How many embryos to transfer? Swedish researchers develop new method to avoid twins while maintaining high live birth rates

Stockholm, Sweden: Swedish researchers have, for the first time, developed a reliable way of deciding whether one or two embryos should be transferred during fertility treatment; the method simultaneously maintains a high chance of women giving birth to a live baby, while reducing the risk of twins.

Dr Jan Holte told the annual meeting of the European Society for Human Reproduction and Embryology, that if this model were to be applied in all fertility clinics, it had the potential to reduce the twin rates to the level of just under two percent seen in the normal population. Already, another four clinics have started to use the model.

Sweden leads the world in attempts to reduce multiple pregnancies by transferring only one embryo to a woman’s womb during fertility treatment whenever possible. In 2007 (the most recent year for which data are available*), 69.9% of embryo transfers were single embryos, 30.1% were double embryo transfers, and only 5.3% of deliveries after fertility treatment were multiple deliveries – the lowest multiple delivery rate in the world.

“However, until now, evidence-based strategies have been lacking for guidance on when to perform single embryo transfer in order to achieve the best possible balance between maintaining a high level of live birth rates, but reducing twin implantation rates,” said Dr Holte, who is medical director, a senior consultant and responsible for research at the Carl Von Linnéklinikken, Uppsala Science Park (Uppsala, Sweden).

Over a four-year period between 1999-2002 Dr Holte and his colleagues analysed a series of 3223 embryo transfers and recorded 80 different factors that played a role in the success of fertility treatment. They found that four variables were significant in predicting pregnancy outcome: the quality of the embryo, the age of the woman, ovarian responsiveness (the number of eggs retrieved in relation to the dose of ovarian stimulating hormones), and information about whether the woman had had previous IVF attempts with either fresh or frozen-thawed embryos, how many, and whether or not they resulted in a pregnancy.

Using these four variables, they constructed a mathematical model that predicted the chances of pregnancy after the transfer of one or two embryos and of the risk of twins. Over a subsequent four-year period between 2004-2007, they applied the model in the clinic for 3410 embryo transfers. They aimed to achieve a risk of twins of no more than 15% and any women that had a higher risk had only one embryo transferred in one cycle. Transfers of embryos that had been frozen, stored and then thawed before implantation followed the same model.

During this period the proportion of single embryo transfers increased to 76.2% (compared to 11.1% in the previous period between 1999-2002), and the rate of twin deliveries was reduced from 26.1% to 1.9%. Live birth rates per fresh embryo transfer fell from 29.1% to 24.6%, but when transfers of frozen-thawed embryos were included the live birth rate was similar during the two periods: 31.1% in the earlier period and 30.7% in the later period.

Dr Holte also made adjustments to take account of the fact that women in the later period tended to be older and have a less favourable prognosis than those in the earlier period and, therefore, would have a lower predicted live birth rate. Once he had done this, the live birth rate (including frozen-thawed embryos) in the later period was higher at 36% versus 31.1% in the earlier period.

Just as significant as the dramatic drop in twin births were the outcomes for babies born between 2004-2007. Average birth weights increased from 3086g to 3412g; the frequencies of babies born prematurely (before week 33) and babies with birth weights below 2500g were reduced by two-thirds; the frequency of babies born small for gestational age was reduced by 26%; and deaths either just before, during or just after birth were reduced by 58%.

Dr Holte said: “These improved outcomes were entirely due to the lower rate of twins. There was no significant differences in outcomes between the two periods when comparing only babies born as a result of a single pregnancy.”

The researchers found that their model correctly predicted the pregnancy rates that occurred in all women, regardless of their chances of becoming pregnant. Dr Holte explained: “The predicted chance of pregnancy for an individual couple ranged between around 5% up to around 60% per attempt. When the treatments were grouped according to their predicted chances into ‘stratas’ or groups of 0-10%, 10-20%, 20-30%, 30-40%, 40-50% and >50%, the corresponding observed clinical pregnancy rates were shown to fit very accurately with the predicted results.

“To our knowledge, this is the first time that a model has been developed that successfully predicts pregnancy and the risk of twin implantation during fertility treatment. The results suggest that application of the model may reduce twin rates to the desired level, in our case to that of the normal Swedish population, while totally preserving pregnancy rates and markedly reducing risks for the offspring.”

Multiple pregnancies, where a woman becomes pregnant with two or more embryos, carry a high risk to both mothers and babies; complications can include miscarriage, premature birth, low birth weight, cerebral palsy and death. For this reason, fertility doctors increasingly try to find ways of avoiding them without jeopardising women’s chances of becoming pregnant.

The model has been validated by another, independent Swedish clinic, with similarly good results over a five-year period, and now three other clinics (two in Sweden and one in Italy) have started to use it. Dr Holte and his colleagues are continuing to refine and test the model further.


Abstract no: O-157, Tuesday 16.00 hrs CEST (Hall A1)

* “Assisted reproductive technology in Europe, 2007: results generated from European registers by ESHRE”, by J. de Mouzon et al. Human Reproduction journal, in press.