全文

Luteal phase support with progesterone is the standard of care for assisted reproductive technology cycles. In the United States, progesterone in oil injected intramuscularly has traditionally been the most popular form of luteal support, whereas in Europe vaginal administration of progesterone has predominated. Several randomized studies and a meta-analysis have now shown that pregnancy outcomes are similar with vaginal and intramuscular progesterone delivery for luteal support in fresh in vitro fertilization (IVF) cycles. There are disadvantages to each delivery method. In some cases intramuscular progesterone injections may be painful, especially if prolonged for up to 10 weeks. Vaginal progesterone may be inconvenient, because it is generally required to be inserted multiple times daily and may cause vaginal irritation in some women. Physicians and patients have been reluctant to change from the route of administration that has proven to be successful in the past.

In a stimulated IVF/intracytoplasmic sperm injection (ICSI) cycle, a large dose of hCG is used to trigger the final stage of oocyte maturation before oocyte retrieval. A 10,000 IU bolus of hCG results in elevated serum levels of hCG for more than 5 days (1) and provides support for the multiple corpora lutea present after retrieval. As a result, there is significant continuous endogenous progesterone secretion underlying the supplementary luteal phase support with either vaginal or intramuscular progesterone. It is therefore not surprising that no difference in pregnancy rates occurs with either route of progesterone administration.

The focus of this “Inklings” is on frozen embryo transfer (FET) or donor oocyte embryo transfer where hormonal therapy is used for endometrial preparation and no functional corpus luteum is present. Because of the absence of a corpus luteum providing underlying progesterone secretion, one can not really extrapolate the luteal support data from stimulated IVF cycles to FET. In addition, there are few, if any, randomized comparative trials between progesterone in oil and vaginal progesterone for luteal phase support in FET or donor oocyte cycles.

Like most clinics in Europe, we have used only vaginal progesterone for luteal phase support over the past 20 years. Our hormonal preparation of the endometrium for FET includes administration of micronized E2 orally or vaginally in doses of 4–8 mg daily for 10 days before vaginal ultrasound for endometrial thickness and pattern. Ideally, E2 administration is continued until a triple-line endometrium of ≥8 mm thickness is observed, at which time vaginal progesterone suppositories (200 mg three times daily) are initiated. Blastocyst transfer is done 6 days after starting progesterone.

We became interested in exploring options for luteal support recently as a result of performing mock embryo transfer cycles in women who had failed to conceive with good-quality blastocyst transfers. We performed endometrial biopsies in cycles we prepared with estrogen and progesterone as described above, and we determined that the endometrium seemed to be out of phase by more than 2 days in the midluteal phase in some of these women. However, of more interest was the observation that many of the mock cycles were associated with more than four endometrial contractions per minute on the day of the proposed embryo transfer. Multiple endometrial waves in the luteal phase have been associated with a lower pregnancy rate, based on the pioneering work of Fanchin et al. in France (2). In addition, we have experienced some ectopic pregnancies with embryo transfer. Because embryos are placed in the midendometrial cavity under ultrasound guidance, the occurrence of a tubal ectopic pregnancy can be explained only by abnormal endometrial motion that squeezes the embryo up into the proximal fallopian tube. Previous sonographic studies looking at the fate of a bolus of a suspension of soluble galactose microparticles placed in the endometrial cavity under ultrasound guidance have demonstrated movement of the bolus into the cervix or into the fallopian tubes in certain patients, consistent with abnormal uterine contractility (3).

Once we discovered that several patients seemed to have excessive endometrial waves on ultrasound monitoring, we tried switching to intramuscular progesterone in oil. We found that uterine activity ceased or was reduced to one contraction per minute within 24 hours of the progesterone in oil injection. In subsequent cycles, we used 3 days of vaginal progesterone followed by 3 days of progesterone in oil before frozen/warmed blastocyst transfer. Ultrasound on the day before FET and on the morning of FET showed quiet endometrium with no wave activity. Because it is thought that blastocyst attachment occurs a few hours after embryo transfer and that implantation occurs within days, we switched back to vaginal progesterone 1 or 2 days after FET. Some women preferred intramuscular progesterone injections and continued with that treatment until pregnancy was confirmed.

It is known that estrogen increases uterine contractility and subendometrial wave activity and that progesterone antagonizes this action to quiet the uterus and reduce endometrial waves. Endometrial concentrations of progesterone are higher after vaginal administration compared with intramuscular administration, even though serum concentrations are lower with vaginal progesterone use (4). However, the short half-life of natural progesterone dictates that multiple administrations are required vaginally to maintain relatively normal luteal phase levels. In contrast, progesterone in oil results in a depot effect and continuous release of progesterone over time, with an elimination half-life of >1 day, allowing once-daily injections. It is possible that the continuous exposure of the uterus to progesterone with intramuscular injection, rather than the intermittent peaks of absorption with vaginal administration, results in more sustained uterine quiescence. In addition, the long gap in vaginal progesterone use during the night may reduce the uterine progesterone relaxing effect, especially if the morning dose is withheld on the day of embryo transfer, which was our former practice. This long gap in progesterone exposure does not occur with intramuscular progesterone. We think that the uterine relaxation resulting from intramuscular progesterone reduces the chance of displacement of transferred embryos from the midendometrial cavity.

Perhaps a similar situation is related to the use of GnRH antagonist cycles in which a GnRH agonist trigger is used for oocyte maturation. The clinical pregnancy rate with this protocol, often used for prevention of ovarian hyperstimulation syndrome (OHSS), has been reported to be very low as a result of a luteal phase defect that does not seem to be rescued by vaginal progesterone administration. In contrast, one study from the United States has demonstrated clinical pregnancy and live birth rates equivalent to hCG triggering in an antagonist protocol using a GnRH agonist trigger when intramuscular progesterone in oil was used. In that study, the intramuscular progesterone dose was adjusted up to 75 mg/d to maintain circulating progesterone levels of >72 nmol/L (5). In contrast, with the use of intermittent vaginal progesterone it is possible that excessive endometrial contractility occurred, especially with supraphysiologic E2 levels in these hyperresponders at risk of OHSS. The luteolysis resulting from administration of short-acting GnRH agonist would make these cycles more like estrogen/progesterone-prepared FET cycles lacking a corpus luteum. In this situation, a high dose of progesterone in oil with constant release may overcome the estrogen effect and suppress uterine contractility. This speculation could be easily confirmed by ultrasound monitoring of endometrial waves.

In conclusion, for FET or donor oocyte cycles, we propose that the use of progesterone in oil injections may reduce uterine contractility and endometrial wave activity better than vaginal progesterone suppositories. Intramuscular progesterone may be beneficial for luteal support, at least until blastocyst attachment or implantation has occurred, when a switch back to vaginal progesterone could be considered. Similarly, in GnRH antagonist cycles where GnRH agonist use is considered for triggering follicle maturation before oocyte retrieval, luteal phase support with intramuscular progesterone may be beneficial in reducing endometrial wave activity and enhancing implantation and pregnancy rates.