Selection of patients suitable for one-embryo transfer may reduce the rate of multiple births by half without impairment of overall birth rates

A. Strandell1, C. Bergh and K. Lundin

Department of Obstetrics and Gynaecology, Sahlgrenska University Hospital, Göteborg, Sweden


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
The aim of the present study was to identify a subset of patients at high risk of multiple birth after IVF and hence suitable for one-embryo transfer, which undoubtedly would reduce the multiple birth rate. This retrospective study included 2107 IVF cycles in which two embryos were transferred. Factors with possible correlation to multiple birth were studied in a multivariate analysis. The factors included background data (female age, previous pregnancies and births, previous IVF cycles, indication for IVF) and IVF cycle characteristics. The following factors were independently predictive of multiple birth: female age expressed a negative correlation while number of good quality embryos transferred was positively correlated. A subset of patients was identified as being at high risk of multiple birth by including age, cycle number and presence of tubal infertility in a model derived from a logistic regression analysis. The rate of multiple births can be reduced from 26% to 13% of all births if one-embryo transfer is performed in selected cases. The total birth rate will decrease from 29% to 25% but may be completely restored by performing one additional one-embryo freeze transfer in high risk patients who do not achieve a term pregnancy.

Key words: in-vitro fertilization/multiple births/single embryo transfer


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Multiple births have increased dramatically over the past decade in many developed countries and most are associated with treatment of infertility. Multiple pregnancies, including twins, carry extra risks of obstetric and perinatal complications. There is an urgent need to develop strategies which decrease the number of multiple births after assisted reproductive techniques. The change in policy at Nordic IVF centres, transfer of two embryos instead of three, almost eliminated triplets while the twin rate remained fairly constant (20–25%). Results from the large Swedish registry study showed that the main adverse effects of IVF were associated with the high number of multiple births (Bergh et al., 1999Go). Accordingly, the Swedish National Board of Health and Welfare declared that a one-embryo transfer policy was preferable and singleton births should be the goal in IVF. An overall change to transfer of a single embryo would certainly mainly result in singletons, but pregnancy and birth rates would probably decrease substantially and hardly be accepted by the infertile patients and their physicians, both of whom are motivated to maximize the outcome in each cycle. Deliveries of singletons at term would be an appropriate definition of IVF success and an incentive to make efforts to reduce the rates of multiple births, as previously suggested (Martin and Welch, 1998Go).

An alternative to consistent one-embryo transfer would be an individualized embryo transfer policy, as previously discussed (Coetsier and Dhont, 1998Go). A theoretical model was developed in which multiple pregnancies were reduced from 28% to 15% and pregnancy rates from 30% to 26% if one-embryo transfer was applied in selected patients. A Finnish study has also demonstrated satisfactory pregnancy rates (29.7%) after one-embryo transfers in selected groups of patients (Vilska et al., 1999Go).

The aim of the present study was to identify cycles at increased risk of multiple births and to calculate the effects on the overall birth and multiple birth rates in routine cycles if a one-embryo transfer policy was applied to the high risk group.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Study population
A database of IVF cycles was utilized including the period 1995–1997 to identify IVF cycles reaching transfer. A total of 1441 women undergoing 2107 fresh cycles was included. Age limit for admittance to IVF treatment was 40 years. Conventional IVF was performed in 1024 cycles (49%) and intracytoplasmic injection (ICSI) in 1083 cycles. All cycles were performed using a long protocol including down-regulation with a gonadotrophin-releasing hormone analogue and ovarian stimulation with gonadotrophins [highly purified or recombinant follicle stimulating hormone (FSH)]. No frozen–thawed cycles were included. During the entire study period, two embryos were routinely transferred, which allowed us to study the impact of the different variables without any influence from the number of embryos transferred. Exceptions to this routine, transfers of one or three embryos, were uncommon (n = 115, n = 183 respectively) and represented a poor prognosis group, since one-embryo transfers were performed when only one embryo was available and three embryos were transferred occasionally after several previous cycles had failed. These cycles were not included in the analysis.

The mean maternal age was 32.7 years (SD 3.7, range 22–40). The primary indications for IVF treatment were tubal disease (29%), endometriosis (9%), male factors (33%), unexplained infertility (16%), hormonal factors (2%) and mixed causes (11%). During the study period, 61.4% of the women underwent one cycle, 31.8% underwent two, 6.4% underwent three and 0.4% underwent four stimulated transfer cycles, although our IVF programme offers routinely only three stimulated cycles. Among the 2107 cycles, first cycles constituted 49.3%, second cycles 32.0%, third cycles 13.6% and fourth or higher rank cycles 2.0%.

Definitions
Embryo quality was assessed by considering the number of blastomeres, day of transfer and degree of fragmentation. Embryos with a normal cleavage rate (3–4 cells on day 2 and 6–8 cells on day 3) and <20% fragmentation were defined as good quality embryos. Embryos with >50% fragmentation and/or cleavage arrest were not transferred.

Birth and multiple birth were analysed as primary outcomes. Implantation was defined as an implanted embryo resulting in birth. Calculations of birth rates after theoretical one-embryo transfers were made as if the same implantation rates were present regardless of the number of embryos transferred.

Embryo reduction of triplets to single pregnancies was performed in two cases. Their data were recorded as if they had a triplet pregnancy reaching delivery. No more triplet pregnancies occurred.

Analysed variables
The factors for inclusion were chosen because they were either known to be an important predictor of outcome or there was a theoretical rationale for the factor to be predictive (Templeton et al., 1996Go; Croucher et al., 1998Go). The following factors were analysed as possible predictors of birth and multiple birth: the woman's age, previous pregnancy and previous childbirth (spontaneous and as a result of previous IVF), previous IVF attempts, the indication for IVF, FSH dosage and duration, number of retrieved oocytes, fertilization method (conventional IVF or ICSI), number of fertilized oocytes, proportion of fertilized oocytes (number of fertilized oocytes divided by the number of retrieved oocytes), number of good quality embryos available for transfer, day of transfer (day 2 or 3 after oocyte retrieval), number of good quality embryos transferred and number of embryos suitable for freezing.

Statistical analysis
Each of the above-listed variables was included in a primary logistic regression analysis, with adjustment only for number of previous IVF cycles, with the dependent variables birth and multiple birth. A P value < 0.10 was chosen for inclusion of predictors in the secondary analysis. For both of the dependent variables, those predictors were included in a multivariate forward step-wise logistic regression analysis. Adjustment of the variance of the ß-coefficients was made, in order to allow for each individual to contribute with more than one cycle. No bias of the ß-coefficients is introduced when a patient is included more than once in the logistic regression analysis and the order of the IVF cycle is used as a variable. However, the dependence will imply an underestimation of the variance of the ß-coefficients and thus influence the P values and the confidence intervals. An adjusted variance was achieved by multiplying the variance with n2/n1, where n1 denotes the number of patients and n2 the number of cycles. This conservative method was applied for calculation of P values and confidence intervals. A P value < 0.05 was considered to be significant.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
The clinical pregnancy rate was 38.3% and the rate of childbirth was 29.0%. Multiple births occurred in 25.8% (158/612) of all births and in 7.5% (158/2107) of all cycles.

Analyses of correlation between the predictors and the outcome variables
The results of the correlation analysis are shown in Table IGo. Each variable is adjusted for number of previous IVF cycles. Age, tubal infertility, FSH dosage expressed as negative correlation and number of good quality embryos transferred showed a positive correlation to both birth and multiple birth. In addition, number of retrieved oocytes, number of fertilized oocytes, number of good quality embryos available for transfer and number suitable for freezing were all positively correlated with birth.


View this table:
[in this window]
[in a new window]
 
Table I. Factors analysed for possible correlation to birth and multiple birth, adjusted for number of previous IVF cycles
 
Multivariate analyses of predictors for the outcome variables
Three variables were related to the need for FSH and were positively correlated with age. Among these variables, the total dose of FSH showed the highest predictive capacity, but it was omitted from the final model, which resulted in female age being included in both models; age was considered to be a more practical clinical variable.

The number of good quality embryos transferred and female age were independently predictive of both birth and multiple birth. In addition, tubal indication and number of previous IVF cycles were negatively predictive of birth. These results are further presented in Table IIGo as odds ratios and 95% confidence intervals for each level of variables that were independently predictive of birth and multiple birth.


View this table:
[in this window]
[in a new window]
 
Table II. Variables predictive of birth and multiple birth, adjusted for all other variables
 
Implications of the multivariate birth models
To illustrate the probability of birth and the risk of multiple birth, a fertility index based on the independently predictive variables was calculated by means of the logistic regression equation. The distribution of the actual births and multiple births in the different pentiles of fertility index for multiple birth is shown in Figure 1Go, as is the fitness of the predictive model. The equations for calculation of the fertility indices (birth and multiple birth) are shown in Figure 2Go.



View larger version (58K):
[in this window]
[in a new window]
 
Figure 1. Distribution of childbirth in pentiles of 2107 IVF cycles according to fertility index for multiple birth. Each bar represents the observed distribution of single and multiple births in one fifth of the study population, the further to the right side the higher the fertility index and increased risk of multiple birth. The probability of multiple birth among cycles within each pentile is calculated and presented as the median value within that subpopulation by *.

 


View larger version (15K):
[in this window]
[in a new window]
 
Figure 2. Equations for calculations of fertility indices by means of logistic regression.

 
The probability of birth and multiple birth after transfer of one or two good quality embryos is shown in Figure 3Go. The curved slope of the multiple birth probability, compared with the linear shape of the birth probability slope, illustrates the benefit of identifying patients at high risk of multiple birth at younger ages. To identify a subgroup of patients suitable for one-embryo transfer, we decided upon goals for reduction of multiple births and limits for impaired success rates. Our aim was to reduce the multiple birth rate by half, from 26% to 13% of all births. The equation of fertility index to identify patients at high risk of multiple birth resulted in ages <33, 32 and 31 years for the first, second and third cycle respectively, assuming at least two good quality embryos were available (Table IIIGo). Tubal indication was considered to be a clinically important negative predictor for birth and was included in the model, although statistical significance was not quite achieved (P = 0.06). This decision was supported by a calculation excluding the variable tubal factor from the model, which implied that a larger number of patients undergoing one-embryo transfer was necessary to reach the goal of reducing the multiple birth rate by half, than if tubal factor as a predictor was included. The age limits were 29, 28 and 27 years if a tubal factor was present. If one-embryo transfer was applied to the high risk group defined by the multiple birth model (37% of cycles), the birth rate was calculated to decrease from 29% to 25%. If we instead aimed at reducing the multiple birth rate to 10% and applied one-embryo transfer to the high risk group (50% of cycles), the birth rate would not have decreased more than to 23%. A third scenario including a 7% multiple birth rate is also presented in Table IIIGo.



View larger version (23K):
[in this window]
[in a new window]
 
Figure 3. Probability of birth and multiple birth in relation to female age, number of previous transfer cycles and presence of tubal infertility if one (lower panel) or two (upper panel) good quality embryos are transferred to a specified group of patients with high risk for multiple birth. The probability estimations refer to the total population including patients without high risk for multiple births, receiving two embryos.

 

View this table:
[in this window]
[in a new window]
 
Table III. Age limits for possible one-embryo transfer if at least two embryos are available. The calculations are based on the model using `multiple birth' as outcome
 
The same type of calculation can be applied using the birth model including a condition of lowest acceptable birth rate. Both models yielded similar high risk groups although the multiple birth model identified, as expected, a few additional multiple births at the same birth rate.


    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Multiple births cause both medical and economic problems and pregnancies derived from assisted reproduction treatment constitute the vast majority of these cases. There is an urgent need to develop strategies which reduce the rate of multiple births. The present study shows that the variables affecting multiple birth also influence the total birth rate. It seems possible, however, to identify a subset of patients at significantly higher risk of multiple birth and apply one-embryo transfer without impairing total birth rates considerably. Previous studies analysing variables predictive of multiple birth have focused on reduction from three to two transferred embryos (Nijs et al., 1993Go; Staessen et al., 1993Go; Bassil et al., 1997Go; Murdoch, 1997Go; Roest et al., 1997Go). In a large study using the national British IVF database, it was demonstrated that only the multiple birth rate and not total birth rate increased when three instead of two embryos were transferred, if more than four eggs had been fertilized (Templeton and Morris, 1998Go). A recent study focusing on embryo quality demonstrated a significantly higher twin rate if both embryos transferred, as opposed to one of two, were of top quality (57 versus 21%) (Van Royen et al., 1999Go). The authors suggested in a subsequent randomized trial, single-embryo transfer to women younger than 34 years with at least one top quality embryo available in the first cycle. Other studies have also demonstrated the impact of embryo quality on the risk of multiple pregnancy rates, although two was the minimum number of embryos transferred and one-embryo transfer was not considered (Hu et al., 1998Go; Minaretzis et al., 1998Go).

The present study analysed predictors of both birth and multiple birth, with the intention of identifying patients suitable for one-embryo transfer. The study population has the advantage of being very homogeneous concerning stimulation protocols and transfer policies, since all cycles were performed at the same centre.

We identified age and quality of embryos as the two most important predictive variables of birth and multiple birth. The odds ratios for all variables were generally close to 1.0, indicating that the impact of each variable was very low, demonstrating the difficulties in finding individual variables of substantial importance. The very close connection between factors influencing birth and multiple birth was demonstrated by the possibility of using both models, predicting birth and multiple birth with similar results.

One reason to analyse multiple pregnancy in addition to multiple birth is that a multiple pregnancy can be considered for fetal reduction. In a study with a similar design and size as ours, although with a higher rate of multiple gestation, the predictive variables for multiple gestation were identified as: female age, cycle number, number of embryos available for transfer and number of transferred embryos of good quality (Coetsier and Dhont, 1997Go).

Tubal infertility as a negative predictor for multiple birth did not reach statistical significance, but was considered to be a clinically important factor for predicting birth. Well known factors, such as the presence of hydrosalpinges (Strandell et al., 1999Go), may contribute to the negative influence of tubal infertility, but also severe tubal damage, regardless of hydrosalpinges, is associated with poor IVF outcome (Csemiscky et al., 1996). Patients with hydrosalpinges accounted for <10% of tubal infertility in the present study. The model was improved by the inclusion of the variable tubal factor, thus allowing for lower age limits in the recommendation of one-embryo transfer to tubal infertility patients.

It has to be emphasized that our calculations on birth rates after one-embryo transfers are theoretical and based on the assumption that implantation rates will be the same whether one or two embryos are transferred. However, only implantations resulting in a birth are included and since a one-embryo transfer always includes the highest quality embryo, birth rates are likely to be even higher than calculated. In the near future, when one-embryo transfers will be performed more often, we will have unique opportunities to gain knowledge about embryo characteristics and their importance to implantation.

Our initial aim is to reduce the multiple birth rate by half. This scenario includes a reduction of the total birth rate from 29% to 25% if no additional cycles are performed. It should be possible to reach the original birth rate by performing additional frozen–thawed cycles. Based on the data in Table IIIGo and an estimated implantation (resulting in birth) rate of 15% at frozen–thawed one-embryo transfer, one freeze cycle has to be performed in three out of four patients in the high risk group (269 births if two embryo transfers – 179 births if one embryo transfers = 90 missing births, 90/0.15 = 600 needed freeze cycles, 600/786 women in the high risk group = 76%). Assuming a 50% survival rate of frozen–thawed embryos, there is a minimum demand for at least three good quality embryos from the stimulated cycle. Consequently, we should be able to reach the same birth rate by transferring one embryo in a fresh cycle to the identified high risk group and one embryo in a freeze cycle to the three quarters of the high risk group that did not have a term pregnancy from the fresh transfer. A prerequisite, though, is a successful high quality freezing programme. This scenario would increase the work load and the initial cost at the IVF centre, but undoubtedly decrease the rate of multiple birth substantially and consequently reduce the costs of neonatal and handicap care for society.

In a similar theoretical analysis, costs of sick leave and hospitalization during pregnancy, delivery, neonatal intensive care and handicap care were considered after transfer of one or two embryos. Although more treatments were needed to achieve the same take-home baby rate, the one-embryo transfer policy was more cost-efficient (Wølner-Hanssen and Rydström, 1998Go).

Based on the results of the present study, a prospective randomized multicentre study in Sweden has been initiated. It should provide us with guidelines and support us in our effort to reduce the rate of multiple births.


    Acknowledgments
 
We would like to thank Nils-Gunnar Pehrsson for valuable statistical advice and for assistance in performing the analyses. Bo Stawreberg is thanked for his help with data extraction. The study was supported by grants from the Regional Health Care Authority of Western Sweden.


    Notes
 
1 To whom correspondence should be addressed at: Department of Obstetrics and Gynaecology, Sahlgrenska University Hospital,S-413 45 Göteborg, Sweden.E-mail: annika.strandell{at}medfak.gu.se Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Bassil, S., Wyns, C., Toussant-Demylle, D. et al. (1997) Predictive factors for multiple pregnancy in in vitro fertilization. J. Reprod. Med., 42, 761–766.[ISI][Medline]

Bergh, T., Ericson, A., Hillensjö, T. et al. (1999) Deliveries and children born after in-vitro fertilisation in Sweden 1982–95 – a retrospective cohort study. Lancet, 354, 1579–1585.[ISI][Medline]

Coetsier, T. and Dhont, M. (1997) Multiple pregnancy rates in IVF: three embryos is too much for good prognosis. Am. J. Obstet. Gynecol., 178, 1368–1389.

Coetsier, T. and Dhont, M. (1998) Avoiding multiple pregnancies in in-vitro fertilization: who's afraid of single embryo transfer? Hum. Reprod., 13, 2663–2664.[Free Full Text]

Croucher, C.A., Lass, A., Margara, R. and Winston, R.M. (1998) Predictive value of the results of a first in-vitro fertilization cycle on the outcome of subsequent cycles. Hum. Reprod., 13, 403–408.[ISI][Medline]

Csemiczky, G., Landgren, B.M., Fried, G. et al. (1996) High tubal damage grade associated with low pregnancy rate in women undergoing in-vitro fertilization treatment. Hum. Reprod., 11, 2438–2440.[Abstract]

Hu, Y., Maxson, W.S., Hoffman, D.I. et al. (1998) Maximizing pregnancy rates and limiting higher-order multiple conceptions by determining the optimal number of embryos to transfer based on quality. Fertil. Steril., 69, 650–657.[ISI][Medline]

Martin, P.M. and Welch, H.G. (1998) Probabilities for singleton and multiple pregnancies after in vitro fertilization. Fertil. Steril., 70, 478–481.[ISI][Medline]

Minaretzis, D., Harris, D., Alper, M.A. et al. (1998) Multivariate analysis of factors predictive of successful live births in in vitro fertilization (IVF) suggests strategies to improve IVF outcome. J. Assist. Reprod. Genet., 15, 365–371.[ISI][Medline]

Murdoch, A. (1997) Triplets and embryo transfer policy. Hum. Reprod., 12, 88–92.[Abstract]

Nijs, M., Geerts, L., van Roosendaal, E. et al. (1993) Prevention of multiple pregnancies in an in vitro fertilization program. Fertil. Steril., 59, 1245–1250.[ISI][Medline]

Roest, J., van Heusden, A.M., Verhoeff, A. et al. (1997) A triplet pregnancy after in vitro fertilization is a procedure-related complication that should be prevented by replacement of two embryos only. Fertil. Steril., 68, 746–748.[Medline]

Staessen, C., Janssenswillen, C., Van den Abbeel, E. et al. (1993) Avoidance of triplet pregnancies by elective transfer of two good quality embryos. Hum. Reprod., 8, 1650–1653.[Abstract]

Strandell, A., Lindhard, A., Waldenström, U. et al. (1999) Hydrosalpinx and IVF outcome: a prospective, randomized multicentre trial in Scandinavia on salpingectomy prior to IVF. Hum. Reprod., 14, 2762–2769.[Abstract/Free Full Text]

Templeton, A. and Morris, J.K. (1998) Reducing the risk of multiple birth by transfer of two embryos after in vitro fertilization. N. Engl. J. Med., 339, 573–577.[Abstract/Free Full Text]

Templeton, A., Morris, J.K. and Parslow, W. (1996) Factors that affect outcome of in-vitro fertilisation. Lancet, 348, 1402–1406.[ISI][Medline]

Van Royen, E., Mangelschots, K., De Neubourg, D. et al. (1999) Characterization of a top quality embryo, a step towards single-embryo transfer. Hum. Reprod., 14, 2345–2349.[Abstract/Free Full Text]

Vilska, S., Tiitinen, A. and Hydén-Granskog, C. (1999) Elective transfer of one embryo results in an acceptable pregnancy rate and eliminates the risk of multiple births. Hum. Reprod., 14, 2392–2395.[Abstract/Free Full Text]

Wølner-Hanssen, P. and Rydström, H. (1998) Cost-effectiveness analysis of in-vitro fertilization: estimated cost per successful pregnancy after transfer of one or two embryos. Hum. Reprod., 13, 88–94.[Abstract]

Submitted on June 18, 2000; accepted on September 12, 2000.