Risk of adverse pregnancy and perinatal outcomes after high technology infertility treatment: a comprehensive systematic review

An excellent meta-analysis of individual data from RCTs demonstrated that the elective single embryo transfer (SET) is less effective (of about 50 %) than double embryo transfer (DET) in women aged more than 33 years but not in younger, and that the odds of a multiple live birth in women randomized to eSET were significantly smaller than for women who received DET [28]. After eSET, the adjusted risk to deliver a LBW baby (aOR 0.36, 95 % CI 0.15 to 0.87) and to have a PTB (aOR 0.33, 95 % CI 0.20 to 0.55) was significantly reduced, whereas the adjusted odds of a term singleton birth after eSET were almost five times higher than those after DET (aOR 4.93, 95 % CI 2.98 to 8.18) [28]. Data obtained from the US National Assisted Reproductive Technology Surveillance System, a retrospective population-based study analyzed the 82,508 fresh autologous ART cycles, confirmed that a higher chance of a good perinatal outcome, defined as a live birth at or after 37 weeks of gestation of a normal birth weight (2,500 g or greater) singleton, is associated with SET in patients with a favorable prognosis who were aged younger than 35 years [29]. The same conclusions were obtained after the analysis of the 2013 data, i.e. increasing use of elective SET, when clinically appropriate (typically age < 35 years), might reduce multiple births and related adverse consequences of ART [7]. Moreover, the average rate of SET was 21.4 % also among good prognosis patients and varied considerably among States (from 4 to 77.5 %) [7].

Irrespective of the risk related to the multiple pregnancies, the number of embryos transferred can be an independent and potential factor influencing the obstetric and perinatal outcomes. Available data were very heterogeneous showing better perinatal outcomes [30, 31] or no difference [32, 33] between SET singletons and DET singletons. A meta-analysis demonstrated no difference in the incidence of PTB (aOR 0.83, 95 % CI 0.64 to 1.06) between singletons born after SET and singletons born after DET [9]. However, an effect of vanishing twin syndrome (VTS) [34], as consequence of DET or multiple embryo transfer, cannot be excluded. This is particularly true in consideration of the higher risk of pregnancy complications observed in VTS pregnancies from IVF/ICSI cycles in comparison with VTS pregnancies from NC [35]. A large cohort study including 7,757 deliveries following IVF/ICSI procedures found more adverse outcomes among VTS survivors occurring after 8 weeks of gestation, even after adjustment for maternal age and parity [36]. However, that study did not differentiate between “true” (first trimester) cases of the VTS and those occurring at the second trimester since embryonic losses from 9 weeks until midgestation were grouped together, and a case-mix of di-and monochorionic twins is probable. On the other hand, similar maternal and perinatal complications, such as gestational duration and birth weights, were observed between VTS survivors and singletons in a retrospective series including IVF/ICSI patients when only well selected patients with initial heart beat on both embryos and with clear chorionicity were included [37].

In singletons with a ‘vanishing co-twin’, the risk of PTB (aOR 1.73, 95 % CI 1.54 to 1.94) and of other adverse perinatal outcomes (aOR ranging from 1.73 to 2.88) was significantly higher than in singletons from a single gestation [38]. The obstetric and perinatal risks in VTS seem to increase in accordance with the number of reduced fetuses [38]. Other case–control and cohort studies on singleton deliveries after IVF/ICSI showed that VTS was associated with higher risk of vaginal bleeding and preterm premature rupture of membranes, and that VTS survivors had a birth weight significantly lower, and a rate of LBW and SGA about double when compared to singleton controls [39‐41]. Of note, no difference was observed in terms of duration of gestation suggesting a direct effect of the VTS on the survived embryos/fetus [40].

A well designed retrospective study on a very large cohort of 252,994 singleton deliveries evaluated the pregnancy and perinatal outcomes, adjusted for confounders such as fertility treatment and maternal age, in VTS in comparison with those recorded in singletons and dichorionic twins [42]. VTS was found to be an independent risk factor for several adverse perinatal outcomes [42]. Specifically, the risk of GDM (aOR 1.4, 95 % CI 1.01 to 2.0), intrauterine growth restriction (IUGR) (aOR 2.7, 95 % CI 1.7 to 4.3), VLBW (aOR 6.9, 95 % CI 4.7 to 10.2), low Apgar scores (aOR 1.9, 95 % CI 1.1 to 3.3), and perinatal mortality (aOR 2.4, 95 % CI 1.2 to 4.6) was significantly increased in VTS when compared with singletons. In addition, VTS significantly influenced also the risk of malpresentation, placental abruption, and cesarean section [42]. However, that study [42] included mainly NC, and the effect of VTS in IVF/ICSI vs. NC pregnancies was not tested. To this regard, a recent retrospective matched (for maternal age, previous gravidity, parity, and prepregnancy BMI) analysis demonstrated that the incidence of the VTS in infertile IVF/ICSI patients is lower than in NC (12.6 % vs. 18.2 %, respectively; estimated for twin pregnancies) but the perinatal outcomes resulted worse. In fact, the risk of GDM (aOR 3.0, 95 % CI 1.6 to 5.6 vs. aOR 0.46, 95 % CI 0.2 to 1.1) and of PE (aOR 1.6, 95 % CI 0.7 to 6.1 vs. aOR 1.00, 95 % CI 0.8 to 1.8) in VTS pregnancies was very high in IVF/ICSI patients but lower or unchanged in NC women [35]. Also the incidence of IUGR (aOR 3.0, 95 % CI 1.8 to 5.2 vs. aOR 9.2, 95 % CI 5 to 22) and of LBW (aOR 4.0, 95 % CI 1.8 to 7.1 vs. aOR 2.1, 95 % CI 1.6 to 4.0) in VTS pregnancies was higher in IVF/ICSI patients than in NC women [35]. Very interesting data from multiple logistic regression analyses that identified in VTS pregnancies obtained after IVF/ICSI cycles a risk of IUGR 28-fold higher (aOR 28.2, 95 % CI 2.2 to 14.5) [35].

In comparison with children born after natural-cycle IVF, IVF children conceived after stimulation with clomiphene citrate (CC) and with CC plus gonadotropins had a higher risk of LBW [43]. On the other hand, IVF children conceived after ovarian stimulation with gonadotropins alone did not have a higher risk of LBW compared to natural-cycle [43] confirming IUI data [44]. In addition, no effect of the duration of stimulation, amount of drug administered, and number of oocytes retrieved was detected on SGA or LBW in a cohort study on 32,000 singletons born from gonadotropin IVF/ICSI cycles [45]. A retrospective cohort study on 392 women less than 40 years of age demonstrated the safety, in terms of obstetric and neonatal complications, of the use of gonadotropin-releasing hormone (GnRH) agonist to trigger ovulation in antagonist cycles. Specifically, there were no significant differences in the rate of maternal (27.6 vs. 20.8 %) or neonatal complications (19.7 vs. 20.0 %) between the GnRH agonist and hCG [46]. In addition, luteal GnRH antagonist administration in women with severe early ovarian hyperstimulation syndrome (OHSS) is associated not only with not different live birth rates but also with similar duration of gestation (36.9 ± 0.9 vs. 36.9 ± 2.4 weeks) and neonatal weight (2392 ± 427 vs. 2646 ± 655 g) when compared to patients who do not receive the treatment [47]. However, in patients undergoing COH using the GnRH antagonist protocol, GnRH agonist trigger were found to be independent risk factors for ectopic pregnancy [48].

The ovarian response to COH has been also related to pregnancy complications. A recent observational United Kingdom (UK) study [49] on 402,185 IVF cycles and 65,868 singleton live birth outcomes detected a risk of PTB (aOR 1.15, 95 % CI 1.03 to 1.28), EPTB (aOR 1.30, 95 % CI 1.03 to 1.64), LBW (aOR 1.17, 95 % CI 1.05 to 1.30), and VLBW (aOR 1.23, 95 % CI 0.97 to 1.55) higher in women with a high number (>20) of oocytes retrieved when compared with women with a normal response (10–15 oocytes) [49]. Moreover, the observation that higher rates of LBW infants in singleton gestations after IVF cycles was only present in patients who had a fresh ET when compared with those with frozen-thawed cycles [50] has suggested that the gonadotropin stimulated multifollicular development and the production of supraphysiologic levels of sex steroid hormones immediately before embryo implantation might represent an independent contributing factor to that increased risk (and not only an effect of fresh or frozen embryos as below detailed). In fact, in IVF cycles the elevated peak serum estradiol (E₂) levels on the day of the human chorionic gonadotropin (hCG) trigger were closely related to the risks of SGA and PE in singleton pregnancies [51, 52]. Thus, the supraphysiologic hormonal milieu at the time of implantation and placentation during a fresh IVF cycle may modulate trophoblast invasion and lead to abnormal placentation, whereas hormonal levels in a frozen ET cycle are much more akin to the endocrine environment of NC. This hypothesis has been tested by a small cohort study on a population at high risk for OHSS for high E₂ levels; the elective cryopreservation of all embryos with subsequent thawed ET reduced significantly the risk of PE (21.9 % vs. 0 %, respectively; OR not calculable), and to deliver SGA infants (OR 0.09, 95 % CI 0.01 to 0.65) as compared with the patients who had fresh ET [53]. However, the published case–control studies on the obstetric and perinatal outcomes in patients with OHSS showed divergent results [54‐56]. In line with previous data, no increased risk of the adverse outcomes among women with a poor ovarian response (≤ 3 oocytes) compared with women with a normal response was observed [49], even if several important confounders, such as smoking, body mass index (BMI), polycystic ovarian syndrome (PCOS), and gonadotropin doses were not available for an optimal data adjustment. On the other hand, a retrospective cohort study of the Society for Assisted Reproductive Technology (SART) on 14,086 patients demonstrated an inverse relationship between maximal basal follicle stimulating hormone (FSH) levels and the risk for PTB and LBW in singleton IVF gestations with the lowest risk of PTB (aRR 0.87, 95 % CI 0.76 to 1.01) and of LBW (aRR 0.89, 95 % CI 0.73 to 1.04) in patients with the lowest ovarian reserve, defined as highest serum basal and/or CC-induced FSH levels [57].

Finally, the hitherto published literature regarding the relation between OHSS and pregnancy complications is limited and controversial [56]. This disparity in the observed outcomes probably results from the inclusion of patients with OHSS, not exclusively limited to those with significant increase in vascular permeability, as reflected by third space fluid accumulation necessitating drainage. A recent study [58] has demonstrated that severe OHSS, complicated by third space fluid sequestration necessitating drainage, is not associated with adverse late pregnancy outcome (i.e. IUGR and PIH/PE), except for PTB. Therefore, following resolution of the OHSS, pregnancies should be regarded as any pregnancy resulting from IVF treatment, with special attention to identify and treat PTB.

In a first systematic review with meta-analysis [9], the risk of PTB was not significantly different in singletons conceived after day 5 vs. day 2 embryo culture (aOR 1.14, 95 % CI 0.80 to 1.64). A further meta-analysis [59] aimed to assess specifically the risk of pregnancy complications in singleton pregnancies resulting from ET at the blastocyst stage (days 4–5 or 5–6) compared with those resulting from ET at the cleavage stage (days 2–3 or day 3 or days 2–4) demonstrated, after data synthesis of 7 retrospective cohort studies, that pregnancies occurring as a result of ET at the blastocyst stage were associated with a higher risk of PTB (RR 1.27, 95 % CI 1.22 to 1.31) and VPTB (RR 1.22, 95 % CI 1.10 to 1.35) delivery and a lower risk of SGA [59]. However, these data were not adjusted for confounders and included very different study protocols. Another more recent meta-analytic study [60] aimed to evaluate the perinatal outcomes among singleton births following blastocyst stage (day 5 or 6) ET compared with cleavage stage (day 2–4) ET and included 6 observational studies with low to moderate risk of bias. It demonstrated, after adjusting data for main confounders, that only the risk of PTB (aOR 1.32, 95 % CI 1.19 to 1.46) was significantly higher among births after blastocyst transfer than after cleavage stage transfer [60]. No further difference in perinatal/neonatal outcome was observed [60].

At the moment, the most complete data on maternal and child health adjusted for confounders were from a large retrospective cohort study [61] comparing 46,288 neonates born after cleavage-stage ET to 22,751 neonates born after blastocyst stage ET, and showing an increased risk for PTB (aOR 1.39, 95 % CI 1.29 to 1.50) and VPTB (aOR 1.35, 95 % CI 1.13 to 1.61) in singletons neonates born after transfer of embryos at blastocyst stage vs. cleavage stage [61]. These results were confirmed also in twins (aOR 1.81, 95 % CI 1.63 to 2.00; and aOR 1.75, 95 % CI 1.50 to 2.04 for PTB and VPTB, respectively) [61]. However, pregnancies that did not result in a live birth were not analyzed and the number of embryos transferred was not considered. To this regard, a retrospective cohort study analyzed the transfer of 693 fresh single cleavage embryos comparing to 850 fresh single blastocysts [62]. A higher live birth rate (33.5 % vs. 13.8 %) was detected after single blastocysts transfer and after adjusting data for several confounders, while no effect of the extended culture was observed on obstetric or perinatal outcomes [62]. These data have been confirmed also more recently in a retrospective matched case–control study by the same researchers [63] and by a large retrospective population-based study including a total of 50,788 infants (43,952 singleton and 3,418 twin deliveries) [64]. Conversely, there were lower odds of PTB among twins (aOR 0.80, 95 % CI 0.70 to 0.93) conceived after blastocyst transfer than after cleavage stage ET [64].

An important issue observed in the studies comparing the transfer of embryos at blastocyst vs. cleavage state was the discrepancy between the increased incidence in PTB and the similar incidence of LBW neonates in singletons [61]. Regarding this, an effect of the extended culture on the neonatal weight has been demonstrated. In fact, the incidence of SGA in neonates born after extended embryo culture compared with cleavage stage ET was significantly lower (OR 0.80, 95 % CI 0.74 to 0.87) [61]. The effect of extended embryo culture on fetal growth has been specifically assessed in a cohort study on 1,079 singleton neonates born after fresh ET [65]. The risk of large-for-gestational age (LGA) (but not for SGA) adjusted for mother’s age, BMI, parity, type of treatment (IVF or ICSI), main cause of infertility and embryo culture period was increased after extended embryo culture (aOR 2.22, 95 % CI 1.14 to 4.34), and the length of the embryo culture was a strong and significant factor determining the gestation and gender-adjusted birth weight of the IVF babies [65].

On the other hand, a recent retrospective registry study [66] (including 4,819 singletons born after blastocyst ET, 25,747 after cleavage stage ET, and 1,196,394 after NC) demonstrated, after adjusting data for several confounders, a risk of perinatal mortality (aOR 1.61, 95 % CI 1.14 to 2.29) higher in singletons born after blastocyst ET vs. cleavage stage ET, although a lower incidence of LBW (aOR 0.83, 95 % CI 0.71 to 0.97) and SGA (aOR 0.71, 95 % CI 0.56 to 0.88) children was confirmed. The risk of placenta previa (aOR 2.08, 95 % CI 1.70 to 2.55) and placental abruption (aOR 1.62, 95 % CI 1.15 to 2.29) was higher in pregnancies after blastocyst ET vs. pregnancies after cleavage stage ET [66].

The findings of obstetric and perinatal complications in pregnancies achieved after ET at the blastocyst stage are unexpected because good quality embryos from women with the best prognosis tend to be selected for extended culture. The specific mechanism underlying the association between extended embryo culture and PTB is unknown. It could be the result of the transfer at blastocyst stage inducing a defective implantation for asynchrony between the endometrium and embryo [67], trigger genetic and epigenetic modifications in the pre-implantation embryos due to the culture medium of the extended culture [67, 68] and/or in decidual/trophodermal cells due to abnormal embryo-endometrium interaction [69]. In fact, extended culture up to blastocyst stage increases the incidence of monozygotic twins (OR 3.04, 95 % CI 1.54 to 6.01) and the male-to-female ratio (OR 1.29, 95 % CI 1.10 to 1.51 [70], and both are associated with PTB [71]. Although an effect of the procedure of embryo freezing and thawing cannot be excluded (see below), all these factors, alone and/or in concert, can be responsible for the higher risk of PTB detected in children born after blastocyst transfer [72].

Well-designed studies [73‐75] have recently evaluated the effect of different types of in vitro culture medium and of the duration of in vitro culture on obstetric and perinatal outcomes in IVF/ICSI populations showing conflicting results, suggesting that data cannot be generalized for all commercial culture media. The different culture medium (Medicult vs. Vitrolife) and the duration of culture (day 3 vs. day 5) had no effect on singleton newborns birthweight, as well as on the incidence of PTB and VPTB, was detected on a large population of 2,098 infertile women treated with fresh IVF and ICSI cycles after controlling for confounders [75]. Unfortunately, all these studies had a retrospective design and the risk of residual confounding cannot be excluded [72].

The proportion of frozen ET has increased in consideration of the better reproductive outcomes compared to IVF fresh cycles [76, 77]. Analyzing the SART database from 2008 to 2011, odds of extra-uterine pregnancy 65 % lower in women who had a frozen compared with a fresh transfer in autologous cycles were observed [78], that risk is lower for day 5 blastocyst vs. day 3 embryos [79].

The first systematic reviews [80] available in the literature, including mostly pregnancy studies achieved by using slow-frozen embryos and only a few vitrification studies, concluded that the risk of adverse obstetric outcome was similar between cryopreserved embryos/oocytes and NC children or fresh ET. A more recent systematic review with meta-analysis [81], including 11 observational studies, suggested that singleton pregnancies arising from frozen embryos seem to offer better obstetric and perinatal outcome than those obtained after fresh oocyte cycles. The clinical benefits regarded the risk of antepartum hemorrhage (RR 0.67, 95 % CI 0.55 to 0.81), PTB (RR 0.84, 95 % CI 0.78 to 0.90), SGA (RR 0.45, 95 % CI 0.30 to 0.66), LBW (RR 0.69, 95 % CI 0.62 to 0.76) and perinatal mortality (RR 0.68, 95 % CI 0.48 to 0.96) [81]. The use of frozen ET was also subsequently confirmed to be a predictor of reduced risk of PTB when compared to fresh ET after adjusting data for main confounders (aOR 0.85, 95 % CI 0.76 to 0.94) [9]. However, a significant heterogeneity in the technique of freezing (slow freezing vs. vitrification), in embryo stage (cleavage stage vs. blastocyst stage vs. both), and in regimen in replacement cycles (natural cycles vs. hormone replacement cycles) was observed among studies. In addition, a bias due to difference in prognosis between populations cannot be excluded. In fact, although the risk of pregnancy complications was higher in the paired comparison of the same patient who conceived after fresh vs. subsequent frozen-thawed ET (including donor cycles) [82], it is likely that the frozen embryo populations are more likely composed of the better prognosis patients. Similarly, also the selection bias due to the physical effects of freezing and thawing on the worse embryos cannot be excluded.

In a large retrospective cohort study [83], the perinatal outcome of 6,647 singletons conceived after frozen-thawed ET were compared with 42,242 singletons born after fresh ET and 288,542 singletons born after spontaneous conception. Data, adjusted for several confounding factors, confirmed that singletons conceived after frozen-thawed ET had better perinatal outcomes than after fresh ET [83]. Specifically, albeit the perinatal outcomes were worse when compared to NC, singletons born after frozen-thawed ET had a lower risk of LBW (aOR 0.81, 95 % CI 0.71 to 0.91), PTB (aOR 0.84, 95 % CI 0.76 to 0.92) and SGA (aOR 0.72, 95 % CI 0.62 to 0.83) when compared with singletons born after fresh ET. However, after frozen-thawed ET, the perinatal mortality resulted increased when compared with both fresh ET (aOR 1.49, 95 % CI 1.07 to 2.07) and NC (aOR 1.39, 95 % CI 1.03 to 1.87) [83]. The increased perinatal mortality rate has been showed only in few studies [84] but not in others [85‐88]. Specifically, the perinatal mortality rate resulted 25.2 per 1,000 births for fresh IVF/ICSI vs. 17.5 per 1,000 for frozen-thawed cycles [88].

A Japanese large-scale registry-based study not only confirmed that SET of frozen-thawed blastocyst was associated with significantly lower odds of PTB, LBW and SGA, but, for the first time, suggested a risk three-fold higher of placenta accreta (aOR 3.16, 95 % CI 1.71 to 6.23) after frozen-thawed ET than after fresh ET [89]. A study [90] explored specifically this complication in patients using IVF/ICSI, with autologous or donor oocytes, undergoing fresh or cryopreserved transfer. After multivariate analysis (including many predictors such as non-Caucasian race, uterine factor infertility, prior myomectomy and placenta previa), a significant and specific association between frozen-thawed ET and placenta accreta (aOR 3.20, 95 % CI 1.14 to 9.02) was detected [90]. On the other hand, a decreased risk of placenta previa (RR 0.71, 95 % CI 0.53 to 0.95) and placental abruption (RR 0.44, 95 % CI 0.24 to 0.83) in pregnancies after frozen thawed ET was reported [81]. Unfortunately, the demographics of patients having these complications is lacking in these studies.

The frozen-thawed embryo cycles showed an increased association with PIH/PE (aOR 1.58, 95 % CI 1.35 to1.86) [89], similarly to data from the Swedish IVF registry [84]. In comparison with pregnancies following fresh cycles, those following frozen-thawed ET had a higher risk of PIH/PE with a risk difference of 1.8 % (95 % CI 1.2 to 2.8) and 5.1 % (95 % CI 3.0 to 7.1) in singleton and twin pregnancies, respectively [91]. These data were confirmed when compared with fresh cycle pregnancies by the same mother after sibling analysis (OR 2.63, 95 % CI 1.73 to 3.99) [91] and, interestingly, also in PCOS patients [92]. In fact, a large multicenter RCT on a total of 1,508 infertile women with PCOS assigned to undergo either fresh-embryo transfer or embryo cryopreservation (vitrification) followed by frozen-embryo transfer demonstrated an incidence of PE significantly higher with frozen-embryo transfer [rate ratio (ReRa) 3.12, 95 % CI 1.26 to 7.73] [92].

The risk of post-term birth (aOR 1.40, 95 % CI 1.27 to 1.55), LGA (aOR 1.45, 95 % CI 1.27 to 1.64), and macrosomia (aOR 1.58, 95 % CI 1.39 to 1.80) also increased [83, 89]. Of note, even if not statistically significant, the risk of LGA was higher in boys than in girls (41 % vs. 17 %) [83] confirming previous data that showed an odd for LBW in frozen-thawed ET significantly higher in female than that in male neonates [93]. The first paper reporting the increased risk of LGA newborns in frozen-thawed ET compared with fresh ET, irrespective of maternal BMI and abnormal oral glucose tolerance test (OGTT), was published in 2010 [85]. At the moment, there is not a conclusive hypothesis for explaining that finding and many potential explanations of the increased rate of LGA in frozen-thawed embryos have been proposed. Different media used for culturing of human embryos can differently affect the birth-weight of the newborns [94] with an effect of the embryo freezing and thawing [95]. However, the effect on human embryos of the many components of the culture media is not known. A clinical study [96] demonstrated no difference in birth-weight between two culture media but a higher birth-weight after embryo freezing and thawing in both groups when compared with fresh cycles suggesting generic effect of cryoprotectants on enzymatic pattern involved in epigenetic programming [97]. However, a specific effect of media for extended culture cannot be excluded since babies born after day 5–6 transfer had significantly higher risk of LGA compared with babies born after day 2 transfer [65] (see before).

Recently, a national register-based controlled cohort study with further meta-analysis was designed in order to clarify whether the “large baby syndrome” was caused by intrinsic maternal factors or related to the freezing/thawing procedures [98]. In order to avoid biases due to different frozen techniques and embryo culture/stage, only singletons born after slow freezing and ET of cleavage stage embryos (day 2 or 3) were included in the meta-analysis [98]. The increased risk of LGA in frozen-thawed embryos singletons was firstly confirmed also in a sibling cohort after adjusting data for maternal age, parity and year of birth (aOR 3.45, 95 % CI 1.33 to 8.33) [98]. The pooled risk for LGA (aOR 1.54, 95 % CI 1.31 to 1.81) and macrosomia (aOR 1.64, 95 % CI 1.26 to 2.12) was confirmed in frozen-thawed embryos vs. fresh embryos [98]. These risks were the same extension as when compared to NC [98]. Unfortunately, these data were not adjusted for maternal BMI and GDM that can act as residual confounders [98].

A recent retrospective UK cohort study [99] analyzed a total of 112,432 cycles (95,911 fresh and 16,521 frozen cycles) using multivariate logistic regression and detected no association between type of embryo transferred (frozen vs. fresh) and PTB (aRR 0.96, 99.5 % CI 0.88 to 1.03) and VPTB [aRR 0.86, 99.5 % CI 0.70 to 1.05). On the other hand, the odds of LBW (aRR 0.73, 99.5 % CI 0.66 to 0.80) and VLBW (aRR 0.78, 99.5 % CI 0.63 to 0.96) were lower after frozen ET, and those of having a LGA was higher (aRR 1.64, 99.5 % CI 1.53 to 1.76) [99]. Moreover, that data were anonymized and regarded outcomes of IVF cycles and not of IVF patients, and did not included pregnancy complications. Thus, a bias due to more cycles of the same patient cannot be excluded, and the influence of the obstetric complications on perinatal outcomes cannot be assessed.

Finally, two retrospective recent cohort studies suggest that frozen cycles are associated with a reduction of the incidence of ectopic pregnancies and monozygotic twins [100, 101]. Specifically, a registry study [100] based on a total of 44,102 pregnancies from Australian and New Zealand Assisted Reproduction Technology Database demonstrated that the SET of frozen blastocyst is associated with lowest risk of ectopic pregnancy (aOR 0.70, 95 % CI 0.54 to 0.91; vs. fresh blastocyst transfer), and that the characteristics influencing that risk were the number of embryo transferred, and the use of fresh and/or cleavage stage embryos. In addition, the analysis of 6,103 cycles revealed that frozen ET was associated with a significant reduction in monozygotic twins incidence (aOR 0.48, 95 % CI 0.29 to 0.80) [101].

Initially, there were some safety concerns about old vitrification procedures because the use of very high concentrations of cryoprotectants required to optimize the efficacy could induce toxic effects [102]. New vitrification methods requiring very small volumes and small concentrations of cryoprotectants and extreme cooling rates seem safer [103, 104]. However, new generation devices may require direct contact between samples and liquid nitrogen during vitrification with the risk due to the potential exposure to contaminants.

Specific data on the influence of oocyte vitrification on obstetric risk are few and poorly reported, and head-to-head studies are lacking. A systematic review intercepted only uncontrolled or case reports reporting very limited data on the health of the newborns from vitrified and slow frozen oocytes. Also successive studies [105] reported reassuring pregnancy and neonatal data, even if the population studied were very little and poorly or not controlled. Only more recently, a retrospective study [106] on 954 pregnancies achieved after ET from slow frozen (n. 197) or fresh (n. 757) oocytes reported data on pregnancy and perinatal health. No differences were found between the use of fresh and frozen oocytes in the rates of ectopic pregnancies (3.6 % vs. 2.9 %) and spontaneous abortions (17.6 % vs. 26.9 %), and in gestational age at delivery [106]. However, the mean birth weights were significantly lower with fresh oocyte than with frozen-thawed oocytes, both in singleton (2.725 ± 727 g vs. 3.231 ± 615 g) and twins (2,128 ± 555 g 2,418- ± 92 g) [106]. Of interest, the analysis of the 63 patients who obtained pregnancies both in fresh and thawed cycles (138 pregnancies) showed no differences in the abortion rate and in the mean birth weight [106].

Only in 2014, a retrospective cohort study [107] compared the obstetric and perinatal outcomes of infertile patients who received IVF cycles after oocyte vitrification compared with those who received fresh oocytes. In particular, data from a total of 804 and 996 pregnancies, and from 1,027 and 1,224 children, respectively for vitrified and fresh oocyte group, were analyzed. No differences were found between the two techniques of cryopreservation in the rate of pregnancy complications, including obstetric, perinatal and puerperal problems [107]. In the vitrified oocytes group, the incidence of invasive procedures (aOR 2.12, 95 % CI 1.41 to 3.20), such as chorionic villous sampling or amniocentesis, and of urinary tract infections (aOR 0.51, 95 % CI 0.28 to 0.91) were respectively increased and reduced [107]. The proportion of female offspring was significantly higher in the vitrification group (53.8 % vs. 47.4 %) [107]. Main data did not change after sub-analysis for singleton and multiple pregnancies [107].

Comparable gestational age at delivery, weight at birth, and congenital anomaly rate were observed analyzing 435 blastocyst vitrification cycles using an open system, with the fresh oocyte control cycles [108]. Other researchers have confirmed those results in further studies after the vitrification of both cleavage stage embryos [109‐112] and blastocysts [113, 114], or a combination of both [87]. One large study including 6,623 infants conceived after vitrification, found no difference also in terms of gestational perinatal mortality, although the incidence of SGA was lower and the weight at birth was heavier after vitrification in comparison with fresh ET [87]. Of note, all pregnancies and babies analyzed were conceived after SET.

One retrospective, single-center study of children born after day 3 ET from fresh, slow frozen or vitrified embryos showed that the incidence of PIH/PE, GDM, placenta previa and abruptio placenta were similar in all groups [112]. Specifically, no difference between vitrified and slow freezing day 3 embryos was observed in gestational ages at delivery, PTB, and perinatal mortality, even if the birth-weight of the babies born from vitrified embryos was higher in comparison with those of the babies born from slow frozen and fresh embryos, and the rate of LBW in vitrified embryos arm was significantly lower than in fresh embryos arm [112]. In twins, the birth-weight for children born from vitrified day 3 embryos was higher than that from the slow freezing, even if the difference cannot be considered clinically significant [112]. Unfortunately, a crucial recall bias cannot be excluded because the obstetric and perinatal data were obtained by questionnaires sent to the parents without checking medical records. Conversely, another recent retrospective study [113] demonstrated that the freezing methods of cleavage stage embryos did not exert any effect on the neonatal weight.

No difference in almost all pregnancy and perinatal/neonatal outcomes was observed in a Swedish retrospective study [114] comparing pregnancies achieved from vitrified blastocysts with those achieved from fresh blastocysts and slow frozen cleavage stage embryos. However, in pregnancies from vitrified blastocysts the incidence of SGA was lower (12.1 vs. 3 %) in comparison with fresh blastocyst, and the rate of major postpartum hemorrhage was higher (25 vs. 6 and 7.5 %) in comparison with fresh blastocyst and slow frozen cleavage stage embryos [114].

The largest and most recent population-based cohort study compared obstetric and perinatal outcomes of 16,845, 2,766 and 6,537 clinical pregnancies, and 13,049, 2,065 and 4,955 live deliveries of fresh, slow frozen and vitrified blastocysts [115]. Singletons born after transfer of vitrified but not slow frozen blastocysts had a lower risk of PTB (aRR 0.86, 95 % CI 0.77 to 0.96), LBW (aRR 0.67, 95 % CI 0.58 to 0.78) and SGA (aRR 0.60, 95 % CI 0.53 to 0.68) in comparison with singletons born after transfer of fresh blastocysts, [115]. The beneficial effect of vitrification on the birth weight of babies (and on the pregnancy duration) have been confirmed in a smaller retrospective study on 1,209 infertile patients who received a total of 1,157 fresh and 645 vitrified-warmed single blastocyst (day 5) transfers [116]. A higher infant birth weight has been demonstrated not only in singletons but also in twins after retrospective analysis of 784 fresh transfers and 382 vitrified-warmed DET at blastocyst stage [117].

An interesting retrospective study [118] analyzed the perinatal and neonatal health, adjusted for treatment variables and maternal characteristics, after 960 vitrified cycles and 1,644 fresh cycles according to the day of vitrification or fresh transfer, i.e. day 3 (cleavage stage) vs. day 5 (blastocyst stage). Singletons, but not twins, born after vitrification showed only lower SGA rate (aOR 0.55, 95 % CI 0.34 to 0.90) in comparison with peers born after fresh ET. The embryonic stage at vitrification or at fresh transfer did not influence any perinatal outcome [118]. These data seem to suggest that both vitrification and extensive culture at blastocyst stage can act and interact on fetal/neonatal weight.

Any adverse effects of the IVM procedure on pregnancy and perinatal outcomes are a partially unknown field of research. From a mechanicistic point of view, IVM could affect the oocyte development, as any intervention in the growth phase could affect oocyte maturation, fertilization and subsequent embryo development [119]. On the other hand, the minimal gonadotropin stimulation, which very mildly stimulates the endometrium mimics natural environmental.

A systematic review [120] of head-to-head RCTs aimed to compare IVM followed by IVF/ICSI vs. conventional IVF/ICSI in terms of live birth rate and/or other efficacy and safety maternal/perinatal outcomes did not find any such study. Cha et al. [121] firstly reported on obstetric and perinatal outcomes in 41 IVM pregnancies suggesting no adverse effect of the technique. However, that data were uncontrolled and obtained in PCOS patients. Subsequent controlled studies confirmed the maternal and perinatal safety of IMV in comparison to IVF and/or ICSI, even if only few obstetric and perinatal data have been reported [122‐124]. A slightly increased rate of GDM (17 % vs. 11 % and 10 % for IVM vs. IVF and ICSI, respectively) and of PE (12 % vs. 5 % and 8 % for IVM vs. IVF and ICSI, respectively) was observed in 55 IVM patients when compared with BMI and parity matched IVF/ICSI subjects (n. 217 and n. 160, respectively) [123]. No difference was observed all other pregnancy and perinatal outcomes among ART groups [123]. In singleton IVM pregnancies, the cesarean section (39 vs. 26 %) and instrumental delivery (9.5 vs. 6.5 %) rates were higher than in NC pregnancies but no difference between IVM and NC was detected in LBW (3 vs. 9 %), VLBW (0 vs. 2 %), gestational age (39 ± 3 vs. 39 ± 6 weeks), VPTB (0 vs. 2 %) [123]. Of interest, contrarily to IVF/ICSI neonates, the birth weight at delivery resulted heavier in IVM neonates (3,482 g vs. 3,260 g) than in NC neonates, and the PTB incidence (5 vs. 5 %) was not different from NC pregnancies [123]. A possible role of COH in determining a reduced birthweight in IVF/ICSI children has been also more recently suggested in a small but well controlled retrospective cohort study [125] performed on 196 IVM cycles compared with those of ICSI treatments. Although couples with a maternal age higher than 39 years or affected by azoospermia were excluded, the sample was heterogeneous as IVM cycles had different priming regimens and maturation techniques [125]. Only a difference in birth weight was observed between IVM and ICSI babies (3091 ± 669 g vs. 3269 ± 619 g) [125]. That difference may represent an inevitable influence of drug administration [125] but also an increased incidence of imprinting disorders [119, 126].

Although many studies on the safety of IVM are available in the literature, all have a retrospective design, very small sample sizes, and are not controlled and include a number of biases and (unadjusted) confounders. Obviously a high prevalence of PCOS cases in these studies could influence the results [11, 127]. In fact, a retrospective study [124] on 1,581 positive pregnancy tests demonstrated a rate of miscarriage after IVM significantly higher than after IVF/ICSI (25.3 % vs. 15.7 % and 12.6 %, respectively) but directly related to the PCOS rather than to the IVM procedure [124].

Although hatching of the blastocyst is a crucial step for embryo implantation, and the failure to hatch can be considered one of limiting factors of the embryo development, the artificial disruption of the zona pellucida (assisted hatching) can potentially affect embryo competence and, thus, increase the risk of pregnancy complications. Although each technique used for assisted hatching can potentially exert specific effects on maternal and perinatal/neonatal health, very few data are available in literature about the effects of assisted hatching on obstetric and perinatal outcomes.

In a double-blind RCT [128], no effect on assisted hatching using acidic Tyroide’s solution was detected in women younger than 38 years on neonatal complications. Similarly, a retrospective cohort analysis [129] including a total of 699 women found no statistically significant effect of laser-assisted hatching, performed on day 3 in frozen (slow frozen and vitrified) embryos on gestational age, birth weight and Apgar score. The safety of the laser-assisted hatching was confirmed in singleton and twin pregnancies. Moreover, these data were limited by the heterogeneous subject selection due to inclusion criteria and patient choice.

Finally, the use of assisted hatching has been associated with an increased incidence of monozygotic twinning [130]. Moreover, available data are contrasting [101].

The European Society of Human Reproduction and Embryology (ESHRE) PGD Consortium reported that pregnancies and babies born after ICSI/PGD are not different from pregnancies obtained and babies born after ICSI treatment [131]. Similarly, gestational age, birth weight, and perinatal death were not different in a study analyzing 995 children born after ICSI/PGD and 1,507 children born after ICSI [132]. Interestingly, fewer multiples born after PGD presented a LBW [132]. These results, unfortunately, were not controlled for main confounders and did not have a NC group as reference.

A recent study [133] compared the perinatal outcomes of 245 neonates born after ICSI/PGD with those of neonates born to mothers matched for age, BMI and parity during the same period after ICSI (n. 242) and after NC (n. 733). The incidence of pregnancy complications after ICSI/PGD was low and not different from control groups whereas PTB and IUGR babies were more frequent in ICSI conceptions than in NC [133]. In singleton pregnancies, the weight of ICSI/PGD neonates, the rate of IUGR and LBW neonates and of PTB were not different from NC neonates but higher in comparison with ICSI neonates [133]. LGA was more frequent in the PGD group than after NC [133]. In twin pregnancies, no difference in LBW was detected between ICSI/PGD and NC twins, whereas significantly more ICSI twins presented with LBW [133]. That result did not change after sub-analysis for fresh and cryopreserved ET or for type of biopsy [133]. Of note, the mechanical partial zona dissection was the method for zona breaching in all cases of PGD.

A very recent retrospective cohort study, using data from the U.S. National ART Surveillance System on fresh autologous cycles with blastocyst transfer, compared 97,069 non-PGD procedures with other 9,833 PGD procedures demonstrating the influence on perinatal outcome of the specific indication for PGD [134]. Specifically, in women aged less than 35 years who had a live birth after ICSI/PGD the odd ratio of LBW infant was lower (aOR 0.73, 95 % CI 0.54 to 0.98) and higher (aOR 1.25, 95 % CI 1.01 to 1.54) in comparison with non-PGD cycles, when the indication was the detection of genetic disorders and aneuploidy screening, respectively [134]. On the other hand, no difference in any perinatal outcome was observed in women aged more than 35 years according to PGD indication [134]. Another recent multicenter-cohort study confirmed that the risk of adverse perinatal outcomes was mainly related to the underlying parental conditions rather than the PGD procedure [135]. In fact, after stratification of data for PGD indication, adverse obstetric and neonatal outcomes were only present in children conceived by PGD due to parental monogenetic disorder but not in children conceived after PGD due to parental chromosomal aberrations [135]. Nevertheless, there was a consistent risk of placenta previa (aOR 9.1, 95 % CI 3.4 to 24.9) after PGD and IVF/ICSI, suggesting that parental factors cannot explain all adverse outcomes [135].

Notwithstanding these available scientific data, definitive conclusions on the safety of the PGD cannot be drawn since couples undergoing PGD are usually fertile and in good health, data on fetal/perinatal growth and weight can be biased by timing of the biopsy, extended embryo culture, type of ET, etc., and long-term data on offspring are totally lacking. To this regard, a recent retrospective questionnaire analysis [136] showed a higher incidence of PIH (aOR 4.85, 95 % CI 1.34 to 17.56) in singleton pregnancies after blastocyst-stage biopsy and frozen ET than after cleavage-stage biopsy and fresh ET. In this regard, the Society of Obstetricians and Gynecologists of Canada [137] underlined that the data on an improved pregnancy outcome are inconsistent.

A systematic review on gestational surrogacy has been very recently published, and the pregnancy and perinatal outcomes of gestational carriers compared, when possible, with those of standard IVF and OD cycles [162]. The incidence of PIH ranged from 4.3 to 10 % and from 2.9 and 7.4 %, and the incidence of placenta previa and/or placental abruption from 1.1 to 7.9 % and from 1.1 to 3.7 %, respectively, in singleton and twin surrogate pregnancies [162], and resulted not different from those observed in IVF pregnancies and lower than that usually reported in OD pregnancies (ranged from 16 to 40 %) [163]. Three cases of hysterectomies related to delivery were also reported in gestational carriers and were due to uterine atony, placenta accreta and uterine rupture [162]. In surrogate singletons, the incidence of PTB and of LBW resulted, respectively, ranging from 0 to 11.5 % and from 0 to 11.1 % [162]. When compared to control groups, the risk of PTB and LBW was not different from IVF singletons (incidence of PTB of 14 % and of 13.6–14.0 %, respectively) and the risk of LBW was also not different from OD singletons (incidence of 14.0 %) [162]. However, a very recent US cohort study [164] underlined that the increased risk of PTB (aRR 1.14, 95 % CI 1.05 to 1.23) observed in gestational carriers is significantly influenced not only by multiple pregnancies but also by OD.