Chapter  167:  Effectiveness of Assisted Reproductive Technology

Ovulation Induction

Clomiphene is an effective first-line therapy for women with PCOS. Metformin is, at best, no more effective, and, based on a large multi-center trial, less effective than clomiphene alone.

Although a statistically significant effect is not observed in individual studies, meta-analyses do demonstrate a significant increase in pregnancy rates in clomiphene-resistant women treated with metformin, a finding which should be confirmed in large studies. There is insufficient evidence to draw conclusions about the relative efficacy of aromatase inhibitors.

Use of laparoscopic cauterization of the ovaries, followed by ovulation induction if necessary, results in similar pregnancy and live birth rates, with significantly lower multiple gestation rates, compared to immediate gonadotropin use in clomiphene-resistant women; these rates may be further improved by the addition of metformin, although there are no data on possible long-term adverse outcomes of cautery.

Superovulation in Ovulatory Women

Pooled data show significantly higher pregnancy rates with gonadotropins compared to clomiphene or aromatase inhibitors; there are trends toward higher rates of live birth, multiple pregnancy and hyperstimulation with gonadotropins, but study sizes are too small to draw definite conclusions regarding relative efficacies of these ovulation-inducing therapies.

There do not appear to be substantial differences in pregnancy rates between different gonadotropin preparations. Higher doses increase the risk of multiples and hyperstimulation without significant improvement in pregnancy rates. The addition of gonadotropin-releasing hormone (GnRH) antagonists to superovulation protocols may increase both pregnancy rates and twin gestation rates. Further studies adequately powered for the outcome of live birth per couple are needed.

ART-the Female Partner

No clear superiority of any specific protocol for pituitary down-regulation with GnRH agonists was identified.

Although only one individual study comparing GnRH agonists to antagonists found a significant difference in pregnancy or live birth rates (in favor of agonists), published meta-analyses show significantly higher pregnancy and live birth rate with the use of agonists. Antagonists do result in significant decreases in gonadotropin requirements, and a significant decrease in the risk of ovarian hyperstimulation syndrome (OHSS).

Pooled results of individual trials of gonadotropin preparations suggest that human menopausal gonadotropins are superior in terms of pregnancy and live birth rates compared to recombinant follicle stimulating hormone (rFSH) in long protocol GnRH agonist regimens, with higher multiple pregnancy rates, and that the addition of recombinant luteinizing hormone (rLH) to rFSH improves live birth rates in poor responders. Based on differences in the amount of gonadotropin required, there may be economic advantages to some formulations.

Timing of human chorionic gonadotropin (hCG) administration for triggering oocyte maturation is important for optimizing live birth rates, but the optimal timing and threshold relative to follicular growth have not been determined. There does not appear to be any difference in pregnancy or live birth rates, or other major outcomes, between recombinant hCG and urinary hCG, although injection site reactions are more common with urinary hCG. In cycles using a GnRH antagonist for pituitary down-regulation, use of hCG is superior to use of a GnRH agonist.

There is insufficient evidence to determine the optimal method for endometrial preparation for frozen-thawed embryo transfer.

Ultrasound-guided embryo transfer consistently results in substantially improved (40 percent relative increase) pregnancy and live birth rates compared to various “clinical touch” methods. The consistency of this finding and the size of the effect are striking considering that the majority of interventions evaluated in this review do not show significant differences.

Some form of luteal support is necessary with ART, since both progesterone and hCG result in improved pregnancy rates compared to no treatment. Although there is no detectable difference between oral progesterone and the various formulations of vaginal progesterone, both result in lower pregnancy and live birth rates compared to intramuscular progesterone. The addition of estrogen to progesterone may improve outcomes, although additional larger studies are needed to confirm these findings.

The non-steroidal anti-inflammatory drug (NSAID) piroxicam significantly improved pregnancy and live birth rates in a general ART population, and further studies of NSAIDs are warranted. Randomized trials of intercessory prayer and acupuncture showed benefit, but there are remaining methodological questions (particularly the most appropriate control intervention) which need to be addressed.

ART-the Embryo

ART results in much higher birth rates within 90 days than watchful waiting in eligible patients, although cumulative pregnancy rates were similar in one trial comparing ART to intrauterine insemination (IUI) and IUI after ovarian stimulation. There is no evidence of benefit for intracytoplasmic sperm injection (ICSI) compared to ART in patients with non-male factor infertility. Laboratory procedures used during fertilization, such as media and equipment used, may have significant impact on outcomes.

Assisted hatching improves pregnancy and live birth rates in couples with previous ART failure, but there is insufficient evidence to draw inferences about benefits in other groups.

Blastocyst transfer results in better live birth rates than day 3 transfer, especially in patients with a good prognosis. The disadvantage of delaying transfer is a reduction in the number of embryos available for transfer and for cryopreservation, and an increased risk of monozygotic twinning.

Although double embryo transfer results in higher pregnancy and live birth rates compared to single embryo transfer, multiple rates - almost all twins - are consistently higher. Strategies involving alternative methods for pituitary down-regulation, or involving multiple cycles with fewer embryo transfers per cycle, appear to result in similar live birth rates with fewer multiples.

Long-Term Outcomes

Review of the literature on this topic included the inherent limitations of observational studies compared to randomized trials, difficulty in identifying appropriate controls, changes in clinical practice which may make findings about older treatments obsolete, and issues relating to generalizability of findings between countries.

Loss of the entire pregnancy is more common for singleton pregnancies than for twins after ART, suggesting that factors associated with successful implantation and placentation contribute to the likelihood of both multiple gestation and a successful pregnancy outcome.

False positive results for maternal testing for chromosomal abnormalities after assisted reproduction are more likely for second trimester serum screening, resulting in an increased false positive rate with combined screening strategies that incorporate both modalities.

Preterm delivery is approximately twice as likely in women pregnant with singleton pregnancies after infertility treatment compared to spontaneous singleton pregnancies. The evidence is most consistent for ART, but the risk was also increased in a large study of women pregnant after ovulation induction alone. The proportion of preterm deliveries that are indicated due to maternal/fetal complications versus those due to spontaneous preterm labor is unclear. Conversely, the risk of preterm birth in ART twins compared to spontaneous twins is either not elevated, or elevated to a lesser extent than in singletons, in the majority of studies.

Much of the elevated risk of low birth weight is due to the increased risk of preterm birth. However, studies that examined gestational age-specific weights found an increased risk of small-for-gestational age (SGA) infants among singleton, but not twin, pregnancies after infertility treatment.

Women pregnant after infertility treatment are at increased risk for disorders potentially related to abnormal implantation, including preeclampsia, placenta previa, and placental abruption. The extent to which specific treatments or underlying maternal/embryonic characteristics contribute to this risk is unclear.

Risks for major congenital anomalies are increased after infertility treatment, but much of this risk appears to be related to maternal and/or paternal characteristics, including a history of subfertility or infertility. Given the relative rarity of specific birth defects or syndromes, identifying an association between a specific exposure and subsequent risk is difficult.

In the neonatal period, although there is evidence of an increased risk for adverse outcomes, especially among singletons, it is unclear to what extent this is due to the observed increased preterm delivery rate. Large-scale studies that control for gestational age and birth weight are needed. In later infancy, there is a significantly increased hospitalization rate among children born after ART compared to the general population, but rates are similar when compared to children born to couples with a history of treated and untreated subfertility.

Children born after assisted reproduction have an increased risk of hospitalization and surgery compared to general population controls. There does not appear to be an increased risk of childhood cancers in children conceived after infertility treatments.

The available evidence suggests that there is not an increase in the risk of adverse neurodevelopmental outcomes in children born after infertility treatment that is not associated with the underlying condition of infertility or the well-established increased risk of prematurity and SGA. The available evidence on learning and other developmental outcomes is reassuring, but larger studies across a wider population are needed.

In general, infertility treatments involving ovarian stimulation do not appear to be associated with an increased risk of breast cancer, although non-significantly elevated risks were seen 20 years after exposure in one study, suggesting that continued monitoring is warranted.

Ovarian cancers are strongly associated with an infertility diagnosis; use of ovulation stimulating drugs does not appear to increase the risk above baseline levels in this patient population. As with breast cancer, increasing risk with increased duration with treatment cannot be ruled out with confidence.

Based on the available literature, there are no differences in psychological outcomes, including parenting skills, when comparing singleton pregnancies resulting from ART to spontaneous conceptions. If anything, mothers of infants resulting from ART have better outcomes, although there is some evidence that fathers may do worse on some scales. Multiple gestations significantly increase stress and depressive symptoms, especially for mothers of infants with chronic disabilities; to the extent that women undergoing ART are more likely to experience multiples, especially preterm multiples, they are more likely to experience these symptoms.

I. Sources

The primary source of literature was MEDLINE^® (1966–January Week 4 2008). Searches of this database were supplemented by a search of the Cochrane Database of Systematic Reviews, and by a review of the reference lists of included articles and relevant review articles and meta-analyses.

II. Search Strategies

The basic MEDLINE^® search strategy used the National Library of Medicine's Medical Subject Headings (MeSH) key word nomenclature. Searches were limited to articles published in English. The exact search string used is given in Appendix A.^* Relevant reviews in the Cochrane Database of Systematic Reviews were identified by hand searching the list of reviews published by the Menstrual Disorders and Subfertility Group, which covers all topics relevant to this report. All search strategies combined yielded a total of 5294 citations, whose records are maintained in a ProCite (Thompson ISI ResearchSoft, Berkeley, CA) database.

III. Screening of Abstracts

Paired clinicians from the Duke research team independently reviewed abstracts and classified each as included or excluded according to project-specific criteria, which they also developed. An abstract was included for full-text review if at least one of the paired reviewers recommended that it be included.

The inclusion criteria applied at the abstract screening stage were:

• N ≥ 50 if not a randomized controlled trial (RCT; smaller RCTs were acceptable); and

• Female age ≤ 45; and

• Study relevant to at least one of the key questions, as follows:

  • Compares outcomes of ovulation induction or ART based on presence/absence or differing levels of biological, environmental, or other factors (Question 1b); and/or

  • Reports sensitivity/specificity of diagnostic tests for predicting the likelihood of different outcomes of ovulation induction or ART; or study reports “associations” or “correlations” between test results and outcomes (Question 1c); and/or

  • Reports benefits and risks of treatment with Clomid^®, Pergonal^®, other injectable super-ovulatory drugs, or Glucophage^® in various populations (Question 2); and/or

  • Reports pregnancy and/or live birth rates of ART (Question 3); and/or

  • Reports adverse outcomes (including quality-of-life measures) of ovulatory drug-induced pregnancies and of pregnancies achieved with IVF based on either (i) history of infertility or (ii) treatment (Question 4).

When these screening criteria were applied, a total of 2712 citations were included for further review at the full-text stage.

IV. Screening of Full Texts

At the full-text screening stage, paired researchers independently reviewed the articles that had passed the abstract screening and indicated a decision to include or exclude them for data abstraction for one or more of the key questions. When the two reviewers arrived at different decisions about inclusion/exclusion or about question assignment for a given article, they were asked to reconcile their differences. The question-specific screening criteria applied at the full-text stage are described in Table 1.

Summaries of the results of the abstract screening and full-text review are provided in Tables 2 and 3. A list of excluded articles, with reasons for exclusion, is provided in Appendix B.

I. Research Question

Among women of reproductive age, what are the benefits and risks of Clomid^® and Pergonal^® (or other injectable super-ovulatory drugs) and Glucophage^®, and how do they vary in different patient populations? Different patient populations include racial/ethnic groups and age by decade (or age groups comparable to those in the Centers for Disease Control and Prevention [CDC]-Society for Assisted Reproductive Technology [SART] national assisted reproductive technology [ART] success rates reports¹⁴). Risks include high rates of higher order multiples and ovarian hyperstimulation syndrome. Benefits include reduced time to achieve pregnancy, correction of ovulatory dysfunction, possible decreased miscarriage rates, and decreased gestational diabetes risk with Glucophage^®.

II. Approach

Agents that promote ovulation are used in two specific subgroups of infertile patients. First, the single most common etiology for infertility in the United States is anovulation or oligo-ovulation, most commonly as part of the polycystic ovarian syndrome (PCOS).⁴⁰ Without ovulation, conception and pregnancy cannot occur; in these patients, use of techniques that stimulate ovulation is oriented towards correcting the primary etiology of infertility. We focused on treatment of anovulation solely in women seeking pregnancy: correction of endocrine abnormalities, including anovulation, in women not seeking pregnancy is clearly an important therapeutic goal, but the considerations in deciding on optimal therapy may be quite different.⁴¹ We did not include studies of women with anovulation due to hypothalamic amenorrhea or premature ovarian failure.

A second group of patients includes couples with unexplained infertility, mild male factor infertility, or other non-tubal etiologies. In theory, given patent fallopian tubes, normal uterine anatomy, and functional tubes, increasing the number of eggs produced in a given cycle increases the probability of conception. In these patients, use of ovulation-inducing agents is aimed at producing multiple eggs in a given cycle (superovulation), in order to increase the chances of conception. Given these very different patient populations and therapeutic goals, we began our review by separating included studies between those which specifically corrected anovulation in women with PCOS and those which involved superovulation in women with normal ovulatory function.

For each category of patient, we further divided studies by the types of intervention used. For anovulatory women, these were: (a) inhibitors of estrogen action (including anti-estrogens such as clomiphene citrate, e.g., Clomid®, and aromatase inhibitors such as letrozole; as a group, we refer to these as estrogen inhibitors); (b) insulin sensitizers (such as metformin, or Glucophage^®); (c) gonadotropins (such as human menopausal gonadotropins, e.g., Pergonal®); (d) combination therapies; and (e) surgical therapies. For ovulatory women, we used the same categories, with the exception of insulin sensitizers. Since intrauterine insemination (IUI) is often included as part of the ovulation induction or superovulation regimen, we also included studies which addressed specific aspects of IUI in each group.

As described in the Methods chapter, we excluded all non-randomized studies, as well as “quasi-randomized” studies (such as those where treatment assignment was based on alternate history numbers or clinic days). For this topic, the primary outcome of interest was the cumulative number of clinical pregnancies or, preferably, live births per couple; wherever possible, we used the number of women/couples randomized as the denominator. We excluded any study where these outcomes were not reported or calculable from the presented results. Some studies used crossover designs. Because a crossover design requires the assumption that all cycles are equivalent, and ignores the implications of different pregnancy rates in the first cycle on the subjects in the second cycle, interpretation of the results of crossover studies of infertility treatments is extremely problematic.³⁶ Therefore, we included crossover studies only if the results for the first cycle were presented separately.

For the primary outcomes, relative risks (RRs) with 95 percent confidence intervals (CIs) were calculated from the presented results. Because of substantial clinical heterogeneity in the studies in terms of patient characteristics (such as body mass index [BMI] in studies of PCOS) and treatment regimens, we did not perform formal meta-analyses.

Results for other outcomes, such as multiple pregnancy or spontaneous abortion rates, are summarized in the text. The majority of included studies were extremely limited in power to detect differences in the primary outcomes, let alone any differences in other less common outcomes. Outcomes related to later pregnancy and longer term maternal and child outcomes are discussed under Question 4.

Please note that in the summary tables throughout this chapter, estimates of relative effect with CIs that do not cross 1 (i.e., estimates that are statistically significant) are bolded for emphasis.

III. Search Results

The flow of articles on this topic through the literature search and screening process is depicted in Figure 3.

IV. Induction of Ovulation in Anovulatory Women

A. Drugs for inducing ovulation-estrogen inhibitors. PCOS is a condition marked by anovulation, hyperandrogenism, and insulin resistance. Common clinical manifestations include oligo- or amenorrhea, acne, hirsutism, and obesity.⁴² The mainstay of treatment for many years has been clomiphene citrate (CC); clomiphene is a non-steroid which chemically resembles tamoxifen, and, like tamoxifen, it has both estrogen agonist and antagonist effects at the level of the estrogen receptor; it promotes the release of follicle-stimulating hormone (FSH) from the pituitary, with subsequent follicular development and ovulation in the ovary.⁴³ Trials prior to 2000 demonstrated that clomiphene is superior to placebo in achieving pregnancy in anovulatory women.⁴⁴

Recently, another class of estrogen inhibitors, aromatase inhibitors, has been explored as an alternative for ovulation induction. These agents, which have been shown to have efficacy in breast cancer patients, work by preventing the conversion of testosterone to estrogen via the enzyme aromatase.

This section reviews studies where estrogen inhibitors were the sole treatments for infertile women with PCOS. Studies where they are compared to other classes of agents, or studies with combination therapies, are described below.

1. Included studies. Five studies met our inclusion criteria (Table 4). All five had fewer than 50 subjects per arm, only two followed subjects for more than one cycle, and none reported live births.

In direct comparisons of estrogen inhibitors, the small sample sizes of comparisons of clomiphene to tamoxifen,⁴⁵ anastrozole,⁴⁶ and letrozole⁴⁷ result in wide confidence intervals for treatment efficacy.

Based on one small study, administration of clomiphene on cycle days 1–5 results in a significantly higher cumulative pregnancy rate than administration on cycle days 5–9 (RR 2.08; 95 percent CI 1.00–4.33).⁴⁸

None of the studies had sufficient numbers to draw any conclusions regarding other outcomes such as spontaneous abortion or multiple pregnancies.

2. Other published systematic reviews. In one published systematic review of clomiphene versus tamoxifen⁴⁹ involving four studies (three pre-2000) with a total of 243 subjects and 743 cycles, there was no significant difference in pregnancy rate per cycle (RR 1.06; 95 percent CI 0.58–1.91); pregnancy or live birth per couple were not calculable.

3. Cochrane reviews. The most recent Cochrane update was in November 2004.⁴⁴ Other than showing superiority of clomiphene to placebo, no comparison (tamoxifen vs. clomiphene, clomiphene plus tamoxifen vs. clomiphene alone, or letrozole vs. anastrozole) had sufficient numbers of patients to be able to reach any conclusions regarding relative efficacy in achieving pregnancy (Table 5).

4. Conclusions. Clomiphene citrate is superior to placebo in achieving pregnancy in anovulatory women; as such, it is a reasonable reference treatment for evaluation of other methods for induction of ovulation in this patient population. There is insufficient evidence to allow any inferences regarding the relative efficacy of other estrogen inhibitors compared to clomiphene.

B. Drugs for inducing ovulation - insulin-sensitizers. Interventions that improve insulin resistance, such as weight loss or treatment with specific drugs in women with PCOS can also lead to decreases in circulating androgens and ovulation. The most commonly used agent has been metformin; the most recent Cochrane review found significantly increased rates of ovulation with metformin compared to placebo (odds ratio [OR] 3.88; 95 percent CI 2.26–6.69).⁵⁰ A different class of insulin sensitizers, the thiazolidinediones, have also been investigated, although one agent that increased ovulation rates in PCOS patients in a randomized controlled trial (RCT), troglitazone,⁵¹ has subsequently been removed from the market due to hepatic toxicity. Potential advantages of insulin sensitizers for induction of ovulation compared to estrogen inhibitors or gonadotropins include correction of underlying metabolic abnormalities which may have adverse longer term cardiovascular consequences⁵² and reduced rates of multiple gestation. Although neither class of drugs is approved for use in pregnancy, there are enough data available for metformin to be placed in the U.S. Food and Drug Administration (FDA) Pregnancy Category B (human data reassuring), while thiazoledinediones are in Category C (insufficient data).⁵³

Although efficacy in establishing ovulation has been established, at least for metformin, the evidence available at the time of the Cochrane review was limited for pregnancy and live birth.⁵⁰ This section reviews the literature meeting our search criteria that provided data on pregnancy and live birth rates.

1. Included Studies. The following sections describe studies comparing metformin to placebo, metformin to other insulin sensitizers, and metformin to clomiphene. Studies that compared metformin in combination with other agents are described in the section on combination therapy.

We identified three studies⁵⁴–⁵⁶ comparing metformin to placebo that met our search criteria (Table 6). All three studies were small, ranging in size from 20 to 56 subjects. Two studies, one in new patients⁵⁴ and one in patients who had previously failed to ovulate with clomiphene treatment,⁵⁵ had non-significant increases in pregnancy rates; the third trial⁵⁶ had only three pregnancies in 20 subjects.

Two small studies compared metformin to rosiglitazone⁵⁷ or pioglitazone⁵⁸(Table 6). Neither study had sufficient power to demonstrate any difference in pregnancy or live birth rates, and the study by Ortega-Gonzalez and colleagues⁵⁸ was not designed as an infertility trial.

Two RCTs provided data which allowed direct comparison of metformin to clomiphene^6,59(Table 6). Both studies used a double-blind, double-dummy design, where women received either clomiphene plus placebo “metformin,” or metformin plus placebo “clomiphene,” and continued treatment for up to 6 months.

In a single center study, Palomba and colleagues randomized 50 women to each arm. The primary outcome was pregnancy rate, and the study was powered to detect a 30 percent absolute difference. Both ovulation and pregnancy rates were higher in the first two cycles with clomiphene, but higher with metformin in subsequent cycles.⁵⁹ Cumulative ovulation rates were similar (62.9 percent with metformin vs. 67 percent for clomiphene), but cumulative and ongoing pregnancy rates were significantly higher with metformin (RR for cumulative pregnancy rates 3.10; 95 percent CI 1.71–5.62; for ongoing pregnancy, RR 2.80; 1.53–5.13). Spontaneous abortion rates were higher in the clomiphene group. There were no multiple pregnancies in either arm, and no clear difference in pregnancy complications.

Contrasting results were found in a larger multi-center trial, the Pregnancy in Polycystic Ovary Syndrome (PPCOS) study, conducted by Legro and colleagues.⁶ This trial also included a third arm of active clomiphene plus metformin; these results are discussed separately in the combination therapy section. Randomization was stratified by center and history of prior therapy with either metformin or clomiphene (approximately 60 percent of subjects had previously received at least one of the experimental treatments, with 18 percent having received both). The primary outcome was live birth, powered to detect an absolute difference of 15 percent. Six hundred twenty-six women were randomized. Ovulation rates were significantly higher in the clomiphene only group compared to metformin (49 percent vs. 29 percent), and both pregnancy and live birth rates were substantially higher in the clomiphene only group (RR for live birth 0.33; 95 percent CI 0.19–0.57). There were three multiple pregnancies in the clomiphene-only group, none in the metformin group, with a non-significant trend towards higher pregnancy loss rates in the metformin group; there were no clear differences in pregnancy complications. Overall side effects were similar, with hot flashes and vaginal symptoms more common with clomiphene, and gastrointestinal symptoms more common with metformin.

From the published data, there is no clear explanation for the discrepant results of these two similarly designed studies. The main differences in the subject populations were prior treatment (none in the Palomba study, 60 percent in PPCOS) and BMI (restricted to less than 30 kg/m² in the Palomba study, while almost 20 percent of the PPCOS subjects had a BMI between 30 and 34 kg/m², and almost 50 percent had a BMI of 35 kg/m² or above). However, because of the large sample size and randomized design, these factors were equally distributed between treatment arms. In addition, post-hoc analyses based on BMI and history of prior treatment showed similar results for the comparison of metformin to clomiphene alone. Given the single center European setting versus the multi-center U.S. setting, and subsequent findings of genetic variability in response to metformin,⁶⁰ it is possible that variations in the distribution of relevant genes in different patient populations contributed to some of the difference.

2. Other published systematic reviews. We identified one published non-Cochrane review by Kashyap and colleagues.⁶¹ This review identified two studies with a total of 65 subjects comparing metformin to placebo, with a summary odds ratio of 1.07 (95 percent CI 0.20–5.74).

3. Cochrane reviews. The most recent Cochrane update was in December 2002.⁵⁰ Based on five studies with a total of 172 subjects, pregnancy rates were increased non-significantly with metformin compared to no treatment or placebo (OR 2.76; 95 percent CI 0.85–8.98); only two of these studies (n = 50) reported live birth rates (OR 1.00; 0.13–7.79).

4. Conclusions. Although the majority of randomized studies suggest that pregnancy rates are increased with metformin compared to placebo, the small number of trials, along with the small size of the trials, means that the results are non-significant for both individual studies and meta-analyses performed to date.

There is insufficient evidence to compare the efficacy of available thiazolidinediones to placebo, metformin, or any other currently used agent for induction of ovulation in women with PCOS.

Results of the two direct randomized comparisons of metformin to clomiphene are contradictory. The smaller single center study found metformin superior to clomiphene in achieving pregnancy, while a much larger multi-center study found clomiphene superior to metformin in achieving both pregnancy and live birth, results that were consistent regardless of BMI or history of prior therapy. Results for spontaneous abortion rates were similarly discrepant. Multiple pregnancies were only observed in women treated with clomiphene. Based on this evidence, we conclude that metformin is, at best, not superior to clomiphene in achieving pregnancy and live birth, and, based on the largest study, is inferior. Sample sizes are too small in the randomized trials to draw conclusions about spontaneous abortion or other pregnancy-related outcomes.

C. Drugs for inducing ovulation - gonadotropins. Approximately 20–40 percent of women with PCOS will fail to conceive in response to clomiphene.^62,63 One option for treating these women is stimulation with exogenous gonadotropins. Although effective in achieving pregnancy, there is an increased risk of both multiple pregnancies and ovarian hyperstimulation syndrome (OHSS).⁶⁴ The purpose of studies of variation in the type and/or dosing of gonadotropin is to determine optimal pregnancy and live births while minimizing multiple births and OHSS. This section reviews the existing evidence on the efficacy of various approaches to ovulation induction using gonadotropins in PCOS patients.

1. Included studies. The six identified studies are shown in Table 7. None of the studies had adequate power to detect differences in pregnancy rate. Because multiples and OHSS will be even less frequent than pregnancy, these studies were not able to provide any conclusive evidence regarding any gonadotropin-based method.

2. Other systematic reviews. We did not identify any other non-Cochrane published reviews.

3. Cochrane reviews. There are three relevant Cochrane reviews. The first⁷¹ was most recently updated in May 2000 and reviewed studies of gonadotropin therapy in PCOS. All studies were published prior to 2000, and neither pregnancy nor live birth per couple was reported or calculable. In five studies, FSH alone resulted in lower OHSS compared to human menopausal gonadotropins (hMG) when no gonadotropin-releasing hormone (GnRH) analog was used (OR 0.20; 95 percent CI 0.08–0.46); when GnRH agonists were used, overstimulation requiring cycle cancellation was significantly more frequent. OHSS was increased, but the confidence intervals for the OR include 1.0.

The second review⁷² was most recently updated in February 2001 and compared recombinant (rFSH) versus urinary FSH (uFSH) preparations. Using urinary FSH as the reference, there was no significant difference in pregnancy rate (OR 0.95; 95 percent CI 0.64–1.41), multiple gestations (0.44; 0.16, 1.21), or OHSS (1.55; 0.50, 4.84). Only one study (pre-2000) of different dosing regimens was included in the review. It compared a conventional regimen guided by ovarian response versus chronic low-dose rFSH and found non-significant differences in pregnancy rates (OR 1.62; 95 percent CI 0.65–4.07).

The third review of pulsatile GnRH administration⁷³ included only the study of Timmerman et al.;⁷⁰ with only 30 subjects, this study, like the majority of the others, was not powered to detect meaningful differences in pregnancy rates.

4. Conclusions. Based on pre-2000 studies included in the Cochrane review,⁷¹ use of FSH results in a lower incidence of OHSS compared with hMG, particularly if there is no concomitant pituitary suppression. There is insufficient evidence to determine the most effective form or regimen for administration of FSH for ovulation induction in women with PCOS who do not respond to clomiphene.

D. Drugs for inducing ovulation - combinations. Combinations of all three of the major classes of medical treatments for PCOS have been tested, along with other adjunctive therapies, both as primary treatment for PCOS and in women who fail to respond to a trial of clomiphene. This section describes studies that tested combinations of medical therapies, divided broadly by studies of first-line treatment and treatments in clomiphene-resistant women.

1. Included studies: first-line treatment. Summary RRs for included studies are shown in Table 8. Two studies compared metformin plus clomiphene to monotherapy in patients receiving initial therapy for infertility associated with PCOS. Moll and colleagues⁷⁴ randomized 225 women to clomiphene plus placebo or clomiphene plus metformin and found no difference in pregnancy rates (RR 0.87; 95 percent CI 0.64–1.18). In the previously described PPCOS study,⁶ clomiphene plus metformin was significantly more effective in achieving both pregnancy and live birth than metformin alone; live birth rates were increased, but not significantly, compared to clomiphene alone (RR 1.19; 0.85–1.67). This effect was seen in women with and without prior therapy. In another subgroup analysis, any benefit of adding metformin to clomiphene was limited to women with a BMI greater than or equal to 35, although the sample size was not sufficient to show statistical significance.

Two studies compared clomiphene alone to clomiphene with ultrasound monitoring of the ovaries and triggering of ovulation with human chorionic gonadotropin (hCG), followed by intercourse.^75,76 Pregnancy rates were increased in both, but not significantly (Table 8).

In one small study, the addition of ketoconazole to clomiphene resulted in significantly more live births (RR 2.24; 95 percent CI 1.01–4.95), with a trend towards reduced multiple pregnancies. This study was published in 2001, and we did not identify any subsequent similar studies in our search.

Because clomiphene has both agonist and antagonist effects on the estrogen receptor, depending on the target tissue, failure to conceive or early pregnancy loss in some women receiving clomiphene may be due to estrogen inhibiting effects in other sites in the reproductive tract. Two studies evaluating the addition of estrogens, either ethinyl estradiol⁷⁷ or phytoestrogens,⁷⁸ found significantly increased live birth rates compared to clomiphene alone (RRs of 4.6 and 6.0), with decreased spontaneous abortion rates. Again, we did not identify any other studies that would confirm these results.

2. Included studies: second-line treatment after initial failure with clomiphene. Summaries of study size and RRs are presented in Table 9.

Two small studies^80,81 suggest an improvement in pregnancy rates with the addition of metformin in women who have previously failed clomiphene treatment, although individual differences were not statistically significant. Another small study failed to show a significant difference with the addition of rosiglitazone.⁸²

Metformin also non-significantly increased pregnancy rates in two studies of gonadotropin use.^83,84

Three studies of different adjunct therapies demonstrated large and statistically significant improvements in pregnancy rates in clomiphene-resistant women compared to clomiphene alone: pre-treatment with oral contraceptives⁸⁵(RR 13.0; 95 percent CI 1.84–97.0); co-administration of n-acetyl-cysteine⁸⁶(RR 28.0; 1.7–488); and co-administration of dexamethasone⁸⁷(RR 8.00; 1.97–32.5). Of note, multiple gestation rates were increased with all three approaches. As is evident from the width of the confidence intervals, the combination of relatively small study size and lower event rates prevents precise estimates of efficacy, but the effect size for all suggests that further studies of each of these approaches with a focus on minimizing multiple gestation risk are warranted.

3. Other systematic reviews. One published non-Cochrane systematic review⁶¹ found an increased pregnancy rate with clomiphene plus metformin compared to clomiphene plus placebo in clomiphene-resistant women (OR 3.65; 95 percent CI 1.11–12.0).

The relevant Cochrane review⁴⁴(Table 10) showed significantly increased pregnancy rates with use of clomiphene plus dexamethasone (OR 11.3; 95% CI 5.33–24.1) and clomiphene after pre-treatment with oral contraceptives (OR 26.7; 4.91–145); both of these treatments also had substantial increases in multiple pregnancy rates, although confidence intervals included 1.0. The addition of metformin to gonadotropins was also superior to gonadotropins alone for pregnancy (OR 4.88; 2.46–9.67).

4. Conclusions. Based on two large randomized trials, the addition of metformin to clomiphene as first-line therapy does not appear to significantly increase pregnancy or live birth rates, although a subgroup analysis of the largest trials suggests that there may be benefit in women with a BMI greater than or equal to 35, a finding which should be confirmed in a larger study.

The addition of ketoconazole (one study) and estrogens (two studies) to clomiphene in first-line therapy resulted in significantly increased live birth rates due to decreased spontaneous abortion rates, findings which should be confirmed in larger trials.

Although a statistically significant effect is not observed in individual studies, meta-analyses do demonstrate a significant increase in pregnancy rates in clomiphene-resistant women treated with metformin. Whether these results translate into improved live birth rates should be confirmed in larger studies, although the lower overall birth rate in this population will require large studies.

Pre-treatment with oral contraceptives, co-treatment with n-acetyl-cysteine, and co-treatment with dexamethasone all resulted in large and statistically significant increases in pregnancy rates in combination with clomiphene in clomiphene-resistant anovulatory women, along with increased multiple gestation rates. These findings warrant further investigation, particularly if multiple gestation can be avoided.

E. Surgical procedures for inducing ovulation. One of the earliest treatments for PCOS was wedge resection of the ovary, which, while effective in inducing ovulation, had attendant surgical risks, as well as the risk of developing adhesions.⁹⁰ With the advent of laparoscopic surgical procedures, both short- and long-term risks are theoretically lower. Several studies have investigated the role of laparoscopic “drilling” of the ovary using electrocautery.

1. Identified studies. Identified studies are summarized in Table 11. The largest study, by Bayram and colleagues,⁹¹ compared a strategy of immediate gonadotropins to laparoscopic electrocautery, followed by ovulation induction agents only if pregnancy did not occur. The electrocautery strategy resulted in similar pregnancy and live birth rates (live birth RR 1.14; 95 percent CI 0.94–1.39) with significantly lower multiple gestation rates (RR 0.11; 0.01–0.88). In another study in a similar population, Palomba and colleagues found significantly higher pregnancy and live birth rates with the addition of metformin after laparoscopic cautery.⁹² None of the studies had sufficient followup to assess the risk of longer term complications such as adhesions or premature ovarian failure.

2. Other systematic reviews. We did not identify any non-Cochrane published reviews.

3. Cochrane reviews. The relevant Cochrane review⁹⁶ concluded that laparoscopic drilling, with or without stimulation, resulted in essentially equivalent pregnancy (OR 1.08; 95 percent CI 0.69–1.71) and live birth rates (OR 1.04; 0.59–1.85), with a significantly reduced risk of multiple gestation (OR 0.13; 0.03–0.52).

4. Conclusions. Use of laparoscopic cautery, followed by ovulation induction if necessary, results in similar pregnancy and live birth rates, with significantly lower multiple gestation rates, compared to immediate gonadotropin use in clomiphene-resistant women. The addition of metformin may result in further improvements in pregnancy and live birth rates. There are no data on the long-term sequelae of laparoscopic ovarian cautery.

F. Aspects of intrauterine insemination in anovulatory women. Intrauterine insemination (IUI) may be used as an adjunct to ovulation induction in women with PCOS, although we did not identify any recent randomized trials that directly compared ovulation induction with and without IUI.

1. Identified studies. We identified one study that addressed aspects of IUI in this population. Lewis and colleagues⁹⁷ compared two methods for the timing of IUI - one with home monitoring of urinary luteinizing hormone (LH), with IUI after detection of the LH surge, versus ultrasound monitoring of follicular development and triggered ovulation using hCG, followed by IUI. Pregnancy rates were increased with hCG triggering, but not significantly (RR 1.73; 95 percent CI 0.88–3.38).

2. Other systematic reviews. Kosmas and colleagues,⁹⁸ in a systematic review of timing of IUI based on LH monitoring versus hCG triggering, found non-significantly increased pregnancy rates with hCG triggering after clomiphene treatment in anovulatory patients (OR 2.00; 95 percent CI 0.84–4.77)

3. Cochrane reviews. There were no relevant Cochrane reviews.

4. Conclusions. Although the available studies suggest an increase in pregnancy rates with hCG triggering for IUI after ovulation induction with clomiphene in women with PCOS, sample sizes have been too small to demonstrate statistically significant differences. Given the large differences in cost, patient convenience, and the fairly high relative rates (1.7–2.0) observed between these two treatments, definitive determination of superiority should be a research priority.

V. Superovulation in Ovulatory Women

For couples where the female partner has normal ovulatory function and at least one patent fallopian tube, and the male partner has motile sperm, superovulation (use of gonadotropins to induce development of more than one follicle in a given cycle), followed by IUI, is the most efficient method of treatment, resulting in 2–3 times higher pregnancy and live birth rates within 6 months of treatment compared to IUI alone, intracervical insemination (ICI) alone, or superovulation with ICI.⁹⁹ However, this increased probability is associated with an increased risk of multiple gestations, which are at risk of multiple complications, including preterm birth and its sequelae; in the trial cited above, 16 percent of the live births in the two superovulation arms were preterm, compared to 6 percent of those in the other two arms (RR 2.60; 95 percent CI 0.79–8.61).

This section reviews publications subsequent to this study that address methods for superovulation, largely with IUI, as therapy in infertile couples where the female partner has normal ovulatory function and tubal patency, and where the male partner has motile sperm.

A. Drugs for superovulation-estrogen inhibitors. In theory, estrogen inhibitors should produce similar hypothalamic and pituitary responses in ovulatory women as they do in anovulatory women, leading to the development of multiple follicles and an increased probability of conception. Because estrogen inhibitors are oral agents with a lower risk of higher order multiples than the injectable gonadotropins, and cost significantly less, they are a potentially attractive candidate for superovulation. This section reviews the evidence on the efficacy of estrogen inhibitors and aromatase inhibitors compared to no treatment, to each other, and to gonadotropins.

1. Identified studies. Table 12 summarizes the identified studies. In general, significant differences were not observed in pregnancy rates for any comparison, with the exception of 2.5 mg versus 5.0 mg of letrozole, where the higher dose resulted in large and significant increase in pregnancy rate (RR 4.47; 95 percent CI 1.05–19.0). Although no differences were observed in rates of multiple pregnancy or OHSS, the number of these events in individual studies was small.

2. Other systematic reviews. We did not identify any non-Cochrane reviews.

3. Cochrane reviews. There are three relevant Cochrane reviews. The first,¹⁰⁶ most recently updated in November 2006, reviewed studies of clomiphene versus placebo or no treatment in couples with unexplained infertility; statistically significant differences were not observed, but the overall sample sizes were small, and there was a trend towards higher pregnancy rates when clomiphene was used with IUI (OR 2.40; 95 percent CI 0.70–8.19) or with hCG triggering (OR 1.66; 0.48–4.80). Multiple pregnancy rates were similar (OR 0.99; 0.14–7.12).

The second review,¹⁰⁷ updated in May 2002, compared clomiphene to gonadotropins. In three studies with a total of 200 subjects, clomiphene had a significantly lower pregnancy rate (OR 0.44; 95 percent CI 0.19–0.99) and a trend towards lower live births (OR 0.51; 0.18–1.47). There was also a trend towards fewer multiple gestations (OR 0.37; 0.06–2.43).

Finally, a review updated in January 2007 compared a variety of protocols for superovulation combined with IUI.¹⁰⁸ Compared to estrogen inhibitors, gonadotropins resulted in higher pregnancy rates (OR 1.76; 95 percent CI 1.16–2.66) based on seven studies, but there was no difference in live birth rates in the single study that allowed estimation of live birth rates (OR 0.94; 0.44–1.98). Both multiple pregnancy (OR 1.85; 0.53–6.44) and OHSS (OR 4.44; 0.48, 41.3) were more likely with gonadotropins, but, again, because of the relatively low number of these events, confidence intervals include 1.0. In five studies comparing aromatase inhibitors to clomiphene, there was no significant difference in pregnancy rates (OR 0.15; 95 percent CI 0.64–2.08).

4. Conclusions. The available literature does not allow any conclusions about the relative efficacy of different estrogen inhibitors, although 5 mg of letrozole appears to be superior to 2.5 mg. Pooled data show significantly higher pregnancy rates with gonadotropins compared to estrogen inhibitors, but data are too limited to draw conclusions about live birth rates. There is a trend towards higher rates of multiple pregnancies and OHSS with gonadotropins compared to estrogen inhibitors, but the number of events, even in pooled studies, prevents definite conclusions.

B. Drugs for superovulation - gonadotropins. Given the finding that superovulation with gonadotropins plus IUI results in the highest pregnancy rates along with higher multiple pregnancy rates, the obvious next step is to identify a protocol that optimizes the chances of a live birth while minimizing the multiple gestation risk. This section summarizes studies that address this issue.

1. Identified studies. Identified studies that met our inclusion criteria are summarized in Table 13. Individual studies show no significant difference between urinary and recombinant FSH, although fewer vials are used with rFSH, which may result in reduced treatment costs. Significant differences were not observed between lower and higher dose protocols, although hyper-response, a potential surrogate for OHSS, was higher. Pregnancy rates were consistently higher when GnRH antagonists were used in conjunction with gonadotropins in four studies (significantly in one¹⁰⁹), while twin rates were 4- to 5-fold higher in three of the four studies.

2. Other systematic reviews. We did not identify any non-Cochrane published reviews.

3. Cochrane reviews. Results of the relevant Cochrane review,¹⁰⁸ updated in January 2007, are summarized in Table 14. As has been seen with all of the study reviews, live birth is rarely reported and overall study numbers are small, with no consistent difference in pregnancy rates. Elevated pooled estimates for the risk of multiples and OHSS were observed with higher doses compared to lower doses (multiples 3.11; 95 percent CI 0.48–20.13; OHSS 5.52; 1.85–16.5), and with gonadotropins and GnRH agonists compared to gonadotropins alone (multiples 2.86; 95 percent CI 1.03–7.94; OHSS 2.02; 0.70–5.87). Pooled estimates of multiple pregnancy rates were not elevated with gonadotropins plus GnRH antagonists, but two of the studies noted above which did observe a significant increase in twins were published after this review.

4. Conclusions. There do not appear to be substantial differences in pregnancy rates between different gonadotropin preparations. Higher doses increase the risk of multiples and OHSS without significant improvement in pregnancy rates. The addition of GnRH antagonists to superovulation protocols may increase both pregnancy rates and twin gestation rates. Further studies adequately powered for the outcome of live birth per couple are needed.

C. Surgical adjuncts. Surgical procedures to address minor abnormalities detected during the infertility evaluation may result in improved outcomes for those couples who go on to superovulation and IUI.

1. Identified studies. We identified one study¹²⁰ that assessed the utility of diagnosis and treatment of minor abnormalities. Women who were candidates for superovulation and IUI who had small endometrial polyps (mean diameter 16 mm) detected on ultrasound were randomized to hysteroscopy with either biopsy (to rule out malignancy) or resection of the polyps. Polypectomy resulted in significantly higher pregnancy rates (RR 2.23; 95 percent CI 1.57–3.15); data on live birth rates were not presented. Time to pregnancy was substantially shorter in the polypectomy group; of note, 65 percent of the pregnancies in this group occurred before the first IUI.

2. Other systematic reviews. We did not identify any other published or relevant Cochrane reviews.

3. Conclusions. Hysteroscopic resection of ultrasound-detected endometrial polyps results in improved pregnancy rates for women undergoing superovulation and may even obviate the need for further treatment; this would likely result in a decrease in multiple pregnancy rates.

D. Aspects of intrauterine insemination after superovulation. Finally, we reviewed studies that addressed various aspects of IUI after superovulation.

1. Identified studies. We did not identify any studies that met our inclusion criteria.

2. Other systematic reviews. One published systematic review of hCG triggering of ovulation versus urinary LH monitoring for timing of IUI after clomiphene found no significant differences in pregnancy rates in couples with male factor infertility (OR 0.66; 95 percent CI 0.35–1.21) or unexplained fertility (OR 0.79; 0.38–1.64), although hCG triggering did significantly increase rates in anovulatory women, as noted above.

3. Cochrane reviews. In a review updated in July 2007,¹²¹ three studies published prior to 2000, with a total of 202 subjects, suggest a higher pregnancy rate with IUI compared to timed intercourse with superovulation, but confidence intervals cross 1.0 (OR 1.67; 95 percent CI 0.83–3.37). A review updated in July 2007 found no evidence for superiority of any semen preparation techniques, but the number of subjects was small.¹²² Finally, in a review updated in November 2002,¹²³ no differences were observed when comparing single versus double IUI (total number of subjects 355, OR 1.45; 95 percent CI 0.78–2.68).

4. Conclusions. There is insufficient evidence to identify any aspect of IUI that significantly affects pregnancy rates, let alone live birth rates or other less common outcomes.

I. Research Question

Among women of reproductive age, which laboratory, clinical, and other practice approaches result in the highest successful singleton pregnancy (or live-born) rates, and what practices lead to high multiple rates? Laboratory practices include intracytoplasmic sperm injection (ICSI), different types of embryo culture, fresh versus frozen embryo transfer, and day 2 to 3 versus day 5 to 6 transfer. Clinical practices include number of embryos transferred and selection criteria for eligible patients, as well as using the implantation rates from previous unsuccessful cycles to inform subsequent embryo transfer. Other practices include insurance coverage strategies.

II. Approach

Some infertile couples are either not candidates for the interventions described in the preceding section (because of tubal disease, for example) or have failed a trial of ovulation induction or superovulation. In all of the interventions described in the previous section, the ovaries are exposed to increased levels of endogenous or exogenous gonadotropins, and may or may not receive additional agents to trigger ovulation (the extrusion of the egg[s] from the ovary), but the individual steps of ovulation, exposure to sperm, fertilization, and initial development of the embryo all take place within the patient's body. The interventions described in this section involve direct intervention with at least one, and most commonly all, of these individual steps.

The review is organized around interventions applied to the individual steps in the process, based on the most commonly used protocols. Interventions are divided into those used in the female partner, in the male partner, and in the embryo.

For the female partner, interventions include:

• a)

  Suppression of endogenous pituitary gonadotropin secretion (pituitary down-regulation);

• b)

  Stimulation of follicular development with exogenous agents (controlled ovarian hyperstimulation);

• c)

  Triggering of ovulation;

• d)

  Retrieval of oocytes;

• e)

  Replacement of gametes (relevant only for gamete intrafallopian transfer [GIFT]);

• f)

  Transfer of the embryo;

• g)

  Luteal support;

• h)

  Other adjunctive therapies; and

• i)

  Strategies for prevention of ovarian hyperstimulation syndrome (OHSS).

For the male partner, interventions include:

• a)

  Methods for sperm retrieval; and

• b)

  Methods for sperm preparation.

For the embryo, interventions include:

• a)

  Methods for fertilization;

• b)

  Methods to support early embryonic growth;

• c)

  Methods for preparation for transfer;

• d)

  Methods for embryo storage for future transfers;

• e)

  Selection of embryos for transfer;

• f)

  Timing of embryo transfer;

• g)

  Number of embryos to transfer.

Our focus here is on interventions that can feasibly be evaluated using randomized trials; as mentioned in the Introduction, there was almost no literature on the male partner, so this section focuses on interventions focusing on the female partner and the embryo. The effect of broader interventions, such as insurance coverage for specific procedures, is more difficult to evaluate. Although there are some data on the effects of varying insurance policies on outcomes, the evaluation of the effectiveness of these policies involves completely different methods. The available data, and their implications for clinical care and policy, are discussed in the final chapter of this report.

Our general approach to study inclusion and summarization was similar to the one used for studies of ovulation induction and superovulation. As described in the Methods chapter, we excluded all non-randomized studies, as well as “quasi-randomized” studies (such as those where treatment assignment was based on alternate history numbers or clinic days). For this topic, the primary outcome of interest was the cumulative number of clinical pregnancies or, preferably, live births per couple; wherever possible, we used the number of women/couples randomized as the denominator. We excluded any study where these outcomes were not reported or calculable from the presented results.

For the primary outcomes, relative risks (RRs) with 95 percent CIs were calculated from the presented results. Because of substantial clinical heterogeneity in the studies in terms of patient characteristics (such as BMI in studies of PCOS) and treatment regimens, we did not perform formal meta-analyses.

Results for other outcomes, such as multiple pregnancy or spontaneous abortion rates, are summarized in the text. The majority of included studies were extremely limited in power to detect differences in the primary outcomes, let alone any differences in other less common outcomes. Outcomes related to later pregnancy and longer term maternal and child outcomes are discussed under Question 4.

III. Search Results

The flow of articles on this topic through the literature search and screening process is depicted in Figure 4.

IV. The Female Partner

Up to and including embryo transfer, the overall immediate short-term goal of each step in the IVF process is to maximize the probability of success at the next step, with the ultimate goal of maximizing the likelihood of a healthy live birth. This is usually achieved by maximizing the number of “units” available for the subsequent step. Thus, controlled ovarian hyperstimulation aims at maximizing the number of follicles suitable for oocyte retrieval, where as many eggs as possible are retrieved, after which as many embryos as possible are cultured. All other things being equal, increasing the number of embryos improves the likelihood that at least one will develop and progress to a live birth.

Unfortunately, this “maximization” strategy increases the risk of multiple pregnancies, as well as the risk of OHSS. As a rule, the ultimate goal for comparative trials of these steps is to identify interventions that maximize the chances of a healthy live birth while minimizing the risks of multiple pregnancy and complications such as OHSS.

A. Methods for pituitary down-regulation. In the normal menstrual cycle, ovulation is triggered by a surge of luteinizing hormone (LH) in response to feedback mechanisms involving ovarian hormones at the level of the hypothalamus and pituitary. Hyperstimulation of the ovaries with exogenous gonadotropins in women with a normal hypothalamic/pituitary/ovarian axis alters these feedback mechanisms and, potentially, the timing of the LH surge. Since the goal of hyperstimulation in the setting of IVF is to have as many eggs as possible to retrieve through the development of as many follicles as possible, a premature spontaneous LH surge may lead to ovulation prior to retrieval, forcing the cancellation of the entire IVF cycle.¹²⁴

Two general approaches have been used. The “classic” technique involves the use of a gonadotropin-releasing hormone (GnRH) agonist, given beginning 2 to 3 weeks before the IVF cycle. More recently, direct antagonists of the GnRH receptor, which do not require pre-treatment, have been introduced.

1. Included studies. We identified nine studies comparing different aspects of GnRH agonist administration that met our inclusion criteria (Table 15). In general, none of the comparisons of timing, dose, or type of agonist showed significant improvements in pregnancy or, when reported, live birth rates. The one exception was a comparison of a reduced dose of triptorelin compared to the standard dose, which showed significant improvement in both cycle-specific pregnancy rates and cumulative rates when using subsequent frozen embryo transfer.¹²⁵

We identified 14 studies directly comparing GnRH agonists and antagonists (Table 16). Pregnancy rates did not differ significantly in any of the individual studies, although none were adequately powered or designed as equivalency studies. In studies where relative OHSS rates were calculable, rates were consistently lower with antagonists, although this was statistically significant in only one.¹³⁴

We identified one other randomized trial comparing a GnRH long agonist protocol to a protocol of pre-treatment with oral contraceptives, clomiphene citrate plus rFSH, and rLH plus prednisolone in 194 subjects;¹⁴⁹ pregnancy rates were not significantly different (RR 1.20; 95% CI 0.86–1.67), and OHSS rates were lower with the clomiphene-based regimen (RR 0.23; 0.07–0.79). We did not find any additional studies evaluating this regimen.

Studies that compared different dosing, timing, or types of GnRH antagonists did not show significant differences in pregnancy rates (Table 17). However, three studies of pre-treatment with oral contraceptives (in order to allow scheduling of the beginning of the stimulation cycle) followed by an antagonist suggest, at best, no benefit and possibly worse outcomes with this regimen. Oral contraceptives followed by an antagonist had similar pregnancy rates compared with long protocol GnRH agonist in a small study of PCOS patients who had previously failed clomiphene,¹⁵⁰ and non-significantly lower rates in a larger trial (which excluded PCOS subjects).¹⁵¹ In the Rombauts study¹⁵¹ and two others comparing the addition of pre-treatment with OCPs to GnRH antagonists alone,^152,153 pregnancy rates were lower, significantly so in one.¹⁵²

We identified six studies in patients with either a history of a poor response to standard hyperstimulation protocols,¹⁵⁷–¹⁵⁹ a low likelihood of a good response based on age or basal FSH levels,^160,161 or endometriosis¹⁶²(Table 18). The five studies comparing antagonists to agonists did not show significant differences or a consistent pattern of one type of agent being superior to the other. In the one study comparing two GnRH agonist protocols, a short protocol was significantly inferior to a long protocol.

2. Other systematic reviews. We did not identify any relevant non-Cochrane reviews.

3. Cochrane reviews. There are three relevant Cochrane reviews, which are summarized in Table 19. The first, updated in September 2004, focuses on comparisons of a long-acting depot form of a GnRH agonist to daily administration.¹⁶³ No significant differences in pregnancy or live birth rate were found, although the gonadotropin requirement was lower with daily administration.

The second review¹²⁴ performed a meta-analysis of studies comparing GnRH agonists to antagonists. Pooled data showed a significant reduction in both pregnancy (OR 0.83; 95 percent CI 0.72–0.95) and live birth (OR 0.82; 0.68–0.97), multiple pregnancy rates were not significantly different (OR 0.82; 0.57–1.18). Antagonists significantly lowered the risk of severe OHSS (OR 0.61; 0.42–0.89), as well as the dosage and duration of gonadotropin required.

Finally, a review of interventions for poor responders¹⁶⁴ did not find sufficient evidence to draw conclusions about efficacy for any of the regimens reviewed.

4. Conclusions. Only a few of the studies we identified had adequate power to detect differences in pregnancy or live birth rates, let alone less common outcomes such as multiple pregnancy or OHSS. We did not identify clear evidence of the superiority of any specific protocol involving GnRH agonists. In the setting of endometrial preparation for frozen-thawed embryo transfer, two relatively large studies had conflicting results regarding the benefit of adding an agonist; further research is needed.

Although only one individual study comparing GnRH agonists to antagonists found a significant difference in pregnancy or live birth rates (in favor of agonists), formal meta-analysis shows a significantly lower pregnancy and live birth rate with the use of antagonists; antagonists do result in significant decreases in gonadotropin requirements, and a significant decrease in the risk of OHSS.

Pre-treatment with an oral contraceptive to assist with scheduling GnRH antagonist cycles resulted in decreases in pregnancy rates in all three identified studies; this reduction was statistically significant in one.

Finally, although there is no clear evidence for superiority of any strategy for improving outcomes in patients with a history of poor response, a long GnRH agonist protocol was superior to a short GnRH protocol in women over 40 in one trial.

B. Methods for ovarian stimulation. Once endogenous gonadotropin down-regulation has occurred, exogenous gonadotropins need to be administered in order to stimulate follicular development. A variety of preparations are available. The classic method uses human menopausal gonadotropin (hMG), which contains both LH and FSH; in addition to hMG, pure FSH, derived either from urine (uFSH) or as a recombinant form (rFSH), is also available. All three of these can stimulate follicular development alone. Because LH is part of normal follicular development in ovulating women, adding recombinant LH (rLH) to protocols using rFSH theoretically may improve outcomes.¹⁶⁵ In addition, some women do not produce multiple follicles (usually defined as three or more) in response to standard stimulation protocols and are classified as “poor responders;” women who are above age 35, or who have elevated levels of FSH early in a spontaneous cycle, are at increased risk of poor response.¹⁶⁴

1. Included studies. We identified 38 studies meeting inclusion criteria. Results are summarized in tables for comparisons of rFSH versus hMG, rFSH versus uFSH, and different rFSH preparations (Table 20); rFSH alone versus rFSH plus rLH (Table 21); various gonadotropin dosing regimens (Table 22); methods of administering gonadotropins (Table 23); and protocols for stimulation in poor responders (Table 24). Of all the studies, only two individual studies showed a significant improvement in pregnancy rates: individualized dosing protocol based on a nomogram was superior to a fixed dose regimen,¹⁶⁶ and a regimen of urinary FSH for 6 days followed by rFSH was superior to FSH alone.¹⁶⁷ Only one study¹⁶⁸ was explicitly designed as an equivalence trial. From both a statistical and regulatory perspective, demonstrating equivalence or non-inferiority requires specific a priori hypotheses about the degree of difference in efficacy, and in general requires a larger sample size than studies designed to demonstrate superiority.³⁶ This means that, in spite of a lack of demonstrable superiority of one preparation or another, it is not possible to conclude that the preparations are in fact equivalent in efficacy.

2. Other systematic reviews. We did not identify any relevant non-Cochrane reviews.

3. Cochrane reviews. There are two relevant Cochrane reviews^165,205(Table 25). In the review of hMG versus rFSH, last updated in August 2002,²⁰⁵ hMG was significantly superior to rFSH in terms of pregnancy rates (OR 1.28; 95 percent CI 1.11–1.54), and nearly so for live birth rates (OR 1.27; 0.98–1.64). hMG required significantly more medication, however, and the rate of multiple gestations was higher (OR 1.48; 0.98–2.16). In the review of rFSH versus rFSH plus rLH,¹⁶⁵ the addition of rLH to rFSH significantly increased live birth rates in previous poor responders (OR 1.85; 95 percent CI 1.10–3.11).

4. Conclusions. Trials of methods for ovarian stimulation in the setting of IVF, like those of methods for pituitary down-regulation, are consistently underpowered to detect differences in pregnancy rates or live birth rates, and few are specifically designed to demonstrate equivalence in these outcomes. Power to detect less common outcomes such as multiple pregnancy or OHSS is even lower. There is evidence from one trial that pregnancy rates are superior with an individualized dosing regimen of rFSH compared to fixed dosing. Pooled results of individual trials suggest that hMG is superior to rFSH in long protocol GnRH agonist regimens, with higher multiple pregnancy rates, and that the addition of rLH to rFSH improves live birth rates in poor responders.

C. Methods for follicular maturation. In a spontaneous ovulatory cycle, final maturation and rupture of the follicle, resulting in release of the ovum, is triggered by a surge in LH; this surge also promotes luteinization, resulting in production of the progesterone necessary for endometrial preparation for implantation and early placentation.²⁰⁶ In controlled hyperstimulation, although ovum release is not needed (or desirable), human chorionic gonadotropin (hCG), which has biological activity similar to LH, has traditionally been given to induce final maturation prior to oocyte retrieval. Recent developments that might theoretically improve outcomes are the development of recombinant hCG (rhCG), which would provide a purer, more consistent product than urinary LH (uLH), and recombinant LH (rLH), which, because of a shorter duration of action, might reduce the risk of OHSS. An alternative approach in patients treated with a short-acting GnRH antagonist could be induction of an endogenous LH surge through administration of a GnRH agonist.

1. Included studies. Studies meeting inclusion criteria are shown in Table 26. One study evaluated two different protocols for timing of administration of hCG.²⁰⁷ Under ultrasound monitoring beginning on day 6 of stimulation, subjects were randomized to administration of hCG as soon as at least three follicles had reached at least 17 mm in diameter, or 2 days after this point. Live birth rates were significantly lower in the late hCG group (RR 0.72; 95 percent CI 0.53–0.98); including biochemical pregnancies and miscarriages, early pregnancy loss was two-fold greater in the late hCG group.

Three studies randomizing women to urinary versus recombinant hCG showed no difference in pregnancy or live birth rates,²⁰⁸–²¹⁰ although minor adverse events, especially injection site reactions, were more common with urinary hCG. In the one study that included two different doses of rhCG, there was a trend towards an increased rate of OHSS at the higher dose (RR 2.93; 95 percent CI 0.75–11.4).²¹⁰

Two studies comparing uhCG to rLH did not demonstrate significant differences in pregnancy or live birth rate.^211,212

Finally, four studies compared hCG to a GnRH agonist in women receiving a GnRH antagonist for down-regulation.²¹³–²¹⁶ Three showed significantly decreased pregnancy rates with the use of the agonist, with significantly higher early loss rates. A fourth, conducted in women considered at high risk of OHSS because of PCOS or prior response to stimulation, showed no difference in pregnancy rates, but significantly lower OHSS rates; this study used a different GnRH agonist and included suppression with oral contraceptives and GnRH agonist before beginning GnRH antagonists.

2. Other systematic reviews. We did not identify any other non-Cochrane reviews.

3. Cochrane reviews. The relevant Cochrane review (Table 27),²⁰⁶ updated February 2005, quantitatively found no difference in pregnancy or live birth rates between uhCG or rHCG, with a significant decrease in any adverse event, particularly injection site reactions (OR 0.47; 95 percent CI 0.32–0.70). Similarly, there was no difference in pregnancy or live birth rates between uhCG and rLH; an unpublished trial showed that doses of rLH required to prevent OHSS led to decreased pregnancy rate, and further development of the product for this indication was halted.

4. Conclusions. Timing of hCG administration for follicular maturation is important for optimizing live birth rates - delays of 48 hours after one ultrasound threshold (at least three follicles of at least 17 mm) resulted in significant decreases in live births. The optimal time and threshold have not been determined. There does not appear to be any difference in pregnancy or live birth rates, or other major outcomes, between rhCG and uhCG, although injection site reactions are more common with uhCG. In cycles using a GnRH antagonist for pituitary down-regulation, use of hCG is superior to use of a GnRH agonist in most women, although agonists significantly lowered the risk of OHSS without affecting pregnancy rate in one trial of high-risk women.

D. Methods for oocyte retrieval. The current standard of care for oocyte retrieval is transvaginal aspiration under ultrasound guidance.

1. Included studies. We identified one trial of different techniques for retrieval in PCOS patients, and seven trials comparing different methods for analgesia (Table 28). Branigan and colleagues²¹⁷ compared a standard protocol, where only follicles with a diameter of at least 10 mm (those believed to have the greatest likelihood of a fertilizable ovum) were aspirated, to a “thorough” protocol, where any “possible” follicle, down to 4 mm, was aspirated, in women with PCOS scheduled for IVF; those women who did not conceive after IVF were followed. The “thorough” protocol resulted in a higher pregnancy rate (RR 15.1; 95 percent CI 0.91–250) subsequent to the IVF cycle. Results for the entire randomized group, which includes 31 women who conceived during the IVF cycle, were not presented. Cumulative pregnancy and live birth rates for both the IVF and non-IVF cycles would be preferable.

Choice of analgesia did not significantly affect pregnancy rates in any of the studies. In general, overall pain scores were similar between the interventions, although variations in the scales, as well as types and dosing of analgesic agents and doses used, prevent any between-study comparisons. In studies where one arm did not include some kind of sedation,^218,219 or used a lower level of sedation,²²⁰ peri-procedural pain was significantly higher, although this did not appear to have any impact on overall patient preferences.

2. Other systematic reviews. We did not identify any non-Cochrane reviews.

3. Cochrane reviews. The relevant Cochrane review²²⁵ found no difference in pregnancy rates. Intraoperative pain scores by visual analog scale were significantly higher with electroacupuncture compared to standard treatment, as well as with patient controlled sedation compared to physician controlled sedation.

4. Conclusions. Choice of analgesia for oocyte retrieval does not appear to affect pregnancy rates. Techniques involving some form of sedation result in lower intraoperative pain, but this does not appear to adversely affect overall patient perceptions and satisfaction.

E. Methods for endometrial preparation in frozen-thawed transfer. In the setting of transfer of frozen-thawed embryos from previous cycles, controlled ovarian hyperstimulation is obviously not necessary, but methods to improve preparation of the endometrium for implantation are frequently used. Since frozen embryo transfer from previous cycles is one potential way to maximize cumulative pregnancy rates while minimizing the risk of multiple gestations (see the section on the number of embryos transferred [section G under “The Embryo”], below), identifying the optimal method for preparation should be a high priority.

1. Included studies. Two studies compared the use of estrogen with and without a GnRH agonist (Table 29). Both were relatively large. In one,²²⁶ the GnRH agonist used did not significantly improve pregnancy rates; in the other,²²⁷ both pregnancy and live birth rates were significantly improved with the use of the agonist (RR for live birth 2.30; 95 percent CI 1.15–4.62). Both the type of agonist and the estrogen formulation used differed between the two studies. A third, smaller study²²⁸ compared regimens in women with unsuppressed cycles and found no difference in rates with oral estradiol followed by vaginal progesterone when endometrial thickness reached 7 mm compared with FSH on cycle days 6, 8, and 10 plus hCG to trigger ovulation.

2. Other systematic reviews. We did not identify any other systematic reviews.

3. Cochrane reviews. The most recent Cochrane review, published in January 2008,²²⁹ is summarized in Table 30. The effectiveness of no intervention (natural cycle) transfer compared to endometrial preparation was evaluated in only one small trial, with subsequent wide confidence intervals. There was insufficient evidence to draw conclusions about other regimens, although there was an overall trend to higher pregnancy rates with the addition of GnRH agonists to estradiol/progesterone.

4. Conclusions. There is insufficient evidence to determine the optimal method for endometrial preparation for frozen-thawed embryo transfer.

F. Methods for embryo transfer. Methods for fertilization, embryo culture, selection and timing of transfer are discussed below. In the majority of procedures in the United States, embryos are transferred back into the uterus using a thin transcervical catheter.

1. Included studies. Studies meeting inclusion criteria are shown in Table 31. Berkkanoglu and colleagues randomized patients to either standard transfer protocol or irrigation with embryo culture media.²³⁰ Although reported rates were similar for the two arms, a much larger number of randomized subjects were excluded from the flushing arm (48 vs. 12) in the analysis, a difference that seems unlikely to be random. When analyzed by intention-to-treat, pregnancy and live birth rates were significantly lower in the flushing group (live birth RR 0.67; 95 percent CI 0.47–0.95).

A Swedish study found no differences in pregnancy rates after ultrasound-guided transfer by a trained midwife or physician.²³¹

A study of prophylactic antibiotics found no difference in pregnancy rates, despite a significantly reduced rate of bacterial contamination of the catheter.²³²

Two studies of different catheter types detected no difference in pregnancy rates.^233,234 The third, comparing a catheter with a fixed metal obturator to a soft catheter where use of a metal obturator was optional, found significantly higher pregnancy rates with the soft catheter (RR 1.32; 95 percent CI 1.08–1.60).²³⁵

Timing of catheter withdrawal did not affect pregnancy rates.²³⁶

Three studies of embryo transfer media containing hyaluronic acid compared to standard media²³⁷–²³⁹ all showed improved pregnancy rates with media containing hyaluronic acid, with one²³⁷ showing significantly increased rates.

Ultrasound guidance of the transfer resulted in higher pregnancy rates in all but one of the studies identified (Table 32); this difference was significant in five of the eight studies. The one study which did not show any difference²⁴⁰ varied from the others in several ways. First, a single operator performed all of the procedures - an overall benefit of ultrasound guidance among multiple practitioners does not rule out the possibility of no difference for individuals. Second, there were two unplanned interim analyses involving the investigators rather than a separate statistical or data and safety monitoring board, a process which is somewhat unorthodox for clinical trials.

2. Other systematic reviews. We did not identify any other non-Cochrane reviews.

3. Cochrane reviews. The relevant Cochrane review, updated November 2006, concluded that ultrasound guidance significantly improved both pregnancy (OR 1.49; 95 percent CI 1.29–1.72) and live birth rates (OR 1.40; 1.18–1.66).²⁴⁸ Multiple pregnancy rates were increased, but not significantly (OR 1.26; 0.91–1.75) and ectopic rates non-significantly decreased (OR 0.64; 0.25–1.61).

4. Conclusions. Pre-transfer irrigation does not improve pregnancy or live birth rate, and, based on an intention-to-treat analysis of the one study identified, significantly reduces both rates. There is no evidence that type of provider changes outcomes. Although pre-treatment with antibiotics significantly lowers measurable bacterial contamination, this does not translate into improved pregnancy or live birth rates. Hyaluronic acid containing media may result in higher pregnancy rates compared to other media.

Ultrasound-guided embryo transfer consistently results in substantial improvements (40 percent relative increase) in pregnancy and live birth rates compared to various “clinical touch” methods. The consistency of this finding and the size of the effect are striking considering that the majority of interventions covered in this review do not show significant differences.

G. Methods for luteal support. Aspiration of follicular cells during oocyte retrieval and suppression of GnRH can inhibit luteinization, which is necessary for progesterone production. The use of exogenous progesterone significantly increases pregnancy rates compared to placebo or no treatment.²⁴⁹ This section reviews studies published since 2000 that evaluate different progesterone-based regimens; varying routes of administration and timings of these regimens; alternatives to progesterone; and adjunctive treatments.

1. Included studies. Nine studies evaluated different formulations of progesterone (Table 33). In two studies, one with 205 subjects²⁵⁰ and another with 734,²⁵¹ intramuscular progesterone resulted in higher pregnancy and live birth rates, with lower miscarriage rates in the larger study (RR 0.33; 95 percent CI 0.20,0.55), compared to vaginal progesterone. One study did not detect a significant difference between vaginal and oral progesterone.²⁵² The remaining studies compared various formulations for vaginal administration; none detected a significant difference in pregnancy rates.

Four studies evaluated hCG (Table 34). Compared to a standard GnRH agonist long protocol with no supplementation, hCG substantially increased pregnancy rates. This increase was not significant, probably due to the small sample size.²⁵⁹ In three studies comparing hCG to progesterone, there were no significant differences in pregnancy or live birth rates.^256,260–²⁶²

Four studies evaluated different regimens for the timing of beginning or ending progesterone supplementation (Table 35). None found a significant difference.

Finally, we reviewed studies of adjuncts to progesterone (Table 36). The addition of hCG on days 1, 4, and 7 after transfer significantly increased pregnancy rates (RR 2.31; 95 percent CI 1.06–5.03) in a subsequent cycle in poor responders.²⁶⁷ The addition of estrogens significantly increased pregnancy and live birth rates in GnRH agonist suppression protocols in two of three studies.^268,269 Finally, a single administration of GnRH agonist added to progesterone and estrogen support increased pregnancy rates in patients using either a GnRH agonist or antagonist suppression protocol; the increase was significant in the antagonist group (RR 1.41; 95 percent CI 1.04–1.91).

2. Other systematic reviews. We did not identify any other non-Cochrane reviews.

3. Cochrane reviews. The most recent Cochrane review was most recently updated in May 2004 (Table 37).²⁴⁹ Quantitative findings were largely similar to the qualitative findings described above. Intramuscular progesterone resulted in higher pregnancy and live birth rates compared to either oral or vaginal progesterone, although this was significant only for live births in the vaginal versus intramuscular group, likely because of the small number of subjects in the oral progesterone studies. Interestingly, multiple pregnancies were significantly increased with intramuscular compared to oral progesterone, even with the small sample size (OR 7.88; 95 percent CI 1.10–56.2), consistent with some implantation advantage. Significant differences were not detected between the different vaginal progesterone formulations.

hCG was significantly better than placebo in terms of live birth rates (OR 1.94; 95 percent CI 1.25–3.01) and miscarriages (OR 0.27; 0.11–0.61). Rates of multiple gestation (OR 2.77; 0.47–16.5) and moderate/severe OHSS (OR 11.17; 1.45– 86.2) were higher.

The addition of hCG to progesterone did not significantly increase pregnancy or live birth rates. In the two studies included in the meta-analysis, the addition of estrogen did not improve pregnancy or live birth rates; however, all three of the studies published subsequent to the Cochrane review do show improved rates.

4. Conclusions. Some form of luteal support is necessary with IVF, since both progesterone and hCG result in improved pregnancy rates compared to no treatment. Although there is no detectable difference between oral progesterone and the various formulations of vaginal progesterone, both result in lower pregnancy and live birth rates compared to intramuscular progesterone. The addition of estrogen to progesterone may improve outcomes, although additional larger studies are needed to confirm these findings. Finally, adding stimulation with a GnRH agonist to progesterone and estrogen in patients down-regulated with a GnRH antagonist improves live birth rates.

H. Other adjunct treatments. A variety of adjunctive treatments have been proposed to help improve pregnancy and live birth rates, decrease multiple pregnancy rates, or prevent complications related to IVF, in both first-line treatment and in patients who either have a worse prognosis or have failed previous therapy.

1. Included studies. We identified seven studies of medical therapy (Table 38). Two involved vasoactive agents^273,274 and did not detect any significant differences. Five other studies involved the use of aspirin, with or without a corticosteroid, or a non-steroidal anti-inflammatory drug (NSAID). Only one showed a significant effect: in a placebo-controlled trial, administration of the NSAID piroxicam 1 day prior to embryo transfer increased pregnancy rates by almost 70 percent (RR 1.69; 95 percent CI 1.14–2.50).²⁷⁵

Six studies evaluated non-medical adjuncts (Table 39). Cha and Wirth found a two-fold higher pregnancy rate in subjects randomized to receiving intercessory prayer, where strangers prayed specifically for success (RR 2.07; 95 percent CI 1.34–3.22).²⁸⁰ We did not identify any similar studies, and this particular one raised multiple methodological questions, including issues regarding informed consent. Three studies of acupuncture all showed improvement in pregnancy and/or live birth rates.²⁸¹–²⁸³ The three studies differed in the nature of the intervention, as well as the nature of the control - ranging from no acupuncture to acupuncture with a “sham” needle to active acupuncture of points thought to be unrelated to reproduction - making interpretation of the results difficult. Finally, a large Australian study found no differences in pregnancy rates between couples who were asked to abstain from intercourse around the time of embryo transfer and those who were encouraged to engage in intercourse at this time.²⁸⁴

Finally, several trials of treatments in patients with a lower probability of a successful pregnancy because of known co-conditions or previous ART failure showed benefit (Table 40). Treatment with nitroglycerin,²⁸⁶ heparin and aspirin,²⁸⁷ IV immunoglobulin,²⁸⁸ or letrozole²⁸⁹ did not improve pregnancy rates in women with previous poor ovarian response. However, in patients without previous endometrial imaging, hysteroscopy and treatment of any discovered pathology significantly improved both pregnancy and live birth rates compared to repeat treatment without hysteroscopy (RR for live birth 1.70; 95 percent CI 1.22–2.37).²⁹⁰

In women aged 40 or older, the addition of dexamethasone²⁹¹ or growth hormone²⁹² both significantly improved outcomes.

In women with PCOS, the addition of metformin reduced the incidence of OHSS and increased pregnancy and live birth rates.^293,294 Both studies were small (52 or fewer subjects/arm), but the differences were significant in the study by Tang and colleagues (RR for live birth 2.67; 95 percent CI 1.15–6.22; for OHSS, 0.48; 0.23, 0.98).²⁹³

In women with known endometriosis, pre-treatment with a GnRH agonist for 3–6 months prior to initiating an IVF cycle increased pregnancy rates three-fold, although both studies were too small to detect a significant difference.^295,296 The study by Rickes and colleagues²⁹⁵ is also notable as one of the few IVF studies where cumulative rates over several cycles were used as the endpoint. Laparoscopic removal of endometriomas detected prior to IVF did not improve pregnancy rates significantly.²⁹⁷

In patients with hydrosalpinges detected prior to IVF, laparoscopic occlusion or salpingectomy increased live birth rates five- to six-fold.²⁹⁸ The lower bound of the 95 percent CIs crossed 1.0 for both surgeries combined, but there were only 15 subjects in the no treatment arm, as opposed to 50 in each of the surgical arms. Ectopic pregnancy rates were not evaluable.

2. Other systematic reviews. We did not identify any other non-Cochrane reviews.

3. Cochrane reviews. There are five relevant Cochrane reviews on adjuncts for IVF (Table 41). Reviews of low-dose aspirin (7 studies with over 1200 subjects)³⁰¹ and glucocorticoids (13 studies with over 1700 subjects)³⁰² did not find significant treatment effects.

The review of growth hormone³⁰³ did not find an overall significant treatment effect (OR 1.18; 95 percent CI 0.41–3.37). However, three studies of growth hormone in poor responders published prior to 2000 with a total of 74 subjects had a significant improvement in live birth rates (OR 4.37; 95 percent CI 1.06–18.3). This is consistent with the study by Tesarik and colleagues,²⁹² which found a five-fold higher live birth rate with growth hormone in women over 40.

Prolonged pre-IVF down- regulation with a GnRH agonist significantly improved pregnancy and live birth rates (OR 9.19; 95 percent CI 1.08–78.2) in three studies with a total of 165 subjects.³⁰⁴

Surgical treatment of hydrosalpinges significantly improved pregnancy and live birth rates based on three pre-2000 studies with a total of 295 subjects (OR for live birth 2.13; 95 percent CI 1.24–3.65).³⁰⁵ This is consistent with the findings of Kontoravdis and colleagues described above.²⁹⁸

4. Conclusions. Based on the available evidence, vasoactive agents such as nitroglycerin, beta-agonists, or l-arginine do not improve pregnancy or live birth rates in either first-time or poor prognosis IVF patients. Low-dose aspirin does also not appear to have any effect. The NSAID piroxicam significantly improved pregnancy and live birth rates in a general IVF population, and further studies of NSAIDs are warranted. Randomized trials of intercessory prayer and acupuncture showed benefit, but there are remaining methodological questions which need to be addressed.

Dexamethasone and growth hormone both improved pregnancy and live births in women over 40 undergoing IVF; the growth hormone findings are consistent with earlier studies showing a benefit in poor responders. Metformin reduced the incidence of OHSS, and showed evidence of improvement in pregnancy and live birth rates, in women with PCOS undergoing IVF. Pre-treatment of women with endometriosis with a GnRH agonist for several months prior to IVF improves pregnancy and live birth rates, as do hysteroscopic removal of endometrial lesions and surgical removal or occlusion of hydrosalpinges.

I. Prevention of ovarian hyperstimulation syndrome.

1. Included studies. We identified two studies of interventions designed specifically as prophylaxis against OHSS (Table 42). Gokmen and colleagues³⁰⁶ found significant reductions in OHSS, with no difference in pregnancy rates, with the use of both hydroxyethyl starch and albumin. In contrast, in a much larger study, Bellver and colleagues³⁰⁷ found no differences, although the width of the confidence intervals cannot rule out benefit. This may represent differences in patient populations: the rate of OHSS in the no-treatment arm in the Gokmen study was 19.2 percent (16/83) compared to 6.9 percent (21/307) in the Bellver study. There are no other obvious sources for the differences - neither study used placebo or unblended assessment of the endpoints.

2. Other systematic reviews. We did not identify any other non-Cochrane reviews.

3. Cochrane reviews. There are three relevant Cochrane reviews. The first reviews the use of intravenous albumin³⁰⁸ and was most recently updated in December 2001. In five studies with a total of 378 subjects, the pooled OR for prevention of OHSS was significantly lower with albumin (OR 0.28; 95 percent CI 0.11–0.73), with no difference in pregnancy rates (OR 1.09; 0.65–1.83). The calculated number-needed-to-treat (NNT) to prevent one case of moderate to severe OHSS based on these estimates was 18. This may explain the difference between the previous studies and that of Bellver and colleagues: although the overall study was much larger, the rate was much smaller. The observed number of cases in the control group, 21, was close to the NNT, meaning that only one or two fewer cases would be expected to be observed in the albumin arm, a difference that would be very unlikely to be detectable.

Two other reviews addressed embryo freezing³⁰⁹ and coasting (withholding gonadotropins in patients judged to be at risk).³¹⁰ There was insufficient evidence to draw any conclusions (two studies of embryo freezing with 26 and 125 subjects that did not show differences, and one study of coasting with a sample size of 30).

4. Conclusions. In one large study published subsequent to the last Cochrane update, IV albumin was not effective in reducing the incidence of moderate to severe OHSS in patients at risk, in contrast to the pooled analysis in the Cochrane review. This difference may be due to the low event rate in the larger study, which resulted in an absolute number of events too small to detect the estimated effect of albumin. Another study with a larger absolute number of subjects would be needed to resolve the issue. Given that many of the interventions discussed above, such as GnRH antagonists, may reduce the risk of OHSS, this may be difficult to accomplish.