- Email: [email protected]
Factors predicting IVF treatment outcome: a multivariate analysis of 5310 cycles
RBMOnline - Vol 10. No 5. 2005 645–649 Reproductive BioMedicine Online; www.rbmonline.com/Article/1518 on web 18 March 2005
Article Factors predicting IVF treatment outcome: a multivariate analysis of 5310 cycles Shai E Elizur is a tutor in Obstetrics and Gynaecology at the Sackler Faculty of Medicine, Tel Aviv University, Israel. Born in 1965, he obtained his MD magna cum laude in 1997 as a graduate of the Sackler Faculty of Medicine in Tel Aviv. He received his qualification in Obstetrics and Gynaecology in 2004. Since then he is working in the IVF unit at Sheba Medical Centre, Tel Hashomer, Israel. Current clinical research interests include fertility preservation in women with cancer.
Dr Shai E Elizur Shai E Elizur1,3, Liat Lerner-Geva2, Jacob Levron1, Adrian Shulman1, David Bider1, Jehoshua Dor1 Unit, Department of Obstetrics and Gynecology, Chaim Sheba Medical Centre, Tel-Hashomer 52621; 2Women and Children’s Health Research Unit, Gertner Institute for Epidemiology and Health Policy Research, Tel-Hashomer; both affiliated to the Sackler School of Medicine, Tel-Aviv University, Tel-Aviv, Israel 3Correspondence: Tel: +972 3 5302882; Fax: +972 3 5341589; e-mail: [email protected]
1IVF
Abstract The objective of this study was to analyse factors predicting live birth rate following IVF. A computerized database of 1928 women who underwent 5310 consecutive IVF cycles in a single IVF unit was evaluated. Data on the women’s age, number of retrieved oocytes, performance of intracytoplasmic sperm injection (ICSI), aetiology of infertility, number of transferred embryos and option of choosing embryos for transfer were evaluated. There were 1126 pregnancies that resulted in 689 live births. Transferring two embryos doubled the chances of delivery compared with one embryo, but transferring three embryos was not significantly superior to two embryos. Moreover, following a three-embryo transfer, the multiple delivery rates were significantly higher (P < 0.01) compared with transferring two embryos. Optimal delivery rates were observed in women aged 26–30 years, with gradual decline with advanced age. The performance of ICSI resulted in higher delivery rates compared with conventional insemination. According to these data, the best live birth results following IVF treatment were achieved when the maternal age was 26–30 years, in couples with male factor infertility undergoing ICSI, and when two embryos were transferred. Keywords: delivery rate, ICSI, IVF
Introduction The introduction of IVF and embryo transfer in the late 1970s revolutionized the treatment of infertility and spawned an entire new field of assisted reproduction technology. Traditional IVF was the optimal solution to infertility due to female mechanical factors. The next major breakthrough in treating infertility came with the introduction of intracytoplasmic sperm injection (ICSI) (Palermo et al., 1992), a procedure that enables males with severe semen impairment to father their own children. The assessment of success of IVF treatment has, however, long been controversial. It is usually evaluated by using crude pregnancy or delivery rates per treatment cycle (Alsalili et al., 1995), or by calculating the cumulative pregnancy or delivery rates using timetable analyses (Dor et al., 1992, 1996; Tan et al., 1992).
This study evaluated a computerized database of all the women who underwent IVF during a 6-year period, and performed a multivariate analysis in order to delineate which factors could best predict live birth rate.
Materials and methods Between 1995 and 2001, 1928 women attended the IVF unit at the Chaim Sheba Medical Centre and underwent a total of 6336 consecutive cycles. These women were admitted to the IVF programme following thorough infertility investigation or failure of conventional infertility treatments or IVF in other centres. Two main protocols were used for the induction of follicular growth: (i) approximately 85% underwent a long protocol of gonadotrophin-releasing hormone analogue (GnRHa) (D-trip-6LHRH microcapsules: Decapeptyl Depot 3.75 mg microcapsules: Ferring Ltd, Malmö, Sweden) or Suprefact (Aventis Pharma Ltd., West Malling, Kent; nasal spray 600–900 µg/day) given either in
645
Article - Multivariate analysis to predict IVF treatment outcome - SE Elizur et al.
the mid-luteal or early follicular phase of the menstrual cycle followed by human menopausal gonadotrophin (HMG, Menogon: Ferring) or recombinant FSH (rFSH) (Gonal-f: Serono, Boulogne, France), 225 IU/day 15 days later after verification of complete ovarian suppression; (ii) approximately 15% underwent ovarian stimulation using the flare-up protocol: daily Decapeptyl 0.1 mg or Superfact 600–900 µg was used from day 1 of the menstrual cycle and stimulation by HMG or rFSH was administered from day 3. Follicular growth was monitored by measurements of serum oestradiol and progesterone and by performing transvaginal sonography. Human chorionic gonadotrophin (HCG) (Chorigonal: Teva Pharmaceutical Industries Ltd, Jerusalem, Israel), 10,000 IU, was administered when there were at least two follicles with a mean diameter of ≥18 mm and a serum oestradiol concentration of at least 1500 pmol/l was noted. Oocyte retrieval, culture, fertilization, embryo culture and transfer were carried out as previously described (Dor et al., 1992). Delivery rates were calculated according to cycle characteristics, including the number of oocytes retrieved, whether ICSI has been performed, the number of transferred embryos and whether or not there had been the option of choosing embryos for transfer. The latter opportunity was available when there were more than three fertilized oocytes or when the number of fertilized oocytes was less than or equal to three, but greater than the number of transferred embryos. In addition, successful delivery rates were calculated according to the couple’s characteristics, i.e. the age of the women at first treatment and the aetiology of infertility. Data were analysed using SAS software (SAS Institute Inc., 1996). The SAS procedure PHREG (proportional hazard regression) was used to calculate adjusted success rate in terms of hazard ratio and 95% confidence interval (95% CI). Because of the discrete time (i.e. cycle) scale, the discrete logistic model was applied. The woman’s age at treatment, number of oocytes retrieved, performance of ICSI, number of embryos transferred and whether there was the possibility of choosing embryos for transfer were included in the model as time-dependent (cycledependent) variables.
Results From 1 January 1995 to 30 June 2001, 6336 IVF cycles were performed on 1928 women at the IVF Unit, Chaim Sheba Medical Centre. There were 5605 cycles of ovarian stimulation and 731 cycles of frozen embryo transfer. In all, 295/5605 cycles (5.3%) were cancelled (Figure 1). All analyses were performed for the 5310 cycles with embryo transfer. The mean numbers of retrieved and fertilized oocytes were 10.7 ± 7.3 and 6.0 ± 4.3 respectively, and the mean number of transferred embryos was 3.5 ± 1.6. ICSI was performed in 2795 cycles (52.6%), mostly indicated by severe male factor infertility. A total of 1126 pregnancies (21.2%) were diagnosed and 689 (61.2%) ended in live birth. The delivery rate according to cycle characteristics is summarized in Table 1. The highest delivery rates were observed when ≥10 oocytes were retrieved (16.6%), in cycles where ICSI was performed (15.4 compared with 10.3%), with transfer of three embryos (17.1%) and when a choice of embryos was available (17.0 compared with 7.8%). The mean age of the women who attended the IVF programme was 32.7 ± 5.9 years. A total of 235 women (12.3%) were older than 40 years. The mean number of IVF cycles per woman was 2.8 ± 2.3 (range 1–28). Women who were 26–30 years of age had the highest delivery rate (48.0%) (Table 2). The main cause of infertility was male factor in 905 couples (46.9%), mechanical factor in 514 couples (26.7%), anovulation in 252 couples (13.0%), and unexplained infertility in 257 couples (13.3%). In male factor infertility, the delivery rate was 42.1% compared with 31.3% in other aetiologies of infertility (Table 2). Table 3 summarizes the results of multivariate analysis of success rate adjusting for both the woman’s and the cycle characteristics. The adjusted success ratio was significantly lower in women who were >30 years of age (31–35 years, P = 0.04, 36–40 years, P = 0.003, 41–52 years, P < 0.001) and in couples for whom the aetiology of infertility was other than male factor infertility
6336 cycles
731 (frozen embryos transferred)
5310 (ET was performed)
5605 (ovulation induction)
295 (no ET)
Figure 1. IVF cycles performed from 1 January 1995 to 30 June 2001 at the IVF Unit, Chaim Sheba Medical Centre, Israel. ET = embryo transfer.
646
Article - Multivariate analysis to predict IVF treatment outcome - SE Elizur et al.
Table 1. Successful delivery rate according to cycle characteristics (5310 cycles). No. IVF cycles
Table 2. Successful delivery rate according to women’s characteristics (1928 couples).
No. successful cycles (delivery rate %)
Number of retrieved oocytes 1–5 1392 6–9 1302 10–14 1310 15+ 1306
88 (6.3) 165 (12.7) 218 (16.6) 218 (16.7)
Performance of ICSI No 2515 Yes 2795
259 (10.3) 430 (15.4)
No. successful pregnancies (delivery rate %)
Age at first treatment (years) 17–25 195 80 (41.0) 26–30 569 273 (48.0) 31–35 499 183 (36.7) 36–40 374 116 (31.0) 41+ 235 29 (12.3) Unknown 56 8 (14.3) Aetiology of infertility Male factor 789 only Mechanical 448 Anovulation 220 Unexplained 224 Unknown 247
Number of transferred embryos 1 495 22 (4.4) 2 816 93 (11.4) 3 1309 224 (17.1) 4+ 2184 346 (15.8) Unknown 506 4 (0.8) Choice of embryo No 1404 Yes 3371 Unknown 535
No. women
332 (42.1) 131 (29.2) 80 (36.4) 71 (31.7) 75 (30.4)
109 (7.8) 574 (17.0) 6 (1.1)
Table 3. Multivariate analysis of success rates. NS = not significant. Variables
Women’s age (years) 17–25 26–30 31–35 36–40 41–52 Aetiology of infertility Male factor only Other aetiologies
Adjusted success ratio (95% CI)
P-value
0.84 (0.62–1.15) 1.0 0.88 (0.71–1.08) 0.80 (0.63–1.02) 0.38 (0.26–0.56)
NS
1.0 0.81 (0.67–0.97)
Number of retrieved oocytes 1–5 1.0 6–9 0.89 (0.65–1.22) 10–14 1.00 (0.72–1.36) 15+ 0.88 (0.63–1.23) Performance of ICSI Conventional IVF 1.0 ICSI 1.32 (1.09–1.59) Number of transferred embryos 1 1.0 2 1.97 (1.20–3.23) 3 2.69 (1.65–4.39) 4+ 2.14 (1.29–3.54) Choice of embryos No 1.0 Yes 1.96 (1.46–2.64)
0.04 0.003 <0.001
0.03
NS NS NS
0.004
0.008 <0.001 0.003
<0.001
Table 4. Multiple deliveries according to number of embryos transferred (n = 630)a. No. embryos transferred
Deliveries Singletonb (n) n (%)
Twin n (%)
1 2 3 4+
22 79 206 323
–
22 (100) 62 (78.5) 138 (67.0) 196 (60.7)
Triplet n (%)
– 17 (21.5) – 57 (27.7) 11 (5.3) 107 (33.1) 20 (6.2)
aThe number of embryos transferred was unknown in four deliveries and multiplicity was unknown in 55 deliveries. bP for trend <0.01.
647
Article - Multivariate analysis to predict IVF treatment outcome - SE Elizur et al.
(adjusted success ratio = 0.81; 95% CI 0.67–0.97, P = 0.03). The performance of ICSI and infertility aetiology were assessed independently. Delivery rates following ICSI were significantly higher compared with conventional IVF (success ratio = 1.32; 95% CI 1.09–1.59, P = 0.004). Transferring two embryos doubled the chance of successful delivery compared with one embryo (success ratio = 1.97; 95% CI 1.20–3.23), but transferring three embryos was not significantly better than two embryos. There was a significantly higher delivery rate when there was a choice of embryos (success ratio = 1.96; 95% CI 1.46–2.64, P < 0.001). Table 4 shows the number of multiple deliveries according to the number of transferred embryos. Multiple delivery rates following two transferred embryos (21.5% twins) were significantly lower (P < 0.01) compared with the transfer of three embryos (27.7% twins, 5.3% triplets) and four or more embryos.
Discussion The current study is unique by its use of a comprehensive multivariate model to simultaneously assess many variables that are known to influence delivery rate following IVF treatment. Evaluation of different variables on treatment success has been published previously (Speirs et al., 1996; Templeton et al., 1996; Baker et al., 2000), but so far as is known, there are only very few publications (Hunault et al., 2002) that take into account so many variables at the same time, as does the present study. The performance of ICSI or the aetiology of infertility, for instance, is lacking from most of the previous publications. In the current study, the highest live birth rates were among women aged 26–30 years, confirming the findings of Templeton et al. (1996). There was a sharp decline for older women and, surprisingly, in the youngest women as well. This may relate to the more severe causes of infertility that are present in the reproductive years among younger women. The lowest delivery rates were seen, as expected, in women older than 40 years, although 12.3% did succeed in delivering a child. Univariate analysis for evaluating the influence of the number of oocytes retrieved at each cycle on delivery rates revealed that the lowest delivery rates (6.3%) were recorded in the cycles in which 1–5 oocytes were retrieved. This is in contrast to previous reports (Lashen et al., 1999; Biljan et al., 2000) that found similar implantation and pregnancy rates following cycles with up to 3–5 follicles compared with cycles with more follicles, although these studies included only women with normal day 3 FSH or women aged <40 years. After adjustment for the woman’s age, aetiology of infertility, number of transferred embryos and whether or not there was a choice of embryo, however, the number of retrieved oocytes was no longer significantly associated with successful delivery.
648
ICSI was performed in 2795 cycles (52.6%) in the present study, mainly for severe male factor infertility. In previously published studies, couples with male factor infertility had the lowest pregnancy rates compared with other couples (Tan et
al., 1992; Alsalili et al., 1995). The data are more optimistic: the delivery rates following ICSI (15.4% per cycle) were significantly higher compared with those following conventional IVF (10.3%). Moreover, the data did not support the findings of studies that showed the rates of on-going pregnancies after repeated ICSI cycles as being lower than those after IVF (Stalf et al., 1999). The findings of the current study indicate that the introduction of ICSI for the treatment of couples with male factor infertility, a treatment that dramatically changed the previously poor prognosis of those couples, enhances the likelihood of successful pregnancy outcome. This study has shown that transferring two or three embryos is superior to single-embryo transfer (adjusted success ratio 1.97 and 2.69 respectively). However, transferring three embryos was not significantly superior to two embryos. In addition, the multiple delivery rates following a two-embryo transfer (21.5% twins) were significantly lower compared with those associated with transferring three embryos (27.7% twins, 5.3% triplets) and four embryos (33.1% twins, 6.2% triplets) (P < 0.01 for trend). These findings are consistent with previous studies (Vauthier-Brouzes et al., 1994; Dean et al., 2000; Licciardi et al., 2001) demonstrating that transfer of two embryos does not decrease pregnancy rate (although it did reduce the risk of multiple pregnancies). Most of these findings, however, come from studies conducted on women ≤35 years of age who had good quality embryos. The current study suggests that regardless of the women’s age, transferring two embryos is the preferred option in IVF treatments as a way to reduce multiple pregnancies without decreasing delivery rates. Hunault et al. (2002) suggested a predictive model, taking into account women’s age and embryo quality for selecting patients for elective single embryo transfer in order to reduce the chance of twin pregnancy. Others (De Neubourg and Gerris, 2003; Gurgan and Demirol, 2004) have shown that when a good prognosis group is chosen for a single-embryo transfer, pregnancy rate can be as high as 40%. However, in the double-embryo transfer group, pregnancy rates were higher compared with the single-embryo transfer group but there was a greater number of multiple pregnancies. ZegersHochschild et al. (2003) suggested that multiple pregnancies are not solely dependent on the number of embryos transferred, but rather may reflect a subgroup of highly fertile women. In the present study, a single embryo was transferred only in those cycles where only one embryo was available for transfer, whereas other investigators (Gerris et al., 1999, 2002) transferred the best quality single embryo chosen from a number of embryos. This is probably the reason that transferring one embryo had the lowest delivery rate in the current analysis. The findings apparently reflect the lower delivery rates in poor responders, and not an overall assessment of a single-embryo transfer protocol. This predicting model does not take into account IVF cycles in which embryos were not transferred due to cycle cancellation (e.g. poor ovarian response, no fertilization) or frozen embryo transfer. However, only 5.3% of the cycles were cancelled, and therefore the model provides information for the vast majority of IVF cycles.
Article - Multivariate analysis to predict IVF treatment outcome - SE Elizur et al.
In conclusion, multivariate analysis of a large number of patients undergoing IVF suggests that transferring two embryos achieves the best results in IVF in terms of delivery rates and multiple pregnancies. In addition, the highest delivery rates were observed in women aged between 26–30 years of age and for couples in whom ICSI was performed for male factor infertility.
Acknowledgement Esther Eshkol is thanked for editorial assistance.
References Alsalili M, Yuzpe A, Tummon I et al. 1995 Cumulative pregnancy rates and pregnancy outcome after in-vitro fertilization: >5000 cycles at one centre. Human Reproduction 10, 470–474. Baker HWG, Saunders DM, Tyler JPP et al. 2000 Difficulties in comparison of results between assisted reproductive technology clinics: an attempt to standarize reporting. Reproductive Technologies 10, 103–111. Biljan MM, Buckett WM, Dean N et al. 2000 The outcome of IVF–embryo transfer treatment in patients who develop three follicles or less. Human Reproduction 15, 2140–2144. Dean NL, Phillips SJ, Buckett WM et al. 2000 Impact of reducing the number of embryos transferred from three to two in women under the age of 35 who produced three or more high-quality embryos. Fertility and Sterility 74, 820–823. De Neubourg D, Gerris J 2003 Single embryo transfer – state of the art. Reproductive BioMedicine Online 7, 615–622. Dor J, Ben-Shlomo I, Levran D 1992 The relative success of gonadotropin-releasing hormone analogue, clomiphene citrate, and gonadotropin in 1,099 cycles of in vitro fertilization. Fertility and Sterility 58, 986–990. Dor J, Seidman DS, Ben-Shlomo I et al. 1996 Cumulative pregnancy rate following in-vitro fertilization: the significance of age and infertility aetiology. Human Reproduction 11, 425–428. Gerris J, De Neubourg D, Mangelschots K et al. 1999 Prevention of twin pregnancy after in-vitro fertilization or intracytoplasmic sperm injection based on strict embryo criteria: a prospective randomized clinical trial. Human Reproduction 14, 2581–2587. Gerris J, De Neubourg D, Mangelschots K, et al. 2002 Elective single day 3 embryo transfer halves the twinning rate without
decrease in the ongoing pregnancy rate of an IVF/ICSI programme. Human Reproduction 17, 2626–2631. Gurgan T, Demirol A 2004 Why and how should multiple pregnancies be prevented in assisted reproduction treatment programmes? Reproductive BioMedicine Online 9, 237–244. Hunault CC, Eijkemans MJC, Pieters MHEC et al. 2002 A prediction model for selecting patients undergoing in vitro fertilization for elective single embryo transfer. Fertility and Sterility 77, 725–732. Lashen H, Ledger W, Lopez-Bernal A, Barlow D 1999 Poor responders to ovulation induction: is proceeding to in-vitro fertilization worthwhile? Human Reproduction 14, 964–969. Licciardi F, Berkeley AS, Krey L et al. 2001 A two- versus threeembryo transfer: the oocyte donation model. Fertility and Sterility 75, 510–513. Palermo G, Joris H, Devroey P, Van Steirteghem AC 1992 Pregnancies after intracytoplasmic injection of single spermatozoon into an oocyte. Lancet 340, 17–18. SAS Institute Inc. 1996 SAS/STAT Software: Changes and Enhancements through Release 6.11. SAS Institute Inc., Cary, NC, USA. Speirs AL, Baker HW, Abdullah N 1996 Analysis of factors affecting embryo implantation. Human Reproduction 11 (suppl. 1), 187–191. Stalf T, Herrero J, Turley H et al. 1999 Different cumulative pregnancy rates in patients with repeated IVF or ICSI cycles: possible influence of a male factor. Andrologia 31, 149–156. Tan SL, Royston P, Campbell S et al. 1992 Cumulative conception and livebirth rates after in-vitro fertilisation. Lancet 339, 1390–1394. Templeton A, Morris JK, Parslow W 1996 Factors that affect outcome of in-vitro fertilisation treatment. Lancet 348, 1402–1406. Vauthier-Brouzes D, Lefebvre G, Lesourd S et al. 1994 How many embryos should be transferred in in vitro fertilization? A prospective randomized study. Fertility and Sterility 62, 339–342. Zegers-Hochschild F, Bravo M, Fernandez E et al. 2004 Multiple gestation as a marker of reproductive efficacy: learning from assisted reproductive technologies. Reproductive BioMedicine Online 8, 125–129.
Received 26 August 2004; refereed 21 September 2004; accepted 24 February 2005.
649
Article Factors predicting IVF treatment outcome: a multivariate analysis of 5310 cycles Shai E Elizur is a tutor in Obstetrics and Gynaecology at the Sackler Faculty of Medicine, Tel Aviv University, Israel. Born in 1965, he obtained his MD magna cum laude in 1997 as a graduate of the Sackler Faculty of Medicine in Tel Aviv. He received his qualification in Obstetrics and Gynaecology in 2004. Since then he is working in the IVF unit at Sheba Medical Centre, Tel Hashomer, Israel. Current clinical research interests include fertility preservation in women with cancer.
Dr Shai E Elizur Shai E Elizur1,3, Liat Lerner-Geva2, Jacob Levron1, Adrian Shulman1, David Bider1, Jehoshua Dor1 Unit, Department of Obstetrics and Gynecology, Chaim Sheba Medical Centre, Tel-Hashomer 52621; 2Women and Children’s Health Research Unit, Gertner Institute for Epidemiology and Health Policy Research, Tel-Hashomer; both affiliated to the Sackler School of Medicine, Tel-Aviv University, Tel-Aviv, Israel 3Correspondence: Tel: +972 3 5302882; Fax: +972 3 5341589; e-mail: [email protected]
1IVF
Abstract The objective of this study was to analyse factors predicting live birth rate following IVF. A computerized database of 1928 women who underwent 5310 consecutive IVF cycles in a single IVF unit was evaluated. Data on the women’s age, number of retrieved oocytes, performance of intracytoplasmic sperm injection (ICSI), aetiology of infertility, number of transferred embryos and option of choosing embryos for transfer were evaluated. There were 1126 pregnancies that resulted in 689 live births. Transferring two embryos doubled the chances of delivery compared with one embryo, but transferring three embryos was not significantly superior to two embryos. Moreover, following a three-embryo transfer, the multiple delivery rates were significantly higher (P < 0.01) compared with transferring two embryos. Optimal delivery rates were observed in women aged 26–30 years, with gradual decline with advanced age. The performance of ICSI resulted in higher delivery rates compared with conventional insemination. According to these data, the best live birth results following IVF treatment were achieved when the maternal age was 26–30 years, in couples with male factor infertility undergoing ICSI, and when two embryos were transferred. Keywords: delivery rate, ICSI, IVF
Introduction The introduction of IVF and embryo transfer in the late 1970s revolutionized the treatment of infertility and spawned an entire new field of assisted reproduction technology. Traditional IVF was the optimal solution to infertility due to female mechanical factors. The next major breakthrough in treating infertility came with the introduction of intracytoplasmic sperm injection (ICSI) (Palermo et al., 1992), a procedure that enables males with severe semen impairment to father their own children. The assessment of success of IVF treatment has, however, long been controversial. It is usually evaluated by using crude pregnancy or delivery rates per treatment cycle (Alsalili et al., 1995), or by calculating the cumulative pregnancy or delivery rates using timetable analyses (Dor et al., 1992, 1996; Tan et al., 1992).
This study evaluated a computerized database of all the women who underwent IVF during a 6-year period, and performed a multivariate analysis in order to delineate which factors could best predict live birth rate.
Materials and methods Between 1995 and 2001, 1928 women attended the IVF unit at the Chaim Sheba Medical Centre and underwent a total of 6336 consecutive cycles. These women were admitted to the IVF programme following thorough infertility investigation or failure of conventional infertility treatments or IVF in other centres. Two main protocols were used for the induction of follicular growth: (i) approximately 85% underwent a long protocol of gonadotrophin-releasing hormone analogue (GnRHa) (D-trip-6LHRH microcapsules: Decapeptyl Depot 3.75 mg microcapsules: Ferring Ltd, Malmö, Sweden) or Suprefact (Aventis Pharma Ltd., West Malling, Kent; nasal spray 600–900 µg/day) given either in
645
Article - Multivariate analysis to predict IVF treatment outcome - SE Elizur et al.
the mid-luteal or early follicular phase of the menstrual cycle followed by human menopausal gonadotrophin (HMG, Menogon: Ferring) or recombinant FSH (rFSH) (Gonal-f: Serono, Boulogne, France), 225 IU/day 15 days later after verification of complete ovarian suppression; (ii) approximately 15% underwent ovarian stimulation using the flare-up protocol: daily Decapeptyl 0.1 mg or Superfact 600–900 µg was used from day 1 of the menstrual cycle and stimulation by HMG or rFSH was administered from day 3. Follicular growth was monitored by measurements of serum oestradiol and progesterone and by performing transvaginal sonography. Human chorionic gonadotrophin (HCG) (Chorigonal: Teva Pharmaceutical Industries Ltd, Jerusalem, Israel), 10,000 IU, was administered when there were at least two follicles with a mean diameter of ≥18 mm and a serum oestradiol concentration of at least 1500 pmol/l was noted. Oocyte retrieval, culture, fertilization, embryo culture and transfer were carried out as previously described (Dor et al., 1992). Delivery rates were calculated according to cycle characteristics, including the number of oocytes retrieved, whether ICSI has been performed, the number of transferred embryos and whether or not there had been the option of choosing embryos for transfer. The latter opportunity was available when there were more than three fertilized oocytes or when the number of fertilized oocytes was less than or equal to three, but greater than the number of transferred embryos. In addition, successful delivery rates were calculated according to the couple’s characteristics, i.e. the age of the women at first treatment and the aetiology of infertility. Data were analysed using SAS software (SAS Institute Inc., 1996). The SAS procedure PHREG (proportional hazard regression) was used to calculate adjusted success rate in terms of hazard ratio and 95% confidence interval (95% CI). Because of the discrete time (i.e. cycle) scale, the discrete logistic model was applied. The woman’s age at treatment, number of oocytes retrieved, performance of ICSI, number of embryos transferred and whether there was the possibility of choosing embryos for transfer were included in the model as time-dependent (cycledependent) variables.
Results From 1 January 1995 to 30 June 2001, 6336 IVF cycles were performed on 1928 women at the IVF Unit, Chaim Sheba Medical Centre. There were 5605 cycles of ovarian stimulation and 731 cycles of frozen embryo transfer. In all, 295/5605 cycles (5.3%) were cancelled (Figure 1). All analyses were performed for the 5310 cycles with embryo transfer. The mean numbers of retrieved and fertilized oocytes were 10.7 ± 7.3 and 6.0 ± 4.3 respectively, and the mean number of transferred embryos was 3.5 ± 1.6. ICSI was performed in 2795 cycles (52.6%), mostly indicated by severe male factor infertility. A total of 1126 pregnancies (21.2%) were diagnosed and 689 (61.2%) ended in live birth. The delivery rate according to cycle characteristics is summarized in Table 1. The highest delivery rates were observed when ≥10 oocytes were retrieved (16.6%), in cycles where ICSI was performed (15.4 compared with 10.3%), with transfer of three embryos (17.1%) and when a choice of embryos was available (17.0 compared with 7.8%). The mean age of the women who attended the IVF programme was 32.7 ± 5.9 years. A total of 235 women (12.3%) were older than 40 years. The mean number of IVF cycles per woman was 2.8 ± 2.3 (range 1–28). Women who were 26–30 years of age had the highest delivery rate (48.0%) (Table 2). The main cause of infertility was male factor in 905 couples (46.9%), mechanical factor in 514 couples (26.7%), anovulation in 252 couples (13.0%), and unexplained infertility in 257 couples (13.3%). In male factor infertility, the delivery rate was 42.1% compared with 31.3% in other aetiologies of infertility (Table 2). Table 3 summarizes the results of multivariate analysis of success rate adjusting for both the woman’s and the cycle characteristics. The adjusted success ratio was significantly lower in women who were >30 years of age (31–35 years, P = 0.04, 36–40 years, P = 0.003, 41–52 years, P < 0.001) and in couples for whom the aetiology of infertility was other than male factor infertility
6336 cycles
731 (frozen embryos transferred)
5310 (ET was performed)
5605 (ovulation induction)
295 (no ET)
Figure 1. IVF cycles performed from 1 January 1995 to 30 June 2001 at the IVF Unit, Chaim Sheba Medical Centre, Israel. ET = embryo transfer.
646
Article - Multivariate analysis to predict IVF treatment outcome - SE Elizur et al.
Table 1. Successful delivery rate according to cycle characteristics (5310 cycles). No. IVF cycles
Table 2. Successful delivery rate according to women’s characteristics (1928 couples).
No. successful cycles (delivery rate %)
Number of retrieved oocytes 1–5 1392 6–9 1302 10–14 1310 15+ 1306
88 (6.3) 165 (12.7) 218 (16.6) 218 (16.7)
Performance of ICSI No 2515 Yes 2795
259 (10.3) 430 (15.4)
No. successful pregnancies (delivery rate %)
Age at first treatment (years) 17–25 195 80 (41.0) 26–30 569 273 (48.0) 31–35 499 183 (36.7) 36–40 374 116 (31.0) 41+ 235 29 (12.3) Unknown 56 8 (14.3) Aetiology of infertility Male factor 789 only Mechanical 448 Anovulation 220 Unexplained 224 Unknown 247
Number of transferred embryos 1 495 22 (4.4) 2 816 93 (11.4) 3 1309 224 (17.1) 4+ 2184 346 (15.8) Unknown 506 4 (0.8) Choice of embryo No 1404 Yes 3371 Unknown 535
No. women
332 (42.1) 131 (29.2) 80 (36.4) 71 (31.7) 75 (30.4)
109 (7.8) 574 (17.0) 6 (1.1)
Table 3. Multivariate analysis of success rates. NS = not significant. Variables
Women’s age (years) 17–25 26–30 31–35 36–40 41–52 Aetiology of infertility Male factor only Other aetiologies
Adjusted success ratio (95% CI)
P-value
0.84 (0.62–1.15) 1.0 0.88 (0.71–1.08) 0.80 (0.63–1.02) 0.38 (0.26–0.56)
NS
1.0 0.81 (0.67–0.97)
Number of retrieved oocytes 1–5 1.0 6–9 0.89 (0.65–1.22) 10–14 1.00 (0.72–1.36) 15+ 0.88 (0.63–1.23) Performance of ICSI Conventional IVF 1.0 ICSI 1.32 (1.09–1.59) Number of transferred embryos 1 1.0 2 1.97 (1.20–3.23) 3 2.69 (1.65–4.39) 4+ 2.14 (1.29–3.54) Choice of embryos No 1.0 Yes 1.96 (1.46–2.64)
0.04 0.003 <0.001
0.03
NS NS NS
0.004
0.008 <0.001 0.003
<0.001
Table 4. Multiple deliveries according to number of embryos transferred (n = 630)a. No. embryos transferred
Deliveries Singletonb (n) n (%)
Twin n (%)
1 2 3 4+
22 79 206 323
–
22 (100) 62 (78.5) 138 (67.0) 196 (60.7)
Triplet n (%)
– 17 (21.5) – 57 (27.7) 11 (5.3) 107 (33.1) 20 (6.2)
aThe number of embryos transferred was unknown in four deliveries and multiplicity was unknown in 55 deliveries. bP for trend <0.01.
647
Article - Multivariate analysis to predict IVF treatment outcome - SE Elizur et al.
(adjusted success ratio = 0.81; 95% CI 0.67–0.97, P = 0.03). The performance of ICSI and infertility aetiology were assessed independently. Delivery rates following ICSI were significantly higher compared with conventional IVF (success ratio = 1.32; 95% CI 1.09–1.59, P = 0.004). Transferring two embryos doubled the chance of successful delivery compared with one embryo (success ratio = 1.97; 95% CI 1.20–3.23), but transferring three embryos was not significantly better than two embryos. There was a significantly higher delivery rate when there was a choice of embryos (success ratio = 1.96; 95% CI 1.46–2.64, P < 0.001). Table 4 shows the number of multiple deliveries according to the number of transferred embryos. Multiple delivery rates following two transferred embryos (21.5% twins) were significantly lower (P < 0.01) compared with the transfer of three embryos (27.7% twins, 5.3% triplets) and four or more embryos.
Discussion The current study is unique by its use of a comprehensive multivariate model to simultaneously assess many variables that are known to influence delivery rate following IVF treatment. Evaluation of different variables on treatment success has been published previously (Speirs et al., 1996; Templeton et al., 1996; Baker et al., 2000), but so far as is known, there are only very few publications (Hunault et al., 2002) that take into account so many variables at the same time, as does the present study. The performance of ICSI or the aetiology of infertility, for instance, is lacking from most of the previous publications. In the current study, the highest live birth rates were among women aged 26–30 years, confirming the findings of Templeton et al. (1996). There was a sharp decline for older women and, surprisingly, in the youngest women as well. This may relate to the more severe causes of infertility that are present in the reproductive years among younger women. The lowest delivery rates were seen, as expected, in women older than 40 years, although 12.3% did succeed in delivering a child. Univariate analysis for evaluating the influence of the number of oocytes retrieved at each cycle on delivery rates revealed that the lowest delivery rates (6.3%) were recorded in the cycles in which 1–5 oocytes were retrieved. This is in contrast to previous reports (Lashen et al., 1999; Biljan et al., 2000) that found similar implantation and pregnancy rates following cycles with up to 3–5 follicles compared with cycles with more follicles, although these studies included only women with normal day 3 FSH or women aged <40 years. After adjustment for the woman’s age, aetiology of infertility, number of transferred embryos and whether or not there was a choice of embryo, however, the number of retrieved oocytes was no longer significantly associated with successful delivery.
648
ICSI was performed in 2795 cycles (52.6%) in the present study, mainly for severe male factor infertility. In previously published studies, couples with male factor infertility had the lowest pregnancy rates compared with other couples (Tan et
al., 1992; Alsalili et al., 1995). The data are more optimistic: the delivery rates following ICSI (15.4% per cycle) were significantly higher compared with those following conventional IVF (10.3%). Moreover, the data did not support the findings of studies that showed the rates of on-going pregnancies after repeated ICSI cycles as being lower than those after IVF (Stalf et al., 1999). The findings of the current study indicate that the introduction of ICSI for the treatment of couples with male factor infertility, a treatment that dramatically changed the previously poor prognosis of those couples, enhances the likelihood of successful pregnancy outcome. This study has shown that transferring two or three embryos is superior to single-embryo transfer (adjusted success ratio 1.97 and 2.69 respectively). However, transferring three embryos was not significantly superior to two embryos. In addition, the multiple delivery rates following a two-embryo transfer (21.5% twins) were significantly lower compared with those associated with transferring three embryos (27.7% twins, 5.3% triplets) and four embryos (33.1% twins, 6.2% triplets) (P < 0.01 for trend). These findings are consistent with previous studies (Vauthier-Brouzes et al., 1994; Dean et al., 2000; Licciardi et al., 2001) demonstrating that transfer of two embryos does not decrease pregnancy rate (although it did reduce the risk of multiple pregnancies). Most of these findings, however, come from studies conducted on women ≤35 years of age who had good quality embryos. The current study suggests that regardless of the women’s age, transferring two embryos is the preferred option in IVF treatments as a way to reduce multiple pregnancies without decreasing delivery rates. Hunault et al. (2002) suggested a predictive model, taking into account women’s age and embryo quality for selecting patients for elective single embryo transfer in order to reduce the chance of twin pregnancy. Others (De Neubourg and Gerris, 2003; Gurgan and Demirol, 2004) have shown that when a good prognosis group is chosen for a single-embryo transfer, pregnancy rate can be as high as 40%. However, in the double-embryo transfer group, pregnancy rates were higher compared with the single-embryo transfer group but there was a greater number of multiple pregnancies. ZegersHochschild et al. (2003) suggested that multiple pregnancies are not solely dependent on the number of embryos transferred, but rather may reflect a subgroup of highly fertile women. In the present study, a single embryo was transferred only in those cycles where only one embryo was available for transfer, whereas other investigators (Gerris et al., 1999, 2002) transferred the best quality single embryo chosen from a number of embryos. This is probably the reason that transferring one embryo had the lowest delivery rate in the current analysis. The findings apparently reflect the lower delivery rates in poor responders, and not an overall assessment of a single-embryo transfer protocol. This predicting model does not take into account IVF cycles in which embryos were not transferred due to cycle cancellation (e.g. poor ovarian response, no fertilization) or frozen embryo transfer. However, only 5.3% of the cycles were cancelled, and therefore the model provides information for the vast majority of IVF cycles.
Article - Multivariate analysis to predict IVF treatment outcome - SE Elizur et al.
In conclusion, multivariate analysis of a large number of patients undergoing IVF suggests that transferring two embryos achieves the best results in IVF in terms of delivery rates and multiple pregnancies. In addition, the highest delivery rates were observed in women aged between 26–30 years of age and for couples in whom ICSI was performed for male factor infertility.
Acknowledgement Esther Eshkol is thanked for editorial assistance.
References Alsalili M, Yuzpe A, Tummon I et al. 1995 Cumulative pregnancy rates and pregnancy outcome after in-vitro fertilization: >5000 cycles at one centre. Human Reproduction 10, 470–474. Baker HWG, Saunders DM, Tyler JPP et al. 2000 Difficulties in comparison of results between assisted reproductive technology clinics: an attempt to standarize reporting. Reproductive Technologies 10, 103–111. Biljan MM, Buckett WM, Dean N et al. 2000 The outcome of IVF–embryo transfer treatment in patients who develop three follicles or less. Human Reproduction 15, 2140–2144. Dean NL, Phillips SJ, Buckett WM et al. 2000 Impact of reducing the number of embryos transferred from three to two in women under the age of 35 who produced three or more high-quality embryos. Fertility and Sterility 74, 820–823. De Neubourg D, Gerris J 2003 Single embryo transfer – state of the art. Reproductive BioMedicine Online 7, 615–622. Dor J, Ben-Shlomo I, Levran D 1992 The relative success of gonadotropin-releasing hormone analogue, clomiphene citrate, and gonadotropin in 1,099 cycles of in vitro fertilization. Fertility and Sterility 58, 986–990. Dor J, Seidman DS, Ben-Shlomo I et al. 1996 Cumulative pregnancy rate following in-vitro fertilization: the significance of age and infertility aetiology. Human Reproduction 11, 425–428. Gerris J, De Neubourg D, Mangelschots K et al. 1999 Prevention of twin pregnancy after in-vitro fertilization or intracytoplasmic sperm injection based on strict embryo criteria: a prospective randomized clinical trial. Human Reproduction 14, 2581–2587. Gerris J, De Neubourg D, Mangelschots K, et al. 2002 Elective single day 3 embryo transfer halves the twinning rate without
decrease in the ongoing pregnancy rate of an IVF/ICSI programme. Human Reproduction 17, 2626–2631. Gurgan T, Demirol A 2004 Why and how should multiple pregnancies be prevented in assisted reproduction treatment programmes? Reproductive BioMedicine Online 9, 237–244. Hunault CC, Eijkemans MJC, Pieters MHEC et al. 2002 A prediction model for selecting patients undergoing in vitro fertilization for elective single embryo transfer. Fertility and Sterility 77, 725–732. Lashen H, Ledger W, Lopez-Bernal A, Barlow D 1999 Poor responders to ovulation induction: is proceeding to in-vitro fertilization worthwhile? Human Reproduction 14, 964–969. Licciardi F, Berkeley AS, Krey L et al. 2001 A two- versus threeembryo transfer: the oocyte donation model. Fertility and Sterility 75, 510–513. Palermo G, Joris H, Devroey P, Van Steirteghem AC 1992 Pregnancies after intracytoplasmic injection of single spermatozoon into an oocyte. Lancet 340, 17–18. SAS Institute Inc. 1996 SAS/STAT Software: Changes and Enhancements through Release 6.11. SAS Institute Inc., Cary, NC, USA. Speirs AL, Baker HW, Abdullah N 1996 Analysis of factors affecting embryo implantation. Human Reproduction 11 (suppl. 1), 187–191. Stalf T, Herrero J, Turley H et al. 1999 Different cumulative pregnancy rates in patients with repeated IVF or ICSI cycles: possible influence of a male factor. Andrologia 31, 149–156. Tan SL, Royston P, Campbell S et al. 1992 Cumulative conception and livebirth rates after in-vitro fertilisation. Lancet 339, 1390–1394. Templeton A, Morris JK, Parslow W 1996 Factors that affect outcome of in-vitro fertilisation treatment. Lancet 348, 1402–1406. Vauthier-Brouzes D, Lefebvre G, Lesourd S et al. 1994 How many embryos should be transferred in in vitro fertilization? A prospective randomized study. Fertility and Sterility 62, 339–342. Zegers-Hochschild F, Bravo M, Fernandez E et al. 2004 Multiple gestation as a marker of reproductive efficacy: learning from assisted reproductive technologies. Reproductive BioMedicine Online 8, 125–129.
Received 26 August 2004; refereed 21 September 2004; accepted 24 February 2005.
649