pregnancy rates were lower in the progesterone-only group (OR 0.67, 95% CI 0.56 to 0.81, seven RCTs, 2113 women, I(2) = 69%, low-quality evidence). Statistical heterogeneity for this comparison was high because of unexplained variation in the size of the effect, but the direction of effect was consistent across studies. OHSS was not reported by any of the studies. Progesterone regimens (45 RCTs, 13,814 women)The included studies reported nine different comparisons between progesterone regimens. Findings for live birth or ongoing pregnancy were as follows: intramuscular (IM) versus oral: OR 0.71, 95% CI 0.14 to 3.66 (one RCT, 40 women, very low-quality evidence); IM versus vaginal/rectal: OR 1.24, 95% CI 1.03 to 1.5 (seven RCTs, 2309 women, I(2) = 71%, very low-quality evidence); vaginal/rectal versus oral: OR 1.19, 95% CI 0.83 to 1.69 (four RCTs, 857 women, I(2) = 32%, low-quality evidence); low-dose versus high-dose vaginal: OR 0.97, 95% CI 0.84 to 1.11 (five RCTs, 3720 women, I(2) = 0%, moderate-quality evidence); short versus long protocol: OR 1.04, 95% CI 0.79 to 1.36 (five RCTs, 1205 women, I(2) = 0%, low-quality evidence); micronised versus synthetic: OR 0.9, 95% CI 0.53 to 1.55 (two RCTs, 470 women, I(2) = 0%, low-quality evidence); vaginal ring versus gel: OR 1.09, 95% CI 0.88 to 1.36 (one RCT, 1271 women, low-quality evidence); subcutaneous versus vaginal gel: OR 0.92, 95% CI 0.74 to 1.14 (two RCTs, 1465 women, I(2) = 0%, low-quality evidence); and vaginal versus rectal: OR 1.28, 95% CI 0.64 to 2.54 (one RCT, 147 women, very low-quality evidence). OHSS rates were reported for only two of these comparisons: IM versus oral, and low versus high-dose vaginal. No evidence showed a difference between groups in either case. Progesterone and oestrogen regimens (two RCTs, 1195 women)The included studies compared two different oestrogen protocols. No evidence was found to suggest differences in live birth or ongoing pregnancy rates between a short and a long protocol (OR 1.08, 95% CI 0.81 to 1.43, one RCT, 910 women, low-quality evidence) or between a low dose and a high dose of oestrogen (OR 0.65, 95% CI 0.37 to 1.13, one RCT, 285 women, very low-quality evidence). Neither study reported OHSS.
Progesterone appears to be the best method of providing luteal phase support, as it is associated with higher rates of live birth or ongoing pregnancy than placebo, and lower rates of OHSS than hCG. Moreover, addition of one or more doses of GnRH agonists to progesterone is associated with higher live birth and ongoing pregnancy rates than progesterone alone. Overall, addition of other substances such as oestrogen or hCG does not seem to improve outcomes, and hCG is associated with higher risk of OHSS.The route of progesterone administration does not seem to matter.