Multiple Pregnancy in ART Then & Now
In 2003, Jain and Hornstein reviewed trends in multiple pregnancy in the USA from 1980 to 2001 and demonstrated a steady increase in US twin rate. High order multiple birth rate (defined as the rate of triplets or more) had increased by over 400% between 1980 and 1998, although it then plateaued. In 2004, these authors continued their studies and demonstrated that some portion of this increase was associated with a simultaneous increase in ART procedures. Although these procedures and their concomitant live birth rates increased over this time period, the number of embryos transferred began declining in 1998 and this was associated with the plateau in high order multiples.
Multiple birth is well known to result in health problems for mother and babies. Maternal complications include increased rates of pre-eclampsia, gestational diabetes and preterm labor. Risks for the baby include low birth weight and prematurity as well as an increased risk of long-term disability and death. The risks are greater in high order multiples than they are in twin pregnancies. However, multiple birth is not a problem unique to ART. A substantial portion of multiple births are known to be related to non-ART treatments. These non-ART gonadotropin and, to a lesser extent, clomiphene simulation cycles introduce a greater risk of multiple birth, particularly high order multiple birth. Practitioners have less control of multiples in these procedures than they do in ART, where they can manipulate the embryo transfer number. These other treatments, however, are not the focus of this review.
The introduction of clomiphene citrate and human menopausal gonadotropin ovarian hyperstimulation to ART was one of the earliest and most effective protocol adjustments leading to increased live birth rates in ART. Stimulation of the ovaries allowed for retrieval of multiple eggs resulting in multiple embryos for transfer and this in turn dramatically increased procedure effectiveness such that use of stimulation was rapidly adopted. Studies demonstrated that pregnancy rates could be significantly increased when the number of embryos transferred increased, but also that this would result in increases in multiple pregnancy. Contributing to this problem were low implantation rates – defined as number of gestational sacs divided by the total number of embryos transferred. One early study reported an overall implantation rate of 8.0%, less than we would expect today in women aged 41–42 years.
In 2007, Stern et al., using data from the national Society for Assisted Reproductive Technology Clinic Outcome Reporting System (SART CORS) database, which collects data on over 90% of ART cycles in the USA, evaluated changes in the number of embryos transferred and associated multiple birth rates with respect to the guidelines promulgated by the American Society for Reproductive Medicine (ASRM) and SART. Similar to the study of Jain et al., this study found a decline over time in the number of embryos transferred and the associated high order multiple birth rates. Reductions were dramatic in women aged <35 years but only minor in women >40. Women aged 35–40 years showed trends between the two. The study also found that these reductions followed publication of increasingly strict guidelines by ASRM/SART. An additional finding was that clinics transferring fewer embryos in women <35 were also those with higher delivery rates. Clinics were divided into four groups (quartiles) based on mean number transferred in women aged <35 years. In 2003, for example, those in the group with the lowest transfer number (mean of 2.18 embryos) had live birth rates per cycle of 40% and those with highest transfer number (mean 3.37 embryos) had a birth rate of 35.5% suggesting that increases in live birth rate were associated with greater acceptance of reducing number transferred. Concomitantly, clinics with the higher delivery and lower number transferred had lowered multiple birth rates (triplet rates of 2.2 and 6.6%, respectively). Although high order multiple birth rates decreased over time, those of the higher performing clinics decreased most dramatically. This and other studies make clear that even when overall delivery rate is lower, an occasional transfer of many embryos leads to implantation of more than one embryo and thus to increases in multiple delivery.
As Jain et al. demonstrated, pregnancy and live birth rates in ART have risen steadily since the mid 1990s. As of 2011, the most recent data reported to the SART, the overall clinical pregnancy rate per cycle for women using their own eggs was 32.8% and that for women under 35 years of age was 40.1%. Implantation rate is critical. Implantation rate tells us the likelihood that any one embryo will implant. The overall implantation rate for all 2011 ART cycles using fresh embryos made from their own eggs reported by SART was 24.5% but this parameter is heavily influenced by age ranging from 36.0% in women aged <35 years to 4.0% in women aged >42 years. As the implantation rate decreases in each age group, the number transferred is increased to compensate for the decline.
Nationally reported live birth rate per cycle and mean number transferred, for women aged <35 years, as well as implantation rates, twin rate and high order multiple birth rate for reporting years 1995 through 2011, are shown in Table 1. As the implantation rate has increased, the live birth rate has been maintained or even increased despite a decline in the number transferred. More importantly, as number transferred declined for this age group in this time period, there has been more than sevenfold reduction in the high order multiple birth rate. Twin rate has not changed.
A less dramatic picture emerges for women over 40 years of age. This group has been a much more difficult one in which to reduce the number transferred since a high proportion of embryos in these women fail to continue development. While there was little change in number transferred in this group between 1996 and 2003,Table 2 shows that since 2003, the number transferred in the 41- to 42-year-old age group has declined slightly. Some progress is also being made in high order multiple birth rate as a result (Table 2). Women between 35 and 40 years of age trend between the <35 and >40 groups.
Cycles using donated eggs have historically had higher success rates than cycles using autologous eggs. Results of these cycles are reported nationally as live births per embryo transfer, rather than per cycle, and these are not differentiated in national data by age of the recipient. A look at these national data reveals the same pattern of increasing live birth rate with time (live birth per transfer: 50.9% in 2003 and 54.9% in 2011) and decreasing number transferred (mean embryos transferred: 2.6 in 2003 and 1.9 in 2011). Cycles using frozen embryos, have historically had lower success rates than those with fresh embryos although thaw cycles from donor eggs or embryos tend to result in higher pregnancy rates. Multiple birth rates are unfortunately not reported for either donor egg or thaw cycles.
In spite of progress on high order multiple pregnancies, twin rates continue to remain unchanged (Table 1 & Table 2). As the number of embryos has declined and implantation rates risen, the balance of one to the other has shifted the high order multiple birth rate, but not the twin rate. Gibbons et al. previously reviewed this situation that has many causes, some of which include patient preference and insurance, as will be discussed in later sections of this paper. Unfortunately, a high twin rate will likely persist until the number of embryos transferred is one.
Recent data have also suggested the possibility that transfer of more than one embryo could contribute to adverse outcomes even in singleton deliveries. In 2009, Luke et al. used national SART CORS data to describe when the number of fetal hearts on an early pregnancy ultrasound exceeded the number of delivered babies (singletons or twins), these 'vanishing twins' were associated with increased preterm birth and low birth weight in the remaining babies. Multiple fetal hearts arise from multiple transferred embryos. A subsequent analysis also suggested that transfer of more than one embryo, even where there was a singleton delivery and a single fetal heart on ultrasound, may itself result in fetal growth restriction. However, number transferred might also be influenced by embryo quality which could in turn have resulted in the observed fetal growth restriction and this possibility could not be ruled out in the study.
There has been a recent resurgence in interest in unstimulated and minimally stimulated ART cycles. Minimal stimulation allows for retrieval of fewer eggs at lower cost. There are also fewer embryos for transfer and transfer of a single embryo is more likely than in hyperstimulated cycles. Improved embryo culture conditions have made a return to these methods possible without completely sacrificing live birth rates, however, rates still tend to be much lower in unstimulated than in stimulated cycles.